Designing and Testing an Innovative Hydrogen Combustor for Gas Turbines

Abstract

Hydrogen-fueled gas turbines face challenges related to flashback risk, nitrogen oxide (NO_x) emissions, and operational flexibility. In this study, a Center-Graded Spiral Micromixing (CGSM) combustor was designed and experimentally investigated to enhance the robustness of fuel–air mixing under hydrogen-rich conditions. The proposed CGSM concept employs spiral microtubes to induce curvature-driven secondary flows, promoting mixing through airflow-controlled mechanisms rather than relying solely on fuel jet momentum. Numerical simulations were conducted to qualitatively analyze the internal flow and mixing characteristics of the spiral microtubes, followed by pressurized combustor experiments at an inlet pressure of 0.3 MPa and elevated air temperatures. The experimental results demonstrate stable combustion of pure hydrogen under lean conditions, with NO_x emissions being maintained below 25 ppm, corrected to 15% O₂, without observable flashback or combustion oscillations within the designated operating range (from ignition to full load). The combustor further exhibits stable operation with blended hydrogen–methane and hydrogen–ammonia fuels, enabling online fuel switching without hardware modification. Application tests on an 80 kW micro-gas turbine indicate that the CGSM combustor can support stable operation across the full range of load conditions, from ignition to full-load operation, under both simple- and reheat-cycle modes, with performance characteristics that are consistent with established operational standards for micro-gas turbines. These results suggest that the CGSM concept provides a feasible micromixing strategy for hydrogen and hydrogen-rich fuels at a moderate pressure and micro-gas turbine scale.

1. Introduction

Hydrogen is anticipated to play a pivotal role in carbon-free energy systems, serving as both an energy storage option and fuel for peaking power to address the intermittency of renewable energy sources [1,2]. In these scenarios, gas turbines are acknowledged as crucial components owing to their promising efficiency [3]. However, the physical and combustion characteristics of hydrogen differ significantly from those of conventional hydrocarbon fuels, as it is marked by rapid molecular diffusion, a high flame propagation speed, and resistance to stretching. This makes hydrogen susceptible to flashback and combustion instability [4,5]. Consequently, the maturity of pure hydrogen combustion with dry low-NO_x (DLN) technology lags behind that of hydrogen blend fuels. Therefore, developing safe, clean, and highly reliable technologies for hydrogen combustion organization under gas turbine conditions is crucial. To address this challenge, a range of scientific investigations have been conducted, such as moderate or intense low-oxygen dilution [6], sequential combustion [7,8,9], and micromixing [10,11,12,13]. Some authors have also reviewed research advances and technologies for low-pollution hydrogen gas turbines [14,15,16]. Micromixing combustion has been demonstrated to be a promising solution for low-NO_x combustion in pure hydrogen gas turbines [11].

Micromixing combustion focuses on reducing thermal NO_x with multiple small-scale and lean premixed flames, rather than a single flame sheet as in conventional DLN combustors, thus reducing the high-temperature region caused by an inhomogeneous equivalence ratio distribution, as well as the residence time of flue gas in the high-temperature region. Currently, methods for micromixed combustion can be categorized into microdiffusion and micropremixing. The primary purpose of both combustion organization methods is to simultaneously address mixing uniformity and flashback resistance. Several gas turbine manufacturers have developed micromixing nozzles. For instance, Kawasaki Heavy Industries, in collaboration with Aachen University of Applied Technology in Germany, developed a microdiffusion combustion structure based on transverse jets [17,18]. Test results demonstrated that NO_x emissions, under a 1500 K firing temperature, could reach 35 to 40 ppm@16% O₂. However, the combustion structure of microdiffusion inevitably leads to a high local equivalence ratio region in the combustion zone, making it challenging to further reduce NO_x. The approach chosen by General Electric involves constructing micropremixing combustion using a cross-jet flow to promote fuel and air mixing [19,20]. Their bench test results suggested that with a 60%H₂-40%N₂ fuel, NO_x emissions of a full-can model combustor were below 10 ppm@15% O₂ under F-class conditions. Multiple premixed flames were demonstrated to be well established downstream of the nozzle, as was expected; however, it was relatively easy to induce a flame-holding phenomenon in the premixing chamber. Mitsubishi Heavy Industries designed a multi-cluster combustor with several small and short cyclonic premixing passages to mix fuel and air rapidly and form a lifting flame at the stationary point of the recirculation zone. A seven-nozzle configuration was designed to suit their annular tube-type combustor [21]. Ambient pressure tests revealed that NO_x emissions did not exceed 25 ppm at a hydrogen content of 65%. However, no data on pure hydrogen fuel has been obtained. Recently, there has been some progress in the research on micro-hydrogen nozzles. These research efforts primarily focus on investigating the effects of the swirl number, tube diameter, and fuel jet parameters in micromixing nozzles on NO_x reduction, along with an assessment of their application prospects [22]. Prior studies [23,24] introduced the process of converting a commercial micro-gas turbine to run on a blended methane/hydrogen fuel. Testing results revealed that the highest NO_x emissions recorded for pure hydrogen and full-load operations were 62 ppm@15% O₂.

In this study, we propose an Enhanced Micromixing concept. This approach represents a functional evolution designed specifically for hydrogen fuels: it retains the multi-point injection characteristic of micromixing but integrates spiral-induced Dean vortices and flow contraction to achieve a superior balance between NO_x reduction and flashback resistance.

Although micromixing combustion excels in low NO_x emissions and flashback resistance, its fuel–air mixing efficiency is highly dependent on the fuel injection momentum [25,26,27] (i.e., the injection velocity and pressure). Research results exhibit compromised mixing uniformity below the 50% fuel momentum threshold, limiting operational flexibility. This leads to a rapid decline in mixing uniformity under significant variations in load and fuel composition, resulting in increased NO_x emissions and reduced combustion stability [28,29,30]. Therefore, enhancing the mixing and combustion robustness of micromixing concepts over a wide range of loads and fuel compositions remains a critical challenge.

To address the problem of efficient mixing in the organization of low-NO_x hydrogen combustion and broaden the understanding of matching between hydrogen combustors and gas turbines, this paper proposes an airflow-driven turbulence-controlled mixing enhancement strategy, based on an innovatively designed spiral-integrated micromix unit. Based on this design, a Center-Graded Spiral Micromixing (CGSM) nozzle was tested and confirmed. This work explores the use of Dean vortices (secondary flow accounting for approximately 15% of the main flow) to achieve blending uniformity that is weakly dependent on the fuel momentum, enabling low NO_x emissions under variable load and fuel composition conditions.

Unlike conventional micromix combustors that primarily rely on fuel jet penetration to achieve rapid mixing, the proposed CGSM concept introduces an airflow-driven mixing mechanism. By deliberately generating Dean-vortex-induced secondary flows within spiral microtubes, the mixing process is weakly dependent on the fuel momentum. This fundamental decoupling between fuel momentum and mixing uniformity represents a key distinction from existing micromixing and swirl-based hydrogen combustor concepts.

2. Model Design and Numerical Analysis

2.1. Spiral Microtube Design

2.1.1. Design Concept and Geometric Model of Microtube

The mixing performance of conventional micromix nozzles often depends heavily on the penetration depth of the fuel jet, which limits their low-emission operating range. To mitigate this dependency and further reduce emissions, this study proposes an aerodynamic mixing concept that utilizes mainstream air to generate controlled turbulence, thereby improving the mixing process. Accordingly, a spiral microtube mixing unit was developed, leveraging geometry-induced secondary flows to enhance the mixing efficiency and broaden the operational range. Through the integration of spiral flow trajectories and a converging variable cross-section, the microtube promotes curvature-induced secondary flows and sustained vortical activity in the mainstream air. As evidenced by the increased turbulence level and coherent vortical structures discussed in Section 2, these flow features are associated with stronger interfacial disturbance and enhanced micromixing between the fuel jet and the air mainstream, thereby reducing the dependence of mixing effectiveness on jet penetration alone and enabling more uniform mixing over a wide load range.

From the perspective of mixing mechanisms, the spiral tube incorporates a spiral angle into the conventional straight tube configuration, enabling the fluid to undergo circumferential motion while advancing axially. This design utilizes centrifugal force to induce secondary flows and generate Dean vortices, which serve to enhance radial mixing. As the fluid flows through the microtube, the high-velocity fluid in the core region experiences a greater centrifugal force than the low-velocity fluid in the near-wall region. This centrifugal force gradient drives the fluid to form a pair of counter-rotating vortices on the cross-section perpendicular to the mainstream direction, known as Dean vortices. It should be noted that in this study, the Dean number is employed to qualitatively interpret the role of curvature-induced secondary flows in enhancing fuel–air mixing, rather than as an optimization parameter. The Dean number is defined as:

$${\mathrm{D}}_{\mathrm{e}}={\mathrm{R}}_{\mathrm{e}}\sqrt{{\displaystyle \frac{{\mathrm{D}}_{\mathrm{h}}}{{2\mathrm{R}}_{\mathrm{c}}}}}$$

(1)

This parameter characterizes the relative strength of curvature-induced centrifugal effects compared with viscous forces, where Re is the Reynolds number, D_h is the hydraulic diameter, and R_c is the radius of curvature of the flow path. According to this definition, an increase in the Dean number intensifies secondary flows and associated radial transport, thereby promoting scalar homogenization and improving the mixing uniformity, while simultaneously leading to increased frictional pressure losses. For the present spiral microtube, the hydraulic diameter (D_h) and curvature radius (Rc) were obtained directly from the CAD geometry. Under the representative operating condition used for the microtube-flow analysis (Uout = 120 m/s, Tin = 483 K, Pin = 0.3 MPa), the Reynolds number is Re ≈ 26,700 and the corresponding Dean number is De ≈ 2460. These quantified parameters indicate that the present design operates in a regime where curvature-induced secondary flows are sufficiently developed to promote radial transport, while avoiding excessive frictional penalties.

Based on this physical consideration, the spiral microtube geometry adopted in the present CGSM design intentionally operates at De ≈ 2460, where stable Dean vortices are established to provide mixing enhancement with a controlled pressure-loss level. This design choice is consistent with the observed secondary flow intensity and the experimentally demonstrated combustion stability and flashback resistance. A systematic parametric optimization over De (via D_h and Rc) is not the primary objective of the present study; however, it will be considered in future work to further quantify the design sensitivity and extend the database for practical combustor development.

In this work, the spiral tube induces a stable Dean vortex system through the combined effects of the spiral angle and variable cross-section. The presence of Dean vortices propels the central fluid toward the outer wall while “scraping” the wall-bounded fluid toward the inner wall, creating a spiral secondary flow that persists throughout the entire pipeline. Although the velocity magnitude of the secondary flow is relatively modest (approximately 15% of the mainstream velocity), its continuous radial transport capability will play a critical role in enhancing both the mixing efficiency and the robustness of the mixing performance.

Figure 1 presents a three-dimensional schematic of the helical microtube configuration. The circumferential deflection angle of the microtube centerline is designed to be 28°. The cross-section transitions linearly from a rectangular shape at the inlet to a circular shape at the outlet, with the cross-sectional parameters—height, width, and curvature—varying gradually. The axial length of the spiral microtube is 60 mm, the characteristic diameter at the outlet is 3 mm, and the area ratio of the inlet to the outlet of the microtube is 1.5. Two hydrogen fuel inlets with a diameter of 0.5 mm were arranged 3 mm away from the microtube inlet. The designed flow velocity at the outlet is 120 m/s. The geometric design is synergistically optimized for both combustion performance and industrial manufacturability. Among the key parameters, the 0.5 mm diameter of the hydrogen fuel inlets is the most tolerance-sensitive feature for multi-tube repeatability, because small deviations in hole size can directly alter the initial fuel split among parallel micromixing passages and thus increase tube-to-tube non-uniformity. The selected diameter (~0.5 mm) represents a practical compromise among the required fuel delivery capacity, the desired injection momentum level, and the achievable dimensional repeatability of commercial SLM manufacturing. Therefore, strict tolerance control of this dimension is emphasized in fabrication and quality inspection. The critical dimensions are well within the processing capability of commercial Selective Laser Melting (SLM) systems, and the feasibility of the manufacturing method will be elaborated in detail in Section 3.1.

The spiral microtube exhibits the following key characteristics:

• Continuous effective turbulence: Its geometric configuration facilitates the generation of controllable intense turbulence via air, incorporating quantifiable Dean vortices with measurable vorticity magnitudes. These vortices collaborate synergistically with the fuel jet to augment the efficiency of fuel–air mixing.
• Boundary layer modulation: The convergent cross-sectional profile sustains a favorable pressure gradient, thereby effectively suppressing flow separation at the outlet.

2.1.2. Numerical Simulation Method

This study employed ANSYS Fluent 2023 R1 (Version: 2023 R1; ANSYS, Inc., Canonsburg, PA, USA) to simulate the flow, mixing, and combustion processes, incorporating the combustion reaction mechanism proposed by Li et al. [31], which comprises nine species and 19 elementary reactions. The realizable k−ε turbulence model was adopted, along with the flamelet-generated manifold (FGM) combustion model. To accurately resolve the flow field characteristics within the microtube, a refined mesh was applied in regions exhibiting high turbulence intensity. The mesh resolution ranged from a minimum size of 0.01 mm to a maximum of 0.1 mm, with a smooth transition that was governed by a size function and a growth rate of 1.05. The computational domain of the microtube consists of approximately 4.2 million cells. This numerical approach has been validated and demonstrated to accurately capture the flow and mixing behavior in a microtube [32]. In the analysis of flow characteristics, the turbulence intensity (I) is defined as the ratio of the root-mean-square of the turbulent velocity fluctuations (u′) to the mean velocity (${\mathrm{U}}_{\mathrm{a}\mathrm{v}\mathrm{g}}$), calculated as follows:

$$\mathrm{I}={\displaystyle \frac{{\mathrm{u}}^{\prime }}{{\mathrm{U}}_{\mathrm{a}\mathrm{v}\mathrm{g}}}}={\displaystyle \frac{\sqrt{\frac{1}{3}({{\mathrm{u}}_{\mathrm{x}}^{\prime }}^2+{{\mathrm{u}}_{\mathrm{y}}^{\prime }}^2{{+\mathrm{u}}_{\mathrm{z}}^{\prime }}^2)}}{{\mathrm{U}}_{\mathrm{a}\mathrm{v}\mathrm{g}}}}$$

(2)

This parameter is used to quantify the level of mixing-enhancing turbulence generated within the spiral structures. In the present study, CFD is used as qualitative support to elucidate dominant flow structures (e.g., secondary flows and vortical activity) and mixing tendencies, while quantitative prediction of emission levels is beyond the scope of the model. The inlet air temperature used in the numerical simulations corresponded to simple-cycle operating conditions, whereas the experimental tests were performed under regenerative-cycle conditions with elevated inlet air temperatures. The numerical simulations focused on elucidating the dominant flow and mixing mechanisms, which are primarily governed by geometric features and flow structures, while the experimental conditions represented practical combustor operation. Therefore, this difference in inlet temperature did not alter the qualitative conclusions regarding flow-structure-driven mixing enhancement and flame stabilization, which are the primary targets of the present numerical analysis.

2.1.3. Flow and Mixing Characteristics

The boundary conditions for the microtube are presented in

Table 1. Key parameters of numerical calculation.

Parameter	Value
air velocity of the outlet (m/s)	120
equivalence ratio	0.32
air temperature (°C)	210
air pressure (MPa)	0.3

. Figure 2a illustrates the flow field characteristics of the spiral microtube’s nozzle. Typical flow field characteristics can be observed through the four representative microtube cross-sections shown in this figure. Here, Z represents the axial length from the inlet at the given cross-section. Notably, the initial section of the spiral microtube induces a robust secondary flow owing to complex changes in the wall structure, considerably enhancing hydrogen–air mixing under the combined impact of square-to-circle, contraction, and rotation transitions. Meanwhile, Figure 2a shows that the vortex intensity within the axial length of the spiral microtubes exhibits no significant attenuation, and the radial velocity reaches a maximum of approximately 15% of the mainstream velocity, which helps sustain intensified fuel–air mixing. The distribution of hydrogen equivalence ratios at the exit of the spiral microtubes, shown in Figure 2b,c, indicates that the proportion of equivalence ratios below 0.53 at the outlet is as high as 88%. This ensures that the temperature of the adiabatic flame does not exceed 1950 K. In comparison, for straight microtubes with the same mixing length, this value decreases by 21%. This comparison underscores the advantages of the spiral micromixing structure.

Figure 2 serves as a baseline comparison to isolate the effect of the spiral trajectory. Unlike the straight microtube configuration, the spiral geometry induces intense Dean vortices that drive cross-stream mixing. This aerodynamic mechanism is supported by the significantly narrower equivalence-ratio distribution in the spiral tube compared with the straight baseline (Figure 2c), and is consistent with the strengthened cross-stream transport induced by curvature-driven secondary flows. To quantitatively support the above interpretation, turbulence statistics extracted from the same CFD dataset are provided in the paper. Specifically, Figure 2d compares the turbulence intensity between the straight and spiral microtubes under identical boundary conditions. The spiral microtube maintains a higher turbulence level along the micromixing length and at the outlet plane, which is consistent with sustained vortical activity and enhanced scalar homogenization.

Under identical geometric and operating conditions (axial length of 60 mm and outlet diameter of 3 mm), Figure 2e compares the coherent vortical structures in a straight circular microtube and the spiral converging microtube using the normalized omega (ω) criterion. To further clarify the internal flow physics, Figure 2f additionally provides two representative cross-sectional fields at selected axial locations, showing that the spiral geometry sustains a stable secondary-flow pair along the micromixing length, whereas the straight tube exhibits rapid decay downstream of the injection region. The Omega (ω) method is adopted to identify the vortex structures. It is defined as

$$\omega ={\displaystyle \frac{{\Vert B\Vert }_F^2}{{\Vert B\Vert }_F^2+{\Vert A\Vert }_F^2+\epsilon }}$$

(3)

where A and B represent the symmetric and antisymmetric parts of the velocity gradient tensor, respectively, and ω is a small positive parameter to avoid division by zero. ω is a scalar bounded between 0 and 1, representing the relative contribution of rotational motion to the total deformation. And the ω = 0.52 iso-surface is adopted to delineate vortex boundaries more clearly [30]. In Figure 2e,f, the ω distribution is used to delineate the boundaries and core concentration of the vortices. While effective for identification, this scalar field does not carry directional information. The gray surface denotes the ω = 0.52 iso-surface, and the cross-section shows the local ω distribution. As shown in Figure 2d, the straight tube exhibits pronounced vortical structures only within a short axial distance downstream of the fuel injection region, indicating that turbulence-enhanced mixing decays rapidly along the tube. In contrast, Figure 2d demonstrates that the spiral converging microtube sustains strong vortical structures over the entire tube length, implying persistently enhanced turbulence and continuously promoted fuel–air mixing along the full micromixing path. Additionally, the high-concentration zone at the nozzle exit is centered relative to the cross-section, and the low-velocity zone near the circumferential wall has a relatively low equivalence ratio. This concentration distribution is conducive to the suppression of boundary layer flashback.

To explicitly visualize the counter-rotating nature of the Dean vortices and exclude the interference of wall shear vorticity, Figure 2g presents the distribution of the streamwise vorticity component on the cross-sections. A distinct dipole structure is observed, featuring symmetric regions of positive (red) and negative (blue) vorticity. This confirms that the spiral geometry successfully induces a stable pair of counter-rotating Dean vortices, which are responsible for the convective transport of fluid between the core and wall regions. For comparison, a straight tube under identical conditions would exhibit a near-zero streamwise vorticity field (dominated solely by axisymmetric wall shear), lacking the organized secondary flow structures observed here. This contrast highlights the mechanism by which the spiral geometry actively enhances radial mixing through convective transport.

As demonstrated by the numerical results, the localized flow structures effectively enhance the fuel–air interaction. Consequently, this ‘Enhanced Micromixing’ strategy represents a synergistic design at the single-tube level: the spiral geometry induces localized secondary flow structures (Dean vortices) to promote rapid fuel–air homogenization for NO_x reduction, while a downstream contracting nozzle configuration is implemented to increase local flow velocity and suppress the risk of flashback. This coordinated approach optimizes the trade-off between ultra-low emissions and operational safety under high-hydrogen conditions. These fundamental principles of single-tube design provide the basis for the integrated nozzle configuration discussed in the following section.

2.2. CGSM Nozzle

Based on the spiral microtube, the CGSM nozzle (depicted in Figure 3a) is designed with pilot, inner main, and outer main stages in radial order, with an 80 kW micro-gas turbine as its application object. In this design, the swirl number (S) is employed to quantify the rotation intensity, defined as the ratio of the axial flux of the angular momentum (G_Φ) to the axial flux of the axial momentum (G_X) multiplied by the equivalent nozzle radius (R). The swirl number is defined as:

$$\mathrm{S}={\mathrm{G}}_{\mathsf{\Phi }}/\left({\mathrm{G}}_{\mathrm{X}}\mathrm{R}\right)$$

(4)

Based on this definition, the pilot stage adopted a direct-injection diffusion nozzle, combined with a single-stage air swirl (swirl number 0.87) to form a pilot flame. The inner and outer main combustion stages feature a spiral microtube structure, which facilitates lean premixing of hydrogen and air through variable cross-sections, shrinkage, and a spiral microtube design.

The characteristic diameter of the microtube outlet is 3 mm. The main combustion stage is further divided into an inner and outer ring to expand the load regulation range. The inner ring is arranged with 2 circles of microtubes, totaling 44, with swirl numbers of 0.7 and 0.6 from the inside to the outside, and the microtube outlets are located on a conical surface with a half-cone angle of 33.7°. The outer ring is arranged with 4 circles of microtubes, totaling 148, with a swirl number of 0.5 for each circle and an axial length of 60 mm. In this configuration, the inner ring uses relatively higher swirl numbers to strengthen the near-axis recirculation, thereby enhancing flame anchoring and improving stability during ignition and low-load operation. In contrast, the outer ring adopts lower swirl levels to mitigate strong vortex-vortex interactions among the shear layers of adjacent microtube outlets and to reduce their coupling with the corner recirculation zone. Such coupled vortical structures can intensify unsteady heat-release oscillations, which may increase the propensity for flashback. The reduced outer-ring swirl may also help avoid excessive residence time in localized high-temperature regions. Geometric and flow parameters used to quantify Re and Dean number (D_e) for the spiral microtube can be seen in

Table 2. Geometric and flow parameters used to quantify Re and Dean number (De) for the spiral microtube (Serial numbering proceeds from the innermost to the outermost).

Number	Dh (mm)	Rc (mm)	Uout (m/s)	Tin (°C)	Pin (MPa)	Re	De
No.1	3	54.0	120	210	0.3	26,700	4440
No.2	3	91.6	120	210	0.3	26,700	3410
No.3	3	150.5	120	210	0.3	26,700	2660
No.4	3	177.0	120	210	0.3	26,700	2460
No.5	3	208.1	120	210	0.3	26,700	2260
No.6	3	234.5	120	210	0.3	26,700	2130
No.1	3	54.0	120	280	0.3	21,300	3550
No.2	3	91.6	120	280	0.3	21,300	2730
No.3	3	150.5	120	280	0.3	21,300	2130
No.4	3	177.0	120	280	0.3	21,300	1960
No.5	3	208.1	120	280	0.3	21,300	1810
No.6	3	234.5	120	280	0.3	21,300	1700

. As the microtube row number increases from the inner to the outer rings, the radius of curvature (R_c) is adjusted, leading to a corresponding variation in the Dean number (D_e), which highlights the enhancement of secondary flows across different radial positions. This geometric grading, in conjunction with the radial distribution of swirl intensity discussed above, is specifically designed to manage the turbulent kinetic energy distribution across the nozzle face. By coordinating the curvature-induced secondary flow with the prescribed swirl levels, the design effectively balances the mixing enhancement in the core region with the mitigation of excessive vortex-vortex interactions and pressure losses in the outer radial span. To prevent hydrogen flashback, the flow area of the microtube channels contracts in the flow direction, with an area contraction ratio of about 1.5. The flow area of each circle of microtubes shrinks along the flow direction, as shown in Figure 3b. In the figure, the circles of microtubes are numbered in sequence from the inside to the outside in the radial direction. The designed flow velocity inside the spiral microtube is 120 m/s. The distribution of air across various stages in the nozzle is detailed in

Table 3. Airflow distribution scheduling scheme.

Nozzle Stage	Air Distribution Ratio, %
Pilot	10
Inner main	20
Outer main	70

.

Each microtube of the central staged micromixing nozzle has a pair of fuel injection holes along the radial direction. According to the different flow areas of each microtube, the fuel hole diameters range from 0.43 to 0.45 mm. The axial positions of the fuel holes of the nozzle are located 3 mm from the end of the microtube inlet, corresponding to fuel–air mixing lengths of 57 mm. It is worth noting that three individually adjustable fuel supplies enable multiple combustion modes to adapt to the load regulation requirements of gas turbines. Figure 3c presents the numerical calculation results for the nozzle combustion flow field under the aforementioned conditions. Isosurfaces with a reaction progress variable of 0.8 were employed to delineate the flame structure. The results indicate that the micromixed flame in the main combustion stage of the nozzle exhibits an independent flame structure with essentially no interaction. Figure 3d shows that the pilot stage exhibits a larger angle of inclination, compressing the downstream flow field of the inner ring in the main stage. This will facilitate burnout downstream of the microtube flame under low-load conditions.

Although a direct quantitative validation of the numerical model is challenging due to the micro-scale geometry and limited optical access under pressurized conditions, the predicted flame structure shows clear qualitative consistency with experimental observations. This qualitative agreement between the simulated flame structure (Figure 3c) and the experimentally observed OH distribution further supports the feature-level qualitative agreement between the numerical model and experimental data. A similar flame topology was observed experimentally in the OH radical distribution captured by the UV camera, where spatially separated and uniformly distributed reaction zones were identified. This feature-level qualitative agreement supports the use of the numerical results as interpretive evidence for dominant flow–flame structures, rather than as a quantitative validation of emissions.

3. Experimental Setup

Combustor tests at elevated pressures and application tests of a micro-gas turbine were conducted to evaluate the overall combustion performance of the CGSM nozzle for hydrogen combustion, as well as its compatibility with a gas turbine. The experimental setup and procedures are detailed in this section.

3.1. Hydrogen Combustor

A low-pollution hydrogen combustor test piece was designed based on the CGSM nozzle, as shown in Figure 4a. It comprises a CGSM nozzle, flame tube, outer liner, and observation window, with a transparent design to allow for flame structure observation. Pressurized and heated air is introduced into the combustor, with three fuel paths (pilot stage, inner main stage, and outer main stage) facilitating adjustment of the combustion mode. The combustor test piece was manufactured using Selective Laser Melting (SLM) technology with a dimensional tolerance of ±0.05 mm, which ensures the geometric fidelity of the 3 mm micro-tubes and 0.5 mm fuel holes. For multi-tube nozzle assemblies (44 inner main-stage and 148 outer main-stage microtubes), the dimensional consistency between individual microtubes is controlled within ±0.05 mm, ensuring uniform airflow and fuel distribution across the entire nozzle. Reproducibility validation with two identical combustor test pieces confirms that their NO_x emissions and combustion stability differ by <4% under identical operating conditions (0.3 MPa, 280 °C inlet air, pure hydrogen fuel), validating the reliability of the manufacturing process.

3.2. Experimental System

Figure 4 illustrates the hydrogen DLN combustor (Figure 4a), experimental bench setup, and arrangement of measurement equipment (Figure 4b). Pressurized, preheated air enters the head nozzle counter along the combustor ring cavity, while fuel is delivered through the cylinder assembly and measured by the flow meter in the pilot, inner main, and outer main stages. Flue gases resulting from fuel combustion and air in the combustor are mixed with air in the mixing holes, cooled, and then discharged.

Key measurement systems include a mass flow meter for fuel mass flow rate, an S-type Pitot tube for total air flow rate measurement, a differential pressure transmitter for total pressure loss in the combustor, and a Testo 350 flue gas analyzer for NO_x emission measurement. A sampling probe extracted flue gas downstream of the primary combustion zone, which was then cooled and fed into the flue gas analyzer for compositional measurements. The sampling probe was made of tungsten alloy, with an inner diameter of 1 mm and four 0.5 mm diameter micropores that were evenly distributed radially in the primary combustion zone. Dynamic pressure sensors (Kulite XTEH-10L-190M series) were installed in the combustor end caps, primary combustion zone, and tertiary fuel supply tubes to monitor the dynamic pressure. The flame structure was measured simultaneously by a charge-coupled device (CCD) and a UV camera. A Sony CCD camera (ILCE-7S and visible-light capability) was used to monitor the flame structure and capture the dynamic characteristics of the flame. The UV camera, a Lucid ATX204S-MC with a standard UV fixed-focus lens with a focal length of 25 mm, captured the heat released in the combustion zone by monitoring the OH distribution. A 310 nm filter was also added.

3.3. Gas Turbine Application Test

A complete machine application test was conducted on an 80 kW micro-gas turbine. Figure 5 shows the micro-gas turbine (Figure 5a) and the arrangement of its system (Figure 5b). This gas turbine is equipped with a regenerator and has two operating modes: simple cycle and regenerative cycle. In the experimental setup, an auxiliary fan was employed to facilitate two key operational phases: the startup of the test bench and the turning gear operation post-test. The main airflow cycle proceeded as follows: ambient air was drawn in and pressurized by a centrifugal compressor and then directed to an air preheater, where it absorbed heat from flue gas. The preheated air was subsequently introduced into the combustor’s test piece, where it participated in fuel combustion and mixing processes. The resulting combustion products (working fluids) were channeled into a radial turbine, whose rotational output drove the centrifugal compressor to sustain its normal operation. Exhaust gas discharged from the turbine passed through the air preheater (serving as the heat source for air preheating) before being vented to the atmosphere.

To support diverse experimental objectives, the test system was configured with multiple fuel supply lines, enabling the simultaneous delivery of hydrogen, ammonia, and methane. Downstream of the combustion test section, equipment for injections of secondary fuel and cooling water was integrated; by adjusting the injection rates of secondary fuel or cooling water, the turbine inlet conditions (e.g., temperature, pressure, and equivalence ratio) could be precisely tuned to match various test scenarios. Additionally, a bypass pipeline was installed on the flue gas side of the air preheater, and a control valve on this pipeline was used to regulate the fraction of flue gas entering the preheater—this mechanism allowed for accurate control of the inlet air temperature of the combustor.

Comprehensive instrumentation was deployed to monitor critical parameters throughout the system:

• Pressure and temperature transducers at key aerodynamic and thermodynamic nodes (e.g., compressor inlet/outlet, turbine inlet/outlet);
• Fuel pressure regulators and mass flow meters for each fuel line;
• A flue gas composition analyzer at the turbine exhaust to quantify combustion products (e.g., NO_x, unburned hydrocarbons);
• A non-contact rotor speed sensor to measure the turbine/compressor’s rotational speed;
• Temperature and pressure gauges for cooling water and lubricating oil circuits to ensure operational safety.

This setup was used to validate the combustion characteristics of the hydrogen combustor configured with a CGSM nozzle and gas turbine.

4. Results and Discussion

4.1. Analysis of Combustion Characteristics of the Hydrogen-Fueled Combustors

4.1.1. Combustion Performance of the Combustor

Fuel supply strategies serve as a core technological approach for balancing combustion performance and operational reliability in combustors. The grading strategies constitute key parts of performance validation tests of combustors. To identify optimal fuel supply schemes, this study systematically investigated NO_x emission characteristics and dynamic pressure pulsation properties within combustors under varying fuel grading ratios, using CGSM nozzles as test subjects. The fuel supply strategies employed in these experiments are shown in

Table 4. Fuel grading strategies.

Test	Nozzle Stage
Mode 1	Pilot + Outer main
Mode 2	Pilot + 5% Inner main + Outer main
Mode 3	Pilot + 10% Inner main+ Outer main

. Throughout the tests, a constant total fuel flow rate was maintained. Under the premise of a fixed fuel proportion for the inner ring of the primary fuel stage, the fuel allocation ratio between the duty stage and the outer ring of the primary fuel stage was adjusted in a gradient manner. This allowed for analysis of the coupled effects of fuel staging patterns on NO_x formation patterns and dynamic pressure characteristics of the combustion. The specific parameter settings for the tests are detailed in

Table 5. Key parameters of the test.

Parameter	Value
air velocity of the outlet (m/s)	120
equivalence ratio	0.32
air temperature (°C)	280
air pressure (MPa)	0.3

.

The effects of varying fuel grading strategies on NO_x emissions and pressure pulsations in the primary combustion zone under the condition of combusting pure hydrogen in a CGSM nozzle are presented in Figure 6a. To evaluate the influence of fuel staging, we varied the inner-main-stage ratio (IMSR = 0%, 5%, and 10%) and the outer-main-stage ratio (OMSR = 55–90%) under otherwise identical conditions. NO_x emissions remained at ultra-low levels (10–15 ppm@15% O₂) when IMSR = 5% across the investigated OMSR range, whereas higher emissions were observed for IMSR = 0% and 10%. The sound pressure level (SPL) increased gradually with OMSR up to ~80%, but rose more rapidly when OMSR exceeded ~80%, indicating intensified unsteady heat-release behavior near the operating limit. Based on these observations, IMSR = 5% with OMSR ≤ 80% provides a practical trade-off between low NO_x emissions and suppressed combustion noise.

Furthermore, the CGSM nozzle exhibited good flame stability during our experiments, and no significant combustion pulsation was observed under the designed conditions. The dynamic pressure pulsation values in the graph are significant, representing the acoustic energy that is distributed at each frequency. The peak values of the pressure pulsations are maintained at a low level, and the spectral characteristics of the pressure pulsations indicate consistency with the increase in the proportion of the outer main stage (Figure 6b). These test results establish a foundation for combustion adjustment based on the tradeoff relationship between NO_x emissions and pressure pulsations.

4.1.2. Flame Structure Characteristics

Figure 7 illustrates the flame structure of the CGSM combustor during pure hydrogen combustion under the designed operating conditions: Figure 7a presents CCD camera-captured visualizations, while Figure 7b displays the OH radical distribution, obtained using a UV camera. From the CCD imagery (Figure 7a), each microtube outlet in the main combustion stage exhibits an independent, stable, and small conical flame structure: the flame is short (≈4D, where D denotes the microtube diameter), characterized by a blue-violet hue and a rotational orientation, with the flame root being firmly attached near the micromixing tube nozzle, indicating excellent flame stability. The front of the pilot flame appears pale red; this is primarily attributed to the presence of hydrogen combustion products within the Venturi section, which alters the emission spectrum of the flame. The enhanced stability of the nozzle flames stems from the microtube unit’s rotating-ejection design: this configuration introduces a mild swirling stabilization effect, complementing the inherent sudden-expansion stabilization of the combustor and thereby improving flame anchoring under pure-hydrogen conditions. The uniform fuel distribution at the center of the microtube outlet maintains its low velocity and low equivalence ratio at the outlet wall (Figure 3c), thereby effectively suppressing boundary layer flashback.

4.1.3. Flashback Boundary and Phenomena

To explore the flashback boundary of the CGSM nozzle, overloading experiments were performed in the main stage: the fuel proportion in the main stage was incrementally increased until flashback initiation. Figure 8 maps the flashback boundary of the nozzle, defined by the characteristic velocity of the outlet (abscissa) and adiabatic flame temperature of the premixed gas (ordinate), for different configurations (inner main stage: version1; outer main stage: version1/2) and air preheating temperatures (210 °C, 290 °C, and 410 °C, i.e., 483 K, 563 K, and 683 K). Herein, versions 1 and 2 correspond to axial distances (between the fuel injection orifice and the microtube outlet) of 19D and 10D, respectively, where D denotes the nozzle outlet diameter (3 mm). The flashback boundaries of the microtubes in both the inner and outer rings of the main combustion stage converge near a premixed flame temperature of 1550 K (at a 563 K air temperature) and exhibit negligible dependence on increases in the nozzle’s characteristic velocity or the air preheating temperature. For the outer main stage, the baseline flashback equivalence ratio is 0.387; notably, this equivalence ratio expands with increasing characteristic velocity—even under elevated preheating temperatures. This phenomenon arises from a velocity-dominated flashback suppression mechanism: as the nozzle’s characteristic velocity increases, the flow velocity in localized low-velocity zones (e.g., boundary layers at the nozzle outlet) is enhanced, which elevates the flow velocity to above the laminar burning velocity of the premixed hydrogen–air mixture, thereby inhibiting flame propagation upstream (i.e., flashback). However, this benefit is offset by a proportional increase in pressure loss, as higher flow velocities induce greater frictional and inertial losses in the micromixing tubes.

Figure 9a shows the flashback event captured by a visible-light CCD camera, and Figure 9b presents the corresponding UV image that makes the evolution of the flame front during the transition more clearly discernible. Experimental observations indicate that the flame undergoes pronounced high-frequency fluctuations as it approaches the flashback limit. These violent oscillations in the flame front serve as a precursor to the loss of flame stabilization, reflecting the intensified thermoacoustic coupling that eventually drives the flame to propagate upstream into the micro-mixing tubes. During flashback, an intense yellow luminosity was observed in the visible images, suggesting severe wall heating and possible surface degradation under high-temperature flame impingement. Since no dedicated spectroscopy was performed in this study, we do not attribute this luminosity to specific atomic species. We therefore report it strictly as an observed optical feature that is consistent with extreme thermal loading of the nozzle surface and potential material heating/ablation effects. These observations suggest that monitoring upstream pressure fluctuations (in the microtube plenum and combustion zone) can provide an early indication of impending flashback. Accordingly, operating conditions can be adjusted through targeted fuel-supply strategies to mitigate flashback.

4.1.4. Fuel Compatibility

To evaluate the fuel compatibility of the designed nozzle and its adaptability to both current mainstream gas turbine fuels and future carbon-free alternatives, combustion performance experiments were conducted for natural gas–hydrogen blends and hydrogen–ammonia blends. All tests were executed under standardized operating conditions: an air pressure of 0.3 MPa, an air preheating temperature of 483 K, and thermal power outputs of 260 kW for units (a)–(c) (Figure 10a–c) and 220 kW for units (d)–(f) (Figure 10d–f). Figure 10 presents the flame morphology (upper row) and OH radical spatial distribution (lower row) across six fuel configurations: (a) pure hydrogen, (b) methane blended with 60 vol% hydrogen, (c) pure methane, (d) pure hydrogen, (e) hydrogen doped with 5 vol% ammonia, and (f) hydrogen doped with 15 vol% ammonia. Specifically, configurations (a)–(c) were designed to investigate the evolution of flame characteristics when the gas turbine fuel is switched from pure hydrogen to pure methane without modifying the nozzle hardware. In contrast, configurations (d)–(f) aimed to explore the variations in flame structure during the combustion of hydrogen–ammonia blends, corresponding to different ammonia cracking ratios.

Notably, seamless online fuel switching between hydrogen and methane was achieved with sustained combustion stability—evidenced by the uniform flame topology (Figure 10a–c) and continuous, well-distributed OH radical profiles—accompanied by NO_x emissions remaining consistently below 25 ppm. For hydrogen–ammonia blends, stable combustion was maintained, even with 15 vol% ammonia addition (Figure 10f), although sporadic NO_x surges were observed. This phenomenon originates from ammonia’s stepwise thermal decomposition, producing reactive NH_x radicals (NH₂/NH/N), which drive rapid fuel–NO_x formation within locally oxidizing zones of the combustor.

The nozzle’s exceptional fuel flexibility derives from two core design attributes:

• Momentum-insensitive mixing performance: The micromixing channel’s geometry is tailored to decouple the fuel–air mixing efficiency from the fuel momentum. For low-momentum (e.g., methane) and high-momentum (e.g., hydrogen) fuels, the channel’s internal flow field sustains near-homogeneous premixing (reflected in the consistent OH distribution across all cases in Figure 10), eliminating fuel-specific mixing inhomogeneities.
• Synergistic flame stabilization mechanism: The integration of sudden-expansion recirculation and swirling flow provides robust flame anchoring across fuel types. For high-burning-velocity fuels (e.g., hydrogen), the swirling flow enhances the flow shear to suppress flashback; for low-burning-velocity fuels (e.g., methane), the sudden-expansion recirculation zone sustains flame residence time-both effects manifest in the attached, spatially uniform flames observed in Figure 10. This fuel adaptability holds critical technological significance: it enables gas turbines to transition seamlessly from conventional hydrocarbon fuels to carbon-free hydrogen–ammonia blends, thereby supporting the scalable deployment of low-carbon power generation systems.

To further characterize the nozzle’s fuel flexibility and combustion performance, analyses of the flame length (derived from the upper-row flame morphology in Figure 10) and uniformity of the OH distribution (from lower-row OH images in Figure 10) were conducted across the six fuel cases. The metrics for flame length and OH distribution uniformity were quantified via image-processing-based semi-quantitative methods: For each upper-row flame morphology image, the axial distance from the nozzle outlet to the flame tip was measured using pixel calibration (scaled to the nozzle diameter, D) and then normalized to D for cross-case comparison. Lower-row OH images were converted to grayscale, and their uniformity was calculated as the reciprocal of the coefficient of variation in pixel intensity. A value of 1 corresponds to a flat intensity profile (fully uniform), while values approach 0 for highly heterogeneous intensity distributions. The results of these analyses are summarized as follows:

• Flame length trends: The flame length increases slightly with lower fuel reactivity (e.g., pure methane > hydrogen–ammonia blends > pure hydrogen), which is consistent with the lower burning velocity of less reactive fuels (methane < ammonia-doped hydrogen < pure hydrogen). The small variation indicates the nozzle’s ability to constrain the flame size, regardless of fuel type.
• OH distribution uniformity: All cases exhibit high uniformity, confirming the nozzle’s momentum-insensitive mixing performance. Minor reductions for blended fuels (e.g., 15 vol% NH₃-H₂) arise from subtle differences in fuel diffusion rates but remain within a range that ensures stable combustion.

This analysis further confirmed that the combustion of this nozzle under various fuel combinations has excellent robustness and stability, mainly due to the mixing effect, which reduces the dependence on fuel momentum.

4.2. Application Test of CGSM Nozzles in Micro-Gas Turbines

4.2.1. General Situation

To evaluate the load adaptability of the Center-Graded Spiral Micromixing (CGSM) nozzle and assess the effectiveness of the proposed fuel staging control strategy, loading tests were conducted using 100% hydrogen over the full operating cycle. These tests demonstrated stable load-following operation with limited impacts on the working line of the target micro-gas turbine, supporting controllable operational risk. Furthermore, Figure 11 compares the operating characteristic lines of the CGSM nozzle and a conventional swirled nozzle under the same control schedule. The CGSM nozzle exhibits an operating line comparable to the conventional case and shows similar characteristic-line behavior for methane and hydrogen, indicating fuel flexibility without requiring major changes in operating-map control.

4.2.2. Variable Load Performance

Figure 12 illustrates the correlations among the outer main-stage flame temperature, NO_x emissions (corrected to 15% O₂), and dominant combustion sound power level (SPL) during load variation tests. Here, the abscissa denotes the outer main-stage flame temperature (K), a proxy for the intensity of combustion heat release; the left ordinate represents the NO_x concentration (ppm@15% O₂), with solid squares and circles corresponding to the load-increase and load-decrease paths, respectively. The right ordinate reports the dominant combustion SPL (dB, open markers), which reflects the intensity of combustion pressure fluctuations.

Integrated with the control strategies validated in the pressurized tests, the observed hysteresis between the load-increase and load-decrease paths is primarily attributed to (i) thermal-state memory of the combustion zone (residual temperature/heat release) and (ii) the availability of reactive radicals (e.g., OH and H) that facilitate flame stabilization, which together determine ignition/extinction limits and NO_x formation. Specifically, during the load-increase path, limited initial heat feedback and a smaller radical pool require a higher equivalence ratio to achieve stable conical flame anchoring (Φ ≈ 0.194). Along the load-decrease path, residual heat (>1100 K) and accumulated radicals sustain combustion at leaner conditions (Φ ≈ 0.116), while extended exposure to high-temperature regions promotes thermal NO_x formation via the extended Zeldovich mechanism, resulting in 7–12% higher NO_x along the load-decrease path at the same flame temperature.

To mitigate this, adjusting the fuel split (reducing pilot fuel by >25% while increasing premixed fuel) disrupts radical-rich pockets, lowering local temperatures by >150 K and reducing NO_x by 15–22%. Notably, the system maintained stable and favorable performance during load variation tests. NO_x < 20 ppm@15% O₂ (900–1400 K) and SPL < 120 dB (with <3 dB abrupt variations). This synergy stems from the CGSM nozzle’s micromixing design: microtube-induced turbulence ensures a uniform equivalence ratio distribution (eliminating hot spots), while spiral flames attenuate heat release fluctuations—enabling concurrent control of NO_x and combustion noise under the tested control strategy.

It should be clarified that the reported wide load operability refers to stable combustion and emissions performance within the tested 80 kW micro-gas turbine platform. Representative operating points at approximately 45%, 60%, 78%, 90%, and 100% of the rated load were experimentally investigated. The term “load” refers to the power generation capacity of the engine, not the heat release rate of the combustor.

4.2.3. NO_x Emission Performance Under Base Load

Figure 13 illustrates the effect of the main-stage fuel ratio on NO_x emissions (corrected to 15% O₂) across different air preheating temperatures (210 °C, 290 °C, 410 °C), encompassing both simple- and reheat-cycle operating conditions. Here, the abscissa denotes the main-stage fuel ratio (%), representing the fraction of total fuel that is allocated to the main combustion stage, while the ordinate indicates the NO_x concentration (ppm@15% O₂)—a core metric for pollution control. Distinct markers and curves correspond to air preheating temperatures (210 °C: solid squares; 290 °C: solid circles; 410 °C: solid triangles), reflecting the thermal conditions of different cycle configurations.

Across all preheating temperatures, NO_x emissions exhibit a non-linear trend with increasing main-stage fuel ratios: emissions decrease gradually to a minimum as the main-stage fuel ratio rises from 50% to 70% and then increase thereafter. This behavior stems from the synergistic influence of the main-stage fuel ratio on the premixing uniformity and combustion temperature: at low main-stage ratios, fuel is concentrated in the pilot stage, forming local fuel-rich zones that elevate temperatures and promote NO_x formation; as the main-stage ratio increases, the fraction of premixed fuel rises, improving mixing uniformity and reducing NO_x emissions. However, excessively high main-stage ratios intensify heat release in the main combustion zone, raising flame temperatures and re-enhancing thermal NO_x production (via the Zeldovich mechanism). Concurrently, higher air preheating temperatures (from 210 °C to 410 °C) elevate both the minimum NO_x emissions (e.g., ~10 ppm at 210 °C vs. ~25 ppm at 410 °C) and the overall emission level, while narrowing the low-pollution operating range. This is attributed to elevated preheating temperatures increasing the initial premixed gas temperature, which lowers the threshold for thermal NO_x formation and enhances the emission sensitivity to fuel distribution. Collectively, these results suggest that a main-stage fuel ratio of around 75% yielded the minimum NO_x levels under the tested conditions, indicating the importance of precise fuel split control—particularly for reheat-cycle (high-preheat) conditions—and highlighting the coupled influence of fuel distribution and preheating temperature on low-emission operation.

5. Conclusions

This study investigated a Center-Graded Spiral Micromixing (CGSM) combustor concept, designed to address key challenges in hydrogen-fueled gas turbines, including flashback resistance, low NO_x emissions, and fuel flexibility. Based on experimental validation, with qualitative numerical analysis as interpretive support for the internal flow and mixing mechanisms, the following conclusions can be drawn:

• Spiral microtube micromixing strategy was demonstrated to effectively enhance fuel–air mixing through Dean-vortex-induced secondary flows. This airflow-driven mechanism reduces the dependence of mixing uniformity on the fuel jet momentum, contributing to improved robustness under varying operating conditions.
• Pressurized combustor experiments at a moderate pressure (0.3 MPa) confirmed stable combustion of pure hydrogen under lean conditions, without observable flashback or combustion instabilities under the designated operating conditions. Low NO_x emissions were achieved across the investigated operating range, indicating effective suppression of localized high-temperature regions.
• The CGSM combustor exhibited strong fuel adaptability, maintaining stable flame structures and acceptable emission levels during operation with blended hydrogen–methane and hydrogen–ammonia fuels. Seamless online fuel switching without structural modification of the combustor was demonstrated.
• Application tests on an 80 kW micro-gas turbine verified that the combustor can support stable operation across a wide range of load conditions, both in simple- and reheat-cycle modes, with combustion behavior that is consistent with the established operational characteristics of micro-gas turbines.

Overall, the present work validates the feasibility of the CGSM combustor concept at the laboratory and micro-gas turbine scales under moderate-pressure conditions. Further studies at higher pressures and with extended fuel compositions are required to assess its scalability and applicability to large industrial gas turbines.