The Combustion of Droplets of Liquid Fuels and Biomass Particles
K.D.Bartle1, E.M.Fitzpatrick1, J.M.Jones1, M.L.Kubacki1, R Plant2, M.Pourkashanian1, A.B.Ross1 and
A.Williams1*.
1
Energy and Resources Research Centre, Leeds University, LS2 9JT, UK
2
Present address: Alstom Power Ltd, Derbyshire, DE74 2TZ, UK
Abstract
Studies have been made of the combustion of droplets of liquid biofuels such as FAME and the alcohols, especially
ethanol and n-butanol, and of pulverised solid biomass materials such as wood and Miscanthus which burn in an
analogous fashion. Information is given on the burning rates of both the liquids and the solids and data given on
soot formation yields for the different fuels. The mechanism of soot formation is discussed in relation to (1) volatile
liquid fuels such as n-heptane, alcohols and aviation fuels, (2) liquid fuels having higher aromatic levels such as
diesel fuels, and (3) biomass particle combustion.
1. Introduction
For half a century extensive research has been
undertaken on the combustion of sprays of conventional
fuels for engines and for power generation, and
extensive information has been obtained on the rates
and mechanism of combustion, on soot and on NOx
formation. Generally it is agreed that after the initial
stage droplets burn approximately at a rate give by the
d2 law even in the case of multi-component fuels such
as commercial fuels which have similar physical
properties [for example 1-5]. However in the case of
multi-component fuels the sooting propensity is a strong
function of the aromatic content even though the
burning rates do not vary significantly [8-10].
Recently attention has been directed to alternative liquid
fuels such as FAME, bio-oil and the alcohols especially
bio-ethanol and bio-butanol, only a few measurements
have been made previously [1, 11].
At the same time attention has been directed to the use
of finely pulverised biomass materials such as wood and
Miscanthus for power station co-firing applications and
a number of fundamental studies have been made [for
example, references [12, 13]. Comparison can be made
of droplet combustion with the combustion of small
solid biomass particles where a large fraction of the cell
wall material decomposes to give a gaseous product,
and then combustion proceeds in an analogous fashion
to the evaporation of a liquid droplet.
The combustion of these fuels has much in common and
these features are examined in this paper.
3. Experimental methods
The burning rate experiments consist of photographic
examinations of suspended burning droplets and
suspended wood particles. These gave information on
the decrease in size as a function of time, and hence the
burning rates. Relative indications of the amount of
soot produced were determined using direct
photography and measurements of the soot radiance
from the image.
An examination has been made of soot produced which
was heated using py-GC-MS temperature programmed
analysis. A similar technique has been used for the soot
from the wood samples.
Experiments were also undertaken using the porous
sphere technique to give information on the mass
burning rates and the soot index. A number of fuels
were used including pure hydrocarbons, commercial
aviation kerosene turbine fuel (Avtur) with an aromatic
content of 19.9% and a diesel fuel (D1) with an
aromatic content of 29.9%.
4. Experimental results
4.1. Burning Rates for Droplets for a Range of
Conventional Fuels and Biofuels.
The burning rate constant, K, is defined by the
equation d(d2)/dt =K and a considerable amount of
experimental data are available in the literature on the
burning rate constants for both suspended and moving
droplets under a variety of conditions.
Only data for a range of suspended single droplets are
given here but this is suitable for comparison purposes.
These are for fuels with an initial size of 1mm or less as
a function of temperature and are given below in Table
1. These are mainly for single components but most
distillate commercial fuel behave as single component
fuels as far as the physically controlled burning rates are
concerned [7].
2. Specific objectives
Whilst the formation of smoke and NOx from burning
hydrocarbon fuels has been extensively studied much
less is known about the formation of these from both
biofuels and solid biomass combustion. Biomass
generated smoke is of considerable importance both
from a health point of view and because it has
significant atmospheric chemistry implications [eg 14].
1∗ Corresponding author: [email protected]
Proceedings of the European Combustion Meeting 2009
1
Table 1 Burning rate constants (mm2s-1) for a range of
n-heptane, FAME and for a wood particle are shown in
Figs 1-3 respectively.
single droplets as a function of temperature (K).
Fuel
Kerosene/AVTUR
n-Heptane
FAME
Palm oil/rapeseed
Sunflower oil
Benzene
Vegetable oil
Ground nut
Ethyl alcohol
Diesel fuel (D1)
n-Butyl alcohol
290K
720K
1000K
1110
0.92
0.92
1.0
1.0
0.90
0.88
0.86
0.82
0.80
0.79
0.72
1.0
1.2
1.2
1.2
1.1
1.2
0.99
1.25
1.0
Fig. 1. Photograph of a single suspended droplet
(1 mm diam) of n-heptane.
The burning rate constants for all these fuels do not
differ much. The values were taken from a number of
sources [1, 2, 6, 11-13, and unpublished work in this
laboratory] and thus using different experimental
techniques thereby increasing the error. A good set of
data taken under identical conditions is desirable. It is
interesting to note that the burning rate constants for
kerosene > vegetable oils> diesel oil. But in most spray
combustion applications the main difference in their
combustion behaviour is determined by the energy
density of the fuel and the rate of heat release.
Data were also obtained on the mass burning rates of
single spheres of liquid fuels under steady burning
conditions. Typical of these it was found that for a 2
mm diameter sphere, the burning rate (dm/dt) for diesel
fuel = 1.0 mg s -1 and 1.5 mg s -1 for both n-dodecane
and for n-heptane. The addition of small amounts of
aromatic compounds to any of these fuels did not
change the burning rates by any significant amount.
It is not possible to directly determine burning rate
constants for solid materials such as biomass but
equivalent mass burning rates of the volatile component
of wood particle can be determined. This is because
approximately 90% of the material is given off as
volatile material. Combustion of biomass particles,
where the cellular material effectively depolymerises,
gives off a gas analogous to the evaporation of a liquid
droplet. Thus the data can be interpreted in terms of
single droplet combustion theory.
In the case of wood and miscanthus dm/dt = 0.25 mg s -1
for the period of volatiles release, which is about a
quarter of the value found for hydrocarbons, and which
equates to about one eight of the rate of release in terms
of the calorific value.
Fig. 2. Photograph of a single suspended droplet
(1 mm diam) of FAME.
Fig. 3. Photograph of a single suspended particle
(1mm long, 0.3 mm diam) of Miscanthus.
A range of suspended droplets or particles of different
fuels were studied. They ranged from fuels that
produced flames that were completely non-sooting (blue
flames) ranked as 0, to those that produced extensive
soot and ranked as 10.
The range of soot luminosities for the liquid fuels were:
ethyl alcohol, 1; n-butanol, 3; n-heptane, 5; n-dodecane,
8; rapeseed oil, 9; fatty acid methyl ester (FAME), 9;
diesel oil (D1), 10; and for the solids, pine
particle/miscanthus, 9; Miscanthus particle, 9.
4.2 Sooting tendencies of different fuels including
biofuels
Photographs were taken of the burning droplets and
biomass/wood particles and the extent of the yellow
intensity was measured by a densiometric method.
Typical photographs taken of a slightly sooting flame,
2
Some information is available on the concentrations of
acetylene, benzene and soot in the inner flame zones for
a burning droplet of aviation fuel (the flame looks very
similar to the n-heptane droplet shown in Fig 1), and
this is given below:
At 0.5 mm distance from the surface the concentration
of benzene was 33.9 ppm, acetylene 1476 ppm, at 1 mm
13.3, 1289, and at 1.5 mm 14.5, 1035,and at 2 mm 0, 0,
respectively.
Butadiene was also present.
The
maximum temperature was at 1mm and the maximum
rate of soot formation at 0.5 mm.
Measurements were made of the total amount of soot
produced during the combustion of single droplets, that
is, the soot index which is given in terms of mass yield
of soot/unit of fuel burned. Soot indices were obtained
for droplets with an initial diameter of 1.6 mm over a
range of temperatures as shown in the Fig. 4 below.
Fig. 5. Plot of soot index for various fuels burning on
porous spheres of different radius, rL: O, benzene; ●,
Dieso; Δ, Avtur; □, Premium paraffin; n-heptane; ■, ndodecane; ▲, n-hexadecane and ▼, Derv, D1.
Information is also given on the effect of influence of
aromatic additives to n-dodecane on smoke formation
and this is given in Fig. 6.
Two of the mixtures were attempts to replicate the
sooting capability of real fuel using a mixture of (a)
butyl benzene, an olefin (indene) and n-decane, and (b)
butyl benzene, tetralin and n-hexadecane. It is clear that
the attempts failed.
Fig. 4. Plot of soot index for burning droplets of various
fuels: ∆, Dieso; O, Avtur; □, 2 % indane with ndodecane; ▲, 2% tetralin with n-dodecane; ●, 2 %
indene with n- dodecane; ■ 20 % toluene + n- dodecane
A marine diesel fuel (Dieso) and an aviation fuel
(Avtur) were burned. It is clear that the soot index for
Dieso is higher than that for the aviation fuel. Attempts
were made to replicate the soot yields and that it is not
possible to generate as much soot from the synthetic
fuel mixtures selected here.
Measurements have also been made of the smoke
formation from diesel and other fuels during
combustion under steady-state conditions using a porous
sphere technique and data recorded as the soot index.
This method produces larger yields of soot and is more
accurate. A number of fuels were burned including a
marine diesel fuel (Dieso), an aviation fuel (Avtur) a
premium paraffin heating oil (Esso Blue) and some pure
hydrocarbons.
Fig. 5 shows the soot indices as a function of porous
sphere diameter for the range of fuels. A number of
features can be observed: (1) There is a correlation with
the aromatic content, (2) Low sooting fuels only
produce soot above a critical diameter.
Mol fraction added aromatic compound
Fig. 6. Influence of added aromatic compounds (mol
fraction) on smoke formation from n-dodecane: Δ,
benzene; O, ethyl benzene; □, toluene, ▲, naphthalene
and ●, tetralin.
3
Clearly the aromatics, especially naphthalene had the
largest effect and tetralin had the smallest influence.
Measurements were made of PAH formation from
aviation fuel wood and aviation fuel and chromatograms
of products are given in Figs 7 and 8 respectively.
The aviation fuel soot contains small concentrations of
fuel constituents and fragments because sampling was
made of the soot from the whole of the burning droplet,
so that some of the original remained in the wake flame.
It is interesting to note that smaller fragments are not
present in the wood soot (possibly due to differences in
the methods used and particularly that slightly different
columns were used), which also contains higher MW
oxygenated PAH.
5.2 General issues regarding soot formation.
Soot precursors may be formed during the burning of
fuel particle and liquid droplet volatiles by a number of
mechanisms which can occur individually or in parallel,
depending on the chemical nature of the fuel:
(a) the formation of PAH from monocycles by
hydrogen abstraction/ carbon addition (HACA). This
requires the presence of acetylene and/or butadiene and
is indicated by the identification of appropriate PAH
series by GC-MS, and of PAH series with members
differing by 24 Da (C2) and 50 Da (C4H2) by probe MS.
The contribution of the HACA route may be determined
by modelling using procedures such as CHEMKIN [15].
(b) through cyclopentadienyl (CPDyl) radicals
generated from the phenoxy radical which arises from
oxidation of the benzene ring, or from phenols; CPDyl
generates naphthalene and indene from which higher
PAH may be produced either by HACA or further
reaction with CPDyl radicals. In the case of biomass,
CPDyl may also be formed from the lignin. The higher
PAH resulting from reactions of CPDyl may be
condensed or, by reaction with five-membered ring
PAH, have more 'open' structures. Evidence for the
involvement of CPDyl radicals is provided by the
presence of high concentrations of naphthalene and
indene in the soot.
( c) by direct coupling of PAH either liberated from
the fuel or generated by the above routes to give
'aromers'. The resulting compounds have high molecular
weights.
1
8
3
2
4
5 7
6 9
10
Fig. 7. Py-GC-MS of soot from single droplet
combustion of AVTUR: 1, benzene; 2, naphthalene; 3,
methylnaphthalene; 4, phenanthrenes; 5, pyrene; 6,
benzofluorene; 7, benz[a]anthracene; 8, chrysene; 9,
benzopyrenes/benzofluoanthenes;10, coronene.
(x1,000,000)
4.0
TIC
7
3.5
5.3 Soot formation during the combustion of volatile
liquid fuels.
Light hydrocarbon fuels such as n-heptane and
alcohols contain an alkane polymethylene chain from
which acetylene is produced on pyrolysis, so that soot
precursors are likely to be generated by the HACA route
[17].
Aviation fuels such as kerosene are also
dominated by n-alkanes, and hence a HACA route is
likely to be important here. However, these fuels also
contains significant concentrations of monocyclic
aromatics which give rise to CPDyl and hence higher
PAH. This is consistent with the conclusions of D'Anna
and Violi [18], which were determined by a kinetic
study.
The mechanism of soot formation for aviation fuel is
thus likely to have contributions from both HACA and
CPDyl routes and analysis by Py-GC-MS of the soot
collected during droplet combustion is consistent with
this. The PAH profile (Fig. 7) shows significant
contributions from benzene, a CPDyl precursor, and
then naphthalene and methylnaphthalenes, initial
products of CPDyl reactions; products of further
reaction of CPDyl identified by Mulholland and coworkers [19, 20, 21] such as benzofluorenes,
benz[a]anthracene and chrysene are also prominent.
The simultaneous operation of the HACA route to PAH
is suggested both by the concentrations of acetylene and
butadiene in the surrounding gas during the combustion
of kerosene droplets and by the presence of PAH with
9
3.0
2.5
2.0
35
4
1.5
1
1.0
2
6
8 10 11
0.5
10
20
30
40
50
60
Fig. 8. Py-GC-MS of soot from wood combustion. 1,
naphthalene; 2, vanillin; 3, 1,6-anhydro-.beta-Dglucopyranose (levoglucosan); 4, 9H-fluoren-9-one; 5,
phenanthrene; 6, 2-dibenzofuranol; 7, fluoranthene; 8,
cyclopenta(def)phenanthrenon, 9, pyrene; 10, benzo[b]
naphtho[2,3-d]furan; 11, benzo[ghi]fluoranthene.
5. Discussion
5.1 Combustion of droplets and particles
It is proposed that during the combustion of solid
particles of biomass (and also coal or heavy oil)
volatiles are first generated and that this then burns in a
way analogous to the combustion of vapour from
droplets of liquid fuels.
4
was determined (Fig. 6). As expected n-hexadecane had
little effect on the HACA mechanism for soot
formation, but much larger (and similar) SI values were
observed for the monocyclic aromatics. This is
consistent with the progressive substitution of CPDyl
radical reactions for HACA.
‘The very rapid growth of soot when naphthalene (Fig.
6 ) was added to n-dodecanecan be primarily attributed
to the formation of ‘aromer’ precursors; binaphthyl
dimers are the most abundant products of naphthalene
pyrolysis (15% yield at 850oC) although other major
products include phenylacetylene and indene [21].
Pyrolysis of the latter [20] produces a variety of fourring PAH.
MWs separated by 24 Da (e.g. phenanthrene/ pyrene)
and 50 Da (pyrene/ benzopyrenes). Both CPDyl and
HACA routes could lead to the observation of higher
PAH compounds.
Thus burning kerosene as approximately 1 mm sized
droplets generates a mixture which can be thermally
desorbed from the soot of mainly monoalkylated and
unsubstituted PAH with between 2 and 7 rings (Fig. 8)
whereas the particulate PAH from (piston) engine
combustion [22] of a similar fuel are mainly
polyalkylated derivatives of 2- and 3-ring aromatics. A
comparison of these combustion products leads to the
conclusion that the precursors of soot generated in high
yield from droplets arise mainly by synthesis, whereas
the engine-soot PAH are mainly the unburned residue of
those in the original fuel.
5.6 Solid fuels.
In principle the constituents of the tar volatilised from
coal particles prior to combustion can generate soot
precursors by any of the three routes (a) - (c) above: the
polymethylene chains present in coal can generate
acetylene, the HACA reagent; the liberated phenols can
lose CO to produce CPDyl and hence produce the
copious amounts of naphthalene observed during coal
pyrolysis; and aromers may arise by hydrogen
abstraction from higher PAH present in the tar followed
by combination of the resulting radicals.
However, application of a model in which fractional
soot volumes produced by the HACA mechanism were
calculated suggested that the latter was less important
than the ‘aromatic' routes involving either CPDyl and
then naphthalene and indene or, as suggested by Wornat
[23, 24, 25 ], aromers, from the tar.
5.4 Liquid fuels with higher contents of aromatics.
Diesel fuel contains n- and branched chain alkanes
centred approximately at C16 which can be presumed to
produce on pyrolysis the C2 and C4 radicals necessary
for the HACA process. Like kerosene, monocyclic
aromatics are also present in Diesel fuel, and although
many of these have longer chain alkyl substituents they
are likely precursors of the soot PAH intermediates via
CPDyl. Diesel fuels also contain both unsubstituted and
alkylated PAH which are available for reaction through
the HACA mechanism and by reaction with CPDyl; nor
can reaction of these PAH by condensation through
hydrogen abstraction followed by combination of
resultant radicals to form aromers be excluded.
The lower MW fraction of heavier oils can be
considered to contain the same chemical types of
precursors as Diesel fuel, but may also contain
asphaltenes for which the chemical structure and MW
range are still the subject of controversy. Asphaltenes
can, however directly follow a route to smoke or
cenospheres in a way similar to smoke formation from
coal.
5.7 The role of fuel-nitrogen
All the liquid fuels have low fuel-N contents and the
major generation is via prompt NO and thermal routes,
the former depending on the C/H ratios. These can be
modelled fairly accurately using chemical models that
are currently available.
Solid biomass fuels are more complicated. The fuel-N
content can be high depending upon the fuel (e.g.
agricultural residues, and pits and shells). Secondly,
some of the nitrogen is released as ammonia and some
as HCN [26], the latter being normally the only product
with hydrocarbon fuels. It is necessary to be able to
predict the N-distribution between the volatiles and the
char. However, information on this is now becoming
available, and it is clear that the distribution is more
complex than for coal [27].
5.5 Soot formation during the combustion of liquid fuels
using a porous sphere
Measurements of soot index (SI) as a function of
droplet-sphere radius at constant burning rate (Figs. 5
and 6) are consistent with the above arguments and with
the proposal that SI is increased when soot is generated
by the CPDyl route. n-alkanes (C7, C10, and C16) show
little soot-forming tendency, although this increases
with carbon number.
The presence of a small
concentration of monoaromatics and with them the
contribution of the CPDyl radicals increases the SI of
domestic heating fuel paraffin, which further increases
for aviation fuel (20% aromatics), and there is another
marked increase for the Diesel fuels where the CPDyl
mechanism is supplemented by the formation of
aromers. The largest values of the SI are observed for
the burning of benzene where CPDyl formation is
predominant [18].
The effect on the SI of dodecane by the addition of a
variety of (mainly) aromatics for a 3.2 mm radius sphere
6. Conclusions
The burning rates for oxygenated fuels are quite
similar to those for hydrocarbon fuels and the relative
rates are capable of being calculated from their physical
properties. The burning rates of the volatiles released
during the burning of small particles of biomass are also
similar to those for hydrocarbons-both being determined
by heat transfer considerations.
5
[14] Y.B.Yang,, V.N.Sharifi , J. Swithenbank, L. Ma,
L.I. Darvell, J.M. Jones, M. Pourkashanian, A.
Williams, A. Energy Fuel 22, (2008) 306-316.
The rate of the release of the thermal energy is quite
different because the calorific values of solid biomass
volatiles is lower. A similar feature holds for the
biofuel droplets.
The soot forming tendencies vary very significantly
depending on the aromatic content of the fuel being
considered. For aromatic hydrocarbons, the sooting
tendency generally increased with an increase in the
number of carbon atoms per molecule. The influence of
the fuel-oxygen on sooting behaviour appears to follow
the rules set out by Pepiot-Desjardins et al. [28].
[15] E.M. Fitzpatrick, J.M. Jones, M. Pourkashanian,
A.B. Ross, A. Williams, K.D. Bartle, Energ. Fuel, 22
(2008) 3771-3778.
[16] J.M Jones, A.B. Ross, A. Williams, J Energy Inst.
78 (2005) 199-200.
[17] H. Richter, J.B. Howard, Prog. in Energy and
Combust. Science 26 (2000) 565–608
Acknowledgements
We wish to thank EPSRC for support for RP and EMF.
We also thank MOD Pyestock for suppoting some of
this work and for giving permission for it to be
published.
[18] A. D’Anna, A. Violi, Energy. Fuel 19 (2005) 7986.
References
[19] J.A. Mulholland, M. Lu, D. H. Kim, Proc.
Combust. Inst. 28 (2000) 2593-2599.
[1] H. Wise, G A Agnoston Burning Liquid droplets
Advances in Chemistry Series No 20, Am., 116135.Chem. Soc Washington DC 1958
[20] M.Lu, J.A. Mulholland, Chemosphere 42 (2001)
625-633.
[2] A. Williams, Combust. Flame 21 (1973) 1-31.
[21] M. Lu, J.A. Mulholland, Chemosphere 55 (2004)
605-610.
[3] A. Williams, Combustion of Liquid Fuel Sprays.
Butterworths, London, 1989.
[22] I.L. Davies, M. Raynor, P.T. Williams, G.E.
Andrews, K.D. Bartle, Anal. Chem. 59 (1987) 25792585.
[4] C. K. Law and F. A. Williams, Combust. Flame 19
(1972) 393-405.
[23] M.J. Wornat, E.B. Ledesma, N.D. Marsh, Fuel 80
(2001) 1711-1726.
[5] W L. H. Hallett, Combust. Flame 121, (2000) 334344
[24] S. Thomas, M.J. Wornat, Fuel 87 (2008) 768-781.
[6] R.H Sioui, L.H.S Roblee, Combust and Flame, 13,
(1969) 447-454.
[25] S. Thomas, E.B. Ledesma, M.J. Wornat, Fuel 86
(2007) 2581-2595.
[7] S.S. Sazhin, Progress in Energy and Combustion
Science 32 (2006) 162–214
[26] K-M. Hansson, L-E. Åmand, A. Habermann, F.
Winter, Fuel 82, (2003) 653-660.
[8] T. Kadota, H Hirorasu,, Combust Flame 55 (1984)
195-201
[27] J.M.Jones et al., Unpublished information,(2009).
[9] K. Nakanishi, T.Kadota, H. Hiroyasu, Combust.
Flame, 40 (1981) 247-262.
[28] P Pepiot-Desjardins, H. Pitsch, R Malhotra, S.R.
Kirby, A.L. Boehman Comcust. Flame, 154 (2008) 191205.
11.
[10] C.T. Avedisian, B.J. Callahan,, Proc Combust Inst.
28 (2000) 991-997.
[11] M.A.A Nazha, R.J Crookes, Applied Energy
Research Institute of Energy, Adam Hilger, London,
1989
[12] R J. Crookes, F Kiannejad, M A. A. Nazha, Energy
Conversion and Management, 38 (1997) 1785-1795.
[13] C. Morin, C. Chauveau , P. Dagaut , I. Gökalp , M.
Cathonnet, Combust. Sc. and Tech, 176 (2004) 499-529.
6