Professional Documents
Culture Documents
Hyrdoge Gas Turbin
Hyrdoge Gas Turbin
HOSTED BY http://ppr.buaa.edu.cn/
www.sciencedirect.com
a
B&B-AGEMA GmbH, Jülicher Str. 338, Aachen 52070, Germany
b
FH Aachen University of Applied Sciences, Hohenstaufenallee 6, Aachen 52064, Germany
c
RWTH Aachen University, Templergraben 55, Aachen 52062, Germany
KEYWORDS Abstract Combined with the use of renewable energy sources for its production, hydrogen
represents a possible alternative gas turbine fuel within future low emission power generation.
Micromix combustion;
Hydrogen gas turbine; Due to the large difference in the physical properties of hydrogen compared to other fuels such as
Dry-low-NOx (DLN) natural gas, well established gas turbine combustion systems cannot be directly applied for dry-
combustion; low-NOx (DLN) hydrogen combustion. Thus, the development of DLN combustion technologies
Hydrogen combustion; is an essential and challenging task for the future of hydrogen fuelled gas turbines.
High hydrogen The DLN micromix combustion principle for hydrogen fuel has been developed to significantly
combustion reduce NOx emissions. This combustion principle is based on cross-flow mixing of air and
gaseous hydrogen which reacts in multiple miniaturized diffusion-type flames. The major
advantages of this combustion principle are the inherent safety against flash-back and the low
NOx emissions due to a very short residence time of reactants in the flame region of the micro-
flames.
The micromix combustion technology has been already proven experimentally and numerically
for pure hydrogen fuel operation at different energy density levels. The aim of the present study is
to apply and compare different combustion models for the characterization of the micromix flame
structure, its interaction with the flow field and its NOx emissions.
The study reveals great potential for the successful application of numerical flow simulation to
predict flame structure and NOx emission level of micromix hydrogen combustion, help
n
Corresponding author. Tel.: þ49 241 5687830.
E-mail address: ayed@bub-agema.de (A. Haj Ayed).
Peer review under responsibility of National Laboratory for Aeronautics and Astronautics, China.
2212-540X & 2015 National Laboratory for Aeronautics and Astronautics. Production and hosting by Elsevier B.V. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.jppr.2015.07.005
124 A. Haj Ayed et al.
understanding the flow phenomena related with the micromixing, reaction zone and NOx
formation and support further optimization of the burner performance.
& 2015 National Laboratory for Aeronautics and Astronautics. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Figure 1 Micromix prototype combustor for gas turbine Honeywell/Garrett Auxiliary Power Unit APU GTCP 36-300.
Experimental and numerical investigations of the DLN 125
Nomenclature
Greek letters
A area (unit: mm2)
BR blockage ratio Φ equivalence ratio
d diameter/inner diameter (unit: mm2) ν kinematic viscosity (unit: m2/s)
D outer diameter (unit: mm2)
ED energy density (unit: MW/(m2 bar))
Subscripts
ṁ mass flow (unit: kg/s)
p pressure (unit: bar)
3 combustor inlet
T temperature (unit: K)
4 combustor outlet
u friction velocity (unit: m/s)
AGP air guiding panel
y wall distance (unit: m)
crit critical
yþ dimensionless wall distance
fuel fuel/hydrogen
H2-seg hydrogen burner
The micromix hydrogen combustion research is done using 2. Micromix hydrogen combustion
an interactive optimization cycle including experimental and
numerical studies on test burners, full scale combustion 2.1. Micromix description
chamber investigations and the feasibility is proven in real
gas turbine operation (Figure 2). Gaseous hydrogen is injected through miniaturized injec-
Based on these studies the impact of different geometric tors perpendicularly into an air cross-flow through small air
parameters on flow field, flame structure and NOx formation guiding panel (AGP) structures. This leads to a fast and
are identified and the micromix combustion principle is intense mixing, which takes place simultaneously to the
continuously optimized. combustion process. As a result, miniaturized micro flames
Within the present study, 3 different combustion models develop and anchor at the burner segment edge downstream
have been applied within a simplified numerical approach to of the injector nozzle. Multiple micro flames instead of
simulate the flow, the combustion reaction and the thermal large scale flames lower the residence time of the NOx
NO emission inside the micromix test burner. The compar- forming reactants and consequently the averaged molar
ison of the different combustion models should support fraction of NOx can be reduced significantly as has been
identification of well suited models for the accurate shown in Ref. [6]. The main influence on the low NOx
simulation of the micromix combustion principle and NO characteristic can be ascribed to the key design parameters
emission behavior. blockage ratio BR of the air guiding panel AGP (Figures 3
Figure 2 Interactive optimization cycle of micromix combustor research and development for APU GTCP 36-300.
126 A. Haj Ayed et al.
3. Numerical investigation
For the numerical evaluation of the burner geometry in terms
of flame stabilization and NO emission level a simplified
numerical approach has been applied. The numerical approach
Figure 4 Experimental results of test burners APU 030 ED 6.7. includes the 3D numerical simulation of the flow field within
the burner and combustion chamber based on a RANS solver,
three different combustion models and a simplified thermal NO
formation model. The comparison of 3 different combustion
models will show the sensitivity of the micromix flame
simulation to model selection and will support the selection
of an appropriate model for the further numerical investigation
of micromix burners.
In this section of the paper the simplified numerical
approach is described and its application to calculate the
reactive flow in the burner configuration is presented.
Table 1 Calculated thermal NO emissions. The Eddy Dissipation Concept (EDC) model also combines
turbulent mixing effects with chemical kinetics. In each volume
EDM ED-FR EDC Measurement cell the space is divided into two zones: the fine structure and
the bulk gas zone. All the chemical reactions are assumed to
NO emission 6 0.65 1.02 1.3
[ppmv @ 15% O2] occur in the fine structures which represent the smallest
turbulence scales, where all turbulence energy is dissipated into
heat. The reactions are neglected in the bulk gas zone, and the
mass transfer exchanges the reactant and product gases between
the fine structures and the bulk zone. The reaction is driven by
simplified geometric model derived from the test burner
the residence time of each species in the fine structured zone.
configuration APU 030 ED 6.7. The geometric model is
Therefore the mass exchange rate between the fine structures
shown in Figure 5 and covers a longitudinal burner slice,
and the bulk gas as well as the mass fraction of the fine
which makes use of the symmetric nature of the burner in
structures in every cell is calculated by the turbulence dissipa-
both lateral and vertical directions. The symmetric bound-
tion and kinematic viscosity information.
aries along the lateral direction are set on the cross section
The calculation of reaction rates and species consump-
through the center of one hydrogen injection hole and on
tion/formation is achieved by the use of the DARS-CFD
the cross section between two hydrogen injection holes,
library provided within the CFD code. DARS-CFD pro-
respectively. Along the vertical direction the symmetry
vides a detailed hydrogen reaction mechanism including 19
planes are set on the center section through one air guiding
reversible reaction rates and 9 species (see Table 1) and
panel and on the center section through one hydrogen
calculates the reaction rates based on detailed chemistry
segment. Thus, the slice model contains one half of a
Ref. [14]. Thereby, it makes use of the ISAT tabulation
hydrogen injection hole and one half of an air guiding gate.
(in situ adaptive tabulation), which approximates computa-
3D steady RANS calculations have been performed. The
tionally expensive functions based on previously calculated
realizable k, ε turbulence model with all yþ wall treatment
and stored solutions instead of solving them in every
has been applied. The wall treatment is decided depending
calculation step, hence, reducing the calculation effort
on the local dimensionless wall distance yþ, which is
and time.
calculated from the wall distance y, the friction velocity u
The 19 reactions forming the detailed hydrogen reaction
and the kinemativ viscosity ν:
mechanism are listed as follows:
yþ ¼ y u=ν
Figure 9 Calculated temperature distribution. Figure 10 Calculated temperature distribution (modified color scale).
a higher temperature of the outer vortex pair as marked in Considering that the numerical result only includes
Figure 9. The ED-FR and the EDC calculations provide more thermal NO and excludes any other possible NO formation
similar temperature distribution and flame structure. The ED-FR routes, e.g. the NNH formation route, the numerical result
model provides slightly higher temperatures along the micromix of the EDC model shows very good agreement with the
flame, whereas the EDC model provides slightly higher experimental findings and can be considered as reliable
temperature inside the outer cortex pair. The maximum CFD model for the proper modeling of the micromix flames
temperature level is found below 1700 1C for both models. and prediction of their NOx emission behavior.
Significantly higher peak temperatures are reached with the
EDM calculation. These exceed 2000 1C in the reaction zone
and result from fast reaction that leads to concentrated local heat 4. Conclusion
release.
The micromix test burner with an injector diameter of
Figure 10 shows the temperature distributions with
0.3 mm has been tested successfully under atmospheric
modified color scale (1200 1C to 2000 1C), which confirm
conditions and has proven its dry low NOx ability over a
the significant differences in the temperatures calculated in
wide operating range. Numerical investigations with differ-
the EDM calculation compared to the ED-FR and the EDC
ent combustion models revealed the ability of the applied
calculations. The figure also confirms similarity of the ED-
numerical approach to simulate the micromix combustion
FR and the EDC results.
and to capture the typical micromix flame anchoring and
The different reaction speed resulting from different
structure. The comparison of different combustion models
combustion models is shown by comparison of the reaction
has shown that the results obtained with the EDC model
rates for all three calculations.
based on the DARS-CFD detailed chemistry model of the
Figure 11 shows the distribution of the production rate of
CFD code STAR-CCMþ show better agreement compared
H2O in the burner, which also stands for the reaction rate, in
to the EDM and the ED-FR model results. A further
comparison to the experimentally captured visual flame
validation of the EDC model regarding the prediction of
appearance. Highest reaction rates are found in the EDM
NOx emissions along the combustor operation range will be
calculation. Thereby, the reaction starts to take place just
investigated within a next research step, in order to establish
downstream the hydrogen injector, indicating fast and early
an efficient numerical method for the systematic optimiza-
heat release. Further, most of the heat release takes place in
tion of the micromix combustion principle.
a short distance downstream the hydrogen injector, which
explains the higher temperature in the outer vortex pair.
Contrarily, the reaction rates calculated by the ED-FR Acknowledgements
and the EDC models show that the reaction mainly starts at
the typical micromix flame anchoring point, which is at the The numerical flow and combustion simulations pre-
edge of the hydrogen segment, and the reaction zone is sented in this paper have been carried out with the STAR-
thinner and takes place within a longer axial distance CCMþ Software of CD-adapco. Their support is gratefully
compared to the EDM calculation, which explains more acknowledged.
homogenous temperature distribution inside the combustion
chamber and lower peak temperatures. The position and References
structure of the micromix flames calculated by the ED-FR
and the EDC model show good agreement with the visual [1] T. Lieuwen, V. Yang, R. Yetter, Synthesis Gas Combustion:
Fundamentals and Applications, CRC Press Taylor & Francis
flame appearance captured during experimental testing,
Group, Boca Raton, 2010.
whereas the flame structure resulting from the EDM model
[2] G. Dahl, F. Suttrop, Engine control and low-NOx combustion
shows significant deviations as shown in Figure 11. for hydrogen fuelled aircraft gas turbines, International
The thermal NO emission has been calculated separately Journal of Hydrogen Energy 23 (1998) 695–704.
for each combustion model based on the extended Zeldo- [3] G. Dahl, R. Dorneiski, Low NOx-potential of hydrogen-
vich mechanism as explained above. The required concen- fuelled gas turbine engines, In: Proceedings of the 1st
tration of O and OH radicals is calculated based on the International Conference on Combustion Technologies for
Warnatz approach [16] as these are not directly provided by Clear Environment, Villamoura, Portugal, 3-6 September
the EDM and the ED-FR models. 1991.
Table 1 shows the calculated thermal NO emissions for [4] H.H.-W. Funke, S. Börner, P. Hendrick, E. Recker,
all three calculations and the measured NOx emission Modification and testing of an engine and fuel control system
for a hydrogen fuelled gas turbine, Progress in Propulsion
(corrected to 15% O2). Due to the higher temperatures in
Physics III, 2009.
the reaction zone, the EDM model provides the highest
[5] H.H.-W. Funke, S. Börner, P. Hendrick, E. Recker, Control
thermal NO emission of approx. 6 ppmv @ 15% O2. The system modifications for a hydrogen fuelled gas-turbine, In:
ED-FR and the EDC models provide lower thermal NO The 13th International Symposium on Transport Phenomena
emission with approx. 0.65 and 1.02 ppmv @ 15% O2, and Dynamics of Rotating Machinery, 2010.
respectively. Thereby, the result of the EDC model is found [6] H.H.-W. Funke, S. Börner, P. Hendrick, E. Recker,
closest to the measurement value of 1.3 ppmv @ 15% O2. R. Elsing, Development and integration of a scalable low
Experimental and numerical investigations of the DLN 131
NOx combustion chamber for a hydrogen fuelled aero gas and experimental characterization of low NOx micromix
turbine, In: Proceedings of the 4th European Conference for combustion principle for industrial hydrogen gas turbine
Aeronautics and Space Sciences, 2011, accepted for: applications, ASME Turbo Expo 2012, GT2012-69421,
Advances in Propulsion Physics. Copenhagen, DK, 2012.
[7] S. Börner, H.H.-W. Funke, F. Falk, P. Hendrick, Control [12] H.H.-W. Funke, S. Börner, A. Robinson, P. Hendrick,
system modifications and their effects on the operation of a E. Recker, Low NOx H2 combustion for industrial gas
hydrogen-fueled auxiliary power unit, In: Proceedings of the turbines of various power ranges, In: Proceedings of the
20th International Symposium on Air Breathing Engines, 2011. 5th International Conference the Future of Gas Turbine
[8] F. Shum, J. Ziemann, Potential use of hydrogen in air Technology, ETN-2010-42, Brussels, Belgium, 2010.
propulsion, Euro-Québec Hydro-Hydrogen Pilot Project [13] H.H.-W. Funke, E. Recker, S. Börner, W. Bosschaerts, LES
(EQHHPP), European Union, Contract No. 4541-91-11 EL of jets in cross-flow and application to the micromix
ISP PC, Final Report, 1996. hydrogen combustion, In: Proceedings of the 19th Interna-
[9] A. Westenberger, Liquid hydrogen fuelled aircraft – system
tional Symposium on Air Breathing Engine, ISABE-2009-
analysis, CRYOPLANE, European Commission Final Tech-
1309, Montreal, Canada, 2009.
nical Report No. GRD1-1999-10014, 2003.
[14] CD-adapco, Userguide Star-CCMþ 9.02. CD-adapco, 2014.
[10] H.H.-W. Funke, S. Börner, J. Keinz, P. Hendrick, E. Recker,
[15] G.A. Lavoie, J.B. Heywood, J.C. Keck, Experimental and
Low NOx hydrogen combustion chamber for industrial gas
theoretical study of nitric oxide formation in internal com-
turbine applications, In: Proceedings of the 14th International
Symposium on Transport Phenomena and Dynamics of bustion engines, Combust. Sci. Tech 1 (4) (1970) 313–326.
Rotating Machinery, Honolulu, Hawaii, 2012. [16] J. Warnatz, Rate coefficients in the C/H/O system, Combus-
[11] H.H.-W. Funke, S. Börner, J. Keinz, K. Kusterer, tion Chemistry, 1984.
D. Kroniger, J. Kitajima, M. Kazari, A. Horikawa, Numerical