Assessment of Current Capabilities and Near-Term Availability of Hydrogen-Fired Gas Turbines Considering a Low-Carbon Future

Due to ever changing demand needs along with implications of carbon reductions, gas turbine combustion faces a game changing future that will continue to drive major technological developments in the coming years. This paper assesses the state of the art capability of gas turbine combustion technologies for use with Hydrogen (H₂) fuels, which are applicable for both combined cycle (as seen in Fig. 1) and simple cycle units, all the while, keeping in mind the importance of gas turbines in the future for a resilient and reliable power grid.

With this in mind, according to the EIA (U.S. Energy Information Administration), about 35% of electricity produced in 2018 was produced via natural gas (NG) combustion in gas turbines (GT) or boilers, which is up from just 26% in 2017 [1]. As coal retirements continue to occur, this number will continue to increase as utilities seek to replace baseload requirements. Continued growth of GT capacity is predicted in the USA [2]. Thus, GTs will continue to perform an important role in power generation for many decades through baseload, cyclic, and peaking operations.

Thus, GTs will be a part of electricity generation for decades to come, but the question to start answering now is —how does the energy industry reduce, or even eliminate, carbon output? To start the process of answering this question requires an examination of trends in the industry with future predictions and possible strategies for continued reduction. To this point, NG burning GTs, along with solar/wind renewable installations, maintaining nuclear units, coal retirements, as well as increased consumer energy use efficiency, have paved the current U.S. electricity sector path to lower carbon emissions. Specifically, from 2005 to 2019, there was an overall U.S. power industry CO₂ emissions reduction of about 15%; this includes industry/buildings, transportation, and electricity sector contributions. During this time, the specific electricity generation sector realized an approximate 28% reduction alone even though during this same time reduction in sectors such as transportation were partially feasible through electric vehicles that increased demand on the power grid [3]. The continued path that EPRI would like to see for 2030 would be an additional twofold decrease in CO₂ emissions. Realistically this can be accomplished from continued coal retirements, new renewable installations, and new, more efficient GT installments—even with the possibility of additional nuclear retirements. However, as the industry looks further out, to continue overall U.S. energy CO₂ emissions reductions to targets such as 80%+ by 2050, bigger changes are needed, specifically related to cleaner dispatchable generation.

For the electricity generating sector, advanced nuclear and carbon capture and sequestration are additional avenues for accomplishing the continued reduction, but an additional path is eliminating carbon from the fuel altogether through the utilization of H₂ or ammonia (NH₃). Hydrogen combustion is not a new topic. It has for many decades been investigated to understand combustion characteristics and emissions from a range of H₂ containing fuels. Most prominently in recent history (c. 2000) was the directive to understand and use synthesis gas [syngas, coal derived fuel usually containing CO, H₂, CH₄, along with inerts like H₂O, N₂, CO₂, etc.…)] from an integrated gasification combined cycle (IGCC) [4,5]. During this time, the U.S. Department of Energy (DOE) invested millions of dollars to utilize H₂ containing fuels all the way to the current advanced combustion turbines program [6]. This R&D focus laid the groundwork for new designs and technologies that we see today such as micromixers and fuel staging. These technologies are emerging as the focus of many new engine designs.

The purpose of this paper is to overview the impact of decarbonized fuels on gas turbine combustion systems. The paper begins with a presentation of the challenges of combustor operability with decarbonized fuels. The paper closes with an overview of the openly published research and development that the original equipment manufacturers (OEMs) are exploring to respond to these challenges. The primary focus of the paper is on H₂ combustion, which the authors identify as the most likely decarbonized gas turbine fuel. This paper provides a broad overview of the many engineering challenges and solutions surrounding H₂ combustion with an extensive literature survey referenced for a deep dive into the details of these topics.

The infrastructure for widespread H₂ fuel usage in gas turbines is presently unavailable. Implementation of a H₂ infrastructure must overcome several key safety related hurdles. The H₂ molecule has a greater tendency to leak than other gases. This is because the small H₂ molecules can squeeze through small cracks, gaps, and other tolerances that cannot accommodate a larger CH₄ molecule. Furthermore, H₂ embrittles many fuel plumbing materials [7], compromising many metals and materials commonly included in combustion and gas plumbing hardware. These compromised materials can introduce new leaks as well as cause premature failure of flow control and instrumentation equipment. A H₂ leak also brings greater hazard than a NG leak for several reasons. First, H₂ has a lower lean flammability limit (4% by volume at standard temperature and pressure (STP) in air) compared to CH₄ (5% by volume at STP in air). By mass, this translates to a lean flammability limit for H₂ in air that is only 10% that of CH₄. This means that a H₂/air mixture becomes flammable and can be ignited with lower fuel gas concentrations (i.e., an explosive mixture will build up sooner in a confined space such as enclosures and heat recovery steam generator (HRSGs)) than NG. This will have important purge considerations for failed starts. It might also require ventilation solutions for HRSG attics which could easily accumulate highly buoyant H₂ gas.

Hydrogen leaks are also hazardous because the chemiluminescence from premixed H₂ flames is nearly invisible. A common misconception is that H₂ flames are completely invisible; however, the hot gases that are produced in nonpremixed H₂ air flames in many leaks will emit some light [8]. The lower observability of H₂ flames relative to NG flames can make the avoidance of hot gases difficult for those who respond to or who happen upon a leak.

While the lack of H₂ fuel usage for GTs can be primarily attributed to limited H₂ availability, the complexity of using and burning H₂ is a close second. While H₂ combustion does have technological barriers, it boasts the smallest technological barriers to storage and clean use of renewable generation with existing infrastructure relative to other alternative carbon neutral fuels. It is already possible to use H₂ as a fuel, no matter the fuel constituent percentage as evidenced by current installations [10–12]. However, current challenges exist with dry,low NO_x (DLN) technologies pushing past 10 to 50% depending upon the hardware. Most high H₂ capabilities exist with either diffusion or wet-low NO_x type technologies, which are not comparable to the NO_x levels of DLN. Figure 2 outlines a brief comparison between nonpremixed (diffusion type) and premixed (DLN type) combustion systems.

Overall, these challenges primarily exist because of the vast differences in combustion characteristics in H₂ enriched fuels as compared to typical NG. For example, for a given set of combustor inlet conditions, the adiabatic flame temperature (or postflame temperature) is increased from 5 to 10% [13] with high H₂ fuels. This has two implications: (i) potential for higher NO_x and (ii) potential for material/coating issues. Another marked difference and one of the most problematic is that the flame speed is up to nearly an order of magnitude higher than NG [14]. Generally speaking, using the same conditions, a premixed, DLN nozzle would need 10 times higher flow velocity to prevent the flame from flashing back and damaging the hardware. Moreover, with the low flammability limits as compared to NG, as visualized in Fig. 3, at high H₂ content (>90% H₂), and no change in nozzle velocity, the lean blowout (LBO) and flashback limits allow for a narrow stable operating range [13]. From a detailed turbulent combustion perspective, lean H₂ combustion is fundamentally different than combustion of other lean mixtures. For example, lean H₂ flames are highly thermo-diffusively unstable, meaning that corrugations and wrinkles in the flame will tend to grow and become more pronounced. This is partially responsible for the enhanced turbulent flame speed of lean H₂ flames relative to other lean mixtures.

The lower heating value (LHV) of H₂ compared to methane (CH₄) is another consideration for the development of H₂ capable systems, see Table 1. With this difference in mind, a H₂ fired gas turbine would have to flow less than half of the mass flow of fuel compared to a CH₄-fired gas turbine. However, H₂ is eight times less dense than CH₄ at the same pressure and temperature. Therefore, its volumetric LHV is roughly a third of CH₄. Therefore, a H₂ fired gas turbine would need three times more fuel volume flow rate than a CH₄ fired plant. The volume flow rate sets the size of the plumbing, so the H₂ plant would need piping, valves, and instrumentation that are sized roughly three times bigger than the CH₄ plant.

Alternatively, the fuel gas pressure could be scaled as a function of H₂ content. For example, a pure H₂ fuel gas would need to be suppled at roughly three times the pressure to have the same volume specific heating value as CH₄ at the same temperature. With increased supply pressure or volume flow rate for a H₂ capable GT does have substantial cost and safety considerations especially considering retrofitting existing assets.

Another challenge is the desire for H₂ composition flexibility in power generation depending on the availability/cost of H₂. This leaves the same problems as illustrated previously but exacerbated due to the need to operate at both low and high flame speed conditions depending on fuel choice. This leads to the challenge to design a robust combustor that can operate with acceptable emissions on either H₂ or CH₄ (or more practically on a blend of CH₄ and H₂ with widely variable blend ratios). This challenge exists for premixed combustors because low flame speeds promote lean blowout, and high flame speeds promote flashback (only a concern for premixed combustors like those implemented in all lean premixed systems). However, premixed combustors (versus diffusion or nonpremixed combustors) are needed for their low NOx characteristics. Therefore, a premixed combustor with H₂ fuel flex must be designed to support a H₂ flame with flashback margin in the same hardware that can support a CH₄ flame with LBO margin. This leaves a thin, or possibly nonexistent, operating margin. The operating regime can be maximized by limiting the percentage of H₂ that can be implemented but that limits the benefits of a H₂ fuel flex capable GT. Additionally, from the earlier stated issue of supply the plumbing in a fuel flex system that is sized appropriately for H₂ would be over-sized/over-pressurized for CH₄ at the same conditions.

Combustion instabilities (also known as combustion dynamics) will also be a factor for H₂ combustion and fuel flexibility. The combustion instability phenomenon has the challenging feature that its characteristics are nonmonotonic with operating parameters (such as combustor inlet pressure, combustor inlet temperature, fuel gas composition, etc.). Therefore, it is not possible to say that H₂ combustion will make combustion dynamics “better” or “worse.” Instead, the H₂ effect on combustion instabilities is anticipated to vary on a case by case basis. For example, combustion instabilities typically show up as “islands” in the operating space. Hydrogen addition may move these islands in the operating space relative to where they appear with NG combustion. This is likely to present a challenge for fuel flexibility, since different combustor tunes (i.e., fuel splits) will have to be established for different H₂ fractions to balance combustion instabilities against emissions.

The combustion properties of H₂ that this section has outlined for combustor design (flame speed) and safety (lean flammability limit and observability) are nonlinear with mixture fraction [16]. This means that H₂ addition to the gas supply can have minimal impact on these properties at low H₂ fractions, but that small increments of H₂ fraction can have drastic impacts on these properties at higher H₂ fractions. In reality, the effect of H₂ fraction on the turbulent burning properties of H₂/NG mixtures is still an area of active research.

Also, note this section is about carbon free fuels and while the discussion has been centered around H₂, another fuel type that is getting attention in a more limited purview is ammonia (NH3). Liquid NH3 is a much easier storage and transportation media than that of gaseous H₂. In fact, the agricultural industry already operates NH3 pipelines. However, the combustion considerations are much less favorable. Flame temperature and flame speeds are much lower for ammonia than NG as shown in Table 2 [17]. However, the Nitrogen content of the fuel has increased NOx potential, even in premixed conditions. Therefore, the primary focus for the remainder of this paper will be upon H₂ only. For additional works on ammonia usage in gas turbine combustors have been published by Kurata et al. [18] and Valera-Medina et al. [19].

The last and most important series of questions to consider are a) what are the OEMs current capabilities with H₂ and b) what are they doing to enable H₂ capable GTs of the future? Over the next 10 years (2018 to 2027) it is estimated that an additional $107 billion in new GTs will be sold worldwide [2]. These projected numbers would equate to over 900 new GTs installed in North America during this time with a majority (based on capacity and value) being General Electric (GE) assets (See Fig. 4).

This section will present several emerging technologies that the various manufacturers have released in various publications. These technologies include various fuel staging and advanced mixing concepts. Utilizing manufacturer-published materials (including conference papers, reports, and patents), these complex combustion system concepts will be showcased below. The main objective of this section is to emphasize the recurring theme of fuel staging and advanced mixing technologies in the various forms that are surfacing among the different OEMs.

Recently, GEs CMO (GE Power, Greenville, SC) Brian Gutknecht during a panel at the 2019 Edison Electric Institute Conference stated that he sees H₂ as a key factor for 100% carbon-free generation [20]. In fact, to this point GE has the largest in-service operating experience with H₂ containing fuels at 70+ in operation dating back over 30 years from IGCC to refinery and steel mill supplied gases [11]. This experience includes 25 GTs utilizing fuels with at least 50% H₂ by volume and examples of units such as a 6B.03 using fuel blends of 70 to 90% H₂. One specific example is a site in Italy (Fusina) utilizing a 41.6% efficient GE10 (a Baker Hughes GE gas turbine) combined cycle gas turbine plant (12 MW) operating with 96 to 100% H₂ [21]. This site accounts for an annual reduction of 17,000 tonnes of CO₂ for approximately 20,000 households. This site has been operated by ENEL since July 2010 and is equipped with diffusion style burners with abatement strategies (steam injection) for NOx reduction. Overall, this was a approximately 50 Million Euro investment generating an estimated 60 million kWh annually.

Moreover, U.S. DOE National Energy Technology Laboratory (NETL) funds have been utilized to develop micromixers that can handle high H₂ fuels. This work dates initially to IGCC work that began in 2005 [22,23]. This has culminated in a new burner design designated DLN 2.6e that uses micromixers (see Fig. 5) and can reportedly handle up to 50% H₂ [11,24,25]. The micromixer allows for increased axial velocity in exchange for the swirling flow component that is traditionally associated with DLN type technologies, thus the pressure loss is equivalent and not increased.

Less directly impactful, but just as holistically important, is the recent incorporation of axial fuel staging (AFS), which was introduced with the HA class units for increased turndown and lower emissions [27]. The AFS for H₂ systems would serve the same purpose as for NG systems: it allows for leaner main combustor operation, therefore, lessening high NOx production and flashback concerns for the main combustion zone (Fig. 6).

Likewise, Siemens has demonstrated H₂ capability in its current gas turbine fleet. Siemens advertises its fleet's H₂ capabilities in its July 2019 Hydrogen Combustion Sales Information brochure [28]. It claims up to 80% H₂ capability in its dry low emission (DLE) systems with the SGT-600, and up to 100% H₂ capability in some of its Wet Low Emissions aeroderivative engines (SGT-A65, SGT-A45, SGT-A35). Siemens aeroderivatives have operated with more than 100,000 h on fuels containing up to 78% H₂. These units can handle substantial Wobbe index (NG composition) variation, including real-time gas composition swings. However, many of these units require water injection for NO_x control. Siemens' small industrial gas turbines have extensive H₂ combustion experience, including refinery and coke oven gas with high H₂ contents in conventional burners and H₂ blends in DLE combustors. Their midsize gas turbines have more than 10 years of H₂ combustion experience, drawing from R&D activities and operations with high H₂ refinery fuel gas. Finally, their large gas turbines have more than 45 years of H₂ experience from IGCC projects with up to 60% H₂ tests. The H₂ capabilities that Siemens currently advertises are summarized in Table 3. Siemens also advertises the possibility of augmented H₂ capability on a project by project basis.

Siemens' future H₂ plans as illustrated in their July 2019 Hydrogen Combustion Sales Information brochure [28] are driven by H₂ cofiring in Europe. Siemens is targeting 20% H₂ capability in all models by 2020 (aeroderivative through frame units, with many already at 30%) and 100% H₂ capability in all models by 2030 [29]. Some of these anticipated developments will be implemented in new units and retrofits. For example, the 10 to 30% H₂ range will require combustor upgrades which include modified burners, higher temperature materials, an advanced gas detection system, and leak checks after maintenance. The 50 to 70% range will require new burner designs and bigger fuel pipe diameters and may require start up and shut down on conventional fuels. Siemens' past and future roadmap to 100% H₂ capable low NO_x gas turbines begins with full pressure testing soon after 2019, followed by 100% H₂ capability in aeroderivatives, then industrial gas turbines, and finally heavy-duty gas turbines. The roadmap also identifies future development of the DLE technology for low NOx, high temperature combustion extended to 100% H₂.

The U.S. DOE Office of Scientific and Technical Information has published a final report from Siemens' Advanced Hydrogen Turbine Development program [30]. This program was sponsored by the National Energy Technology Laboratories and overviews several new technologies, including combustion system technologies. Siemens discusses many of the key physics and engineering challenges in this report. For example, the report details the challenge of flashback for H₂ and residence time for NO_x, and the intersection of these challenges with Siemens' ultra low NO_x (ULN) combustion system. The report discusses the development of an advanced ULN head end combustor with modified fuel injectors to combat flashback, and a fuel staging concept which is referred to as the “distributed combustion system (DCS).” The report shows results from university testing, including micromixing nozzles, and full-scale testing with greater than 80% H₂. They also demonstrate NG operation at greater than J class temperatures with reasonable NO_x emissions. The advanced head end combustor and Distributed Combustion System are reproduced from the report in Fig. 7.

Siemens has also released patents that show the DCS concept on their combustors [31] and the associated fuel ducting [32]. Figure 8 presents an example of one of these patents. Like the drawing and photograph of the DCS in Fig. 7, Fig. 8 portrays a secondary fuel injection system downstream of the main combustor. The section view in the patent shows six discrete fuel injection holes, with an assumed ten total injectors for the combustor. The fuel injectors are numbered differently, suggesting a degree of fuel staging (i.e., different distributions of fuel between the staged injectors).

Mitsubishi Heavy Industries (MHI) is currently working to accommodate larger fractions of H₂. Their strategy is a multicluster diffusion burner, consisting of many small diffusion burners. Each nozzle can mix fuel and air. The result is many small nonpremixed (or partially premixed) flames which are robust against flashback, but with rapid air mixing to reduce temperatures and control NO_x emissions. This project is reported to be in early stages. The technology has been tested on small gas turbines and is planned as a retrofit to replace the combustion components (fuel nozzle, combustor basket, transition piece, igniters, and flame detectors). In addition, a fuel line modification will be required to accommodate the higher flow that will be required for high H₂ fuels. To date, the technology has been tested at 30% H₂ in a 1600 °C (J class) DLN combustor. A fuel gas composition with 30% H₂ provides a 10% CO2 reduction relative to pure NG fuel. A 100% H₂ goal is slated for 2030.

Mitsubishi Heavy Industries has published the technical developments that are being incorporated into the multicluster diffusion burner. The burner concept consists of many small, very short premixing passages with rapid mixing to minimize flashback risk. Each burner clusters these passages together so that they inject fuel and air with swirl to induce flow recirculation (see Fig. 9 for an outline of the general concept).

The multicluster diffusion burner is designed to establish a lifted flame at the stagnation point of the recirculation zone. For example, Fig. 10 presents a cartoon of the flame shape for a single burner published by MHI [34]. The lifted flame is key to the low NOx strategy so that the separate fuel and air streams from the cluster can complete mixing upstream of the flame.

Mitsubishi Heavy Industries has released several publications that show multiple fuel stages for this concept, similar to modern day lean premixed systems.

Figure 11 identifies three general fuel stages, including a liquid fuel nozzle, a pilot burner (F1), and main burners. The six main burners are further partition into four additional fuel stages according to Fig. 12. These stages consist of two groups of three outer burners, with each group on a separate fuel stage. The groups are named F2 and F3. Furthermore, each of these groups has an inner fuel stage and an outer fuel stage.

Figure 12 also illustrates the general strategy for fuel stage manipulation during loading. The figure shows oil-only operation on the left, and hints that oil-only operation can access low loads. The figure also shows fueling to the F1 (pilot stage) and F2-1, F3-1 (inner stages) at lower loads with fueling to all stages at high loads. The figure also hints at an “Advanced” versus “Initial” mode. The authors speculate from the figure that the advanced mode uses more outer fuel for better premixing and lower NO_x.

Extensive testing has already been performed on this concept, which demonstrates significant investment in and maturation of the technology. For example, Fig. 13 shows the load and operating mode dependence of emissions and combustion dynamics. The figure suggests that the manufacturer has been successful at mitigating combustion dynamics and limiting NO_x emissions with the available fuel splits. The figure also shows that today's challenges of fuel system tuning to balance emissions and dynamics will likely remain relevant in next generation systems.

In addition to the multicluster diffusion burner, MHI has developed a combustion system for H₂-rich syngas fuels in oxygen-blown IGCC plants [34]. The nozzles in this combustion system somewhat resemble the multicluster diffusion burner. In this case, the concept is a transition from a premixed combustor to a diffusion combustor for flashback prevention, and injection of diluents (water, steam, and/or nitrogen) for NOx abatement.

A recent Turbomachinery International article reports an interview with Ansaldo, where the manufacturer explains that a high H₂ fuel supply will require attention to the fuel skid and controls to avoid leakage and maintain safe operation [35]. Ansaldo discusses the need for rapid mixing to enable high H₂ technology. They describe a strategy for flashback mitigation that requires high premixer exit velocities with appropriate boundary layer control. Furthermore, they discuss the need for a more robust flame stabilization mechanism that can handle changes in flame speed (for example, a stabilization mechanism that does not blow out when H₂ content is reduced). Ansaldo Thomassen's low emission combustion combustor is currently capable of up to 25% H₂. The Power Systems Manufacturing (PSM) Flamesheet™ combustor can handle up to 40% H₂. According to Ansaldo Thomassen CEO Peter Stuttaford “tests are ongoing to demonstrate 80% H₂ capability for Flamesheet, and development work is ongoing for a 100% H₂ demonstration.” Moreover, test data has been published showing various successes for Ansaldo Energia with testing standard GT36 hardware up to 70% H₂ and essentially full range capability is possible with corresponding derates [36].

A recent article from Power Magazine in May of 2019 nicely summarizes the development efforts of the different OEMs to boost H₂ capabilities, with some of the noteworthy efforts expanded here [8]. Mitsubishi Hitachi Power Systems (MHPS), Siemens Energy, and Ansaldo Energia are all working diligently to secure a place in the market with the technologies talked about previously. MHPS has 29 units with H₂ capabilities ranging from 30 to 90% but notes the key challenges of NOx emissions and flashback as the technology evolves. MHPS has several conversion projects in the works. One is in the Netherlands at Nuon's Magnum power plant that will be using a diffusion combustor (CCGT M701F 440 MW) firing 100% H₂ by 2024 [37]. This project is estimated to reduce CO₂ emission by 2 Mt/year. Siemens and Ansaldo have H₂ capabilities noted as well, with both focusing on new technologies to reduce emissions levels and push toward 100% H₂ capabilities congruent with NG DLN type technologies. Table 4 outlines the capabilities known to be current as of publication with any future capabilities that are in development.

Decarbonization is happening at different levels throughout the world. Specifically, in the USA, carbon taxes and carbon caps have just been discussions; however, campaigns like that of “beyond coal” have played a major role in fully shuttering coal-fired power plants since 2011 with aims for all units by 2030 [40]. Along with this June 2019 news report, Michael Bloomberg announced a renewed investment of $500 M USD for “beyond coal,” which is setting eyes on preventing new gas plant construction. Thus, having a path for decarbonized gas turbines in the USA is necessary now. Current technologies are available for certain circumstances, but not as a broad solution. Lower percentage H₂ containing fuels will work with most systems; however, high percentage to pure H₂ fuels pose challenges and still need robust, low-NO_x solutions.

Original equipment manufacturers are working in these areas with the help of U.S. DOE funding along with other market drivers. Hydrogen combustion was a big topic at the 2019 ASME Turbo Expo, where top officials from the major OEMs, along with EPRI's Tom Alley, outlined the need for decarbonized fuel use and pushed the GT R&D community to keep working toward solutions for H₂ use. The GT community is well positioned but needs to keep pushing in order to be ready as the H₂ production community continues to increase the availability of the carbon-free fuel. Overall, most DLN/DLE/ULN systems can currently handle approximately 60% or less H₂ with many OEMs having test data pushing to 80 or 90%. Continued development and experience are needed with holistic considerations for other combined cycle systems such as the HRSG around temperatures, purge requirements, and exhaust water content, for example.

AFS =

axial fuel staging

DCS =

distributed combustion system

DLE =

dry-low emission

DLN =

dry-low NO_x

F =

flammability limit equivalence ratio

HRSG =

heat recovery steam generator

IGCC =

integrated gasification combined cycle

LBO =

lean blowoff

LHV =

lower heating value

MW =

megawatt

OEM =

original equipment manufacturer

STP =

standard temperature and pressure

ULE =

ultra-low emission

ULN =

ultra low NO_x

WLE =

wet-low emission

Φ =

equivalence ratio