Department of Chemical Engineering The University of Newcastle Australia “THE CONTRIBUTION TO ATMOSPHERIC PARTICULATES OF ASH EMITTED FROM COAL FIRED POWER STATIONS” A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY By JAMES TREVOR HINKLEY BE (Hons) MEngSci JANUARY 2005
220
Embed
2005 HINKLEY PhD Thesis Atmospheric Particulates from Coal Fired Power Stations
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Department of Chemical Engineering
The University of Newcastle
Australia
“THE CONTRIBUTION TO ATMOSPHERIC PARTICULATES OF ASH EMITTED FROM COAL FIRED POWER STATIONS”
A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
By
JAMES TREVOR HINKLEY
BE (Hons) MEngSci
JANUARY 2005
ii
I hereby certify that the work embodied in this thesis is the result of original research and has not been submitted for a higher degree to any other University or Institution.
(Signed): ____________________________
James Hinkley
iii
ACKNOWLEDGEMENTS
The study described in this thesis was carried out in the Cooperative Centre for Coal in
Sustainable Development (CCSD) at the University of Newcastle. I wish to thank the
CCSD for their funding of the project, and the University for provision of a
postgraduate research scholarship.
I would especially like to express my sincere appreciation to my original triumvirate of
supervisors, Professor Terry Wall, Professor Peter Nelson and Associate Professor
Howard Bridgman. The wide variety of experience in this group was essential for the
successful development of a realistic project framework and our regular review
meetings were both a terrific stimulus and an invaluable focussing tool. Dr Raj Gupta
was also an important participant in these review meetings, and subsequently inducted
into my supervisory panel for his efforts.
I would also like to thank Dr John Carras from CSIRO Energy Technology for always
taking an interest in the project and his significant contributions while we were
developing the project scope. I am also indebted to many other CSIRO personnel,
notably Dr Brendan Halliburton for his experimental expertise and Drs Moetaz Attala
and Denys Angove for their helpful suggestions. Dr Bill Physick and Peter Hurley of
CSIRO DAR were also more than helpful with the TAPM modelling.
To my fellow fine particle postgraduate student Bart Buhre, many thanks for your
intellectual and social interaction, which has helped make returning to University so
enjoyable and rewarding. Thanks also to Joe Auberger, exchange student from Vienna,
for your contributions to the CCSD in general and my experiences in particular.
Thanks also to Dave Phelan, from the EM Unit of the University of Newcastle, for his
instruction and advice on the SEM analysis, which forms a significant part of this work,
and to Gary Weber, also of the UNEMU, who was a great help with the TEM imaging.
I am also grateful to Katie Levick, at the EMU at the University of New South Wales,
for a very productive and enjoyable session on the TEM at UNSW. I am also thankful
for the gentle guidance and assistance provided by a couple of friendly mathematicians
iv
at the University of Newcastle, Mr Kim Colyvas and Mr Frank Tuyl, who helped point
me through the statistical jungle to find information from mere data.
I wish to also thank the staff of the Discipline of Chemical Engineering University of
Newcastle for their help and support, particularly Robin D’Ombrain for building my
conditional power supply, Con Safouris, Gillian Hensman, John Wanders, Neil Gardner,
Steve “Richo” Richardson, Jenny Martin and Jane Hamson. And of course Leonie
Fuller who was so helpful in printing out my final drafts and my final document. And
particular thanks to Chris Wensrich, who pulled me out of a computational hole by
allowing me to use his super hyper threaded multi-chip P4 to get my TAPM modelling
completed before my scholarship ran out!
I would also like to express my sincere thanks to all at ANSTO, and especially Dr Ivo
Orlic, Eduardo Stelcer and Dr David Cohen for their dedication to work, enthusiasm
and friendly manner.
This project was reliant on support from power generators, and I would like to thank
Malcolm Rothe from Macquarie Generation, and Nino di Falco and fellow Piled high
and Deep student Mick Jensen of Delta Electricity for their open supply of historical
data as well as access to existing monitoring sites.
My thanks also to many at HLA Envirosciences, who facilitated my access to
Macquarie Generation’s monitoring sites and provided assistance with access to the gas
monitoring data and equipment. Key personnel were Graham Taylor and Paul Voigt at
the Warabrook office and Colin Davies, Dee Murdoch and Ben de Somer at the
Singleton office.
And finally, special thanks to my wife Tracey for your encouragement and support
throughout my PhD project. It’s been a rich and rewarding experience.
v
ABSTRACT
A comprehensive study has been conducted to assess the contribution of emissions from
coal fired power stations to ambient particles near two large stations in the Hunter
Valley region in New South Wales, Australia. Fine particles are an active area of
research with many studies revealing statistical associations between concentrations of
atmospheric ambient particles and mortality and other health impacts. A review of the
wide body of literature in this area concluded that, while coal fired power stations were
a significant anthropogenic source of fine particles, little information was available on
the contribution that these emissions make to ambient particles. Key characteristics of
interest identified were the contribution that emissions make to fine particulate mass,
aerosol chemistry and ultrafine particles.
Sampling was conducted at an existing monitoring site at Ravensworth, approximately
11 km to the SE of the stations. Primary particulate emissions were targeted as minimal
gas to particle conversion was expected so close to source. Samples were collected
between June 2002 and March 2004 using a Burkard spore sampler to estimate the
contribution to particulate mass, a cascade impactor to assess the contribution to aerosol
chemistry and a Nanometer Aerosol Sampler to allow the characterisation of ultrafine
particles in the air. Sulphur dioxide measurements at the site were used as a plume
indicator for conditional sampling to target the contribution of emissions.
Air pollution modelling using a commercial package (The Air Pollution Model or
TAPM) was also used to estimate the maximum contributions of power station
particulate emissions to the ambient aerosol and to study dispersion patterns. This was
complemented by a review of historical data; both sets of information sources indicated
that events were episodic and related to the breakdown of overnight atmospheric
stability due to solar heating of the ground.
Power station primary particulate emissions were found to make only a minor
contribution to ambient particulate mass, with episodic events of comparatively minor
significance. Maximum contributions to PM10 (particulate matter with an aerodynamic
diameter less than 10 µm) predicted by TAPM at the Ravensworth site were 2.3 µg m-3.
Results from the Burkard spore sampler were consistent and indicated a maximum
vi
estimated contribution from particulate emissions between 1 and 10 µm of 0.4 µg m-3.
The aerosol at the site was dominated by other sources such as windblown dust.
However, it was found during analysis of the cascade impactor results that power
station emissions also contributed significantly to the mass of particles less than 1 µm,
and that this mass was potentially more significant than primary particulate emissions.
Six size fractions from the cascade impactor ranging from 2.5 µm to less than 0.3 µm
were analysed using Ion Beam Analysis to provide high sensitivity analysis over a wide
elemental suite. The resulting elemental mass concentration data was interpreted using
factor analysis to extract 5 sources, including soil, salt, diesel and an industrial source.
A coal fired power station source was also extracted, concentrated primarily in the size
fraction less than 0.3 µm, and associated with the elements sulphur, chlorine, chromium,
nickel and copper. The average mass contribution of the power station component to
the samples collected at an average sulphur dioxide concentration of 46 ppb was 2.0 µg
m-3, approximately three times the estimated contribution of primary particulate
emissions based on pro rata dilution of the stack emissions. Note that these impacts are
the direct impact of the plume, and do not include background contributions of prior
emissions to secondary particulates.
Transmission Electron Microscopy studies of the fine particles confirmed the presence
of significant quantities of particles which were unstable under the electron beam,
consistent with literature descriptions of sulphate species. The appearance and nature of
the residues of these sublimated particles indicated varying neutralisation and water
association. Calculations based on source emission data suggested that this material
was probably formed from primary emissions of sulphuric acid due to the presence of
SO3 in the power station stack gases rather than through gas to particle conversion.
While emissions are therefore expected to have only a minor and intermittent
contribution to the ambient aerosol even relatively close to the power stations, some
uncertainty remains in the contribution of emissions to the minus 0.3 µm size fraction.
Additional characterisation work is recommended to clarify the extent, composition and
nature of this material and understand the relevant atmospheric chemistry in terms of
the rate of conversion to sulphuric acid, ammonium sulphate and other sulphate species.
Larger sample masses would reduce analysis uncertainties and permit investigation of
vii
the indicated but statistically unproven association of transition metals with the fine
particulate mass derived from power station emissions.
viii
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ............................................................................................. iii
ABSTRACT ...................................................................................................................... v
TABLE OF CONTENTS ............................................................................................... viii
LIST OF TABLES ......................................................................................................... xiii
LIST OF FIGURES ........................................................................................................ xv
1 INTRODUCTION AND STUDY OBJECTIVES ................................................... 1
Despite significant improvements in emissions controls, air pollution poses a major risk
to human health even today. It has been estimated that as many as 2,800,000 people die
annually from exposure to high concentrations of suspended particles in the indoor
environment, mainly associated with domestic cooking and heating with poor
ventilation in developing countries (WHO, 1999). Ambient air quality is also a
significant problem, with the excess mortality due to suspended particles and sulphur
dioxide in ambient air estimated at around 500,000 (WHO, 1999).
These estimates are based on correlations between health statistics and measured
ambient concentrations of pollutants. Fine airborne particulate matter has been linked
with disease and mortality in many studies. One of the earliest and best known studies
was the “Six Cities Study” (Dockery et al., 1993), which showed statistically significant
increases in mortality with increasing airborne fine particulate mass. Recent reviews of
available data have suggested that particles with an aerodynamic diameter less than
2.5 µm (PM2.5) have the most acute impacts (WHO, 2003). The European Union and
the United States are currently reviewing proposals to tighten PM2.5 air quality
guidelines (USEPA, 2003b; CAFE, 2004).
These findings have focussed attention on the various sources that contribute fine
particles to the atmosphere. Particles can be grouped into “primary” particles formed at
source, and “secondary” particles formed by transformation of gaseous pollutants in the
atmosphere e.g. ammonium sulphate. Primary particles are generally formed by either
combustion or mechanical processes. Mechanical processes – for example erosion,
agriculture and mining - tend to produce coarser particles than combustion processes
(Wilson and Suh, 1997).
Combustion aerosols have received particular attention as they are generally very fine:
80-90% of the particulate matter (PM) mass emitted from agricultural burning, wood
stoves, diesel trucks and crude oil combustion is less than one micrometer in diameter
(Lighty et al., 2000). Limited data is available for coal-fired power station emissions,
2
but it would appear that primary particulate emissions are coarser than for other
combustion processes (McElroy et al., 1982; Meij et al., 1985); this will be discussed in
greater detail in the literature review. One of the key differences is that emissions from
most combustion processes are unburnt carbon whereas primary emissions from power
generation are largely particles derived from the mineral matter in the coal which have
evaded emission control devices (Meij et al., 1985).
Internationally, coal combustion has received considerable attention, as large-scale
electricity generation is one of the major anthropogenic sources of airborne particulates
(Wolf and Hidy, 1997). It has also been shown that the surfaces of these particles are
enriched in potentially toxic elements (Linton et al., 1976; Linton et al., 1977; Mamane
et al., 1986), and that combustion particles in the fraction smaller than 2.5 microns from
mobile and coal combustion sources, but not fine crustal particles, are associated with
increased mortality (Laden et al., 2000).
In NSW, anthropogenic PM10 emissions are dominated by fugitive emissions from coal
mining, with coal fired electricity generation accounting for 8-12% of the total in recent
years (NPI, 2002). Emissions from coal fired power stations are therefore potentially
significant contributors to ambient particulate matter. While the contribution of power
stations to the emission inventory can be readily estimated, the significance in terms of
ambient particulate matter is comparatively poorly understood.
The Upper Hunter Valley was selected as the preferred location for a case study to
assess the contribution of power station emissions to ambient particulate matter. The
area has two large, modern coal fired power stations which supply approximately 40%
of the electricity for the state of New South Wales (DUAP, 1997). These stations are
both equipped with current best practice emission control devices in the form of fabric
filters (Heeley, 2001). Several previous studies have considered other pollutants in the
area such as fugitive dust from mining and sulphur dioxide, and one study was found
when one of these stations was equipped with less efficient electrostatic precipitators
(ESPs) for emission control.
3
1.2 OBJECTIVES
This study assesses the contribution of power station particulate emissions to ambient
particulate matter, in an Australian context. The study was based on sampling in the
Upper Hunter Valley, with dispersion modelling used to validate the sampling site and
extrapolate the results to nearby urban areas of interest.
The objectives of the study were to:
• Summarise the current state of knowledge about the contribution of coal fired
power station emissions to ambient particulate matter;
• Determine key aerosol characteristics of interest;
• Develop and implement an experimental program including air sampling to
determine the contribution of power station emissions to ambient particulate
matter;
• Interpret results to evaluate the significance in urban centres near to the
sampling site.
1.3 THESIS OUTLINE
The thesis has been organised into 7 chapters, which can be briefly summarised as
follows:
• Chapter 1 provides an overview of the thesis
• Chapter 2 reviews the current understanding of the contribution of coal fired
power generation to ambient particulate matter and refines the study objectives.
• Chapter 3 reviews potential experimental equipment and explains the
experimental methodology employed in this study.
• Chapter 4 describes the results of analysis of historical monitoring data at the
study site and compares measured data with the predictions of dispersion
modelling.
• Chapter 5 presents the results obtained from the various facets of the sampling
program.
• Chapter 6 integrates the findings of the individual sampling programs and
assesses the implications of the work including dispersion modelling.
• Chapter 7 concludes the thesis and makes recommendations for future research.
4
2 LITERATURE REVIEW
2.1 INTRODUCTION
This chapter provides a review of the current state of knowledge regarding the
contribution of power station emissions to atmospheric fine particles. Six main areas
will be discussed to both provide sufficient background for the issue and review the
available literature in the specific area. These are:
• Health studies and air quality legislation;
• Characteristics of airborne particulate matter – sources, chemistry, size
• Characteristics of power station emissions – formation, chemistry, size
• Dispersion of emissions from point sources
• Previous studies on the significance of power station emissions
• Overview of Upper Hunter Valley – sources, previous studies
This will be followed by a summary of the gaps in knowledge and the refined thesis
objectives.
2.2 HEALTH STUDIES AND AIR QUALITY LEGISLATION
2.2.1 Fine Particles and Human Health
Air pollution is by no means a modern phenomenon. Classical writers report the
oppressive fumes of Rome, while 19th Century London was once infamous for its “great
stinking fogs” (Brimblecombe, 1987). In more recent times, well publicised
catastrophes such as the 1952 London fog event, when an estimated 4,000 extra deaths
were attributed to an extreme pollution event (Brimblecombe, 1987), have resulted in
significant changes in the perception of air pollution and its potential for effects on
human health.
Airborne particulate concentrations are usually expressed as one of the following
(WHO, 1999):
• Total Suspended Particulates (TSP) – all airborne particles;
• PM10 – particles with an equivalent aerodynamic diameter of 10µm or less;
• PM2.5 – particles with an equivalent aerodynamic diameter of 2.5µm or less.
5
The most common measures in current use are PM10 and PM2.5. Monitoring networks
have been established around the world for some years now to monitor ambient
concentrations of particles either or both of these sizes, enabling correlation with health
statistics. One of the first studies to do this was the “Six Cities Study” mentioned
previously, which examined the correlation between mortality and disease rates with
particulate matter in six US cities (Dockery et al., 1993). Statistically significant
associations were found between exposure to PM10 and PM2.5 and increased mortality.
Subsequent studies have confirmed both the findings of this study (Krewski et al., 2000)
and have reported similar findings for other populations e.g. (Wichmann et al., 2000).
While epidemiological studies have demonstrated correlations between disease and the
amount of particulate matter in the air, understanding of the toxicology and even
conclusive causality remain incomplete (Smith and Sloss, 1998). However, some
studies have concluded that the finer particles are more significant in terms of health
effects – mortality in six US cities was found to be more strongly correlated with PM2.5
than PM10 (Schwartz et al., 1996). Recent appraisal of available data by a WHO
working group concluded that “fine particles (commonly measured as PM2.5) are
strongly associated with mortality and other endpoints such as hospitalization for
cardio-pulmonary disease” although it was also noted that a “smaller body of evidence
suggests that coarse mass (particles between 2.5 and 10 µm) also has some effects on
health.” (WHO, 2003)
The increased risks associated with PM2.5 are believed to be a result of the finer
particles being more able to elude the body’s protection mechanisms and penetrate into
the lungs. Most particles larger than10µm and 60-80% of 5-10µm are trapped in the
nose and upper respiratory tract and are expelled naturally from the body (Smith and
Sloss, 1998). About 60% of particles less than 0.1µm are deposited in the lung, where
they accumulate because the lung is unable to clean itself. What happens to particles
from there is the subject of much debate and research. A recent review of selected data
on the major and minor component composition of PM2.5 and PM10 concluded that there
was “little support for the idea that any single major or trace component of the
particulate matter is responsible for the adverse effects” (Harrison and Yin, 2000). The
authors also concluded “there are, if anything, too many plausible mechanisms and too
little established fact”.
6
Lighty et al. (2000) note that the “epidemiology and toxicology of ambient PM is an
active area of research”. They summarised the demonstrated and suspected “bad
actors” in atmospheric particulate matter into 11 categories, as shown in Table 2-1
(Lighty et al., 2000). These range from measures of overall mass concentrations to
particular compounds and species.
Table 2-1:Toxicological hypotheses for impacts of airborne particulate matter on human health (Lighty et al., 2000).
1. PM Mass Concentration . The initial epidemiologic studies correlated effects with mass as measured by ambient monitoring procedures.
2. PM Particle Size / Surface Area . Stronger associations are seen with fine particle mass, and the body interacts with the surface of an insoluble particle, not the volume.
3. Ultrafine PM. Particles smaller than 0.1 µm dominate the total number of particles in urban aerosols. Ultrafine particles are deposited deep in the lung by diffusion.
4. Metals . Transition metals including Fe, V, Cu and Ni can act as catalysts in the formation of reactive oxygen species (ROS) or activate biochemical processes.
5. Acids . Inhalation studies have shown toxic responses that are associated with the amount of H+ delivered to respiratory surfaces.
6. Organic Compounds . Volatile and semi-volatile organic chemicals associated with particles can act as irritants/allergens. Many aromatic compounds are carcinogenic.
7. Biogenic Particles . Pollen, spores and proteins are known allergens. Ambient PM also includes viable bacteria and viruses, as well as other biologically generated compounds.
8. Salt and Secondary Aerosols . Soluble salts formed by ocean spray and by gas-to-particle conversion are thought relatively benign, although implicated indirectly by mass.
9. Peroxides . Ambient peroxides associated with particles may be transported into the lung and may cause oxidant injury.
10. Soot . Carbon black has been shown in laboratory studies to cause tissue irritation and promote toxic formation. Soot particles also act as carriers for organic compounds.
11. Cofactors . The combination of two or more pollutants may cause greater or different effects than the individual pollutants acting separately.
Given the uncertainty over the “bad actors” in fine particulate matter, any study into the
contribution of power station emissions must endeavour to provide information about as
many of these potential hypotheses as possible. The most pertinent potential “bad
actors” from the above table fall into three main categories:
•••• airborne particulate mass;
•••• airborne particulate chemistry; and
•••• particle size distribution, particularly the ultrafine component.
2.2.2 Air Quality Legislation
Air pollution legislation dates back to at least as early as the 13th century
(Brimblecombe, 1987). However, legislation has evolved rapidly over the last 30 years
or so as epidemiologic studies have progressively linked health outcomes with TSP,
PM10 and PM2.5. Some relevant International and Australian air quality standards,
7
showing the trend towards tighter controls and finer particle sizes, are shown in Table
2-2. Note that the recent proposals to tighten PM2.5 limits and introduce PM2.5-10
standards in the US (USEPA, 2003b) and EU (CAFE, 2004) have not been legislated as
yet and the figures quoted are guidelines put forward by working groups after reviewing
available epidemiologic data.
Table 2-2: Selected international and Australian ambient air quality standards.
Country Year Type Details USA 1970 Total Suspended
Particulates (TSP) limits prescribed
TSP: primary (health based) annual arithmetic mean 75 µg m-3
24 hr maximum 260 µg m-3 USA 1987 PM10 replaces TSP PM10 annual arithmetic mean 50 µg m-3
PM10 24 hr maximum 150 µg m-3 India
1994 TSP and “respirable particles” i.e. PM10
(CPCB, 1994)
Residential, rural and other areas: TSP annual arithmetic mean 140 µg m-3 TSP 24 hr maximum (98th percentile) 200 µg m-3 PM10 annual arithmetic mean 60 µg m-3 PM10 24 hr maximum (98th percentile) 100 µg m-3
Thailand 1995 TSP & PM10 standards (PCD, 1995)
TSP annual geometric mean 100 µg m-3 TSP 24 hr average 330 µg m-3 PM10 annual geometric mean 50 µg m-3 PM10 24 hr average 120 µg m-3
USA 1997 PM10 requirements retained, PM2.5 standards added 40 CFR Part 50
PM10 annual arithmetic mean 50 µg m-3
PM10 24 hr maximum (99th percentile) 150 µg m-3
PM2.5 annual arithmetic mean 15 µg m-3
PM2.5 24 hr maximum (98th percentile) 65 µg m-3 Australia 1998 Australian NEPM limit for
PM10 Goal: PM10 24 hr maximum 50 µg m-3 exceeded no more than 5 times per annum by year 2008. Equates approx to 99th percentile adopted by USEPA.
European Union
1999 PM10 target levels set (CEU, 1999)
PM10 annual arithmetic mean 40 µg m-3 by 2005 and 20 µg m-3 by 2010 PM10 24 hr average not to exceed 50 µg m-3
more than 35 times a calendar year by 2005 and 7 times by 2010.
Australia 2003 Australian NEPM varied to include monitoring and reporting of PM2.5
Monitoring and reporting to commence 2004 Goal: PM2.5 annual average 8 µg m-3 Goal: PM2.5 24 hr average 25 µg m-3
USA 2003 Staff paper recommends modified PM metrics – not legislated (USEPA, 2003b)
Recommended changes: PM2.5 limits to be reviewed downwards 24 hr maximum (98th percentile) 30-50 µg m-3 Annual standard 12-15 µg m-3 PM10 to be replaced by PM2.5-10 as coarse mode metric 24 hr maximum (98th percentile) 30-75 µg m-3 Annual standard 13-30 µg m-3
European Union
2004 CAFE working group recommends PM2.5 as principal metric (CAFE, 2004)
Goals to be refined but approximate values as follows: PM2.5 long term mean in range 12 to 20 µg m-3
PM2.5 24 hr maximum around 35 µg m-3 (not to be exceeded more than 10% of the days of the year) PM10 target to be reassessed
In Australia, PM10 standards were set as late as 1998: a 10-year goal to exceed a daily
average of 50 µg m-3 no more than 5 times per year (NEPC, 1998). Australia has also
undertaken other initiatives to improve air quality including the following:
8
• A network of 24 PM2.5 monitors was established in the Newcastle, Sydney and
Wollongong regions in 1994 with funding from the Energy Development
Research Council. Samples from this network continue to be collected under the
Australian Nuclear Science and Technology Organisation’s (ANSTO) Aerosol
Sampling Program, providing data for chemical analysis of the aerosol and
source apportionment (Cohen et al., 1996)
• The Commonwealth government through Environment Australia now requires
industries to predict and or measure and report criteria pollutants including
PM10. This information is held in an on-line database: the National Pollutant
-) 20d a From land clearing and agricultural management practices b Includes steel, alumina, lime, gypsum and coke production; petroleum refining and municipal waste incineration c If secondary production as nitrate and condensed organic material is added, this total has an upper limit of about 370 Tg yr-1 d If sulphate in the atmosphere is assumed to be NH4HSO4 rather than SO4
2-, the sulphate contribution would amount to about 140 Tg yr-1; nitrates as NH4NO3 would be about 2 6 Tg yr-1.
In Australia, electricity supply (dominated by coal fired power stations) accounted for
12.1% and 7.7% of total anthropogenic PM10 emissions in NSW for the 2000/2001 and
2001/2002 reporting years respectively according to the NPI (www.npi.gov.au). Coal
mining was the single greatest source, accounting for 43.4% and 50.9%, while motor
10
vehicles contributed 9.5% and 8.5% of the PM10 for the same periods (note that mining
emissions are not included in Table 2-4). As will be discussed below, these two sources
produce very different particle sizes: motor vehicle emissions are typically less than 1
µm (Wilson and Suh, 1997) while dust from mining is significantly coarser with only 3-
6% of total emissions in the PM2.5 fraction (DUAP, 1997).
2.3.2 Characteristics of Different Sources
Different sources often have characteristic size distributions and chemistries. Figure 2-1
shows an idealised mass distribution of particle sizes found in the atmosphere (Watson
and Chow, 1994).
Figure 2-1: Idealised mass distribution of particle sizes found in the atmosphere (Watson and Chow, 1994).
Three modes of particles are identified in Figure 2-1:
• Nucleation: particles with diameters less than 0.08 µm that are emitted directly
from combustion sources. These particles rapidly coagulate or serve as nuclei for
cloud or fog droplets (Watson and Chow, 1994). This size range is detected only
when fresh emissions sources are close to the measurement site, for example close
to traffic corridors (Molnar et al., 2002).
• Accumulation: particles with diameters between 0.08 and ~2 µm. These particles
result from the coagulation of smaller particles emitted from combustion sources,
from condensation of volatile species, from gas-to-particle conversion, and from
finely ground dust particles. Figure 2-1 shows two modes within the accumulation
11
range attributed to a "condensation" mode with a peak at 0.2 µm containing gas-
phase reaction products, and a "droplet" mode that results from reactions that take
place in water droplets with a peak at 0.7 µm (Watson and Chow, 1994).
• Coarse: Particles larger than ~2 µm are called coarse particles; these result from
mechanical processes such as crushing and grinding and are dominated by material
of geological origin. Pollen and spores also inhabit the coarse size range, as do
ground-up trash, leaves, and tires. The finer end of the coarse particle size mode
includes particles formed when clouds and fog droplets form in a polluted
environment, then dry out after having scavenged other particles and gases (Watson
and Chow, 1994). Fly ash may also be found in this mode.
Note that the cut off between the accumulation and coarse modes is perhaps more
properly placed at around 1 µm as this is the upper size for particles formed through the
formation and growth of particles in the accumulation mode, as well as being the
minimum size for particles formed by breakage due to energetic limitations (CAFE,
2004). However, by convention, coarse particles are defined as particles larger than 2.5
µm and fine particles as those less than 2.5 µm.
Table 2-5 (adapted from (Wilson and Suh, 1997)) explains the formation and properties
of particles in more detail. Note the accumulation and nucleation modes referred to
above have consolidated into a “fine” mode, as normally found in urban ambient
particulate matter (Wilson and Suh, 1997).
Table 2-5: Sources and Properties of Fine and Coarse Mode Particles (Wilson and Suh, 1997).
Fine Mode Coarse Mode Formed from: Gases / combustion processes Large solids/droplets Formed by: Chemical reaction or vaporisation.
Nucleation, condensation on nuclei, and coagulation. Evaporation of fog and cloud droplets in which gases have dissolved and reacted
Mechanical disruption (crushing, grinding, abrasion of surfaces etc.). Evaporation of sprays. Suspension of dusts
Composed of: Sulphate, nitrate, ammonium and hydrogen ions. Elemental carbon. Organic compounds (e.g. PAHs). Metals (e.g. Pb, Cd, V, Ni, Cu, Zn, Mn, and Fe). Particle bound water
Crustal material. Coal and oil fly ash. Oxides of crustal elements (Si, Al, Ti, Fe). CaCO3, NaCl, sea salt. Pollen, mould, fungal spores. Plant/animal fragments. Tire wear debris
Solubility Largely soluble, hygroscopic and deliquescent
Largely insoluble and non-hygroscopic
Sources Combustion of coal, oil, gasoline, diesel Resuspension of industrial dust and soil
12
Fine Mode Coarse Mode fuel, and wood. Atmospheric transformation products of NOx, SO2 & organic compounds. High temperature processes e.g. smelters, steel mills
tracked onto roads and streets. Suspension from disturbed soil (e.g. farming, mining, unpaved roads). Biological sources. Construction and demolition. Coal and oil combustion Ocean spray
Atmospheric Half Life
Days to weeks Minutes to hours
Travel Distance
100s to 1000s of km <1 to 10s of km
Naturally the size distribution of the ambient aerosol is very dependent on location,
meteorology and local sources, and can vary substantially with season (e.g. due to the
use of domestic fires for winter heating). Figure 2-2 shows measured size distributions
at four cities in Australia, with varying contributions of the three particle size modes
(Ayers et al., 1999b). Most of the distributions show only two modes, although the
Canberra aerosol has some suggestion of a nucleation mode, possibly from wood fires
as these samples were taken in winter.
Figure 2-2: Mass distribution of urban aerosol in four Australian cities. Taken from (Ayers et al., 1999b).
2.3.3 Previous Studies on Aerosol Composition
2.3.3.1 International Studies
There are many studies in the literature on the composition of PM10 and PM2.5 for
various countries and regions; data from some of these studies have been collated by
Harrison and Yin (Harrison and Yin, 2000). This data indicates that PM2.5 in urban
13
areas tends to be dominated by carbon (typically 20-40%) and soluble species such as
NO3-, SO4
2- and NH4+ (highly variable, commonly ~40%), with relatively small
amounts of crustal material (4-15%). In contrast, the coarser components (PM2.5-10 or
PM2.5-15) tend to be dominated by crustal material (50-90%).
Similar results are reported in a recent review of EU monitoring (CAFE, 2004). PM2.5
in urban areas was found to consist mainly of elemental and organic carbon (20-35%)
and secondary organic aerosols (20-40%), with smaller amounts of crustal material (2-
20%) and marine aerosol (1-3%). Roadside sampling was broadly similar, although
higher levels of elemental and organic carbon were noted due to vehicle emissions.
PM2.5-10 was not separately reported, although PM10 samples contained more marine (5-
12%) and crustal material (10-30%) indicating that these particles were more prevalent
in the PM2.5-10 fraction.
2.3.3.2 Australian Aerosol Composition Studies
Studies in Australian urban areas have produced broadly similar results, although a
study in Perth, Western Australia indicated that suburban air contains relatively less
material of industrial origin than some urban areas in the US, with ammonium sulphate
contributing only a few percent to TSP (O'Connor et al., 1981).
A pilot study was commissioned by Environment Australia to better understand the
measurement techniques and characteristics of PM10 and PM2.5 in six urban areas in
different states (Ayers et al., 1999c). This study found that the results fell into three
subsets. The aerosol in Sydney, Melbourne and Adelaide was dominated by estimated
organic matter or EOM (60-80%), which increased in the finer fractions. Crustal
material accounted for a further 15-30%, decreasing in the finer fractions. Elemental
carbon made up 10% of the mass and increased in the finer fractions. EOM was
estimated as the difference between the total gravimetric mass and the sum of the
inorganic matter and elemental carbon. Canberra and Launceston showed a strong
effect of wood heating, with 75-85% of the mass reported as EOM and lower elemental
carbon (<10%). Brisbane was found to have approximately equal contributions from
EOM and inorganic matter (Ayers et al., 1999c).
14
A number of other studies have also been conducted in Brisbane. Most particulate
matter during typical high asthma incidence periods (autumn) was found to be less than
2 µm and composed of carbon from vehicle emissions, crustal material and some spores
and soil bacteria (Glikson et al., 1995). Fungal spores were found to dominate the 2 µm
to 10 µm size range (Mastalerz et al., 1998). Subsequent studies with dichotomous
samplers and a cascade impactor have shown that the PM2.5 and PM2.5-10 size fractions
show similar patterns to overseas studies, in that the crustal signature is most
pronounced in the larger size fraction while the fines are dominated by combustion
products and soluble salts (Chan et al., 1999b; Chan et al., 2000).
PM2.5 has also been sampled and analysed at Muswellbrook in the Upper Hunter Valley,
New South Wales to determine composition and origin (MSC, 2003). Figure 2-3 shows
the average composition of the aerosol over the 2002-2003 monitoring period for a site
near the local water treatment plant. The aerosol is dominated by ammonium sulphate
and organics with some soot, crustal material and salt. It is interesting to note the
significant contribution (24%) of ammonium sulphate to PM2.5 mass (total ~ 7 µg m-3).
Figure 2-3: Breakdown of PM2.5 for Muswellbrook, 2002/2003 (MSC, 2003).
2.4 CHARACTERISTICS OF POWER STATION EMISSIONS
Power station emissions contribute to airborne particulate matter through both primary
and secondary particles. Primary emissions are derived from the mineral matter in the
coal that has formed ash and escaped to the atmosphere by evading the emission control
devices. These emissions are quite different to particles formed by other combustion
processes, which are dominated by unburnt carbon or soot (Lighty et al., 2000).
15
Power stations also emit significant quantities of the oxides of sulphur and nitrogen,
which form secondary aerosols through oxidation in the atmosphere after emission. At
least 90% of the sulphur in the coal enters the gas phase during combustion as SO2
(Hewitt, 2001), with around 1-3% emitted as SO3 (Graham and Sarofim, 1997). High
temperature combustion processes also produce NOx (NO and NO2) from nitrogen in
the coal (fuel nitrogen) and from oxidation of N2 at combustion temperatures (Pershing
and Wendt, 1979). Oxidation rates of SO2 and the formation of secondary aerosols will
be discussed further in Section 2.6 which reviews previous studies assessing the impacts
of power stations.
Primary particulate emissions can be expected to have an impact on local ambient
particulate matter. In contrast secondary aerosols formed from power station gaseous
emissions are expected to have a more regional impact on background PM levels, as
these gases travel tens or hundreds of kilometres before being oxidised to produce
secondary particles (Hidy, 1994; Hidy, 2002).
The current section will concentrate on primary particulates and review ash formation
mechanisms and emission controls before presenting a summary of studies that have
characterised stack emissions.
2.4.1 Ash Formation Mechanisms
Power stations utilise coal by combustion with air in a furnace to heat water and
produce steam, which is used to drive turbines that generate electricity. Coal contains
mineral matter, which forms ash upon combustion. Some of the ash deposits in the
furnace, but around 80% (Wall et al., 1982) is carried out of the furnace with the
combustion gases. This ash is termed “fly ash”, which is a slightly misleading term as it
is not the part of the ash emitted to the environment. Most of the fly ash is removed
from the waste gases by emission control devices such as electrostatic precipitators or
fabric filters before reaching the stack (Carr and Smith, 1984; Meij et al., 1985).
Ash characteristics vary greatly for different coals, due to the impact of mineral matter
composition and distribution within the raw coal and its subsequent behaviour upon
combustion (Wibberley and Wall, 1986). Most coal burnt in NSW power stations is
sub-bituminous (“Black coal”); lower rank coals such as lignite are not as mature and
16
have much higher moisture and lower calorific value (Smoot, 1991). The current
understanding of ash formation mechanisms from black coal is illustrated in Figure 2-4.
Excluded Minerals
Fragmentation
Coal Particle with included minerals
Vaporisation
Char Fragmentation
Fusion, melting
Coalescence
Cenosphere Formation
Fragmentation
Proces s during Combustion
Cooling
Boiler input
Swelling Char
Non - Swelling Char
Heterogeneous Condensation
Homogeneous Condensation
< 30 µ m
10 - 90 µ m
30 µ m
0.02 – 0.2 µ m
Surface Enrichment
< 30 µ m
Shedding
< 1 µ m
Figure 2-4: Mechanisms of Fly ash Formation (Buhre et al., 2001).
The mechanisms can be summarised as follows (Raask, 1985; Wibberley and Wall,
1986):
• Excluded mineral matter (no combustible material)
o particle may fragment into smaller particles
o mineral matter fuses to form single particle (silica may not fuse)
o particle may be hollow due to gas formation from decomposition
• Included mineral matter (contained within coal particles) – this is released
during combustion of the char after pyrolysis and volatile release
o non-swelling coals:
17
� char tends to burn as shrinking core, ash droplets form on particle
surface, some coalesce
o coals that swell and fragment (form finer ash):
� porous char, may fragment
� individual fragments burn as shrinking core (smaller than with
non-swelling)
� may produce hollow char cenospheres, where agglomeration of
ash droplets is delayed, producing numerous finer ash particles
o vaporised elements – re-condense as temperature decreases
� heterogeneous condensation - on existing particles (surface
enrichment)
� homogeneous nucleation/coalescence – fume production
Physical characterisation of fly ash particles has been conducted by a number of
authors. Ramsden and Shibaoka (1982) identified seven categories of particles
Figure 2-6: Cumulative particle size distribution of emissions for dry bottom boilers burning pulverised bituminous and sub-bituminous coal with various
controls. Data sourced from US EPA Table 1.1-6 (1995).
Data published by the US EPA relating to the size distribution of emitted particulates
from utilities using ESP and fabric filter plants is shown in Figure 2-6, together with the
21
uncontrolled emissions or feed fly ash to the emission control device (USEPA, 1995).
This data differs from that of McElroy et al. (1982) in that the fabric filter emissions are
finer than those with ESP, with 25% less than 1 µm compared to 14% for ESP.
Figure 2-6 also illustrates the relative enrichment in the finer particles resulting from the
efficiency limitations of both control devices. For example, 92% of fabric filter
emissions were less than 10 µm compared to 23% upstream of the filter. Note that the
emissions for the ESP case in Figure 2-6 are considerably coarser than those found in a
survey of Dutch power plants, where the 90% passing size was found to be between
around 3 and 5 µm (Meij et al., 1985).
The impact of pollution control equipment on the size and chemistry of emissions will
be discussed further in the following sections.
2.4.3 Characteristics of Emitted Particulates
Power station particulate emissions have been extensively studied and characterised
over the years in and around ESPs. Only one study has been found reporting size and
chemistry information for a fabric filter (McElroy et al., 1982). Table 2-6 summarises
the key findings of a number studies on samples of particulate emissions from various
installations, most equipped with ESPs. Significant variations are reported in the mean
particle size, although the existence of the evaporation/condensation mode has been
confirmed by several studies. Only one Australian study was found; this study
presented morphology and size information only for a plant burning lignite, with an
unusually high mean diameter (Zou, 2000). The particle size, morphology, chemistry
and surface enrichment of particular elements will be discussed separately in subsequent
sections.
Table 2-6: Particle Size and Morphology of Emitted Particulates.
Reference: Study location; Coal type
Emission Control; sample point & method
Particle Size Information Morphology
(Zou, 2000) Australian; Victorian lignite
ESP; Outlet
Reported mean 21 µm (Malvern laser analyser).
Variable – glassy spheres to irregular aggregates. Aggregates ~10 µm the most common particles.
(Kauppinen and Pakkanen, 1990)
Finnish; Polish bituminous
ESP; In-stack; 11 Stage Cascade impactor
Range 0.01 to 11 µm (Stokes diameter). Bimodal with geometric mass means about 0.05 and 2 µm.
Not reported (trace element study)
22
Reference: Study location; Coal type
Emission Control; sample point & method
Particle Size Information Morphology
(Mamane et al., 1986)
US; Unspecified
ESP; In stack; Dichotomous sampler with Teflon filters
Two fractions: minus 2.5 µm and 2.5 to 5-10 µm aerodynamic diameter. Mass split approx. 15/85%
>95% spherical with rather smooth surfaces
(Meij et al., 1985)
Netherlands; Bituminous (US&Aust)
ESP; In-stack; Anderson Mk III Cascade Impactor
Aerodynamic diameter of all particles < 10 µm; 90% less than 6 µm; mass median 1-2 µm
Spherical, density about 2.7g cm-3
(Lichtman and Mroczkowski, 1985)
US; two plants using high/low sulphur coal
ESP; In-stack; Anderson cascade impactor
High S: peak ~ 2 µm; 4% larger than 8 µm. Low S: bimodal, peaks at 6 and 0.7 µm. 4% larger than 30 µm.
Submicron examined with SEM/EDA. Spherical, solid particles for both coals, some surface nodules (more common with High S coal).
(McElroy et al., 1982)
US, 25MW boiler, sub bituminous
Fabric Filter 8% of emissions <2 µm, 0.5% <0.3 µm
Not reported
(McElroy et al., 1982)
US, 5 other boilers from 113-540MW
ESP Bimodal size distribution measured at 540MW boiler. 4 to 20% of emissions <2 µm, 0.2 to 2.2% <0.3 µm
Not reported
(Fisher et al., 1978)
US; Unspecified low S, high ash & TM
ESP; Stack; Cyclone, centripeter & filter.
Range 1 µm to 60 µm; classified into 4 size fractions with cyclones
winter. Overnight drainage flows still occur but are weaker with average speed of
1.6 m s-1. Sea breezes from NE and E occur on about 1/3 of days, starting in the late
morning and lasting up to 13.5 hours. Irregular cool changes shift wind direction to
the SW (“southerly buster”).
2.7.2 Hunter Valley Studies – Sulphur Dioxide & Aci d Rain
There have been a number of studies over the last 20 years looking at the impact of
various industries and activities in the Hunter Valley on the environment. Studies on
power station emissions have been primarily concerned with sulphur dioxide emissions
and will be reviewed in this section. Studies on dust emissions in the Hunter Valley
have concentrated on emissions from coal mining operations and will be reviewed in the
next section.
Several studies were conducted to assess the impact of power station emissions around
the commissioning of Bayswater power station in the mid 1980s. Table 2-10
summarises some of the publications from these studies, which focussed primarily on
ground level concentrations (“glc”) of SO2. These studies confirm the importance of
inversions and trapping of pollutants for high concentration episodes, and indicate the
inability of Gaussian models to accurately model such events.
Table 2-10: Upper Hunter SO2 Emission Studies.
Study & Area Summary and key findings (Chambers et al., 1982) Liddell, middle Hunter Valley
Various models used to determine SO2 glc’s from Liddell including Gaussian and trapping model. Concluded that understanding of prevalent atmospheric boundary layer conditions critical to determine which model appropriate. Highest glc’s found under trapping (inversion) conditions, underestimated by Gaussian approach. Trapping model better, but underestimated decrease with distance.
(Chambers and Bridgman, 1983) Liddell, middle Hunter Valley
Gaussian model with local spreading coefficients most appropriate for predicting weekly glc’s from Liddell (58% of the time within a factor of 2). Pasquill-Gifford spreading coefficients found to be unrepresentative of middle Hunter region (38%). Error attributed to uncertainty in wind speed, plume rise and other terms.
39
Study & Area Summary and key findings (Jakeman and Simpson, 1987) Hunter Valley
Gaussian model with trapping used to assess potential locations for further power stations. Bayswater/Liddell plume produced highest concentrations in line with prevailing winds i.e. NW-SE.
(Physick et al., 1991) middle Hunter Valley
Prognostic wind field/Lagrangian particle model approach (a la TAPM) used to predict SO2 glc’s and results compared to Gaussian model. Found that the prognostic model predicted wind fields well and performed considerably better than Gaussian model under fumigation conditions in particular.
(Carras et al., 1992) Hunter Valley & Central Coast
Plumes from Bayswater and Liddell mainly travelled down valley under influence of NW wind in winter. Plumes normally merged within ~10 km from sources. Central Coast plume behaviour very complex and poorly described by simple models due to terrain and presence of sea breezes in summer months. Gaussian plume models generally OK - Plume spreading coefficients developed for stable and convective conditions; plume rise conformed to Briggs’ formula in stable but not convective conditions. Bayswater/Liddell in-plume peak SO2 ~40 ppb at Muswellbrook, <25 ppb at Newcastle.
Factor analysis has been used to evaluate the contribution of various sources to
rainwater contamination in the Hunter Valley (Bridgman, 1992). Soil and
animal/fertiliser sources were found to be the main sources that determined water
quality over most of the Hunter Region. Industrial sources contributed 10 to 47% of
observed variance, with the highest results in the mid Hunter (between Singleton and
approximately 20 km to the east). It was also concluded that local sources were more
significant than long-range transport of pollutants from the Sydney basin 175 km to the
south.
In summary, the meteorology of the Hunter Valley has been well characterised and SO2
has been shown in a number of studies to be a suitable indicator of emissions from
power stations. High SO2 concentrations appear to arise from trapping of pollutants
through overnight inversions and solar heating bringing the plume to ground (Chambers
et al., 1982).
2.7.3 Hunter Valley Studies – Airborne Dust
A significant amount of research was conducted in the mid to late 1980’s to assess the
impact of coal mining operations on airborne dust in the Hunter Valley. The most
comprehensive of these was the NERDDC “Air Pollution from Surface Coal Mining”
study, published in three volumes in 1988-1989 (NERDDC, 1988). Key findings of this
study, which included extensive community surveys, were:
• Air pollution from surface coal mining was a significant community concern;
40
• There was a significant correlation between community perceptions of dust
problems and dust deposition rates, although inconsistencies were noted in
“nuisance” thresholds between individuals;
• Some survey respondents blamed dust pollution for health complaints such as
asthma, nasal congestion, sinus problems and lung complaints;
• A number of respondents who had moved to the area said they believed their
health had deteriorated as a result of the dust;
• A review of published data on size distribution of particulates emitted by various
mining activities indicated approximately 6% was less than 2.5 µm, 52% lay in
the range 2.5 to 15 µm and 42% was larger than 15 µm;
• Dust deposition rates (measured and modelled) decreased rapidly with distance
from the source, due to the rapid fall out of coarse particles;
• Power station particulate emissions were not considered.
The only two studies dealing with power station primary particulate emissions
specifically were discussed in Section 2.7. These studies are not believed to reflect the
current impact of emissions on ambient air quality for the following reasons:
• One of the studies considered dust deposition rather than ambient air quality
(Malfroy et al., 1993)
• The other study used dilution estimates rather than sampling to determine
contributions to airborne particulate mass, and was based on significantly higher
mass emission rates than current (Jakeman and Simpson, 1987).
In summary, while airborne dust has received considerable attention in the Hunter
Valley, the main focus has been on the contribution from mining activities. Past studies
on power station emissions have been limited to dust deposition and estimation of mass
contributions based on assumptions that are no longer valid. This study will address
this deficiency and assess the contribution of power station emissions to air quality in
the context of other sources. While mining emissions may be more significant in mass
terms, they are coarser and less likely to travel long distances.
41
2.8 GAPS IN KNOWLEDGE AND THESIS OBJECTIVES
2.8.1 Summary of Literature Review
While mechanisms remain the topic of intense research and debate, it is now widely
accepted that airborne fine particulate matter causes increased mortality and morbidity
(Lighty et al., 2000). Current legislation around the world usually prescribes air quality
guidelines for either or both PM2.5 and PM10. Combustion aerosols have received
considerable attention as they are relatively fine compared to other sources, with a high
proportion of particles less than one micron (Lighty et al., 2000).
Coal combustion is recognised as a major anthropogenic source of both primary and
secondary particulate matter in the air (Wolf and Hidy, 1997). Primary particulate
emissions from coal fired power stations have received particular attention due to the
decreased collection efficiencies of air pollution control equipment on the particles
formed by evaporation and condensation of certain elements under combustion
conditions (McElroy et al., 1982).
Previous studies into the significance of power station emissions in terms of ambient air
quality can be summarised as follows:
• Primary particulate emissions have been found to be a significant contributor to
TSP in one study (Fung and Wong, 1995) and a minor component in several
other studies including one in Brisbane, Australia (Chan et al., 1999a);
• Secondary particulates formed from the oxidation of power station emissions in
the form of SO2 and NOx can be a significant component of the aerosol (Querol
et al., 1999; Khosah and McManus, 2001), although oxidation rates in Australia
are slower than overseas due to lower levels of background pollution (Carras
and Williams, 1988; Ayers et al., 1999a).
• Power station particulate emissions show bulk and surface enrichment of
potentially toxic elements, notably transition metals (Linton et al., 1976;
Mamane et al., 1986);
• Soil sampling and analysis of biomonitors have indicated that power station
emissions can have a significant local impact on the alkalinity of the soil
(Padmanabhamurty and Gupta, 1977; Mehra et al., 1998) and the uptake of
transition metals and sulphur (Bohm et al., 1998; Sawidis et al., 2001).
42
Relatively few studies have considered the impact of coal fired power generation on air
quality within Australia. A number of indirectly related studies have looked at the
impacts of dust from coal mining (NERDDC, 1988), while most of the interest on the
utilisation side has been on sulphur dioxide (Physick et al., 1991). Only three studies
were found where power station emissions have been specifically assessed within
Australia:
• A study which assessed fly ash deposition rates in the vicinity of power stations
and concluded that the maximum contribution was less than 5% (Malfroy et al.,
1993);
• A study which assessed the contribution of power station emissions to trace
element deposition, which concluded that emissions were a significant but not
detrimental source of some elements (Swaine, 1994);
• A study which estimated the contribution of power station particulate emissions
to TSP, which concluded that the maximum hourly contribution of Liddell
power station emissions when equipped with less efficient ESPs was 90 µg m-3
(Jakeman and Simpson, 1987). Modern emissions controls have significantly
reduced mass emission rates.
In contrast, the Hunter Valley has been extensively studied to examine the impact of
both emissions from open cut coal mining (Bridgman, 1998) and the impact of gaseous
emissions from power stations, in particular SO2 (Carras et al., 1992). Understanding of
the meteorology of the area is relatively mature (Bridgman and McManus, 2000).
2.8.2 Gaps in Knowledge
While one study was found that identified fly ash in urban areas (Chan et al., 1999b),
there has as yet been no systematic study into the significance of particulate emissions
in regions adjacent to power stations in Australia since the installation of fabric filters at
Liddell. Epidemiological studies have identified the following as potential “bad actors”
in atmospheric particulate matter (Lighty et al., 2000):
Primary particulates appear to be of greater relevance to the Australian context due to
expectations of relatively slow oxidation rates near to the power stations (Williams et
al., 1981; Ayers and Granek, 1997). The goal of this project can therefore be refined to
develop and implement techniques and methodologies to enable the contribution of
power station primary particulates to the above areas to be assessed, and conduct a case
study. The Hunter Valley appears to be a suitable site for such a study given the body
of previous research into meteorology and dispersion of various pollutants (NERDDC,
1988; Carras et al., 1992), with the notable exception of primary particulates.
The goal of this project can therefore be redefined as:
To assess, through a case study in the Hunter Valley, the significance of the
contribution of primary power station particulate emissions to airborne particulate
matter as follows:
• Contribution to total particulate mass
• Contribution to aerosol chemistry
• Contribution to ultrafines
The study will also need to address the issue of temporal variations which are likely to
be critical given the episodic nature of events.
44
3 EXPERIMENTAL AND ANALYTICAL TECHNIQUES
3.1 OBJECTIVES & EXPERIMENTAL COMPONENTS
The objective of this study is to identify or develop and implement techniques to assess
the significance of the contribution of power station primary particulate emissions to
ambient particles. Three key aspects were identified in the literature review:
• Contribution to total particulate mass
• Contribution to aerosol chemistry
• Contribution to ultrafines
The selection process for the experimental equipment used will be dealt with briefly in
the next section to provide an overview of the project, followed by a more detailed
discussion of the methodology employed with each component.
3.2 REVIEW AND SELECTION OF SAMPLING TECHNIQUES
3.2.1 Determination of Contribution to Total Partic ulate Mass
3.2.1.1 Standard Methods for Determining Airborne P articulate Mass
Table 3-1 reviews standard methods employed to determine airborne particulate mass.
Table 3-1: Standard methods for determining airborne particulate mass.
Technique
Principle of Operation
Advantages
Disadvantages
Filter Based Sampling
Air is sucked through a filter, retaining virtually all particles (John and Reischl, 1978). Standard gravimetric method for determining mass concentration of airborne particulates (Sloss, 1998). Glass fibre filters are the most commonly used medium for mass determinations as they are robust, have low moisture retention and have high collection efficiencies (Sloss, 1998). Membrane filters are more suitable when subsequent microscopic or chemical analysis is required as they are thinner, have lower levels of trace elements and some media can be dissolved in organic solvents or nitric acid. These filters are widely used for receptor modelling studies (Sloss, 1998).
Can pre-classify to sample only the size fraction of interest. Samples can be subjected to chemical or SEM analysis if suitable membranes are selected. Widely used for receptor modelling.
Matching of flow and sample requirements: e.g. PM10 measurements are usually for a 24 hour period and so are not sensitive to individual events. SEM imaging is problematic at high loading, due to the inability to distinguish individual particles. Potential artefacts from interaction between reactive particles, gas-particle or gas-filter media reactions, loss of volatile compounds.
45
Technique
Principle of Operation
Advantages
Disadvantages
TEOM – Tapered Element Oscillating Microbalance
The system is based on an oscillating filter attached to the tip of a hollow, tapered, oscillating glass rod (Sloss, 1998). Accumulation of material on the filter changes frequency of oscillation, allowing a direct measurement of mass on the filter over time. By relating the increase in mass to the flow rate, the dust concentration can be determined every two seconds.
Provides a continuous measure of TSP, PM10 or PM2.5 allowing individual events to be identified as well as longer term trends.
Unable to differentiate between sources, limited to one size fraction at a time. Can be affected by loss of volatiles such as VOCs and ammonium nitrate (Ayers et al., 1999b; Green et al., 2001)
Light Scattering
Particles interact with monochromatic light or laser beams, reflecting, absorbing, diffracting and refracting them depending on the particle size and the wavelength of the incident light. Laser analysers use peak analysis to separate the particles into different size ranges (Sloss, 1998).
Provides a continuous measure of TSP, PM10 or PM2.5 allowing individual events to be identified as well as longer term trends.
Unable to differentiate between sources. Calibration of these devices is critical and based on gravimetric sampling; the calibration is therefore only valid while the nature of the particles does not change.
Beta Attenuation
Based on measurement of the reduction in intensity of beta particles passing through a dust laden filter, due to absorption of beta particles by the dust collected (and the filter material). The relationship between radiation absorbed and mass of dust collected closely follows an exponential relationship which is reasonably independent of the chemical composition of typical particulate material found in the atmosphere. Widely used in EU. (DEFRA, 2004)
Provides a continuous measure of TSP, PM10 or PM2.5 allowing individual events to be identified as well as longer term trends.
Unable to differentiate between sources, limited to one size fraction at a time. May be susceptible to water or VOC loss as with TEOM. Radioactive source. (DEFRA, 2004)
Cascade Impactors
Inertial impaction accelerates an aerosol through a nozzle directed at a flat plate. Particles with sufficient inertia are unable to follow the streamlines of the deflected airflow and impact on the plate. Smaller particles avoid hitting the plate and flow on with the air. A cascade impactor uses a progression of decreasing nozzle widths to progressively remove finer particles, producing a number of size fractions (Marple and Willeke, 1976). Backup filter used behind final stage.
Produces physical samples of various size fractions which can be weighed to provide mass loadings. Simple and robust.
Sample mass is a function of the volume sampled, which can be low compared to high volume gravimetric filter sampling.
The main disadvantages of the above methods in terms of this study can be summarised
as follows:
• While filter based sampling and cascade impactors produce physical samples
which can be analysed in bulk to determine sources, they offer limited temporal
resolution which is essential to investigate individual events;
• TEOM, Light Scattering and Beta Attenuation instruments offer superior
temporal resolution but only measure mass concentrations and cannot be
apportioned to sources.
46
These techniques were therefore not considered sufficient for the study and a new
approach was sought. A limited number of relevant studies were found in the literature
using the Burkard 7-day spore sampler, which will be discussed below.
3.2.1.2 Burkard Spore Sampler (Burkard, 2000)
The 7-Day Burkard Spore Sampler is widely used for collecting spores and pollen for
immunology (Razmovski et al., 1998). Particles are collected by inertial impaction on a
tape mounted on a rotating drum which completes one turn per 7 days. The tape moves
at a rate of 2 mm per hour or 48 mm per day, and is normally Vaseline coated cellulose
acetate. A picture of the sampler is shown in Figure 3-1.
Inlet
Figure 3-1: Burkard Spore Sampler(Burkard, 2000).
The large vane at the right is used to orient the inlet orifice towards the prevailing wind
direction. The sampler has a pump which draws air at a nominal 10 litres minute-1
(LPM) through an orifice 14 mm in length and 2 mm in width. The orifice can be
reduced to 0.5 mm to improve trapping efficiency in the 1-10 µm range (Burkard,
2000). The pump can be run off either mains supply or 12 V batteries for field use, and
the flow can be adjusted manually. The major advantage of this equipment is that the
particles are collected on a time resolved basis, allowing individual events to be studied.
The sampler has been used with the standard 2 mm slot to sample the urban aerosol in
London to provide temporal resolution of particulate loadings (Battarbee et al., 1997;
Mackay and Rose, 1998). Analysis based on light microscopy clearly showed an
47
increase in particulates in the morning and afternoon associated with traffic density at
rush hours (Mackay and Rose, 1998).
The sampler has also been used at the University of North Dakota, using the standard 2
mm slot and double sided carbon tape (Benson et al., 2001; Erickson et al., 2001).
Tapes were transferred to glass microscope slides for analysis by SEM analysis of
individual particles. Direct impaction on carbon tape offers significant advantages over
other sample preparation methodologies which could contaminate or alter the samples
(O'Keefe et al., 2000; Benson et al., 2001).
There is limited information available about the capture efficiency at different sizes, as
the sampler has primarily been used to collect spores and pollen, which are typically 10
to 35 µm (Frenz, 1999). Unpublished calculations by one of the co-authors of the
London study suggest that the device capture efficiency falls to below 50% for particles
less than 2 µm in diameter, although smaller particles are still captured - 59% of
particles counted were less than 0.5 µm (Mackay and Rose, 1998).
3.2.1.3 Analysis of Burkard Samples
As mentioned above, the samples from the spore sampler are best suited to microscopic
analysis. The two studies referred to above used quite different approaches to this. The
London traffic studies used optical microscopy (Battarbee et al., 1997; Mackay and
Rose, 1998) while the University of North Dakota study used SEM analysis (Benson et
al., 2001; Erickson et al., 2001).
Optical microscopy uses transmitted or reflected light to generate a visible light image
of the sample, and provides information about colour, surface texture and optical
properties (Cheng et al., 1976). However, resolution at fine particle sizes is limited by
the wavelength of visible light (0.4 to 0.7 µm), with the best light microscopes limited
to a resolution of about 0.2 µm (Culling, 1974).
Scanning electron microscopy (SEM) uses a very narrow, high energy electron beam
which scans across the surface of the sample (Swift, 1970). The electron beam interacts
with the sample and generates three emissions of interest:
48
• secondary electrons (SE): commonly used for imaging, these electrons produce
an image relating to the surface topography of the sample (Swift, 1970). Each
high energy primary electron in the incident beam produces many slow moving
(secondary) electrons as the primary electron collides with numerous atoms
along its path. Some of these electrons diffuse to the surface and are detected
and converted to an image by a scintillator/photomultiplier system (Swift, 1970).
• back scattered electrons (BSE): less commonly used for imaging, these
electrons are high energy primary electrons scattered with little loss of energy by
the sample. BSE images have been preferred in several previous fly ash studies
due to superior contrast between particles and background and some sensitivity
to atomic number due to the increased likelihood of interaction with larger
nuclei (Jalkanen et al., 2000; Benson et al., 2001).
• X-rays: some of the energy of the primary electrons is absorbed by the sample
through electron orbital transitions – decay back lower orbitals produce x-rays of
characteristic wavelengths which provide gross information about the chemical
composition of the sample (Swift, 1970).
SEM has several major advantages over optical microscopy for identifying particles:
• Image resolution is up to several orders of magnitude better than optical
microscopy, with resolution down to 10 nm possible with secondary electrons
(Swift, 1970);
• SEM offers superior textural resolution and has a much greater depth of field
enabling different sized objects to be in focus even though they are not on the
same plane (Goldstein, 2003);
• SEM-EDX chemistry information gives valuable data on particle composition
and possible origin. This information has been used at the University of North
Dakota to classify particles into groups (Benson et al., 2001).
However, SEM also has a several drawbacks which should also be noted:
• The SEM image is greyscale; it is not possible to see the natural colour of the
sample;
• SEM requires a vacuum and special sample preparation compared to simply
viewing a sample with an optical microscope;
49
• SEM imaging is time consuming and needs to be performed objectively for valid
results.
3.2.1.4 Summary of Burkard Sampler
The Burkard Spore sampler was identified as a potentially useful instrument to assess
the contribution of power station particulate emissions to the bulk of the aerosol mass:
• Established standard monitoring equipment for spores and pollen;
• Sampler is robust and can be left in the field for long periods;
• Produces a time resolved record of airborne particulates;
• Rotating drum completes one revolution per 7 days;
• Employs inertial impaction to collect particulates on a sticky tape mounted on a
rotating drum behind a slotted nozzle;
• Double-sided SEM carbon tape can be used to facilitate SEM analysis without
further treatment other than mounting on glass slides;
• Collection efficiency needs to be investigated, but sampler collects particles
smaller than 0.5 µm (Battarbee et al., 1997);
• Most of the fly ash mass (75-98%) is expected to be larger than 1 µm (McElroy
et al., 1982; USEPA, 1995);
• Individual particle analysis likely to allow identification of fly ash.
3.2.2 Determination of Contribution to Aerosol Chem istry
3.2.2.1 Methods for Collecting Samples for Measurem ent of Aerosol
Chemistry
Table 3-2 reviews some potential methods to determine aerosol chemistry; note that the
first two techniques produce samples for subsequent analysis (discussed in the next
section) while the second two techniques are direct on-line determinations.
Table 3-2: Potential methodologies for determining aerosol chemistry.
Technique
Principle of Operation
Advantages
Disadvantages
Filter Based Sampling
As above; can use sequential filters to produce more than one size fraction – typically a PM2.5 and a PM2.5-10 fraction.
Simple, robust, can use size selective inlets.
Insensitive to short duration events. Can be subject to artefacts (both positive and negative) as discussed in Table 3-1.
50
Technique
Principle of Operation
Advantages
Disadvantages
Cascade Impactors
As described above. Produces physical samples of various size fractions which can be analysed to determine chemistry variations with particle size.
Sample mass a function of the volume sampled, which can be low compared to high volume gravimetric filter sampling.
ATOFMS - Aerosol Time of Flight Mass Spectroscopy
Individual particles are blasted into component atoms by a high power laser and subsequently analysed by mass spectroscopy.
High sensitivity, can attribute to sources by assigning source chemistry (Noble and Prather, 1996)
Expensive. Source attribution requires considerable calibration. Not readily available in Australia as yet.
Determines amount of elemental and organic carbon present by determining weight loss of a periodic sample through oxidation at different temperatures.
On line measurement; carbon is not easily measured through conventional wet chemical analysis.
Sensitive laboratory instrument suitable for short field campaigns only. Limited value for power station emissions. Significant variations (up to factor of 3) reported between different methodologies (Muller et al., 2004)
Both ATOFMS and EC/OC analysis were discounted from this study due to availability
of equipment in the first instance and limited applicability in the second. Sampling with
a cascade impactor was preferred to sampling with a filter due to its ability to readily
generate size-segregated samples of the aerosol, enabling the exploration of variations
in chemistry with size. Size segregated samples were expected to be useful to help
resolve sources, as crustal material is more likely to fall into the coarser sizes and
combustion products normally report to finer sizes (Wilson and Suh, 1997).
3.2.2.2 Wet Chemical Methods for Determination of C hemistry for
Cascade Impactor Samples
These techniques involve digesting the sample and then analysing the solution to detect
elemental concentrations. Table 3-3 lists the three main methods of interest.
Table 3-3: Commonly used wet chemical analytical methods (Christian and O'Reilly, 1986; Bettinelli et al., 1998).
Method Acronym Remarks Atomic absorption spectrometry AAS Sample solutions aspirated into a flame: absorption
of monochromatic light measured. Can give accurate data but elements determined individually (characteristic wavelength). Sample size 50 mg to 1 g
ICP-AES Sample solutions aspirated into high temp flame or plasma; emitted spectra measured. Multi elemental capability (20-30 elements) Poor sensitivity for some elements; interference problems from spectral overlap
51
Method Acronym Remarks Inductively coupled plasma mass spectrometry
ICP-MS Sample solutions aspirated into high temp flame or plasma; ionised particles analysed for mass to charge spectrum. Multi elemental capability (30-40 elements) Accurate & high sensitivity Can analyse low sample masses (0.5-10 mg)
ICP-MS has been extensively used for trace element determinations due to the high
sensitivity and the ability to analyse very small samples. However, the need to generate
a solution from particulate samples poses some technical and practical issues. Samples
are typically collected on a filter medium such as quartz or glass fibre or Teflon;
particulates can be removed from the filter medium by ultrasonification or by acid
digestion of the filter medium (Bettinelli et al., 1998; Querol et al., 2000). Allowances
for the chemical composition of the unexposed filters have to be made as “impurities in
glass-fibre filters affect most of the minimum detection limits” (Bettinelli et al., 1998).
An alternative approach well suited to extremely low particle masses is Ion Beam
Analysis (IBA), discussed below.
3.2.2.3 Ion Beam Analysis: IBA
Ion Beam Analysis techniques use particles from accelerators to energise the sample
and generate various emissions which can be analysed to infer the chemistry of the
sample. Such techniques have been used for some years now to analyse aerosol
particulates because they are fast, relatively cheap, non-destructive and very sensitive to
a broad range of elements over a wide range of concentrations (Cohen, 1992). They are
therefore ideally suited to the bulk analysis of filter papers from aerosol sampling,
where sample sizes may be only 100 or 200 µg in mass (Cohen, 1998). There are four
principal particle accelerator based techniques that can be used simultaneously on a
single sample, summarised in Table 3-4:
Table 3-4: Accelerator Based Techniques Applied to Particle Analysis.
Acronym Technique Elements Detection range 3 Relative Error 2 PIXE Proton induced X-ray
These techniques are ideal when applied to membrane filter samples, with subsequent
mathematical interpretation as discussed below. A number of studies have been
conducted in Australia using these techniques e.g. (Cohen, 1992; Chan et al., 1997).
3.2.2.4 Mathematical Techniques Used for Interpreti ng Results
Because different sources often have characteristic chemical compositions, it is possible
to use mathematical techniques to determine the contribution of the various sources to
the aerosol. There are two fundamental approaches that can be taken depending on
whether the individual sources are known or not – in a sense working forwards from
known source profiles or working backwards from observed chemical compositions.
3.2.2.4.1 Chemical Mass Balance (CMB) Techniques
In CMB, a number of defined or measured source profiles are used to determine the
contribution of each to the overall chemistry of individual samples through least squares
analysis. Typically, total elemental deposition is determined through the analysis of
many samples taken over a wide area encompassing a number of known or suspected
emitters. The resulting matrix of data is then analysed using complex mathematical
techniques to resolve a number of source characteristics, which can then be plotted on a
map to show regional impacts. This approach is termed receptor modelling as the
interpretation is based on information gained from the analysis of the samples at each
receptor. The value of receptor modelling can be greatly enhanced by obtaining local
source samples to allow more specific “fingerprinting” of the chemical characteristics of
specific sources (Stern, 1986).
3.2.2.4.2 Factor Analysis
In contrast, factor analysis mathematically derives a number of vectors or “sources”
from a large body of chemical analysis data from one or more monitoring sites – these
are then related to prospective sources. Source vectors are determined by multivariate
principal component analysis followed by matrix methods such as orthogonal
transformations to maximise distinction between sources (Henry, 1991). Additional
information on source profiles and relative mass contributions can also be derived using
matrix methods (Thurston and Spengler, 1985). Both CMB and factor analysis require
considerable data and reasonably distinct source chemistry. There is a potential issue in
53
using such approaches to differentiate between power station emissions and crustal
material (i.e. soil, overburden etc) as they have similar chemistry: both are composed
mainly of oxides of silicon, aluminium, iron and calcium with varying levels of other
elements (Dzubay and Mamane, 1989).
3.2.2.5 Summary of Cascade Impactor Application
It was decided that the contribution of power station emissions to aerosol chemistry
would be best investigated using a cascade impactor. Key features of cascade impactors
and relevant analysis and interpretation can be summarised as follows:
• Cascade impactors generate size segregated aerosol samples through inertial
impaction;
• Chemical analysis of the different fractions allows the variations in aerosol
chemistry with size to be determined;
• Ion beam analysis (IBA) appears well suited to cascade impactor samples with a
wide elemental suite and high sensitivity;
• Sophisticated mathematical techniques have been shown to be effective in
delineating the contributions of different sources;
• Size-chemistry data is likely to assist in resolving various sources.
It was also intended to develop a methodology capable of providing samples in the
presence and absence of the plume from the power stations, with SO2 measurements
thought to be the most likely candidate based on past research (Jakeman and Simpson,
1987; Carras et al., 1992). This would enable two approaches to be undertaken on the
analysis: comparison of plume with non-plume aerosol chemistry as well as the
potential use of receptor modelling to resolve sources. A potential confounder in
determining power station particulate emissions was insufficient differentiation between
power station particulates and crustal sources that have very similar chemistry.
3.2.3 Determination of Contribution to Ultrafines
3.2.3.1 Methods for Determining Ultrafines
Sampling of ultrafine particles is an active area of research with many new approaches
in the literature. Table 3-5 reviews some potential methods to measure and/or
characterise ultrafine particulates.
54
Table 3-5: Potential methodologies for assessing ultrafine particulates.
Technique
Principle of Operation
Advantages
Disadvantages
Filters As above – using a size selective inlet as a pre-cutter
Easy to collect Difficult to resolve individual particles or differentiate between sources.
Low Pressure Cascade Impactors (Hillamo and Kauppinen, 1991)
As with cascade impactors, uses inertial impaction; use low pressure to reduce cut off size in final stages.
Produce physical samples of various size fractions which can be analysed to determine chemistry variations with particle size.
Better suited to chemical analysis than individual particle analysis.
ATOFMS - Aerosol Time of Flight Mass Spectroscopy
Individual particles are blasted into component atoms by a high power laser and subsequently analysed by mass spectroscopy.
High sensitivity, can attribute to sources by assigning source chemistry (Noble and Prather, 1996)
Expensive. Source attribution requires considerable calibration. Not readily available in Australia as yet.
TSI Nanometer Aerosol Sampler (NAS)
Collects positively charged particles using a high voltage electric field. Particles are collected on a substrate for SEM or TEM analysis (TSI, 2001).
Can uniformly deposit particles as small as 2 nm on substrate.
Nascent technology: largely untested in field sampling
Scanning mobility particle sizer (SMPS)
Uses differential mobility of particles in an electrostatic field separate out and count particular size ranges; size distribution built up by varying field intensity to select a range of size bins from 0.01 to 1 µm
Sensitive, able to size very small particles. On-line, continuous.
Limited equipment availability within Australia. Unable to differentiate between particles on basis of chemistry etc.
The NAS was selected in preference to the other approaches described above for the
following reasons:
• The SMPS provides information on size distribution alone and can only be used
implicitly to examine source contributions (e.g. by cross-correlation with SO2
monitoring data);
• Filters and low pressure impactors offer minimal improvements over the
cascade impactor approach to aerosol chemistry in that individual particles can
be difficult to discern;
• ATOFMS offers significant potential for charactering individual particles but is
currently unavailable;
• The NAS is largely untested but analogous in some respects to the Burkard
sampler in that it collects samples directly on a suitable medium for individual
particle analysis. Integration with SO2 monitoring would potentially enable the
assessment of plume impacts.
55
3.2.3.2 Analysis of Ultrafine Particulate Samples
Ultrafine particulates are too small to be readily analysed using SEM and transmission
electron microscopy (TEM) is preferred. TEM also uses high energy electrons to form
images of the sample, although the key difference is that the electron beam passes
through the sample. TEM often involves complicated sample preparation to ensure that
samples are thin enough to permit transmission of some electrons (Gibbon, 1979;
Glikson et al., 1988; Clausnitzer and Singer, 1999). A different approach was used in a
study at the University of Plymouth where the minus 1 µm fraction of the urban aerosol
was impacted directly on a porous carbon film, which could then be analysed without
further sample preparation by TEM (Dye et al., 2000). This is analogous to the ability
of the NAS to collect samples on TEM grids which do not require further treatment
before analysis.
3.3 PROJECT OVERVIEW
The project can be summarised in Figure 3-2. At the core of the project is the three
faceted experimental program, while added value is gained by complementary activities
involving the analysis of historical data and air pollution modelling to understand the
study results in the context of nearby urban areas
Figure 3-2: Diagrammatic representation of project scope.
3.4 SELECTION OF STUDY AREA
As discussed in the literature review the Upper Hunter Valley is well suited for a case
study assessing the impact of modern coal fired electricity generation. It is the site of
TSI NAS + TEM
Impactor + IBA
Burkard + SEM
Mass Chemistry
Ultrafine
Analysis of Historical
data
TAPM Dispersion Modelling
56
two large coal fired power stations and limited other industry apart from extensive open
cut coal mining activities. It is also relatively remote from the coast, which reduces
coastal impacts on the local meteorology. The area also has the advantage of having a
reasonable body of previous research. It was recognised at the outset that both
Bayswater and Liddell stations are fitted with fabric filters and the amount of material
emitted could prove insufficient to allow successful source recognition, although this
possibility was thought a significant potential finding in itself.
The next step was to select and validate an appropriate sampling site. Criteria used for
the selection of the site included:
• proximity to power stations
• security
• infrastructure for housing weather sensitive equipment and power supply
• access to historical monitoring data
• other particulate sources
• expected dispersion patterns based on air pollution modelling.
This process resulted in the selection of an existing air quality monitoring site at
Ravensworth for field sampling. The site was initially set up by Macquarie Generation
(then the Electricity Commission of NSW) at the request of the NSW EPA to monitor
air quality impacts of power stations emissions, and subsequently the potential impacts
of fly ash disposal and rehabilitation activities at the nearby Ravensworth void. The site
is located approximately 11 km to the south east of the power stations and was expected
to experience relatively frequent plume events, particularly during the winter months
when NW flows dominate. It was therefore expected that the site would provide a
suitable location to determine the impact of emissions from power stations.
An additional benefit of selecting an existing monitoring site was the availability of
historical data, which was interrogated to determine dilution factors and features of
plume behaviour relevant to the site. As will be demonstrated in Chapter 4, this data
yields valuable insights into seasonal and diurnal patterns as well as the timing and
duration of individual events.
57
Figure 3-3: Satellite image of study area.
Figure 3-3 shows the location of the Ravensworth monitoring site (“R”) relative to
Bayswater (“B”) and Liddell (“L”) power stations and townships Muswellbrook (“M”)
and Singleton (“S”). Areas disturbed by mining activities are clearly seen as white,
while the ranges bordering the valley can be seen in the top right and bottom left of the
figure. It will be noted that open cut mining activities are widespread, although the
nearest mine to the NW is approximately 15 km away. The site is 50 m from the New
England Highway, one of the principal roads in the area, and a railway line passes
approximately 160 m to the east.
3.5 SAMPLING WITH BURKARD SPORE SAMPLER
3.5.1 Details of Spore Sampler Set-up
One of the key features of the Burkard Spore Sampler is that it is capable of collecting
particles directly on a substrate suitable for SEM analysis without further sample
manipulation. In contrast, samples from cascade impactors require resuspension to
separate particles prior to SEM analysis of individual particles. This was not considered
a viable proposition after initial experiments indicated that the suspension medium
(alcohol or water) dissolved soluble salts and therefore modified the sample.
10101010 kmkmkmkm
58
Samples were collected on 20 mm wide double sided carbon tape sourced from
ProSciTech (PO Box 111, Thuringowa QLD 4817). The exposed tape was transferred
to standard glass microscope slides in 48 mm (1 day) sections for analysis. The spore
sampler was ideally deployed for 6 day periods, as this allowed some blank tape at the
end of the sample to facilitate handling during the transfer process.
3.5.2 Details of Field Sampling
Details of the various sampling periods and flow measurements are summarised in
Table 3-6. Samples were collected at the Ravensworth site between May and December
2002, with several outages due to equipment failures. Samples were also collected at
other Hunter Valley sites prior to the final selection of the Ravensworth site, and two
short runs were conducted near Lithgow in October 2003.
Table 3-6: Details of Burkard Spore Sampler Deployment.
13.12 / 12.08 -7 / -11 8.1 / 6.6 OK 6 days, new drum -9 mm
Ravensworth 22/10/02 to 28/10/02
13.06 / 12.10 -9 / no reading
7.3 / 0 Loose connection
Ravensworth 28/10/02 to 4/11/02
13.27 / 11.86 -5 / -10 8.8 / 7.0 OK 7 days; new drum -9 mm
Ravensworth 4/11/02 to 11/11/02
12.23 / 0.28 -8 / no reading
7.7 / 0 Battery not fully charged, 2nd motor failure
Ravensworth 11/12/02 to 18/12/02
240 V supply -3.5 / -5 9.3 / 8.8 OK 7 days. Note using mains supply
Ravensworth 18/12/02 to 23/12/02
240 V supply -5 / -7 8.8 / 8.1 OK 5 days
Blackmans Flat
2/10/02 to 7/10/02
12.63 / 11.89 -7 / -8 8.1 / 7.7 Lithgow area sampling near Mt Piper PS
Wallerawang 7/10/02 to 13/10/02
12.51/ 11.67 -7 / -8.5 8.1 / 7.5 Lithgow area sampling near Wallerawang PS
Most of the sampling was conducted using 12 V car batteries, while two samples from
December 2002 were collected using mains supply. Battery voltage was measured at
the beginning and end of each run using a portable multimeter. Initial runs indicated
that the batteries did show some voltage drop and that maximum charging was required
to last for a full 6 days in the field. Several runs were affected by the capture of small
insects in the inlet slot, resulting in strips of unexposed tape in the wind shadow.
The flow readings were made using a rotameter like flow tube supplied by Burkard
Scientific. The device has a foam seal that fits over the inlet and only three markings –
a 10 LPM line and a “+” and “-” line approximately 5 mm above and below. The flow
was recorded at the start and end of each run by estimating the distance between the top
of the float and the 10 LPM line. These readings were subsequently converted to
flowrates by a cross-calibration of the Burkard meter against a calibrated 10 LPM
rotameter (the latter calibrated using a bubble tube). Details of the cross calibration
used to determine the indicated flowrates shown in Table 17 can be found in Appendix
A. Note also that there were two 12 V motor failures – the motors were really only
60
suitable for shorter durations and 240 V supply would be the recommended option for
any future campaigns. It will also be seen in Table 3-6 that there was some variation of
the flow with the same battery with the old and new drum. This was thought to be
slight changes in the clearance between the drum and the inlet due to stretching or
swelling of the tape. This effect was noted for both battery and mains power supply.
The sampler was located on the roof of the gas monitoring shed at Ravensworth (and
other sites) to reduce the impact of windblown coarse material close to ground. The
inlet to the sampler was approximately 2.9 metres above ground level as shown in
Figure 3-4.
Figure 3-4: Location of Burkard Spore sampler on gas shed roof at Ravensworth.
3.5.3 Predicted Cut Point of Spore Sampler
The spore sampler was fitted with a narrower slot sourced from the manufacturer (0.5
mm compared to the standard 2 mm) to improve collection of smaller particles, in line
with manufacturer’s recommendations. The cut size of an impactor is usually
determined using the following equation (Marple and Willeke, 1976):
CV
WStd
pρµ50
50
9= Equation 3-1
where: d50 = size of particle with 50% chance of collection
61
St50 = Stokes number at 50% efficiency
µ = air viscosity (1.81 x 10-5 kg m-1 s-1)
W = width of impactor slot (0.5 mm)
ρp = density of particle (assumed 1900 kg m-3)
C = Cunningham slip correction factor (calculated)
V = mean velocity at throat of slot
The value of St50 varies depending on the geometry of the impactor, and is particularly
sensitive to the ratio of the stopping distance S (distance from exit of slot to impaction
point) to the slot width W. The spore sampler has a clearance between slot and drum of
0.6 mm (Burkard, 2002), with the double sided adhesive tape having a thickness of 0.22
mm as measured with a micrometer. This gives an S/W ratio of 0.76 for the 0.5 mm
slot; the corresponding √St50 for a rectangular impactor according to the plots of Marple
and Willeke (1976) is approximately 0.65 – iterative calculation of the slip factor and
solution of Equation 3-1 above yields a solution for d50 of 0.82 µm at a flowrate of 9.5
LPM (see Appendix B for a the spreadsheet used for these calculations). The S/W ratio
for the 2 mm slot is 0.19, which is beyond the limits of the Marple and Willeke (1976)
plot; a conservative value of 0.50 for √St50 yields a calculated d50 of 2.7 µm at the same
flowrate. While the cut size of the larger slot is difficult to estimate with confidence, it
is clear that the smaller slot is essential for sampling particles around 1 µm.
It should also be noted, however, that Marple in an earlier paper (Marple and Liu, 1974)
found significant differences in the values obtained by various authors for √St50 with
rectangular slot impactors. A conservative upper limit of √St50 from these data would
be around 0.80, which would give a calculated d50 of 1.02 µm at a flowrate of 9.5 LPM.
The impact of collection efficiency on the mass estimates determined using the spore
sampler will be discussed in Section 3.5.10, as this effect will tend to underestimate the
mass contribution of fly ash.
3.5.4 Analysis of Burkard Spore Sampler Tapes
Figure 3-5 shows a low magnification SEM micrograph of tape from the spore sampler.
Time can be thought of as movement in the vertical direction while particles in the same
horizontal line are essentially contemporaneous. Two high particulate matter events can
62
be clearly seen as horizontal bands in the figure. The total time represented by the figure
corresponds to approximately 1.8 hours.
The spore sampler generates a considerable area of tape each week when one considers
that the collection area is effectively the 14 mm slot width multiplied by the 336 mm of
tape exposed through the rotation of the drum. As it was impractical to manually
analyse such large quantities of tape at the magnifications required to distinguish
between individual particles, sections of the tape were selected based on SO2
concentrations measured by the gas monitoring equipment at the site.
Figure 3-5: Low magnification SEM image of tape exposed at Ravensworth site.
The tapes were analysed by scanning electron microscopy (SEM) using the University
of Newcastle’s JEOL XL30 using a combination of imaging with back-scattered
electrons and EDX analysis for bulk elemental composition. The detector has a
beryllium window and can detect elements from sodium on in the periodic table. The
images were saved in high definition mode as TIFF files (size 1424 x 1064 pixels) at
standard contrast and brightness settings to reduce between run variability. All images
were saved without the scale bar to maximise the available area for analysis (and avoid
artefacts from analysis of this); a selection of images in each session were also saved
with a scale bar to enable spatial calibration for subsequent image analysis.
Time
63
Superior differentiation against the tape background was found in the image generated
from back scattered electrons (BSE) compared to that from secondary electrons (SE). It
was also found that the tapes did not require carbon coating but could be imaged
adequately as they were. A further advantage of the BSE image is that it is more
sensitive to atomic mass, with the brightness of the image providing some indication of
chemistry: elements with higher atomic mass (and hence larger atomic nuclei) have an
increased likelihood of interaction with the electron beam. For example, biological
particles are dull while sodium chloride crystals are relatively bright. However, because
BSE are generated from further in the sample than SE, resolution is not as good and
particles less than 1 µm are difficult to image adequately. This was not considered to be
a major issue as it is comparable to the particle size cut off of the sampler and the
increased complexity of identifying power station emissions less than 1µm.
Figure 3-6:SE (left) and BSE images of large coal and silica particles.
Carbonaceous material such as coal is not readily identified using the BSE image as the
bulk of such particles does not have sufficient atomic mass to generate a bright enough
signal to be recognised as a particle. This is shown in Figure 3-6 – note how the large
coal particle is almost invisible in the BSE image, while other particles are readily
recognised in both images. However, this was not considered a major limitation as the
images were used primarily for the identification of fly ash rather than to fully
characterise other airborne particulates (and in any case relatively few carbonaceous
particles were observed). Also apparent is the significant reduction in the intensity of
the tape background in the BSE image compared to the SE image.
64
3.5.5 EDX Analysis
As noted earlier, the bombardment of a sample with high energy electrons during SEM
analysis can be used to derive information about its chemical composition. This is
achieved by focussing the 15 kV electron beam on a particular spot of the sample (it
normally scans across the field of view) and collecting the X-ray emission spectrum
over a period of approximately 45 seconds, depending on the count rate (number of X-
rays detected per unit time). This analysis is most suitable for “coarse” particles larger
than 1-2 µm because although the electron beam is approximately 1 µm in diameter, it
penetrates and disperses within the target generating X-rays from a larger area termed
the interaction volume. A 15 kV electron beam will have an interaction volume with a
diameter of around 2 µm, depending on the elemental composition (Goldstein, 2003). A
sample spectrum is shown in Figure 3-7, with the elemental peaks identified and
labelled using Link ISIS software at the time of acquisition.
0 5 10 15 20Energy (keV)
0
1000
2000
3000
4000
5000
Counts
Na
Al
Si
PSCl
K
Ti Fe
Figure 3-7: Typical fly ash EDX Spectrum with elemental peaks labelled. Horizontal axis is the energy of the emitted electrons (characteristic for particular
orbital transitions), while the vertical axis is the count rate.
It should be noted that this is only a qualitative indication of elemental composition, as
quantitative EDX analysis requires a flat surface and additional calibration; however it
was decided after initial testing that this information was sufficient to identify some key
particle classes when combined with morphology. EDX spectral information was not
routinely obtained due to the excessive demands this would have placed on acquisition
65
times. After initial confirmation of fly ash chemistry, fly ash particles were identified
on the basis of morphology alone for the purposes of determining mass concentrations.
This will be discussed in greater detail in Section 5.1.2.
3.5.6 Selection of Magnification for Imaging
It was suspected that the magnification used to acquire the images could bias the results
by failing to adequately represent certain particles – if the magnification was too great,
larger particles could be underrepresented as they would have an increased likelihood of
touching the edge of the image and being excluded from analysis. Conversely, if the
magnification was insufficient, smaller particles would not be large enough to be
adequately recognised. This issue was assessed by repeating the image acquisition
process at two magnifications, 500x and 2000x, for several time steps at 5 positions
across the tape. The resulting images were then analysed as described in Section 3.5.9
using Image Tool to identify and measure the particle size of all particles; the resulting
particle size distributions are compared in Figure 3-8.
Figure 3-8: Particle size distributions for all particles counted for images acquired at two magnifications, 500x and 2000x. Distributions are expressed as the number
of particles per mm2 reporting to a log series of size bins.
Figure 3-8 shows that the number of particles counted for a particle size greater than 2
µm is essentially independent of the magnification used; however, the images collected
at 500x magnification have inadequate resolution for smaller particles. The images
acquired at 2000x magnification were adequate for coarser particles and allowed
66
particles as small as 0.3 µm to be counted. This was the standard magnification used in
this study to determine mass concentrations. It is interesting to note that a significant
number of particles smaller than 1 µm are collected, although these are expected to be
collected at reduced efficiency.
3.5.7 Determination of Fly Ash Mass Loading
Selected areas of the tapes were analysed to assess the mass contribution of “coarse” fly
ash using the following methodology:
• Relevant tape sections were identified using SO2 data to indicate probable plume
presence;
• 5 BSE images were acquired at each time step at a magnification of 2000x for a
number of time steps at intervals of 15 minutes to 1 hour depending on the
duration of the event;
• Images were opened in an image analysis package (Image Tool) and fly ash
particles were manually identified by eye (see Section 3.5.9 for further details);
particle size and shape data were generated for both the identified fly ash and
other particles;
• The mass of individual fly ash particles was estimated by calculating the volume
from the major and minor feret diameters:
Mi = 4/3πab2ρ Equation 3-2
Where Mi = mass of ith particle
a,b = major and minor axes (from Image Tool)
ρ = assumed fly ash density (1900 kg m-3)
• Airborne concentrations (of fly ash only) were determined using the volume
flow rate (adjusted for battery voltage drop effects) and the exposed area:
QA
A
MiC
e
a
∑= Equation 3-3
Where C = estimated airborne concentration (µg m-3)
Aa = area of tape analysed (m2)
Ae = area of tape exposed in 1 hour (m2)
Q = volume sampled in one hour, corrected for voltage drop (m3)
67
It should be noted that overall particle mass loadings (as opposed to fly ash) are more
difficult to estimate due to the presence of agglomerates and to particle shape. The sizes
determined are in two dimensions only and, while this can be extrapolated to three
dimensions relatively easily for spherical or near spherical particles, this is certainly not
the case for more irregularly shaped particles and for agglomerates. Overall airborne
mass concentrations were not estimated from the images.
3.5.8 Sources of Error and Uncertainty for Mass Con centrations
Three main sources or error and uncertainty have been identified in the mass estimates:
1. Uncertainty due to thresholding of images, discussed in Section 3.5.9.
2. Potential bias due to using the spherical fly ash particles only for the mass
determinations. This is discussed further in Section 3.5.10.
3. Bias due to variable flowrates and reduced collection efficiency at small particle
diameters – this is best discussed using real data and will be dealt with in the
appropriate results section (Section 5.1.6.2), although the principles will be dealt
with in general terms in Section 3.5.10.
4. Uncertainty due to counting statistics - the uncertainty due to the number of
particles counted relative to the population distribution can be estimated using
statistical methods and the principles will be discussed in Section 3.5.12.
3.5.9 Image Analysis Details
Images were processed using a freeware software package downloaded from the
University of Texas website (http://ddsdx.uthscsa.edu/dig/itdesc.html) called UTHSCA
Image Tool (Version 3.00). The first step was to manually scan through the image and
identify probable fly ash particles. These were readily identified using particle shape
and brightness – the particles are smooth and nearly spherical, and also relatively
uniformly bright due to their composition (typically an alumino-silicate glass).
The software required a number of parameters to be set up before use. Firstly, a spatial
calibration was required so that the output from the program was in microns rather than
pixels. The software has a function that allows spatial calibration against the scale bar
of the SEM image. Checks against multiple images from a session showed no
significant variation in the calibration were required for a given magnification, even for
68
small changes in the working distance (e.g. if the tape was not quite smooth or the glass
slide slightly tilted).
Additional parameters that were manually selected in the program are in the
Settings\Preferences\Find Objects menu; the maximum number of particles to be
counted was set to 500 and the minimum number of pixels was set to 11. These
settings were chosen based on experience with the maximum number of particles
observed in any one image and to give a minimum particle size of around 0.3 µm, to
avoid artefacts from small numbers of pixels. Note that this lower limit is well below
both the calculated d50 of the spore sampler and the size of particles which can be
readily identified as fly ash.
Particles were then identified using the “Find Objects” command, which allows the user
to specify the brightness threshold between the background (collection tape) and
particles. Increasing the lower threshold can be thought of as “peeling off” the edges of
duller particles. After initial testing, the threshold was set at an arbitrary brightness of
40 (out of a scale of 0-255). This gave a reasonable compromise between separating
nearby objects (which deteriorates with a lower threshold) and accurate estimation of
the particle size. This process is shown in Figure 3-9. Note how with a lower limit of
20, the background is treated as belonging to particles and a large number of spurious
“objects” are found (total number is 128). As the lower limit is increased, these
artefacts are eliminated and the main issues are separation of nearby particles and loss
of small particles (less than 1 µm).
(a) Original image (scale bar = 5 µm)
(b) Objects found brightness range 20-255 (128)
69
(c) Objects found brightness range 30-255 (24)
(d) Objects found brightness range 40-255 (20)
Figure 3-9: Effect of lower limit of thresholding on number of objects found by Image Tool “Find Objects” function.
The sensitivity of the diameter determined to thresholding is difficult to assess, mainly
because it is difficult to know where the particle truly finishes and the background
begins. However, because fly ash is relatively bright, the particle edge is generally
easily delineated – trials with varying the thresholding by 10 brightness units either way
were found to influence the diameter by only 0.4%. While the impact on mass is greater
(1.2%), the uncertainty from thresholding will be shown to be negligible compared to
other sources.
3.5.10 Potential bias due to fly ash morphology ass umptions
Implicit in the above calculations is the assumption that all or nearly all of the mass of
primary particulate emissions is readily recognisable as spherical alumino silicate glass.
Although it was not possible to sample and characterise the emissions directly, an
sample of hopper ash collected earlier from Bayswater power station was available. A
small amount of ash was shaken in a plastic jar and the fume (or more correctly fine
particle “mist” that wafted out when the lid was removed) was sampled using the
Burkard sampler on double sided carbon tape as normal. An image of these particles is
shown in Figure 3-10. Nearly all particles are spherical, although a small percentage is
irregular; this is expected to consist of unfused material such as quartz and uncombusted
char. Spherical particles are expected to account for at least 95% of the mass; this is
consistent with earlier literature studies (Fisher et al., 1978; Mamane et al., 1986). It is
therefore considered that basing the mass estimates on spherical particles only will give
slight underestimates of the true value; this bias is estimated at 5%.
70
Figure 3-10: SEM image of fine component of hopper ash from Bayswater power station.
3.5.11 Impact of Flowrate Variations
The impact of voltage drop and the resulting decrease in flowrate is three fold. The first
impact is volumetric – the reduction in the volume sampled needs to be factored into the
determinations of mass concentrations. Secondly, the decrease in nozzle velocity
affects the cutpoint of the sampler – for a drop from 9.3 LPM to 7.5 LPM typical of the
field sampling, the cut size increases from 0.82 µm to 0.93 µm (Appendix B). The third
impact is through the efficiency of collection, which decreases with size, as discussed
above. Unfortunately the design of the spore sampler precludes the direct measurement
of the relationship between particle size and efficiency, as the discharge stream exits
through the suction fan and cannot be readily sampled. Literature values and a
sensitivity analysis will be used in Section 5.1.6.2 to demonstrate the impact of
collection efficiency on the mass estimates.
71
3.5.12 Estimation of Uncertainty for Mass Concentra tions (Counting
Statistics)
Because it was impractical to measure large numbers of fly ash particles at each time
step, there are uncertainties associated with the limited sample populations which have
been estimated using statistical methods (Hall, 1983). A pragmatic approach has been
developed at the University of Newcastle which uses the number of observations and
the observed variability in the population to estimate the uncertainty using a modified
inverted Edgeworth expansion (Tuyl, 2003). This approach uses parameters which
describe both the spread and skewness of the distribution, and a z score based on the
number of observations to estimate the width of the confidence interval. The pragmatic
adaptation extends this approach using the Student’s t statistic as a more conservative
estimate of the extent of the confidence interval, as below:
UCLa = { }1)/6)(2zγ̂(ntsnx 2α-1
1/21-nα,-1
1/2 +++ −−
Equation 3-4
where: UCLa = adjusted upper control limit on mass estimate
x = mass estimate based on mean particle mass
n = number of fly ash particles counted in estimate
s = sample standard deviation (particle mass)
t = Student’s t statistic
α = for 1-α confidence interval (e.g. α = 0.05 for 95% CI)
γ = skewness parameter (calculated by excel)
z = standardised normal variate
3.6 CASCADE IMPACTOR AND IBA ANALYSIS
3.6.1 Cascade Impactor Details and Predicted Cut-po ints
The cascade impactor used in the field sampling campaigns was borrowed from CSIRO
Energy Technology, formerly at Ryde in Sydney, NSW. The impactor is a custom
manufactured unit made of stainless steel with 5 stages. The impaction surface was a
20 mm diameter glass microscope cover slip which sat in the recessed lip of a stainless
steel holder, held in place with a few dabs of vacuum grease. The samples were
collected on a substrate consisting of discs cut from Nucleopore Polycarbonate AP track
etched filters (pore size 8.0 µm, Apiezon coated by manufacturer to assist with particle
retention). These filters were recommended by ANSTO personnel as they are routinely
used for IBA of the ASP PM10 samples. Small circles of filter were cut out and secured
72
to the cover slip using small dabs of vacuum grease. Care was taken to keep the grease
away from the impaction point to avoid possible sample contamination. The back-up
filter was a 47 mm diameter cellulose acetate filter with a pore size of 0.10 µm
(Millipore type VC); this was used in preference to track etched filters due to potential
issues with pore blockages.
Dimensions of the impactor apertures were determined using an optical microscope
with a graticule for the final 3 stages and a vernier calliper for stages 1 and 2. The
apertures are 2, 1.2, 0.97, 0.65 and 0.45 mm, with nominal d50’s for a particle density of
1500 kg m-3 for the stages of 2.57, 1.17, 0.83, 0.43 and 0.21 µm at a flowrate through
the impactor of 1.07 LPM (see Appendix C for details of spreadsheet calculations based
on equations of Marple and Willeke (1976)). This flowrate was measured in laboratory
calibrations and appeared to be determined by the final stage rather than the applied
vacuum – the same flowrate was obtained using the vacuum pump employed for field
sampling as with a larger vacuum pump.
3.6.2 Calibration of Cascade Impactor
The cascade impactor was calibrated at CSIRO Energy Technology in Ryde using a
condensation aerosol generator with sodium chloride seeding and sebacic acid ester
condensation (di-2-ethyl ethylhexyl-sebacate). The size distribution of the largely
monodisperse aerosol was measured using an APS either “unchallenged” or after
passing through a single stage of the impactor. By increasing the temperature of the
sebacic acid ester, the vapour pressure was increased to produce larger aerosol particles.
The classification behaviour of individual impactor stages was assessed by running a
series of experiments with the mean aerosol size above and below the apparent cut
point. This process was conducted for stages 1 to 3; the cut points of stages 4 and 5
were below the lower limit of the APS. The SMPS at CSIRO was not functional at the
time of calibration and therefore the last stages could not be calibrated. Results of the
calibration will be presented in Section 5.2.1.
The molecular weight of the sebacic acid ester is 426.68 and it has a density of 0.912
and a boiling point of 256°C (Weast et al., 1986).
73
3.6.3 Conditional Sampling Methodology
As discussed above, SO2 was used as a plume indicator to investigate the chemistry of
various aerosol size fractions in the presence and absence of power station impacts. For
this approach to be valid, the power stations need to be the dominant local SO2 source;
this will be discussed using historical data in Chapter 4.
The 10 minute SO2 concentration at Ravensworth was used as a conditional switch for
the power supply to the vacuum pump for the impactor, with a threshold value of 20
ppb. There were two modes of operation – “SO2 high” (i.e. >20 ppb) and “SO2 low”
(<=20 ppb) – which allow sized fractionated aerosol samples to be collected under
conditions where the power station influence was expected to be greatest and least. The
threshold of 20 ppb was selected to provide both a significant deviation from the
background and reasonable run time for sampling under “SO2 high” conditions over a
period of one month. The base case for comparison was obtained by sampling in “SO2
low” mode for between one and two days. 8 sets of samples were collected under each
mode of operation over the period August 2002 to June 2003, as shown in Table 3-7.
Table 3-7: Details of sampling campaigns with cascade impactor at Ravensworth.
Date Range Regime Run Hours 08/08/02-28/08/02 SO2 hi 14 28/8/02-30/08/02 SO2 lo 48 30/8/02-16/09/02 SO2 hi 33 23/9/02-22/10/02 SO2 hi 36 22/10/02-24/10/02 SO2 lo 46 24/10/02-28/10/02 SO2 lo 95 28/10/02-26/11/02 SO2 hi 141 26/11/02-28/11/02 SO2 lo 42 28/11/02-16/01/03 SO2 hi 55 16/01/03-28/01/03 SO2 hi 16 10/03/03-11/03/03 SO2 lo 24 11/03/03-05/05/03 SO2 hi 36 05/05/03-06/05/03 SO2 lo 23 06/05/03-08/05/03 SO2 lo 42 08/05/03-10/06/03 SO2 hi 39 10/06/03-11/06/03 SO2 lo 25
The cascade impactor was mounted at the south-western edge of the gas shed roof with
the inlet approximately 2.5 m above ground level. The inlet was directed away from the
shed and a short silicone tube was used in most runs so that the sample point was not
directly over the shed roof but in “free air”. A test conducted to see whether particles
were being retained in this tube indicated that only particles large than 10 µm were
74
collected (by washing the tube out with ethanol and measuring the particle size using a
Malvern Mastersizer). The suction from the cascade impactor (approximately 2.5
metres long) was also silicone tubing which was passed through the hole in the gas shed
wall for the air conditioning unit to the vacuum pump inside. The last few runs were
conducted without the inlet tubing to see whether any noticeable effects were observed.
3.6.4 Ion Beam Analysis of Cascade Impactor Samples
Results of preliminary sampling at power stations and in the field indicated that the
amount of material collected on the impactor stages was insufficient for wet chemical
techniques and consequently IBA techniques would be most appropriate. Discussions
with ANSTO personnel lead to the use of polycarbonate membranes for sample
collection, and initial runs were analysed to confirm sufficient sample mass had been
collected for IBA analysis.
The samples from individual runs and stages were analysed using PIXE and PIGE at
ANSTO (Cohen et al., 1996) for multiple elements to enable reconstitution of the
aerosol. It was not possible to analyse for some key elements using these techniques
e.g. C, H, O and N. Note that it was also not possible to measure individual masses
before and after exposure due to the need to use a small amount of grease to secure the
filter substrate to the impaction surface. A photograph of the samples in the sample
holder “stick” is shown in Figure 3-11. Reference samples for calibration purposes are
in the large holders at the left of the stick, while the cascade impactor samples are in the
smaller holders to the right. This stick was inserted into the IBA machine where the
samples were individually bombarded by protons with an energy of 2.6 MeV, and the
X-Ray and gamma-ray spectra measured. The beam had a diameter of 4 mm for the
first group of analysis in September 2002 and 3 mm in June 2003.
75
Figure 3-11: Photograph of IBA stick showing reference materials (left) and samples from cascade impactor (smaller holders on right).
Figure 3-12: Typical PIXE spectrum showing peaks for various elements.
A sample PIXE spectrum is shown in Figure 3-12. Peaks corresponding to
characteristic energies associated with electron orbital transitions allow the
concentration of individual elements to be determined. The processing of the spectrum
to derive concentrations was carried out by ANSTO personnel. The results were
provided in spreadsheet form as micrograms of element per cm2 and were converted to
elemental masses before interpretation of the results. Results are expressed in this form
because the technique is typically applied to filter samples, and the beam analyses only
a part of the full sample. For the back-up filter sample, the total amount of each
76
element collected was determined by assuming uniform deposition and multiplying the
measured concentration by the exposed area (diameter 26 mm). For the cascade
impactor plate samples, the entire sample was irradiated by the beam, and the elemental
masses are determined by multiplying the measured concentrations by the beam area
(diameter 3 or 4 mm as noted above).
PIGME spectra were also obtained but most of the indicated concentrations were not
used in subsequent analysis due to comparatively high uncertainties, as discussed in the
results section.
3.7 NANOMETER AEROSOL SAMPLER
3.7.1 Collection of Samples from Ambient Air at Rav ensworth
The NAS is designed to collect very small particles uniformly on a substrate for
subsequent analysis. The manual indicates that it will only collect positively charged
particles, and suggests that a TSI particle sizer is used to precondition the feed aerosol.
However, initial trials with the NAS indicated that sufficient particles could only be
collected in the time frame of individual events using a much broader size distribution.
It was therefore decided to collect samples using stages 1 through 4 of the cascade
impactor with greased impaction plates to prevent bounce as a pre-cutter so that only
particles less than 0.4 µm or so would be presented to the NAS.
The issue of particle charging was addressed through careful examination of the
operating manuals as well as contact with TSI representatives. Because the NAS will
only collect positively charged particles, it is important to understand the charge
distribution of particles in the air. The equilibrium charge distribution is governed by
well known rules and is shown in Table 3-8.
Table 3-8: Equilibrium distribution of charges on aerosol particles (TSI, 2003).
Percent of Particles Carrying Np Elementary Charge Un its Dp(µm) Np=–6 –5 –4 –3 –2 –1 0 +1 +2 +3 +4 +5 +6
The cumulative distribution for TSP concentrations both at Ravensworth and the local
maximum in a 5x5 sub-grid region around the monitoring site (i.e. within 1.25 km of
the monitoring site) are shown in Figure 4-12. This data is based on the assumption that
there is minimal loss of particles through the effects of gravitational settling – this is
likely to be valid for the finer particles (almost certainly true for the PM2.5, and possibly
up to around 10 µm), although larger particles will tend to settle out – these estimates
therefore will tend to overestimate the contribution of PS emissions. The plot suggests
94
that the contribution of primary power station emissions to ambient particulate matter
would seldom exceed 1 µg m-3 at either Ravensworth or in the nearby area.
1
10
100
1000
10000
0 0.2 0.4 0.6 0.8 1
Hourly Conc. of APM from PS Emiss ionsHourly Conc. of APM from PS Emiss ionsHourly Conc. of APM from PS Emiss ionsHourly Conc. of APM from PS Emiss ions
# of hourly periods conc.>
C# of hourly periods conc.>
C# of hourly periods conc.>
C# of hourly periods conc.>
C TAPM - Rav
TAPM - Loc
Figure 4-12: TAPM predictions of hourly concentrations of TSP from power station primary emissions in µg m-3.
Table 4-4 summarises the maximum percentiles expected at the Ravensworth site for
both SO2 and TSP based on the TAPM modelling. The 99th percentile value for the
contribution to TSP is 1.6 µg m-3 (comparable to the 1.9 µg m-3 in Table 4-1 based on
2001/2002 data), while the practical maximum or 99.9th percentile is 4.6 µg m-3. The
ratio of TSP to SO2 is 0.0304 µg m-3 per ppb, while the ratio of PM10 to SO2 is 0.0152
µg m-3 per ppb. The data in Table 4-4 also confirms that TAPM appears to be
overpredicting SO2 at the site at high concentrations as noted above, giving conservative
estimates of the contribution of power station primary particulates to aerosol mass.
Table 4-4: Descriptive statistics for hourly concentrations predicted by TAPM (unscaled) for Ravensworth site and SO2 monitoring data for comparison.
TAPM has been used to assess the dispersion patterns and relative concentrations at the
Ravensworth monitoring site in the context of the nearby urban areas of Singleton and
Muswellbrook. It is believed that TAPM is modelling the expected SO2 concentrations
reasonably well, with good agreement in the timing of events. Individual events are not
always well correlated although this is perhaps an unrealistic expectation of the model.
The temporally dissociated correlation of concentrations shows reasonable agreement,
although TAPM does appear to over-predict extreme events. The Ravensworth site was
confirmed as a suitable mid impact site for assessments, with a predicted 99th percentile
contribution of power station PM10 emissions to ambient particulate matter of 0.8 µg m-
3, and a maximum (99.9th percentile) contribution of 2.3 µg m-3. Maximum power
station derived concentrations at Singleton and Ravensworth are expected to be of the
order of 50 to 70% of these values based on relative SO2 concentrations from the TAPM
modelling, although it should be recognised that these estimates are based on steady
state emissions and short term variability in emissions adds some uncertainty to these
estimates.
96
5 RESULTS
5.1 ANALYSIS OF BURKARD 7-DAY SPORE SAMPLER TAPES
This section will present the results of analysis of the spore sampler tapes using SEM
microscopy. The section will begin by assessing the performance of the sampler with
respect to particle collection on the tape, before proceeding to the identification of
particles from different sources. This will be followed by more specific details on the
nature and character of the particles identified as fly ash, before results are presented
from the analysis of various SO2 events including estimated mass concentrations. This
data will be used to test the fundamental assumption that SO2 can be used as a plume
indicator, and the uncertainties from various sources will be discussed. The section will
conclude with a discussion of potential impacts from ash disposal activities at the
nearby Ravensworth Void.
5.1.1 Assessment of Spore Sampler: Deposition Patte rns
5.1.1.1 Evenness of Loading across Tape
The mass estimates determined from the spore sampler tapes are taken from 5 images
acquired across the tape (x axis) at a particular time (y axis). The width of the slot is 14
mm and the width of the tape 20 mm. Images were acquired at the centre of the tape
and at 2 mm and 4 mm either side, as shown in Figure 5-1.
Figure 5-1: Schematic showing location of the 5 images (not to scale) used for mass determinations; grey area indicates slot dimensions (14x0.5 mm).
Y
X
Edge of
Tape
Edge of
Tape
Time
3mm 3mm 2mm 2mm 2mm 2mm 3mm 3mm
Images
97
Size distribution data were generated using Image Tool as a check on the evenness of
deposition across the width of the tape. Each line in the plots in Figure 5-2 below
represents the distribution of all identified particles at a given X position on the tape,
calculated by summing the number of particles in each size bin over the various time
steps (and normalising in the case of plots (a) and (c)). While there is some variation in
the numbers of particles counted at the different positions, there is no evidence of any
systematic bias. The drop off in the number of particles observed below 0.5 µm is
attributed to a combination of reduced collection efficiency at smaller particle sizes and
the difficulty of adequately imaging these particles.
0%
5%
10%
15%
20%
25%
30%
35%
40%
<0.5 0.5-1.0 1.0-2.0 2.0-4.0 4.0-8.0 8.0-16 >16
Size Bin (microns)
Number % in Size Bin
-16500
-14500
-12500
-10500
-8500
X Co-ordinateX Co-ordinateX Co-ordinateX Co-ordinate
(a) Job 23 (5 time steps, 25 images)
0
20
40
60
80
100
120
140
160
180
200
<0.5 0.5-1.0 1.0-2.0 2.0-4.0 4.0-8.0 8.0-16 >16
Size Bin (microns)
Number in Size Bin (total)
-16500
-14500
-12500
-10500
-8500
X Co-ordinateX Co-ordinateX Co-ordinateX Co-ordinate
(b) Job 23 (5 time steps, 25 images)
0%
5%
10%
15%
20%
25%
30%
35%
<0.5 0.5-1.0 1.0-2.0 2.0-4.0 4.0-8.0 8.0-16 >16
Size Bin (microns)
Number % in Size Bin
-16500
-14500
-12500
-10500
-8500
X Co-ordinateX Co-ordinateX Co-ordinateX Co-ordinate
(c) Job 24 (3 time steps, 15 images)
0
20
40
60
80
100
120
140
160
180
<0.5 0.5-1.0 1.0-2.0 2.0-4.0 4.0-8.0 8.0-16 >16
Size Bin (microns)
Number in Size Bin (total)
-16500
-14500
-12500
-10500
-8500
X Co-ordinateX Co-ordinateX Co-ordinateX Co-ordinate
(d) Job 24 (3 time steps, 15 images)
Figure 5-2: Size distributions of particles at different positions across tape. Plots (a) and (c) are normalised number distributions; plots (b) and (d) are raw particle
number data (X-co-ordinate refers to stage position in microns).
5.1.1.2 Time Uncertainty – Event Horizon
While one of the key attributes of the spore sampler is the ability to record temporal
variations in particulate loadings, there is in fact some “blurring” of the time of
98
collection. Particles are not collected on a single line on the tape at any given time, but
rather over an area delineated by the slot dimensions plus any fanning out of the air
stream between the nozzle and impaction point. These effects were investigated by
exposing fresh tape to puff events and examining the width of the collection zone, as
shown in Figure 5-3. Most of the particles are collected within a band approximately
0.70 mm wide, indicating some fanning out of the stream from the 0.5 mm slot. This
corresponds to a time horizon of approximately 20 minutes. The fanning out appears to
be more evident for finer particles, which are more prominent at the edges of the main
impaction zone. Some larger material is also collected outside the main impaction zone.
This is thought to be due to a combination of overloading and imperfect collection.
Image Tool was used to assess the area (in pixels) of particles inside and outside the
main collection zone in Figure 5-3. Around 2000 of the 2470 particles (81%) identified
in all were found in the main impaction zone. On a pixel basis, the main impaction zone
accounts for 89% of the area assigned to particles, indicating acceptably high collection
efficiencies in this main zone even with such massive overloading.
Figure 5-3: Image of “puff” event showing extent of impaction area.
5.1.2 Identification of Particulates from Different Sources
Figure 5-4 shows the tremendous range of particle shapes and sizes present during a
relatively high particulate matter event. While it is not possible to definitively assign
sources to all particles, it is certainly possible to recognise a single large fly ash particle
(“F”) in the lower right of the image. Also recognisable are a number of biological
particles including two donut shaped particles (“B”) at the top of the image; note the
99
relatively dull image due to lower average atomic mass. The majority of the particles
are comparatively bright and irregular – these are believed to be derived from crustal
material.
Figure 5-4: SEM image of a high particulate matter event.
Figure 5-5 shows an image containing several large crystalline particles. Three groups
of commonly occurring crystalline particles were identified using a combination of
particle morphology and EDX chemistry:
• NaCl: bright, often cubic, strong Na and Cl peaks. Particle size is sometimes
quite large (over 10 µm) which suggests local sources (for example irrigation or
water spraying for dust control in nearby mines) are responsible for some of the
salt crystals observed.
• CaSO4: slightly duller, elongated – columnar, sometimes pill shaped, strong Ca
and S peaks
• Dark Crystals: dull image, variable morphology, low signal to noise ratio with
no major elemental peak. The spot used for EDX analysis “drilled” a hole in
these crystals, indicating relatively poor thermal stability – this is consistent with
the low melting and boiling points of ammonium nitrate, which are 170°C and
210°C respectively (Weast et al., 1986). The absence of a sulphur peak
indicates these crystals are not ammonium sulphate.
B
F
100
Figure 5-5: SEM images of several unusually large crystalline particles.
Table 5-1 summarises the key particle categories along with a collage of typical images
and a sample X-ray spectra. It should be noted that categories are not included for two
common types of atmospheric particles, ammonium salts (nitrate and sulphate) and soot.
These particles were not identified in the images due to a combination of the size
limitations of SEM imaging and the analysis limitations of EDX chemistry (only
elements from Na on can be identified). The potential contribution of these species to
ultrafine particles is discussed in Section 5.3.
Table 5-1: Key (coarse) particle categories identified using morphology and spectral data.
Partic le Class, Typical SEM Images and Details. Typical EDX Spectrum
Crustal Material (soil, overburden)
Morphology: Usually irregular or angular. Often as agglomerates. Highly variable particle size Brightness: Varied but usually relatively bright. Often variable within a particle (chemical variations).
Main peaks: highly variable, commonly Si and Al, but many others – Fe, Ti, Ca, K, Na
Fly ash
Morphology: Spherical or near spherical with smooth surface. Sometimes occurs as multiples (usually 2).
NaCl
NaCl CaSO4
NH4NO3?
CaSO4
NH4NO3?
101
Partic le Class, Typical SEM Images and Details. Typical EDX Spectrum Brightness: uniform and relatively bright. Main peaks: Si, Al; often S, K,
Ca, Fe Salt crystals
Morphology: Cubic or right angled corners, straight edges. Often in agglomerates. Brightness: very bright.
Main peaks: Na, Cl
Calcium sulphate crystals
Morphology: crystalline, corners often appear rounded. Often elongated or pill shaped. Brightness: relatively bright.
Main peaks: S, Ca
Biological
� �
Morphology: variable, commonly cigar shaped. Some round, some with surface texture as shown. Brightness: dull – often variable due to structure.
Main peaks: no consistent peaks, sometimes Si, S, and K. High noise to signal ratio.
Coal
Morphology: angular. Brightness: dull overall, though often with bright patches due to mineral inclusions or adhering particles.
Main peaks: Al, Si, Fe, S. High noise to signal ratio.
0 5 10 15 20Energy (keV)
0
500
1000
1500
2000
2500
Counts
NaSiPS
102
5.1.3 Selection of Events for Mass Assessments
Events were selected for SEM investigation primarily on the basis of the 10 minute SO2
data, although a number of high SO2 events were unsampled due to equipment failures.
The periods selected are shown in Table 5-2, along with the corresponding maximum
10 minute SO2 concentration. It is believed that these data reflect the “worst case”
scenarios in that they are from areas of the tape corresponding to highest SO2
measurements over the 8 months study period (concentrating on winter 2002). Each
event is designated by the job number of the SEM session in which it was analysed.
Table 5-2 also shows the number of time steps for which images were acquired as well
as the number of images analysed with image tool (5 per time step with the exception of
Job 30 which covered a 24 hour period) and the number of particles identified as fly
ash. Note that the discussion of results will be largely restricted to the Ravensworth
data, as very limited data was obtained from the Blackmans Flat sampling.
In summary, while cut sizes are difficult to determine with certainty, it is likely that
Stage 1 will have a d50 of around 2.5 µm and that the final stage will have a d50 of
around 0.2-0.3 µm.
5.2.2 SO2 Concentrations During High SO 2 Campaigns
The average SO2 concentrations for the high SO2 cascade impactor samples are
summarised in Table 5-7. The average SO2 concentration was fairly consistent over the
various sample periods, with a weighted average value of 46 ppb (the threshold was 20
ppb).
Table 5-7: Average SO2 concentrations during high SO2 sampling campaigns.
Date Range Regime Run Hours Average SO2, ppb
08/08/02-28/08/02 SO2 hi 14.20 53.4 30/8/02-16/09/02 SO2 hi 32.66 53.8 23/9/02-22/10/02 SO2 hi 35.88 45.4 28/10/02-26/11/02 SO2 hi 140.65 38.4 28/11/02-16/01/03 SO2 hi 55.43 42.0 16/01/03-28/01/03 SO2 hi 16.16 56.1 11/03/03-05/05/03 SO2 hi 35.50 50.6 08/05/03-10/06/03 SO2 hi 39.17 38.5 All periods SO2 hi 369.65 46.3
5.2.3 Factor Analysis of IBA Chemistry Results
The raw data from the IBA results was converted to airborne concentrations by firstly
calculating the total mass collected on an elemental basis. Elemental concentrations
provided by ANSTO in µg cm-2 were converted to elemental masses by multiplying by
the beam diameter for the stages and the collection area for the back-up filter.
Unexposed samples of the filter and isopore membrane were also analysed to ensure
that the results were compensated for the composition of the blank filters.
117
The elemental masses were converted to airborne concentrations in ng m-3 by dividing
by the sampled volume, determined by multiplying the flowrate through the impactor
(measured at 1.07 LPM during laboratory calibrations) by the number of hours indicated
by the run clock on the conditional sampling power supply. Elemental concentrations
were preferred over oxides to avoid biasing the analysis on potentially erroneous
stoichiometry. The full, unchecked data set consisted of a matrix of 96 rows (16 runs,
each with 6 impactor stages) and 23 columns (elements F, Na, Mg, Al, Si, P, S, Cl, K,
Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Br, Se, Sr and Pb). F, Na and Mg were
determined using PIGE, while the other elements were measured with PIXE.
5.2.3.1 Data Validation
The data was validated before analysis by comparing the data values with the quoted
error values from ANSTO. The error estimates from ANSTO are comprised of a
precision error of around 3% for most elements and a variable error based on counting
statistics – this second component is heavily dependent on how close the measured
concentration is to the minimum detection limit (Cohen, 1997). The errors range from
11.2% for elements present in relatively high concentrations to 100% of the measured
value for other elements. F, Mg and P all have significant errors compared to the data;
these elements were excluded from subsequent analysis (see Appendix E for details).
Significant errors were also noted in many of the other elements present in low
concentrations, but these data were retained with the proviso that errors would need to
be considered in any subsequent evaluation. Note that the subtraction of analysis blanks
also increases the errors.
Table 5-8 summarises the key descriptive statistics (mean and standard deviation) for
the high and low SO2 samples on an element by element basis. The total masses are
also included for interest, both as the sum of all elemental masses and as an indicative
mass when converted to oxides (but not allowing for any water of hydration). Note that
these totals are effectively the average mass per stage, and thus an estimate of the
average reconstituted airborne concentration can be made by multiplying by 6. The
indicated concentrations are hence around 6 µg m-3 for the low SO2 cases and 11 µg m-3
for the high SO2 samples – considerably less than the average PM10 measured at the
monitoring site of 25 µg m-3. This is probably due to the elements that were not able to
be measured, particularly C and N; organics and elemental carbon were found to
118
contribute 46% of PM2.5 at Muswellbrook, with soil and salt only accounting for 14% of
PM2.5 mass (MSC, 2003). It is probable that the coarser particles were not collected as
efficiently due to the inlet tube and the fact that sampling was not isokinetic. This is
consistent with the stage masses, which are not dramatically higher for Stage 1 than the
other stages, as might be expected given the cut size of around 2.5 µm.
Table 5-8: Overview of Data Integrity – all stages (“High Integrity” data has an error of less than 25%, “Lower Confidence” from 25-100%).
Figure 5-15: Contribution of identified components to different particle sizes.
123
Enrichment in the finest fraction is consistent with sources where particles are formed
during combustion or at high temperatures, as would be expected for Components 2, 4
and 5. Component 5 also shows enrichment in the coarser sizes under low SO2
conditions, although the reasons for this are unclear. Component 1 – the crustal
signature – is strongly associated with the coarser size fractions i.e. larger than about 1
µm. Perhaps the most surprising result is the strong association of Component 3 – salt –
with the finest sizes; other studies indicate that salt is more strongly associated with the
coarser sizes (Thurston and Spengler, 1985; Pio et al., 1996; Chan et al., 1999b). The
increased loading of this component in the high SO2 samples suggests that some of this
component may also be originating from the power stations, perhaps from dissolved
salts in the water fed to cooling towers or from stack emissions.
5.2.3.3 Robustness of Factor Analysis (see Appendix G for details)
As noted above, several alternative analyses were conducted with various data excluded
to see how sensitive the rotated PCA solutions were to the input data. This is important
as factor analysis is unable to weight data according to its integrity because all variables
are normalised prior to component extraction; it was therefore necessary to demonstrate
that the “lower confidence” data was not forcing the solution. These analyses
(summarised in Appendix G) produced broadly similar results with some minor changes
in the association of some elements.
The most conservative analysis that excludes all elements with significant numbers of
low confidence data is restricted to Na, Al, Si, S, Cl, K, Ca, Ti and Fe – this resulted in
only two factors being extracted with significant lumping together. Component 1
(explaining 52% of variance) remains a soil signature and is characterised by Al, Si, K,
Ca, Ti and Fe as before. Component 2 (explaining 25% of variance) becomes a lumped
sea salt / CFPS signature and is associated with Na, Cl and S.
A slightly less conservative analysis includes the data for Ni, Mn and Zn, the elements
with around 40 high integrity data values; this resulted in the extraction of 3
components. Again, Component 1 is readily identified as soil, explaining 39% of
variance and associated with Al, Si, K, Ca, Ti and Fe. Component 2 (explaining 22% of
variance) is associated with S, Cl, Ni and Zn, and appears to be a CFPS signature.
124
Component 3 explains 16% of variance, is associated with Na, Cl, Mn and Zn, and is
readily identified as salt. The three components account for 76% of the variance in the
data.
The final reduced data set adds in the data for Cr, Co, Cu and Br, the elements which
mainly have “lower confidence” data, but fewer values below the detection limits than
the final group V, Se, Sr and Pb. PCA on this data set resulted in the extraction of five
components, with essentially the same extraction of components as the full data set
(excepting the excluded elements). Component 1 remains associated with Al, Si, K, Ca,
Ti and Fe - it is again identified as soil, explaining 30% of variance. Component 2,
explaining 21% of variance, is identified as CFPS emissions and is associated with S,
Cl, Ni, Cr, Cu and to a lesser extent Zn and Co. Component 3 (13% of variance) is salt,
associated with Na, Cl, Mn and Co. S is also weakly associated with this component, as
with the solution for the full data set. Component 4 (12% of variance) appears to be a
diesel signature, and is associated strongly with Zn and Br, and to a lesser extent Na, Cr
and Cu. Component 5 (10% of variance) is similar to the Indust 1 source, and is most
strongly associated with Mn and Cu, and more weakly with K, Fe and Co, as well as a
negative correlation with Si. Overall variance explained is superior to any of the other
analyses at 86%, indicating that the components explain the variations in the original
data quite well.
Including the lower confidence data for V, Se, Sr and Pb into the data set, but excluding
the three outlier results, results in a decrease in the predictive power of the solution.
Factor analysis of this data set results in a 6 component solution which explains 81% of
the variance. The components are summarised below:
Component 1 (Soil): 24% of variance, elements Al, Si, K, Ca, Ti, V and Fe
Component 2 (CFPS): 17% of variance, elements S, Cl, Cr, Ni, Cu, Zn
Component 3 (Indust): 11% of variance, elements V, Mn, Fe and Cu
Component 4 (Salt): 24% of variance, elements Na, Cl, Mn and Co
Component 5 (Diesel): 10% of variance, elements Cr, Zn, Br and Se
Component 6 (New): 8% of variance, elements Sr and Pb
125
This solution is not favoured as Component 6 is only associated with low confidence
data and the overall variance explained is inferior to the 5 component solution. This is
believed to be due to the extra elements bringing more noise to the data set than
meaningful information.
It is interesting to note that the inclusion of the three outlier results improves the model
fit in terms of variance explained to a level comparable (86%) with the 5 factor solution
using 16 elements. However, it is suspected that this may be due to these outlier results
forcing the solution rather than providing a meaningful improvement in source
extraction. Seven components are extracted, with a similarly strange association of Sr
and Pb on one component. The main difference is a new component associated with Si,
Cr, Co, Ni and Cu – it is possible this component is fly ash, although the overall
solution is more difficult to interpret.
The 5 component solution is preferred as it provides the best explanation of variance
once the outlier results are excluded, and yields associations which can be meaningfully
interpreted. Other solutions offer broadly similar associations but can be difficult to
interpret. The good explanation of variance suggests the analysis results are generally
valid and consistent, despite the retention of considerable data which was identified as
having reasonable uncertainty.
5.2.3.4 Impact of Plume on Aerosol Chemistry
This was investigated by comparing the chemistry of the two groups of samples
collected under high and low SO2 regimes. Independent samples t-tests were used to
assess whether there were statistically significant differences in the elemental
concentrations between the means of the high and low SO2 groups (each consisting of
48 samples). These tests were performed on both the entire data set and on a stage by
stage basis, excluding the 3 outliers. Table 5-10 shows selected results for the overall
data set; the “enrichment” is the difference between the means of the high and low SO2
groups, while the t-statistic and significance indicate the likelihood of the two means
being the same (see Appendix H for the full t-test results). A significance of 0.05
means that there is a 95% probability that the means are not equal and the enrichment is
genuine (with an associated confidence interval).
126
Table 5-10: Independent samples t-test comparing means of overall high and low SO2 data sets (summarised from Appendix H).
Element Enrichment t-statistic Significance Si 252 ng m-3 2.06 0.044 S 102 ng m-3 2.10 0.041 Cl 37 ng m-3 2.00 0.161 Cr 1.1 ng m-3 1.28 0.202 Ni 3.6 ng m-3 0.979 0.332 Cu -0.01 ng m-3 -0.035 0.972 Zn -1.12 ng m-3 -1.279 0.204
Statistically significant enrichments were found in the high SO2 cases compared to the
low SO2 cases for Si and S only when all stages were compared. While some of the
other elements associated with the CFPS component - Cl, Cr and Ni - were also
enriched in the high SO2 cases, there was no evidence that this enrichment was
statistically significant. This is hardly surprising given the low concentrations and noise
in the data set: the standard deviations of the data were considerably higher than the
mean values as shown in Table 5-8. Much larger samples with smaller associated
analysis errors would be required to statistically demonstrate any potential enrichment.
Cu and Zn, the other elements associated with the CFPS signature, were depleted in the
high SO2 cases compared to the low SO2 cases, although once again the mean difference
was not statistically significant.
The analysis was also repeated on a stage by stage basis, given the strong association of
a number of components with particle size. The enrichments of Si, S and Cl are shown
in Table 5-11 (see Appendix H for all results). The only other elements to show
statistically significant mean differences were small depletions found for Fe and Cu in
Stage 5, although this may be due to random variation (bearing in mind that on average
one would expect one in 20 observations to show a “significant” difference at a 95%
confidence interval). Si was enriched in almost every stage, although these differences
were not statistically significant. In contrast, S and Cl showed significant enrichment in
the finer fractions; these differences were found to be statistically significant for S in
both Stage 5 and the filter, and for the filter only in the case of Cl. Stage 1 was found to
be depleted in Cl during high SO2 sampling, suggesting a possible reduction in coarse
salt. This data indicates that the power station emissions are causing an enrichment of
both Cl and S in the finest fractions, particularly in the minus 0.3 µm fraction.
127
Table 5-11: Summary of independent samples t-tests comparing means of high and low SO2 data sets for individual stages. Significance is likelihood of observed
enrichment being due to random error with means equal.
There are a number of conclusions that can be drawn from this data:
• The 5 components together explain 98.6% of the mass, compared to 85.5% of
the variance. This is partially due to the higher noise in elements present in low
concentrations, and the fact that all elements are equally weighted through
normalisation.
• While the sums of the chemical profiles are all close to 100%, the profile for
Component 5 includes values which are difficult to interpret - negative values
and values greater than 100% which are possible in a mathematical solution to a
regression problem, but not in physical reality. However, this component
explains very little of the mass and these anomalies are not considered
detrimental to the integrity of the major sources. Note that the source profiles
for the 6 component solution (i.e. including V, Se, Sr and Pb) are considerably
worse in this respect, with the extra noise resulting in more anomalous values
(see Appendix G for details).
• The chemical profiles for soil, salt and CFPS emissions look reasonable in
terms of their elemental composition, although the Ni concentration appears
quite high for the CFPS component (but is consistent in both 5 and 6
component solutions).
130
• The mass attributed to diesel emissions (6%) is highly questionable as the
principal mass component of diesel (i.e. carbon) was not in the elemental suite.
These data allow the average contribution of the CFPS component source to the high
SO2 cases to be determined (at a mean SO2 of 46 ppb). The major contribution is in the
minus 0.3 µm fraction, with an average elemental mass attributed to this component of
1138 ng m-3 This equates to around 2.0 µg m-3 when converted to oxides and accounts
for about 56% of the total mass of this size fraction. The major contributors to the
CFPS mass are 1.1 µg m-3 of sulphur (assumed present as sulphate), 0.6 µg m-3 of Si
(assumed present as SiO2) and 0.2 µg m-3 of Cl assumed present as chloride. These
values are comparable to the mean differences from the t-tests, although the Cl
contribution from CFPS is less from PCA due to some Cl being derived from the salt
component. Uncertainties in the PCA estimates cannot be readily estimated, although
large variations were noted in the mass assigned to this component, despite
comparatively minor variations in the average SO2.
5.2.4 Summary of Cascade Impactor Results
These results provide an interpretation of the composition of the ambient aerosol at
Ravensworth and the likely sources of particles. The calibration results, while slightly
different to the calculations based on theory, confirm that cut sizes for the various stages
range from around 2.5 µm on the first stage to around 0.3 µm on the final stage.
Conditional sampling was used to generate two data sets, each with 8 sets of 6 stages, a
total of 96 samples. IBA analysis of this data was successful in identifying up to 23
elements in varying concentrations. Examination of the errors associated with this data
set led to the exclusion of 3 elements from further analysis, and the identification of a
number of other elements which were present in such low concentrations that the errors
were significant relative to the measured concentrations.
Principal component analysis with varimax orthogonal rotation was conducted on the
resulting data set, after the removal of 3 outliers and the exclusion of 4 further elements
with high uncertainties. Five components were extracted, with four of these in good
agreement with other studies and consistent with what is known about local sources,
despite low concentrations and significant errors in many of the elements. The
components explained 86% of the variance in the original dataset, indicating both that
131
much of the data in the original dataset was valid and that the components adequately
represented the aerosol chemistry, even though the individual samples represented
different size fractions. The elemental associations and assumed interpretation of these
components is as follows:
• Soil – elements Al, Si, K, Ca, Fe and Ti, possibly Cu
• CFPS – elements S, Cl, Ni, Cr and Cu, possibly Ca, Co and Zn
• Salt – elements Na and Cl, also Mn and Co
• Diesel – elements Zn and Br, possibly Na, Cr and Cu
• Industrial – elements Mn and Cu, possibly K, Fe and Co
Analysis of the contributions of these components to the various size fractions
confirmed the expected predominance of the soil component in the coarser sizes (plus 1
µm). However, the salt component was found to be most strongly associated with the
minus 0.3 µm fraction, which was against expectations from the literature. The other
three components were also mainly associated with the minus 0.3 µm particles,
consistent with a combustion or high temperature origin. Confirmation that the
component referred to as CFPS emissions was indeed correctly identified was provided
by two further inter-related pieces of information:
• The factor loadings for this component were greatest in the finest particle sizes
for the high SO2 cases only; the component was not strongly represented in the
low SO2 cases and hence is associated with the plume;
• Statistical tests on the differences between the means of the high and low SO2
data confirmed that the high SO2 cases were significantly enriched in S for Stage
5, while the filter samples were enriched in both S and Cl. These enrichments
were consistent with mass estimates from the PCA solutions, although some of
the Cl in the minus 0.3 µm fraction is associated with the salt component.
Comparison of the high and low SO2 datasets also suggested enrichment of Si, although
not to a statistically significant extent. It is likely that any alumino silicate fly ash
present would be assigned to the soil component as the chemistry is similar. The
possible enrichment of transition metals Cr, Ni, Cu and Zn in the high SO2 samples due
132
to the CFPS source was not found to be statistically significant with considerable noise
evident in the data.
5.3 RESULTS FROM NANOMETER AEROSOL SAMPLER (NAS)
As noted previously, it is believed that this is one of the first attempts to use the NAS
for atmospheric sampling. 12 separate sampling campaigns were conducted with the
NAS, as summarised in Table 5-13 below. As shown in the table, it was found that a
reasonably long sampling period was required for the ambient samples, even with the
smaller electrode, between 15 and 40 hours. It was also noted that the prevailing
weather conditions had a major influence on the amount of material collected, with run
N5 being far more heavily loaded than run N3 despite being somewhat shorter. Field
notes record the second period as significantly more humid, and it is suspected that the
higher loadings were due to increased atmospheric chemistry, even without the
precursor species from the power station plume.
Table 5-13: Summary of NAS campaigns and quality of sample loading in terms of suitability for TEM assessment.
ID Date Location Regime Duration Comments K2 14/10/03 ANSTO Testing – 25 mm
electrode; not neutralised 1 hr Very light loading
K4 14/10/03 ANSTO Testing – neutralised 2 hrs Very light loading K10 7/11/03 Rave SO2 lo – 6 mm electrode 1 hr Light loading N1 19/1/04 Rave SO2 lo 2 hr Light loading N3 2/3 - 5/3 Rave SO2 lo 42 hr Well loaded N5 5/3 - 6/3 Rave SO2 lo 30 hr Overloaded R1 21/1/04 Rave SO2 hi 1 hr Light loading R3 22/1 - 5/2 Rave SO2 hi 26 hr Overloaded R5 5/2 - 10/2 Rave SO2 hi 16 hr Well loaded R7 10/2 - 2/3 Rave SO2 hi 164 hr Overloaded T1 11/3/04 UN Diesel idle 1 min Well loaded T3 11/3/04 UN Diesel start-up 20 sec Well loaded
This section will discuss in detail the results of TEM investigations into two of the
Ravensworth samples, one collected during low SO2 concentrations (N3) and the other
collected during high SO2 concentrations (R5). The two diesel exhaust samples T1 and
T3 will also be discussed in reasonable detail as it was found that they were highly
significant in understanding the samples collected at Ravensworth.
5.3.1 Diesel Samples
The diesel reference samples were collected directly from the tailpipe emissions of a
Ford Transit van using a short section of silicone tubing and the cascade impactor as a
133
pre-cutter (similar to the standard high and low SO2 sampling at Ravensworth). T1, the
idle sample, was collected while the truck had no obvious emissions, whereas T3 was
collected when the emissions were quite clearly visible. Typical images from the two
samples are shown in Figure 5-17. Note that the small, dark angular objects common
to all images are artefacts due to contaminants in the imaging system, not particles.
(a) T1 – idle – scale bar is 100 nm
(b) T1 – idle – scale bar is 100 nm
(c) T3 – start-up – scale bar is 500 nm
(d) T3 – start-up – scale bar is 200 nm
Figure 5-17: Images from diesel exhaust samples T1 and T3.
Emissions from diesel vehicles are known to consist almost entirely of unburnt carbon,
with some enrichment relative to crustal sources of elements such as Zn, Mo, Ni, Cu,
Ag, Cd, Sb, Se and Br (Weckwerth, 2001). Morphological studies have shown the
emissions be chain like agglomerates of primary particles which range from 5-50 nm
and are commonly around 20-25 nm (Wentzel et al., 2003; Braun et al., 2004). The
primary particles in both the idle and start-up images above are in the range 20-45 nm;
134
the agglomerates appeared larger in the start-up sample than in the idle sample. This is
consistent with other findings in the literature, although great interest is also noted in the
fractal dimensions of the agglomerates (Kim et al., 2001; Virtanen et al., 2004). This
was not explored further as the principal aim of this characterisation was to be able
recognise diesel soot in the ambient samples if present. However, it would appear that
the NAS is well suited for studies of diesel particulates, as collection on the TEM
substrate would prevent further agglomeration or modification.
EDX spectra were also collected for several of the soot particles for comparison with
the atmospheric samples (the spot size was 25 nm). Three spectra are shown in Figure
5-18 – one of a blank section of the sample (i.e. the formvar film) and two where soot
particles were observed. All three spectra show what are believed to be system peaks
for the elements Cu, Fe and Co emanating from the copper grid – the varying intensity
of these peaks is believed to be related to the proximity of the analysis point to the edge
of the grid. Also apparent are carbon peaks of varying intensity – the small peak in the
blank spectrum is believed to be due to the carbon in the formvar film. The particles
identified as soot both show strong carbon peaks.
.
It is interesting to note that a small peak is also found for Si, which is unlikely to be due
to other elements (X-Ray emission energies for Si are 1.739 keV and 1.829 keV; see
Appendix J for a table of energies for most elements). There is also evidence of a peak
for O in one of the spectra, and this peak could be masked by the high carbon peak in
the other. The origin of the Si peak is not clear, but it is suspected to be genuine as its
presence in diesel soot has been previously noted, together with a peak for O as found
here (Wentzel et al., 2003). These authors were also unable to identify the origin of the
Si, but ruled out measurement error or contamination (Wentzel et al., 2003).
135
Figure 5-18: EDX spectra from UNSW TEM of blank film and soot particles from sample T1 – horizontal axis is the energy of the detected X-rays, vertical axis is
total counts.
(a) Blank
(b) Soot
(c) Soot
136
5.3.2 Character of Particles Collected Under Low SO 2 Conditions
A selection of the images acquired during the TEM inspection of the N3 sample is
shown in Figure 5-19. The sample was dominated by soot particles, which were readily
identified from the diesel exhaust characterisation described above. EDX spectra were
acquired for a number of these particles as an additional check, yielding similar results
to those shown in Figure 5-18. It was noted that the soot particles found in the ambient
aerosol appeared to be more compact than in the emissions samples, although this was
not systematically investigated. This is suspected to be due to folding in of some of the
longer chains, which were noted to be quite flexible and moved about under the electron
beam if not attached to the surface of the TEM grid.
(a) Diesel soot (scale bar is 100 nm)
(b) Salt residue, probably sulphate (scale is 50 nm)
(c) Dendritic particle, plus soot (scale bar is 100 nm)
(d) Large solid, stable particle, probably crustal (scale bar is 200 nm)
Figure 5-19: Sample images from TEM analysis of low SO2 sample N3.
137
Two other principal types of particles were observed in this sample:
• volatile species which decomposed under the electron beam and sometimes left
a residue, as shown in Figure 5-19 (b);
• stable particles of variable morphology and composition suspected to be derived
from crustal material, as shown in Figure 5-19 (c) and (d).
It was not possible to individually identify particles off-line via image analysis due to a
significantly lower difference in brightness between objects and the background than
with the SEM images. As a result, the images collected were restricted to adequately
describing the range of objects observed rather than attempting to image large numbers
of particles. An indication of the number distribution of the various particle types was
provided by manually counting particles in 2 fields of view for 10 grid areas randomly
chosen from the overall sample. The magnification selected was 40,000 times, which
corresponded to a 250 nm object appearing as 10 mm on the display. This enabled
identification of particles down to around 20-30 nm. Three broad categories of objects
were identified in the low SO2 sample, as shown in Table 5-14; uncertainties have been
estimated using the standard deviation of the ten frame counts and reflect counting
statistics rather than an attempt to consider the effect of volatilisation of particles. Most
of the objects were identified as soot, with small amounts of both residues from unstable
aerosols and solid particles likely to be very fine crustal particles.
Table 5-14: Approximate distribution of particle types in low SO2 sample (N3).
Particle Class and Identification Number % of Total Uncert, % Adj % Chain like agglomerates, identified as soot 929 94 1.2 87 Unstable species; decompose to amorphous residue 36 4 0.2 10 Stable, solid particles, possibly crustal in origin 27 3 0.2 3
Note: Uncertainty estimated to be one standard deviation of individual frame counts divided by total particle count.
The fourth column in Table 5-14 presents a simple sensitivity analysis which considers
what happens when only one out of three unstable particles leaves a visible residue. In
either case, it is clear that the ultrafine aerosol in this case is strongly dominated by soot
particles, with very little crustal some unstable material. The unstable material is
difficult to identify conclusively but is probably secondary particulate matter consisting
138
of ammonium sulphate. This has been widely observed in other studies both overseas
and in Australia (Mamane and Dzubay, 1986; Querol et al., 1999; MSC, 2003).
Sulphates have been observed in a number of TEM studies, although they can be rapidly
volatilised by the electron beam within a matter of seconds (Posfai et al., 1994). Acidic
particles, where the sulphuric acid has not as yet been neutralised, are more hygroscopic
and tend to spread further on the TEM grid (Mamane and Dzubay, 1986; Buseck and
Posfai, 1999). Residues from evaporated particles have been observed to range from
empty halos to crystalline residues (Posfai et al., 1994; Buseck and Posfai, 1999;
Wentzel et al., 2003).
5.3.3 Character of Particles Collected Under High S O2 Conditions
This sample was quite different to the N3 sample acquired under low SO2 conditions,
with a far higher incidence of unstable particles, believed to be secondary aerosols. As
with the previous sample, the stable particles were almost exclusively soot. A sample
of the images acquired during TEM inspection of the R5 sample is shown in Figure
5-20. These images also show much greater variety in the appearance of the residues –
while dull amorphous patches were most common in the low SO2 sample, residues in
the high SO2 sample included halos, amorphous patches, small crystals and
heterogeneous residues.
(a) Diesel soot showing halo of suspected ammonium sulphate particle (scale bar is 100 nm)
(b) Crystalline residue from evaporated acidic sulphate particle (scale bar is 50 nm)
139
(c) Amorphous, heterogeneous residue from scavenger droplet (scale bar is 200 nm)
(d) Capsule shaped residues from unidentified sublimated particles (scale bar is 200 nm)
Figure 5-20: Sample images from TEM analysis of high SO2 sample R5.
The halo observed touching the soot particle in Figure 5-20 (a) is very similar to an
image in the literature of a mixed particle consisting of a sublimated ammonium
sulphate particle and a soot agglomerate (Wentzel et al., 2003), while Figure 5-20 (b) is
similar to images of sublimated ammonium sulphate particles with un-neutralised
sulphuric acid (Buseck and Posfai, 1999). Figure 5-20 (c) appears to be the residue left
by a scavenger water droplet, and contains a range of regions suspected to be of
different species. The grainy appearance of some of the residue is due to the presence
of small crystals as in Figure 5-20 (b). Figure 5-20 (d) shows a comparatively common
capsule shaped residue, with an empty centre. These particles were observed to
decompose rapidly under the electron beam, but it is unclear why the residues are
capsule shaped rather than spherical as reported in the literature for droplets. One
possibility is that the original particles were elongated crystals, producing a capsule
shaped “melt” during sublimation. It is also possible that these objects could have been
biological, although it seems unlikely that they would be as unstable as observed.
It is suspected that the amount of visible residue is influenced by the degree of
neutralisation and the amount of water associated with the original aerosol particle, as
previous studies have indicated that more acidic particles spread out further on the TEM
grid (Mamane and Dzubay, 1986; Buseck and Posfai, 1999). Sulphuric acid droplets,
which are likely to be present at the site, would be expected to evaporate leaving very
little or no residue.
140
EDX spectra were acquired for several of the residues to assist in their identification.
Two typical spectra from the residues are shown in Figure 5-21. Identification of the
elements present is difficult due to the relatively low density of the residues. Spectral
overlap also makes it difficult to determine if there is any N left in the residues from
ammonium ions, as the peak is on the shoulder of the carbon peak. No definite N peaks
were observed, suggesting that any ammonium ions present had decomposed. The C
peaks are more pronounced than in the blank spectrum shown in Figure 5-18,
suggesting that the residue contains carbon, possibly from secondary organic aerosol
formation. Laboratory studies have shown that the formation of low vapour pressure
compounds from common biogenic organics is catalysed by acidic species (Jang and
Kamens, 2001; Czoschke et al., 2003). Additional systematic analysis would be
required to examine this possibility further.
The Cu and Fe/Co peaks are almost certainly system peaks from the grid itself as
before. The combined Cu/Na peak at around 1 keV is believed to be mainly due to the
presence of Na; blank checks (confirmed by the soot spectra in Figure 5-18) indicated
that the Cu peaks at around 0.94 keV are typically around 10% of the height of the peak
at 8.0 keV. It is also possible that some samples showed genuine Fe peaks, although
these were also observed in some of the blank spectra. Probable, non-system peaks are
typically O, Na, Si, S and K, with significant variation between residues in peak height.
This combination of elements was reasonably consistent between the residues, and is
consistent with combustion signature of coal (Cohen, 1998). K is often also associated
with biomass burning, as well as unburnt carbon (Chan et al., 1999b; Song et al., 2001)
141
Figure 5-21: EDX spectra from UNSW TEM of residues from unstable particles – horizontal axis is the energy of the detected X-rays, vertical axis is total counts.
Manual particle counting was also used to provide an indication of the distribution of
the various particle types for this sample. Extra categories were added to the “unstable”
particle class, with these being classified as either capsule shaped, round or amorphous.
Table 5-15 summarises the results from counting of 10 different grid fields, taken over
up to 6 fields of view (counting continued until around 50 soot particles were identified
in each grid field). The data confirms that soot is the dominant aerosol component
observed, accounting for around two thirds of all counted objects. Nearly all the other
objects were residues from unstable secondary particles, with around 21% of these as
either a capsule shaped or approximately circular halo. 8% of the residues were
amorphous or had crystals that were too small to be seen at the magnification used for
the counting exercise (small crystals were observed when examined at higher
magnifications). Only 2% of the observed particles were thought to be derived from
crustal material.
142
Table 5-15: Approximate distribution of particles in high SO2 sample (R5).
Particle Class and Identification Number % of Total Uncert, % Adj % Chain like agglomerates, identified as soot 493 68 0.3 43 Unstable particles; capsule shaped residue 83 12 0.6 22 Unstable particles; round residue 66 9 0.4 17 Unstable particles; amorphous residue 61 8 0.6 16 Stable, solid particles, possibly crustal in origin 17 2 0.2 1
Note: Uncertainty estimated to be one standard deviation of individual frame counts divided by total particle count.
The fourth column in Table 5-15 is a sensitivity analysis (as in Table 5-14) to see how
the distribution changes if only a third of the secondary particles left a residue. If this
were the case, soot would account for 43% of all particles, with secondary particles
more numerous at 55% on a number basis. Note that it is extremely difficult to estimate
the relative mass concentrations of the various particles due to the variable and complex
size and morphology of the soot particles and the short life of unstable species in the
electron beam. An additional complication is the varying degrees of water association
that can be expected with different sulphate species in particular depending on their
degree of neutralisation (Posfai et al., 1998; Buseck and Posfai, 1999).
It is interesting to note that primary particulates from power station emissions were not
identified in the sample. This may be because primary particles were present in only
low concentrations or alternatively because they were difficult to identify. The
concentration of ultrafine primary particles can be estimated from the expected total
mass of primary particles under sampling conditions by making an assumption about
the relative contribution of ultrafine particles. The average SO2 concentrations during
the high SO2 sampling for the cascade impactor sampling (46 ppb) must be used as no
SO2 data is available for the high SO2 sample N5. At this SO2 concentration, the total
expected contribution of primary particulate matter is 1.4 µg m-3 (using dilution factors
from Table 4-1). If around 2% of this material is assumed to less than 0.4 µm (see
Figure 2-6), this would indicate a concentration of around 0.028 µg m-3. This should be
compared with the estimated sulphate component from the factor analysis of the cascade
impactor data of 1 µg m-3. Similarly, the Muswellbrook PM2.5 data can be used for a
first estimate of the amount of soot likely to be present – the data indicates around 14%
of the total loading of 7 µg m-3 is soot, i.e. also approximately around 1 µg m-3. On this
basis, one might expect 1 or 2% of the particle mass to be power station primary
143
particulate matter. The morphology of such particles is not well known, although it is
known that they are formed by evaporation, condensation and coagulation, much in the
same way as soot. It is possible that some of the particles identified as soot purely on
the basis of morphology (given that the counting statistics were derived on the basis of
morphology alone) were in fact power station primary particulates. Further sampling
and more systematic analysis would be required to conclusively answer this question.
5.3.4 Summary of TEM Investigations of NAS Samples
This study has confirmed that the NAS is useful for collecting samples for TEM
assessment, although the impact of particle stability under the electron beam needs to be
considered. Collection of ambient samples was found to require relatively long
exposures to acquire sufficient numbers of particles; sampling for individual events is
therefore unlikely to result in sufficient material being collected for characterisation.
Relatively short times were ideal for generating well loaded samples for diesel exhaust
characterisation, however.
TEM is the most suitable method for analysing these samples due to the superior
resolution over SEM. Particles from various sources can be readily recognised,
although secondary particulates such as ammonium sulphate are unstable under the
electron beam. Analysis of a limited number of samples from Ravensworth indicates
that the ultrafine component of ambient aerosol is composed of soot, minor amounts of
crustal material and variable amounts of secondary particles. Soot was found to
dominate in the low SO2 samples, with very minor contributions from secondary
particles and some fine crustal material. The high SO2 sample studied also contained
significant numbers of soot particles, but considerable quantities of secondary particles
were noted as well. The behaviour of this material under the beam and the nature of the
residues were similar to that described in the literature for ammonium sulphate and
other sulphate species. Significant carbon peaks were noted in the EDX spectra
obtained from these residues compared to the blank film, suggesting the presence of
organic aerosols, perhaps formed through the catalytic oxidation of VOCs in the
presence of acidic seed species as suggested in the literature. However, the amount of
data available is very limited and this can only be a tentative hypothesis without further
investigation.
144
It is interesting to note that primary particulate emissions from power stations were not
observed in the high SO2 samples. While calculations suggested that such particles
should be present in low concentrations, perhaps accounting for 1-2% of the mass, no
such particles were identified. However, it should be noted that the identification of
particles was conducted on the basis of morphology alone, and it is quite possible that
sub micron particulates from power stations would have been mis-identified as soot in
the absence of chemical analysis data.
While the ultrafine component appears to be heavily impacted by traffic emissions,
there is strong circumstantial evidence from these analyses that power station emissions
can also make a significant contribution though the formation of secondary particulates
such as ammonium sulphate and other sulphate aerosols, and potentially through the
catalysis of other reactions.
145
6 INTEGRATED ASSESSMENT OF RESULTS
This chapter will evaluate and integrate the data from the experimental programs in
terms of the study aims and use the results from air pollution modelling to estimate
likely impacts on the nearby townships of Singleton and Muswellbrook. The results
presented in the previous chapter indicate that the contribution of power station
emissions can be divided into two components: a “coarse” fraction of primary
particulate matter or fly ash larger than 1 µm, and a fine component concentrated in the
size fraction less than 0.3 µm. The mass of this fine component is dominated by sulphur
assumed present as sulphate, and this chapter will include a discussion of the formation
of this material.
6.1 CONTRIBUTION OF EMISSIONS TO PARTICULATE MASS
6.1.1 Expectations from Historical Monitoring Data and Air
Pollution Modelling
Air pollution modelling and calculations based on dilution of SO2 suggest that the
contribution of primary emissions from coal fired power stations to ambient particulate
matter is likely to be intermittent and minor compared to other sources. As a
comparison to the estimates below, annual average PM10 measurements at the
Ravensworth site are approximately 25 µg m-3 while TSP measurements are 75 µg m-3.
TAPM modelling predicts that the practical maximum contribution (the 99.9th percentile
value – i.e. only 0.1% of values are higher) of primary particulate emissions to TSP at
the Ravensworth site is 4.6 µg m-3. These estimates assume equivalent dispersion of
gases and particulates and are based on the ratio of TSP and SO2 emissions at the power
station stack; they are therefore likely to be an overestimate as they ignore the impact of
gravitational settling. The maximum contribution to PM10 is expected to be about 50%
of this value i.e. 2.3 µg m-3. The nearby urban areas of Singleton and Muswellbrook are
less impacted by the power station emissions, with estimated maximum contributions to
PM10 of around 1.6 µg m-3 and 1.2 µg m-3 respectively.
The concentration of primary particulate emissions is both minor and highly episodic –
most of the time the contribution of power station emissions is negligible. It should be
146
noted that these estimates also assume that the source emission rates are constant in
time and the same for all point sources; in reality, emissions are likely to show some
variation which will also affect the concentrations. The variations may increase or
decrease concentrations experienced in individual episodes, but are unlikely to greatly
impact the overall distribution of concentrations.
6.1.2 Measurements of “Coarse” Primary Particulate Contributions
Analysis of the samples collected by the Burkard spore sampler indicated a maximum
contribution of “coarse” (>1 µm) power station primary particulates of 0.4 µg m-3 at the
Ravensworth site when the SO2 concentration was 220 ppb. The estimates have a
significant uncertainty due to the significant leverage of coarser (>4-5 µm) particles on
the mass estimates, with a 95% CI from 0.4 to 1.1 µg m-3. The absence of particles
larger than 5-6 µm suggests that coarser particles in the emissions may be settling out of
the plume before it reaches the site. It was not possible to compare the results of the
cascade impactor sampling with these results because the analysis was unable to extract
a separate fly ash source, due to the similarity in the probable chemical profiles for soil
and fly ash.
The experimentally determined concentrations are therefore consistent with - if slightly
lower than – the estimates described above based on TAPM modelling. The maximum
concentrations are low compared to the background concentrations at the site, which are
dominated by crustal material. Although there is no threshold concentration for fine
particulates in terms of health impacts, it is considered that the contribution of primary
particulate emissions from power stations is of little concern given their intermittent
frequency and low maximum levels. The expected concentrations are a small fraction
of the Australian target for PM10, which is not to exceed a 24 daily average of 50 µg m-3
more than 5 times per year by 2008 (NEPC, 1998). Similarly, the concentrations are
well below recently introduced advisory reporting levels for PM2.5, with a 24 hour
average goal of 25 µg m-3 (NEPC, 2003).
6.1.3 Measurements of the Contribution to Fine (Sub micron)
Particulate Matter
Factor analysis was applied to the data from the cascade impactor to derive the
contribution of various components, or potential sources, to the composition of the
147
measured aerosol. These results indicated that the power station emissions can make a
significant contribution to the minus 0.3 µm size fraction, accounting for an estimated
56% of the mass of this size fraction for samples collected when the average SO2 was
46 ppb. The average airborne concentration of particles attributed to this coal fired
power station (CFPS) component for these cases when converted to oxides was 2.0 µg
m-3, approximately three times the concentration of primary particulate PM10 expected
from dilution estimates (0.7 µg m-3 using the dilution factors established from the
TAPM modelling - 0.0152 µg m-3 PM10 per ppb of SO2). The factor analysis indicated
that this mass was largely composed of 1.1 µg m-3 of sulphur assumed present as
sulphate, 0.6 µg m-3 of Si assumed present as SiO2 and 0.2 µg m-3 of Cl assumed present
as chloride.
The Nanometer Aerosol Sampler (NAS) results on particles less than 0.4 µm are
consistent with these observations, with considerable quantities of unstable particles
assumed to be sulphate species observed during the TEM examination of samples
collected at elevated SO2 concentrations. It was found to be difficult to quantify the
mass contribution of these particulates due to their instability during analysis; the
sublimated particles left varying degrees of residues suspected to be related to the
degree of sulphate neutralisation and hence the extent of associated water. The
chemistry of the residues was consistent with literature profiles for coal combustion,
and their appearance was consistent with literature studies of sulphate species using
TEM. Primary particulate emissions were not identified in the high SO2 sample;
similarly, no particles containing chlorine were observed to explain the measured
enrichment in the cascade impactor samples in particles of similar size. However, there
were indications that the residues may have contained carbon from acid seed catalysed
formation of secondary aerosols.
While the sampling program was not targeted towards this material, results from both
the cascade impactor and Nanometer Aerosol Sampler (NAS) therefore indicate the
presence of significant quantities of sulphate species. Such species are commonly
associated with secondary particulate matter, formed by gas to particle conversion in the
atmosphere. Common species in both urban and non-urban environments are sulphuric
acid, ammonium bisulphate, ammonium sulphate and ammonium nitrate (Watson and
Chow, 1994). Coal fired power stations are one of the largest sources of the precursor
148
gases for these species, SO2 and NOx. Other sources include metal smelting and motor
vehicle emissions (Ayers and Granek, 1997), while some sulphur is also derived from
sea salt (Keywood et al., 2000). Ammonium salts are formed from progressive
neutralisation of acid aerosols by atmospheric ammonia, derived largely from livestock
and fertiliser (ApSimon et al., 1987). Possible explanations for the sulphate and other
species observed are discussed below.
6.1.4 Contribution of Power Station Acid Emissions and Sulphur
Dioxide Oxidation
While there were strong indications that the sulphate and chloride material in the minus
0.3 µm size fraction was derived from power station emissions, it was not clear why the
sulphate in particular was reporting to this size fraction. Two possible pathways will be
explored in this section: sulphuric acid emissions from the power stations and
atmospheric gas to particle (or gas to acid) conversion.
The power stations emit considerable quantities of both sulphuric acid and hydrochloric
acid, as shown in Table 2-9. The reported emissions of sulphuric acid are perhaps more
correctly termed emissions of SO3, which are rapidly transformed in the atmosphere to
droplets of H2SO4 (Hewitt, 2001); however they will be referred to here as sulphuric
acid emissions for simplicity. These reported emissions can be used to estimate
expected concentrations during the sampling period to infer the amount of SO2
oxidation and HCl capture required to explain the observed concentrations. The
equivalent concentrations of H2SO4 and HCl at an SO2 concentration of 46 ppb (the
average SO2 concentration measured at the sampling location during the high SO2
sampling periods) are calculated as follows:
3
222
4242
3.16.246000,000,119
000,300,1
6.2
−==
=
gmxxkg
kg
ppbperSOgxSOConcxSOemissions
SOHemissonsSOHMass
µ
µ
Similarly, the approximate concentration of HCl at 46 ppb SO2 can be calculated as 3.0
µg m-3. Comparing these with the effects attributed to the power station emissions (1.1
149
µg m-3 sulphate and 0.2 µg m-3 chloride) suggests that this material could be largely if
not entirely derived from primary acid emissions.
It is also possible that some of the sulphate is produced through oxidation of SO2,
although daily oxidation rates are expected to be less than 1% per hour (Ayers and
Granek, 1997; Hewitt, 2001), and possibly considerably less since most events are due
to overnight accumulation of emissions. Assuming an average winter drainage flow
wind speed of 3 m s-1 (Bridgman and Cameron, 2000), the travel time of the plume to
the monitoring site would be approximately 3.7 hours. An upper estimate of the amount
of sulphuric acid that could be formed by this pathway is as follows:
3
222
4242
8.6037.06.2460628.64
07754.98
%7.36.2
−==
=
mgxxx
xppbperSOgxSOConcxSOtMolecularW
SOHtMolecularWSOHMass
µ
µ
Conversely, the oxidation rate required to explain (all of) the observed sulphate is
approximately 0.15% per hour, although it is not possible to differentiate between
sulphate formed from sulphuric acid in the emissions and sulphate formed through the
oxidation of SO2. However, it is clear that the amount of sulphate observed is
consistent with power station emissions, and probably dominated by primary sulphuric
acid emissions with possibly some additional sulphate formed in the atmosphere. Both
sulphuric acid droplets and particulate matter formed from the reaction with
atmospheric ammonia are likely to be retained by the back up filter. In contrast, most
HCl would be expected to remain in gaseous form with limited conversion to droplet or
solid form (the 0.2 µg m-3 observed suggests a collection of about 7% of the emitted
HCl). It is therefore debatable whether this material is properly termed secondary
particulate matter, as it is likely to consist mainly of the primary emissions as opposed
to the products of atmospheric transformations. This mass is reported separately to the
contribution of primary particulate (fly ash) emissions.
6.2 CONTRIBUTION OF EMISSIONS TO AEROSOL CHEMISTRY
The cascade impactor results indicate that the main impact of power station emissions
on aerosol chemistry is in the minus 0.3 µm size fraction. As discussed above, the bulk
150
of this contribution is attributed to sulphate and chloride species, which are believed to
be derived from sulphuric and hydrochloric acid emissions from the power stations.
The other major component in terms of mass is Si assumed present as SiO2, which is
more difficult to interpret. While some Si vaporisation is expected under combustion
conditions, this is unlikely to account for the 0.6 µg m-3 indicated by the factor analysis,
as the total contribution of primary emissions to PM10 emissions is expected to be of the
order of 0.7 µg m-3 as described above. Similarly, the reason for the observed
enrichment of Si in most size fractions in the high SO2 samples from the cascade
impactor is unclear, and larger than the expected effect due to power station fly ash
emissions.
The CFPS component was also associated with several transition metals, particularly
Ni, Cr and Cu. These elements are expected to be enriched in the emissions due to their
relatively high volatility, and have been noted in other studies (Helstroom et al., 2002).
However, larger samples and reduced analytical errors are required to confirm whether
these associations are significant. It is noted, however, that Cr and Ni in particular were
strongly associated with this component, with the estimated source profile containing
0.8% Cr and 3.6% Ni. Other elements expected to be associated with this source
include several “trace” and “matrix” elements, and some were weakly associated with
the component. Additional trace elements that could be expected from other studies
include Zn, Cd, As, V, Pb and Mn (Helstroom et al., 2002); of these Cd and As were
not detected in significant quantities to be included in the elemental suite, while V and
Pb were present in such low concentrations that uncertainties were considerable and
they were eliminated from the final data set for factor analysis. Zn and Co were weakly
associated with the CFPS component, while Mn was not. The general absence (i.e.
weak to no association) of the matrix elements such as Al, Si, Ca, K, Fe and Ti with this
component is thought to be due to the inability of the analysis to differentiate between
soil and primary particulate matter derived from the inorganic constituents in the coal.
6.3 CONTRIBUTION OF POWER STATION EMISSIONS TO
ULTRAFINE PARTICULATES
The minus 0.4 µm size fraction of the aerosol was studied using TEM, allowing
particles as small as 20-30 nm to be examined and identified. Diesel soot was found to
be a major component of the aerosol under both high and low SO2 conditions, with
151
comparatively little crustal material observed. The major difference between samples
collected under high and low SO2 conditions was in the amount of material present that
was found to be unstable under the electron beam. The sample collected under higher
SO2 conditions had significant quantities of unstable particles assumed to be sulphate
species; while these particles could not be characterised in detail due to their instability
in the electron beam, their behaviour and the nature of the residues was consistent with
literature expectations. There were also indications that the particles may have had
some water associated with them, which is consistent with sulphuric acid droplets and
partially neutralised sulphate species, which are strongly hygroscopic (Watson and
Chow, 1994). No particles were observed in the high SO2 samples that could be
identified as particulate emissions from power stations, although it was expected that
most of the mass of this material would be in the larger size fractions.
6.4 SUMMARY OF RESULTS
The contribution of power station emissions to atmospheric particulate matter at the
Ravensworth monitoring site can be summarised as follows:
Contribution to Mass:
• TAPM modelling assuming equivalent dispersion of gases and particles from the
power station stacks indicated a practical maximum (99.9 percentile)
concentration to PM10 of 2.3 µg m-3 for primary particulates (fly ash >1 µm) at
an SO2 concentration of 150 ppb
• Spore sampler determinations indicated that the highest observed mass
contribution of particles larger than 1 µm was 0.4 µg m-3 with a 95% confidence
interval of 0.40-1.12 µg m-3 for SO2 concentrations from 68 to 220 ppb
• Cascade impactor analyses indicated that species derived from the primary
sulphuric and hydrochloric acid emissions made a significant contribution to the
minus 0.3 µm size fraction. The average mass attributed to power station
emissions at a SO2 concentration of 46 ppb was 2.0 µg m-3; the contribution of
primary particulate matter to PM10 at 46 ppb is estimated from dilution
calculations at 0.7 µg m-3
152
Contribution to Aerosol Chemistry:
• Main impact of power station emissions is in the minus 0.3 µm size fraction,
consisting of an average (oxide) mass of 1.1 µg m-3 sulphur as sulphate, 0.3 µg
m-3 chloride and 0.6 µg m-3 Si as SiO2.
• Possible enrichment of transition metals, particularly Cr, Ni and Cu.
Contribution to Ultrafines (minus 0.4 µm):
• Primary particulate emissions not a significant component
• Unstable material thought to be derived from sulphuric acid emissions
significantly enriched in high SO2 conditions.
6.4.1 Assessing Impacts on Nearby Urban Areas
Results of the TAPM modelling can also be used to estimate potential impacts at the
nearby urban areas of Singleton and Muswellbrook. These townships are further from
the power stations than the Ravensworth monitoring site and are expected to experience
comparatively fewer events and lower maximum concentrations. The predicted
maximum contributions of power station primary emissions to PM10 are around 1.6 µg
m-3 and 1.2 µg m-3 at Singleton and Muswellbrook respectively compared to 4.6 µg m-3
at Ravensworth. The contribution of emissions to the submicron size fraction is more
difficult to estimate as additional secondary particulate formation can be expected
compared to the Ravensworth site given the extra travel time. It is possible that the
dilution of the primary acid gas emissions will be offset by additional secondary aerosol
formation, although this has not been modelled as the main focus of this project was on
primary particulate emissions.
The results of this study are consistent with results from international studies, both in
terms of the magnitude of primary coal fired power station particulate contributions and
the significance of the sulphate component in the finer size fractions of the ambient
aerosol. The mass contributions are, however, significantly lower than the estimates
from dilution calculations made when Liddell was equipped with less efficient ESPs for
emission controls (Jakeman and Simpson, 1987). The indications of incomplete
sulphate neutralisation are also consistent with other studies.
153
7 CONCLUSIONS & RECOMMENDATIONS
This chapter reviews the basis, methodology and findings of the study and makes
recommendations for future research.
7.1 CONCLUSIONS FROM LITERATURE REVIEW
This study was conducted to improve the understanding of the contribution that power
station particulate emissions make to ambient particulate matter, in an Australian
context. A comprehensive literature survey to review the current state of knowledge
and identify appropriate objectives for this work concluded:
• There is a substantial body of evidence indicating fine airborne particulate
matter has negative impacts on human health;
• There are many sources, both natural and anthropogenic, that contribute to the
ambient aerosol. Combustion aerosols are of particular concern due to their
relatively fine size compared to other sources;
• Coal combustion is responsible for a significant percentage of anthropogenic
particulate emissions less than 10 µm (PM10);
• While source emission data is readily available from the National Pollutant
Inventory (NPI), the contribution of power station emissions to the ambient
aerosol is less clearly understood;
• The Hunter Valley has two large coal fired power stations (total capacity 5.6
GW) fitted with fabric filters for emission control;
• Understanding of the meteorology in the Hunter Valley is relatively mature due
to past studies of fugitive dust from coal mining and sulphur dioxide emissions
from the stations;
• Sulphur dioxide can be expected to be a useful indicator species for the presence
of the plume;
• Contributions to the ambient aerosol can be expected to be dominated by
primary particulate emissions, with slow gas to particle conversion rates;
• Primary particulate emissions are formed from mineral matter in the coal, and
consist largely of the oxides of silicon and aluminium;
• Fine particulate emissions from coal combustion have been shown to be
enriched in potentially toxic trace elements including transition metals;
154
• Few studies have characterised fabric filter emissions, but most of the mass (75-
98%) of emissions is expected to be larger than 1 µm
Three key areas were identified as goals for the experimental program. They were to
develop and implement methodologies for assessing the contribution of power station
primary particulate emissions in terms of their:
o contribution to PM mass;
o contribution to aerosol chemistry; and
o contribution to ultrafine particulates
7.2 SAMPLING PROGRAM AND METHODOLOGY
7.2.1 Study Site Selection
An integrated approach was developed to meet these objectives using a combination of
field sampling, assessment of historical data and air pollution modelling. The two
power stations are located between the townships of Singleton and Muswellbrook, with
several existing monitoring sites in the townships and closer to the stations. Existing
monitoring sites were preferred for the sampling campaign due to the security of the
sites, existing infrastructure and the availability of historical data. The selection and
validation of the Ravensworth site, 11 km to the south east of the two power stations, is
discussed below.
7.2.2 Conclusions from Historical Data and Air Poll ution Modelling
Analysis of historical air quality monitoring data and air pollution modelling were used
to assess the various existing sites. This exercise confirmed that the Ravensworth site
was the best of the established monitoring sites for the assessment of power station
impacts, experiencing moderate impacts from power station emissions. Results
indicated that the Ravensworth site was more impacted than the nearby urban areas of
Singleton and Muswellbrook, as it was closer to the stations and more impacted by
down valley drainage flows during winter when dispersion of emissions is generally
weaker. Scaling factors were developed to allow the experimental results to be
extended to a wider area.
155
The historical data also showed the episodic nature of events, with highly variable
impacts depending on the prevailing weather conditions. Events were most common
between 8am and 12 noon, due to trapping of pollutants above a stable layer overnight,
which was mixed to ground under the influence of solar heating. The frequency and
duration of events was highly variable; thus the contribution of emissions to the ambient
aerosol was also expected to be intermittent and variable.
7.2.3 Experimental Program
The three parameters of interest were tackled with separate sampling methodologies as
follows:
o Contribution to mass: a time resolved record of super-micron particles was
collected on carbon tape using a Burkard spore sampler; this tape was analysed
by Scanning Electron Microscopy (SEM) to provide estimates of airborne
concentrations of fly ash particles.
o Contribution to aerosol chemistry: size segregated aerosol samples were
collected using a cascade impactor, in the presence and absence of the plume as
indicated by SO2 monitoring. This apparatus uses inertial impaction to collect
particles of progressively finer sizes by throttling the gas flow through
progressively smaller apertures to increase the velocity and likelihood of
impaction on a surface. Ion Beam Analysis (IBA), where the samples were
bombarded with high energy protons to generate emission spectra, was used to
generate a broad elemental suite on the comparatively small masses collected.
The data was explored using statistical methods to apportion sources and
estimate the contribution of power station emissions.
o Contribution to ultrafines: samples of minus 0.4 µm particles were collected
using a relatively new instrument, the Nanometer Aerosol Sampler (NAS).
Samples were collected in the presence and absence of the plume, and analysed
using Transmission Electron Microscopy (TEM).
While the three methodologies are independent, there is a degree of overlap in terms of
the sizes collected.
156
7.3 SUMMARY OF RESULTS
The key findings from the various experimental campaigns are summarised below. The
results from the different aspects of the experimental program are complementary and
consistent where overlap does occur.
7.3.1 Burkard Spore Sampler
o Fly ash was readily recognised and typically associated with plume events as
indicated by SO2 monitoring data;
o Events ranging from 68 to 220 ppb SO2 were assessed; while the incidence of fly
ash correlated with SO2, mass concentrations were more noisy due to the sensitivity
of the determinations to larger (>4-5 µm) particles;
o Coarse fly ash (>1 µm) contributions to atmospheric PM were episodic and variable,
with a maximum estimated contribution of 0.4 µg m-3 in the samples (with a 95% CI
of 0.4 to 1.1 µg m-3)
7.3.2 Cascade Impactor
o 16 different sampling campaigns were conducted, 8 when the SO2 concentration
exceeded 20 ppb (with an average of 46 ppb) and 8 under low SO2 conditions. Each
run generated 6 size fractions, with the stage cut sizes ranging between about 2.5 µm
and 0.3 µm (i.e. 96 samples for analysis);
o The validated IBA results provided airborne concentration data for 20 elements,
with varying associated uncertainties depending on concentrations;
o Factor analysis (Principal Component Analysis with varimax orthogonal rotation)
was used to characterise the sources that contributed to the size fractionated aerosol
samples. 5 components were extracted with 4 of these were in good agreement with
literature profiles for soil, salt, diesel and CFPS emissions. An unidentified
industrial component (perhaps from metal smelting) was also extracted. The
components generally showed expected size associations e.g. soil enriched in the +1
µm size fraction and diesel primarily in the minus 0.3 µm size fraction;
o The CFPS component was associated with S and Cl as well as the transition metals
Cr, Ni, and Cu. The profile was missing some elements found by others due to
partly to the elemental suite available, while other elements were more highly
correlated with the soil and salt components. The CFPS component was
significantly enriched in the high SO2 cases as expected.
157
o Mass contributions for the CFPS component were estimated from the factor analysis
results and found to be consistent with direct comparison of the means of the high
and low SO2 samples. Enrichments in the minus 0.3 µm size fraction were
estimated at 2.0 µg m-3 for all elements associated with this component (on an oxide
basis), while comparison of the means indicated a contribution of 1.4 µg m-3 for S
and Cl alone.
o The transition metals associated with the CFPS component were not found to be
enriched to a statistically significant effect, although these elements were present in
low concentrations and uncertainties were considerable.
7.3.3 Nanometer Aerosol Sampler (NAS)
o A limited number of samples were collected at the Ravensworth site in both high
and low SO2 conditions, as well as reference samples from diesel exhaust. Striking
differences were noted in the TEM examination of the high and low SO2 cases;
o A low SO2 sample was dominated by diesel soot, with minor contributions from fine
crustal material and small amounts of unstable species believed to be secondary
particulate matter such as ammonium sulphate;
o A high SO2 sample was also found to contain considerable quantities of diesel soot,
as well as significantly more unstable material which was difficult to characterise
and quantify as the particles were vaporised by the TEM beam almost
instantaneously. Residues from sublimated particles were consistent with literature
accounts of sulphate particles. The residues also had chemistry consistent with a
coal combustion signature, and morphology indicative of variable hydration
suspected to be related to the degree of sulphate neutralisation. The chemistry data
from the residues suggested the presence of secondary organic aerosols; it has been
noted in laboratory studies that acidic aerosols can catalyse the oxidation of VOCs.
7.4 INTEGRATED ASSESSMENT OF RESULTS
7.4.1 Contribution of Particulate Emissions to Mass
The contribution of primary particulate emissions from power stations has been found
to be highly episodic in nature and generally low in significance. TAPM modelling
predicted maximum expected contributions to PM10 of around 2.3 µg m-3 at the
Ravensworth monitoring site, at an SO2 concentration of 150 ppb. Less significant
158
impacts were predicted at the nearby urban areas of Muswellbrook (maximum 1.2 µg m-
3) and Singleton (maximum 1.6 µg m-3). Analysis of the samples from the Burkard
spore sampler indicated comparable if slightly lower maximum concentrations, with the
highest determination for plus 1 µm particles around 0.4 µg m-3 at an SO2 concentration
of 220 ppb. Uncertainties were significant due to the large impact of 4-5 µm particles
on the mass concentrations, yielding a 95% confidence interval from 0.40-1.12 µg m-3.
Annual average PM10 measurements at Ravensworth are 25 µg m-3, dominated by non-
power station sources such as wind blown soil and emissions from traffic.
Source apportionment of the aerosol based on the size fractionated chemistry data from
the cascade impactor samples indicated that the coal fired power stations (CFPS)
emissions were making a significant contribution to the minus 0.3 µm size fraction.
This material is believed to be largely formed from the emissions of sulphuric and
hydrochloric acid from the power stations. The estimated contribution of CFPS
emissions to this size fraction (on an oxide basis) for the high SO2 samples was 2.0 µg
m-3 at an average SO2 of 46 ppb. This is around 2.8 times the expected contribution of
primary particulates of 0.7 µg m-3 based on TAPM modelling. However, it should be
noted that this material is also subject to the same episodic nature as the primary
particulate contribution as it is the direct impact of the plume. Emissions may also
contribute to background sulphate concentrations due to regional scale impacts.
7.4.2 Contribution to Aerosol Chemistry
The CFPS component in the minus 0.3 µm size fraction was composed mainly of
sulphur assumed present as sulphate (oxide mass 1.1 µg m-3), with some chloride (0.2
µg m-3) and silicon assumed present as SiO2 (0.6 µg m-3). The measurements of
sulphate and chloride concentrations are consistent with the emissions of sulphuric acid
and hydrochloric acid from the power stations; it was not possible to establish the extent
of neutralisation of these species and the degree of post stack transformation. It is also
likely that some of the observed sulphate was due to atmospheric gas to particle
conversion of SO2, although it is not possible to distinguish between the two possible
pathways.
The observed silicon was not readily explained, but may be due in part to fume
emissions from the vaporisation and condensation of silica under combustion
159
conditions. Low concentrations of chromium, nickel, copper and zinc were also found
to be associated with the fine mass originating from power station emissions, although
this was not statistically significant due to the low sample masses obtained and the
comparatively large analysis errors on these elements.
7.4.3 Contribution to Ultrafine Particles (minus 0. 4 µm)
The presence of significant quantities of sulphate species in the aerosol under high SO2
conditions was consistent with the results of TEM investigations of samples collected
by the NAS. Although these particles were unstable under the electron beam and
difficult to characterise, the appearance and chemical composition of the residues from
the sublimated particles were consistent with literature studies of various sulphate
species commonly present in the atmosphere. While it was not possible to determine
the chemical composition of these species, there were indications of variable hydration
consistent with incomplete neutralisation of acid species. This is consistent with the
hypothesis that these particles are derived from the primary acid emissions from the
power stations. Similarly, indications of carbon in the residues were consistent with the
potential acid seed catalysed formation of secondary organic aerosol.
7.5 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE
RESEARCH
The contribution of primary power station particulates is believed to have minimal
impact on ambient particulate mass even within 10-15 km of the power stations, with
episodic events of comparatively minor significance. The impact of particulate matter
derived from power station sulphuric and hydrochloric acid emissions appears to be
slightly more significant, although subject to the same episodic nature. The
composition of this material was not able to be conclusively determined, but is likely to
consist of sulphate and chloride species with some silica and possibly traces of
transition metals. Results suggested that these species were only partly neutralised.
While emissions are expected to have only a minor and intermittent contribution to the
ambient aerosol even close to source, some uncertainty remains in the contribution of
power station emissions to the minus 0.3 µm size fraction. Additional characterisation
work is recommended in terms of the composition and nature of this fine particulate
matter (minus 0.3 µm) attributed to power station emissions as follows:
160
• Characterisation of sulphate species and rate of conversion to sulphuric acid,
ammonium sulphate and other sulphate species;
• Investigation of potential acid seed catalysed secondary organic aerosol
formation;
• Clarification of the occurrence and nature of silica;
• Investigation of the potential association of this fine particulate with transition
metals, notably chromium and nickel.
161
REFERENCES
Alliksaar, T. and Punning, J. (1998), The spatial distribution of characterised fly-ash particles and trace metals in lake sediments and catchment mosses: Estonia, Water, Air and Soil Pollution, 106, 219-239.
Andrade, F., Orsini, C. and Maenhaut, W. (1994), Relation between aerosol sources and meteorological parameters for inhalable atmospheric particles in Sao Paulo City, Brazil, Atmospheric Environment, 28, 2307-15.
ApSimon, H. M., Kruse, M. and Bell, J. N. B. (1987), Ammonia emissions and their role in acid deposition, Atmospheric Environment (1967-1989), 21, 1939-46.
Artaxo, P., Oyola, P. and Martinez, R. (1999), Aerosol composition and source apportionment in Santiago de Chile, Nuclear Instruments & Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms, 150, 409-416.
Ayers, G. P., Carras, J. N., Gillett, R. W., Granek, H., Hibberd, M. F., Lilley, W. E., Manins, P. C., Merry, R., Mullins, P. J., Munksgaard, N., Smith, C., Szemes, F., Parry, D. and Williams, D. (1999a), MIM emissions and the environment final report to Mount Isa Mines Limited, 5/2/03, CSIRO Atmospheric Research, CSIRO Energy Technology, CSIRO Land and Water, Northern Territory University, http://www.env.qld.gov.au/environment/environment/air/Execsummary.pdf
Ayers, G. P. and Granek, H. (1997), Atmospheric sulfur and its relevance to acid deposition over Australia, Clean Air (Aust), 31, 38-46.
Ayers, G. P., Keywood, M. D. and Gras, J. L. (1999b), TEOM vs. manual gravimetric methods for determination of PM2.5 aerosol mass concentrations, Atmospheric Environment, 33, 3717-3721.
Ayers, G. P., Keywood, M. D., Gras, J. L., Cohen, D., Garton , D. and Bailey, G. M. (1999c), Chemical and physical properties of Australian fine particles: a pilot study, CSIRO Division of Atmospheric Research and Australian Nuclear Science and Technology Organisation (ANSTO), pp. 99.
Bargagli, R., Cateni, D., Nelli, L., Olmastroni, S. and Zagarese, B. (1997), Environmental impact of trace element emissions from geothermal power plants, Archives of Environmental Contamination & Toxicology, 33, 172-181.
Barr, E. B., Newton, G. J. and Yeh, H. C. (1982), Nonideal collection characteristics of a cascade impactor with various collection substrates, Environmental Science and Technology, 16, 633-5.
Bates, T. S., Lamb, B. K., Guenther, A., Dignon, J. and Stober, R. E. (1992), Sulfur emissions to the atmosphere from natural sources, Journal of Atmospheric Chemistry, 14, 325-337.
Battarbee, J. L., Rose, N. L. and Long, X. Z. (1997), A continuous, high resolution record of urban airborne particulates suitable for retrospective microscopical analysis, Atmospheric Environment, 31, 171-181.
Benitez, J. (1993), Process engineering and design for air pollution control, Prentice Hall, Englewood Cliffs, New Jersey.
Benson, S. A., McCollor, D. P., Eylands, K. E., Laumb, J. D. and Jensen, R. R. (2001), Characterization of particulate matter collected with a Burkard sampler, Preprints of Symposia - American Chemical Society, Division of Fuel Chemistry, 46, 296-301.
162
Bettinelli, M., Spezia, S., Baroni, U. and Bizzarri, G. (1998), Determination of trace elements in power plant emissions by inductively coupled plasma mass spectrometry, comparison with other spectrometric techniques, Microchemical Journal, 59, 203-218.
Biswas, J., Hogrefe, C., Rao, S. T., Hao, W. and Sistla, G. (2001), Evaluating the performance of regional-scale photochemical modeling systems. Part III-Precursor predictions, Atmospheric Environment, 35, 6129-6149.
Bohm, P., Wolterbeek, H., Verburg, T. and Musilek, L. (1998), The use of tree bark for environmental pollution monitoring in the Czech Republic, Environmental Pollution, 102, 243-250.
Braun, A., Huggins, F. E., Seifert, S., Ilavsky, J., Shah, N., Kelly, K. E., Sarofim, A. and Huffman, G. P. (2004), Size-range analysis of diesel soot with ultra-small angle X-ray scattering, Combustion and Flame, 137, 63-72.
Bridgman, H. (1998), Particulate air pollution and meteorology in the upper Hunter Valley, In Geotechnical engineering and engineering geology in the Hunter Valley(Eds, Fityus, S., Hitchcock, P., Allman, M. and Delaney, M.), Conference Publications, Springwood NSW 2777, University of Newcastle, pp. 185-197.
Bridgman, H. A. (1989), Acid rain studies in Australia and New Zealand, Archives of Environmental Contamination and Toxicology, 18, 137-146.
Bridgman, H. A. (1992), Evaluating rainwater contamination and sources in southeast Australia using factor analysis, Atmospheric Environment, 26A, 2401-2412.
Bridgman, H. A. and Cameron, C. (2000), Air quality, In Advances in environmental management in the Hunter region, (Eds: Whitehead, J. H., Geary, P. M. and Kidd, R. W.), Hunter Environmental Institute, Marrickville, NSW 2204, pp. 188-209.
Bridgman, H. A. and McManus, P. (2000), Weather, climate and air quality: the Hunter's atmosphere, In Journeys: The making of the Hunter Region, (Eds: McManus, P., O'Neill, P. and Loughran, R.), Allen & Unwin, St Leonards, Sydney, NSW, Australia, pp. 42-66.
Briggs, G. A. (1975), Plume rise predictions, In Lectures on air pollution and environmental impact analysis; presented at the AMS Workshop on Meteorology and Environmental Assessment, American Meteorological Society, Boston, Mass., pp. 59-111.
Brimblecombe, P. (1987), The big smoke: a history of air pollution in London since medieval times, Methuen, London; New York.
Buhre, B. J. P., Hinkley, J. T., Gupta, R. and Wall, T. F. (2001), Fine particles from coal combustion: formation and characterisation, In 18th International Pittsburgh Coal Conference, Vol. cd-rom, (Ed, Conference, P. C.), Pittsburgh Coal Conference, Newcastle, NSW, Australia, pp. Paper # 22-3.
Burkard (2000), Agronomic product data sheet - Burkard 7 day recording volumetric spore sampler, 12/12/01, Burkard Scientific Agronomics Division, http://www.burkardscientific.co.uk/Agronomics/Agronomics_Product_Spore_sampler.htm
Burkard (2002), Personal communication, email of 25/1/02, (Ed, Hinkley, J.), pp. email from Bryan Willi and Burkard Scientific.
Buseck, P. R. and Posfai, M. (1999), Airborne minerals and related aerosol particles: effects on climate and the environment, Proceedings of the National Academy of Sciences of the United States of America, 96, 3372-3379.
CAFE (2004), Second position paper on particulate matter, final draft, Clean Air for Europe Working Group on Particulate Matter, 24/01/05, European Commission,
Carr, R. C. and Smith, W. B. (1984), Fabric filter technology for utility coal-fired power plants, Journal of the Air Pollution Control Association, 34, 79-89.
Carras, J. N. (1995), The transport and dispersion of plumes from tall stacks, In Environmental aspects of trace elements in coal, (Eds: Swaine, D. J. and Goodarzi, F.), Kluwer Academic Publishers Ltd, Dordrecht, The Netherlands, pp. 146-177.
Carras, J. N., Lange, A. L., Thomson, C. J. and Williams, D. J. (1992), Behaviour of the power station plumes in the Hunter Valley/Central Coast region of NSW, Volume I: results of the intensive field campaigns, CSIRO, Sydney, pp. 120 (approx).
Carras, J. N. and Williams, D. J. (1988), Measurements of relative σy up to 1800km from a single source, Atmospheric Environment, 22, 1061-1069.
Cereda, E., Marcazzan, G. M. B., Grime, G. W. and Mattachini, F. (1996), Macroscopic and microscopic approach to the study of enrichment phenomena in the electrostatic precipitation of coal fly ash particles, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials & Atoms, 109, 696-700.
CEU (1999), COUNCIL DIRECTIVE 1999/30/EC of 22 April 1999 relating to limit values for sulphur dioxide, nitrogen dioxide and oxides of nitrogen, particulate matter and lead in ambient air, In Official Journal L163, 29.06.1999.
Chambers, A. J. and Bridgman, H. A. (1983), Air quality in the middle Hunter: the extensive study periods, University of Newcastle, Department of Mechanical Engineering, Newcastle, Australia, pp. 87.
Chambers, J., Bridgman, H. A. and Long, G. (1982), Dispersion of a buoyant plume from tall stacks in the middle Hunter Valley, Clean Air (Aust.), 16, 68-73.
Chan, Y., Simpson, R. W., McTainsh, G. H. and Vowles, P. D. (1999a), Sources of Brisbane airborne dusts from electron microprobe analysis of individual particles, Clean Air (Aust.), 33, 26-31.
Chan, Y. C., Simpson, R. W., McTainsh, G. H., Vowles, P. D., Cohen, D. D. and Bailey, G. M. (1997), Characterisation of chemical species in PM2.5 and PM10 aerosols in Brisbane, Australia, Atmospheric Environment, 31, 3773-3785.
Chan, Y. C., Simpson, R. W., McTainsh, G. H., Vowles, P. D., Cohen, D. D. and Bailey, G. M. (1999b), Source apportionment of PM2.5 and PM10 aerosols in Brisbane (Australia) by receptor modelling, Atmospheric Environment, 33, 3251-3268.
Chan, Y. C., Vowles, P. D., McTainsh, G. H., Simpson, R. W., Cohen, D. D., Bailey, G. M. and McOrist, G. D. (2000), Characterisation and source identification of PM10 aerosol samples collected with a high volume cascade impactor in Brisbane (Australia), Science of the Total Environment, 262, 5-19.
Cheng, R. J., Mohnen, V. A., Shen, T. A., Current, M. and Hudson, J. B. (1976), Characterization of particulates from power plants, Journal of the Air Pollution Control Association, 26, 787-790.
Chow, J. C. and Watson, J. G. (1998), Guideline on speciated particulate monitoring, 2/9/02, Desert Research Institute, www.epa.gov/ttn/amtic/files/ambient/pm25/spec/drispec.pdf
Christian, G. D. and O'Reilly, J. E. (Eds.) (1986), Instrumental analysis, Allyn and Bacon, Newton, Massachusetts, USA.
164
Clarke, L. B. and Sloss, L. L. (1992), Trace elements - emissions from coal combustion and gasification, IEA Coal Research, London.
Clausnitzer, H. and Singer, M. J. (1999), Mineralogy of agricultural source soil and respirable dust in California, Journal of Environmental Quality, 28, 1619-1629.
Cohen, D. D. (1992), Ion beam analysis techniques in aerosol analysis, Clean Air, 26, 113-121.
Cohen, D. D. (1997), Ion beam analysis methods in aerosol analysis, quality assurance and inter-technique comparisons, Harmonization of Health Related Environmental Measurements Using Nuclear and Isotopic Techniques, Proceedings of an International Symposium, Hyderabad, India, Nov. 4-7, 1996, 457-472.
Cohen, D. D. (1998), Characterisation of atmospheric fine particles using IBA techniques, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials & Atoms, 138, 14-22.
Cohen, D. D., Bailey, G. M. and Kondepudi, R. (1996), Elemental analysis by PIXE and other IBA techniques and their application to source fingerprinting of atmospheric fine particle pollution, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials & Atoms, 109/110, 218-226.
Cohen, D. D., Stelcer, E., Hawas, O. and Garton, D. (2004), IBA methods for characterisation of fine particulate atmospheric pollution: a local, regional and global research problem, Nuclear Instruments & Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms, 219-220, 145-152.
CPCB (1994), National ambient air quality standards (India), 12/03/04, Central Pollution Control Board, Government of India, http://www.cpcb.delhi.nic.in/standard2.htm
Culling, C. F. A. (1974), Modern microscopy; elementary theory and practice, Butterworths, London.
Czoschke, N. M., Jang, M. and Kamens, R. M. (2003), Effect of acidic seed on biogenic secondary aerosol growth, Atmospheric Environment, 37, 4287-4299.
De Nevers, N. (1995), Air pollution control engineering, McGraw-Hill, New York. DEFRA (2004), Expert group report on particulate matter. Chapter 5: Methods for
monitoring particulate concentrations, 24/1/05, Department for Environment, Food and Rural Affairs (UK), http://www.defra.gov.uk/corporate/consult/particulate%2Dmatter/
Desrosiers, R. E., Riehl, J. W., Ulrich, G. D. and Chiu, A. S. (1979), Submicron fly-ash formation in coal-fired boilers, In Seventeenth Symposium (International) on Combustion. The Combustion Institute, pp. 1395-1403.
Dockery, D. W., Pope, C. A., Xu, X. P., Spengler, J. D., Ware, J. H., Fay, M. E., Ferris, B. G. and Speizer, F. E. (1993), An association between air pollution and mortality in six U.S. cities, New England Journal of Medicine, 329, 1753-1759.
DUAP (1997), Upper Hunter cumulative impact study and action strategy, Department of Urban Affairs and Planning (DUAP), Sydney, Australia, pp. 154.
Dye, A. L., Rhead, M. M. and Trier, C. J. (2000), The quantitative morphology of roadside and background urban aerosol in Plymouth, UK, Atmospheric Environment, 34, 3139-3148.
Dzubay, T. G. and Mamane, Y. (1989), Use of electron microscopy data in receptor models for PM-10, Atmospheric Environment, 23, 467-476.
Erickson, T. A., Benson, S. A., Laumb, J. D., Jensen, R. R., Eylands, K. E. and D.P., M. (2001), Particulate matter: trends and analysis using multiple sampling and
165
characterisation methods, In 18th International Pittsburgh Coal Conference, Pittsburgh Coal Conference, Newcastle, NSW, Australia, pp. CD-ROM Paper # 22-1.
Fisher, G. L., Prentice, B. A., Silberman, D., Ondov, J. M., Bierman, A. H., Ragaini, R. C. and McFarland, A. R. (1978), Physical and morphological studies of size classified fly ash, Environmental Science and Technology, 12, 447-451.
Fisher, N. S. and Wang, W. X. (1998), Trophic transfer of silver to marine herbivores - a review of recent studies, Environmental Toxicology & Chemistry, 17, 562-571.
Frenz, D. A. (1999), Comparing pollen and spore counts collected with the rotorod sampler and Burkard spore trap (Review), Annals of Allergy, Asthma, & Immunology, 83, 341-347.
Fung, Y. S. and Wong, L. W. Y. (1995), Apportionment of air pollution sources by receptor models in Hong Kong, Atmospheric Environment, 29, 2041-2048.
Garty, J., Weissman, L., Cohen, Y., Karnieli, A. and Orlovsky, L. (2001), Transplanted lichens in and around the Mount Carmel National Park and the Haifa Bay industrial region in Israel: Physiological and chemical responses, Environmental Research, 85, 159-176.
Ghosal, S. and Self, S. A. (1995), Particle size-density relation and cenosphere content of coal fly ash, Fuel, 74, 522-529.
Gibbon, D. L. (1979), Microcharacterisation of fly-ash and analogs: the role of SEM and TEM, Scanning Electron Microscopy, 1, 501-510.
Glikson, M., Rutherford, S., Simpson, R. W., Mitchell, C. A. and Yago, A. (1995), Microscopical and submicron components of atmospheric particulate matter during high asthma periods in Brisbane, Queensland, Australia., Atmospheric Environment, 29, 549-562.
Glikson, M., Simpson, R., Polach, H. and Taylor, J. (1988), Microscopical TSP studies comparing a city centre and suburban sites in Canberra, Australia, Atmospheric Environment, 22, 1745-1758.
Goldstein, J. I. (2003), Scanning electron microscopy and X-ray microanalysis, Kluwer Academic/Plenum, New York.
Gonzalez, C. M. and Pignata, M. L. (1997), Chemical response of the lichen Punctelia Subrudecta (nyl.) krog transplanted close to a power station in an urban-industrial environment, Environmental Pollution, 97, 195-203.
Graham, K. A. and Sarofim, A. F. (1997), Inorganic aerosols and their role in catalyzing sulfuric acid production in furnaces, Journal of the Air and Waste Management Association, 48, 106-112.
Gras, J. L., Keywood, M. D. and Ayers, G. P. (2001), Factors controlling winter-time aerosol light scattering in Launceston, Tasmania, Atmospheric Environment, 35, 1881-1889.
Green, D., Fuller, G. and Barratt, B. (2001), Evaluation of TEOMTM 'correction factors' for assessing the EU Stage 1 limit values for PM10, Atmospheric Environment, 35, 2589-2593.
Guthrie, A. C. M. and Lamb, A. N. (1976), Ground level concentrations from power station emissions, Clean Air (Aust), 10, 25-32.
Hall, P. (1983), Inverting an Edgeworth expansion, The Annals of Statistics, 11, 569-576.
Hanna, S. R., Briggs, G. A. and Hosker, R. P. J. (1982), Handbook on atmospheric diffusion, Technical Information Center, U.S. Dept. of Energy, Oak Ridge, Tenn.
Harris, R. J. (1985), A primer of multivariate statistics, Academic Press, Orlando.
166
Harrison, R. M. and Yin, J. X. (2000), Particulate matter in the atmosphere: which particle properties are important for its effects on health? Science of the Total Environment, 249, 85-101.
Heeley, P. (2001), The use of fabric filters on low sulphur coals, In 18th International Pittsburgh Coal Conference, Vol. cd-rom, (Ed, Conference, P. C.), Pittsburgh Coal Conference, Newcastle, NSW, Australia, pp. Paper # 49-2.
Helble, J. J. (2000), A model for the air emissions of trace metallic elements from coal combustors equipped with electrostatic precipitators, Fuel Processing Technology, 63, 125-147.
Helble, J. J., Mojtahedi, W., Lyyranen, J., Jokiniemi, J. and Kauppinen, E. (1996), Trace element partitioning during coal gasification, Fuel, 75, 931-939.
Held, T., Chang, D. P. Y. and Niemeier, D. A. (2003), UCD 2001: an improved model to simulate pollutant dispersion from roadways, Atmospheric Environment, 37, 5325-5336.
Helstroom, R., Carras, J. N., Prokopiuk, A. J., Chase, D. R., Sapounas, J., Buhre, B. and Nelson, P. F. (2002), Fine particle emissions from power stations, Cooperative Research Centre for Coal in Sustainable Development (CCSD), Pullenvale, QLD 4069, Australia, pp. 26.
Henry, R. C. (1991), Multivariate receptor models, In Receptor modelling for air quality management, (Ed, Hopke, P. K.), Elsevier; Distributors in the United States and Canada Elsevier Science Pub. Co., New York, NY, USA, pp. 117-147.
Henry, R. C. and Hidy, G. M. (1979), Multivariate analysis of particulate sulfate and other air quality variables by principal components. Part I. Annual data from Los Angeles and New York, Atmospheric Environment (1967-1989), 13, 1581-96.
Hewitt, C. N. (2001), The atmospheric chemistry of sulphur and nitrogen in power station plumes, Atmospheric Environment, 35, 1155-1170.
Hidy, G. M. (1994), Atmospheric sulfur and nitrogen oxides: eastern North American source receptor relationships, Academic Press, San Diego.
Hidy, G. M. (2002), Multiscale impact of fuel consumption on air quality, Energy & Fuels, 16, 270-281.
Hillamo, R. E. and Kauppinen, E. I. (1991), On the performance of the Berner low pressure impactor, Aerosol Science and Technology, 14, 33-47.
Hock, J. L. and Lichtman, D. (1983), A comparative study of in-plume and in-stack collected individual coal particles, Atmospheric Environment, 17, 849-852.
Huang, C. H. and Tsai, C. J. (2001), Effect of gravity on particle collection efficiency of inertial impactors, Journal of Aerosol Science, 32, 375-387.
Huang, X., Olmez, I. and Aras, N. K. (1994), Emissions of trace elements from motor vehicles: potential marker elements and source composition profile, Atmospheric Environment, 28, 1385-91.
Hurley, P. (2000), The air pollution model (TAPM): summary of some recent verification work in Australia, In 15th International Clean Air & Environment Conference - CASANZ, Vol. 1, CASANZ, Sydney, Australia, pp. 98-103.
Hurley, P., Manins, P., Lee, S., Boyle, R., Ng, Y. L. and Dewundege, P. (2003), Year-long, high-resolution, urban airshed modeling: verification of TAPM predictions of smog and particles in Melbourne, Australia, Atmospheric Environment, 37, 1899-1910.
Hurley, P. J. (1999), The air pollution model (TAPM) version 1: technical description and examples, CSIRO Atmospheric Research, Aspendale, Melbourne.
Hurley, P. J., Blockley, A. and Rayner, K. (2001), Verification of a prognostic meteorological and air pollution model for year-long predictions in the Kwinana
167
industrial region of Western Australia, Atmospheric Environment, 35, 1871-1880.
IPCC (1996), Climate change 1995: the science of climate change, Cambridge University Press, New York.
Jacko, R. B. and Neuendorf, D. W. (1977), Trace metal particulate emission test results from a number of industrial and municipal point sources, Journal of the Air Pollution Control Association, 27, 989-994.
Jakeman, A. J. and Simpson, R. W. (1987), Air quality and resource development: a risk assessment in the Hunter region in Australia, Centre for Resource and Environmental Studies, Canberra.
Jakeman, A. J., Simpson, R. W., Butt, J. A. and Daly, N. J. (1985), The effect of atmospheric pollutant emissions in the Upper Hunter region. Part 1: analysis and modelling of sulfur dioxide, Clean Air (Aust), 19, 9-16.
Jalkanen, L., Makinen, A., Hasanen, E. and Juhanoja, J. (2000), The effect of large anthropogenic particulate emissions on atmospheric aerosols, deposition and bioindicators in the eastern Gulf of Finland region, Science of the Total Environment, 262, 123-136.
Jang, M. and Kamens, R. M. (2001), Atmospheric secondary aerosol formation by heterogeneous reactions of aldehydes in the presence of a sulfuric acid aerosol catalyst, Environmental Science and Technology, 35, 4758-4766.
John, W. and Reischl, G. (1978), Measurements of the filtration efficiencies of selected filter types, Atmospheric Environment, 12, 2015-2019.
Johnson, D. E. (1998), Applied multivariate methods for data analysts, Duxbury Press, Pacific Grove, CA.
Jylha, K. (1995), Deposition around a coal-fired power station during a wintertime precipitation event, Water, Air, & Soil Pollution, 85, 2125-2130.
Kauppinen, E. I. and Pakkanen, T. A. (1990), Coal combustion aerosols: a field study, Environmental Science and Technology, 24, 1811-1818.
Keywood, M., Ayers, G., Gras, J. and Cohen, D. (2000), Size-resolved chemistry of Australian urban aerosol, In 15th International Clean Air and Environment Conference(Ed, Zealand, C. A. S. o. A. a. N.), Mitcham, Sydney, Australia, pp. 550-555.
Khosah, R. P. and McManus, T. J. (2001), Ambient fine particulate matter (PM25) sampling and analysis in the Upper Ohio River Valley (Pennsylvania, USA), Proceedings - Annual International Pittsburgh Coal Conference, 18th, 1043-1084.
Khosah, R. P., McManus, T. J., Agbede, R. O. and Feeley, T. J., III (2000), Ambient fine particulate matter (PM2.5) sampling and analysis in the upper Ohio River Valley, Prepr. Symp. - Am. Chem. Soc., Div. Fuel Chem., 45, 45-49.
Kiely, G. (1998), Environmental Engineering, McGraw Hill Europe. Kim, W.-s., Kim, S. H., Lee, D. W., Lee, S., Lim, C. S. and Ryu, J. H. (2001), Size
analysis of automobile soot particles using field-flow fractionation, Environmental Science and Technology, 35, 1005-1012.
Krewski, D., Burnett, R. T., Goldberg, M. S., Hoover, K., Siemiatycki, J., Jerrett, M., Abrahamowicz, M. and White, W. H. (2000), Reanalysis of the Harvard six cities study and the American Cancer Society study of particulate air pollution and mortality, 5/9/01, Health Effects Institute, http://www.healtheffects.org/Pubs/Rean-ExecSumm.pdf
168
Kuik, P. and Wolterbeek, H. T. (1995), Factor analysis of atmospheric trace-element deposition data in the Netherlands obtained by moss monitoring, Water, Air, & Soil Pollution, 84, 323-346.
Laden, F., Neas, L. M., Dockery, D. W. and Schwartz, J. (2000), Association of fine particulate matter from different sources with daily mortality in six US cities, Environmental Health Perspectives, 108, 941-947.
Lauf, R. J. (1985), Characterisation of the mineralogy and microchemistry of fly ash, Nuclear and Chemical Waste Management, 5, 231-236.
Lichtman, D. and Mroczkowski, S. (1985), Scanning electron microscopy and energy dispersive X-ray spectroscopy analysis of submicrometer coal fly ash particles, Environmental Science and Technology, 19, 274-277.
Lighty, J. S., Veranth, J. M. and Sarofim, A. F. (2000), Combustion aerosols: factors governing their size and composition and implications to human health, Journal of the Air and Waste Management Association, 50, 1565-1618.
Linton, R. W., Loh, A. and Natusch, D. F. S. (1976), Surface predominance of trace elements in airborne particles, Science, 191, 852-854.
Linton, R. W., Williams, P. and Evans, C. A. (1977), Determination of the surface predominance of toxic elements in airborne particles by ion microprobe mass spectrometry and auger electron spectrometry, Analytical Chemistry, 49, 1514-1521.
Liu, B. Y. B., Srinivasachar, S. and Helble, J. J. (2000), The effect of chemical composition on the fractal-like structure of combustion-generated inorganic aerosols, Aerosol Science & Technology, 33, 459-469.
Luhar, A. K. and Hurley, P. J. (2003), Evaluation of TAPM, a prognostic meteorological and air pollution model, using urban and rural point-source data, Atmospheric Environment, 37, 2795-2810.
Mackay, A. W. and Rose, N. L. (1998), Characterisation and temporal distribution of atmospheric particulates (PM10s) in central London, Advances in Air Pollution, 5, 947-957.
Malfroy, H., Davies, P. and Bell, D. (1993), Ambient dust depositions in the Hunter Valley, Pacific Power, Newcastle, pp. 29.
Mamane, Y. and Dzubay, T. G. (1986), Characteristics of individual particles at a rural site in the eastern United States, Journal of the Air Pollution Control Association, 36, 906-11.
Mamane, Y., Miller, J. L. and Dzubay, T. G. (1986), Characterisation of individual fly ash particles emitted from coal- and oil-fired power plants, Atmospheric Environment, 20, 2125-2135.
Manz, M., Weissflog, L., Kuhne, R. and Schuurmann, G. (1999), Ecotoxicological hazard and risk assessment of heavy metal contents in agricultural soils of central Germany, Ecotoxicology & Environmental Safety, 42, 191-201.
Marple, V. A. and Liu, B. Y. H. (1974), Characteristics of laminar jet impactors, Environmental Science and Technology, 8, 648-654.
Marple, V. A. and Willeke, K. (1976), Inertial impactors: theory, design and use, In Fine particles: aerosol generation, measurement, sampling and analysis, (Ed, Liu, B. Y. H.), Academic Press, New York, pp. 411-446.
Mastalerz, M., Glikson, M. and Simpson, R. W. (1998), Analysis of atmospheric particulate matter - application of optical and selected geochemical techniques, International Journal of Coal Geology, 37, 143-153.
McElroy, M. W., Carr, R. C., Ensor, D. S. and Markowski, R. (1982), Size distribution of fine particles from coal combustion, Science, 215, 13-19.
169
Mehra, A., Farago, M. E. and Banerjee, D. K. (1998), Impact of fly ash from coal-fired power stations in Delhi, with particular reference to metal contamination, Environmental Monitoring & Assessment, 50, 15-35.
Meij, R., Kooij, J. V. D., Sloot, H. A. V. D., Koppius-Odink, J. M. and Clement, L. J. (1985), Emissions and control of particulates of coal-fired power plants, In Second US-Dutch International Symposium on Aerosols(Eds, Lee, S. D., Schneider, T., Grant, L. D. and Verkerk, P. J.), Lewis Publishers, Chelsea, Michigan, Williamsburg, VA, USA, pp. 427-440.
Molnar, P., Janhall, S. and Hallquist, M. (2002), Roadside measurements of fine and ultrafine particles at a major road north of Gothenburg, Atmospheric Environment, 36, 4115-4123.
MSC (2003), State of the Environment Report 2002/2003, Muswellbrook Shire Council, Muswellbrook, pp. 88.
Muller, K., Spindler, G., Maenhaut, W., Hitzenberger, R., Wieprecht, W., Baltensperger, U. and ten Brink, H. (2004), INTERCOMP2000, a campaign to assess the comparability of methods in use in Europe for measuring aerosol composition, Atmospheric Environment, 38, 6459-6466.
Natusch, D. F. S., Wallace, J. R. and Evans, C. A. (1974), Toxic trace elements: preferential concentration in respirable particles, Science, 183, 202-204.
NEPC (1998), National Environment Protection Measure for ambient air quality, 7/9/2001, National Environment Protection Council, http://www.nepc.gov.au/
NEPC (2003), Variation to the national environment protection (ambient air quality) measure, 20/11/03, National Environment Protection Council (NEPC). http://www.ephc.gov.au/pdf/Air_Variation_PM25/PM2_5_Variation.pdf
NERDDC (1988), Project No 921: Air pollution from surface coal mining: measurement, modelling and community perception, National energy research development and demonstration council, Sydney, NSW, pp. 174.
Neville, M., McCarthy, J. F. and Sarofim, A. F. (1983), Size fractionation of submicrometer coal combustion aerosol for chemical analysis, Atmospheric Environment, 17, 2599-2604.
Noble, C. A. and Prather, K. A. (1996), Real-time measurement of correlated size and composition profiles of individual atmospheric aerosol particles, Environmental Science & Technology, 30, 2667-2680.
NPI (2002), National Pollutant Inventory, 12/06/2002, Environment Australia, http://www.npi.gov.au/index.html
O'Connor, B. H., Chang, W. J. and Martin, D. P. (1981), Chemical characterisation of atmospheric aerosol in Perth, Western Australia, In 7th International Clean Air Conference(Eds, Webb, K. A. and Smith, A. J.), Ann Arbor Science Publishers, Adelaide, Australia, pp. 639-652.
O'Keefe, C. A., Watne, T. M. and Hurley, J. P. (2000), Development of advanced scanning electron microscopy techniques for characterization of submicron ash, Powder Technology, 108, 95-102.
Padmanabhamurty, B. and Gupta, R. N. (1977), Particulate pollution in Delhi due to Indraprastha power plant, Indian Journal of Meteorology Hydrology and Geophysics, 28, 375-384.
170
Patel, C. B. and Pandey, G. S. (1986), Alkalinization of soil through thermal power plant fly ash fallout, The Science of the Total Environment, 57, 67-72.
PCD (1995), Ambient Air Standards of Thailand (1995), 12/3/04, Pollution Control Department (PCD), Ministry of Natural Resources and Environment, http://www.pcd.go.th/Information/Regulations/air_noise/Airquality.htm#AirQuality
Pershing, D. W. and Wendt, J. O. L. (1979), Relative contributions of volatile nitrogen and char nitrogen to NOx emissions from pulverized coal flames, Industrial and Engineering Chemistry Process Design and Development, 18, 60.
Physick, W. (2002), Stack characteristics and Liddell stack co-ordinates, (Ed, Hinkley, J.), pp. series of emails.
Physick, W. L., Noonan, J. A. and Manins, P. C. (1991), Air quality modelling study of the Hunter Valley. Phase 1: emitters in the Upper Hunter, CSIRO Division of Atmospheric Research, Aspendale (Melbourne), Victoria, pp. 102.
Pio, C. A., Castro, L. M., Cerqueira, M. A., Santos, I. M., Belchior, F. and Salgueiro, M. L. (1996), Source assessment of particulate air pollutants measured at the southwest European coast, Atmospheric Environment, 30, 3309-3320.
Posfai, M., Anderson, J. R., Buseck, P. R., Shattuck, T. W. and Tindale, N. W. (1994), Constituents of a remote Pacific marine aerosol: a TEM study, Atmospheric Environment, 28, 1747-56.
Posfai, M., Xu, H., Anderson, J. R. and Buseck, P. R. (1998), Wet and dry sizes of atmospheric aerosol particles: an AFM-TEM study, Geophysical Research Letters, 25, 1907-1910.
Quann, R. J., Neville, M. and Sarofim, A. F. (1990), A laboratory study of the effect of coal selection on the amount and composition of combustion generated submicron particles, Combustion Science and Technology, 74, 245-265.
Querol, X., Alastuey, A., Lopez-Soler, A. and Plana, F. (2000), Levels and chemistry of atmospheric particulates induced by a spill of heavy metal mining wastes in the Donana area, Southwest Spain, Atmospheric Environment, 34, 239-253.
Querol, X., Alastuey, A., Lopez-Soler, A., Plana, F., Mantilla, E., Juan, R., Ruiz, C. R. and La Orden, A. (1999), Characterisation of atmospheric particulates around a coal-fired power station, International Journal of Coal Geology, 40, 175-188.
Querol, X., Fernandezturiel, J. L. and Lopezsoler, A. (1995), Trace elements in coal and their behaviour during combustion in a large power station, Fuel, 74, 331-343.
Raask, E. (1985), Mineral impurities in coal combustion, Hemisphere Publishing, Washington.
Ramsden, A. R. and Shibaoka, M. (1982), Characterisation and analysis of individual fly-ash particles from coal fired power stations by a combination of optical microscopy, electron microscopy and quantitative electron microprobe analysis, Atmospheric Environment, 16, 2191-2206.
Razmovski, V., O'Meara, T., Hjelmroos, M., Marks, G. and Tovey, E. (1998), Adhesive tapes as capturing surfaces in Burkard sampling, Grana, 37, 305-310.
Reis, M. A., Alves, L. C., Wolterbeek, H. T., Verburg, T., Freitas, M. C. and Gouveia, A. (1996), Main atmospheric heavy metal sources in Portugal by biomonitor analysis, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials & Atoms, 109, 493-497.
Rogers, G. F. C. and Mayhew, Y. R. (1980), Thermodynamic and transport properties of fluids (SI units), Basil Blackwell, Worcester.
Rothe, M. (2003), Personal communication, (Ed, Hinkley, J.).
171
Rubow, K. L., Marple, V. A., Olin, J. and McCawley, M. A. (1987), A personal cascade impactor: design, evaluation and calibration, American Industrial Hygiene Association journal, 48, 532-8.
Sawidis, T., Chettri, M. K., Papaioannou, A., Zachariadis, G. and Stratis, J. (2001), A study of metal distribution from lignite fuels using trees as biological monitors, Ecotoxicology & Environmental Safety, 48, 27-35.
Schwartz, J., Dockery, D. W. and Neas, L. M. (1996), Is daily mortality associated specifically with fine particles?, Journal of the Air & Waste Management Association, 46, 927-939.
Seames, W. S. (2003), An initial study of the fine fragmentation fly ash particle mode generated during pulverised coal combustion, Fuel Processing Technology, 81, 109-125.
Sivacoumar, R., Bhanarkar, A. D., Goyal, S. K., Gadkari, S. K. and Aggarwal, A. L. (2000), Air pollution modeling for an industrial complex and model performance evaluation, Environmental Pollution (Oxford, United Kingdom), 111, 471-477.
Sloss, L. L. (1998), Sampling and analysis of PM10/PM2.5, IEA Coal Research, London, UK.
Smith, I. M. and Sloss, L. L. (1998), PM10/PM2.5 - emissions and effects, IEA Coal Research, London, UK.
Smith, R. D., Campbell, J. A. and Nielson, K. K. (1979), Characterization and formation of submicron particles in coal-fired plants, Atmospheric Environment, 13, 607-617.
Smoot, L. D. (1991), Coal and char combustion, In Fossil fuel combustion: a source book, (Eds: Bartok, W. and Sarofim, A. F.), Wiley, New York, pp. 653-781.
Song, X. H., Polissar, A. V. and Hopke, P. K. (2001), Sources of fine particle composition in the northeastern US, Atmospheric Environment, 35, 5277-5286.
Stern, A. C. (1986), Air pollution vol VI: supplement to air pollutants, their transformation, transport and effects, Academic Press, Orlando, Florida.
Suarez, A. E. and Ondov, J. M. (2002), Ambient aerosol concentrations of elements resolved by size and source: contributions of some cytokine-active metals from coal- and oil-fired power plants, Energy & Fuels, 16, 562-568.
Swaine, D. J. (1984), Environmental consequences of coal combustion, CSIRO, Institute of Energy and Earth Resources, Division of Fossil Fuels.
Swaine, D. J. (1994), Trace elements in coal and their dispersal during combustion, Fuel Processing Technology, 39, 121-137.
Swift, J. A. (1970), Electron microscopes, Kogan Page, London. Tataruch, F. (1995), Red deer antlers as biomonitors for lead contamination, Bulletin of
Environmental Contamination & Toxicology, 55, 332-337. ten Brink, H. M., Janssen, A. J. and Slanina, J. (1988), Plume wash-out near a coal fired
power plant: measurements and model calculations, Atmospheric Environment, 22, 177-187.
Thurston, G. D. (1981), Multivariate analysis of particulate sulfate and other air quality variables by principal components. Part I. Annual data from Los Angeles and New York. Comments, Atmospheric Environment (1967-1989), 15, 424-5.
Thurston, G. D. and Spengler, J. D. (1985), A quantitative assessment of source contributions to inhalable particulate matter pollution in metropolitan Boston, Atmospheric Environment (1967-1989), 19, 9-25.
TSI (2001), Model 3089 nanometer aerosol sampler - instruction manual, 19/9/01, TSI, http://www.tsi.com/particle/product/pdf/3089.pdf
Tuyl, F. (2003), Particle size data, (Ed, Hinkley, J.), Newcastle, Australia, pp. spreadsheet with modified Edgeworth expansion.
USEPA (1995), Compilation of air pollutant emission factors. AP-42, fifth edition, volume I chapter 1: external combustion sources, 4/9/01, USEPA, http://www.epa.gov/ttn/chief/ap42/ch01/final/c01s01.pdf
USEPA (2003a), 40 CFR Part 51 Revision to the guideline on air quality models: adoption of a preferred long range transport model and other revisions; final rule, 9/06/04, http://www.epa.gov/scram001/tt25.htm#guidance
USEPA (2003b), Review of the National Ambient Air Quality Standards for particulate matter: policy assessment of scientific and technical tnformation. OAQPS staff paper - first draft. EPA-452/D-03-001, 24/01/05, Environmental Protection Agency, http://www.epa.gov/ttn/naaqs/standards/pm/data/pm_staff_paper_august2003_1stdraft.pdf
Vann Bush, P., Snyder, T. R. and Chang, R. L. (1989), Determination of baghouse performance from coal and ash properties: Part II, Journal of the Air and Waste Management Association, 39, 361-372.
Virtanen, A. K. K., Ristimaeki, J. M., Vaaraslahti, K. M. and Keskinen, J. (2004), Effect of Engine Load on Diesel Soot Particles, Environmental Science and Technology, 38, 2551-2556.
Wall, T. F., Wibberley, L. J. and Phong-anant, D. (1982), Environmental aspects of coal combustion: fly ash, NOx and SOx, In Coal, properties, analysis and effective use, (Ed, Wall, T. F.), The Institute of Coal Research, University of Newcastle, Newcastle, Australia, pp. 9-1 to 9-29.
Wangen, L. E. and Williams, M. D. (1980), Elemental deposition downwind of a coal-fired power plant, Water, Air and Soil Pollution, 10, 33-44.
Watson, J. G. and Chow, J. C. (1994), Clear sky visibility as a challenge for society, Annual Review of Energy and the Environment, 19, 241-266.
Weast, R. C., Astle, M. J. and Beyer, W. H. (Eds.) (1986), CRC handbook of chemistry and physics, CRC Press.
Weckwerth, G. (2001), Verification of traffic emitted aerosol components in the ambient air of Cologne (Germany), Atmospheric Environment, 35, 5525-5536.
Wentzel, M., Gorzawski, H., Naumann, K. H., Saathoff, H. and Weinbruch, S. (2003), Transmission electron microscopical and aerosol dynamical characterization of soot aerosols, Journal of Aerosol Science, 34, 1347-1370.
WHO (1999), Guidelines for air quality, 19/7/2001, World Health Organisation, Geneva, http://www.who.int/peh/air/Airqualitygd.htm
WHO (2003), Health aspects of air pollution with particulate matter, ozone and nitrogen dioxide. Report on a WHO Working Group. Bonn, Germany 13–15 January 2003. EUR/03/5042688, 24/1/2005, Health Documentation Services, WHO Regional Office for Europe, http://www.euro.who.int/document/e79097.pdf
Wibberley, L. J. and Wall, T. F. (1986), An investigation of factors affecting the physical characteristics of fly ash formed in a laboratory scale combustor, Combustion Science and Technology, 48, 177-190.
Wichmann, H., Spix, C., Tuch, T., Wolke, G., Peters, A., Heinrich, J., Kreyling, W. G. and Heyder, J. (2000), Daily mortality and fine and ultrafine particles in Erfurt, Germany. Part I: Role of particle number and particle mass, 5/9/01, Health Effects Institute, http://www.healtheffects.org/Pubs/Wichmann.pdf
173
Williams, D. J., Carras, J. N., Milne, J. W. and Heggies, A. C. (1981), The oxidation and long-range transport of sulfur dioxide in a remote region., Atmospheric Environment, 15, 2255-62.
Wilson, W. E. and Suh, H. H. (1997), Fine particles and coarse particles - concentration relationships relevant to epidemiologic studies, Journal of the Air & Waste Management Association, 47, 1238-1249.
Wolf, M. E. and Hidy, G. M. (1997), Aerosols and climate - anthropogenic emissions and trends for 50 years, Journal of Geophysical Research-Atmospheres, 102, 11113-11121.
Ziomas, I. C., Tzoumaka, P., Balis, D., Melas, D., Zerefos, C. S. and Klemm, O. (1998), Ozone episodes in Athens, Greece. A modeling approach using data from the MEDCAPHOT-TRACE, Atmospheric Environment, 32, 2313-2321.
Zou, L. (2000), Elemental composition and morphology of size-resolved brown coal fly ash particles, In 15th International Clean Air & Environment Conference - CASANZ, Vol. 1, CASANZ, Sydney, Australia, pp. 428-431.
174
Appendix A: Calibration of Burkard Flow Tube
It was not possible to directly calibrate the Burkard flow tube due to the absence of
sufficient differentiation in the markings on the side of the tube. Instead, field readings
were made based on the distance from the top of the float to the centreline (the nominal
10 LPM marking) and this distance was calibrated against a 10 LPM rotameter, which
was in turn calibrated against a bubble tube. The rotameter calibration is shown in
Figure A-1(a) and the flow tube calibration in Figure A-1(b). The calibration indicates
a good linear response, although there is a slight suggestion of a sinusoidal “wobble”
about the line of best fit. The uncertainties associated with this determination are not
significant, as the flow readings were in general only used to ensure that the sampler
was operating properly and to estimate approximate flows for the capture efficiency
• Scenarios: 12 monthly runs, from July 2002 through to June 2003 (as month
plus last day of preceding month to “spin-up”).
Sample Log File
Only the initial part of the file is shown here as the full .lis file would run to 2385 pages
(it is a 3.83MB text file). |----------------------------------------| | THE AIR POLLUTION MODEL (TAPM V2.0.1). | | Copyright (C) CSIRO Australia. | | All Rights Reserved. | |----------------------------------------| ---------------- RUN INFORMATION: ---------------- NUMBER OF GRIDS= 3 GRID CENTRE (longitude,latitude)=( 150.966705 , -32.3916702 ) GRID CENTRE (cx,cy)=( 0 , 0 ) (m) GRID DIMENSIONS (nx,ny,nz)=( 40 , 40 , 25 ) NUMBER OF VERTICAL LEVELS OUTPUT = 15 DATES (START,END)=( 20020831 , 20020930 ) DATE FROM WHICH OUTPUT BEGINS = 20020901 LOCAL HOUR IS GMT+ 10.1000004 SYNOPTIC WIND SPEED MAXIMUM = 30 (m/s) SYNOPTIC PRESSURE-GRADIENT SCALING FACTOR = 1.00000000 SYNOPTIC PRESSURE-GRADIENT FILTERING FACTOR = 1.00000000 VARY SYNOPTIC WITH 3-D SPACE AND TIME INCLUDE VEGETATION EXCLUDE NON-HYDROSTATIC EFFECTS EXCLUDE RAIN INCLUDE PROGNOSTIC EDDY DISSIPATION RATE EQUATION POLLUTION : 4 TRACERS (TR1,TR2,TR3,TR4) EXCLUDE POLLUTANT CROSS-CORRELATION EQUATION POLLUTANT GRID DIMENSIONS (nxf,nyf)=( 79 , 79 ) TR1 BACKGROUND = 0.00000000E+00 (ug/m3) TR2 BACKGROUND = 0.00000000E+00 (ug/m3)
183
TR3 BACKGROUND = 0.00000000E+00 (ug/m3) TR4 BACKGROUND = 0.500000000 (ug/m3) TR1 DECAY RATE = 0.00000000E+00 (per second) TR2 DECAY RATE = 0.00000000E+00 (per second) TR3 DECAY RATE = 0.00000000E+00 (per second) TR4 DECAY RATE = 0.00000000E+00 (per second) TR1 EMISSION TEMPERATURE VARIATION:NONE TR2 EMISSION TEMPERATURE VARIATION:NONE TR3 EMISSION TEMPERATURE VARIATION:NONE TR4 EMISSION TEMPERATURE VARIATION:NONE --------------------------------- START GRID 1 C:\tapm\run\r100k\r100k GRID SPACING (delx,dely)=( 10000 , 10000 ) (m) POLLUTANT GRID SPACING (delxf,delyf)=( 5000 , 5000 ) (m) NO MET. DATA ASSIMILATION FILE AVAILABLE NO BUILDING FILE AVAILABLE NUMBER OF PSE SOURCES= 4 NO LSE EMISSION FILE AVAILABLE NO ASE EMISSION FILE AVAILABLE NO GSE EMISSION FILE AVAILABLE NO BSE EMISSION FILE AVAILABLE NO WHE EMISSION FILE AVAILABLE NO VPX EMISSION FILE AVAILABLE NO VDX EMISSION FILE AVAILABLE NO VLX EMISSION FILE AVAILABLE NO VPV EMISSION FILE AVAILABLE INITIALISE LARGE TIMESTEP = 300.000000 METEOROLOGICAL ADVECTION TIMESTEP = 300.000000 (s) Deep Soil Moisture Content (kg/kg)= 0.150000006 Deep Soil & Sea Temperatures (K) = 289.600006 289.600006 POLLUTION ADVECTION TIMESTEP = 300.000000 (s) PSE KEY : is = Source Number ls = Source Switch (-1=Off,0=EGM,1=EGM+LPM) xs,ys = Source Position (m) hs = Source Height (m) rs = Source Radius (m) es = Buoyancy Enhancement Factor fs_no = Fraction of NOX Emitted as NO fs_fpm= Fraction of APM Emitted as FPM INIT_PSE is, ls, xs, ys, hs, rs, es, fs_no, fs_fpm 1, 1, -1532., -522., 250.00, 5.28, 1.30, 1.00, 0.50, 2, 1, -1796., -404., 250.00, 5.28, 1.30, 1.00, 0.50, 3, 1, 956., 1990., 168.00, 4.35, 1.40, 1.00, 0.50, 4, 1, 927., 2144., 168.00, 4.35, 1.40, 1.00, 0.50, LAGRANGIAN (LPM) MODE IS OFF FOR THIS GRID
184
Appendix F: Cascade Impactor – Associated Errors RUN SAMPLE F(197) Na(440) Mg(585) AL SI P S CL K CA TI V CR MN FE CO NI CU ZN BR SE SR PBL NAME error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %error %
Salt Na .123 .250 .706 Al .905 -.086 .027 Si .721 .261 -.161 S -.042 .945 .133 Cl .076 .799 .324 K .905 -.078 .288 Ca .861 .269 .123 Ti .935 -.176 .225 Fe .886 -.101 .325 Mn .189 .033 .833 Ni -.034 .803 .081 Zn .079 .407 .581
187
Comments: the inclusion of Mn, Ni and Zn has resulted in the separation of the Salt and
CFPS components.
Case 3: Reduced number of variables (16): Na, Al, S i, S, Cl, K, Ca, Ti, Fe, Mn, Ni, Zn, Cr, Co, Cu and Br.
Total Variance Explained
Cpt Initial Eigenvalues Extraction Sums of Squared