Professional Documents
Culture Documents
ScienceDirect
Article history: In this work, a new 3 kWe flameless combustor for hydrogen fuel is designed and analyzed
Received 8 January 2019 using CFD simulation. The strategy of the design is to provide a large volumetric com-
Received in revised form bustion for hydrogen fuel without significant rise of the temperature. The combustor initial
15 February 2019 dimensions and specification were obtained from practical design procedures, and then
Accepted 19 February 2019 optimized using CFD simulations. A three-dimensional model for the designed combustor
Available online 15 March 2019 is constructed to further analysis of flameless hydrogen combustion and consideration
that leads to disappearance of flame-front and flameless combustion. The key design pa-
Keywords: rameters including aerodynamic, temperature at walls and flame, NOX, pressure drop,
Hydrogen combustion efficiency for the hydrogen flame is analyzed in the designed combustor. To
Microturbine well demonstrate the combustor, the NOX and entropy destruction and finally energy
Flameless combustor conversion efficiency, and overall operability in the microturbine cycle of hydrogen
Low NOX flameless combustor is compared with a 3 kWe design counterpart for natural gas. The
Low carbon findings demonstrate that hydrogen flameless combustion is superior to derive the
microturbines with significantly lower NOX, and improvements in energy efficiency, and
cycle overall efficiency with low wall temperatures guaranteeing the long-term operation
of combustor and microturbine parts.
© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
* Corresponding author. Hamidreza Gohari Darabkhani, Professor of Low Carbon and Renewable Energy Systems.
E-mail addresses: b.bazooyar@staffs.ac.uk, bazooyar.bb@gmail.com (B. Bazooyar), h.g.darabkhani@staffs.ac.uk, h.g.darabkhani@
gmail.com (H.G. Darabkhani).
1
Bahamin Bazooyar, Research Fellow in Turbulent Combustion.
https://doi.org/10.1016/j.ijhydene.2019.02.132
0360-3199/© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4 11135
minimum. In order to improve the energy conversion effi- combustor nozzle are 135 and 264 mm. A 10 cm long entrance
ciency of these devices further, the main attention should be duct for the combustor is considered which is connected to a
given to their combustor. The combustor determines the conic wall angled 60 from axis of the combustor. The swirler
outlet temperature uniformity, the power efficiency, and the and fuel nozzle are embedded to the entrance duct of the
level of pollutant [6,7]. Along with a successful design of the combustor. The body of the linear is considered corrugated
combustor, the application of a renewable high calorific fuel with cylindrical 18 serrations. The stainless-steel material
could benefit the micropower generator by simultaneously with thickness 3 mm overall for the combustor is considered.
increasing the effectiveness and decreasing the pollution The discharge nozzle outlet diameter is 20 cm.
[8e15]. Hydrogen is a good candidate for this purpose as it has To calculate the fuel flow rate to power the turbine and
three times the energy density per weight as methane or combustor nozzle characteristics, the equilibrium combustion
gasoline while does carry no carbon emissions (CO, CO2, VOC, of H2/N2 was simulated and analyzed. Fig. 2 gives the sche-
and PAH). However, the adiabatic flame temperature of matic of the Aspen Plus process simulation flow diagram for
hydrogen is high which creates the potential for thermal NO fuel flow rate calculations. The fuel flow rate was obtained
formation [16,17]. This makes it necessary to come up with 8.042 gr/sec and fuel required pressure is 190.8 kPa (Calculated
strategies to control the temperature in hydrogen combustion. from Pfuel ¼ Pcombustor þ 2 DPpressure drop ).
The application of hydrogen could be considered as either an The fuel nozzle was designed based on the required fuel
additive to other conventional fuels or if it is diluted by an pressure and flow rate using Bernoulli equation. Nine fuel
inert gas. One significant strategy is to use hydrogen with ni- passage were considered for the nozzle, one along the
trogen which leads to a high volumetric flameless combustion combustor axis and eight with 45 inclination angle. The fuel
under the controlled temperature-so called autoignition. nozzle diameter was obtained 1.7 mm for the coaxial hole and
However, the rate of mixing in the combustor should be 0.985 mm (y1) for the inclined holes.
carefully managed to hamper any strong agitation of An axial swirler inclined 40 was considered and embedded
hydrogen with oxygen, and as a result significant NOX for- into the duct of the combustor. The shape of the vanes and an
mation. This could be achieved in a carefully designed area they allocate to the swirl air to enter the combustor were
combustor with ignition occurring via diffusion of hydrogen designed specifically based on the 3% combustor overall
and nitrogen rather than turbulent mixing. The autoignition pressure drop and share of the air considered as the swirl air
of H2/N2 flame in terms of flame stabilization, autoignition, (in this case: 7%). The position of the swirl vanes and nozzle
flame and turbulence structure, swirl dynamic, extinction and both in the combustor duct was achieved through the opti-
ignition [18e26] has been thoroughly researched using mization stage to avoid any flashback and high thermal stress
experimental and numerous numerical methods in a non- on the combustor walls. Four Primary holes with diameter
premixed imitated environment. However, it was rarely uti- 8 mm and sixteen dilution holes with diameter 6 mm were
lized and analyzed in micropower generators for production considered on the liner to stage the air within the combustor.
of power and energy. The UK government also aspires to The primary and dilution air holes were embedded in the liner
decarbonizing, or reducing the carbon content of the UK gas at distances 35 and 200 mm from beginning of the combustor
supply that is one of the options for reducing the emissions duct, respectively. The stainless-steel material is considered
from heat and power generation. In this work, a 3 KWe for the components of the combustor.
microturbine power generator with a recuperated combustor
is chosen as a case study. First, the application of H2/N2 was
verified in terms of the required amount of fuel and the opti- Model formulation
mum proportion of hydrogen to nitrogen to obtain the desired
power. Second, the renewable combustor designed by is The modelling of the combustor was performed using ANSYS
analyzed in the microturbine in terms of emission, energy, 19.2 Software to research the H2/N2 flame in the designed
and exergy. Thirdly, hydrogen is compared with methane in microturbine combustor. The conservation equations of
the designed microturbine combustor. mass, momentum and energy are achieved using a steady-
state solver with no Dufour effects, no work by pressure and
viscous forces, no surface reactions and no gas radiation.
Combustor design
V:ðr!
uÞ ¼ 0 (1)
The primary objective was to design a microturbine
2
combustor that could drive a 3 kWe turbine. Other combustor rð!
u :V!
u Þ ¼ Vp þ V: m V! u ÞT V:!
u þ ðV! uI (2)
parametric design outlooks include 47 kW heat output power, 3
3% overall pressure drop, 0.05 kg/s recuperated air mass flow 2 0 1
X !
rate at 850 K and 180 kPa, and 1100 K combustor outlet tem-
perature. A vortex combustor to provide the following opera- V: u rEf þ p ¼ V:4keff VT @ hj Jj A þ m V!
! u þ ðV!
u ÞT
j
tional conditions is designed and optimized to meet the 3
desired conditions. The dimensions of the combustor was 2 ! ! 5
V: u I : u þ Shf (3)
obtained from the Lefebvre practical formulation [27] and then 3
optimized using CFD calculations. Fig. 1 shows the schematic !
of the combustor. The diameters of the combustor liner and where r, !u , p, m, I, T, Ef , keff , Jj , hj , and Shf denote gas density,
casing are 46 and 58 mm. The length of the liner and velocity vector, absolute pressure, molecular viscosity, unit
11136 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4
tensor, temperature, total fluid energy, effective conductivity, According to second law of thermodynamics, the exergy
diffusion flux of species j, enthalpy of species j, and fluid efficiency is defined as follows:
enthalpy source term, respectively. For the combustor walls,
Edes þ Eloss
the conservation of energy is maintained: hQ ¼ 1 100% (9)
m_ H2 QLHV
V:ðkw :VTÞ ¼ 0 (4)
In this equation, Eloss represents the energy exiting the
where kw is the thermal conductivity of the wall. combustor by the flue gas, m_ H2 and QLHV denote the mass flow
To model the combustion, the conservation of species is rate and lower heating value of hydrogen, 119.96 MJ/kg.
required: The above-mentioned equations are solved in the defined
computational domain using following modelling consider-
!
V r!
u Yj ¼ V Jj þ Rj (5) ation. The shear stress transport turbulence model is
considered to model the highly reactive turbulence flow in the
In this equation, Yj and Rj denote the mass fraction and the
combustor [28]. The flamelet concept with 32 diffusion flames
net production rate of the species in chemical reactions.
in 64 grid points was utilized to model the chemistry of the
A transport equation for entropy is also solved:
flame. For this case, a Li hydrogen combustion mechanism
X
n
[29] is considered with 13 extra reaction among nitrogen ox-
Tds ¼ du þ pdð1=rÞ mj d nj r (6)
ides. Table 1 gives the list of reaction for modelling the com-
j¼1
bustion. Fig. 3 gives the path-line of reaction among nitrogen
where s indicates the specific entropy, u denotes the spe- oxides considered to conduct this simulation.
cific internal energy, mj represents the chemical potential of The density and specific heat of the H2/air are obtained by
species j, and nj is the number density of specie j. The incompressible-ideal-gas law and mixing law, respectively.
entropy generation Sgen in the combustor is defined as The gas mixture thermal conductivity and viscosity are
follows: calculated as a mass fraction-weighted mean of all species,
while the specific heat of species is extracted from a piecewise
_
Sgen ¼ ms (7)
polynomial fitting of temperature.
where m_ is the flow rate of gas. The computational domain and boundaries of the
The exergy destruction Edes in the combustor is germane to combustor is shown in Fig. 4. The boundary conditions are set
the overall entropy generation as follows: as: (1) three mass-flow-rate inlets are defined-one for fuel and
two for swirl and staged air (including primary and dilution) to
Edes ¼ Sgen T0 (8) the liner are considered. The inlet pressure of the fuel and air
T0 represents the ambient temperature, 850 K. were set as 190.8 and 180 kPa, respectively. The turbulent
Fig. 2 e The Aspen Plus simulation flow diagram of the microturbine combustor.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4 11137
intensity for inlets are 5% and hydraulic diameters of the fuel, pressure-outlet boundary condition is employed. The outlet
swirl and staged air inlets are 3.4, 15, and 5 mm. The fuel mass gauge pressure is 174.6 kPa. The hydraulic diameter and tur-
fraction of species is yH2 ¼ 0:02345, yO2 ¼ 0:0313, yH2 O ¼ bulent intensity of the outlet are 0.02 mm and 5%,
0:00126, yAR ¼ 0:01, and yN2 ¼ 0:96215. (2) For combustor respectively.
walls, zero diffusive flux for species and no slip condition for In finite volume part, simple algorithm is employed to
the entire fluid are employed at the gas-solid interfaces. For solve the pressure velocity coupling and discretize the equa-
the liner wall, the radiation and turbulent hear transfer is tions. All the equations are discretized by second order up-
considered, while for other walls the radiation in conjunction wind method: the convergence criterion for continuity,
with natural convection heat transfer is employed. The heat momentum, mixture fraction and its variance is 103 and for
transfer equation with the environment is obtained from the energy equation is as 106 . A monitor for area-averaged-
following: mean-temperature is also considered at the combustor
X X outlet as an assurance for completion of simulation. Before
Qloss ¼ h0 Ai Tw;i To þ εs Ai T4w;i T40 (10) final post processing of the results, several fine to coarse
meshes are tested starting with 524,546 to 5,890,285 cells. Our
where Ai denotes the surface area of grid cells, Tw;i represents
preliminary analysis has shown that the by increasing the
the temperature at the grid cells of the walls, s is the Stephan-
number of grids more than 6 million, the variation in results is
Boltzman constant, 5:67 108 W=ðm2 K4 Þ, h0 denotes the heat
less than 5%. Consequently, the 5.8 million grids are consid-
transfer coefficient, 10 W=ðm2 KÞ for natural convection heat
ered optimum in obtaining the required accuracy and saving
transfer and 32 W=ðm2 KÞ for turbulent convection heat
computational costs for this study.
transfer, and ε indicates the wall emissivity, 0.85. (3) For outlet,
11138 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4
vertical planes are shown in Fig. 5 (a) and (b). The flow topol-
Result and discussion ogies of the H2/N2 reactive jet are particularly suitable for
showing the preciseness and comprehensiveness of the design.
The results are categorized into two parts. In the first part, the The central core vortex mildly interacts with the previous
combustor performance is analyzed in terms of velocity field, impinging primary jets, which drives some part as backflow in
temperature, combustion efficiency, pressure drop. In the the primary region. As a result, a part of H2/N2 coming from the
second part, the combustor exergy, NOX and operation in a angled holes is pushed towards the walls, forming the sec-
3 kW recuperated microturbine power generator are ondary recirculation zone. These features are necessary for
compared for hydrogen and natural gas. sustainable combustion of the fuel in the combustor.
The primary port cross section (Fig. 5 (c)) has shown that a
Part I: combustor characteristics for H2/N2 part of the air pushes back to the combustor walls leading to
establishment of a secondary recirculation zone. This recir-
In this part, the aerodynamic field of the combustor, the
culation zone establishment near the primary holes curtails
temperature in the combustor and at it solid parts, combus-
the flame impingement to the liner, as a result chilling the
tion efficiency and pressure drop at the combustor axis are
combustor walls. A similar radial backflow movement has
evaluated from the CFD results.
also been observed at the dilution port cross section plane
(Fig. 5 (d)). The air jet penetration at the dilution port is also
Velocity field deep enough for the air to reach the center of combustor and
The qualitative feature of the designed combustor is displayed
cooling the fume.
by the velocity field of H2/N2 highly reactive turbulent flow. The
The design dimensions result in establishment of an axial
ensemble average of streamlines in the mid-horizontal and
recirculation zone near the combustor head and four radial
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4 11139
Fig. 5 e Reynolds averaged flow field, (a). Streamline in the vertical midplane Z-X. (b). Streamline in the vertical midplane Z-
Y. (c). Primary port cross section. (d). Dilution port cross section.
recirculation zones near the liner wall. However, the nature of region. The liner wall is completely isolated from the fuel
flow at dilution jet is of different nature. The flow pattern stream by primary jet and no fuel encroachment could be
downstream of the primary jets is not that much under the observed: no reaction approaches the liner wall which is evident
influence of design consideration including geometry of fuel from green layer in proximity of the wall in all the temperature
nozzle and swirl characteristics. The penetration jet to the profile. Also apparent is the high temperature of escaping fume
combustor center at the dilution port is rather symmetrical. from the primary region is fully controlled and went down in the
This is from the prefect circular shape of discharge nozzle of discharge nozzle, leading to uniform outlet temperature dis-
the combustor. tribution with a desired average value in design objectives (Fig. 6
(c)). The effective control of temperature in the combustor leads
Temperature to very low concentration of NOX at the outlet plane.
The occurrence of the combustion in the designed combustor
fully mimics the features of flameless burners without any Combustion efficiency
flame front and significant NOX emission. This is because 1) The combustion efficiency in percentage is reported here to
the hydrogen fuel is mixed with nitrogen, an inert gas to show the capability of the combustor in completion of the
control the combustion 2) the reactive fluid temperature does combustion before the combustor nozzle. Here, combustion
not transcend 1450 K, 3) the combustion takes place over a efficiency is defined the heat loss from combustor outlet via
large volume of gas. In this section, we have shown the the incomplete combustion products:
feature of the designed combustor. Fig. 6 (a) gives the velocity
QH2 ½H2
vectors in mid vertical place of the combustor. The primary h¼1 100%
QH2 ½H2 0
jet momentum is not high enough to penetrate the central
cortex of the fluid. Instead, it moves coaxially with instead of where QH2 ¼ 286 kJ/mol is hydrogen enthalpy of the combus-
agitating the fuel stream. This brings about the large volu- tion and ½H2 is hydrogen spatial concentration in the
metric combustion with low spatial temperatures (Fig. 6 (a)). combustor. The ½H2 0 denotes the average volumetric concen-
The primary jet also helps the swirl air narrowing the flame tration of hydrogen in combustor when there is no
by deviating the fuel stream that moves towards the combustion.
combustor liner, as a result controls the temperature at the Fig. 7 represents the value of combustion efficiency at the
walls (Fig. 6 (b)). The swirl intensity is high enough to carry combustor axis. The length of the axis is normalized in this
the mixture of hydrogen/nitrogen forward without any flame graph giving 0 for the beginning of the liner and 1 for the
flashback and to narrow down the flame without any combustor outlet. The position of primary holes, dilution
impingement to the combustor head. holes and liner end surface are 0.07, 0.3, and 0.39, respectively.
The temperature profile (Fig. 6(a)) demonstrates that hot Fig. 7 can show how the design can proceed the combustion to
region is between the primary and dilution ports in the middle the completion. The combustion efficiency reaches almost
parts of the chamber. Indeed, no combustion occurs after the 100% at 0.1 total combustor length. This length corresponds to
injection of dilution air and the flame is evidently stabilized in a little distance further than primary holes and is well fallen
the head of the combustor and in the wake of the recirculation before the dilution holes.
11140 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4
Fig. 6 e Reynolds averaged (a). Temperature and velocity vector in the vertical midplane Y-Z. (b). Wall Temperature. (c).
Outlet temperature distribution.
Pressure drop benchmarks and specification that was taken into account of
The pressure drop is another key design parameter of the consideration.
microturbine combustors. The fluid acquires pressure drop
because of change of in areas as it passes through the primary, Comparison between hydrogen and methane
secondary and dilution holes and combustor outlet nozzle, if
any. The pressure drop at the combustor axis is illustrated in In this part, the NOX and entropy, and finally the overall
Fig. 8. It shows that how the microturbine fluid experiences operation of the combustor for 3 kWe energy power genera-
pressure drop passing the combustor. tion are evaluated, analyzed and compared for H2 and natural
The s giving the proportion of the spatial pressures to the gas. The natural gas is considered 94% methane, 4% ethane,
inlet air pressure s ¼ P=Pinlet is given in Fig. 8. The s is around 0.3% propane, 0.2 butane, 1% nitrogen, and 0.5% carbon di-
0.97 at the outlet giving the fluid pressure drop in the oxide (by mole). For Natural gas, the same strategy repre-
combustor around 3%. This meets the combustor design sented in section 2 is repeated to obtain the required natural
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 4 ( 2 0 1 9 ) 1 1 1 3 4 e1 1 1 4 4 11141
Entropy
The entropy destruction for both hydrogen and natural gas
fuels are compared in the designed 3 kWe microturbine
combustor. The contour plots of Fig. 10 shows variation of
entropy. For this combustor, the entropy variation is very
much consistent with temperature variations. The entropy at
the combustor outlet for hydrogen and natural gas were 8395
and 8383 J=kg:K, respectively.
Comparing it for hydrogen and natural gas, it can be readily
Fig. 7 e The spatial combustion efficiency at the axis of the
seen that the entropy generation as well as destruction for
combustor (z/L ¼ 0: end of liner head and beginning of the
natural gas is higher than hydrogen. The maximum entropy
liner).
generation for natural gas is (12,509 J=kg:K), while for hydrogen
flameless combustion is (9446.19 J=kg:K). As a result, the en-
tropy destruction of natural gas is 4126J=kg:K, while for
hydrogen fuel it is only 1051 J=kg:K. This less entropy
destruction in H2/N2 combustion indeed leads to more exergy
loss from the combustor when the hydrogen ignites in the MT
combustor.
qregen;act
ε¼ (11)
qregen;max
wturb þ wcomp
hth;regen ¼ (12)
qH
Fig. 9 e Contour plots of NOX [mgr=m3 ]; a) Flameless hydrogen combustion, b) Natural gas.
Fig. 10 e Contour plots of Entropy [J=kg:K]; a) Flameless hydrogen combustion, b) Natural gas.
recuperator. The recuperator duty is to provide heat and in- The T-S diagram of the cycle for both fuels reveals that
crease the temperature of the compressed air from standard cycle efficiency for hydrogen is 3% higher than natural gas,
conditions to 850 K. while the recuperator effectiveness is 0.02 lower.
The cycle efficiency for hydrogen and natural gas were 29% 5) Finally, it can be concluded that the air distribution,
and 26%, respectively. The difference among the cycle effi- orientation of nozzle, its position, and the diameters of
ciency was obtained mainly because of the release of energy primary and secondary holes are all important in estab-
in the combustor. The fact that hydrogen heating density is lishment of the flameless combustion in the combustor for
more than NG gives rise to the higher overall efficiency of hydrogen. These design considerations should be opti-
microturbine with hydrogen fuel compared to NG. mized for whatever the design is to achieve the better the
advantages of renewable fuels.
Conclusion
and emission of biodiesel blends and diesel in an [23] Wang H, Luo K, Yi F, Fan J. Direct numerical simulation study
experimental boiler. Appl Energy 2011;88(12):4725e32. of an experimental lifted H2/N2 flame. Part 2: flame
[16] Bazooyar B, Hashemabadi SH, Shariati A. NOX formation of stabilization. Energy Fuels 2012;26(8):4830e9.
biodiesel in utility power plant boilers; Part B. Comparison of [24] Wang Z, Fan J, Zhou J, Cen K. Direct numerical simulation of
NO between biodiesel and petrodiesel. Fuel 2016;182:323e32. hydrogen turbulent lifted jet flame in a vitiated coflow. Chin
[17] Bazooyar B, Shariati A, Hashemabadi SH. Characterization Sci Bull 2007;52(15):2147e56.
and reduction of NO during the combustion of biodiesel in a [25] Wang Z, Zhou J, Cen K. Visualization of nonpremixed
semi-industrial boiler. Energy Fuels 2015;29(10):6804e14. hydrogen jet flame in a vitiated coflow by DNS. J Visual
[18] Gordon RL, Masri AR, Mastorakos E. Simultaneous Rayleigh 2007;10(2):136.
temperature, OH-and CH 2 O-LIF imaging of methane jets in [26] Wu Z, Masri AR, Bilger RW. An experimental investigation of
a vitiated coflow. Combust Flame 2008;155(1):181e95. the turbulence structure of a lifted H2/N2 jet flame in a
[19] Luo K, Wang H, Yi F, Fan J. Direct numerical simulation study vitiated co-flow. Flow Turbul Combust 2006;76(1):61e81.
of an experimental lifted H2/N2 flame. Part 1: validation and [27] Lefebvre AH. Gas turbine combustion. CRC press; 1998.
flame structure. Energy Fuels 2012;26(10):6118e27. [28] Menter FR. Two-equation eddy-viscosity turbulence models
[20] Michel J-B, Colin O, Angelberger C, Veynante D. Using the for engineering applications. AIAA J 1994;32(8):1598e605.
tabulated diffusion flamelet model ADF-PCM to simulate a [29] Li J, Zhao Z, Kazakov A, Dryer FL. An updated comprehensive
lifted methaneeair jet flame. Combust Flame kinetic model of hydrogen combustion. Int J Chem Kinet
2009;156(7):1318e31. 2004;36(10):566e75.
[21] Mouangue R, Obounou M, Mura A. Turbulent lifted flames of [30] Bazooyar B, Ebrahimzadeh E, Jomekian A, Shariati A. NOx
H 2/N 2 fuel issuing into a vitiated coflow investigated using formation of biodiesel in utility power plant boilers. Part A:
Lagrangian Intermittent Modelling. Int J Hydrogen Energy influence of fuel characteristics. Energy Fuels
2014;39(24):13002e13. 2014;28(6):3778e92.
[22] Patwardhan S, De S, Lakshmisha K, Raghunandan B. CMC
simulations of lifted turbulent jet flame in a vitiated coflow.
Proc Combust Inst 2009;32(2):1705e12.