CO-FIRING OF HYDROGEN AND NATURAL GASES IN LEAN PREMIXED CONVENTIONAL AND REHEAT BURNERS (ALSTOM GT26) Felix Güthe Alstom Baden, Switzerland

Proceedings of ASME Turbo Expo 2014: Turbine Technical Conference and Exposition GT2014 June 16 20, 2014, Düsseldorf, Germany GT2014-25813 CO-FIRING OF HYDROGEN AND NATURAL GASES IN LEAN PREMIXED CONVENTIONAL AND REHEAT BURNERS (ALSTOM GT26) Torsten Wind Alstom Baden, Switzerland Felix Güthe Alstom Baden, Switzerland Khawar Syed Alstom Baden, Switzerland ABSTRACT Addition of hydrogen (H 2 ), produced from excess renewable electricity, to natural gas has become a new fuel type of interest for gas turbines. The addition of hydrogen extends the existing requirements to widen the fuel flexibility of gas turbine combustion systems to accommodate natural gases of varying content of higher hydrocarbons (C2+). The present paper examines the performance of the EV and SEV burners used in the sequential combustion system of Alstom s reheat engines, which are fired with natural gas containing varying amounts of hydrogen and higher hydrocarbons. The performance is evaluated by means of single burner high pressure testing at full scale and at engine-relevant conditions. The fuel blends studied introduce variations in Wobbe index and reactivity. The latter influences, for example, laminar and turbulent burning velocities, which are significant parameters for conventional lean premixed burners such as the EV, and auto-ignition delay times, which is a significant parameter for reheat burners, such as the SEV. An increase in fuel reactivity can lead to increased NO x emissions, flashback sensitivity and flame dynamics. The impact of the fuel blends and operating parameters, such as flame temperature, on the combustion performance is studied. Results indicate that variation of flame temperature of the first burner is an effective parameter to maintain low NO x emissions as well as offsetting the impact of fuel reactivity on the auto-ignition delay time of the downstream reheat burner. The relative impact of hydrogen and higher hydrocarbons is in agreement with results from simple reactor and 1D flame analyses. The changes in combustion behaviour can be compensated by a slightly extended operation concept of the engine following the guidelines of the existing C2+ operation concept and will lead to a widened, safe operational range of Alstom reheat engines with respect to fuel flexibility without hardware modifications. INTRODUCTION Due to the progress in installation of sustainable energy from e.g. wind turbines or solar with unforeseeable power output, several potential energy carriers are in development. The use of hydrogen (H 2 ) as energy storage, produced by electrolysis powered by surplus sustainable energy, gives the challenge to burn a hydrogen blend natural gas in an Alstom gas turbine. The existing natural gas infrastructure could be used for storage of a small amount of hydrogen. The actual limit of 5 vol.-% in the natural gas pipelines could be extended to 15 vol.-% in the mid-term [1]. Furthermore a local electrolysis device with H 2 storage is feasible to increase the sustainable proportion of the fuel within the operational limit of the local gas turbine. Also fuel gas mixtures from biomass or coal gasification require higher fuel flexibility regarding H 2. Extensive single burner high pressure tests at full scale were performed for existing GT26 standard premix and standard reheat burners (GT26 upgrade 2006 and upgrade 2011 [1]) with 15 to 60 vol.-% H 2 -doped natural gas. Emissions and flame instabilities were monitored and several flame position tests were carried out. Aspects of fuel flexibility have been highlighted for the Alstom products and it is perceived as one of the strengths of the Alstom combustion technology. This has been demonstrated several times [2][3]. Fuel flexibility as a feature can offer a competitive advantage. The limits for allowed fuels are determined by several factors and several components in the plant. ALSTOM 2014. All rights reserved. Information contained in this document is indicative only. No representation or warranty is given or should be relied on that it is complete or correct or will apply to any particular project. This will depend on the technical and commercial circumstances. It is provided without liability and is subject to change without notice. Reproduction, use or disclosure to third parties, without express written authority, is strictly prohibited. 1 Copyright 2014 by Alstom Technologie AG

The fuel has to be brought to the GT and controlled within the different fuel lines via the fuel distribution system (FDS). Constraints to the fuels are therefore given by the required pressure at the combustor, which is given by the pipeline and might require expensive additional compression. Condensation of low boiling components has to be avoided. The supply pressure has to be sufficient to allow flow through the fuel injectors into the burner with sufficient momentum to allow good fuel-air mixing. In the combustor finally the composition of the fuel determines the chemical reactivity of the fuel in the combustion process, which, for safety reasons, has to be between the limits of flashback and blow out. Increased NO x production of H 2 containing natural gas flames have been reported before [4]. The composition also influences the emissions of the combustor with respect to NO x and CO emissions, which is determined by the reactivity via the flame position and the degree of premixing and eventually the NO x production chemistry of the fuel itself. While the direct chemical impact on (prompt) NO x by reaction with the fuel plays a minor role (except for trace species like fuel bound nitrogen) the flame shape and position influence the emission noticeably. The parameters on engine level constraining the fuels are therefore the dew point of the fuels, the pressure loss between pipeline and the injector, the jet penetration in the burner as well as the chemical reactivity in the flames of the corresponding combustors. The comparison of different fuels with respect to the pressure requirements is established by the Wobbe index, which assures that a given injector with a given pressure drop is capable of delivering a given energy content through nozzles of given dimensions at given jet penetration. The Wobbe index therefore contains the energy content and the density of the fuel. LHV Chemical reactivity can be assessed by chemical kinetics simulations as explained in a later section. The change of Wobbe index by doping with hydrogen or C2+ is shown in Figure 2, natural H-gas provides a basis for fuel mixtures, which were used for high pressure single combustor tests. NOMENCLATURE C2+ - Higher Hydrocarbons EV burner - EnVironmental burner FDS - Fuel distribution system GT - Gas Turbine LHV mass MJ/kg Lower Heating Value p Pa Pressure PSR - perfectly stirred reactor PFR - plug flow reactor PREMIX - premixed laminar flame model S1R - Staging ratio SEV - Sequential EnVironmental S L - Laminar Flame speed S T - Turbulent Flame speed T K Temperature TAT1 C Temperature After Turbine (High Pressure) TAT2 C Temperature After Turbine (Low Pressure) TFM K SEV Burner Inlet Temperature for fulfilling Flashback margin THG K Hot Gas Temperature at Combustor Exit Tin K Inlet Temperature VG - Vortex Generator WI - Wobbe Index WRL - Wide Range Logic for C2+ ρ kg/m 3 density φ - Equivalence ratio ALSTOM GT 26 The sequential combustion system of the GT26 gas turbine consists of primary combustor (EV) with 24 circumferential EV burners and lances, followed by a turbine and a re-heat combustor (SEV) with 24 burners and lances. A cross section of the GT26 gas turbine is shown in Figure 1. The advantage of the Alstom GT26 gas turbine with its reheat technology compared to single stage combustion is an additional degree of freedom balancing the power of the two combustion chambers. The Alstom GT26 C2+ operation concept utilizes this degree of freedom to optimize combustion behavior for fuels containing C2+ [5][6]. Due to the fact that H 2 and C2+ have a similar behavior in terms of reactivity, the C2+ operation concept can also be applied to H 2 fuel mixtures, which will be shown later in detail. Figure 1: Alstom GT24/GT26 sequential combustion system Part of the fuel is injected within the EV burners into the swirling air and thoroughly mixed before entering the EV combustor. The premix flame is stabilized at the exit of each EV burner by a recirculation zone due to vortex breakdown. The reaction products from the EV combustor are expanded in a single stage high-pressure turbine and subsequently enter the SEV combustor. The temperature at the SEV inlet is defined as TAT1. The fuel is injected into hot gases in the SEV burner and 2 Copyright 2014 by Alstom Technologie AG

well mixed before entering the SEV combustor, which is operating at auto-igniting conditions not requiring an external igniter. The fuel amount between both combustors is optimized for each operating point and approximately equal at full load with natural gas. The WI min and WI max limits show the Wobbe Index range, in which the GT26 can be operated (apart from other fuel restrictions). The intended H 2 doped fuels are within that Wobbe Index range with and without fuel preheating. To take into account the enhanced reactivity of the H 2 containing fuels in a turbulent flame the molecular transport properties of the low Lewis number have to be taken into account [9]. The correction for turbulence accounts for an averaged Lewis number effect and has been described in [10]. The data are normalized to natural gas at 1700K. To maintain an average reactivity of that reference condition (and keep a similar flame position) the temperature would have to be adjusted accordingly. The turbulent reactivity is only estimated here since not all the characteristics of the flow field are accounted for. The relative difference appears to change more with fuel reactivity than the laminar flame speed alone. Validation of the work on turbulent and laminar flame speed is ongoing [11]. Figure 2: Wobbe Index for H 2, C2+, natural gas mixtures The fuel reactivity is increasing (laminar flame speed, auto-ignition delay time) with increase of H 2 / C2+. INVESTIGATION OF CHEMICAL REACTIVITY The challenge for the combustion system for highly reactive fuels can be described and evaluated by chemical kinetics studies. The chemical reactivity can be assessed by chemical kinetics simulations, where usually the laminar flame speed, which takes into account the molecular transport properties of the fuel, is determined for the premix flame (first combustor-ev) or the minimum residence time of a perfectly stirred reactor with heat release (PSR extinction time). To account for the fuel effects that are investigated here, turbulent flame speeds are estimated as a measure of reactivity in the burner as describes elsewhere [11], [12]and [13]. To access assess the sequential combustor (SEV) of the reheat engine [3] auto-ignition delay times (from a plug flow reactor- PFR) are used. To obtain reliable trends the use of a properly validated reaction scheme is crucial. This validation has been part of a research program over several years and has been reported in [7], [10], [11]. The kinetic mechanism was chosen from Curran s group at national university of Gallway (NUIG). For the current paper version NUIG_54.1 was used. The flame position of the first combustor of the GT24/GT26 is determined by the flow field and the chemical reactivity in the turbulent field. A relative comparison of fuels reactivity can be done based on the laminar flame speeds as shown in Figure 3 near GT26/GT24 base load conditions. The flame speeds are calculated as function of flame temperatures (varying φ) at lean conditions near GT operation for various model fuels as indicated. Figure 3: Effect of H 2 addition to the normalized laminar flame speed (top) and turbulent flame speed (middle) and PSR extinction time( bottom) calculated for different flame temperatures and various H 2 and C2+ model fuels as indicated at conditions near GT26 Base load. The reference case is indicated by the red square. 3 Copyright 2014 by Alstom Technologie AG

The flame speeds depend not only on H 2 content but also on the equivalence ratio φ of the mixture of premixed flame. They are shown in a 3D colormap of φ vs. H 2 content in Figure 4 for hydrogen and natural gas mixtures in terms of turbulent flame speeds. The red solid lines indicate the maximum flame speed for given fuel composition. Note that the maximum flame speed is on the rich side near φ ~1.1 for natural gas (red line) and is shifting towards higher equivalence ratios for high H 2 contents. The technical challenge is that the highest risk at high reactivity is at the fuel rich side near the injection, which has to be considered for evaluation of the flashback risk. Figure 5: Effect of H 2 addition to the normalized autoignition time calculated for various SEV inlet temperatures and various H 2 and C2+ model fuels as indicated at conditions near GT26 Base load. EXPERIMENTAL SETUP Full scale engine hardware tests at full engine pressure with H 2 co-firing were performed in the high pressure test facility at DLR Cologne, Germany. For EV and SEV two different sector test rigs were used to simulate the gas turbine conditions as close as possible. The air conditions (e.g. pressure and temperature) at compressor exit of a gas turbine can be adjusted. Different fuel gas compositions can be mixed before fuel injection. Figure 4: Contour plot of turbulent flame speeds (using NUIG_54_1 mechanism) for natural gas/ hydrogen mixtures (horizontal axis) and different equivalence ratios φ (vertical axis). The red line indicates the maximum flame speed for each fuel composition The reactivity of the sequential burner (SEV) is evaluated by calculating auto-ignition times, as shown in Figure 5. The relevance of the calculated auto-ignition times to burner performance within the SEV high pressure data, is demonstrated by the good match of the measured near flashback operation points as shown later (Figure 13) for all tested fuels. The trends of the SEV reactivity (Figure 5) are very similar to the trends for the EV (Figure 3) showing the highest reactivity for 60% H 2 and showing C2+ and H 2 to have comparable effects. GT26 operation with higher reactivity fuel requires a re-balancing of the power from EV to the SEV combustor. For fuel gas with varying C2+ content, Alstom has already a good experience with the wide range logic. This optional GT26 logic allows operation with C2+ content up to 18 vol.-% C2+. EV test rig The EV burner of GT26 upgrade 2006 and 2011 consists of a conical swirler and a long lance. The fuel gas is divided between two stages, as shown in Figure 6. Stage 1 fuel is injected through the lance and stage 2 fuel is injected along the swirler slots. Both stages are continuously in operation, the ratio is adjusted to ensure flame stability and low emissions. The flame is stabilized by a recirculation region generated by vortex breakdown. Figure 6: Staged EV burner with long lance Figure 7 shows the single burner high pressure test rig in flow direction from left to right, in which the EV burner was tested. Emissions were measured at plane 4. Plane 1 and plane 4 were used for optical access. Pulsations were measured at plane 2. 4 Copyright 2014 by Alstom Technologie AG

Figure 7: EV single burner high pressure test rig; standard rig instrumentation SEV test rig The SEV burner and SEV lance are shown in Figure 8. Two different sets of hardware of GT26 upgrade 2006 and upgrade 2011 [1] were tested. The combustion products of the first combustor, mixed with cooling air, enter the burner and pass through the four delta wing vortex generators (VGs), which enhance the mixing. The lance is located directly after the VGs. The fuel is injected perpendicularly, shielded by carrier air, through four nozzles at the lance tip. The hot gas and fuel are then premixed in the mixing zone until finally autoignition takes place in the combustion chamber after sudden expansion. Figure 8: SEV burner (left), SEV lance (right) The described hardware was tested in a reheat test rig [1], shown in Figure 9. A single premix (EV) burner serves as hot gas generator for the SEV burner, providing the requested SEV inlet conditions. A transition piece with blockage rods is in between for decoupling the combustion chambers. These blockage rods also contain the emission probes, for detection of the SEV inlet components. The tested SEV burner and SEV lance are installed after the transition piece. The SEV emissions are measured at the end of the SEV combustor. Figure 9 shows the SEV high pressure test rig with flow direction from left to right. Figure 9: SEV single burner high pressure test rig The SEV combustion chamber features an inner and outer segment like in the gas turbine. The high pressure test rig was operated at nominal GT26 base and part load conditions and at specific off-design conditions for flame position testing. RESULTS The high pressure tests were mainly used to show influence on NO x and flashback margin for higher reactivity fuel like H 2 -doped natural gas and C2+ fuels. EV single burner high pressure test results The high pressure test showed increased flashback robustness of EV burner with staged long lance (upgrade 2006 and 2011) to burn high C2+ and H 2 co-firing mixtures. No signs of flashback appeared at base load condition at a pressure level of 30 bar with a fuel mixture of 45 vol.-% H 2 and 55 vol.-% NG. At part load condition at a pressure level of 16 bar up to 60 vol.-% H 2 were tested. Only during one test at off-design conditions with higher hot gas temperature first flashback indications occurred. The optical device showed sporadic flashback but no indication at the burner thermocouples could be seen. This test showed that flashback risk for up to 30 vol.-% H 2 doped NG for the GT26 is low. Pulsations at idle operation are reduced by increasing the H 2 content in the fuel. For the hot gas mappings the fuel split of stage 1 and 2 was adjusted to minimize the NOx emissions for each type of fuel. The mappings were performed at engine part and base load conditions. Figure 10 shows the NO x emissions over the hot gas temperature for different H 2, C2+ and natural gas mixtures at base load conditions. The black curve shows the reference case with natural gas (7 % C2+). The blue (15% H 2, 6% C2+) curves are below the light grey (18% C2+) curve. This means that the EV NO x emissions of 15% H 2 -doped natural gas do not exceed the NO x emissions of natural gas with 18 % C2+. 5 Copyright 2014 by Alstom Technologie AG

Figure 10: EV NO x over TAT1 for different H 2, C2+, natural gas mixtures at base load pressure level Figure 11 shows the interpolation of the required hot gas temperature reduction to maintain 5 ppm NO x @ 15 % O 2 versus the volume fraction of H 2 + C2+. Fuel mixtures with up to 15 vol.-% H 2 (6 vol.-% C2+) require less reduction of hot gas temperature compared to fuel with similar volume percentages of C2+. Therefore the extension of the Alstom C2+ operation concept (Wide Range Logic / WRL) to H 2 (limited to 9 % C2+) can be used to control EV NOx. At volume fractions of H 2 and C2+ higher than 30%, C2+ fuels require less hot gas temperature reduction compared to H 2 -doped fuels. Flashback analysis method: The propensity for SEV flashback is a major concern when combusting highly reactive fuels. The SEV combustor is operating at auto-igniting conditions. Shorter auto-ignition delay times, respectively higher inlet temperatures and the increase of fuel reactivity shift the flame closer to the burner. To operate with a constant safety margin the inlet temperature has to be altered for reactive fuel. An experimental approach to this is given by the flashback analysis method described here. For a given fuel the test is started from a safe operating point at low TAT1 (SEV -T inlet ) using a burner instrumented with thermocouples. A point of equal reactivity (still with margin to damaging flashback) is defined by the thermocouple being influenced by the flame itself. This is approached by raising TAT1 until the slope of the temperature sensor increases, indicating the presence and direct heat impact of the flame. The difference between the smoothened temperature measurement and the simulated temperature (from a heat balance model) of the burner exit wall were plotted over time, or equivalently over the inlet temperature, see Figure 12. Figure 12: Difference between the smoothened temperature measurement and the simulated temperature of the burner exit wall as function of SEV inlet temperature for different fuel types during flame position tests Figure 11: Delta flame temperature for EV NO x over H 2 + C2+ content. Interpolation of required flame temperature reduction for constant NO x of 5 ppm @ 15 % O2 at GT26 base load conditions for different fuel types SEV single burner high pressure test results Off-design flame position tests were performed at base load pressure level for both hardware of upgrade 2006 and upgrade 2011 [1] with several fuel gas mixtures. The SEV fuel mass flow was kept constant during that test. The SEV inlet temperature was set to a lower starting value. The inlet temperature was increased at a constant rate until critical conditions were reached; all other parameters were held constant. The base case of testing was with fuel natural gas. All other cases were compared to it. While increasing TAT1 by a given rate at a certain inlet temperature the measured values start to deviate from the heat balance simulation (increases faster than the air temperature), which shows that heat radiation from the flame influences the thermocouple at the burner exit. In that case the heat release zone is expected to be at the burner exit and the flashback margin is very narrow. The temperature at the difference of 10 K defines the inlet temperature for comparison of flashback margin (TFM) of the different types of fuel. This temperature is the basis for the required TAT1 de-rating (with respect to the base fuel) which is shown below for both SEV hardware types independently. The tests yield a set of fuel compositions and operating conditions (TAT1) of comparable reactivity. To compare this with the theoretical approach a simple 1 D model is used. The assumption is that the auto ignition time describes the reactivity sufficiently. The auto-ignition delay 6 Copyright 2014 by Alstom Technologie AG

times at TFM, which refers to the same reactivity or to the same simulated ignition time, were calculated. These auto-ignition delay times for all fuels (containing H 2, C2+) at TFM were converted into a distance by multiplying by the mean burner velocity and plotted as probability distribution (Figure 13). All calculations for all tested fuels fall in a very narrow range (accuracy of < 10%) and lie close to the measured physical position of the thermocouple used for the analysis. For visualization the measured values are plotted on top of a Gaussian probability distribution. The procedure proves the relevance and the reliability of the PFR modeling for flashback safety prediction and NUIG kinetic model to describe the engine condition. It combines the experimental approach with a proper kinetic model to predict engine behavior. The corresponding analysis using the GRI3.0 model was carried out and resulted in very scattered results (STD +-25% and a total spread of 100%) for the different fuels, highlighting the importance of proper chemical kinetics basics to operate the reheat gas turbine. Figure 14: Required TAT1 reduction for fulfilling sufficient flashback margin vs. H 2 + C2+ content (upgrade 2006). Measured de-rating and simulated (NUIG) de-rating, reference natural gas with 7 vol.-% C2+ The NO x vs. flame temperature curves of different fuel types and different SEV inlet temperatures are very close together. NO x emissions of natural gas doped with 15 vol.-% H 2 up to 45 vol.-% H 2 and also 33 vol. % C2+ are comparable and 40 % higher than the reference case with NG with 6 vol.-% C2+, see Figure 15. Figure 13: Histogram of ignition distance calculated from the flashback condition for different fuels and burner temperature (log spacing) With the experimental verification of the prediction methodology (using validated kinetics) the operation of the highly reactive fuels can be adjusted. SEV hardware upgrade 2006: The required TAT1 de-rating for H2, C2+ and natural gas mixtures, which maintains the flashback margin of reference natural gas is shown in Figure 14 for SEV hardware upgrade 2006. Up to 30 vol.-% H 2 + C2+ the flashback margin is higher for the same amount of total H 2 + C2+, the higher the H 2 /C2+ ratio. Figure 14 shows the normalized required TAT1 de-rating for fuel variation based on flashback tests (symbols) and based on chemical kinetic simulations (curves). Figure 15: SEV NOx emissions vs. SEV THG for different fuel types of GT26 upgrade 2006 SEV hardware upgrade 2011: The flashback margin of upgrade 2011 hardware for natural gas is already higher than for upgrade 2006 hardware. Therefore operation with higher TAT1 is possible compared to upgrade 2006. The entire TAT1 scale is shifted to a higher temperature. Beside that bonus the required TAT1 de-rating with natural gas as reference is shown in Figure 16. Up to 30 vol.-% H 2 + C2+ the flashback margin is higher for the same amount of total H 2 + C2+, the higher the H 2 /C2+ ratio. Figure 16 shows the normalized required TAT1 de-rating for fuel variation based on flashback tests (symbols) and based on chemical kinetic simulations (curves). 7 Copyright 2014 by Alstom Technologie AG

Figure 16: Required TAT1 reduction for fulfilling sufficient flashback margin vs. H 2 + C2+ content (upgrade 2011). Measured de-rating and simulated (NUIG) de-rating, reference natural gas with 7 vol.-% C2+ The effect of 15 vol.-% H 2 doping on NO x is small or even negligible at engine conditions. The compensation of increased SEV NO x for higher H 2 concentration by decreasing the SEV inlet temperature is limited. An overview of NO x for different type of fuels and inlet temperatures is given in Figure 17. on the fact that up to 30 vol.-% H 2 + C2+ the flashback margin is higher and the NOx is lower for the same amount of H 2 + C2+ the higher the H 2 /C2+ ratio. This dependency is switched at around 30 vol.-% H 2 + C2+. Limitations for EV and SEV operation are SEV flashback and overall NO x emissions, the EV flashback risk can be neglected, as it was described above. The SEV flashback margin has to be sufficient for all fuel mixtures. In general fulfilling the NO x emission limit of 15 ppm @ 15 % O 2 is a more stringent requirement than fulfilling a sufficient flashback margin. As the same EV burner is installed in GT26 upgrade 2006 and 2011, the operation with EV only is shown below for both engines. Due to the higher reactivity of H 2 -doped natural gas, the effect of piloting is increased and the flame is much more stable in general, the S1R has to be adapted accordingly. For doing that, H 2 can be weighted as C2+ and the Wide Range Logic [5][6] for C2+ can be used. The SEV could be switched on earlier the higher the H 2 content due to the fact that part load CO is lower the higher the H 2 content. Two scenarios for the modification of the operation concept for H 2 doped natural gas are possible. Scenario 1 stands for operation with constant NOx and Scenario 2 for operation with similar SEV flame position in comparison to the reference case with natural gas. For operation with SEV on, this section is divided into GT26 upgrade 2006 and 2011, because of different hardware used. Operation of GT26 upgrade 2006 GT26 upgrade 2006 operated with 15 vol.-% H 2 -doped natural gas (15 vol.-% H 2, 6 vol.-% C2+) at design conditions relates to a SEV NO x increase of about 1.1 g/kg fuel. At the same time the EV NO x is increased by 0.5 g/kg fuel. For scenario 1 with constant NO x level compared to the reference case with natural gas at design conditions the reduction of TAT1, which is also close to the required flashback margin (scenario 2). Operation at constant NO x level with higher H 2 content than 15 vol.-% is not possible with GT26 upgrade 2006 without reduction of TAT2, and consequently loss of efficiency. Figure 17: SEV NO x emissions vs. SEV THG for different fuel types of GT26 upgrade 2011 ADAPTION OF GT OPERATION CONCEPT The operation with H 2 mixtures should have minimal impact on performance and emissions. Decrease of TAT1 for operation with H 2 doping is acceptable due to the fact that the power and efficiency reduction is minor in Alstom s reheat engines. Decrease of TAT2 should be avoided due to large reduction of power output of the gas turbine and the whole combined cycle power plant. A sufficient flashback margin has to be guaranteed. For H 2 co-firing up to 30 vol.-% H 2 an adapted C2+ Wide Range Logic [5][6] can be used as conservative operation based Operation of GT26 upgrade 2011 The SEV NO x of the GT26 upgrade 2011 hardware is lower compared to the upgrade 2006 hardware. Also the flashback risk is lower, as described in previous sections. The flashback margin of 15 % H 2 -doped natural gas (15% H 2, 6 % C2+) is sufficient already at design conditions, no de-rating would be required for operation with slightly higher NO x, but still below 15 ppm -@ 15 % O 2. For constant NOx (scenario 1) compared to the reference case with natural gas a reduction of EV hot gas temperature is required. Operation with 30 vol.-% H 2 -doped natural gas is still feasible with same gas turbine NO x level compared to the reference case with natural gas by further reduction of TAT1. 8 Copyright 2014 by Alstom Technologie AG

CONCLUSION AND OUTLOOK The combustors of Alstom GT26 upgrade 2011 were tested successfully with H 2, natural gas mixtures. Standard hardware can be applied for H 2 co-firing up to 30 vol.-% H 2. The existing Alstom Wide Range Logic for C2+ operation [5][6] can be adapted to mixtures of natural gas, C2+ and H 2. GT26 upgrade 2006 can be operated with up to 15 vol.-% H 2 -doped natural gas and GT26 upgrade 2011 can be operated with up to 30 vol.-% H 2 -doped natural gas without change of hardware. No reduction of exhaust temperature is required and therefore the power reduction respectively the loss of efficiency caused by EV de-rating is minor. In case of the GT26 upgrade 2011 operation with 15 vol.- % doped natural gas below the NO x level of 15 ppm @ 15 % O 2 is possible without de-rating of the EV. A demonstration plant for H 2 storage and co-firing of GT26 upgrade 2011 with 30 vol.-% H 2 doped natural gas could show the capability for storage of the excess renewable electricity and grid stabilization using fast-responding combined cycle power plant operated with H 2 -doped natural gas. Further tests and minor adaptions of gas turbine hardware are required for operation with up to 45 vol.-% H 2 -doping. Supporting the EU 2020 target of CO 2 reduction, operation with 45 vol.-% H 2 based on sustainable energy would result in 20 % reduction of CO 2 emissions. The Wobbe Index of intended H 2 doped fuels is within the specified range of the GT26. REFERENCES [1] K. M. Düsing, A. Ciani, U. Benz, A. Eroglu, K. Knapp, Development of GT24 and GT26 (upgrades 2011) Reheat Combustors, achieving reduced emissions and increased fuel flexibility, ASME Turbo Expo, GT2013-95437. [2] DVGW Mit Gas-Innovationen in die Zukunft (2012) [3] F. Güthe, J. Hellat, P. Flohr, "The reheat concept: the proven pathway to ultra-low emissions and high efficiency and flexibility", Journal of Engineering for Gas Turbines and Power, 131, 021503 (2009). [4] Therkelsen, P., Werts, T., McDonell, V. Samuelsen, S., Analysis of NOx Formation in a Hydrogen Fueled Gas Turbine Engine, Paper No. GT2008-50841, ASME Turbo Expo 2008, Berlin, Germany, June 9 13, 2008 [5] Oliver Riccius, Richard Smith, Peter Flohr, Felix Güthe"The GT24/26 Combustion Technology and high Hydrocarbon ( C2+ ) Fuels", GT2005-68799. [6] Matthias Hiddemann, Mark Stevens, Frank Hummel Increased operational flexibility from the GT26 (2011) upgrade, Power-Gen Asia (2012) [7] M. Brower, E. Petersen, W. Metcalfe, H. Curran, N. Aluri, F. Güthe, M. Füri, G. Bourque, "Ignition Delay Time and Laminar Flame Speed Calculations for Natural Gas/Hydrogen Blends at Elevated Pressures ", Journal of Engineering for Gas Turbines and Power, 135, 021504-1 (2013). [8] N. Donohoe, A. Heufer, W. K. Metcalfe, H. J. Curran, M. L. Brower, O. Mathieu, E. L. Petersen, G. Bourque, F. Güthe, Ignition delay time experiments and mechanism validation for natural gas/hydrogen blends at elevated pressures", submitted to Combustion and Flame, (2013). [9] Lipatnikov AN, Chomiak J. Molecular Transport Effects on Turbulent Flame Propagation and Structure. Prog. Energy Combust. Sci. 2005;31:1-73. [10] M. L. Brower, O. Mathieu, E. L. Petersen, N. Donohoe, A. Heufer, W. K. Metcalfe, H. J. Curran, G. Bourque, F. Güthe, "Ignition Delay Time Experiments For Natural Gas/Hydrogen Blends At Elevated Pressures ", ASME Turbo Expo, GT2013-95151, (2013). [11] A. Morones, S. Ravi, D. Plichta, E. L. Petersen, N. Donohoe, A. Heufer, H. J. Curran, F. Güthe, T. Wind, Laminar and Turbulent Flame Speeds for Natural Gas/Hydrogen Blends at Elevated Pressures, ASME Turbo Expo, GT2014-26742, (2014). [12] Beerer D., McDonell V., Therkelsen P. and Cheng, R.K, Flashback, Blow Out, Emissions, and Turbulent Displacement Flame Speed Measurements in a Hydrogen and Methane Fired Low-Swirl Injector at Elevated Pressures and Temperatures, Paper No. GT2012-68216, ASME Turbo Expo 2012, Copenhagen, Denmark, June 11 15, 2012 [13] Eichler, C., Baumgartner, G., Sattelmayer, T., Experimental Investigation of Turbulent Boundary Layer Flashback Limits for Premixed Hydrogen-Air Flames Confined in Ducts; Journal of Engineering for Gas Turbines and Power, Vol. 134, No. 1, pages 1-8, 2012. 9 Copyright 2014 by Alstom Technologie AG