Aircraft Engine Speciated Organic Gases:
Speciation of Unburned Organic Gases
in Aircraft Exhaust
          United States
          Environmental Protection
          Agency

-------
Aircraft Engine Speciated Organic Gases:
  Speciation of Unburned Organic Gases
              in Aircraft Exhaust
                Assessment and Standards Division
               Office of Transportation and Air Quality
               U.S. Environmental Protection Agency

                          and

                  AEE-300 - Emissions Division
                Office of Environment and Energy
                 Federal Aviation Administration
                                          EPA-420-R-09-902
                                          May 2009

-------
 1                     Aircraft Engine Speciated Organic Gases:
 2           Speciation of Unburned Organic Gases in Aircraft Exhaust
 3
 4                     W.B. Knighton, S.C. Herndon, and R.C. Miake-Lye
 5
 6
 7    Purpose and Scope: The F AA and EPA are evaluating the methodology to quantify
 8    Hazardous Air Pollutants (HAPs) emissions from commercial aircraft engines, to be used
 9    when an aircraft HAPs emissions inventory is requested. Central to the methodology is a
10    singular HAPs speciation profile. The final HAPs speciation profile will be:
11
12           -   Nationally consistent,
13           -   Supported by state-of-the-science data,
14           -   Representative of today's commercial aircraft fleet, and
15           -   "Living" to continue to reflect the state-of-the-science as studies are
16              conducted and new data becomes available.
17
18    A second outcome of this effort is to evaluate and, if necessary, update the factors needed
19    to convert between total unburned hydrocarbons (HC), volatile organic compounds
20    (VOC), and total organic gases (TOG).
21
22    The scope of this work is to update the current HAPs profile that exists for commercial
23    aircraft engines, using recent HAPs measurements conducted on more modern
24    commercial aircraft engines.  The original HAPs profile has been in existence unchanged
25    for over 2 decades, based upon a single 1984 measurement campaign by Spicer et al.1 To
26    support the update of the existing HAPs profile, consolidated data from Spicer and more
27    recent measurements (EXCAVATE, APEX) will be investigated and discussed in this
28    document. Important questions to address in this scope of work are:  how to combine all
29    of the data sets into a single profile given the various methods used to collect the
30    samples; and how to address combustor technologies, etc. not yet tested.  We still  have
31    very limited  data to work with at this time, which limits the conclusions we can make, so
32    it is necessary to be mindful of these questions as new HAPs data becomes available in
33    the future and we endeavor to  update this methodology.
34
35    Introduction: Aircraft gas turbine engines are designed to burn their hydrocarbon (HC)
36    fuel efficiently, since any inefficiency translates into carrying more fuel, a greater take-
37    off weight, and a steeply rising cost of operation as efficiency decreases.  Because most
38    of the fuel is consumed at higher power settings and most of the operational time is spent
39    at cruise,  for power settings of cruise and above most engines convert significantly more
      1 HAPs profile No. 1098 in EPA'S SPECIATE database, http://www.epa.gov/ttn/chief/software/speciate/index.html
      Composite profile developed from data for a CFM-56 jet engine fired with JP-5 fuel at idle, 30% thrust and
      80% thrust. Data collected by GC/MS and DNPH analyses were combined according to average LTO
      cycle times obtained from AP-42. Spicer, C. W., et al., Battelle Columbus Laboratories, Composition and
      Photochemical Reactivity of Turbine Engine Exhaust, Report No. ESL-TR-84-28, Prepared for Air Force
      Engineering and Services Center (RDVS), Tyndall AFB, FL, September 1984.

-------
                                                   ..  '

 1    than 99% of the fuel through complete combustion to carbon dioxide (CO2) and water
 2    (H2O). At idle conditions, much less fuel is consumed and, in the interest of maintaining
 3    stable combustion at lower power conditions, some sacrifice in combustion efficiency
 4    occurs even though this inefficiency is still only a percent or so. Any combustion
 5    inefficiency of HC fuel will result in emissions of some combination of CO and
 6    incompletely oxidized HCs, as well as some carbonaceous particles.
 7
 8    From the point of view of understanding the combustion process, knowing the
 9    combustion efficiency is important since any HC emission represents an inefficiency in
10    converting fuel to CC>2 and H2O. In order to understand the environmental impact of the
11    emissions, it is important to quantify the amounts of the emitted species, especially those
12    that are deemed highly toxic or carcinogenic. The US EPA considers a number of HCs
13    (among other pollutants) as HAPs, and quantification of levels of these species takes on a
14    special importance.  This report will discuss the emissions of HAPs from aircraft engines
15    and how the speciation of the HC emissions relates to levels of the various HAPs present
16    in aircraft exhaust.2 Since the concentrations of HCs and HAPs are highest in the exhaust
17    at low power conditions, the emphasis will be on measurements under such conditions.
18
19    Components of PM emissions from aviation engines may also be classified as Hazardous
20    Air Pollutants, but PM emissions are measured and analyzed very differently than
21    gaseous emissions, and are not discussed in this document.  Much work is currently being
22    directed at identifying measurement approaches and resolving sampling issues for
23    aviation gas turbine engine PM emissions (e.g. the APEX and related campaigns), and
24    data characterizing PM emissions from a variety of commercial  aircraft gas turbine
25    engines is being accumulated.
26
27    Background: Several studies have attempted to document the speciation of the HCs
28    emitted from aircraft engines. Most notably, in the 1980s Spicer et al. (Spicer,  Holdren et
29    al.  1994) performed a series of studies on a set of military engines using a variety of
30    analytical techniques to quantify a wide range of HCs. Subsequently Gerstle et al.
31    (Gerstle, Virag et al. 1999) examined  another set of military aircraft, with a similar set of
32    analytical techniques.  Most recently,  a set of studies initiated by NASA called  Aircraft
33    Particle Emissions experiment (APEX), and supported by a wide range of sponsors
34    (NASA, FAA, CARB, EPA, DoD ...) has focused attention on commercial aircraft
35    Particulate Matter (PM) emissions (Wey 2004; Onasch, Jayne et al. 2006; Wey,
36    Anderson et al. 2006; Lobo, Hagen et al. 2007), using a wide range of analytical
37    techniques. These studies also included HC gaseous emissions analysis. Some of the
38    techniques employed in APEX1-3 overlap with the earlier tests, but also some more
39    advanced (faster time response/higher sensitivity) techniques were used during APEX.3
      2 It should be noted that because a compound is considered hazardous it does not imply health or welfare
      effects at current levels, or that it is appropriate to adopt controls to limit the emissions of such a compound
      from turbine engine aircraft or their fuels.
      3 APEX was the collaborative research effort of NASA, EPA, DoD, and the FAA. The main objective of
      the APEX research was to characterize both gaseous and paniculate emissions to advance the
      understanding of emissions from commercial aircraft engines. APEX1 was conducted in April of 2004 with
      a NASA-owned DC-8 aircraft equipped with CFM-56-2C1 engines.  APEX2 was conducted in August
      2005 for typical in-use aircraft engines (CFM56 engines on B737 aircraft), APEX3 testing was conducted
      05/27/09

-------
                                            ..

 1    The range of experiments and the variety of techniques employed can be used to provide
 2    greater confidence in the HAPs emissions measurement data, and to allow assessment of
 3    which results can be verified by multiple techniques.  In addition, one engine type, a
 4    CFM56 (a high bypass turbofan engine), was part of both the Spicer and APEX studies,
 5    so that a most direct cross comparison can be made.4
 6
 7    The comprehensive measurements of Spicer et al. (Spicer, Holdren et al. 1994) have
 8    provided valuable data for the CFM56-3 and the TF-39 (forerunner to the General
 9    Electric CF6 class of high bypass turbofan engines).5  These measurements were
10    conducted using a mixture of on-line instrumentation and canister sampling with off-line
11    analysis developed from a prolific program of military engine emissions characterization.
12    This work chose to report values as ppmC present in the exhaust. The data labeled 'idle'
13    in this work was conducted at nominal 'ground idle' and does not reflect the ICAO
14    definition of idle, also called 7% of rated thrust. The Spicer et al. work finds that 40% of
15    the organic gas mass is accounted for by the compounds, ethene, formaldehyde, propene,
16    ethyne and methane.
17
18    In a report to the US Air Force, Gerstle and co-workers (Gerstle, Virag et al.  1999)
19    reported HC emission rates for several engines not included in the ICAO databank, as
20    well as some emissions from auxiliary power units. Some of the military engines
21    addressed in this study represent older engine technologies that are  no longer represented
22    in the commercial fleet and, as such,  there may be issues regarding  combustion
23    efficiencies at low power conditions that may cause significant differences in emissions
24    due to raw  fuel contributions to the HCs emissions at low power (personal
25    communication Will Dodds, GE, and KBE, February 2007 et seq.).
26
27    A more recent series of measurements have focused on commercial engines.  NASA's
28    interest in charactering the emissions from commercial engines in dedicated engine tests
29    was demonstrated during the EXCAVATE campaign. Anderson et al.  (Anderson, Chen
30    et al. 2006) measured the speciated organic gas emissions from a Rolls-Royce RB211-
31    535-E4 engine (another high bypass turbofan engine) for two different fuel sulfur levels.
32    A very comprehensive program continued with the APEX-1 campaign (Wey, Anderson
33    et al. 2006) within which HAPs characterization was  conducted with high time response
34    on-line organic gas speciation using infrared fingerprint absorption spectroscopy and
35    chemical ionization mass spectrometry for a CFM56-2C1 (Knighton, Rogers et al. 2007;
      in October and November of 2005 spanning a range of engines from a small business jet, through a modern
      regional turbofan, a single-aisle transport turbofan, to a large high bypass ratio turbo fan, representing five
      different engine types, some measuring more than one example. In all studies, exhaust plumes were
      sampled at the engine exit plane and several downstream measurement locations.
      4 CFM56 and the CFM logo are for CFM International, which is a joint company of Snecma and General
      Electric. Snecma is a French manufacturer of engines for commercial and military aircraft, and space
      vehicles.
      5 The General Electric TF-39 was the first high bypass turbofan engine, and it was developed for the Air
      Force back in 1965 for a new transport aircraft. Turbofan engines with a bypass ratio of 5 or greater are
      considered to be high bypass turbofan engines (Cumpstv. N., Jet Propulsion, Cambridge University Press,
      2002, p. 46.). Bypass ratio is the ratio between the mass flow rate of the air drawn in by the fan, but
      bypassing the engine core, to the mass flow rate passing through the engine core (Cumpstv. loc. cit).
      05/27/09

-------
                                                  .. '

 1    Yelvington, Herndon et al. 2007). Time integrated LTO cycle data were also collected at
 2    the same time (Kinsey et al,  document in preparation). An analysis of the JETS/APEX-2
 3    (Lobo et al., 2007)6 and APEX-3 datasets is forthcoming (Timko et al., in preparation).
 4
 5    In all of the APEX dedicated engine tests, measurements were made at both the engine
 6    exit plane and in the plume at a downstream location (nominally 30 m for an intermediate
 7    engine size such as a CFM56).  It is important to note that the measured HC profile is
 8    relatively consistent regardless of measurement location. In the ensembles of data
 9    presented below, all  of the various distances, fuels, and power conditions below 30% of
10    rated thrust are combined in demonstrating the tight correlations among HC emissions.
11    Further, in the airport studies discussed next, much further downwind measurements also
12    indicated no change  in the relative concentrations of species, although as the exhaust
13    continues to dilute, the species present as very small fractions of the total profile begin to
14    fall below detection  limits as distances increase further from the emission source.
15
16    In addition to dedicated engine tests,  sampling  from airports during routing operation
17    have also provided useful data for HAPs emissions. Using analysis of wind-advected
18    plumes sampled at Boston Logan International Airport, selected speciated organic gas
19    emissions were characterized from in-use aircraft (Herndon, Rogers et al. 2006).
20    Schurmann et al. (Schurmann, Schafer et al. 2007) also measured volatile organic
21    compounds using canister sampling of diluted exhaust in an operational taxiway area.
22    They found that refueling activity altered the profile of hydrocarbons considerably.  An
23    analysis of the wind  advected data collected at the Oakland GRE and taxiway/runway
24    sampling is forthcoming (Herndon et al., in preparation).
25
26    All of the studies indicate that a wide range of combustion-related emissions are present
27    in aircraft exhaust. Despite the long list of species present, a ranking of the species by
28    concentration indicates that  15-20 species represent most (95% or more) of the emissions
29    on the basis of concentration. A greater number of species are present at a fraction  of a
30    percent or smaller of the total concentration.  Of the overall speciated mixture, a number
31    of species can be considered HAPs, while another set may be significant to the overall
32    level of VOC emissions but  data indicating toxicity are lacking.
33
34    An important point to note is that no instrument measures  all of the HC emissions.  The
35    fast time response instrument (Proton Transfer Reaction Mass Spectrometer: PTR-MS)
36    used for HC measurements in APEX was focused on measuring relevant HAPs, and as
37    such was not capable of measuring alkanes or acetylene. Since the PTR-MS is capable of
38    measuring a wide range of HCs other than alkanes and acetylene, the measurement  focus
39    was on a list of species that were measurable by the PTR-MS, identified EPA HAPs
40    species, and present  in aircraft exhaust. Formaldehyde and ethylene were also not
41    measurable with the PTR-MS, but were measured separately in APEX using IR
42    techniques (Tunable Infrared Laser Absorption Spectroscopy:  TILDAS). In the Spicer
      6 Additional data reported from JETS/APEX2, taken by the UC Riverside team, was not used to develop
      the jet aircraft speciation profile, because the compromised sampling system for that data source prevented
      a complete and high-confidence organic compound data set from being assembled from the UC Riverside
      data.
      05/27/09

-------
                                                  ..  '

 1    studies, a wide range of techniques was used, but no measurement of methanol was
 2    attempted, and none of the trimethylbenzenes nor several of C9-C11 aromatic species
 3    were identified with the techniques employed therein.  In many of the studies, a Flame
 4    lonization Detector (FID) was used to quantify the total "unburned hydrocarbons"
 5    (UHCs), but this is an imperfect estimation of the total emissions due to the FID's non-
 6    uniform response to different carbon-containing compounds.  All this is to note that,
 7    while these several data sets provide very useful data on many individual compounds and
 8    their relationship to one another, arriving at an estimate of a total quantity by mass or by
 9    concentration is dependent on which species are included in the total. And, the measured
10    species are determined by what measurement techniques have been employed.
11
12    Data Comparison: The most direct intercomparison between the earlier studies and the
13    recent APEX mission is accomplished through the overlap with the CFM56 engine.
14    Table 1 reproduces Spicer's speciation data for this engine (Spicer, Holdren et al. 1994)
15    ranked in order of decreasing concentration. The first column lists the species present in
16    the highest concentration, which represent about 95% of the total speciated non methane
17    hydrocarbon (NMHC) emissions on a concentration basis  as measured by Spicer. The
18    highlighted species indicate those species measured by Spicer that were also quantified in
19    APEX by PTR-MS (yellow) or TILDAS (green). In the first  column, only acetylene and
20    ethane are not highlighted.  In subsequent columns, the sum of which represents 5% of
21    the Spicer emissions concentration, a number of other alkanes also are not highlighted.
22    These species, not measured by PTR-MS or TILDAS, represent about 1.4% of Spicer's
23    total, and are not typically considered HAPs. It is worth noting that, of the species noted
24    in the "EPA 14" and "FAA 10" HAPs lists  that were developed based on relevant HC
25    emissions from aviation engines (URS and  FAA 2003), all of those species are in the
26    highlighted (measured in both studies)  elements of Table 1.
27
28
      05/27/09

-------
                           Technical Support Document
1 Table 1. NMHC emission ratios for the CFM56-3 engine reported by Spicer et al. listed
2 in decreasing magnitude. The first column represents 95% of the emissions on a molar
3 basis. Green highlighted cells indicate compounds that are measured by TILDAS. Yellow
4 highlighted cells indicate compounds that are quantified by the PTR-MS.
5
Compound
ethylene
brmaldehyde
acetylene
propene
acetaldehyde
acrolein
1-butene
glyoxal
[, 3 -butadiene
jenzene
methylglyoxal
ethane
)utanal/crotona
dehyde
)ropanal
l-pentene
l-hexene
toluene
ER
(mmole/mole)
0.77
0.572
0.211
0.151
0.135
0.061
0.044
0.044
0.044
0.03
0.029
0.024
0.019
0.017
0.015
0.012
0.0097
Compound
acetone
C5-ene
2-methylpentane
benzaldehyde
1-heptene
naphthalene
C5-ene
cis-2-butene
styrene
n-undecane
n-pentane
n-dodecane
m,p-xylene
2-methyl-2-butene
1-octene
n-decane
phenol
ER
(mmole/mole)
0.0089
0.0072
0.0066
0.0062
0.0061
0.0059
0.0055
0.0052
0.0041
0.0040
0.0038
0.0038
0.0037
0.0037
0.0034
0.0031
0.0029
Compound
l-nonene
)ropane
[ -CH3-naphthalene
lexanal
C5-cyclohexane
ethylbenzene
C4-benzene
o-xylene
2-CH3-naphthalene
C5 -benzene
l-decene
C13-alkane
C14-alkane
n-heptane
n-octane
n-nonane
C12-C18alkanes
ER
(mmole/mole)
0.0027
0.0025
0.0024
0.0023
0.0023
0.0023
0.0023
0.0022
0.0020
0.0020
0.0018
0.0014
0.0013
0.0009
0.0008
0.0007
0.0045
 9
10
11
12
13
14
15
16
17
The highlighted sections in Table 1 indicate that comparisons can be made for the
measurements of those species measured for Spicer's CFM56 and the several CFM56
engines measured in APEX. Those comparisons are listed in Table 2 as mass ratios,
expressed as ratios of Emission Indices (Els). The El of a species is the mass of that
species emitted in grams, divided by the mass of fuel consumed in kilograms (species
g/kg fuel). The unhighlighted elements in Table 1 indicate that the APEX PTR-
MS/TILDAS data set is missing those elements and no direct comparison can be made
and are thus not included as rows in Table 2.  The unhighlighted elements in Table 1
represent approximately 12% of the concentration in Spicer's list.
     05/27/09

-------

                                               .ipo

 1
 2
Table 2. Compound Els normalized to formaldehyde (Elx/Elncno) for low engine powers
(4-15% rated thrust) evaluated as the slopes of plots ofEI(x) vs EI(HCHO)
Compound
Methanol
Propene
Acetaldehyde
Butene + Acrolein
Acetone + Propanal +
Glyoxal
Benzene
Toluene
mass 107
mass 121
mass 135
mass 149
Naphthalene
Methylnaphthalenes
Dimethylnaphthalenes
Phenol
Styrene
Acetic acid
APEX1
EIx/EIHcHo
0.18
0.36
0.32
0.30
0.18
0.15
0.056
0.088
0.074
0.035
0.014
0.018
0.009
0.0026
0.063
0.020
0.16
APEX 2
EIx/EIHcHo
0.14
0.39
0.36
0.45
0.16
0.17
0.082
0.138
0.119
0.074
0.038
0.034
0.023
0.011
0.064
0.035
0.057
APEX 3
EIx/EIHcHo
0.12
0.38
0.36
0.48
0.20
0.16
0.073
0.103
0.085
0.051
0.027
0.020
0.016
0.0083
0.050
0.023
0.084
Spicer et al.
EIx/EIHcHo
~
0.37
0.35
0.36
0.24
0.14
0.052
0.089
~
~
~
0.044
0.037
~
0.016
0.025
~
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
propene - quantified assuming that 68% all of the ion intensity measured at m/z43 originated originates
from propene.
butene + acrolein - quantified assuming the m/z 57 signal is distributed as reported by Spicer et al. 45%
butenes and 55% acrolein.
acetone + propanal + gyloxal - quantified assuming the m/z 59 signal is distributed as reported by Spicer
et al. 12% acetone, 25% propanal and 63% glyoxal.
Mass 107 - quantified as p-xylene and represents the sum of o,m,p-xylene, ethyl benzene & benzaldehyde
Mass 121 - quantified as 1,2,4-trimethylbenzene and represents the sum of C9H12 and C8H8O
Mass 135 - quantified using a single rate constant and represents the sum of Ci0H14 and C9H10O
Mass 149 - quantified using a single rate constant and represents the sum of CnH16 and Ci0H12O

However, as is noted by the first row of Table 2, Spicer did not measure methanol, which
is approximately 5% of Spicer's total concentration, which would increase the HC total
by that amount. There  are also several other aromatic species listed in Table 2, which
were not identified in the Spicer analysis. While these compounds would fall into the
second two columns of Table 1 if they were included, and thus represent only a percent
or so of the total concentration profile, they do represent a significant number of aromatic
compounds. These several differences in the lists of species measured in these
measurement studies highlight the uncertainty in working with any "total" emissions
level: the "total" is only a sum of whatever species are included in the "total".

A longer list of species measured in the APEX campaigns but not measured  by Spicer is
included in the accompanying spreadsheet.  In that spreadsheet, the additional  species are
color coded by blue  (from PTR-MS) and yellow (from EPA's set of integrating
measurements, Kinsey  et al, manuscript in preparation).
      05/27/09

-------
                                                  ..  '

 1    This spreadsheet also provides a normalized emission profile. This profile was
 2    developed based on Spicer's original speciation and carbon balance. Adjustments and
 3    additions were made, based on the new data available (all APEX1-3 data discussed in this
 4    report), but the measured species continue to make use of the original carbon balance.
 5    Thus, because of the longer list of species now quantified, these refinements to the
 6    speciation profile result in a decrease of the unidentified emitted mass from about 35% in
 7    the original Spicer work (34% if methane were included, but as discussed below, Spicer
 8    has shown elsewhere that the methane  measurement was due to background methane and
 9    should not be included in the  sum) to about 29% due to the additionally identified species
10    and refinements to phenol and butyraldehyde/crotonaldehyde. Separate analysis of the
11    total HC emissions by independent measurements during the APEX1-3 campaigns (J.
12    Kinsey, personal communication) used time-integrated  sampling over a range of power
13    conditions. Because that approach  is distinguished from the single power points
14    measured by Spicer and the data presented  here, precise agreement would not be
15    expected due to  different dependence on background levels and related data analysis
16    issues.  However, despite these potential differences, similar  ratios of the sum of
17    identified to the  total HC mass were calculated using the time integrated measurements in
18    APEX1-3 as compared to those of this revised profile, giving increased confidence in the
19    overall HC mass balance presented with these data.
20
21    While the unidentified species mass has been reduced through this process, the
22    composition  of that unidentified mass remains an uncertainty. In the original Spicer
23    profile, which used gas chromatography and various HC capture techniques, the
24    unidentified mass could possibly include contributions from some of the species that
25    were specifically identified.  In other words, based on the original Spicer work,  one could
26    argue that some  of the identified species may have been present in larger  amounts than
27    were reported because they may have also been contributing to the unidentified  mass.
28    That would be a result of some mass "sticking" to a GC column or a HC capture medium.
29    The new additional data reduces that uncertainty considerably, since independent real-
30    time data were collected which largely corroborated the Spicer profile. The combination
31    of the original and new data provide good evidence that the identified species contribute
32    to the profile at the levels measured and have little or no contribution to the unidentified
33    HC mass fraction.
34
35    Table 2 compares the concentration of the particular species of interest to that of
36    formaldehyde, which is one of the most prevalent emissions and serves as a useful
37    reference species.  This is done, rather than directly compare concentrations, since the
38    combustion efficiency is highly dependent  on precise fuel flow and power settings at low
39    engine powers, with ambient temperature also having a significant impact on emission
40    levels (Yelvington, et al). Since engine operating point, ambient temperature, and
41    related details are all slightly different  from test to test,  the combustion efficiency is also
42    likely to vary from data set to data set.  However, the relationship of the various
43    emissions to  each other is quite constant  even though their levels may go up and down
44    together.
45
      05/27/09

-------
                                                  ..  '

 1    Figure 1 shows that this is true not only for the CFM56 measured in APEX1 and
 2    discussed by Yelvington et al. but it is also true for the several CFM56 engines measured
 3    in JETS/APEX2 and, indeed, is equally true for the wider range of different commercial
 4    engine types measured in APEX3.  In fact, not only is it true that the speciation is
 5    invariant as a specific engine varies power and combustion efficiency, but for the range
 6    of commercial engines measured in APEX1 -3 and the range of standard jet fuels used
 7    through those tests, the relationship between the various HC emissions, (i.e. the
 8    speciation profile shown here as individual species plotted versus formaldehyde, HCHO),
 9    is also invariant across these different commercial  engine types: all of the curves lie
10    essentially on top of one another.  The invariance  of the speciation profile across power
11    settings, ambient temperature, and engine types for commercial engines Is very useful
12   for Interpreting HAPs emissions from commercial engines.
13
14    The range of fuels used in the diverse set of tests presented in Figure 1 suggests that fuel
15    also has a minimal impact on the speciation profile. Fuel sulfur and aromatic content
16    spanned a range of values across these tests, particularly when the APEX1 fuel sulfur
17    additions are included. However dramatic changes in the hydrocarbon composition of
18    the fuel, as might be encountered using alternative fuels like Fischer-Tropsch or bio-
19    fuels, have not been explored in the set of data presented here.
20
21    The correlation of each of the individual species versus formaldehyde plotted in Figure 1
22    show that, for the three APEX campaigns, there  are very tight correlations for the several
23    species plotted. Species present in greater concentration (propene and acetaldehyde)
24    have a tighter correlation than species at lower concentrations (benzene and, especially,
25    naphthalene, which is a PAH and may begin condensing on PM emissions soon after
26    leaving the engine, which might affect its gas phase concentration).
27
      05/27/09

-------
                         Technical Support Document
0.4 -,



0.3-
I
en
*"-*
iTi
« 0.2-
^
0)
N
1
0.1-


0.0-

• APEX1
• APEX 2
M APEX 3 ,
I I



*
•
• _ [xl [^ M
IX^-j [Xj
• M M
y*
•sP
Jp'
^P^'
#
I ' I ' I ' I ' I
0.0 0.4 0.8 1.2 1.6
HCHO El (g/kg)

0.8 -,


0.6-
t^vt*,
-S3
•H^
m 04_
c
0)
Q.
2
^ 0.2-


0.0-

.


t^W
•
M
M
JX*
i «^
ff
• •



*
i ' i ' i ' i ' i
0.0 0.4 0.8 1.2 1.6
HCHO El (g/kg)









































0.06 -,

0.05-

"3
§ 0.04-

LU
fit
c 0.03-
rti
V
CO
•4_»
•= 0.02-
CO
c
0.01-

0.00-


"
g
!
'

•
i '|
•" ' M M

g
1_ >^
• I H%^M M »
"IVMM M '
*r • *
itt$
w5^
i ' i ' i ' i ' i
0.0 0.4 0.8 1.2 1.6
HCHO El (g/kg)

0.6 -,
0.5-

"3
^c
ra 0.4-

Lil
0)
•o 0.3-
JC
0)
T3
S 0.2-
8
CO
0.1-
0.0-

•
M [x]
i i
M"
9
M
f
^N
**
>;ji
^
J^»
J¥
Jf
r
/
i ' i ' i ' i ' i
0.0 0.4 0.8 1.2 1.6
HCHO El (g/kg)
1
2
3
4
5
Figure 1.  Correlation scatter plots of selected HC vs. HCHO emission indices measured
on the 1-meter probe under low power, 4-15% rated thrust.

Table 2 provides the comparison between the speciation profile measured by Spicer and
that from PTR-MS/TILDAS from APEX. The three data columns from APEX cover the
CFM56-2C1 measured in APEX1, the several CFM56-3 and -7 engines measured in
    05/27/09
                                    10

-------
                                                  .. '

 1    APEX2, and an average over the set of engines (excluding the AE3007 for this analysis)
 2    measured in APEX3. For many species, especially those at larger concentration ratios,
 3    the variation among the various tests is no greater than the variation between APEX and
 4    Spicer. The speciation for these species appears to be very robust.  Some of the more
 5    minor species show more significant variation, which may be partly due to measurement
 6    uncertainty and may be partly due to sensitivity to other variables such as minor fuel
 7    composition variations and so forth.  It is  worth noting that many of these smaller
 8    contributors represent less than 1% of the speciated concentration mixture.
 9
10    One species of particular note is phenol.  The APEX series of measurements indicate a
11    concentration ratio three times higher than that of Spicer. That is the largest
12    disagreement in Table 2 (excluding cases where a Spicer measurement is not available),
13    and deserves further comment.  While phenol represents only about 0.1% of the
14    speciation concentration profile, it does represent a test of the ability to measure a minor
15    species accurately. While further analysis might be warranted, phenol was measured by
16    Spicer using canister capture to deliver the sample to the gas chromatographic
17    measurement system. Given the significant differences indicated for this compound, wall
18    losses might be suggested as a possible explanation for this unique discrepancy in the HC
19    speciation.
20
21    The overall agreement between the Spicer and the APEX speciation profiles is shown in
22    Figure 2. This is a direct comparison of the overall APEX speciation profile to that of
23    Spicer for those species where the measurements are available in both studies.  Except for
24    phenol and the combination of acetone, propanal, and glyoxal (which, unlike phenol, is
25    still within 2 sigma), all of the data are within the standard deviation of the measurements
26    themselves to the unit line (the unity line represents perfect agreement).
27
      05/27/09                               11

-------
                      Technical Support Document
 1
 2
 3
 4
 5
 6
 1
 8
 9
10
1 1
12
13
14
15
16
17
18
19
20
        0.6

        0.5 -]

        0.4

        0.3 H

        0.2

        0.1 H
          o
                                                                  O propene
                                                                  • acetaldehyde
                                                                  A butene + acrolein
                                                                  X acetone+propanal+glyoxal
                                                                  X benzene
                                                                  O toluene
                                                                  * mass 107
                                                                  D naphthalene
                                                                  A methylnaphthalenes
                                                                  O phenol
                                                                  X styrene
                                                                 	unit line
                          0.2
                                              0.4
0.6
                             Spiceretal.
                           EI(x)/EI(HCHO)
Figure 2.  Correlation plot of normalized compound Els derived from the APEX
experiments versus that reported by Spicer et al. APEX data is derived from the slopes of
the plots ofEI(x) versus EI(HCHO) for data obtained at low power 4% - 15% rated
thrust on the 1 -meter probe. Error bars reflect the standard deviation of the three
measurements.

The comparison of the various PTR-MS/TILDAS measurements across different engines
in APEX3 provides strong support that the speciation profile is invariant  across engine
technologies for commercial engines.  A similar question could be posed for the various
military engines measured by Spicer and Gerstle. Initial analysis (data not shown here:
KBE) indicates that there is much agreement between some of the relative amounts of
relevant HAPs. Detailed analysis of the TF39 (a forerunner to the GE CF-6 engine)
measured by Spicer is shown in Figure 3 comparing the speciation profiles for the TF39
with that of the CFM56 measured by Spicer, in analogy to what was shown in Figure 2
between the many engines of APEX1-3 and the Spicer  CFM56-3.  The speciation profiles
for these two engines measured by Spicer, which received the careful analysis required
for archival publication (Spicer et al. 1994), also support the contention that the
speciation profile from aviation gas turbine engines is invariant across engine types.
05/27/09
                                             12

-------
                                             ?po

 1
 2
 3
 4
 5
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
              10.000
               1.000
               0.100
               0.010
               0.001
                                                                       I All data
                                                                       oxygenates
                                                                       alkanes
                   0.001       0.010      0.100
                                          CFM-56
                                       EI(x)/EI(HCHO)
                                                  1.000
10.000
Figure 3.  Correlation plot of normalized compound Els for the TF39 vs. CFM56 engines
at ground idle derived from the measurements reported by Spicer et al.

Discussion: The emissions of organic gases are controlled by combustion efficiency.
The ICAO datasheets show a very clear trend of decreasing UHC emission indices from
idle and approach to climb-out and take-off.   There are strong dependences of the
magnitude of UHC emissions between different engine models in the ICAO databank.
The Yelvington et al. result from APEX-1 (Yelvington, Herndon et al. 2007) shows there
is a strong dependence of the emissions of HCHO on temperature; that emissions
increase at colder ambient temperatures, particularly for ground idle. This dependence is
greater than estimated in the Boeing Fuel Flow Method-27 correction (DuBois and
Paynter 2006).

Despite these strong dependences of the magnitude of UHC emissions on various factors,
a remarkable and simplifying result is that the relative profile of organic gas emissions
near idle does not have any such significant dependence, as presented in Figure 1. This
has been demonstrated for various engines to be valid for conditions from ground idle up
to -15%  of rated thrust.  This invariant speciation profile demonstrates that despite large
variations in the total amount of emissions, the ratio of benzene to ethylene, for example,
is a relatively constant value among different conditions and engines.
       The Boeing Fuel Flow Method is a theory-based means of obtaining estimated emissions data at power
      conditions other than the ICAO specified power points by interpolating ICAO certification data.
      05/27/09
                                        13

-------
                                                  ..  '

 1    One HC emission of particular note is methane (CfL;).  This compound was measured by
 2    Spicer, but not in more recent studies. While methane is present in the exhaust of aircraft
 3    engines, it is present at levels below ambient levels for most power conditions (Spicer,
 4    Holdren et al.  1992, Wiesen et al, 1994, Vay et al, 1998).  Indeed, in that reference
 5    Spicer notes "At power levels above idle, the exhaust is depleted in methane compared
 6    with the incoming air used for combustion.  The methane concentrations observed in the
 7    engine exhaust are consistent with partial combustion of the atmospheric methane present
 8    in the inlet air, although some methane production during combustion cannot be ruled
 9    out." At idle the methane values in the exhaust during the Spicer were consistent with
10    ambient levels, so any methane production must have been small enough to be within the
11    experimental uncertainty or was balanced by methane consumption. Methane is not
12    considered to be a significant emission from aircraft gas turbine engines burning Jet A,
13    and is not included in the profile information provided here.8
14
15    At engine power conditions significantly higher that -15% rated thrust, the engine
16    combustion efficiency is so close to 100% that measurement of many HCs becomes
17    difficult or impossible due to instrument detection  levels for diluted exhaust gases (either
18    with 1 m dilution probes or downwind sampling): the HC concentrations are too small to
19    measure. Thus, when considering the total emissions contribution from a given aircraft
20    operation, the amount of HCs is dominated by the low power conditions. Since the total
21    emissions burden is the product of an emission index (g pollutant/kg fuel) times the fuel
22    flow rate (kg fuel/sec) times the time in mode  for that power condition, even the high fuel
23    flow rates of take-off and climb-out cannot compensate for the very small emission
24    indices for HCs and the short times in the take-off and climb-out power conditions.
25
26    The dominance of the low power conditions in determining the overall HC  emission
27    loading suggests that any changes to the HAPs profile above 15% power will have
28    limited impact on the net HAPs loading. Since the emissions levels become too small to
29    measure for many of the smallest percentage HAPs in the profile, a bound can be placed
30    on how much their fractional contribution to the HAPs  profile might be increasing  as
31    power increases.  In lieu of specific data for these very  small levels, a default of retaining
32    the same profile as power increases beyond 15% could be  suggested, which would be
33    used for those powers above which the smallest contributors can be measured. An
34    analysis of the potential errors introduced in using  this default could be performed,
35    however Figure 4 suggests that the limits of detection of the instruments, in combination
36    with the rapidly decreasing overall HC emissions, will limit the overall uncertainties  in
37    the overall HC loading when using a low power HAPs profile.
38
39    Figure 4 demonstrates the relative importance of the elements of an LTO cycle by
40    accounting for times in mode and emission indices for UHCs from the ICAO databank
41    for a CFM56-3C1 engine (two engines for a 737-300).  The LTO cycle in the figure
42    reflects times in mode reported in the Boston Logan Airport 2005 Environmental Data
       When using this speciation profile in concert with reported certification HC emission indices, it is worth
      noting that ICAO CAEP Annex 16 Vol II makes no account for corrections to measured HCs due to
      ambient methane concentrations when reporting FID measurements for certification. Presumably ambient
      methane levels may be included in the certification FID measurements of El HC unless otherwise noted.
      05/27/09                               14

-------

 1    Report (Wilkins 2007). Essentially, it is a modestly adjusted set of times but the same
 2    power conditions as a standard ICAO LTO cycle.  Whatever the variation in the
 3    speciation profile at the higher powers, the lower Els at the high powers preclude a
 4    significant impact on the total emissions burden from the complete LTO cycle, at least in
 5    this first attempt to assess the speciated emissions.  Variation in the HC speciation profile
 6    at higher powers are unlikely to have a significant impact on airport air-quality modeling
 7    or to risk assessment from the compounds that are HAPs.
       to
       (J
       m
                        4 mins
                       approach
               0.7 mins
               take-off
                  2.2 mins
                 climb-out
            0.0
             00:00
00:10
00:20  flight   00:30
00:40
10                                        Time within LTO (min)
11   Figure 4. Emission Rate vs. Time in Mode.  The estimated emission rate, coupling fuel
12   flow and emission index for UHCfor a CFM56-3C1.  The LTO profile begins and ends
13   with a 7.5 minute APU interval. In this figure the apparent area of the  'boxes' reflect the
14   total emissions magnitude for the defined modes.
15
16   Relationship of Dedicated Engine Tests to Airport Measurements: Dedicated engine
17   tests allow control of the engine operation.  However emissions at airports are due to
18   airplanes being operated as required to satisfy airline requirements.  Table 3 compares
19   normalized emission ratios (Species concentration/Formaldehyde concentration) for
20   several APEX measurements and for advected plumes measured at Boston Logan,
21   Zurich, and Oakland airports.  While greater uncertainties might be expected in the
22   advected plumes measured in  a non-interference basis at airports, there  is very good
23   agreement between the emissions ratios measured in these disparate studies.  In the
24   advected plumes (last column) the error bars represent the width of the distribution of
25   results. This uncertainty can be taken as an upper limit on the real variability in these
26   ratios. When the detailed analysis of the instrumental contribution to this noise is
27   complete, it will likely narrow the range of species variability, as opposed to instrument
      05/27/09
                 15

-------

                                              .ipo

 1
 2
 3
 4
noise.  This is possible because the observed distribution in this sample is nearly
Gaussian.

Table 3.  Speciated VOC Index Ratio fHCHO relative;
Compound

HCHO
Acetaldehyde
C2H4
Propene
Butenes+Acrolein
Pentenes
Benzene
Toluene
1-ring Aromatics
Styrene
Naphthalene
Spicer
etal.

1
0.35
1.26
0.36
0.36
0.11
0.14
0.05
0.28
0.03
0.04
APEX-1

1
0.24
0.78
0.31
0.45
0.31
0.14
0.06
0.48
0.03
0.01
Logan

1
0.26


0.25

0.11
0.06
0.3


EXC



1*
0.32
0.45

0.08
0.01
-


Zurich



1*
0.32
0.26&
0.11
0.11
0.13
0.39
0.04

APEX-2
Staged
"1
0.37
0.76
0.45
0.49

0.18
0.09
0.73
0.04
0.04
OAK
Advected
1
0.31±0.09
0.85±0.3
0.42±0.2


0.15±0.08




 5
 6
 7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Table Notes:
    All values are in units of grams of VOC per gram of HCHO, except for the EXCAVATE
    column, which is grams of VOC per gram of C2H4.
    *The EXCAVATE and Zurich datasets have been normalized by the emission index for ethene
    in lieu of formaldehyde.
    &The Zurich tabulation for Butenes+Acrolein assumes the ratio of Acrolein to the sum of the
    butene isomers is that found in Spicer et al.
    The APEX-2, Staged aircraft column represents the average result for 'ground-idle' including
    the following engines; 3 CFM56-7B22, 1 CFM56-3B1, 2 CFM56-3B2.
    Tabulated values in the OAK Advected column represent Gaussian fits to the distribution of
    measured compound to HCHO ratios. The error bar is one Gaussian width.


Next Step Recommendations: First, recent work has reinforced the overall speciation for
commercial engine as measured by  Spicer for the CFM56-3 engine. Both in comparison
to the TF39 measured by Spicer and the wider range of commercial engines in APEX1-3,
this speciation profile is insensitive  to engine type, engine power condition, and ambient
conditions, even though those parameters significantly impact the total amount of UHCs
(or VOCs or total HCs, however one wants to add up a total). Other measurements
(Gerstle and airport advected plume studies) are also consistent with the general
invariance, near idle, of this speciation.

Several modest uncertainties are present, particularly for species that are present in small
quantities.  These may be due to measurement uncertainties, or due to actual variations in
emissions numbers themselves. Modest dependences on fuel composition or other
unknown parameters may cause some of this variation. Most of these variations are
within the uncertainties between studies or engines. One notable exception is the
significant variation  associated with phenol. The APEX studies all agree with one
another for phenol, while Spicer is significantly lower. Wall losses in the canister
      05/27/09
                                         16

-------
                                                 ..  '

 1    sampling done by Spicer are a possible explanation for such a loss, but additional studies
 2    may be warranted to resolve this discrepancy.
 3
 4    For future work, two remaining questions should be kept in mind.
 5       1.  On which set of species do we need to focus to further refine the HAPs profile?
 6          (E.g.: [1] phenol discrepancy, [2] methanol and the several aromatics and long list
 7          of species present at a fraction of a percent of total mass not measured by Spicer
 8          et al, [3] questions regarding acrolein/butene etc.)
 9       2.  What is our approach to evaluating (and possibly revising) the UHC-to-VOC-to-
10          TOG conversions?
11
12    With the completion of this analysis of Spicer and  APEX data, we can offer the
13    speciation profile provided in the accompanying spreadsheet for inclusion into the EPA's
14    SPECIATE database.
15
16    Cited References:
17
18    Anderson, B. E., G. Chen and D. R. Blake (2006).  "Hydrocarbon emissions from a
19    modern commercial airliner." Atmospheric Environ. 40(19):  3601-3612.
20
21    DuBois, D. and G. C. Paynter (2006). ""Fuel Flow Method2" for Estimating Aircraft
22    Emissions." SAE Technical Paper Series 2006-01-1987.
23
24    Gerstle, T., P. Virag and M. Wade (1999). "Aircraft Engine and Auxiliary Power Unit
25    Emission Testing: Vol. 1." United States Air Force: IERA RS-BR-TR-1999-0006.
26
27    Herndon, S. C., T. Rogers, E. J. Dunlea, R. C. Miake-Lye and B. Knighton (2006).
28    "Hydrocarbon emissions from in-use commercial aircraft during airport operations."
29    Environmental Science and Technology 40(14):  4406 - 4413.
30
31    Kinsey, J. S., Y. Dong, and D. C. Williams (in preparation 2008). "Characterization of
32    Emissions from Commercial Aircraft Engines during the Aircraft Particle Emissions
33    experiment (APEX)", U. S. Environmental Protection Agency, Office of Research and
34    Development, National Risk Management Research Laboratory, Research Triangle Park,
35    NC.
36
37    Knighton, W. B., T. Rogers, C. C. Wey, B. E. Anderson, S. C. Herndon, P. E. Yelvington
38    and R. C. Miake-Lye (2007). "Application of Proton Transfer Reaction Mass
39    Spectrometry (PTR-MS) for Measurement of Volatile Organic Trace Gas Emissions
40    From Aircraft." Journal of Propulsion and Power 23: 949-958.
41
42    Lobo, P., P. D. Whitefield, D. E. Hagen, S. C. Herndon, J. T. Jayne, E. C. Wood, W. B.
43    Knighton, M. J. Northway, R. C. Miake-Lye, D.  Cocker, A. Sawant, H. Agrawal and J.
44    Wayne Miller  (2007). "The development of exhaust speciation profiles for commercial
45    jet engines." Final Report, Contract No. 04-344,  California Air Resources Board,
46    Sacramento, CA, October 31, 2007.
      05/27/09                               17

-------
                                                 ..  '

 1
 2   Lobo, P., D. E. Hagen, P. D. Whitefield and D. J. Alofs (2007). "Physical
 3   characterization of aerosol emissions from a Commercial Gas Turbine Engine." Journal
 4   of Propulsion and Power 23: 919-929.
 5
 6   Onasch, T. B., J. T. Jayne, S. C. Herndon, P. Mortimer, D. R. Worsnop and R. C. Miake-
 7   Lye (2006). "Chemical Properties of Aircraft Engine Exhaust Aerosols Sampled During
 8   APEX." NAS A/TM-2006-2143 82 ARL_TR_3903: Appendix J.
 9
10   Schiirmann, G., K. Schafer, C. Jahn, H. Hoffmann, M. Bauerfeind, E. Fleuti and B.
11   Rappengliick (2007). "The impact of NOx,  CO and VOC emissions in the air quality of
12   Zurich Airport." Atmospheric Environ. 41:  103-118.
13
14   Spicer, C. W., M. W. Holdren, R. M. Riggin and T. F. Lyon (1994).  "Chemical
15   composition and photochemical reactivity of exhaust  from aircraft turbine engines." Ann.
16   Geophysicae 12: 944-955.
17
18   Spicer, C. W., M. W. Holdren, D. L. Smith, D. P. Hughes and M. D. Smith (1992).
19   "Chemical composition of exhaust from aircraft turbine engines." Journal of Engineering
20   for Gas Turbines and Power 114(1): 111-117.
21
22   Timko, M. T., S. C. Herndon, E. Wood, T. B. Onasch, M. J. Northway, J. T. Jayne, M. R.
23   Canagaratna, R. C. Miake-Lye and W. Berk Knighton (manuscript in prepartion, 2008).
24   "Gas Turbine Engine Emissions Part 1. Hydrocarbons and Nitrogen Oxides" Aerodyne
25   Research, Inc., Billerica, MA.
26
27   Timko, M. T., T. B. Onasch, M. J. Northway, J. T. Jayne, M. R. Canagaratna, S. C.
28   Herndon, E. Wood and R. C. Miake-Lye (manuscript in prepartion, 2008). "Gas Turbine
29   Engine Emissions Part 2. Chemical Properties of Particulate Matter" Aerodyne Research,
30   Inc., Billerica, MA.
31
32   URS and FAA (2003). Select resource materials and annotated bibliography on the topic
33   of hazardous air pollutants (HAPs) associated with aircraft, airports and aviation.
34   Technical Directive Memorandum: DO 1-010. CSSI Contract: DTFA 01-99-Y-01002.
35
36   Wey, C. C. (2004). Overview of the Aircraft Particle Emissions experiment (APEX)
37   Program. Aircraft Particle Emissions Workshop, Cleveland, OH.
38
39   Vay, S.A., B.E. Anderson, G. W. Sachse, J. E. Collins, Jr., J. R. Podolske, C.  H. Twohy,
40   B. Gandrud, K. R. Chan, S. L. Baughcum, H. A. Wallio. (1998) DC-8-based
41   observations of aircraft CO, CfL;, N2O, and H2O(g) emission indices during SUCCESS.
42   Geophysical Research Letters 25 1717-1720.
43
44   Wey, C. C., B. E. Anderson, et al. (2006). "Aircraft Particle Emissions experiment
45   (APEX)." NAS A/TM-2006-214382 ARL-TR-3903.
46
      05/27/09                               18

-------
                                                .. '

 1   Wiesen, P. 1, J. Kleffmann, R. Kurtenbach, K.H. Becker. (1994) Nitrous Oxide and
 2   Methane Emissions form Aero Engines. Geophysical Research Letters 21 2027-2030.
 3
 4   Wilkins, J. (2007). Description of the 2005 Boston Logan Airport Environmental Data
 5   Report. S. Herndon. Boston: LTO times compiled from actual data collected at the airport
 6   during 2005.
 7
 8   Yelvington, P. E., S. C. Herndon, J. C. Wormhoudt, J. T. Jayne, R. C. Miake-Lye, W. B.
 9   Knighton and C. C. Wey (2007). "Chemical Speciation of Hydrocarbon Emissions from a
10   Commercial Aircraft Engine." Journal of Propulsion and Power 23: 912-918.
11
12
13
14
     05/27/09                               19

-------
                                                  ..  '


 1                    Aircraft Engine Speciated Organic Gases:
 2                           Speciation Profile Spreadsheet
 3
 4                                     R.C. Miake-Lye
 5
 6    Introduction: A numerical spreadsheet9 was developed that used both the Spicer data and
 7    APEX data to formulate a speciated profile of HC emissions, as discussed in the main
 8    body of this report.  These several pages that follow are a description of the process used
 9    to develop that spreadsheet and an explanation of how the equations are used to provide
10    the resulting calculated quantities.  The spreadsheet (in Microsoft Excel format) is
11    intended to accompany this documentation, with its data and imbedded equations.
12
13    The initial formulation of the spreadsheet was based on the several data sources (Spicer,
14    and APEX, including both ARI/MSU and EPA contributions).  These data sources are
15    listed in columns, with the rows representing the numerous species measured by the
16    several investigators.  These data are combined to provide a single profile, as described
17    more fully in the main body  of the report. Below, the approach for that combination will
18    be described.
19
20    After the profile was finalized, with many new species added and a few adjustments of
21    specific species values from  the original  Spicer speciation, several additional quantities
22    were calculated. The calculations are all imbedded in the spreadsheet, via the equations
23    used to generate the quantities  in the labeled cells, and the rationale behind the
24    calculations will be presented below. The types of calculations are primarily directed at
25    understanding how the speciated profile relates to the total amount of HCs emitted, which
26    requires some assumptions since no measurement can quantify with complete certainty
27    all of the HC emissions.  As  part of developing that understanding, calculations were also
28    made to address questions of 1) how the  limited measurements (such as that from a
29    Flame lonization Detector or FID, as used in certification testing) can be  corrected to
30    approximate the full HC complement and 2) how to convert HC emissions expressed in
31    terms of methane mass equivalents, the reporting convention for HC Els  measured with a
32    FID, into an estimate of the actual total mass of the full speciation profile, including
33    unmeasured species.  These calculations are all described below, as well.
34
35    An abbreviated version of the speciation spreadsheet is reproduced in Table 4 below.
36    Species that have no mass fraction that are included in the spreadsheet for completeness
37    are not included in Table 4.  Also, the only columns from the spreadsheet that are shown
38    in Table 4 are the profile  species with non-zero mass fractions, their molecular mass and
39    formula numbers, and the profile mass fraction.  In the last two rows of Table 4, the
40    fractions of the total profile mass represented by the identified species (71%) and that
41    represented by the unidentified mass (29%) are listed, based on the total mass estimates
42    given by Spicer and refined with the new measurements, and supported by the EPA
      9 The accompanying Microsoft Excel filename is FAA-EPA_TSD_Speciated_HC_Aircraft_04AUG08.xls and
      all references to columns, rows, or cells can be found in the worksheet titled "Data Summary".


      05/27/09                                20

-------

                                            .ipo

1
2
3
4
5
6
APEX data as discussed in the main report. The approach for calculating these quantities
is discussed below.
                        Table 4.  Aircraft Profile Speciation
Species
Ethylene
Acetylene
Ethane
Propylene
Propane
Isobutene/ 1 -Butene
1,3 -Butadiene
cis-2-Butene
3 -Methyl- 1 -butene
1-Pentene
2-Methyl-l -butene
n-Pentane
trans-2-Pentene
cis-2-Pentene
2-Methyl-2-butene
4-Methyl- 1 -pentene
2-Methylpentane
2-Methyl- 1 -pentene
1-Hexene
trans-2-Hexene
Benzene
1-Heptene
n-Heptane
Toluene
1-Octene
n-Octane
Ethylbenzene
m-Xylene/p-Xylene
Styrene
o-Xylene
1-Nonene
n-Nonane
Isopropylbenzene
n-Propylbenzene
m-Ethyltoluene
p-Ethyltoluene
1,3,5 -Trimethylbenzene
o-Ethyltoluene
1 ,2,4-Trimethylbenzene
1-Decene
n-Decane
1,2,3 -Trimethylbenzene
n-Undecane
n-Dodecane
n-Tridecane
Molecular
Weight
28
26
30
42
44
56
54
56
70
70
70
72
70
70
70
84
86
84
84
84
78
98
100
92
112
114
106
106
104
106
126
128
120
120
120
120
120
120
120
140
142
120
156
170
184
Formula
C
2
2
2
o
J
o
J
4
4
4
5
5
5
5
5
5
5
6
6
6
6
6
6
7
7
7
8
8
8
8
8
8
9
9
9
9
9
9
9
9
9
10
10
9
11
12
13
H
4
2
6
6
8
8
6
8
10
10
10
12
10
10
10
12
14
12
12
12
6
14
16
8
16
18
10
10
8
10
18
20
12
12
12
12
12
12
12
20
22
12
24
26
28
O













































Mass Fraction
0.15459
0.03939
0.00521
0.04534
0.00078
0.01754
0.01687
0.00210
0.00112
0.00776
0.00140
0.00198
0.00359
0.00276
0.00185
0.00069
0.00408
0.00034
0.00736
0.00030
0.01681
0.00438
0.00064
0.00642
0.00276
0.00062
0.00174
0.00282
0.00309
0.00166
0.00246
0.00062
0.00003
0.00053
0.00154
0.00064
0.00054
0.00065
0.00350
0.00185
0.00320
0.00106
0.00444
0.00462
0.00535
     05/27/09
                                       21

-------

                                            .ipo

C14-alkane
C15-alkane
n-tetradecane
C16-alkane
n-pentadecane
n-hexadecane
C18-alkane
n-heptadecane
phenol
naphthalene
2-methyl naphthalene
1 -methyl naphthalene
dimethylnapthalenes
C4-Benzene + C3-aroald
C5-Benzene+C4-aroald
Methanol
Formaldehyde (FAD)
Acetaldehyde (AAD)
Acetone
Propionaldehyde
Crotonaldehyde
Butyraldehyde
Benzaldehyde
Isovaleraldehyde
Valeraldehyde
o-Tolualdehyde
m-Tolualdehyde
p-Tolualdehyde
Methacrolein
Glyoxal
Methylglyoxal
acrolein
Sum of all identified species
Unidentified mass
198
212
198
226
212
226
254
240
94
128
142
142
156
134
148
32
30
44
58
58
70
72
106
86
86
120
120
120
70
58
72
56


14
15
14
16
15
16
18
17
6
10
11
11
12
10
11
1
1
2
3
3
4
4
7
5
5
8
8
8
4
2
3
3


30
32
30
34
32
34
38
36
6
8
10
10
12
14
16
4
2
4
6
6
6
8
6
10
10
8
8
8
6
2
4
4










1






1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
1


0.00186
0.00177
0.00416
0.00146
0.00173
0.00049
0.00002
0.00009
0.00726
0.00541
0.00206
0.00247
0.00090
0.00656
0.00324
0.01805
0.12308
0.04272
0.00369
0.00727
0.01033
0.00119
0.00470
0.00032
0.00245
0.00230
0.00278
0.00048
0.00429
0.01816
0.01503
0.02449
0.70787
0.29213
 1
 2   Development of the Profile: As described in the main report, most of the species
 3   contributions measured by Spicer were supported by the APEX measurements. So, the
 4   data columns in the spreadsheet (not included in Table 4 above) show values for each
 5   measured species, in separate columns for each data source. If the data from APEX were
 6   not significantly different from Spicer's, the value from Spicer's column was used.  This
 7   was true for almost all of the species measured by Spicer.  The two exceptions were
 8   phenol and butyraldehyde (also called butanal). In these two cases, the more recent
 9   APEX data were used to update the values for those species. Then, additional species
10   from either the ARI/MSU team (color coded blue in the spreadsheet) or from EPA
11   (yellow) were also added to the species list and their contributions quantified. All of
12   these quantifications are first entered into the spreadsheet as column L as ratios of
13   emission indices of the species in question to the emission index of formaldehyde.
14   Column L is titled "Revised Ratios" since these are the El ratios of the individual species
15   to formaldehyde accounting for both Spicer and the more recent APEX data.
16
     05/27/09
22

-------
                                                  .. '

 1    Mass Fractions of Identified and Unidentified Species: Since the ratios of Els are really
 2    mass ratios (Els are mass ratios of the species to mass fuel burn, so ratioing Els divides
 3    one El by another and the fuel burn divides out), this column is a set of
 4    species:formaldehyde mass ratios. This is not useful for general application, since
 5    formaldehyde emissions are not generally measured or known.  To be most generally
 6    useful, a profile for the full complement of species was desired.  This could be done for
 7    the identified species, but there may be some HC species which contribute to the mass
 8    but which could not be identified.
 9
10
11    In order to attempt to account for all of the HC species, including those not identified,
12    Spicer's data was reviewed. Spicer attempted to do a complete carbon balance, based on
13    the suite of instruments employed and accounting for corrections for sensitivities.  With
14    the set of species measured by Spicer, an  estimate was made for the mass of the identified
15    species and for those that were not identified.  For the present purposes, we have taken
16    Spicer's values to be correct for the unidentified species based on his measurements.  We
17    have also compared to the time-integrated LTO cycle HC data taken by EPA,  which were
18    taken under a different set of measurement protocols (integrating over power settings,
19    including engine start, and the corresponding  different effects of backgrounds) and have
20    determined that the APEX EPA data is largely consistent with the Spicer data set (see
21    main report).
22
23    So Spicer's unidentified mass fraction was taken as a starting point. However, the
24    additional species included in APEX actually  reduced this unidentified list and
25    unidentified  mass. And the adjustments of phenol and butyraldehyde must also be
26    accounted for.  So, in calculating column  K of the spreadsheet (reproduced for non-zero
27    mass fraction species in Table 4 above), the original species in Spicer's profile were
28    summed (see cell K7  in the spreadsheet).  In the equation in cell K7, the phenol and
29    butyraldehyde values were individually reset to Spicer's original values rather than use
30    the new "Revised Ratio" values in column L,  and the sum was set to Spicer's  original
31    identified mass fraction so that Spicer's carbon balance could be used, albeit with the
32    unidentified  mass reduced due to the newly added species. (Note that Spicer quotes his
33    numbers in terms of parts per million carbon (ppmC) concentrations. However, within
34    the limits of accuracy of these calculations, the fractions of ppmC reported for identified
35    and unidentified are equivalent to the masses identified and unidentified in that the
36    mass/carbon for the two fractions, identified and unidentified, is not significantly
37    different for  these two fractions.)
38
39    With the sum calculated and set equal to the Spicer's identified mass fraction, the
40    individual mass ratios in column L can be scaled such that they can be referenced to a
41    total given by Spicer's carbon balance. Cell K7 of the spreadsheet takes the sum and
42    uses Spicer's identified mass fraction, to give  a scaling factor for each identified species
43    in the  "Revised Ratios" column (L) to give the resulting profile in column K.  Because
44    more species are now identified, the sum  of the identified profile now comes to 70.8% of
45    the total HC  mass (cell Kl 19).  The new additions to the list represent 6.4 of the total HC
46    mass (cell K120), while phenol and butyraldehyde adjust things a little as well from
      05/27/09                               23

-------
                                                  ..  '

 1    Spicer's original identified mass.  The unidentified mass represents 29.2% of the total
 2    mass in this new profile (K126).
 3
 4    Determining Effective Mass of Total Profile and Corrections for FID response:  Since
 5    certification data from commercial engines are available, such data are often used for
 6    estimating HC emissions.  Unfortunately, there are two problems with that certification
 7    data. The measurement device prescribed for this measurement uses a FID, which
 8    essentially "counts carbon atoms". This raises two problems. One is that the FID does
 9    not count carbon atoms that have an oxygen atom attached, and so is not equally sensitive
10    to all HC species.  Second, since the measurement is "counting carbons", it keeps track of
11    a concentration and there is no direct indication of the mass of the species in question.
12    (Mass is determined by the amount of hydrogen and oxygen in the molecule in addition
13    to the number of carbons.) For the certification numbers, a mass/carbon  based on the
14    methane molecule (molecular mass 16.04) is used by convention.  This is purely an
15    assumption and has not been based on measurements to date, as far as the authors know.
16
17    These two problems can be addressed with the detailed profile provided in column K.
18    For the known species, the mass/carbon can be calculated, since both the number of
19    carbons and the molecular mass are known (given in Table 4 above and in spreadsheet
20    columns D and H, as well as hydrogen and oxygen numbers provided in  columns I and J).
21    From the individual mass/carbon numbers and the mass fractions in column K, the mass-
22    weighted mass/carbon can be calculated for the identified profile.  In order to correct the
23    mass for the total profile, one would need to have the mass-weighted mass/carbon for the
24    complete profile. Since we do not have the molecular masses and formulas for the
25    unidentified species (because they are unidentified), that calculation cannot be done
26    rigorously. Thus, an estimate of the mass/carbon for the full profile is required.
27
28    In order to estimate the mass/carbon for the full profile, the mass/carbon  was examined
29    for two classes of species in the identified species.  The first class represents those
30    species present at greater than 1% of the total HC mass in column K.  This includes many
31    light oxygenated HCs, which have a large oxygen contribution to their total mass.  The
32    mass/carbon for these light species is not likely to be similar to those larger HC in the
33    unidentified mass contributions.  These light species have a mass/carbon of 17.6 (cell
34    C130).  The remaining species in the identified list have a mass/carbon of 14.4 (cell
35    C132), which is likely closer to what might be expected for the larger, partially oxidized
36    species in the unidentified component.  Any deviation from 14.4 for the unidentified,
37    while not expected to be large, is also devalued by the modest (29%) contribution of the
38    unidentified to the total. This argument indicates that a good estimate for the
39    mass/carbon for the full profile can be calculated using this approach.  Thus, assigning
40    14.4 to the unidentified and combining with the identified, gives a total mass/carbon for
41    the full profile of 15.97 (cell C137). (This is surprisingly close to the original convention
42    of using methane's molecular mass of 16.04.)
43
44    The problem of the FID's lack of sensitivity to carbons bound to oxygen can also be
45    rectified by the profile information.  By a similar approach to calculating the mass-
46    weighted mass/carbon, the mass-weighted C/H/O ratios for the various profile
      05/27/09                                24

-------
                                                 ..  '


 1    components discussed above can be calculated. These are included in cells EFG130,
 2    EFG132, and EFG137. By ratioing to carbon, in cells EFG131, EFG133, and EFG138,
 3    the number of oxygens/carbon can be determined.  Since each carbon effectively cancels
 4    out the measurement of one carbon by the FID, the FID response for the full profile can
 5    be estimated by subtracting cell G138 from cell El38 (or  1.00 - 0.1365). The FID
 6    response is then 0.8635 of the total carbon number, or the correction  for the FID's lack of
 7    sensitivity due to oxygen containing molecules is 1.16 times the FID output.
 8
 9    To summarize:
10
11    To correct for the FID response to account for the oxygen content, multiply the FID
12    measurement by 1.16.
13
14    To make use of the best estimate of the actual molecular masses of the HC species
15    instead of using the equivalent methane convention, multiply the FID measurement by
16    15.97/16.04 = 0.996.
17
18    The net total correction is 1.16 times 0.996 = 1.16.
19
20
      05/27/09                               25

-------