Hydrogen Cofiring Demonstration at Constellation Hillabee Siemens Energy SGT6-6000G Power Plant

Testing of a Siemens SGT-6000G gas turbine with up to 38.8% by volume hydrogen blending is discussed herein. This configuration is deemed important to the gas turbine and power generation community by the authors as this turbine, like many gas turbines in operation today, was not built with hydrogen fueling in mind. Similar configuration testing has been conducted [1–3], though the testing described here provides unique insights. This testing is higher by 3× in mass flow and 2× in volume % the hydrogen tested relative to similar configurations tested. The high flow requirements of this testing did lead to a challenge of testing time available as the very high flowrate requires a large amount of hydrogen in storage and short testing durations. This limited the amount of testing that could be completed.

The hydrogen supply was also a (relative to other testing) [1,4] reduced order design showing a minimum viable, safe system, which can be utilized for hydrogen blending testing and potentially for long-term operation. The high flow rates and large change in hydrogen pressures (over time and through the supply) resulted in fuel conditions unique to this testing campaign. Hydrogen and natural gas blend emissions, and NO_x emissions specifically, are a subject of current research [5–14]. NO_x emissions with increasing hydrogen blends have been shown to be able to be maintained at relative constant levels to natural gas in dry low NO_x (DLN) [3,15] combustion systems utilizing varying methods of control and with diffusion systems utilizing diluent for emissions control [16,17]. Test results here exhibit the ability to maintain emissions with this system utilizing combustion temperature reduction and fuel delivery adjustments to the premixed fuel circuits in the combustion system. Analysis [14] and testing [12,13,17] have shown that hydrogen containing fuels result in higher flame speeds resulting in a more compact flame and reaction zone. To date, no field data that the authors are aware of has confirmed this with instrumentation. This testing included two instrumented combustion baskets, which confirm this phenomenon in field testing through combustion hot section metal temperature measurements. In addition, combustion dynamics are known to be affected by hydrogen blending. The effect on combustion dynamic amplitudes and frequencies is dependent on the specific frequency range and the system design. Combustion dynamics were measured in this testing and relative dynamics are shown.

Beyond system performance, information about test protocols which provide relevant information for future tests and permanent operation are contained here. This includes safety evaluations, fuel delivery system evaluations, venting, purging, and gas charging procedures, and finally communication protocols and test execution team arrangement.

An objective of the hydrogen blending addition system design was to blend hydrogen safely and effectively into the natural gas stream without including complex and expensive control and monitoring equipment, which may not add specific value to operations.

Figure 1 is a schematic of the high-pressure tube trailer, manifold, and pressure regulation system used. Figure 2 is a schematic of the hydrogen flow control system used. Figure 3 is an image of the hydrogen flow control system.

The major components of the hydrogen supply include the high-pressure tube trailers, which started at 3600 psia and were connected in parallel to a high-pressure manifold. The manifold included a vent for purging and charging the system. The manifold was then connected to three pressure reduction skids connected in parallel to allow for the high mass flow (>6000 lbm/h) of hydrogen required for testing. The outlet of each pressure reduction skid was then connected to a common manifold and supplied to the hydrogen flow control system. The flow control system included vent and nitrogen supply provisions for gas purging and charging. A hydrogen flow control valve was used to meter hydrogen flow to meet the required blending targets. The hydrogen flow was confirmed though a Coriolis mass flowmeter. This flowrate was used in the gas turbine controller calculation to determine the hydrogen percent volume blend.

A specific design decision was to use a hydraulic valve for the hydrogen gas flow control valve. This was selected to allow for quick ramping operation to meet high speed operation demands such as rapid shutdown and change in hydrogen blending that were determined as critical during hazard design reviews.

Another simplification decision taken was to eliminate any dedicated hydrogen and natural gas mixing device. Gas blending studies including that dedicated to hydrogen and natural gas mixing are quite common [18,19].

Confirmation of adequate hydrogen blending without a dedicated mixer was done through computational fluid dynamics (CFD) and via test results that indicated a well-mixed blend for all hydrogen blend ratios. NO_x emissions, shown in subsequent sections, trended as expected with increasing blending suggesting a well-mixed gas stream. Combustion dynamics did not indicate any maldistribution of hydrogen which could result in dynamics signatures, which vary beyond expectations per combustion basket.

The hydrogen gas was supplied to the 12-inch diameter natural gas system via a 3-inch diameter pipe. Post the hydrogen injection, there is a long section of piping (>100 feet) and multiple bends before the gas crosses into the gas turbine enclosure.

Figure 4 shows a CAD view of components also shown in Fig. 3 of the hydrogen supply and blended gas system up to the fuel gas knockout drum.

Reynolds-averaged Navier Stokes CFD analysis was conducted on a 3″ hydrogen supply to a 12″ natural gas system and shows that for the higher volume blends (38% hydrogen volume blend shown) the jet penetration of the hydrogen gas and resulting turbulence leads to a well-mixed blend sufficiently far downstream of the hydrogen injection. This result is shown in Fig. 5, which models a 10 meter straight mixing length post the hydrogen injection.

Lower volume blends, such as 5% by volume shown in Fig. 6, show that the lower hydrogen velocity results in a jet stream, which does not mix as well with natural gas in a straight pipe of the same length.

As the gas system shown in Figs. 3 and 4 has multiple bends a CFD result of a system with such bends is shown in Fig. 7. This shows that after 2 bends the hydrogen stream becomes well mixed. The higher turbulence intensity reaches the across the diameter of the pipe after these bends mixing the gases well. This result matches well to previous research [18,19].

where μ_H represents the mean and σ_H represents the standard deviation of hydrogen volumetric concentration at that cross section. Figure 8 shows the mixing efficiency versus axial location for the straight pipe and pipe with bends for 5% by volume hydrogen. While the straight pipe does see an increase in mixing efficiency with longer mixing lengths, the pipe bends enhance mixing to reach 97% mixing efficiency versus 86% for the straight pipe at the same pipe length.

Controls system updates for hydrogen blending were also kept relatively simple. Control of total fuel flow and fuel circuit delivery was altered to account for the change in fuel properties with hydrogen blending. One decision was to use the flow control/blending valve position for the hydrogen percent blend calculation. This eliminated the requirement of potentially expensive, complicated, and slower responding measurements for the hydrogen blend percentage. The Coriolis flowmeter was used to validate the hydrogen measurement calculation during and after each test day. On-site and off-site gas chromatograph measurements were used to confirm the gas blend via off-line sampling. Gas turbine control system modifications also included hydrogen supply system trip monitoring and the prescribed response of the control system to protect the gas turbine (GT) from damaging operation.

Test preparation and reviews were critical in ensuring the system design and test protocols were adequate. A hazard and operability study (HAZOP) and system design review were completed in December 2022, including the entire team. The system design modifications and procedure plans were reviewed. An failure mode and effects analysis was developed and based on the review a list of actions were assigned to team members. These actions were completed over the course of the months between the review and test execution.

The hazardous review evaluated all system designs and determined all updates to the preliminary design that were required. Plant personnel were actively engaged in executing and enforcing the safety plan, having ownership for the overall plant operation.

Figure 9 shows the hazardous zone classifications that were determined for hydrogen operation. Red areas indicate hydrogen system exclusion zones while the hydrogen delivery systems are shown in green and electrical systems in purple. Zone classifications were based on relevant safety codes. Areas such as the turbine air inlet and areas of permanent personnel presence were critical in setting classifications. Zone classifications, safety protocols, hydrogen gas pressures, venting design and protocols all were reviewed to determine the area classifications and exclusion zones.

Fuel delivery systems for the site were not designed with high blends of hydrogen in mind. Analysis was conducted to determine any risks associated with hydrogen operation on the existing fuel supply system.

All piping exposed to 100% hydrogen gas was designed in accordance with ASME B31.12 for hydrogen service. Existing piping was exposed to hydrogen blends of up to 38.8% but for limited time due to the testing duration. It was determined during materials analysis and HAZOP review the risk for hydrogen embrittlement at the blends tested was low for the hydrogen blend, temperature, pressure, and exposure times of the test campaign.

High temperature hydrogen attack (HTHA) is a function of gas temperature, hydrogen gas partial pressure, and time [20]. Worst case conditions for HTHA from this testing is shown in Fig. 10 relative to the time required for HTHA to affect the carbon steel materials utilized in the gas supply system. The worst-case condition was below the “No attack” line indicating safe operation from high temperature hydrogen attack for the hydrogen blending testing.

Test plans were focused on the objectives and use case of the Hillabee plant. To reduce testing complexity and focus on the most critical goals, a test at only baseload conditions was conducted.

The detailed test plan was owned by Siemens but reviewed with the entire project team. The advantages and disadvantages of multiple test objectives was discussed. The decision on the test conditions was based on the general operation of the plant and the hydrogen available. The judgment was to focus on baseload operation in lieu of any partload operation. DLN emissions and dynamics tuning conducted were minimal and only conducted to keep the machine and combustion system in a safe, reliable operation while attempting to maintain good, if not constant NO_x emissions.

Procedures were reviewed in detail during the HAZOP and over a period of months with the entire team.

Siemens developed and reviewed a purge in and purge out test plan specifically for the Hillabee testing. This procedure included the Certarus hydrogen storage and pressure regulation supplied equipment as well as the Siemens hydrogen metering and blending system. The nitrogen purge used for the Siemens skid was located at the Certarus high pressure manifold between the hydrogen storage and pressure regulation skids (Fig. 2). The procedure detailed the needed steps to purge the hydrogen supply piping jointly provided by both Certarus and by Siemens. Piping was purged from the Certarus gas delivery trailers to the tee into the natural gas line.

Certarus also developed system pressure testing, purging, inerting, and leak detection procedures for the specific to the Constellation testing. Pneumatic high pressure leak tests using an inert gas blend of helium and nitrogen were conducted based on system design and maximum operating pressures.

The purge procedures ensured that oxygen levels in the gas lines were below 1% by volume before the introduction of hydrogen gas. This was done through a high-pressure purge and vent procedure repeated until vented gas was measured below 1% by volume of oxygen. Once this was completed, the system was deemed ready for gas charging.

Procedure equipment and sensors were limited to only a few changes for hydrogen operation. Gas detection equipment specific to the leak and purge tests was used for each test while leak detection in the gas turbine enclosure was updated for the hydrogen blend by reducing the methane lower flammability leak detectors allowable methane limit. Personal hydrogen leak detectors were worn by personnel working on equipment pressurized with hydrogen gas.

Testing was conducted over a 2-week period of five testing days operating with H2 blends ranging from 0 to ∼38.84% (as confirmed by offline gas chromatograph).

There were several criteria critical to test execution success. Communication protocols were developed to ensure that information between parties was controlled and efficient. Testing plans were discussed at a stand up meeting every day before testing was conducted. Site operations maintained leadership and control over general GT operation while Siemens focused on ensuring testing protocols were met. Personnel were located in specific locations on site per communication needs.

The team focus included ensuring that plant reliability was not impacted negatively as all testing was conducted in parallel with commercial operation. This was successful based on the protocols followed by the team.

Testing was conducted without major issue. The machine was able to be adjusted on-line during testing to complete all test goals. The original hydrogen blending target was reached without major issue, allowing the team to stretch to the higher 38.8% hydrogen blend. It should be noted that the hydrogen blend was not limited by the combustion system or gas turbine but by the hydrogen supply limitations.

A point of active research is the impact of increasing hydrogen blends on NO_x emissions. Research on emissions reporting corrections [1] as well as raw, mass/energy based, or noncorrected emissions has all been published by multiple groups in recent years [11,14]. Dedicated hydrogen blending testing has also been conducted recently on numerous gas turbines, which reported NO_x emissions among other results [3,4,15,21]. To add to this canon of data, NO_x emissions from all testing are provided in this section in multiple figures. Figure 11 shows the instantaneous data (not stable emissions points) from all testing conducted. To be able to complete limited combustion tuning and get to the very high levels of hydrogen, stable emissions points were not able to be taken during the testing. However, these results are indicative of the general emissions capability of the system. These data exhibit that NO_x emissions were able to be kept constant as hydrogen blends increased from 0 to over 38% by volume. Figure 12 shows that for some test points, the exhaust temperature (driven by a decrease in total fuel and reduction in turbine inlet temperature) was decreased, which helped to hold NO_x emissions constant. However, this was not the case in all testing as much testing NO_x emissions were held constant through fuel delivery changes into the combustion basket (combustion fuel split adjustments).

Shown relative to testing time the effect of variation of the hydrogen content during the test run on the gas turbine NO_x emissions is also shown in Fig. 13. The marked area shows the effect of burner tuning, i.e., changing the fuel split to the several stages of the burners. The tuning effort, which was done manually during this test, reduced the NO_x emissions practically to the level of natural gas-only operation. The distribution of the H₂-blended fuel within the several stages of the ULN (Ultra-Low NO_x) dry burner system was slightly modified, which resulted in a reduction of NO_x values like the typical level of operation with natural gas. In future, permanent H₂ cofiring applications, this tuning function will be done in real-time by an automated, self-learning autotuner module within the gas turbine combustion management system. Specifics of the fuel split adjustments conducted cannot be shared here; however, the addition of hydrogen both allows for some fuel split adjustments by reduction of some combustion operation boundaries and the typical fuel splits used for natural gas are not the minimum NOx fuel splits capable for the ULN combustion system.

Figure 13 shows a minor increase of out-of-engine NO_x when hydrogen was blended into the natural gas due to the time taken for the manual combustion adjustments. The gas turbine was operating at baseload, and the increase of NO_x is less than 30%. However, before leaving the stack of the plant, this increased NOx level was always reduced to the permit level by the downstream Air Quality Control Equipment.

Hydrogen gas temperatures seen during testing varied significantly. These results may be of importance for those using high pressure hydrogen storage.

As hydrogen was supplied from the high-pressure trailers and that gas decompressed in the trailers, flowed through metering valves, and exchanged heat with the environment the gas temperature changed from near ambient temperatures to significantly colder. Measured hydrogen gas temperatures are shown in Fig. 14. The dashed lines indicate the trend with time of gas temperatures as testing was conducted and hydrogen blending increased then decreased. The temperatures dropped as the hydrogen blending increased. Near the end of testing as the hydrogen blends were decreased, the gas temperatures did not recover completely, or at all depending on the test day.

An analysis was conducted to understand the driving mechanisms behind the gas temperature measurements. Figure 15 shows the results of this analysis. This model included the high-pressure hydrogen storage, a pressure reduction system, and a model of heat transfer to ambient. The model was run transiently. There are three major drivers of the hydrogen temperature. The first is the decompression of the hydrogen gas in the storage tanks, which results in a large temperature reduction. Second, as the hydrogen gas flows through the metering valves the Joule–Thompson [22] effect drives a small temperature increase of the hydrogen gas. This effect is a function of the gas pressure. Finally, there is a temperature increase as the hydrogen gas exchanges heat with the ambient environment. These results can serve as information for others designing hydrogen supply systems and are especially critical for those with a required minimum fuel gas temperature.

Beside the standard acoustic monitors in the combustion midsection of the gas turbine, the test instrumentation also included a set of acoustic sensors, inserted into each combustor basket to measure the undistorted acoustic fingerprint of the individual burners and its variation with added hydrogen. Figure 16 is an image of an instrumented transition duct used for testing.

The result of the acoustic measurements during the H₂ cofiring tests was evaluated with a spectral analysis. The chart in Fig. 17 shows the spectral analysis of the combustion dynamics measured within the combustion zone at the midsection area of the gas turbine during the 30% by volume H₂ cofiring test. It clearly shows that the baseload operation with 30% by volume H₂ cofiring does not increase the overall levels of dynamic amplitudes within this section. For some frequencies, the addition of hydrogen has a damping effect and reduces amplitudes considerably.

Combustion dynamics were not recorded instantaneously for the duration of the test; thus, combustion dynamics amplitudes are not available to exhibit relative to all hydrogen blends as other gas turbine parameters are. However, observations of combustion dynamics bins (the dynamics spectra data separated into frequency bins and represented by the maximum amplitude in each bin) exhibited that combustion dynamics amplitudes were relatively muted for the hydrogen blends tested.

Unique to field testing conducted thus far [3,4,15,21], Siemens added a full suite of temperature instrumentation to two combustion baskets. These data were used to determine and validate any expected hardware impact of the hydrogen gas.

Test instrumentation with temperature sensors were assembled to two combustor baskets and the corresponding transitions. The measurements of the temperature sensors were used to validate impact to the metal temperature of burner and transitions due to the change in flame shape and position with hydrogen blending.

The chart in Fig. 18 shows the influence of hydrogen cofiring on the temperature of these instrumented parts. The measured values show an increase of metal temperature in the upstream parts of the burner of up to 60 K, close to the burner base plate and gas nozzles. This confirms the expected behavior of hydrogen combustion, which due to the higher flame speed, burns with a more compact flame, moving the highest energy density upstream, closer to the burner nozzles. However, the measured temperature increase is within the capability of the material for a short-term test. This result suggests the need for further material investigation and H2 cofiring tests to determine if H2 cofiring influences long term durability of components and if H2 cofiring needs to be addressed with a service factor to cover the higher temperature level. The transition piece, which is downstream of the burner, showed no significant change of temperature. Any temperature change in the transition piece is compensated by the cooling system to them, which is actively regulated and minimizes temperature variations.

An important aspect of the use of hydrogen is the impact on the power generation and efficiency of the gas turbine. As discussed in Ref. [23], the use of hydrogen will impact the gas turbine generated power through the change in combustion products. The higher concentration of water in the combustion products, relative to natural gas only, results in a higher specific heat. The exhaust gas flow decreases with the lower mass flow of hydrogen as hydrogen has a higher heating value per unit mass. This is shown for a simulated machine in Fig. 19.

The change in specific heat results in higher turbine power generated for the same compressor power and fuel heat energy added. This can result in a higher gas turbine power output and reduced heat rate as shown for the same simulation in Fig. 20.

The actual heat rate and gas turbine power generated is a function of the configuration and control used, however. For this testing, the turbine firing temperature was reduced with hydrogen during portions of the test and the turbine is a combined cycle machine resulting in more factors affecting load and heat rate. Given the short nature of the test determining the specific power and heat rate impact is not shown here as the data the team have is not of the accuracy required. Longer hold times at each hydrogen blend rate and potentially more precision instrumentation would be required. It is expected that for the constant firing temperature, hydrogen blend conditions tested an improved heat rate and output would be measurable.

While testing conducted was valuable to all parties and for the community, further testing could be of value. That includes partload testing to evaluate hydrogen emissions compliant turndown impact. Partload would make having further emissions measurements valuable such as CO emissions and specific NO₂ and NO emissions as the constituent ratios of each to total NO_x to measure the impact of hydrogen and hydrogen at different loads. Instantaneous CO emissions were not measured during testing as it was conducted only at baseload where CO emissions were very low for all test points.

Load ramping with hydrogen blending and hydrogen blend ramp testing would also be valuable as it would provide information around more practical operation of the GT in normal conditions.

Testing conducted on the Siemens SGT-6000G unit at the Constellation Hillabee power plant achieved the test planning major goals. Test results, execution, and planning were presented here to advance the knowledge base for other similar hydrogen blending tests. Results such as the trend of NO_x emissions with increasing hydrogen agree well with other tests [16–18]. Test designs were kept as simple as possible and validated to meet the test requirements. This includes items such as the hydrogen blending capability confirmed via CFD analysis.

These results provide additional data to the canon of results already present in the industry and serve as a platform for continued testing.

The authors would like to acknowledge supporting members of the project, including hydrogen gas and pressure regulation system supplier Certarus as well as EPC firm Kiewit for their system and safety design review assistance.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

BSER =

best system of emissions reduction

CAD =

computer aided design

CFD =

computational fluid dynamics

CO =

carbon monoxide

CO₂ =

carbon dioxide

DLN =

dry low NO_x

EPA =

environmental protection agency

FMEA =

failure mode and effects analysis

GT =

gas turbine

HAZOP =

hazard and operability study

HTHA =

high temperature hydrogen attack

H₂ =

hydrogen

NO_x =

nitrous oxide

RANS =

Reynolds averaged Navier Stokes

ULN =

ultra low NO_x