Experimental Campaign on a Hydrogen Fuelled Combustor for a 10 MW Class Gas Turbine with Reduced NOx Emissions

Transcription

1 Experimental Campaign on a Hydrogen Fuelled Combustor for a 10 MW Class Gas Turbine with Reduced NOx Emissions S. Cocchi 1, M. Provenzale 1, S. Sigali 2, L. Carrai 2 1. GE Oil&Gas NUOVO PIGNONE SpA, Firenze - ITALY 2. ENEL Ricerca, Pisa - ITALY 1. Introduction In the frame of a research project launched in 2006 (partly funded by Regione Veneto, a local institution in the North-East of Italy), ENEL and Nuovo Pignone are developing an innovative zero emission gas turbine cycle suitable for power generation. The gas turbine, a GE10-1 model, is manufactured by Nuovo Pignone and will be installed at ENEL s coal-fired Fusina power plant, near Venice. The turbine, rated for 11 MWe, is equipped with a diffusive flame combustor and is suitable for operation with 100% hydrogen as main fuel over the entire load range. Hydrogen is available at Fusina site as by-product of petrochemical plants. Natural gas will be used as start-up and back-up fuel, and NOx emission abatement will be achieved by means of steam injection. Load operation will be possible with hydrogen only, with methane or hydrogen-methane mixtures (in case of reduced availability of hydrogen) and with or without steam injection. In order to support the combustion system s design, experimental activities have been carried over a prototypical combustor, installed on a combustion test rig at ENEL s experimental facility, located in Sesta (Tuscany). The present work is aimed to provide a description of the relevant results achieved from such experimental campaign. 2. General aspects of hydrogen combustion in gas turbine applications Hydrogen combustion by-products are water vapor and nitrogen oxides (NOx). This last is a harmful pollutant, since is responsible for acid rains and photochemical smog in the troposphere, and ozone depletion in the stratosphere. For this reason, local authorities for air quality and environmental protection are imposing more and more stringent requirements in terms of NOx emission from industrial power plants. Accordingly, gas turbine manufacturers have designed advanced combustion systems, based on lean premixing techniques, where the reduced flame temperature (well below stoichiometric values) generates reduced amount of NOx, according to the predominance of Zeldovich mechanism in combustion application where flame temperature can rise beyond 1800K. Such technology, universally known as Dry-Low-NOx (DLN), has been mainly developed for operation with methane or natural gas, since these fuels have been the common choice for the most part of industrial gas turbine applications. In order to decrease NOx emissions through flame temperature reduction, fuel premixing would be the desired choice also for hydrogen combustion, but the high reactivity of hydrogen-air mixtures rises the risk of flashback, flame anchoring in undesired locations and generally unsafe system operation. Besides that, one important requirement in for heavy-duty industrial applications is the ability of long term continuous operation: the possible unavailability of hydrogen supply (due to the lack of hydrogen networks) imposes the need for a back-up fuel, mainly natural gas, which is also necessary in order to ensure safe turbine start-up. Thus, even if a consolidated lean-premixed technology for hydrogen combustion would exist, the different combustion properties of hydrogen and methane (above all, laminar flame speed) make this choice impracticable for operation with both these fuels, and diffusive flame combustion systems, capable for safe and IX-1, 1

2 31st Meeting on Combustion continuous operation whether the fuel is, are widely adopted. So far, there are heavy-duty gas turbine applications equipped with diffusive flame combustors, operating with gaseous fuel mixtures containing hydrogen up to 95% vol. concentration. Since hydrogen stoichiometric flame temperature is roughly K higher than methane, a diffusive flame combustion system fed with hydrogen is expected to generate NOx emission 3-4 times higher than methane, with absolute values in the range of several hundreds of ppm, at typical heavy-duty operating conditions. The reduction of NOx emissions down to level compliant with air quality regulations is generally achieved by means of water or steam injection, with the aim of flame temperature reduction. In the common practice, the diluent is injected in the combustion air stream, or directly in the reaction zone. Higher effectiveness is expectable if the fuel is premixed with some diluents upstream the fuel nozzles (conventional diffusive combustors, operated with CO2-diluted natural gas, showed decreasing NOx emissions with increasing inerts concentration): this choice, however, has the main drawback of increased system s complexity and operating costs. 3. Combustion hardware s description The GE10 gas turbine is equipped with a single-can silos-type combustion chamber. The prototypical diffusive flame combustor developed for operation with 100% hydrogen fuel, featuring steam injection for NOx abatement, is schematically represented in fig. 1. Air from the compressor discharge plenum is directed to the combustor head through the annular section, comprised between the external casing and the liner. Air is injected in the combustion zone through dilution holes, cooling holes (grouped in several rows, located along the entire combustor length) and primary combustion air holes. The diffusive burner is located at the top of the combustor, and consists of a simple multi-hole fuel nozzle. The burner tip accommodates only a small amount of combustion air, and is designed in order to transmit a moderate swirling component to the air fuel stream entering the combustion zone. Fig. 1 Combustion chamber schematic view. IX-1, 2

3 Italian Section of the Combustion Institute Steam is injected directly in the cold-side combustion air flow through the external combustor head, in order to concentrate dilution effect in the primary zone only. Steam injection directly in the reaction zone is avoided, due to the risk of hot gases backflow in the event of steam supply system s malfunctioning. In order to assess combustion performance s sensitivity to some minor modification in liner s and burner s design parameters, different hardware components have been tested during the experimental campaign. Basically, tests have been focused on two configurations: one comprises a dual gas burner (hydrogen and natural gas are injected as separate streams into the reaction zone) and the commercial version of the liner installed through the GE10 fleet, and is called baseline solution ; the other consists of a single gas burner (hydrogen and methane are blended upstream the burner s flange), combined with a modified liner specifically designed to increase the quote of primary combustion air, called alternative solution. In the alternative configuration, the liner s hole pattern is such that some more air is delivered to the primary combustion zone, by means of a significant reduction of both dilution and cooling air, with the aim of lowering the average temperature of the reaction gases in the primary combustion zone. Besides that, the overall liner s effective area has been reduced, with the aim of inducing higher combustor pressure drop, higher air jets penetration and consequently enhanced combustion mixing in the reaction zone. As far as the burner, the choice for a single gas configuration is mainly aimed to simplify the fuel supply system. Actually, fuel holes are sized in order to keep the pressure drop across the nozzle within an optimal range: high pressure drop can cause flame blowout and requires high fuel supply pressure; low pressure drop can cause instabilities and rises the risk of nozzle s overheating. At a given thermal power, pressure drop across the fuel nozzle for hydrogen is expected to be about 40% higher than methane; however, the high reactivity of hydrogen suggests the possibility of using single gas configured burners without risks of flame stretching or blowout due to the increased fuel jet velocity. Besides operability characteristics, the two burners involved in tests differ also as far as the swirler design: the alternative solution has been developed for enhanced fuel-air mixture uniformity, in order to reduce the extension of the stoichiometric reaction zone and achieve lower NOx concentration. 4. Experimental setup The combustion test rig is located at the ENEL s experimental area, in Sesta (Tuscany, Italy). This facility, built in the Nineties, consists of two test rigs and is suitable for full-scale, fullpressure testing of a wide variety of combustion systems. Compressed air is supplied by a train of two compressors, with maximum air flow capability of about 40 kg/s. Variable combustion air temperatures are achievable by means of electric heaters; combustion air pressure can be modulated by means of a water-cooled lamination valve, located in the exhaust duct. A multi-fuel supply system makes possible the production of fuel mixtures of different composition: high-capacity storage of a large variety of fuels and inert (methane, hydrogen, propane, butane, CO, CO2, N2, liquid fuels and demineralized water) is available. Superheated steam can be supplied as well. As far as GE10 combustion hardware s testing, the test rig has been operated at its full capability. Fuel supply system mass flow capability is about 1100 kg/h for hydrogen, more than 2500 kg/h for methane, about 9000 kg/h for superheated steam. The combustion test rig is equipped with resident instrumentation, dedicated to air and fuel systems control, also ensuring accurate mass flow measurements. The test hardware has been equipped with dedicated instrumentation, specifically installed in order to provide a detailed and comprehensive screening of combustion system s performances. Both liners and burners have been instrumented with thermocouples, giving a discrete map of metal temperature IX-1, 3

4 31st Meeting on Combustion profile. Six dynamic pressure probes have been installed on the combustor s casing, with the piezoelectric element directly faced to the cold-side air stream, and used for real-time spectral analysis. Pollutant emissions are measured at two different sections, one located at the combustor exit section, the other located in the exhaust duct: NOx emissions are measured by means of both chemiluminescence and infrared analyzers; non-dispersive infrared analyzers are available for CO emission measurement; oxygen content is measured by means of paramagnetic O 2 analysers. 5. Relevant results Tests have been focused on simulating engine s operation at baseload operation, since this is the operating condition requiring the highest thermal power and operating temperatures. Mainly, 100% hydrogen and 100% methane have been investigated, in order to evidence any potential operating risk. Cycle conditions to be set for combustor s tests have been derived from a specific GE10 turbine s simulator Metal temperatures Both the baseline and the alternative configuration have exhibited weak sensitivity to fuel operation. Liners metal temperature profiles were almost identical, while burners and liners dome showed some difference depending on fuel. However, all metal temperatures were well below safe limits in any operating conditions, for both the hardware configurations Pressure pulsations Pressure pulsation spectra at baseload dry operation are reported in fig. 2, for both methane and hydrogen fuel, on baseline and alternative solutions. Pulsation spectra show a marked difference only in the Hz frequency range: peak amplitudes for the alternative solution are 4-5 times higher than baseline solution, due to the increased amount of primary combustion air. Operation with hydrogen is generally much more quite than burning methane. Fig. 2 Pressure pulsation spectra at baseload dry operation In the event of steam injection, pulsation spectra exhibit some differences only in the same Hz frequency range. Increasing the steam-fuel mass flow ratio, defined as the ratio of steam mass flow over the equivalent methane mass flow at same thermal power, peak amplitude significantly increases on methane operation (fig. 3), while it s almost insensitive on hydrogen operation (fig. 4). IX-1, 4

5 Italian Section of the Combustion Institute Fig. 3 Pressure pulsation spectra at baseload methane operation with steam injection Fig. 4 Pressure pulsation spectra at baseload hydrogen operation with steam injection 5.3. NOx emission NOx emission, recorded during baseload dry operation of baseline configuration at different hydrogen/methane thermal power split, is reported in fig. 5. As expected, 100% hydrogen operation generates about 3 times higher NOx than 100% methane. In order to compare NOx emission from baseline and alternative solution and assess steam injection effectiveness, experimental values have been processed developing a simple semiempirical correlation for NOx emission estimation at generic turbine s operating conditions. A good model s predictability (accuracy within ±10% of point) was achieved over the entire set of experimental data, for both the baseline and the alternative solution. Fig. 5 NOx emission at different hydrogen/methane split during baseload dry operation IX-1, 5

6 31st Meeting on Combustion Such predictive models are based on 4 main assumptions: 1) NOx formation occurs through the Zeldovich mechanism only; 2) NOx are generated only in a portion of the whole combustor volume, approximated as a Perfectly Stirred Reactor (PSR), with no NOx source elsewhere; 3) such PSR operates close to stoichiometric condition, and its volume depends on the overall fuel-air ratio in the primary combustion zone and the fuel type, as shown in fig. 6; 4) the effect of steam injection is evaluated by means of a reduction factor, scaling virtual dry NOx emission evaluated at same cycle operating conditions. Fig. 6 NOx reaction volume s dependence on fuel-air ratio for the alternative configuration Despite expectation, baseline and alternative solutions exhibited almost the same NOx emission levels at baseload dry operation. However, steam injection revealed enhanced effectiveness on the alternative configuration, as shown in fig. 7, due to the more intense mixing of combustion products promoted by the enhanced pressure drop across the liner. Fig. 7 Steam injection effectiveness on NOx emission from baseline and alternative solution 6. Conclusion Data collected during the experimental campaign have demonstrated combustion system s safe operation with either 100% hydrogen or 100% methane fuel mode, with both hardware configurations. The alternative design (increased air in the primary zone) exhibits higher pressure pulsations, as expected, with no significant benefit in terms of NOx emission at dry operation. However, steam injection appears to be more effective on the alternative design. NOx emission data post-processing led to the development of semi-empirical predictive models, characterized by satisfactory accuracy, that can be used as complementary tools in the analysis of combustor s performances at simulated or experimental turbine s operating conditions. IX-1, 6