Professional Documents
Culture Documents
2019 - An Overview On Dry Low NOx Micromix Combustor Development For Hydrogen-Rich Gas Turbine Applications
2019 - An Overview On Dry Low NOx Micromix Combustor Development For Hydrogen-Rich Gas Turbine Applications
2019 - An Overview On Dry Low NOx Micromix Combustor Development For Hydrogen-Rich Gas Turbine Applications
ScienceDirect
Article history: The paper presents a survey of the interactive optimization cycle at Aachen University of
Received 20 November 2018 Applied Sciences, used for the development of a new low emission Micromix combustor
Received in revised form module for application in hydrogen fueled industrial gas turbines. During the development
16 January 2019 process, experimental and numerical methods are applied to optimize a given baseline
Accepted 17 January 2019 combustor with 0.3 mm nozzles with respect to combustion efficiency, combustion sta-
Available online 14 February 2019 bility, higher thermal power output per nozzle and reduced manufacturing complexity.
Within the described research cycle combustion and flow simulations are used in the
Keywords: context of parametric studies for generating optimized burner geometries and the
Combustion phenomenological interpretation of the experimental results. Experimental tests, carried
Gas turbine out on an atmospheric combustion chamber test stand provide the basis for validation of
Hydrogen simulation results and proof of the predicted combustion characteristics under scaled
CFD down gas turbine conditions.
Low emission In the presented studies, an integration-optimized Micromix combustor with a nozzle
diameter of 0.84 mm is tested at atmospheric pressure over a range of gas turbine operating
conditions with hydrogen fuel. The combustor module offers an increase in the thermal
power output per nozzle by approx. 390% at a significant reduced number of injectors when
compared to the baseline design. This greatly benefits manufacturing complexity and the
robustness of the combustion process against fuel contamination by particles.
During atmospheric testing, the optimized combustor module shows satisfactory oper-
ating behavior, combustion efficiency and pollutant emission level. Within the evaluated
operating range, which correlates to gas turbine part-, full- and overload conditions, the
investigated combustor module exceeds 99% combustion efficiency. The Micromix combustor
achieves NOx emissions less than 2.5 ppm corrected to 15 Vol% O2 at the design point.
Based on numerical analyses and experimental low pressure testing, a full-scale gas
turbine combustion chamber is derived. High pressure testing in the auxiliary power unit
Honeywell/Garrett GTCP 36e300 shows stable operation during acceleration of the engine,
during IDLE and during load variations between IDLE and Main Engine Start (MES) mode.
Throughout the investigated operating range, the combustion chamber generates low NOx
emissions under full-scale gas turbine conditions.
© 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
* Corresponding author.
E-mail addresses: funke@fh-aachen.de (H.H.-W. Funke), n.beckmann@fh-aachen.de (N. Beckmann), sylvester.abanteriba@rmit.edu.
au (S. Abanteriba).
https://doi.org/10.1016/j.ijhydene.2019.01.161
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 ) 6 9 7 8 e6 9 9 0 6979
industrial scale gas turbines [11e13]. Micro injection with jet- region, counter rotating vortex pairs are formed that create an
in-crossflow mixing is applied by the DLN Micromix Com- inner and outer recirculation zone during operation of the
bustion Principle, a gas turbine combustion system developed combustor (Fig. 3). Their proportions are designed to facilitate
at Aachen University of Applied Sciences (AcUAS) for flame stabilization and to prevent adjacent flames from
hydrogen fuel (JH2 ¼ 1). The fuel stream is injected perpen- merging with each other. Merging of adjoining flames would
dicular into the airflow and burned without premixing in a result in the formation of a reduced number of large-scale
multitude of miniaturized, diffusion-like flames. Premixing flames, increasing the residence time of NOx precursors in
offers the potential for low NOx emission due to the high the hot reaction zone and significantly promoting NOx
mixture and temperature homogeneity, but comes along with formation.
the major risk of flashbacks from the combustion into the This paper presents a survey of the interactive optimiza-
premixing zone. To eliminate the risk of flashbacks entirely, tion cycle used for the development of a new low emission
the DLN Micromix principle makes use of non-premixed Micromix combustor module for the application in hydrogen
combustion. Due to intense jet-in-crossflow mixing and fueled industrial-scale gas turbines (see Fig. 4).
miniaturization of flames, NOx emissions are effectively Based on an initial Micromix combustion chamber, already
reduced, even if pure hydrogen is applied as fuel. tested under full-scale gas turbine conditions [17], the opti-
In Refs. [14,15], Funke et al. have shown the great NOx mization targets an improvement of the combustion stability
reduction potential for lab scale and industrial scale gas tur- and combustion efficiency as well as a significant increase in
bine combustors with hydrogen fuel. Despite the high reac- the thermal power output per Micromix injector while main-
tivity of pure hydrogen, stable and safe operation under full- taining low NOx emissions.
scale gas turbine conditions is achieved by an appropriate For the initial Micromix combustion chamber the NOx
choice of combustion chamber and nozzle geometries. The characteristic at varying gas turbine load as well as the dy-
flame miniaturization effectively reduces NOx emissions namic operational behavior are presented. For optimization of
while the non-premixed combustion system is inherent safe the combustion process, experimental and numerical
against flashbacks. methods are applied. Experimental parameter studies of the
The combustion process is based on the phenomenon of identified low-NOx key parameters are conducted with test
jet-in-crossflow mixing of a fuel jet into a crossflow of accel- burners at the combustion chamber test stand at AcUAS.
erated air (cf. Fig. 1). By appropriate choice of the combustor Special focus is laid on achieving optimal combustion char-
geometry, an optimized mixing of fuel and air and aero- acteristics (ignition limits, flame stability, combustion effi-
dynamic flame stabilization is obtained. The miniaturization ciency) and low NOx emissions. The results facilitate the
of the flamelets leads to a significant reduction of NOx emis- validation of design laws for the combustion process and of
sions due to the reduced retention time of NOx precursors in the numerical approach. Combustion and flow simulations
the flame region. Due to the absence of flammable fuel-air- are used in the context of parametric studies for generating
mixtures prior to injection, the risk of flashbacks is avoided. optimized burner geometries and the phenomenological
In Fig. 2, the structural layout of a typical Micromix test burner interpretation of the experimental results. Based on the re-
for atmospheric testing is displayed. Fuel is distributed sults of the conducted parameter study, a optimized prototype
through the fuel supply segments and injected through small combustion chamber for the Auxiliary Power Unit (APU) GTCP
nozzles into a crossflow of air that is constricted and accel- 36-300 is derived and manufactured. It offers a low NOx
erated by the air guiding panels. characteristic at significantly increased thermal power output
The geometry of the fuel supply segments and the imple- per nozzle, which also reduces manufacturing complexity and
mented air guiding panels act as bluff bodies. In their wake costs.
Fig. 1 e Typical flame and flow field structure for Micromix combustion.
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 ) 6 9 7 8 e6 9 9 0 6981
Baseline design
Fig. 4 e Summary of the applied interactive optimization cycle for Micromix combustor development.
6982 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 ) 6 9 7 8 e6 9 9 0
Fig. 7 e Structural layout of the APU GTCP 36-300 combustion chamber and turbine in original configuration (top) and
equipped with a Micromix Module (bottom) [19].
Fig. 8 e Operational behavior of the GTCP 36-300 fueled with H2 in Micromix operation, indicated by the rotational speed
(solid lines) and the IGV position (dashed lines) for IDLE (a) and load changes from IDLE to MES (b) and back to IDLE (c) [18].
compressor negligible. The controller enables a stable IDLE initially the speed level overshoots and in the final stages
operation of the gas turbine with the integrated Micromix undershoots before stabilization at 99% speed level takes
combustion chamber at a target rotational speed of 99% with place. Finally, the controller stabilizes rapidly at a rotational
tolerable oscillations. The oscillations are comparable to speed of 99%. With the implemented Micromix combustion
hydrogen combustion with conventional fuel nozzles. chamber, the general operating behavior of the hydrogen
The switch form IDLE to MES (Fig. 8 b) and back to IDLE fueled APU GTCP 36-300 is in the desired range. Compress-
(Fig. 8 c) including the change of the speed level from 99% to ibility effects of the gaseous hydrogen are compensated by the
101% and back, represents the most challenging change in modified engine control software. The set PID-parameters of
operational conditions. By switching from IDLE to MES, the the on-speed governor allow a safe operation of the hydrogen
rotational speed stabilizes in the area of 100%e100.5% before gas turbine over the entire operating range.
achieving the requested speed level of 101%. The stabilization For evaluating the NOx-reduction potential of the Micromix
is caused by a load change, resulting from the IGV movement. combustion principle, exhaust gas samples are extracted
With the implemented Micromix combustion chamber, the during engine testing of the modified APU GTCP 36-300 and
controller is able to traverse rapidly, without over- or under- fed via heated tubing to the continuous gas analysis system
shooting of the rotational speed to MES. By switching back ABB Advanced Optima AO2020. The gas sample is processed
from MES to ILDE operation, the controller has to compensate by a gas dehydrator and transferred to the analyzing modules
the load compressor's decrease in load and to handle the by heated tubes and hoses under controlled pressure condi-
lowering of the rotational speed. From this follows that tions. The applied Advanced Optima exhaust gas analysis
6984 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 ) 6 9 7 8 e6 9 9 0
system determines the amount of unburned hydrogen (ABB number of nozzles decreased. This enhances the robustness
Caldos 27), the concentration of O2 (ABB Magnos 206) and the against fuel contamination by particles and, in combination
concentrations of CO and CO2 (ABB Uras 26). For the deter- with an optimized combustion chamber design, reduces the
mination of NOx (i.e. NO and NO2), an Eco Physics CLD 700 EL is manufacturing complexity significantly. Increasing the ther-
used and directly connected to the hot exhaust gas sample. mal loading leads to the formation of larger flames that benefit
The cross-sensitivity to the remaining water vapor in the combustion efficiency but increase the risk of higher NOx
sample is below 0.5% of the measured value. Before each test emissions.
campaign, all exhaust gas-analyzing devices are calibrated Considering these effects, numerical and experimental
using zero-point and defined reference-point calibration parameter studies are conducted under atmospheric condi-
gases. An analysis of error sources and their impact on mea- tions to identify a combustor geometry at up-scaled thermal
surement accuracy and reproducibility is given in Ref. [20]. power output per nozzle with optimized combustion charac-
Fig. 9 presents the measured NOx-emissions against APU teristics and low NOx emissions. For analyzing the combus-
power output for different types of combustors and fuels. For tion principle's main geometric parameters influencing flame
comparison, the figure contains the resultant NOx-emissions stabilization, combustion characteristics and NOx formation,
when operating the same engine with conventional kerosene numerical analyses are carried out using the commercial CFD
nozzles, with six conventional hydrogen nozzles and with the code STAR-CCMþ. The 3D combustion simulations base on a
baseline Micromix hydrogen combustion chamber. Due to the simplified geometric model derived from the experimentally
high reactivity of hydrogen, NOx emissions increase signifi- investigated test-burner configuration.
cantly when changing from kerosene to hydrogen operation As displayed in Fig. 10, the geometric model benefits from
with six conventional gas nozzles. The conventional nozzles the symmetric nature of the burner in both lateral and vertical
create six large flames with high retention times of NOx pre- direction. A longitudinal burner slice, containing half of a
cursors in the high temperature flame regions, which pro- hydrogen injector and half of an air guiding panel gate is
motes NOx formation. When miniaturizing the flames and simulated. The spatial discretization is performed using an
enhancing mixing of fuel and oxidizer by Micromix combus- unstructured polyhedral mesh with prism layers and solution
tion, a NOx reduction potential of up to 95% is possible at the adaptive mesh refinement along the reaction and flow path
same operating conditions. [21]. Mesh independency has been established with a final
The implemented Micromix combustion chamber and the mesh size of approx. 540 k volume cells. With two mass flow
realized hardware and software modifications allow the inlet boundaries, schematically shown in Fig. 10, the fuel- and
secure and low NOx hydrogen operation of the APU GTPC 36- the dry air-jet (rO2 ¼ 0.20950; rN2 ¼ 0.79016; rCO2 ¼ 0.00035) are
300. The possibility is given to operate the hydrogen-driven introduced separately into the combustor model far enough
engine in the same way as in kerosene operation with no upstream of the combustion zone to avoid any boundary in-
CO/CO2 emissions and significantly reduced NOx emissions. fluences on the mixing and combustion processes. Inlet
boundary conditions are defined according to experimental
inlet conditions of the investigated combustor configuration
Numerical and experimental optimization (p3 ¼ 1 bar; T3 ¼ 560K; TFuel ¼ 298.15 K). The model offers a fluid
region for the combustion and flow simulation and solid re-
After achieving stable operation of the Micromix combustion gions that account for conjugate heat transfer through the
principle under full-scale gas turbine conditions and estab- combustor walls.
lishing the NOx reduction potential, up-scaling to higher For analyzing the combustion characteristics over the
thermal loading of each injector nozzle with regard to indus- entire operating range, the design point of the combustor
trial gas turbine applications is the focus of the presented (F ¼ 0.375) and off-design points at overload and part-load
optimization cycle. For increasing the thermal power output conditions are analyzed. The equivalence ratios are set at
per nozzle, the nozzle diameter is increased and the total constant air mass flow by adjusting the fuel mass flow
accordingly. The reactive flow regime is solved by a three-
dimensional, steady, pressure based RANS solver using the
realizable k-ε-turbulence model with all yþ wall treatment.
The combustion process is solved by a complex chemistry
model using detailed reaction mechanisms, described in
Ref. [16]. To account for turbulence-chemistry interactions,
the Eddy Dissipation Concept (EDC) is applied [22]. A
comprehensive kinetic reaction mechanism composed of 19
reversible elementary reactions and 9 species, proposed by Li
et al. [23] is used to solve the reactive flow regime. Due to the
highly turbulent nature of the Micromix combustion principle,
transport phenomena are predominantly driven by turbulent
diffusion. In a first approach molecular diffusion and associ-
ated Lewis number effects are treated by the simplified
approach of unity Schmidt number. Density and dynamic
Fig. 9 e NOx emissions of different combustor technologies viscosity are calculated for the overall mixture in the indi-
and fuel types against APU power output. vidual cells.
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 ) 6 9 7 8 e6 9 9 0 6985
micro-flames. At every location, an exhaust gas sample is numerically investigated in Ref. [14]. A design trend is
taken when steady state conditions are reached. For each established for the transition between the initial geometry
equivalence ratio, the arithmetic mean based on measure- with a nozzle diameter of 0.3 mm and the final geometry
ments at a defined number of representative locations at a with a nozzle diameter of 0.84 mm. The initial geometry
distance of 90 mm in z-direction from the test-burner fuel offers very low NOx emissions at the design point and part
supply segment is obtained. load conditions. At overload, NOx emissions increase dras-
The extracted samples are supplied to the analysis mod- tically. The slope can be reduced by application of larger
ules of the continuous gas analysis system ABB Advanced nozzles with higher thermal loading per nozzle at the price
Optima AO2020 as described for full-scale gas turbine testing. of increased part load emissions. These are due to the larger
Fig. 14 shows the measured NOx emissions of three generated flames and the increased residence time of NOx
different test burner configurations that have been precursors in the larger flame region.
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 ) 6 9 7 8 e6 9 9 0 6987
Fig. 14 e NOx emissions against equivalence ratio for nozzle diameters between 0.3 mm and 0.84 mm.
When balancing the slightly increased NOx emissions at evaluated. Measurement accuracies of the gas analyzing
the design point and part-load with the improved overload hardware are indicated in the plots as error bars. The un-
behavior, a significant reduction of manufacturing burned fuel emission levels are depicted in Fig. 16. The
complexity and the enhanced robustness against fuel resulting emissions of unburned hydrogen correspond to a
contamination by particles, the step towards a nozzle diam- combustion efficiency of more than 99% throughout the
eter of 0.84 mm is justified. As indicator of the general flame evaluated operating range. When comparing the simulation
structure and flame length for the optimized test burner ge- and experimental results displayed in Fig. 16, the simulated
ometry with a nozzle diameter of 0.84 mm, the simulated exhaust emissions fall considerably below the experimental
temperature distribution on the central symmetry plane is results. By application of steady RANS-simulations, unsteady
given in Fig. 15. flow phenomena like flame quenching due to local aero-
The 0.84 mm design results in a conventional Micromix dynamic effects cannot be resolved adequately. Furthermore,
flame, despite an increase in thermal power output per the applied geometrical model shows no imperfections or
injector by approx. 390% in comparison to a 0.3 mm design. deviations from the base layout that occur within certain
The flame is aerodynamically stabilized in the shear layer tolerances during the manufacturing and assembly process.
between the vortices in the wake regions of the air guiding The numerically evaluated NO and experimentally
panel and the fuel segment. Cooling of the outer recirculation measured NOx emissions are presented in Fig. 17 for part-,
vortex is achieved by a coflow of air that bypasses the flame full- and overload conditions. The Micromix combustor
and enters the recirculation area. module achieves NOx emissions less than 2.5 ppm corrected
In the following, the measured and simulated trace gas to 15 Vol% O2 at the design point and emissions below
emissions of unburned fuel components and NOx are 4 ppm at overload.
Micromix combustor for industrial gas turbine applications. J [23] Li J, Zhao Z, Kazakov A, Dryer FL. An updated comprehensive
Therm Sci Eng Appl 2019;11(1). kinetic model of hydrogen combustion. Int J Chem Kinet
[22] Magnussen BF. The Eddy dissipation concept a bridge 2004;36(10):566e75.
between science and technology. Trondheim, Norway: [24] Lefebvre AH, Ballal DR. Gas turbine combustion: alternative
Norwegian University of Science and Technology fuels and emissions. 3rd ed. Boca Raton: Taylor & Francis;
Trondheim; 2005. 2010.