Applied Energy 111 (2013) 401–415

Contents lists available at SciVerse ScienceDirect

Applied Energy
journal homepage: [Link]/locate/apenergy

Experimental and numerical investigations of an atmospheric diffusion

oxy-combustion flame in a gas turbine model combustor
Medhat A. Nemitallah a,b,c,⇑, Mohamed A. Habib a
a
KACST TIC #32-753, KACST and Mechanical Engineering Department, Faculty of Engineering, KFUPM, Dhahran 31261, Saudi Arabia
b
Mechanical Engineering Department, Alexandria University, Egypt
c
Mechanical Engineering Department, Massachusetts Institute of Technology, USA

h i g h l i g h t s

 Experimental and numerical study of oxy-combustion flame in gas turbine combustor.

 Modified 2-step methane oxycombustion reaction kinetics model is introduced.
 Wide range of operating parameters was considered experimentally and numerically.
 Oxycombustion and emission characterization and flame stabilization are studied.
 Validation of the oxycombustion reaction kinetics model is also presented.

a r t i c l e i n f o a b s t r a c t

Article history: An atmospheric diffusion oxy-combustion flame in a gas turbine model combustor has been investigated
Received 15 December 2012 experimentally and numerically. Oxy-combustion and emission characterization, flame stabilization and
Received in revised form 4 May 2013 oxy-combustion model validation analyses are the main goals of the present research work. The combus-
Accepted 12 May 2013
tor is fuelled with CH4 and a mixture of CO2 and O2 as oxidizer. A modified two-step oxy-combustion
Available online 4 June 2013
reaction kinetics model for methane–oxygen combustion has been used in order to predict accurately
the oxy-combustion characteristics. The conducted experimental results were used to validate the
Keywords:
numerical model. Wide ranges of different operating parameters have been considered including equiv-
Emission characteristics
Flame stabilization
alence ratio, percentage of O2/CO2 in the oxidizer mixture and fuel volume flow rate. The stability of the
Gas turbine model combustor oxy-combustion diffusion flame is also investigated both experimentally and numerically. The experi-
Oxy-combustion model mental and numerical results showed that the stability of the oxy-combustion flame is affected when
Swirl stabilized flames the operating percentage of oxygen in the oxidizer mixture is reduced below 25%. In all cases, flame
was extinct for conditions of less than 21% oxygen in the oxidizer mixture. Flame visualization over a
wide range of operating parameters has been carried out experimentally and comparisons with the
numerical results have been conducted. The flames have been characterized in detail by measuring the
exhaust gas temperatures and emissions and comparing them with those from the numerical model.
The combustion was found to be improved with increasing the percentage of O2 at inlet however there
is a limitation in temperature. Both experimental and numerical results are in good agreement. The mod-
ified two step reaction kinetics model was found to be capable of capturing the trends of temperature and
the overall flame shape of the experimental data. Flame zone is also characterized in details by plotting
the axial and radial temperatures, species concentrations and flow velocities using the numerical model.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction dioxide (CO2). Strong evidence suggests that both the average
global temperature and the atmospheric CO2 concentration have
Global climate change is one of the greatest challenges in the significantly increased since the onset of the industrial evolution,
present century. The greenhouse gas making the largest contribu- and they are well correlated [1]. CO2 emissions from the fossil
tion to global climate change from human activities is carbon fuel-based large power plants are of main concern as they are
the largest sources of CO2 in the coming decades [2]. Concerns over
⇑ Corresponding author at: KACST TIC #32-753, KACST and Mechanical climate change have led to mounting efforts on developing tech-
Engineering Department, Faculty of Engineering, KFUPM, Dhahran 31261, Saudi nologies to reduce carbon dioxide emissions from human activities
Arabia. [3,4]. Technological solutions to this problem ought to include a
E-mail addresses: medhatahmed@[Link], medhat.ahmed35@[Link], substantial improvement in energy conversion and utilization
mahmed@[Link] (M.A. Nemitallah).

0306-2619/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
[Link]
402 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

Nomenclature

CFD computational fluid dynamics CCS carbon capture and storage

3D three dimensional j absorption coefficient
Rl mass rate of creation or depletion RTE radiative transfer equation
uj velocity component U equivalence ratio
UU diffusion coefficient q fluid density
cl constant ebk blackbody spectral emissive power
lt turbulent viscosity rU constant
rs scattering coefficient rV oxygen vacancy conductivity
e epsilon

efficiencies, carbon capture and storage (CCS), and expanding the should have minimal oxygen level of 24% under the oxidizer tem-
use of nuclear energy and renewable sources such as biomass, hy- perature 520 K condition. Kutne et al. [9] have checked the stability
dro-, solar, wind and geothermal energy [3]. Carbon capture from of a swirl stabilized oxyfuel/CH4 flames for O2/CO2 percentages of
large point sources such as power plants gives the possibility of a 20/80–40/60%, equivalence ratios of 0.5–1 and thermal powers of
relatively quick response to climate changes with a reasonable 10–30 kW. They reported that attempts of operating the burner
low cost. Several approaches have been evaluated and reviewed with less than 22% O2 were unsuccessful even at stoichiometric
for capturing CO2 in the utility industry, namely carbon capture conditions. Syngas and methane flames for premixed swirl stabi-
and storage technology (CCS), including pre-combustion capture, lized conditions for two different oxidizers of air and O2/CO2/N2
oxy-fuel combustion, and post-combustion capture [5]. have been studied by Williams et al. [10]. They report lower nitro-
As a promising CCS technology, oxyfuel combustion can be used gen oxides concentrations (NOx) for the quasi-oxyfuel flames and
for existing power plants and also for new ones. Oxyfuel combus- higher carbon monoxide concentrations (CO), suggesting stoichi-
tion gives different characteristics of heat transfer, ignition, burn- ometric operation at 20–24% O2 as ideal for low emissions. Dita-
out as well as NOx emission [6]. During oxyfuel combustion ranto and Hals [11] studied the influence of stoichiometric
process, a mixture of oxygen and recycled flue gases are used for operation and high Oxygen content in the oxidizer mixture on
the combustion of the fuel. The exhaust gases consist of a mixture thermo-acoustic oscillations in sudden expansion jet configura-
of CO2 and H2O. A large portion of the flue gases should be recycled tion. They reported occurrence of thermo-acoustic instabilities as
in order to substitute the removed nitrogen to ensure there is en- O2 content in the oxidizer was increased, characterizing different
ough gas to carry the heat through the boiler and in order to con- instability modes dependent on flow velocity and flame speed vari-
trol the flame temperature. Another way of particular interest is ations. Anderson and Johnsson [12] have performed experiments
the use of oxyfuel combustion in gas turbines. For oxy-fuel gas tur- on a 100 kW test unit for air and two O2/CO2 test cases with differ-
bine cycles, researches have been focused on thermodynamic stud- ent recycled feed gas mixture concentrations of O2 (OF 21 at
ies of system performance. The combustion behavior, e.g. the flame 21 vol.% O2, 79 vol.% CO2 and OF 27 at 27 vol.% O2, 73 vol.% CO2).
dynamics and reaction zone structures in the gas turbine combus- They showed that the fuel burnout is delayed for the OF 21 case
tors, is less addressed. Swirl stabilized flames are used extensively compared to air-fired conditions as a consequence of reduced tem-
in practical combustion systems because they enable high energy perature levels. Instead, the OF 27 case results in more similar
conversion in a small volume and exhibit good ignition and stabil- combustion behavior as compared to the reference conditions in
ization behavior over a wide operating range. In order to reduce terms of in-flame temperature and gas concentration levels, but
the flame temperature and thereby the remaining oxygen in the with significantly increased flame radiation intensity. A detailed
flue gas, it is beneficial to premix the oxygen and CO2 or steam be- literature about the application of oxy-fuel combustion technology
fore introducing them to the combustor. Also, to generate a stable into gas turbines is there in our recent review work [5].
combustion in a gas turbine combustor through oxy-fuel combus- In the present research work, experimental and numerical
tion, certain minimal oxygen level in the oxidizer has to be main- investigations of an atmospheric diffusion oxy-combustion flame
tained. This is because of the need to have the required high in a gas turbine model combustor are presented. Oxy-combustion
temperature inside the reaction zones for the chain reactions to and emission characterization, flame stabilization and oxy-com-
proceed. Heil et al. [7] reported that poor burnout and lifted dark bustion model validation analyses are the main goals of the pres-
flames appeared when the oxygen mole fraction in the O2/CO2 ent work. The combustor was supplied with CH4 and CO2 and O2
stream was set to 21%; when the oxygen volume fraction was in- as oxidizer mixture. Large range of operating parameters was con-
creased to 27% and 34%, full burnout and stable flames were ob- sidered including equivalence ratio, percentage of O2/CO2 in the
tained. In order to burn the fuel with lower oxygen level in the oxidizer mixture, and fuel volume flow rate. The goals for the pres-
oxidizer (O2/CO2) stream the burner had to be modified to allow ent work is to check the stability of the oxy-combustion diffusion
for recirculation of hot gases to the flame. With this high level of flame experimentally and numerically and also to check the mini-
oxygen in the oxidizer mixture the combustion products however mum permissible percent of O2 in the oxidizer mixture required in
become hot and this may lead to high level of oxygen in the flue order to get a stable flame. Visualizations of the flame at the above
gas due to the dissociation reactions that occur at high tempera- mentioned operating conditions were conducted. Different flames
tures. There is an optimal ‘‘window’’ of oxygen/diluent ratio in were characterized by measuring the exhaust gas temperatures
the oxidizer stream [8]. and comparing them with the numerical model results. A modified
Stable combustion and low turbine inlet temperature can be ob- two-step oxy-combustion reaction kinetics model for methane–
tained simultaneously by optimizing the ratio of oxygen to CO2 oxygen combustion has been used in order to predict a clearer
supplies in the oxidizer mixture to the combustion chamber. Liu oxy-combustion characteristics and then more accurate numerical
et al. [8] reported that the primary oxidizer which is supplied in results in order to correctly validate the numerical model using the
the upstream through the dome of the combustion chamber experimental results.
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 403

2. Experimental and numerical setup burner can provide power ranging from 4.1 MW/m3 bar at 6 L/
min fuel volume flow rate up to 6.2 MW/m3 bar at 10 L/min fuel
2.1. Experimental setup volume flow rate. These conditions are comparable to most indus-
trial gas turbines, which are operated at 3.5–20 MW/m3 bar. In or-
2.1.1. Reactor design der to shorten and stabilize the flame, a swirler with a swirl angle of
Fig. 1 shows a schematic drawing of the swirl stabilized gas tur- 60° is used to apply the required swirl for air. The swirler is located
bine model combustor. The non-swirling fuel (CH4) is supplied below the oxidizer inlet to the reactor inside a tube of 25.4 mm in-
through 16 channels made on the round surface of a bluff body with side diameter as shown in Fig. 1. In order to generate high flow
diameter of 5 mm. The bluff body is inserted at the top part of the velocity at the exit section and to prevent the back flow from the
fuel tube (1/400 outside diameter) in order to generate the required surrounding to the reactor, a conically-shaped top steel part with
velocity to stabilize the flame. Each fuel channel has dimensions of a minimum diameter at its top of 50 mm and a length of 100 mm
0.45 mm  0.2 mm. The oxidizer (O2/CO2 mixture at room temper- is used to form the exhaust gas exit. This conical exhaust part is sup-
ature) is supplied to the combustion chamber through an annular ported with holes to enable measuring both the inner reactor tem-
space between the fuel inlet pipe of 6.35 mm diameter and an out- perature and species concentrations.
side diffuser which has inlet diameter, outlet diameters and length
of 10 mm, 20 mm, and 5 mm respectively. The exit plane of the fuel 2.1.2. Experimental facility
channels is assigned to be the reference zero height of the flame Fig. 2 shows a schematic of the experimental test-rig facility
(h = 0 mm) and the exit plane of the oxidizer is located at 5 mm be- used in the present work. The oxygen and carbon dioxide oxidizer
low the fuel exit plane. The overall flow field of the flame is charac- is premixed before being injected into the combustor to reduce the
terized by this conically shaped inflow of fresh gas. An outer flame temperature and thereby the remaining oxygen in the flue
recirculation zone has been created in the lower part of the reactor gas. The oxygen is mixed with carbon dioxide upstream of the
between the flame and the wall of the reactor and as a result, the combustor in a mixing chamber which has a cylindrical shape of
mixing of hot combustion products with fresh gas leads to a contin- 150 mm diameter, 300 mm length, and an internal throat of
uous ignition and stabilization of the flame. The burner is mounted 50 mm diameter. The mixing chamber has three inlets (for O2 plus
in an optically accessible combustion chamber which consists of a CO2 for oxy-combustion case and air for the air–fuel case) and one
quartz tube of 300 mm height and 76.2 mm inside diameter. The outlet. To improve the mixing of O2 and CO2, the O2/CO2 mixture
(feed gas) is forced to pass through a supply pipe of 50.8 mm inside
diameter upstream of the reactor and with a height of 1.2 m before
the inlet to the combustor at the top part as shown in Fig. 2. Meth-
ane is supplied through a 6.4 mm (1/4 in.) tube extended from the
cylinder to the reactor inlet and a bluff body containing slits on its
surface forming the fuel channels to generate the required velocity
of the fuel flow. All gases are supplied from compressed gas cylin-
ders and the flow is controlled through pressure regulators and
valves. To measure the flow rate, a flow rotameter is used in each
gas line and the mass flow rate is controlled by a mass flow con-
troller from Bronkhorest HIGH-TECH (D-type, model number F-
201CI-10K-AGD-22-V for CH4, F202AI-M20-AAD-55-V for O2 and
F-202AI-M20-AGD-55-V for CO2, with uncertainty of ±0.5% of full
scale). The controllers are connected to a computer for data logging
and post-processing.
The exhaust gas temperature is measured at the exit plan of the
Fig. 1. Schematic of the atmospheric pressure swirl stabilized gas turbine reactor (height of 300 mm) with R-type thermocouple from Omega
combustor.
which has a maximum range of 1500 °C and inserted inside a 12 in.

Fig. 2. Schematic of the test-rig facility used for the present study.
404 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

long ceramic probe to protect it from high flame temperature. The burned with a light blue color. The flames also appeared with a
thermocouple probe is connected to a data logger and a computer conically shaped flame zone at all operating conditions. The width
which takes the average temperature over 1 min at each location. of the flame is increased through the combustor height to com-
Similar local temperature measurements are repeated and average pletely close the quartz reactor tube at a certain height. The flames
values are calculated at each location. The thermocouple probe en- are stabilized in this design by two recirculation zones. One recir-
ters the reactor from its top as shown in Fig. 2 and a traverse mech- culation zone is inside the flame itself in the vicinity of the bluff
anism is used to move the probe precisely in the axial and radial body at the end of the fuel pipe which accelerates the fuel and
directions inside the reactor. For higher flame temperatures, a B- cause a recirculation zone inside the flame. The second recircula-
type thermocouple from Omega is used with the ability to measure tion zone is created due to the existence of the conically shaped
the temperature up to 1950 °C. A high speed camera is used to oxidizer inlet which forms a recirculation zone near to the reactor
visualize and compare different flames. The camera is fixed at wall. Flame visualization shows a good radial symmetry. A simpli-
the same level of the reactor. fied equation is used to calculate the swirl number as given in Eq.
Gas composition data consisting of O2, CO, CO2, dissolved N2 (1). The equation was simplified to eliminate the need for static
and total unburned hydrocarbon (HC) concentrations were mea- pressure measurement and the final form is based on the velocity
sured at the exit section of the combustor (h = 300 mm) using a at the reactor inlet just above the nozzle exit.
gas chromatography, model 450-G.C. The sample is supplied to R Ro
the gas chromatograph through a suction pump after quenching 0
V z Wr2 dr
S¼ R Ro 2 ð1Þ
it at the exit section using a specially designed probe to prevent Ro 0 V z rdr
reactions and protect the pump blades. The probe consists of three
where Vz is the axial velocity, W is the tangential velocity, and Ro is
tubes, the sample is sucked through the inner tube (dia. of 1/800 )
the radius of at the nozzle exit. The swirl number is calculated for
and the cooling water is supplied in the annular area between
the investigated cases of the present work and is found to have a
the outer pipe (dia. of 0.5 in.) and the middle tube (dia. of
range of 0.5–0.6. The nozzle is contoured and a combustion cham-
0.25 in.) and leaves from the annulus between the inner and the
ber is used with an expansion factor (D/d) of 3.5 where D is the in-
middle tubes. The flow of water is controlled in order to prevent
ner diameter of combustion chamber and d is the diameter of the
the condensation inside the probe. The exit pipe of the pump is
nozzle at the reactor inlet. Both inner and outer recirculation zones
long and has a sudden drop in diameter to complete the cooling
are expected to break down the vortex motion. The Reynolds num-
process to separate H2O before injecting the sample as dry to the
ber is calculated for each case based on the cold inflow conditions
GC. The total hydrocarbon concentration is measured as methane
and a nozzle diameter of 10 mm. The mass flow rate is calculated
equivalents by means of an FID (flame ionization detector) based
at the standard temperature and pressure while the viscosity of
instrument. The analysis of O2, CO, N2 and CO2 were each per-
the mixture is calculated using the correlation given by Warnatz
formed by a Thermal Conductivity Detector (TCD) which is the
et al. [13].
most universal detector available with different measurement
ranges, adapted for reference and O2/CO2 conditions, respectively.
2.2. Numerical modeling
The gas is dried and filtered before it is fed into the on-line gas ana-
lyzer train.
2.2.1. Conservation equations
The mathematical model is based on the numerical solution of
2.1.3. Operating conditions
the conservation equations for mass, momentum, and energy,
Two sets of experiments are conducted with the first experi-
and transport equations for scalar variables. The equations, which
ment focusing on determining flame stability for equivalence ra-
are elliptic and three-dimensional, were solved to provide predic-
tios ranging from 0.5 to 1, fuel volume flow rates from 5 to 10 L/
tions of the flow pattern, thermal and emissions characteristics
min and at different O2 concentrations in the oxidizer ranging from
of reacting medium inside a gas turbine model combustor. The
100% O2 (0% CO2) to the percent of O2 that leads to flame blow out
equations which govern the conservation of mass, momentum
condition. The considered blends of the oxidizer mixture of O2/CO2
and energy as well as the equations for species transport are found
are 100/0, 80/20, 60/40, 50/50, 40/60, 30/70 and 25/75. The condi-
there in the work done by Chen and Ghoniem [17]:
tions for the flame blow out are determined by reducing the per-
 
centage of O2 from 25% by a step of 1% until blow out condition @   @ @U
q U j U þ q uj / ¼ C/ þ qSU ð2Þ
is reached. A high speed camera is used to visualize the size, shape @xj @xj @xj
and position of the heat release zone. The second set of experi-
ments involves measurements to characterize the flames. This is where U and / are the Reynolds’s averaged and fluctuating values
achieved by measuring the main species concentrations and the of the dependent variable and uj is the velocity component along
exhaust gas temperature at the exhaust section of the combustor. the coordinate direction xj, q
 is the fluid density, UU is the diffusion
As well, axial profiles of gas temperature and gas species concen- coefficient and SU is the source term. Eq. (2) stands for the mass
trations were measured along the flame from the bottom of the conservation equation when U = 1; the momentum conservation
reactor to the exhaust section. These measured data can serve as equation when U is a velocity component; the energy equation
a database for combustion models validation. The inside reactor when U is the stagnation enthalpy; or the transport equation of a
temperature is measured in the radial direction at radii of 0, 10, scalar when U is a scalar variable such as mixture fraction.
20, 30 mm and at heights of 0–10 cm by a step of 1 cm, from The velocity and the pressure fields of the gas mixture are ob-
10 cm to 20 cm with a step of 2.5 cm, and from 20 cm to 30 cm tained from the numerical solution of the mass and momentum
by a step of 5 cm. The fuel volume flow rate is changed for each conservation equations, while the species concentrations distribu-
equivalence ratio from 5 L/min to 10 L/min with a step of 1 L/ tions are obtained from a scalar convection–diffusion equation.
min. The different percentages of O2/CO2 were determined at each The steady-state mass, momentum, energy and species conserva-
value of fuel volume flow rate and at a certain equivalence ratio. All tion equations for Newtonian fluids were considered:
gases are introduced into the combustion chamber at atmospheric r  ðqUÞ ¼ 0 ð3Þ
pressure and at a temperature of 22 °C.
The flames obtained in the present work are established under
r  ðqUUÞ ¼ rP þ lr2 U ð4Þ
globally lean conditions which showed no soot production and
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 405

ðqC P ÞU  rT ¼ r  ðkrTÞ ð5Þ 2.2.3. Radiation model

The solution of the radiative transfer equation (RTE) was ob-
r  ðqUY i Þ  r  ðqDi;m rY i Þ ¼ 0 ð6Þ tained using the discrete ordinates (DO) radiation model. The
RTE equation is written as:
where u is the velocity vector, q is the fluid density, p is the pres-
sure, l is the dynamic viscosity, k is the thermal conductivity, Di,m dIðr; sÞ
¼ jIb  ðj þ rs ÞIðr; sÞ ð14Þ
is the diffusion coefficient and Yi is the scalar species mass fraction. ds
The total radiation intensity I, depends on the position vector, r,
2.2.2. Turbulence model and the path length s. j is the absorption coefficient and rs is the
The present work utilizes the k–e model of Versteeg and Mala- scattering coefficient. The Planck mean absorption coefficient is gi-
lasekera [15]. The Reynolds stresses and turbulent scalar fluxes are ven as:
related to the gradients of the mean velocities and scalar variable, R
jk ðk; T; PÞebk ðk; TÞdk
respectively, via exchange coefficients as follows: jðT; PÞ ¼ ð15Þ
! rT 4
@U i @U j 2 where ebk is the blackbody spectral emissive power. The blackbody
qui uj ¼ lt ¼  qkdij ð7Þ
@xj @xi 3 spectral emissive power is calculated using variables by Zheng et al.
[18].
@U
quj / ¼ CU ð8Þ 2.2.4. Species transport equations
@xj
The mass fraction of each species ml, is predicted through the
where lt is the turbulent viscosity and UU is equal to lt/rU. The tur- solution of a convection–diffusion equation for the lth species.
bulent viscosity is modeled as: The present calculations utilize a modified two-step methane-oxy-
combustion reaction kinetics model as described below which
lt ¼ cl qk2 =e ð9Þ solves the transport equations for the species of O2, CO2, CO and
H2O. The conservation equations can be expressed in the following
where cl and r/ are constants. The turbulent viscosity is thus ob-
form:
tained from the solution of the transport equations for k and e.
RNG (Renormalized group) turbulence model [16] was used to pro- @   @
qU i ml ¼  Jl;i þ Rl ð16Þ
vide better results for vertical flows. The eddy dissipation model @xi @xi
that described turbulence–chemistry interaction in non-premixed
where Rl is the mass rate of creation or depletion by chemical reac-
combustion was utilized in the present work to provide the produc-
tion of the species l, Jl,i is the diffusion flux of species l, which arises
tion rate of species. The conservation equations of the kinetic en-
due to concentration gradients which is given by:
ergy of turbulence and the rate of dissipation of the kinetic  
energy of turbulence are [17]: lt @ml
J l;i ¼  qDl;m þ ð17Þ
 Sct
leff @k 
@xi
@ @
ðqU j kÞ ¼ þ Gk  qe ð10Þ where Dl,m is the diffusion coefficient for species l in the mixture
@xj @xj rk @xi
and Sct is the turbulent Schmidt number, qlDt t is equal to 0.7.

@   @ leff @ e  e e2 An eddy-dissipation model [19] that relates the rate of reaction
qU j e ¼ þ C 1 Gk  C 2 q ð11Þ to the rate of dissipation of the reactant- and product-containing
@xj @xi re @xi k k
eddies is used to calculate the rate of reaction. The source of chem-
where Gk represents the generation of turbulent kinetic energy due ical species i due to reaction, Ri ; is computed as the sum of the reac-
to the mean velocity gradients and is given by tion sources over the NR reactions, thus:

@U j X
NR

Gk ¼ qui uj ð12Þ Ri ¼ M i Ri;k ð18Þ

@xi k¼1

The quantities rk and re are the effective Prandtl numbers for k where M i is the molecular weight of species i and Ri;k is the molar
and e, respectively and C 2 is given as: rate of creation/destruction of species i in reaction k. The reaction
rate, Ri;k is controlled either by an Arrhenius kinetic rate expression
C 2 ¼ C 2 þ C 3 ð13Þ
or by the mixing of the turbulent eddies containing fluctuating spe-
where C3 is a function of the term k=e and, therefore, the model is cies concentrations. The rate of reaction Ri;k is given by the smaller
responsive to the effects of rapid strain and streamline curvature value of the two expressions below:
and is suitable for the present calculations. The model constants e mR
C1 and C2 have the values; C1 = 1.42 and C2 = 1.68. The wall func- Ri;k ¼ mi;k M i Aq ð19Þ
kmR;k MR
tions establish the link between the field variables at the near-wall
P
cells and the corresponding quantities at the wall. A standard wall e p mP
function was used for the near wall treatment. The turbulent kinetic Ri;k ¼ mi;k M i ABq PN ð20Þ
k j mj;k Mj
energy Prandtl number was set to one.

Table 1
Modified two-step methane–oxygen combustion mechanisms with kinetic rate data.[24].

Reaction number Reactions A b Ea (J/kmole) Reaction orders

Reaction 1 CH4 + 1.5O2 ? CO + 2H2O 1.59  1013 0 1.998  108 [CH4]0.7[O2]0.8
Reaction 2 CO + 0.5O2 ? CO2 3.98  108 0 4.18  107 [CO][O2]0.25[H2O]0.5
Reaction 3 CO2 ? CO + 0.5O2 6.16  1013 0.97 3.277  108 [CO2][H2O]0.5[O2]0.25
406 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

Table 2 where mP is the mass fraction of a product species (P), mR is the

The percentage of O2 in the oxidizer mixture at which the flame turn off at different mass fraction of a reactant (R), where R is the reactant species giv-
equivalence ratios and fuel volume flow rates.
ing the smallest value of Ri;k , A is an empirical constant equal to 4.0,
Equivalence ratio, / V of ¼ 6 L=min V of ¼ 9 L=min B is an empirical constant equal to 0.5, mR;k is the stoichiometric
0.5 24 22 coefficient for reactant i in reaction k and mj;k is the stoichiometric
0.55 23 22 coefficient for product i in reaction k. The eddy-dissipation model
0.65 22 21 relates the rate of reaction to the rate of dissipation of the reactant-
0.75 25 25 and product-containing eddies and the term k=e represents the time
1.0 25 25
scale of the turbulent eddies.

2.2.5. Reaction kinetics model

As a matter of fact, numerical modeling of combustion applica-
tions is a computationally demanding process. That’s why it is nec-
essary to apply simplified reaction mechanisms to reduce the
computing cost. Examples of CFD modeling in combustion applica-
tion using the global mechanisms are found in the literature
[20,21]. However, computationally simple mechanisms are cheap,
simple and available; the simplified schemes cannot be expected
to give accurate results especially under oxy-fuel combustion con-
ditions. The replacement of inert N2 with a chemically reactive
compound, CO2, has been shown to change the importance of some
of the elementary reactions governing the combustion, thereby
requiring a modification of the global multistep reaction mecha-
nisms to make them valid under oxy-fuel conditions [22]. In the
present work, the two-step hydrocarbon oxidation mechanism by
Westbrook and Dryer [23] which was modified by Andersen
et al. [24] is used for the calculations of the reaction kinetics, since
this scheme is directly available as default in the commercial CFD
code Fluent. The model was modified to handle the increased CO2
concentration under oxy-fuel conditions. Westbrook and Dryer in-
Fig. 3. Comparison between experimental (photo using a high speed camera) and cluded the reverse reaction step for CO2 decomposition in order to
numerical (taken from temperature contour plots) flame shape at different reproduce the proper heat of reaction and pressure dependence of
equivalence ratios, Vf = 6 L/min and O2/CO2:50/50%. the [CO]/[CO2] equilibrium. The mechanism of Westbrook and

Fig. 4. Visualization of the flame shape using images captured at steady state operation to investigate the effect of concentration ratio of O2 in the oxidizer mixture at fixed
fuel volume flow rate of 6 L/min: (a) at / = 0.55 and (b) at / = 0.75.
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 407

Dryer [23] consists of two step reactions with the second step oxi- of gaseous fuel may be affected by the following features: (1) lower
dation of CO to CO2, is only reversible. The two step reaction kinet- thermal diffusivity of CO2, (2) higher molar heat capacity of CO2,
ics model has been considered in this work for simplification of the (3) chemical effects of CO2 and (4) modified radiative heat transfer
reactions kinetics calculations with reasonable results regarding [26]. Since the molar heat capacity affects the flame temperature,
the species concentrations as compared to the detailed model its effect generally dominates.
[24]. We are considering here a complete 3D simulation for a gas To visualize the flame shape, a high speed camera has been used
turbine combustor which needs more calculation time. So, we have to record the flame photos under steady state operating conditions.
considered this simple 2-step model in order to reduce the calcu- The displayed images cover a field of view of the whole reactor
lation time with this large domain and get reasonable results. If height (300 mm). Fig. 3 shows a comparison between experimental
we would like to consider H2, we have at least to consider the four (photo using a high speed camera) and numerical (taken from tem-
step model as explained below but this will increase the calcula- perature contour plots) flame shapes at Vf = 6 L/min and O2/CO2:50/
tion time much without great differences in the results as reported 50% for different equivalence ratios. As it is clear from the figure,
by Andersen et al. [24]. The mechanism is listed in the form of the calculated flame shapes using the contour plots of temperature
three irreversible steps: distribution is close to the experimental recorded shapes taken
from the steady state flame photos. This is a good agreement how-
CH4 þ 1:5O2 ! CO þ 2H2 O ðreaction1Þ
CO þ 0:5O2 ! CO2 ðreaction2Þ
CO2 ! CO þ 0:5O2 ðreaction 3Þ
For this model, the modified reactions rates data by Andersen
et al. [24] are listed in Table 1. The set of governing equations and
boundary conditions are solved numerically. The details of the
calculation procedure can be found in previous works [14,25].
Gambit 2.2 was used to construct the mesh and as the reactor is
symmetric as shown in Fig. 1, only half part of the reactor was mod-
eled with a mesh of more than 500,000 finite volumes was used
with very small boundary control volumes in the regions of high
gradients. The CFD code Fluent 12.1 was selected for the solution
of the steady-state conservation equations. The grid independence
study was conducted before starting the simulations to obtain
accurate and numerically free error results. The refinement was
conducted for all computational cell sizes from 0.5  0.56 mm2 to
0.25  0.25 mm2 resulted in a difference of around 6% in the results.
Further reduction in the cell size below 0.25  0.25 mm2 did not
result in any considerable variations in the results.

3. Results and discussions

3.1. Flame stabilization behavior

The stability of swirl stabilized oxyfuel/CH4 flames was studied

for O2 mole fractions of 100 till the flame blow out state, equiva-
lence ratios of 0.5–1 and fuel volume flow rate of 5–10 L/min.
However, attempts of operating the burner with less than 21% O2
were unsuccessful in all of the considered operating range result-
ing in unstable operation and blow out. Table 2 presents the values
of the percentages of O2 in the oxidizer mixture (remainder is CO2)
at which the flame has reached a blow out condition. It seems from
those results that the most stable flame for the present design is at
equivalence ratio of 0.65, as the flame blows out at minimum pos-
sible percentage of O2 for stable flame, 22% at 6 L/min fuel volume
flow rate and 21% at 9 L/min fuel volume flow rate. As the equiva-
lence ratio was increased above 0.65, the flame instability condi-
tions increase and the flame turn off at higher percentage of O2
in the oxidizer mixture. This is attributed to the reduced amount
of oxygen that is available for combustion in the flame zone as
the equivalence ratio is increased. In the present model of gas tur-
bine combustor, a great portion of the inlet oxidizer is escaping
near to the combustor wall around the flame and cools down the
exhaust gases coming out from the combustion process. When
the equivalence ratio was decreased lower than 0.65 also another
cause for instability was encountered because of the increased
amount of CO2 available in the combustion chamber which reduces
the burning velocity and results in poor combustion performance Fig. 5. Predicted axial velocity distribution at different operating parameters:
and a modified distribution of temperature and species in the com- (a) U = 0.65 and Vf = 6 L/min, (b) Vf = 6 L/min and O2/CO2:50/50% and (c) U = 0.65
bustion chamber. The lower burning velocity for oxy-combustion and O2/CO2:50/50%.
408 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

ever; the experimental photos are not showing the regions of small in the oxidizer mixture due to its higher molecular weight. Thus,
temperature gradients. At the lower equivalence ratios, the com- the higher flame speed and lower Reynolds number hence assist
bustion is stable due to excess oxygen available in the combustion flame stabilization and lean operation at higher O2 levels. Also,
region. As a result, the flame temperature increases as character- the flames at higher oxidizer flow rates (lower equivalence ratio
ized from the flame shape and color as shown in the figure. As and higher Reynolds number) have similar stability limits.
the equivalence ratio increases, the red part of the flame increases
indicating less oxygen availability in the flame zone and that’s why 3.2. Velocity field
the flame starts to extend for more distance inside the reactor in
order to complete the combustion process. As a result, the temper- The presently developed computational procedure was used to
ature of the flame was reduced as will be discussed in the next sec- predict the characteristics of the flow field which are presented in
tion. At higher equivalence ratios, the combustor exhibits an the following subsections. Comparisons between experimental and
unstable mode; completely unstable combustion was encountered numerical results are also presented. Fig. 4 shows the predicted ax-
in case of equivalence ratio equal to 1.0. The differences between ial velocity distribution at different operating conditions including
experimental and numerical flame shapes increase as the equiva- percentages of O2/CO2 in the oxidizer at U = 0.65 and Vf = 6 L/min,
lence ratio increases. The instabilities are clearer from the experi- equivalence ratios at Vf=6 L/min and O2/CO2:50/50% and fuel vol-
mental results than the numerical results when the equivalence ume flow rates at U = 0.65 and O2/CO2:50/50%. From Fig. 5, it can
ratio was increased from 0.65 to 0.75. This is attributed to the re- be seen that there is an internal recirculation zone along the cen-
duced flow swirl and recirculation by increasing the equivalence terline downstream of the burner outlet. This zone is clearly indi-
ratio from 0.65 to 0.75. However, the numerical model response cated by the negative values of the axial velocity at heights ranging
to small change of the operating equivalence ratio is slow. The vi- between h = 75 mm and h = 120 mm which indicates a reverse flow
deo images show clear signs of an axially symmetric vortex ring along and in the vicinity of the burner axis. The axial velocity peak
structure forming at the sudden expansion and moving through ranges from 20 to 35 m/s, depending on the operating conditions,
the outer recirculation zone near to the reactor wall. at around h = 30 mm in the axial direction. The swirling oxidizer
Fig. 4 shows the percentages of O2 in the oxidizer mixture at con- stream along with the bluff body in the fuel flow pipe creates a re-
stant fuel volume flow rate of 6 L/min considering two operating gion of a severe reverse pressure gradient along the axis and this is
equivalence ratios. The images indicate that the flame can be oper- responsible for establishing the internal recirculation zone. In the
ated stably as the O2 content in the oxidizer is increased. This can be shear layer between the inflow and both the inner and outer recir-
attributed to two effects: the flame speed and/or the Reynolds culation zones, large velocity fluctuations and the low mean veloc-
number. As shown in the figure for both equivalence ratios, the ity are exhibited causing a very high turbulence intensity.
flame is stable for higher concentrations of O2. As the percentage Therefore, intense mixing of the cold fresh gas with hot burned
of CO2 increases in the mixture as the flow rate increases (constant gases can be expected in these regions. Fig. 5 indicates insignificant
equivalence ratio) and the flame starts to appear lift at 40%/60% of changes in the axial velocity distributions for percentages of O2/
O2/CO2. As the percentage of O2 in the oxidizer mixture is reduced CO2 in the oxidizer mixture between 30/70 and 60/40. At lower
more, the flame lift distance increases and occupies smaller area in- values of equivalence ratios, the recirculation is enhanced in the in-
side the reactor until the blow out point is encountered between ner recirculation zone due to the increased amount of oxidizer
25% and 22% of O2 in the oxidizer mixture. In oxyfuel combustion introduced to the reactor with a swirl angle of 60° as shown in
the O2 fraction of the oxidizer can be seen as an additional variable Fig. 5, part (b). This enhancement in flow recirculation can be seen
influencing the combustion behavior. In fact, the O2 concentration easily from Fig. 6 from the distribution of both axial and tangential
has a large effect on increasing the flame speed. The oxidizer com- components of the flow velocity. Also, as the amount of fuel flow
position also influences the flow field. This is due to the increase in increases at fixed operating equivalence ratio, the axial velocity in-
the Reynolds number increases as the percentage of CO2 increases creases and the recirculation is enhanced as shown in Fig. 5, part

Fig. 6. Predicted contour plots of axial and tangential velocities at the x-symmetry plane of the reactor at Vf = 6 L/min and O2/CO2:50/50% for two different equivalence ratios,
0.55 and 0.75.
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 409

(c). An external recirculation zone near to the reactor wall is also as expected. This is attributed to the reduced oxygen availability
noticed from the contour plots of the tangential velocity as shown for combustion at higher values of equivalence ratios and as a re-
in Fig. 6 and this outer flow recirculation is enhanced by lowering sult, the flame length is increased. The temperature is maximum
the operating equivalence ration. in the flame zone close to the reactor center line and it is reduced
in the radial direction due to fresh gases inlet. The higher temper-
3.3. Temperature and species mole fractions distributions atures in the region close to the reactor center line are explained by
the more frequent occurrence of reactions. To identify the differ-
The distributions of the static temperatures are displayed in ences more clearly, Fig. 8 displays the radial temperature profiles
Figs. 7a and 7b at the x-symmetry plane of the reactor at Vf = 6 L/ at different heights, h = 25 mm, h = 50 mm and h = 100 mm, at
min and O2/CO2:50/50% for two different equivalence ratios. It is U = 0.65, Vf = 6 L/min and O2/CO2:50/50%. The high-temperature
seen that the final temperatures of the flames close to exit section region is close to the center line at all heights however; the maxi-
of the combustor are increasing with the global equivalence ratios, mum temperature is reached around h = 50 mm. The increased

Fig. 7a. Predicted contour plots of temperature and mole fraction of O2, CO and H2O at the x-symmetry plane of the reactor at U = 0.55, Vf = 6 L/min and O2/CO2:50/50%.

Fig. 7b. Predicted contour plots of temperature and mole fraction of O2, CO and H2O at the x-symmetry plane of the reactor at U = 0.75, Vf = 6 L/min and O2/CO2:50/50%.
410 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

Fig. 8. Predicted radial species concentration distribution and temperature at different reactor heights for U = 0.65, Vf = 6 L/min and O2/CO2:50/50% (a) h = 25 mm,
(b) h = 50 mm and (c) h = 100 mm.

temperature level at h = 50 mm enhances, of course, the reactivity To identify the differences more clearly, the axial temperature
of the gas mixtures and, thus, the heat release and burnout as dis- profiles at different operating percentages of O2/CO2 in the oxidizer
cussed in the coming section. The temperature is reduced in the re- for U = 0.65 and Vf = 6 L/min are displayed in Fig. 9-part (a). With
gion at r  0–5 mm due to the recirculation in the inner increasing height, temperature values increase strongly reaching
recirculation zone and the reduced temperature at r  10–20 mm a maximum value at around h = 50 mm, and decrease slowly after-
reflects the inlet streams of fresh oxidizer gas. The temperature le- ward. It has been well established experimentally and numerically
vel in the outer recirculation zone close to reactor wall is much that with increasing the percentage of O2 in the oxidizer mixture, a
lower than in the inner recirculation zone, which is due to the lea- significantly higher peak as well as overall combustion tempera-
ner mixtures and heat loss through the wall by convection and ture levels has been encountered (see Fig. 9-part (a)). The lower
radiation mechanisms. As the height is increased more than combustion temperature in the cases of lower oxygen concentra-
50 mm, the temperature is also reduced due to the mixing of hot tion in the oxidizer mixture is mainly a result of high specific heat
burned gases with the excess unburned oxidizer. capacity of CO2. This causes an increase in radiation losses caused
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 411

Oxygen increases in the mixture, the rate of production of CO in-

creases in the inner recirculation zone as CH4 first burns and pro-
duces CO and then CO is consumed in the combustion with CO. For
O2, large concentrations appear at inlet and in the outer recircula-
tion zone and it decreases as the combustion proceeds. As the
equivalence ratio is increased, oxygen becomes unavailable away
from the burner exit as expected. The shapes of the distributions
of mole fractions of H2O resemble strongly those of temperature
for each flame in the same figure, Fig. 7. This result is also con-
firmed from the axial distributions of both temperature and H2O
mole fractions in Fig. 9. One can see that the larger concentrations
of H2O are near to the combustion zone as the H2O is a product of
combustion and its concentrations are increased with increasing
the amount of oxygen in the oxidizer mixture.
To configure more insight on the species concentrations, radial
gas concentration profiles for CH4, O2, CO, CO2 and H2O are shown
in Fig. 8 at different reactor heights for U = 0.65, Vf = 6 L/min and
O2/CO2:50/50%. The considered heights in this figure represent
the most intense combustion zone with the highest temperatures
for all cases. CH4 is consumed in the direction from center line to
the reactor wall because of reactions with the incoming fresh gas
and it completely disappears after h = 50 mm due to the excess
oxygen which is available to complete the combustion of CH4. Sim-
ilar behavior to that of CH4 is found for the distribution of CO inside
the reactor as CO is an intermediate product and it disappears due
to excess oxygen availability in the combustion zone. H2O is max-
imum in the flame zone and it decreases in the radial direction due
to the reduced reaction intensity and increased concentrations of
CO2 as a result of the coming fresh gas inlet. O2 and CO2 concentra-
tions are maximum close to the reactor wall due to fresh gas inlet
(O2 + CO2) and decreases in the direction of the flame zone.

3.4. Heat transfer

Fig. 10 shows the axial distribution of reaction-1(see Table 2)

Arrhenius rate at different operating conditions including percent-
ages of O2/CO2 in the oxidizer at U = 0.65 and Vf = 6 L/min, equiva-
lence ratios at Vf = 6 L/min and O2/CO2:50/50% and fuel volume
flow rates at U = 0.65 and O2/CO2:50/50%. As it is clear from the fig-
ure, part (a), when the percentage of O2 in the oxidizer mixture is
increased, the rate of reaction increases. This is due to the exis-
tence of enough amount of oxygen to complete the conversion of
Fig. 9. Predicted temperature, CO mole fraction and H2O mole fraction distributions
in the axial direction at different operating percentages of O2/CO2 in the oxidizer for CH4 and as a result the rates of reaction-1 and 2 increase. This fig-
U = 0.65 and Vf = 6 L/min. ure further explains the temperature profiles which have maxi-
mum values in the points of maximum rate of reaction. It is
shown that the rate of reaction reaches to nearly 5, 20 and
by the CO2 which increases the gas emissivity. The significance of 50 kg mol/m3 s for percentages of O2 of 30, 40 and 60, respectively
the latter effect obviously depends on the radiative properties of and drops along the length of the reactor. The rate of reaction be-
the flame considered, i.e. to what extent other components such comes zero after a certain height of the reactor and this is due to
as moisture soot and particles contribute to the radiative heat the fact that the fuel is totally consumed. Also, the rate of reac-
transfer. tion-1 decreases as the equivalence ratio is increased due to the re-
The contour plots of the distributions of the mole fractions of duced available amount of oxygen to enhance the reaction. As well,
O2, CO and H2O are presented in Figs. 7a and 7b at the x-symmetry the rate of reaction-1 is also increasing sharply when the amount
plane of the reactor at Vf = 6 L/min and O2/CO2:50/50% for two dif- of fuel volume flow rate is increased at the same operating equiv-
ferent equivalence ratios. As shown in Fig. 7, the distributions of alence ratio.
mole fractions of CO exhibit high values close to the fuel nozzle. The profiles of the reactor wall temperature in the axial direc-
The consumption of CO with increasing distance from the nozzle tion are shown in Fig. 11 at different operating parameters. Due
is in general accordance with the decrease in temperature as to improved combustion as a result of increasing the amount of
shown from the axial distributions in Fig. 9. CO disappears at a cer- oxygen in the fresh gas mixture, the gas temperature inside the
tain height of the reactor, around h = 100 mm. This is due to the reactor increases and reaches its maximum values around
consumption of CO in the reaction with O2 to produce CO2 and h = 100 mm. As a result, the wall temperature is also increased as
due to the extended reactor length which enables good gases mix- shown in Fig. 11, part (a) and the total amount of heat flux across
ing and, as a result, CO disappears form the exhaust gas region. the reactor wall is also increased as shown in Fig. 12, part (a). It
From Figs. 7 and 8, as the equivalence ratio increases as the was also found that the equivalence ratio has a significant effect
amount of CO increases and this is due to the less availability of on both wall temperature and accordingly the rate of heat loss
oxygen to complete the combustion. Also, as the percentages of across the reactor wall. As shown in Figs. 11 and 12, part (b), both
412 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

Fig. 10. Predicted axial distribution of reaction-1 Arrhenius rate at different

operating parameters: (a) U = 0.65 and Vf = 6 L/min, (b) Vf = 6 L/min and O2/CO2:
Fig. 11. Predicted reactor wall temperature distribution in the axial direction at
50/50% and (c) U = 0.65 and O2/CO2:50/50%.
different operating parameters: (a) U = 0.65 and Vf = 6 L/min, (b) Vf = 6 L/min and
O2/CO2:50/50% and (c) U = 0.65 and O2/CO2:50/50%.

wall temperature and wall heat flux are decreased with decreasing h = 300 mm at different operating conditions as shown in the
the equivalence ratio. This may be attributed to the cooling effect figure. For CH4/oxy-combustion, the flames were subject to
of the increased amount of the oxidizer that enters the reactor with thermo-acoustic instabilities over a substantial range of operating
decreasing the equivalence ratio. This amount helps in cooling the conditions. As shown in Fig. 13, part (a), the exhaust gas tempera-
gases close to the reactor wall and then reduces the amount of total ture increases, at the same radial position, with increasing O2 con-
heat flux across the wall. In part (c) of both Figs. 11 and 12, one can tent in the oxidizer mixture from 30% to 40%. This is attributed to
observe the great differences in both wall temperature and wall the higher flame temperature which causes an increase in the reac-
heat flux as the amount of fuel volume flow rate is changed from tivity leading to higher laminar flame speeds and shorter ignition
6 to 9 L/min at the same operating equivalence ratio. By increasing delay times. When the percentage of O2 in the mixture is increased
the fuel flow at the same equivalence ratio, the amount of oxidizer to 60%, the exhaust temperature was reduced again. From
is increased and the flow swirl is improved. As a result, the overall the experimental work conducted for all ranges of the flames, the
mixing and combustion process are improved causing a significant flame lift starts to occur when the O2/CO2 ratio in the mixture
increase in the combustion temperature. Because of this improved reaches 50/50%. Further reduction in the percentage of O2 in the
performance, the wall temperature increases and accordingly the mixture (increasing CO2) leads to an increase in the flame lift
amount of heat flux across the reactor wall increases. and significant influences on the combustion intensity and stabil-
ity. When the O2/CO2 ratio is 60/40%, flame lift is not there as com-
3.5. Exhaust temperature and emissions pared to the cases of 30/70% and 40/60% and that’s why the
exhaust temperature in case of 60/40% is less than the other con-
Fig. 13 shows a comparison between experimental and numer- sidered cases in Fig. 13, part (a). The temperature is maximum at
ical results of the radial profiles of the exhaust gas temperature at the centerline and decreases toward the reactor wall. The change
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 413

Fig. 12. Predicted reactor wall total heat flux distribution in the axial direction at
different operating parameters: (a) U = 0.65 and Vf = 6 L/min, (b) Vf = 6 L/min and
Fig. 13. Comparison between radial experimental and numerical exhaust gas
O2/CO2:50/50% and (c) U = 0.65 and O2/CO2:50/50%.
temperature at h = 300 mm at different operating parameters: (a) U = 0.65 and
Vf = 6 L/min, (b) Vf = 6 L/min and O2/CO2:50/50% and (c) U = 0.65 and O2/CO2:50/
50%.
in chemical reaction rates along with the reduced Reynolds num-
bers are expected to be main causes for the change of the flame
structure when changing the percentage of oxygen in the oxidizer increasing the fuel flow at fixed equivalence ratio improves the
mixture. overall combustion process and so, the exhaust gas temperature
Fig. 13-part (b) presents the effect of equivalence ratio on the is also improved. As shown in the figure, both experimental and
exhaust gas temperature, at Vf = 6 L/min and O2/CO2:50/50%. At numerical results are in a slightly good agreement. This deviation
the same fuel volume flow rate, the amount of corresponding oxi- between experimental and numerical data may be attributed to
dizer is lowered when the equivalence ratio is increased and the slight movement of the temperature probe inside the rector in
availability of oxygen is also reduced. This may justify the lowered addition to the error comes from the reading of the thermocouple
exhaust gas temperature when the equivalence ratio is increased probe itself. There may be also some errors in the calculations of
as shown in the figure from both experimental and thermal tem- the CO2 concentrations that directly affect the temperature distri-
perature data. The flow swirl is a very important parameter and bution inside the reactor. In the present study, only two step reac-
it forces the flow beside the wall to go toward the flame zone. By tion kinetics model has been considered which is a small number
lowering the oxidizer flow (increasing equivalence ratio) the swirl of steps in order to consider most of the intermediate species.
will be affected and a large portion of the oxidizer will escape be- However, because of the calculation time considering the 3D sim-
side the reactor wall without participating in the combustion pro- ulations, the present two step model has been used with expected
cess. That’s why the temperature is reduced by increasing the reasonable errors in the calculations as compared to higher num-
equivalence ratio toward stoichiometry. In part (c) of Fig. 13, the ber of steps models.
effect of fuel volume flow rate on the combustion temperature is A comparison between experimental and numerical average ex-
investigated at fixed equivalence ratio and O2/CO2 ratio. As seen, haust gas emissions at h = 300 mm as a function of percentage of
414 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415

lence ratio and O2/CO2%, the concentrations of all species are not
affected much as shown in Fig. 14, part (c).

4. Conclusions

In this work, both experimental and numerical investigations

were conducted for an atmospheric diffusion oxy-combustion
flame in a gas turbine model combustor. Oxy-combustion and
emission characterization, flame stabilization and oxy-combustion
model validation have been investigated in details. The combustor
is fuelled with CH4 and a mixture of CO2 and O2 as oxidizer. A mod-
ified two-step oxy-combustion reaction kinetics model for meth-
ane–oxygen combustion has been used in order to predict the
oxy-combustion characteristics of the considered flame and valida-
tions haves been done using the experimental results. Wide ranges
of different operating parameters have been considered including
equivalence ratio, percentage of O2/CO2 in the oxidizer mixture
and fuel volume flow rate. The stability limits of the oxy-combus-
tion diffusion flame have been determined experimentally and
numerically. From the above results, one can conclude that the sta-
bility of the oxy-combustion flame is affected when the operating
percentage of oxygen in the oxidizer mixture is reduced below
25%. In addition, ignition was not possible for the burner with less
than 21% oxygen in the oxidizer mixture. Visualizations of the
flame at the above mentioned operating conditions have been car-
ried out experimentally and comparisons with the numerical re-
sults were performed. The flames have been characterized in
details by measuring the exhaust gas temperatures and emissions
and comparing them with those from the numerical model. The
combustion stability was found to be improved with increasing
the percentage of O2 at inlet however there is a limitation in tem-
perature. Also, by lowering the equivalence ratio, improvement in
flow swirl and the overall combustion process has been reached.
Both experimental and numerical results are in good agreement
and the modified two step reaction kinetics model is capable of
predicting very similar trends of temperature and also the overall
flame shape is very close to the recorded experimental shapes.

Acknowledgements

The authors wish to acknowledge the support received from

King Abdulaziz City for Science and Technology (KACST) through
Fig. 14. Comparison between experimental and numerical average exhaust gas the Science and Technology Unit at King Fahd University of Petro-
emissions at h = 300 mm as a function of different operating parameters: leum and Minerals (KFUPM) for funding this work through Project
(a) U = 0.65 and Vf = 6 L/min, (b) Vf = 6 L/min and O2/CO2:50/50% and (c) U = 0.65 No. 09-ENE755-04 as part of National Science, Technology and
and O2/CO2:50/50%.
Innovation plan. The authors would also like to thank the King
Fahd University of Petroleum and Minerals (KFUPM) in Dhahran,
Saudi Arabia, for funding the research reported in this paper
O2 in the oxidizer mixture, equivalence ratio and fuel volume flow
through the Center of Clean Water and Clean Energy at MIT and
rate is shown in Fig. 14. Due to the long length of the combustor,
KFUPM. The support for the student from the mechanical Eng.
the combustion is almost complete and no CO is appearing there
Dep. at Alexandria University is also acknowledged.
experimentally and numerically as shown in the figure. As shown
in Fig. 14, part (a), the concentration of CO2 in the exhaust gas is
References
reduced when the amount of O2 is increased in the oxidizer mix-
ture which consists of O2 and CO2. However, by increasing the [1] IPCC. Contribution of working groups I, II and III to the fourth assessment
amount of oxygen, the combustion process is improved and CO2 report of the intergovernmental panel on climate change. Intergovernmental
created by combustion is increased. But, the effect of CO2 reduction Panel on Climate Change; 2007.
[2] Nemitallah MA, Ben-Mansour R, Habib MA. Numerical investigations of
in the oxidizer mixture is greater than the effect of increase of CO2 methane fueled oxy-fuel combustion model in a gas turbine combustor – 1.
due to improved combustion. As expected, the concentration of Flow fields, temperature, and species distribution; 2. Effect of CO2
H2O is increased in the exhaust by increasing O2 in the oxidizer recirculation. In: Proceedings of the 10th WSEAS international conference on
heat transfer, thermal engineering and environment, Istanbul, Turkey; 2012
mixture due to the improved combustion process as H2O is a com-
[ISSN: 2227-4596, ISBN: 978-1-61804-114-2].
bustion product. Also, O2 concentrations are increased as shown in [3] Ghoniem AF. Needs, resources and climate change: clean and efficient
the figure. By increasing the equivalence ratio at fixed ratio of O2/ conversion technologies. Prog Energy Combust Sci 2011;37:15–51.
CO2% and fuel volume flow, the concentrations of CO2 and O2 are [4] Pacala S, Socolow R. Stabilization wedges: solving the climate problem for the
next 50 years with current technologies. Science 2004;305:968–72.
decreased and H2O concentrations are slightly increased. By [5] Habib MA, Nemitallah MA, Ben-Mansour R. Recent development in oxy-
increasing the fuel volume flow rate at a fixed operating equiva- combustion technology and its applications to gas turbine combustors and ITM
 M.A. Nemitallah, M.A. Habib / Applied Energy 111 (2013) 401–415 415

reactors. Energy Fuels, in press. doi: [Link] [17] Chen L, Ghoniem AF. Simulation of oxy-coal combustion in a 100 kWth test
[Publication Date (Web): 19 November 2012]. facility using RANS and LES: a validation study. Energy Fuels
[6] Wall T. Combustion processes for carbon capture. Proc Combust Inst 2012;26:4783–98.
2007;31:31–47. [18] Zheng Y, Fan J, Ma Y, Sun P, Cen K. Computational modeling of tangentially
[7] Heil P, Torporov D, Stadler H, Tschunko S, Forster M, Kneer R. Development of fired boiler II NOx emissions. Chin J Chem Eng 2000;8(3):247–50.
an oxycoal swirl burner operating at low O2 concentrations. Fuel [19] Magnussen BF, Hjertager BH. On mathematical models of turbulent
2009;88:1269–74. combustion with special emphasis on soot formation and combustion. In:
[8] Liu C, Chen G, Sipocz N, Assadi M, Bai X. Characteristics of oxy-fuel combustion 16th symp (int’l) on combustion. The Combustion Institute; 1976.
in gas turbines. Appl Energy 2012;89:387–94. [20] Brink A, Mueller C, Kilpinen P, Hupa M. Possibilities and limitations of the Eddy
[9] Kutne P, Kapadia B, Meier W, Aigner M. Experimental analysis of the Break-Up Model. Combust Flame 2000;123:275–9.
combustion behavior of oxyfuel flames in a gas turbine model combustor. [21] Saario A, Oksanen A. Comparison of global ammonia chemistry mechanisms in
Proc Combust Inst 2010. [Link] biomass combustion and selective noncatalytic reduction process conditions.
[10] Williams T, Shaddix C, Schefer R. Effect of syngas composition and CO2-diluted Energy Fuels 2008;22:297–305.
oxygen on performance of a premixed swirl stabilized combustor. Combust Sci [22] Glarborg P, Bentzen LL. Chemical effects of a high CO2 concentration in oxy-
Technol 2008;180:64–88. fuel combustion of methane. Energy Fuels 2007;22:291–6.
[11] Ditaranto M, Hals. Combustion instabilities in sudden expansion oxy-fuel [23] Westbrook CK, Dryer FL. Simplified reaction mechanisms for the oxidation of
flames. J Combust Flame 2006;146:493–512. hydrocarbon fuels in flames. Combust Sci Technol 1981;27:31–43.
[12] Andersson K, Johnsson F. Flame and radiation characteristics of gas-fired O2/ [24] Andersen J, Rasmussen CL, Giselsson T, Glarborg P. Global combustion
CO2 combustion. Fuel 2007;86:656–68. mechanisms for use in CFD modeling under oxy-fuel conditions. Energy
[13] Warnatz J, Maas U, Dibble R. Combustion: physical and chemical Fuels 2009;23:1379–89.
fundamentals, modeling and simulation. 4th ed. Springer; 2006. p. 64. [25] Shuja SZ, Habib MA. Fluid flow and heat transfer characteristics in
[14] Habib MA, Elshafei M, Dajani M. Influence of combustion parameters on NOx axisymmetric annular diffusers. Comput Fluids 1996;25(2):133–50.
production in an industrial boiler. Comput Fluids 2008;37(1):12–23. [26] Liu F, Guo H, Smallwood G. The chemical effect of CO2 replacement of N2 in air
[15] Versteeg HK, Malalasekera W. An introduction to computational fluid on the burning velocity of CH4 and H2 premixed flames. J Combust Flame
dynamics; the finite volume method. Longman Scientific and Technical; 1995. 2003;133(4):495–7.
[16] Wilcox DC. Turbulence modeling for CFD. DCW Industries; 2000.