Available online at www.sciencedirect.

com
Proceedings
of the
Combustion
Institute
Proceedings of the Combustion Institute 33 (2011) 3325–3332
www.elsevier.com/locate/proci

Experimental study of the mild combustion

of liquid hydrocarbons
Marco Derudi ⇑, Renato Rota
Politecnico di Milano, Dip. di Chimica, Materiali e Ingegneria Chimica “Giulio Natta”/CIIRCO, 20131 Milan, Italy

Available online 15 September 2010

Abstract

A strong exhaust gas recirculation combined with air preheating generates relatively low maximum flame
temperatures allowing for a better control of the furnace temperature and a lower pollutant production. While
the sustainability of this combustion technology (named mild or flameless combustion) for different gaseous
fuels has been extensively studied, much less information are available for liquid fuels. Consequently, this work
has focused on the investigation of the sustainability of mild combustion for liquid hydrocarbons using a dual-
nozzle laboratory-scale burner. Pure liquid n-octane has been used as a reference fuel, while mixtures involving
also other hydrocarbons commonly found in practical fuels have been investigated to start collecting informa-
tion on surrogate fuels. It has been found that the dual-nozzle configuration allows sustaining mild combustion
conditions by directly injecting different liquid hydrocarbons in a mild combustion environment previously
attained using a gaseous fuel. Similar results for gaseous alkanes (namely, methane and LPG) have been
obtained, while using liquid hydrocarbons (namely, n-octane, n-octane/i-octane and n-octane/i-octane/n-dec-
ane) have led to somewhat different results, even if similar to each other. In particular, the mild combustion
region in the Tavg–KV space enlarges when using liquid hydrocarbons with respect to gaseous ones. However,
in this region very low amounts of NOx, CO, as well as negligible PAH and soot precursors have been found,
therefore supporting the idea that a dual-nozzle mild combustion burner can create a suitable environment for
NOx, PAH and soot depression for a wide range of liquid wastes and low-BTU liquid fuels.
Ó 2010 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Mild combustion; n-Octane; Burner layout; NOx; LPG

1. Introduction tion combined with air preheating generates rela-

tively low maximum flame temperatures since, if
In the last decades, significant efforts have been the combustion air is diluted with a large amount
made to achieve high thermal efficiencies in com- of recirculated exhaust, the oxygen concentration
bustion processes without the adverse high NOx in the reaction zone is strongly reduced with
emissions caused by high temperature flames [1]. respect to the case of conventional burner-stabi-
It was found that a strong exhaust gas recircula- lized combustion allowing for a better control of
the furnace temperature. This combustion tech-
nology is commonly defined as mild combustion
⇑ Corresponding author. Address: Politecnico di [2–6] or flameless oxidation [7].
The essential principles of mild combustion are
Milano, Dipartimento di Chimica, Materiali e Ingegne-
ria Chimica “G. Natta”, Via Mancinelli 7, 20131 Milan, both a high preheating of the combustion air and
Italy. Fax: +39 02 2399 3180. a massive recycle of burnt gases within the furnace.
E-mail address: marco.derudi@polimi.it (M. Derudi). The latter can be done by means of separated or

1540-7489/$ - see front matter Ó 2010 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
doi:10.1016/j.proci.2010.06.120
3326 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332

coflowing high-velocity jets for the injection of fuel key-role in mild combustion furnaces since it
and combustion air into the furnace [6]. These jets allows to obtain turbulence intensity and exhaust
create a low-pressure zone near the burner nozzles recirculation large enough to attain mild combus-
inducing a massive recycle of exhaust into the reac- tion conditions. In this work, a laboratory-scale
tants prior the combustion reactions take place. burner equipped with a single high-velocity jet
The exhaust entrainment raises the temperature nozzle previously developed for mild combustion
of the fresh reactants and increases the inert content of gaseous fuels has been suitably modified and
of the fuel and air mixture; as a result, O2 concentra- used. The original burner characteristics are dis-
tion in the combustion air quickly decreases, lead- cussed in detail elsewhere [3–5,8]; therefore, they
ing to an increase of the characteristic reaction are only briefly summarized in the following.
time that becomes comparable with the characteris- Apart from the burner, the experimental appa-
tic mixing time, which, on the contrary, is lowered ratus also involves an air/fuel flow-rate control sec-
by the high turbulence generated by the high-veloc- tion and a sampling and measurement section for
ity reactants jets. temperature and exhaust compositions measure-
This allows disappearing the flame front, that ment. The laboratory-scale atmospheric burner is
means the elimination of the temperature peak a vertical quartz tube made by two sections: the
always present in conventional diffusion flames, air preheating region and the combustion chamber
therefore producing a fuel oxidation spread in the (height = 350 mm, ID = 50 mm); through the
whole volume of the furnace and consequently a closed top of the burner, thermocouples and the
flattening of the temperature profiles inside the fur- gas sampling line enter the combustion chamber.
nace. In other words, mild combustion occurs in a The combustion air is preheated by an electric oven
diffuse reaction zone; as a consequence, the maxi- up to 1600 K, while a separated electric oven
mum temperature rise results everywhere minimal reduces heat losses from the combustion chamber
allowing for a significant reduction of NOx emis- by keeping its wall temperature no more than about
sions produced via the thermal mechanism. More- 150 K below the average temperature continuously
over, a lower maximum temperature than that detected in the combustion chamber by three type B
found in traditional burner offers the opportunity thermocouples (named TC1, TC2 and TC3 in the
to strongly preheat the combustion air without following). Thermocouples TC1 and TC2 move
increasing the NOx emissions, therefore increasing along the combustion chamber 14 mm apart from
the global furnace yield. the combustion chamber axis, while TC3 moves
The sustainability of mild combustion for dif- along the combustion chamber axis.
ferent gaseous fuels such as methane or ethane Air and gaseous fuel are fed through the bot-
[3,7,8], as well as mixtures of gaseous hydrocar- tom of the combustion chamber, as shown in
bons and hydrogen [4,5,9], has been extensively Fig. 1a, using a single-nozzle (SN) configuration
studied; on the other hand, much less information (ID = 3 mm). Since both fuel and primary air
are available concerning the sustainability of mild (together with diluting nitrogen when required,
combustion for liquids [10,11]. inlet A1 + N2 in Fig. 1a) enter the same nozzle,
Consequently, this work has been focused on a partial premix of air and fuel arises before they
the investigation of the sustainability of mild com- enter the combustion chamber. However, no fuel
bustion for liquid hydrocarbons using a dual-noz- oxidation occurs inside the nozzle due to the short
zle laboratory-scale burner. Pure liquid n-octane, residence time (at 1000 K, it is in the range 103 to
with its simple structure, a relatively low molecu- 104 s, depending on the flow rates). Figure 1a
lar weight and well-known properties has been also shows the secondary air inlet (labeled A2 in
used as a reference fuel. However, since a real Fig. 1a) that is used both for firing the burner as
liquid fuel is constituted by a complex mixture well as for changing the internal exhaust recycle
of hydrocarbons, surrogate fuels are commonly as discussed in the following.
used both for the design of more reproducible This SN configuration is not suitable for liquid
experimental tests and for well-controlled funda- fuels since pyrolysis would arise plugging the noz-
mental and kinetic studies [12]. Along this line, zle in spite of the small residence time; moreover,
mixtures involving also other hydrocarbons com- firing the burner using a liquid fuel is quite cum-
monly found in practical fuels (namely, i-octane bersome. This required changes both in the appa-
and n-decane) have been investigated. ratus configuration and in the firing procedure, as
discussed in the following.
The apparatus has been modified by imple-
2. Experimental section menting the double-nozzle (DN) inlet configura-
tion shown in Fig. 1c, where the preheated air
2.1. Experimental equipment enters the combustion chamber through the bot-
tom nozzle, while the liquid fuel is injected as a
The burner geometry (in particular the nozzles well-dispersed and homogeneous spray through
configuration for fuel and air injection) plays a a lateral water-cooled plain jet airblast atomizer
 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332 3327

Fig. 1. Reactants feeding system: (a) SN gas-fuel feed; (b) DN liquid- and gas-fuels feed; (c) DN liquid-fuel feed. A1:
primary air; A2: secondary air.

(fuel feeding pipe ID = 0.4 mm). The two jets changed by feeding either a secondary air stream
interact perpendicularly and mix to each other in around the bottom jet (which would reduce the
a high turbulence region. exhaust entrainment and consequently the KV
Since firing the laboratory-scale burner in the value) or some inert gas together with the combus-
DN configuration with liquid fuels is not easily tion air (which would mimic a further exhaust
done in practice, the burner has been always fired entrainment and consequently increase the KV
with a gaseous fuel in the SN configuration value).
(Fig. 1a). Once mild conditions have been attained Several calculations previously performed with
in the SN configuration, the transition from the a general-purpose code for computational fluid
SN gas-fuel configuration (Fig. 1a) to the DN dynamics [3,9] showed that the maximum KV
liquid-fuel one (Fig. 1c) has been carried out by value arising from the internal exhaust recycle
reducing progressively the gas flow rate fed to without any secondary air or inert gas addition
the bottom nozzle while increasing the liquid flow (R) is equal to about five for all the conditions
rate from the atomizer (Fig. 1b); for this reason, investigated in this work; this value is reached at
during the transition from a SN to a DN configu- about 5.5 cm from the bottom nozzle tip, where
ration two different fuels have been fed contempo- the lateral spray-jet has been located.
rarily to the burner, a gas from the bottom nozzle It is possible to estimate the maximum KV
and a liquid from the lateral injection point, value in the laboratory-scale burner when either
reducing step-by-step the gas to liquid ratio. an external recycle or a secondary air inlet is pres-
Exhaust have been sampled, dried and analyzed ent as:
with two on-line instruments, namely a gas analyzer
Horiba PG-250 (for NOx, O2, CO and CO2 mea- R  S=A ðI=AÞ  ð1 þ RÞ
KV ¼ þ ð1Þ
surement) and a GC-FID Perkin Elmer Clarus 1 þ S=A ð1 þ F =AÞ  ð1 þ S=AÞ
500 (for C1–C6 unburned hydrocarbons measure-
ment); maximum deviations of about 1 ppm and where A is the flow rate of the primary air, S of
2 ppm from the mean value have been estimated the secondary air, I of the inert gas, and F of
for NOx and CO concentrations, respectively. the fuel. As expected, when neither secondary air
C1–C6 hydrocarbons concentration can give a nor inert gas are fed to the laboratory-scale bur-
rough indication of highly toxic pollutants produc- ner, the previous relation leads to KV = R.
tion since they are PAH and soot precursors [2]. The practical identification of mild conditions
boundaries requires the definition of some thresh-
2.2. Mild combustion key parameters old values either for pollutant emissions or for
temperature gradients in the combustion cham-
This experimental apparatus allows for investi- ber. Following previous studies [4,5], clean mild
gating the influence of two of the main operating conditions have been defined as characterized by
parameters for mild conditions achievement, NOx < 30 ppm and CO < 50 ppm in the exhaust.
namely: average combustion chamber tempera-
ture (Tavg) and dilution ratio (KV), defined as the
ratio between the recycled exhaust and the incom- 3. Results and discussion
ing air and fuel flow rates [3,7,9].
While in real-size burners with internal exhaust 3.1. SN vs. DN configuration
recycle the KV value is defined by the burner
geometry and the air and fuel jets flow-rates, in At first the influence of the new DN configura-
the laboratory-scale burner the KV value can be tion and that of the new firing procedure have been
3328 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332

investigated using methane as a reference fuel for ilar behavior has been found when using n-octane,
the SN configuration. Several experiments have even if in this case the temperature increase is
been carried out performing the transition from partly limited by the cold nitrogen used to spray
SN to DN configuration by injecting from the lat- the fuel as well as by the fuel droplets evaporation;
eral nozzle methane (to investigate the influence in this run, the progressive switch from the CH4/n-
of switching the fuel feed from the bottom to the lat- octane = 3/1 (75:25% by vol.) condition to the
eral nozzle), LPG (to investigate the influence of pure liquid-fuel feed has been done in order to
changing the fuel properties from methane to hea- increase the thermal input only of about 25%,
vier alkanes, which perform quite similarly in the resulting in a slight increase in the average furnace
high temperature combustion region, without temperature. However, in spite of the average
involving droplet evaporation), or n-octane (which temperature in the combustion chamber is almost
is a reference fuel commonly used for building sur- the same for LPG and n-octane, the latter shows a
rogate fuels [12]). For the investigated fuels, higher NOx emission; even if local hot spots can
depending on the experimental conditions to be be created by spray combustion hot spots and
realized, maximum flow rates of about 2.5  104, can contribute to increase the NOx emissions,
1.6  104 and 7.3  105 mol/s have been used the measured trend should be mainly ascribed to
for CH4, LPG and n-octane, respectively; the air the NOx formation through the Prompt mecha-
flow rate has been evaluated according to both nism [13] in the region close to the evaporating
the reaction stoichiometry and the required air droplets, where large fuel/air ratio values are
excess. expected. While this mechanism plays a negligible
In all the experiments the firing procedure role for LPG, it is important for n-octane, as con-
summarized in Fig. 1a–c has been used apart from firmed by simulations carried out in a simple ideal
nitrogen injection from the lateral nozzle, which is reactor configuration with a detailed kinetic
required only for liquid fuels atomization; in this mechanism [14].
case the atomizer has been cooled at 333 K and The results summarized in Fig. 2 clearly indi-
flushed with 2 Nl/min of nitrogen to generate a cate the possibility of operating the fuel switch
spray of small and homogeneous fuel droplets to DN configuration when the combustion cham-
with a short penetration distance. Concerning ber is operating in SN mild combustion conditions
the liquid droplets, an SMD of about 35 lm has without creating any significant instability. This
been estimated for the fuels here investigated. means that a SN laboratory-scale mild burner
Due to the experimental apparatus characteristics, can be easily retrofitted to operate in stable mild
the thermal power values investigated changed combustion conditions with both a different noz-
with the fuel from about 0.2 kW for CH4, to zle layout and a direct liquid fuel injection, there-
about 0.3 kW for LPG, up to about 0.4 kW for fore suggesting the possibility to perform the same
liquid fuels. retrofit on real-size mild burners. This would
Figure 2 shows a comparison among SN to extend their operability in terms of energy saving,
DN transitions for the three different fuels. When fuel availability, emissions control and cost
CH4 is used, stable clean mild conditions can be reduction.
achieved without any significant variation in both Once assessed the reliability of the DN config-
temperatures and NOx emissions. Stable clean uration, the influence of Tavg and KV on the mild
mild conditions can be also achieved using LPG, combustion attainment in this configuration has
even if in this case temperatures and NOx progres- been investigated and compared (in a typical Tavg
sively rise when the LPG/CH4 ratio increases due vs. Kv diagram) with literature data of CH4 mild
to the increase of the thermal power input. A sim- combustion in SN configuration [5] (to ensure a

35 1470
CH4 CH4
30
LPG LPG
25 1370
NOx (ppm)

n-octane n-octane
Tavg (K)

20
1270
15
10 1170
5
0 1070
0 20 40 60 80 100 0 20 40 60 80 100
ary ary
a fuel 2 nozzle (% by vol.) b fuel 2 nozzle (% by vol.)

Fig. 2. NOx emissions (a) and average temperature (b) as a function of the lateral/bottom fuel ratio (0% = SN;
100% = DN). Tpre-heater = 1273 K; excess air = 20% by vol.
 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332 3329

direct comparison of the results, all data have ature (Tpre-heater) on the mild combustion sustain-
been reported @ 3%O2). The results of such a ability of n-octane with reference to CO formation
comparison are summarized in Fig. 3, where the (because NOx emissions, in the conditions here
mild combustion region for the SN configuration reported, are always of the order of 10 ppm).
is bounded in the high temperature region by a Figure 4 summarizes CO emissions as a function
zone where mild conditions can be sustained but of Tavg for three air excess values at Tpre-heater =
thermal NOx production becomes significant, 1223 K, while Fig. 5 shows CO emissions vs. air
while a no combustion (or extinction) zone and excess for two KV values at Tpre-heater = 1123 K.
a flame combustion zone bound the mild combus- Data reported in Fig. 4 are obtained adjusting
tion region for Tavg and low KV values, respec- the thermal input of the system by means of a
tively. The experimental results obtained with reduction of the reactants flow rates. We can see
the DN configuration show that stable mild con- that both air excess and pre-heater temperature
ditions can be sustained also outside the mild have a small influence on the CO formation; the
combustion region for the SN configuration, CO threshold value is always reached when the
enlarging the mild operating ranges in terms of temperature decreases significantly or when the
KV, until about 3, while producing only small dilution ratio increases significantly, leading to a
changes about the lower Tavg threshold (at about too low oxygen concentration to ensure a com-
1130 K). It should be noted that the decrease of plete fuel conversion (for both figures, similar
the lower dilution threshold up to about KV = 3 conclusions can be drawn even if the data are
confirms literature experimental data of multi- expressed in terms of CO yield instead of CO
nozzle burners [7]. This confirms that a SN burner concentration).
configuration represents the worst case in terms of Figure 6 reports the influence of KV on the mild
mild combustion feasibility since combustion can combustion emissions, where the possibility of
start before the reactants dilution is achieved [5]. operating with KV values as low as 1.5, which is sig-
It is worth noting that LPG shows practically nificantly lower than the minimum KV value found
the same threshold values as CH4 in spite of its for CH4 and LPG, is clearly identified. In all the
higher reactivity, flame speed and thermal power
input. This means that the DN configuration
allows for the achievement of an optimal reactants 90
and thermal dilution before the combustion reac- 20% air exc.
tion starts, thus reducing thermal peaks within the 75 14% air exc.
combustion chamber and suggesting that the 10% air exc.
60
CO (ppm)

thresholds found in these experiments can be

probably extended to other small gaseous alkanes 45 CO mild clean limit
and their mixtures, such as Natural Gas. 30
During the transition from SN to DN configu-
ration the fuel conversion has been found to be 15
always complete, with no unburnt hydrocarbons 0
and negligible CO emissions. 1100 1125 1150 1175 1200 1225 1250
Tavg (K)
3.2. Liquid n-octane mild combustion
Fig. 4. CO emissions as a function of air excess and Tavg
Several tests have been dedicated to investigate in DN n-octane mild combustion; Tpre-heater = 1223 K,
the influence of air excess and pre-heater temper- KV = 7.5.

800
Kv=5.7
700 Kv=2.7
600
CO (ppm)

500
400
300
200
100 CO mild clean limit
0
Fig. 3. Experimental results for CH4 (D) and LPG (h) 0 4 8 12 16 20
with DN configuration; solid line represents the mild Air excess (% by vol.)
combustion region boundary for CH4 with SN config-
uration [5]. Empty and full symbols represent mild Fig. 5. CO emissions as a function of air excess and KV
combustion and non-mild conditions, respectively. in DN n-octane mild combustion; Tpre-heater = 1123 K.
3330 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332

50 amount, contributing to the Thermal-NOx forma-

NOx @ 1373 K
CO @ 1373 K tion, can slightly reduce the upper average tem-
Emissions (ppm)

40 NOx @ 1173 K perature limit of the clean mild combustion

NOx mild clean limit CO @ 1173 K region.
30
The difference between gaseous and liquid fuels
20 in the upper temperature threshold cannot be
explained by the presence of the spray-jet since
10 the main structure of the flow field in both cases
is established by the air jet on the bottom of the
0
0 1 2 3 4 5 6 7
combustion chamber. However, the presence of
fuel droplets that must evaporate before burning
Kv
creates a region with a large fuel/air ratio where
Fig. 6. Emissions trends for DN n-octane mild combus- prompt-NOx mechanism could play a significant
tion with different Tpre-heater (Air excess = 14% by vol). role and explain this difference. In fact, while ther-
mal-NOx formation rate strongly depends on tem-
perature and only to a less extent by oxygen
mild combustion DN runs with n-octane, NOx concentration [4], prompt-NOx are formed by
emissions equal to about 10–15 ppm have been the reaction of nitrogen with hydrocarbon radi-
found together with very low CO concentrations cals in the fuel rich regions at temperature below
(small CO fluctuations, in the range of the 1773 K [13]. As shown in Fig. 8, even if a qualita-
experimental error, have been sometimes detected, tive relationship can be found between air pre-
but were always very low compared with the heater temperature and NOx emissions underlying
corresponding threshold limit value, that is the role of the thermal-NOx route, at such moder-
50 ppm) and practically no emissions of unburnt ate temperature values the NOx concentration val-
hydrocarbons. ues are too high to be explained only by the
The mild operating map for n-octane is shown thermal-NOx route, accordingly with the previous
in Fig. 7, where the empty symbols indicate clean discussion of the results reported in Fig. 3 for
mild conditions, while full squares and full trian- LPG and n-octane mild combustion.
gles represent conditions that are above the NOx Almost no differences have been found
and CO threshold limits, respectively. The clean between gaseous and liquid fuels for the lower
mild combustion region is bounded by a lower average temperature threshold (see Figs. 3 and
average temperature value of about 1110 K, a 7), since this limit is due to the CO formation
lower KV value equal to about 1.5, and an upper and it is scarcely influenced by the fuel composi-
average temperature value of about 1290 K. This tion being controlled by the combustion chamber
upper value is due to the NOx formation and it temperature and its gradients along the burner.
is quite lower than the corresponding threshold A somewhat different consideration can be
found for gaseous fuels (see Fig. 3). An increase done for the lower KV threshold that, for both
of the excess air ratio, as shown in Figs. 4 and gas and liquid alkanes, is determined by a gradual
5, can have a slight positive effect on lower bound- increase of the NOx emissions (see Fig. 6) without
ary limits, enhancing the clean mild combustion any increase of the average combustion chamber
stability at low KV and low average temperatures temperatures. The reason is that a decrease of
values; on the other hand, a too large excess air KV involves a lower air/fuel dilution with the
exhaust, thus leading to a local temperature
increase in the lower part of the combustion
1330
chamber, as confirmed by measurements carried
1280
Tavg (K)

1230 20 1290
NOx @3%O2 (ppm)

1180 1265
15
Tavg (K)

1240
1130
10 1215
1080 1190
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 5
1165
Kv
0 1140
650 850 1050 1250
Fig. 7. Mild combustion region for n-octane in DN
configuration (Air excess = 10% by vol.). Empty sym- Pre-heater Temperature (K)
bols: combustion conditions inside the mild combustion
region; Full squares and full triangles represent condi- Fig. 8. NOx emissions and average temperature as a
tions outside the NOx and CO threshold limits, function of the Tpre-heater for n-octane mild combustion;
respectively. KV = 5.7.
 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332 3331

1325 1325
TC1 TC1
1275 TC2 1275 TC2

Temperature (K)
Temperature (K)

TC3 TC3
1225 T avg 1225 Tavg

1175 1175

1125 1125

1075 1075
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350

a Axial position (mm) b Axial position (mm)

Fig. 9. Temperature profiles measured along the combustion chamber at KV = 5.7 (a) and 2.7 (b) for n-octane mild
combustion in DN configuration (lines with symbols) and average temperatures (dashed lines); Tpre-heater = 1173 K.

out along the combustion chamber. Such mea-

surements (which are affected by an uncertainty 14
lower than 2%) are summarized in Fig. 9 for 12

NOx @3%O2 (ppm)

two different KV values (similar trends have been 10
observed for all the investigated KV values). We 8
can see that, as expected, reducing the KV value n-c8
6
leads to larger maximum temperature values while n-c8/i-c8 95/5
keeping almost the same average temperature 4
value. 2 n-c8/i-c8 60/40
However, this threshold value has been found 0
lower for n-octane than for gaseous alkanes. This 0 20 40 60 80 100
difference can be explained considering that the liquid fuel (%)
overall combustion rate of a liquid hydrocarbon
can be significantly reduced by the droplets evap- Fig. 10. NOx emissions as a function of the lateral/
oration rate, therefore reducing the Damkholer bottom fuel ratio (0% = SN; 100% = DN). Tpre-heater =
number (Da = tm/tc), which is the ratio between 1173 K.
a characteristic mixing time, (tm) and a character-
istic combustion time (tc) [4]. Mild combustion
conditions, which require the Damkholer number the surrogate fuels approach [12]), various liquid
to be low enough, are usually attained by increas- mixtures have been tested in the DN configuration.
ing the jet velocity (which reduces the mixing As surrogate liquid fuels commonly contain
time) and lowering the overall combustion rate linear and branched alkanes, the first mixture
by reducing the oxygen concentration trough tested involves n-octane and i-octane. As shown,
exhaust recycle (which increases the combustion for the sake of example, for NOx emissions in
time). In the case of liquid n-octane combustion, Fig. 10, the presence of the branched isomer does
the overall combustion rate is reduced also by
the droplet evaporation kinetics, resulting in the
1330
possibility of sustaining mild combustion condi-
tions even with larger oxygen concentrations, that 1280
is, with lower KV values. 1230
Tavg (K)

This could mean that the injection of a liquid

hydrocarbon in the DN configuration can sustain 1180
mild combustion of liquid wastes and/or low- 1130
BTU liquid fuels at low costs and with low
pollutant emissions. In this respect, it is worth 1080
remembering that no emissions of unburnt hydro- 1030
carbons (that can be roughly considered as the 0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5
precursors of PAH and soot) have been found in Kv
all the experiments.
Fig. 11. Experimental results for a liquid mixture
constituted by n-octane/i-octane/n-decane (36/24/40 by
3.3. Liquid hydrocarbons mild combustion
vol.) with DN configuration; dashed line: mild combus-
tion region boundary for liquid n-octane. Empty sym-
To verify the possibility to obtain a stable mild bols: combustion conditions inside the mild combustion
combustion also for liquid hydrocarbons mixtures region. Full squares and full triangles represent condi-
more similar to commercial liquid fuels (following tions outside NOx and CO threshold limits, respectively.
3332 M. Derudi, R. Rota / Proceedings of the Combustion Institute 33 (2011) 3325–3332

not affect significantly mild combustion condi- ing the same class (that is, alkanes). In this region
tions with respect to the aforementioned n-octane very low amounts of NOx, CO, as well as PAH
results. and soot precursors are produced.
Moving a further step towards real fuels, a These findings support the idea that a DN mild
liquid mixture constituted by n-octane, i-octane combustion burner can create a suitable environ-
and n-dodecane (36/24/40 by vol.) has been tested. ment for NOx, PAH and soot depression, allow-
Also this mixture shows a similar behavior, being ing the use of a wide range of liquid wastes and
its transition from SN to DN configuration char- low-BTU liquid fuels, even with unsteady compo-
acterized by NOx emissions lower than 30 ppm sition or coming from different sources.
and negligible CO emissions. As shown in
Fig. 11, the results obtained using this mixture
fully compare with the previously identified mild References
operating map of n-octane. This means that, as
found for the mixtures constituted by octane iso- [1] G.M. Choi, M. Katsuki, Proc. Combust. Inst. 29
mers, in mild conditions the considered liquid (2002) 1165–1171.
hydrocarbons show a similar reactivity. [2] A. Cavaliere, M. de Joannon, Prog. Energy Com-
bust. Sci. 30 (2004) 329–366.
[3] A. Cavigiolo, M.A. Galbiati, A. Effuggi, D. Gelosa,
4. Conclusions R. Rota, Combust. Sci. Technol. 175 (2003) 1347–
1367.
[4] M. Derudi, A. Villani, R. Rota, Proc. Combust.
In this work the sustainability of mild combus-
Inst. 31 (2007) 3393–3400.
tion for liquid hydrocarbons has been investigated [5] M. Derudi, A. Villani, R. Rota, Ind. Eng. Chem.
using a DN laboratory-scale burner and methane Res. 46 (2007) 6806–6811.
as a reference fuel for the SN configuration. [6] G.G. Szego, B.B. Dally, G.J. Nathan, Combust.
It has been shown that the DN configuration Flame 156 (2009) 429–438.
allows to sustain mild combustion conditions [7] J.A. Wünning, J.G. Wünning, Prog. Energy Com-
by directly injecting different liquid hydrocarbons bust. Sci. 23 (1997) 81–94.
in a mild combustion environment previously [8] A. Effuggi, D. Gelosa, M. Derudi, R. Rota,
attained using a gaseous fuel. Combust. Sci. Technol. 180 (2008) 481–493.
[9] C. Galletti, A. Parente, M. Derudi, R. Rota, L.
DN configuration provides similar results for
Tognotti, Int. J. Hydr. En. 34 (2009) 8339–8351.
gaseous alkanes (namely, methane and LPG), [10] R. Weber, J.P. Smart, W. Kamp, Proc. Combust.
while using liquid hydrocarbons (namely, Inst. 30 (2005) 2623–2629.
n-octane, n-octane/i-octane and n-octane/i-octane/ [11] S.R. Wu, W.C. Chang, J. Chiao, Fuel 86 (2007)
n-decane) leads to somewhat different results, even 820–828.
if they are similar to each other. This means that [12] E. Ranzi, Energy Fuels 20 (2006) 1024–1032.
the mild combustion characteristics seem more [13] C.P. Fenimore, Proc. Combust. Instit. 13 (1971)
influenced by the physical state of the fuel (that 373–380.
is, gas or liquid) than that by the chain length of [14] E. Ranzi, A. Frassoldati, S. Granata, T. Faravelli,
Ind. Eng. Chem. Res. 44 (2005) 5170–5183. Avail-
the hydrocarbon (that is, C1–C4 or C8–C10).
able at: <http://www.chem.polimi.it/CRECKMod-
In particular, the mild combustion region in eling/kinetic.htm>.
the Tavg–KV space enlarges when using liquid
hydrocarbons with respect to gaseous ones shar-