Hydrogen–Methane Blending in Gas Turbine Combustion Chambers: NOx and CO Emissions, Flame Stabilization, and Thermodynamic Integration with Combined-Cycle Power Plants

Abstract

The global push for low-carbon electricity generation has made hydrogen-enriched natural gas an attractive near-term decarbonization option. This paper combines experimental and thermodynamic analyses of H₂–CH₄ combustion in gas turbine combustion chambers. Experiments were conducted on a patented two-stage swirl burner across 240 operating conditions. The effects of hydrogen fraction (γ = 0–40%), swirler vane angle (30°, 45°, 60°), equivalence ratio (φ = 0.17–1.00), and fuel injection strategy were measured against NOx and CO emissions and lean blowout stability. Each 10% increase in hydrogen content raised NOx by 23–24% via the Zel’dovich thermal mechanism, while CO fell by up to 28.5% at φ = 0.3 and 60° due to enhanced OH-radical activity. The minimum recorded NOx was 12.08 ppm (Type 2 injection, 30°, γ = 0%, φ = 0.3). Hydrogen addition improved lean blowout stability by 32–46% per 10% H₂. A parallel thermodynamic analysis showed that integrating an organic Rankine cycle (ORC) and supplementary H₂–CH₄ firing in the heat recovery steam generator cuts specific CO₂ emissions by 7.5–10% and raises net efficiency by 0.79–4.0 percentage points. Critical comparison with 28 published studies identified an optimal operating window: γ = 20–30%, φ = 0.5–0.7, 45° vane angle (SW = 0.8).

1. Introduction

Reducing CO₂ from electricity generation is urgent and technically difficult. The energy sector accounts for roughly 40% of global CO₂ emissions [1], and gas-fired power plants make up a large fraction of that total. Hydrogen is gaining attention as a clean fuel: it produces only water when burned, its lower heating value (LHV = 120 MJ/kg) is about 2.4 times that of methane, and it can, in principle, flow through existing natural-gas infrastructure [2,3]. Blending hydrogen into natural gas is, therefore, a practical bridge strategy—one that cuts carbon without requiring entirely new plant designs.

Kazakhstan makes this problem concrete. Thermal power stations supply about 66% of the country’s installed capacity, with steam plants accounting for 78% and combined-cycle or gas-turbine plants for 21% [4]. Kazakhstan is a major gas producer and has adopted a Carbon Neutrality Strategy to 2060 (Presidential Decree No. 121, 2 February 2023), which designates natural gas as a transition fuel and hydrogen as a long-term target for hard-to-electrify sectors [5]. That policy framing creates a direct technical need: understanding what happens when you blend hydrogen into the combustion chambers of existing gas turbines.

Burning H₂–CH₄ mixtures is not simply burning methane with a hydrogen additive. Hydrogen’s laminar flame speed (S₁ ≈ 2.9 m/s at stoichiometry) is nearly an order of magnitude higher than methane’s (S₁ ≈ 0.38 m/s) [6,7]. That difference matters in combustion chamber design in two opposing ways. On one hand, hydrogen enrichment extends the lean blowout limit and permits operation at lower equivalence ratios, which helps reduce thermal NOx. On the other hand, higher hydrogen fractions raise the adiabatic flame temperature, increase flashback risk, can trigger thermoacoustic instability, and substantially raise NOx emissions [8,9].

NOx formation in gas turbine combustion chambers follows four main pathways: (1) the thermal Zel’dovich mechanism, dominant above 1800 K; (2) the prompt Fenimore mechanism in fuel-rich zones via CH-radical reactions; (3) the N₂O-intermediate pathway at moderate temperatures in lean systems; (4) the NNH mechanism, whose contribution grows with hydrogen fraction [10,11]. The relative weight of these pathways shifts significantly as γ increases, complicating low-emission burner design for hydrogen-blended fuels.

Swirl-stabilized combustion is the standard approach in modern gas turbine combustors. The swirl number (SW)—the ratio of tangential to axial momentum flux—controls recirculation zone size and structure. SW > 0.6 is generally required for a stable recirculation zone, and peak NOx concentrations tend to occur at SW = 0.75–1.4 [12,13]. The interaction between swirl intensity, hydrogen fraction, and fuel injection strategy across the full range of gas turbine operating conditions has not been systematically studied.

Combined-cycle power plants (CCPPs) pair a gas turbine topping cycle with a steam Rankine bottoming cycle and reach net efficiencies up to 64% in state-of-the-art configurations [14]. Adding an organic Rankine cycle (ORC) as a second bottoming stage—a so-called trinary cycle—can extract another 2.2 percentage points [15]. Supplementary firing in the heat recovery steam generator (HRSG) adds operational flexibility and peaking capacity but interacts with cycle thermodynamics in ways that depend on the fuel chemistry [16]. Substituting natural gas with H₂–CH₄ blends in both the gas turbine and the supplementary burner introduces additional complexity that has not been treated comprehensively in the open literature.

This paper addresses those gaps through four connected steps: (1) experimental characterization of LPG-H₂ combustion (LPG as a functionally equivalent methane surrogate) in a two-stage swirl burner across 240 test conditions; (2) thermodynamic analysis of CCPP configurations integrating ORC and supplementary firing; (3) a critical comparison of results against 28 published studies; (4) engineering recommendations for optimal hydrogen blending in gas turbines.

The specific contributions of this work are as follows: (i) a patented two-stage swirl burner with independently adjustable inlet and outlet vanes, which separates mixing intensity control from recirculation zone geometry; (ii) systematic comparison of two premixed H₂–CH₄ injection strategies; (iii) integration of burner-level emission data with system-level CCPP thermodynamics; (iv) a first cross-study synthesis spanning three sequential investigations [17,18] that cover diffusion combustion, Type 1 premixed injection, and Type 2 premixed injection on a common methodological basis.

2. Published Research: Patterns and Gaps

2.1. Hydrogen Enrichment: Combustion Effects

The literature on H₂–CH₄ combustion shows consistent trends across a wide range of experimental configurations, alongside substantial scatter that traces back to differences in burner geometry, pressure, temperature, and mixing strategy.

Table 1. Summary of published studies on hydrogen-containing fuel combustion in gas turbine applications.

Source	Fuel	γ, %	Burner Type	Key Result	NOx	CO	Stability
Dostiyarov et al. [13] 2024	H2 + LPG	0–40	Swirl, diffusion	φLBO = 0.9 at 60°, γ = 40%	↑ with γ, SW	↓ with γ	↑ with γ
Dostiyarov et al. [13] 2024	H2 + LPG	0–40	Swirl, premixed	Min NOx = 12.08 ppm	↑ 24%/10% H2	↓ 28.5%	↑ 46%/10%
Kindra et al. [18] 2025	CH4 + H2	—	CCPP/HRSG	ORC: Δη = +0.79%	—	CO2 ↑ 10%	—
Emre et al. [19] 2024	NG + H2	0–50	Swirl, premixed	H2 reduces flame size	↑ thermal	↓	↑
Bertsch et al. [8] 2024	H2/air	100	Swirl, premixed	Two stabilization regimes	Very high	N/A	Complex
Ji et al. [9] 2024	CH4 + H2	0–60	Swirl, premixed	Secondary H2 damps thermoacoustics	↑ NOx	↓ CO	↑
Porcarelli [10] 2024	H2/air	100	Flat flame	High strain suppresses NOx	↓ with strain	—	—
Howarth et al. [11] 2024	H2/air	0–100	DNS, premixed	EGR improves diffusion effects	Moderate	—	↑
Nemitallah et al. [14] 2024	CH4 + H2	0–40	Swirl, premixed	H2 has no LBO effect near φ = 0.5	Moderate ↑	↑ CO/UHC	Neutral
Xu et al. [6] 2024	H2/air	100	Closed volume	Burning rate ↑ with narrower section	High	—	↑
Gaucherand et al. [20] 2023	H2 + NH3	0–100	Premixed	Instabilities depend on φ and pressure	High at φ > 0.8	Low	Complex
Elbaz et al. [21] 2019	LPG	0	Double swirl	Staged combustion cuts NOx by 60%	↓ staged	Moderate	↑
Wang et al. [22] 2008	H2 + LPG	0–30	Premixed, flat	H2 triples burning rate	↑	↓	↑
Guo et al. [23] 2005	CH4 + H2	0–80	Counterflow	Flammability limit expands with H2	↑ above 50%	↓	↑
Park [22] 2021	CH4 + H2	0–60	Premixed, high P	NOx ↑ with P and γ nonlinearly	↑	↓	↑
Ferrarotti et al. [24] 2021	CH4 + H2	0–50	MILD/flameless	Flameless combustion cuts NOx sharply	↓	Moderate	Stable
Aravindan et al. [25] 2024	LPG + H2	0–60	IC engine	Optimum at 30% H2	↓ then ↑	↓	↑
Kindra et al. [16] 2023	NG	—	Trinary CCPP	ORC: Δη up to 2.2%	—	CO2 ↓	—
Ahmadi & Dincer [21] 2011	NG	—	Dual-pressure CCPP	Thermoeconomic optimization	—	—	—
Abdollahian & Ameri [17] 2021	NG	—	CCPP + firing	Suppl. firing +26.3 MW	—	—	—

Notes: ↑ = increase; ↓ = decrease; SW = swirl number; LBO = lean blowout; CCPP = combined-cycle power plant; HRSG = heat recovery steam generator; EGR = exhaust gas recirculation; DNS = direct numerical simulation; MILD = moderate or intense low-oxygen dilution combustion.

summarizes 20 key studies, organized by thematic focus and main findings.

2.2. NOx Formation Mechanisms Under Hydrogen Enrichment

The literature shows a consistent but mechanistically layered relationship between hydrogen enrichment and NOx. At low hydrogen fractions (γ < 20%), the N₂O-intermediate pathway is a meaningful contributor, especially in the lean conditions typical of dry-low-NOx (DLN) combustors operating at φ = 0.5–0.7 [10]. Above γ ≈ 20%, the rising adiabatic flame temperature (from ~2230 K for pure methane to ~2480 K for pure hydrogen at stoichiometry) activates the Zel’dovich thermal mechanism exponentially. NOx production scales roughly as exp(−Ea/RT), with Ea ≈ 316 kJ/mol [11].

Park [26] measured NOx concentrations in premixed CH₄–H₂ flames at elevated pressures up to 10 atm and found a superlinear synergy between pressure and hydrogen fraction: at γ = 40% and P = 10 atm, NOx was roughly 3.5 times the atmospheric value at the same γ. This pressure effect is directly relevant to gas turbines, where combustion chamber pressures typically run 15–45 bar. The atmospheric results of the present study should therefore be treated as conservative lower bounds for real turbine emissions.

Howarth et al. [11] showed that the NNH mechanism—via the reaction NNH + O → NH + NO—contributes an estimated 10–30% of total NOx above γ = 30% in conditions similar to those studied here. This pathway is less temperature-sensitive than the thermal mechanism and cannot be suppressed purely by leaning out the mixture. That partially explains why equivalence ratio control is less effective at reducing NOx in hydrogen-enriched fuels than it is in pure methane [10,11].

Ferrarotti et al. [24] showed through DNS that high strain rates in lean premixed hydrogen flames can suppress NOx by cutting residence time in the hot reaction zone. This is the regime that the 60° vane angle configuration in the present study approaches. But the companion increase in CO at high strain rates creates a fundamental NOx–CO trade-off that combustion chamber designers cannot ignore [8].

High-pressure swirl combustion data for H₂–CH₄ blends have been reported by groups at DLR [24] and KIT [25], and CFD validation for H₂-rich swirl flames is available in Varunkumar et al. [27]. These studies establish NOx ∝ Pⁿ with n = 0.5–0.7 for lean premixed conditions, consistent with Section 3.5. NNH contribution estimates from DNS configurations [10,11] should be treated as upper bounds for swirl geometries: near-vane regions have short residence times (strain-dominated, NNH suppressed), while the recirculation core has 5–10× longer τ (thermal NOx dominant). The interaction term β × γ in Equation (7) (p = 0.003) provides experimental evidence of this nonlinear coupling between swirl intensity and hydrogen fraction.

2.3. Swirl Number Effects and Recirculation Zone Dynamics

The relationship between swirl intensity and combustion performance is well established in literature and acquires a new layer of complexity for hydrogen-containing fuels. Elbaz et al. [21] showed that staged combustion with optimized air distribution in a double-swirl configuration reduced NOx by 60% relative to unstaged operation for pure LPG, while maintaining acceptable flame stability. That result underscores the potential of geometric optimization independent of fuel composition. The specific geometric configuration of the multi-channel swirl burner and the fuel-hydrogen mixing scheme investigated in this work are systematically designed in accordance with the patented burner device parameters (Patent KZ 36843) [28].

This study examines three swirl numbers: SW = 0.4 (30°), SW = 0.8 (45°), and SW = 1.3 (60°). At SW = 0.4, recirculation is weak and confined to the central region, giving limited flame anchoring. The lean blowout limit without hydrogen is φLBO = 0.75–0.81—marginal for gas turbines that need stable operation down to φ ≈ 0.5. At SW = 0.8, a well-developed toroidal recirculation zone forms, and the 30% hydrogen case pushes φLBO down to 0.39–0.47, approaching the turbine operating range. At SW = 1.3, the recirculation zone is the largest studied: φLBO falls furthest, but the combination of elevated flame temperature, long gas residence time, and high O-radical concentration creates near-ideal conditions for all four NOx pathways simultaneously. This explains the 65% rise in NOx from SW = 0.4 to SW = 1.3 observed in the present work, consistent with the SW = 0.75–1.4 NOx maximum reported in the literature [13].

3. Materials and Methods

3.1. Experimental Setup and Burner Design

Experiments were conducted in the Combustion Laboratory of the Department of Thermal Power Engineering at Gumarbek Daukeev Energy University (Almaty, Kazakhstan). The test rig investigates swirl-stabilized combustion of gaseous fuel mixtures at atmospheric pressure, reproducing the aerodynamic conditions of a gas turbine primary combustion zone. The central element is a patented two-stage swirl burner with independently controlled inlet and outlet vane angles [22,23]. A schematic of the complete experimental setup is presented in Figure 1.

The burner operates on co-axial swirling jets: air passes through two vane sets—fixed inlet vanes at 45° that impose initial swirl, and adjustable outlet vanes that set the total swirl intensity. This two-stage arrangement, analogous to the staged swirlers in Rolls-Royce Trent and GE LM6000 combustors, gives more operational flexibility than single-stage swirlers [8,21]. Fuel is introduced through two interchangeable injection assemblies, enabling direct comparison of injection-strategy effects on mixing quality and combustion behavior.

Type 1 (central channel injection): The premixed LPG–H₂ blend is supplied through a co-axial central tube (inner diameter 9 mm) positioned upstream of the inlet vane section. The fuel enters co-axially with the swirling air; premixing occurs over approximately 80 mm; momentum flux ratio J = ρ_f U_f²/(ρ_a U_a²) = 0.08–0.34. Type 2 (vane-base injection): Fuel is injected through 12 radial holes (diameter 1.5 mm) at the base of the outlet vane leading edges, 10 mm upstream of the vane throat; mixing length ≈ 20 mm; J = 0.24–1.08. Type 2 achieves finer-scale premixing at the maximum-shear point, producing a more spatially uniform equivalence ratio at the combustion zone inlet and suppressing thermal NOx hot spots. See Figure 1 for annotated cross-sectional schematics of both configurations.

LPG (50% propane C₃H₈ + 50% butane C₄H₁₀ by volume) was used as the base fuel in place of pure methane for practical safety reasons. LPG and methane are both paraffinic hydrocarbons with similar LHV (LPG: 46.4 MJ/kg; CH₄: 50.0 MJ/kg), adiabatic flame temperature at stoichiometry (~2260 K and ~2230 K, respectively), and laminar flame speed (within ~20%), so the results transfer directly to natural gas. Hydrogen had 99.96% purity, supplied from 40 L cylinders with needle-valve flow control. However, C₃ and C₄ hydrocarbons produce higher CH radical concentrations than CH₄, enhancing the Fenimore prompt NOx pathway. Based on PSR calculations (GRI-Mech 3.0 for CH₄; San Diego mechanism for C₃H₈), prompt NOx contributes 3–8% of total NOx for CH₄ at φ = 0.5–0.7 versus 5–11% for LPG. The resulting systematic overestimate in absolute NOx is approximately 5–12% relative to an equivalent CH₄ campaign at the same conditions, consistent with the ±13% agreement with published CH₄-based studies Relative trends with γ, φ, and SW are governed by the thermal Zel’dovich mechanism, which has identical sensitivity for LPG and CH₄ at φ ≥ 0.5. Absolute NOx values should be reduced by 5–12% for direct CH₄ comparison. The detailed constructive and geometric parameters of the investigated experimental swirl burner are systematically summarized in

Table 2. Geometric parameters of the experimental swirl burner.

Parameter	Unit	Value
Overall burner length	mm	350
Inlet channel diameter	mm	50
Inner diameter at outlet section	mm	100
Outlet diameter	mm	200
Number of inlet vanes	—	14
Inlet vane height	mm	20
Fixed inlet vane angle	°	45
Number of outlet vanes	—	12
Outlet vane height	mm	40
Adjustable outlet vane angle	°	30, 45, 60
LPG injection holes (Type 1), N/diam.	—	6/2 mm
H2 injection holes, N/diam.	—	12/1.5 mm
H2 supply tube diameter	mm	9
Distance from H2 holes to vanes	mm	10

.

3.2. Instrumentation and Measurement Procedure

Flue gas composition was measured with a Testo 350 analyzer (Testo SE & Co. KGaA, Titisee-Neustadt, Germany) at an axial distance of 5D = 1000 mm from the burner face, ensuring post-flame chemical reactions were complete. Measurements were taken at five radial positions—four evenly spaced points along two orthogonal diameters plus one central point—to obtain spatially representative averages. This scheme was validated against a 13-point grid in preliminary tests; the deviation in averaged values was under 3%. Each operating condition was repeated three times after the system reached steady state (flue gas composition stable within % for 3 min). The detailed specifications of the instruments and the results of the total rig uncertainty analysis are summarized in

Table 3. Instrumentation and uncertainty analysis.

Instrument	Model/Type	Parameter	Range/Accuracy	Rel. Error, %	Std. Dev.
Combustion analyzer	Testo 350	NOx (NO + NO2)	0–2000 ppm/±2 ppm	5.0	±0.2 ppm
Combustion analyzer	Testo 350	CO	0–10,000 ppm/±10 ppm	1.0	±19.34 ppm
Vortex flowmeter	KROHNE OPTISWIRL (KROHNE Messtechnik GmbH, Duisburg, Germany)	Air flow rate	0.19–1.19 m3/s/±1.5%	1.5	—
Rotameter LPG	RMB-A-0.6 (Pribor Ltd., Almaty, Kazakhstan)	LPG flow rate	0–0.020 m3/s/±2%	2.0	—
Rotameter H2	RMB-A-0.1 (Pribor Ltd., Almaty, Kazakhstan)	H2 flow rate	0–0.005 m3/s/±1.5%	1.5	—
Thermocouple (K-type)	K-type thermocouple (Chromel-Copel, Termopribor JSC, Klin, Russia)	Temperature	−40 to +375 °C/±1.5 °C	0.4	—
Laboratory thermometer	TLS-2	Ambient temperature	0–50 °C/±1.0 °C	0.16	—

Total rig uncertainty: σm = √(σthermal² + σTC² + σNOx² + σCO²) = 5.12%.

.

3.3. Experimental Conditions and Test Matrix

A full-factorial design covered all combinations of the independent variables in

Table 4. Test matrix (N = 240 operating points).

Variable	Symbol	Range	Levels	Points	Total
Hydrogen volume fraction	γ	0–40%	0; 10; 20; 30; 40%	5	—
Outlet vane angle	β	30–60°	30°; 45°; 60°	3	—
Equivalence ratio	φ	0.17–1.00	0.17; 0.3; 0.5; 0.7; 0.85; 1.0	6	—
Fuel injection type	Type	1 or 2	Type 1; Type 2	2	—
Reynolds number	Re	50–350 k	50 k; 150 k; 250 k; 350 k	4	240

. Equivalence ratio varied by adjusting air flow at constant fuel flow, reproducing gas turbine part-load operation. For lean blowout tests, the burner was brought to steady-state combustion, then fuel flow was reduced in 0.1 kg/h steps until extinction, and φLBO was recorded.

3.4. Governing Equations

Hydrogen volume fraction γ (Equation (1)):

$$\mathit{gamma}={\displaystyle \frac{V_2}{(V_2+V_{LPG})}}$$

(1)

Equivalence ratio φ (Equation (2)):

$$phi={\displaystyle \frac{\frac{m_{fuel}}{mair}}{(\frac{m_{fuel}}{mair})stoich}}$$

(2)

Swirl number SW by the Beer–Chigier (Equation (3)):

$$\mathit{SW}={\displaystyle \frac{\frac{2}{3}\times [1-{\left(\frac{Rh}{R}\right)}^3]}{[1-{\left(\frac{Rh}{R}\right)}^2]\times tanbeta}}$$

(3)

where Rh is the hub radius, R is the outer radius, and beta is the outlet vane angle. At 30°, 45°, and 60°, SW = 0.4, 0.8, and 1.3, respectively.

Reynolds number Re (Equation (4)):

$$\mathit{Re}={\displaystyle \frac{(\omega \times d)}{\nu }}$$

(4)

where ω is the mean axial velocity at the burner inlet (m/s), d = 0.05 m, and ν = 15.06 × 10⁻⁶ m²/s at 20 °C.

Adiabatic flame temperature of the blend, $T_{\mathrm{a}\mathrm{d},\mathrm{m}\mathrm{i}\mathrm{x}}$ estimated by linear weighting (Equation (5)):

$$T_{\mathrm{a}\mathrm{d},\mathrm{m}\mathrm{i}\mathrm{x}}=\mathsf{\gamma }\times T_{ad,H_2}+(1-\gamma )\times T_{ad,LPG}$$

(5)

Validation of Equation (5): for γ < 30%, linear mole-fraction weighting gives T_ad within 0.6% of NASA CEA software (v2.0; National Aeronautics and Space Administration, Cleveland, OH, USA) and is retained for simplicity. For γ ≥ 30%, the deviation reaches 52–68 K (2.5%); CEA-computed T_ad values are therefore used directly in all calculations for γ = 30% and 40%. CEA values for all tested hydrogen fractions are provided in Supplementary Table S1.

3.5. Pressure Scaling of NOx Results

All combustion experiments were conducted at atmospheric pressure (P₀ = 101.3 kPa). Gas turbine combustion chambers operate at 15–45 bar, where NOx formation rates are substantially higher. Three physical effects drive this increase: the adiabatic flame temperature rises by 30–80 K per pressure decade at constant φ, exponentially accelerating the thermal Zel’dovich mechanism; molar reactant concentrations increase proportionally with pressure, directly speeding the O + N₂ chain; the effective post-flame residence time above 1800 K increases in practical combustors [7] (p. 499), [29]. These effects were first characterized systematically for gas turbine primary zones by Tumanovsky [29] and consolidated by Lefebvre [7] (p. 499) into the empirical power–law relation:

NOx(P) = NOx(P₀) × (P/P₀)ⁿ

(6)

where n = 0.5–0.7 for lean premixed H₂–CH₄ flames at φ = 0.5–0.7, as established by Lefebvre [7] (p. 499) and confirmed for H₂–CH₄ blends by Park [26]. Applying Equation (6) to the global minimum NOx in this study (12.1 ppm at 1 bar, Type 2, 30°, γ = 0%, φ = 0.3) gives 47–75 ppm at 15 bar and 66–118 ppm at 30 bars. For the recommended window (Type 2, 45°, γ = 30%, φ = 0.7: 29.7 ppm at 1 bar), scaled values are 114–184 ppm at 15 bars. Both cases require selective catalytic reduction (SCR) for EU IED compliance (<25 ppm at 15% O₂). Scaled estimates for all configurations at 5, 10, 15, 20, and 30 bars are in Supplementary Table S1.

The direction and relative magnitude of the NOx trends with γ, φ, and SW were validated against high-pressure H₂–CH₄ data of Park [26] (up to 10 atm), confirming that the ranking of configurations and the recommended operating window are preserved at turbine pressures. All NOx values in this manuscript are atmospheric-pressure measurements and represent lower bounds for real gas turbine operating conditions.

4. Results and Discussion

4.1. Flame Stabilization

4.1.1. Effect of Hydrogen Fraction and Vane Angle on Lean Blowout

Table 5 gives the full lean blowout dataset. The results confirm that hydrogen enrichment and swirl intensity both extend the stable operating range, and that the injection strategy has a non-trivial effect that interacts with swirl aerodynamics.

Table 5. Lean blowout equivalence ratio (φLBO) as a function of hydrogen fraction, vane angle, and injection type. Lower values indicate better stability.

Injection Type	Angle, °	SW	γ = 0%	γ = 10%	γ = 20%	γ = 30%	γ = 40%
Type 1	30°	0.4	0.75	0.72	0.67	0.62	0.57
Type 1	45°	0.8	0.52	0.47	0.44	0.41	0.36
Type 1	60°	1.3	0.41	0.35	0.28	0.23	0.18
Type 2	30°	0.4	0.81	0.77	0.71	0.65	0.58
Type 2	45°	0.8	0.58	0.51	0.46	0.43	0.37
Type 2	60°	1.3	0.49	0.40	0.32	0.26	0.20

Mean values of three repeat measurements at v = 10 m/s. Standard deviation < 0.02 for all conditions. The experimental curves of the lean blowout equivalence ratio (phi_{LBO}) as a function of hydrogen volume fraction gamma for all six tested configurations are presented in Figure 2.

Table 5. Lean blowout equivalence ratio (φLBO) as a function of hydrogen fraction, vane angle, and injection type. Lower values indicate better stability.

Injection Type	Angle, °	SW	γ = 0%	γ = 10%	γ = 20%	γ = 30%	γ = 40%
Type 1	30°	0.4	0.75	0.72	0.67	0.62	0.57
Type 1	45°	0.8	0.52	0.47	0.44	0.41	0.36
Type 1	60°	1.3	0.41	0.35	0.28	0.23	0.18
Type 2	30°	0.4	0.81	0.77	0.71	0.65	0.58
Type 2	45°	0.8	0.58	0.51	0.46	0.43	0.37
Type 2	60°	1.3	0.49	0.40	0.32	0.26	0.20

Mean values of three repeat measurements at v = 10 m/s. Standard deviation < 0.02 for all conditions. The experimental curves of the lean blowout equivalence ratio (phi_{LBO}) as a function of hydrogen volume fraction gamma for all six tested configurations are presented in Figure 2.

Figure 2. Lean blowout equivalence ratio (φLBO) vs. hydrogen volume fraction γ for all six configurations tested. Solid lines: Type 1 injection; dashed lines: Type 2 injection. The green shaded region indicates the gas turbine operating range (φLBO ≤ 0.5).

Figure 2. Lean blowout equivalence ratio (φLBO) vs. hydrogen volume fraction γ for all six configurations tested. Solid lines: Type 1 injection; dashed lines: Type 2 injection. The green shaded region indicates the gas turbine operating range (φLBO ≤ 0.5).

Several patterns stand out. The absolute stability gain from γ = 0% to γ = 40% is larger at high swirl numbers: ΔφLBO = 0.18–0.23 at SW = 0.4 versus 0.23–0.29 at SW = 1.3. Hydrogen enrichment, therefore, delivers proportionally more benefit when a well-developed recirculation zone is present to capture the reactive mixture. The stability improvement rate per 10% H₂ is not constant across vane angles: roughly 24%/10% at 30°, ~32%/10% at 45°, and ~46%/10% at 60°—indicating a nonlinear interaction between swirl intensity and hydrogen reactivity in the recirculation zone dynamics.

From an operational standpoint, the critical threshold for gas turbine part-load operation is φLBO ≤ 0.5. Without hydrogen, this is only met at SW = 1.3 with Type 1 injection (φLBO = 0.41). Adding 10% H₂ opens up SW ≥ 0.8 at Type 1 (φLBO = 0.47). At γ = 30%, SW ≥ 0.8 and Type 1 give φLBO = 0.41 at 45°. Hydrogen blending meaningfully expands the combustion chamber’s stable operating window.

4.1.2. Two-Stage Lean Blowout Mechanism

Close observation of the blowout process revealed a two-stage behavior consistent with descriptions in the literature [13,21]. In the first stage, at φ slightly above φLBO, the flame loses visible brightness while CO rises as combustion temperature falls below the CO oxidation threshold. In the second stage—appearing abruptly within Δφ ≈ 0.02–0.05 below the first-stage transition—thermal blowout occurs: heat release falls below heat loss rate.

Hydrogen enrichment shifts the first-stage onset to lower φ, because H₂’s wide flammability limits (4–75% in air versus 5–15% for CH₄) allow part of the non-uniform recirculation zone to sustain combustion even at very lean global equivalence ratios. The second-stage transition sharpens with increasing γ: higher adiabatic flame temperature creates a thermally more sensitive system, where small φ perturbations either sustain combustion or trigger immediate extinction. The critical blowout window narrows from Δφ ≈ 0.05 at γ = 0% to ≈ 0.02 at γ = 40%.

4.2. NOx Emissions

Quantitative NOx Dependencies

Table 6. NOx concentration (ppm) at φ = 0.7 as a function of hydrogen fraction, vane angle, and injection type.

Injection Type	Angle, °	SW	γ = 0%	γ = 10%	γ = 20%	γ = 30%	γ = 40%
Type 1	30°	0.4	14.2	18.1	22.6	28.2	35.0
Type 1	45°	0.8	17.5	22.4	27.8	34.6	43.0
Type 1	60°	1.3	28.4	36.2	46.0	58.5	72.0
Type 2	30°	0.4	12.1	15.5	19.4	24.1	30.0
Type 2	45°	0.8	15.2	19.2	24.0	29.7	37.0
Type 2	60°	1.3	25.0	31.5	40.2	51.0	64.0

Mean values of three measurements; standard deviation < 1.8 ppm. Minimum NOx = 12.08 ppm at Type 2, 30°, γ = 0%, φ = 0.3 [13].

gives the full NOx dataset at φ = 0.7, a representative gas turbine lean-premixed operating condition. Multiple linear regression across all three independent variables (vane angle, hydrogen fraction, injection type) gives Equation (7).

NOx [ppm] = 12.8 + 0.56 × β [°] + 0.42 × γ [%] − 2.8 × Type

(7)

where Type = 1 for Type 1 injection and Type = 0 for Type 2 (R² = 0.97).

The regression coefficients are revealing. The vane angle term (0.56 ppm/°) is numerically larger than the hydrogen fraction term (0.42 ppm/%H₂). This challenges the common industrial assumption that reducing hydrogen content is the most direct lever for NOx control. Geometric optimization of the combustion chambers, specifically swirl intensity, can be equally or more effective. Type 2 injections consistently produced about 11% lower NOx than Type 1 across all conditions, which is attributable to better premixing and a more uniform flame root temperature distribution. The systematic behavior of NOx concentration (ppm) versus equivalence ratio at a fixed hydrogen fraction of gamma = 30% is detailed in

Table 7. NOx concentration (ppm) versus equivalence ratio at γ = 30% H₂.

Configuration	φ = 0.3	φ = 0.5	φ = 0.7	φ = 0.85	φ = 1.0	ΔNOx (0.3 → 1.0)	Trend
Type 1, 30°, SW = 0.4	12.5	19.4	28.2	35.1	42.8	+342%	Nonlin. ↑
Type 1, 45°, SW = 0.8	15.8	23.7	34.6	43.2	52.0	+329%	Nonlin. ↑
Type 1, 60°, SW = 1.3	26.0	38.4	58.5	74.2	89.0	+342%	Nonlin. ↑
Type 2, 30°, SW = 0.4	10.5	16.2	24.1	29.8	36.0	+343%	Nonlin. ↑
Type 2, 45°, SW = 0.8	13.0	19.8	29.7	37.0	44.5	+342%	Nonlin. ↑
Type 2, 60°, SW = 1.3	22.8	33.2	51.0	64.5	78.0	+342%	Nonlin. ↑

Notes: ↑ = increase.

.

The near-constant relative NOx increases from φ = 0.3 to φ = 1.0 (~340% across all configurations) points to a single governing mechanism: the rise in adiabatic flame temperature with equivalence ratio. This finding matters practically. Effective NOx management in hydrogen-blended gas turbines must rely primarily on geometric design and injection strategy, not on equivalence ratio adjustment, which is constrained by turbine inlet temperature requirements.

4.3. CO Emissions

Table 8. CO concentration (ppm) at φ = 0.7 as a function of hydrogen fraction, vane angle, and injection type.

Injection Type	Angle, °	SW	γ = 0%	γ = 10%	γ = 20%	γ = 30%	γ = 40%
Type 1	30°	0.4	285	252	220	197	178
Type 1	45°	0.8	248	217	190	168	151
Type 1	60°	1.3	198	173	152	134	120
Type 2	30°	0.4	312	278	245	218	195
Type 2	45°	0.8	270	240	210	187	167
Type 2	60°	1.3	228	200	175	154	137

EU Industrial Emissions Directive (IED) limit: ~90 ppm at 15% O₂. All φ = 0.7 values exceed this limit. At φ = 0.3 with γ ≥ 30%, values of 101–130 ppm approach the IED limit. Minimum measured CO: 101 ppm [13].

presents CO data at φ = 0.7. The dependence on equivalence ratio and hydrogen fraction is visualized in Figure 3.

CO drops roughly 37–40% as γ rises from 0% to 40%, driven by elevated OH-radical concentration and higher combustion temperature. The reduction is slightly more pronounced at higher swirl numbers, reflecting better CO oxidation conditions in the larger recirculation zone. The reduction rate (~8–10%/10% H₂ at φ = 0.7, or 10–13% at φ = 0.3) suggests that the hydrogen fraction optimal for CO minimization in lean gas turbine combustors may be lower than in rich-burn systems.

Note on CO measurement uncertainty: the Testo 350 CO module has a stated standard deviation of ±19.34 ppm. At φ = 0.3, where measured CO values are 90–135 ppm, this corresponds to a relative uncertainty of 14–22%. All NOx–CO trade-off analyses, the NOx–CO Pareto frontier, and the combustion efficiency calculations are therefore based on φ ≥ 0.5 data, where CO values are 150–800 ppm and CO relative uncertainty is below 10%. Data points at φ = 0.3 with CO uncertainty above 15% are marked with (*) in Table 8 and

Table 9. NOx–CO performance map at φ = 0.5. EU regulatory assessment: NOx ≤ 25 ppm and CO ≤ 150 ppm.

Configuration	γ = 0% NOx/CO	γ = 10% NOx/CO	γ = 20% NOx/CO	γ = 30% NOx/CO	Compliance Assessment
T1, 30°	16.1/290	20.5/256	25.6/222	31.8/199	NOx exceeds at γ ≥ 20%
T1, 45°	20.2/252	25.8/219	32.0/191	39.7/171	CO above limit throughout
T1, 60°	32.6/202	41.5/176	52.8/154	67.0/137	NOx above; CO marginal at γ = 30–40%
T2, 30°	13.8/318	17.5/282	21.9/248	27.1/221	NOx marginal at γ = 20%; CO above
T2, 45°	17.4/276	22.0/244	27.4/213	33.8/191	NOx exceeds at γ ≥ 20%
T2, 60°	28.8/234	36.0/205	46.0/180	58.4/162	NOx above throughout

Simultaneous compliance with both NOx ≤ 25 ppm AND CO ≤ 150 ppm is not achievable within the studied parameter space without additional abatement. At φ = 0.3 with γ ≥ 30%, CO drops below 150 ppm, but lean blowout stability becomes critical at low SW.

and Figure 4.

4.4. Thermodynamic Integration with Combined-Cycle Power Plants

4.4.1. CCPP Configuration and Simulation Methodology

Section 4.1, Section 4.2 and Section 4.3 present direct experimental measurements from the atmospheric-pressure swirl burner test rig. Section 4.4 presents results from a steady-state thermodynamic simulation of combined-cycle power plant configurations using the GTE-170 gas turbine model (Section 3.4). No combined-cycle power plant experiment was conducted in this study. All efficiency values, specific CO₂ figures, and net output estimates in Section 4.4 and Section 5.2 are model predictions. Experimental validation of the full CCPP system is beyond the scope of this work and is identified as a research priority in Section 5.2.

The thermodynamic analysis used a GTE-170 gas turbine as the topping cycle generator (

Table 10. GTE-170 technical parameters used in the thermodynamic model.

Parameter	Unit	Value
Net electrical output	MW	155.3
Net efficiency	%	34.1
Exhaust gas temperature	°C	538
Exhaust gas mass flow rate	kg/s	509
Fuel LHV (CH4)	MJ/kg	50.03
Fuel HHV (CH4)	MJ/kg	55.515
Mechanical transmission efficiency	%	99
Generator efficiency	%	99
Fuel compressor power	kW	850

). Mass and energy balance equations govern each cycle component; thermophysical properties of combustion products, water/steam, and organic working fluids were taken from the NIST REFPROP DATABASE (V10.0; NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY, GAITHERSBURG, MD, USA). The model was validated against published data for the GTE-160 and GTE-110 units; maximum deviation in net efficiency was 0.5%, confirming model applicability [18].

4.4.2. Binary and Trinary Cycle Efficiency Comparison

Net CCPP efficiency with ORC and supplementary firing (Equation (8)):

η_net = (NGTU + NSTU + NORC)/[(1 + β) × BCC × QLHV]

(8)

where NGTU, NSTU, and NORC are net outputs of the gas turbine, steam turbine, and ORC units, respectively; β is the supplementary firing ratio; BCC is the specific primary fuel consumption; QLHV is the lower heating value. The comprehensive thermodynamic performance comparison of the investigated CCPP configurations at beta = 0 and T_0 = 515 °C is summarized in

Table 11. Thermodynamic performance comparison of CCPP configurations at β = 0, T₀ = 515 °C.

CCPP Configuration	Net Output, MW	Net Eff., %	Δη vs. Base, pp	CO2, kg/kWh	CO2 Reduction	Rating
Single-pressure HRSG (baseline)	231.8	48.57	0.00	0.430	0.0%	Baseline
Dual-pressure HRSG	241.5	50.78	+2.21	0.408	−5.1%	Good
Single-pressure HRSG + RH	246.0	50.94	+2.37	0.405	−5.8%	Good
Dual-pressure HRSG + RH	253.2	52.77	+4.20	0.384	−10.7%	Very Good
Trinary cycle (ORC, no RH)	249.1	51.57	+3.00	0.397	−7.7%	Very Good
Trinary cycle (ORC + RH)	261.4	52.57	+4.00	0.385	−10.5%	Excellent

pp = percentage points; RH = reheat; ORC working fluid: R236ea; calculations based on GTE-170; net efficiency per Equation (8) [18].

.

The trinary cycle with ORC and reheat outperforms the dual-pressure reheat configuration by 1.9 percentage points in net efficiency at comparable CO₂ reduction. Two complementary effects drive this. First, the ORC recovers residual low-grade heat in the 100–200 °C temperature range with higher thermodynamic effectiveness than a low-pressure steam circuit, because organic working fluids have lower critical temperatures. Second, the improved regeneration in the steam section that the ORC enables reduces condenser load [18].

4.4.3. Effect of Supplementary Firing Ratio on CCPP Performance

The data shows four distinct thermodynamic response types. Type 1 (single pressure, no reheat): Monotone efficiency declines with β, because supplementary firing introduces heat at a lower effective temperature than the gas turbine combustion chamber. Type 2 (single pressure + reheat): Non-monotone, with a peak near β = 0.37, from competition between falling high-pressure evaporator effectiveness (negative) and improved reheat expansion conditions (positive). Type 3 (dual-pressure, no reheat): Monotone decline, faster than Type 1. Type 4 (trinary ORC + reheat): Slight efficiency rise at small β, then decline, because the ORC recovers heat that would otherwise be unavailable at higher exhaust temperatures. The calculated net efficiency (%) and net output (MW) data of the selected CCPP configurations versus the supplementary firing ratio beta are presented in Table 12.

Table 12. Net efficiency (%) and net output (MW) of selected CCPP configurations versus supplementary firing ratio β at T₀ = 515 °C.

Configuration	β = 0	β = 0.1	β = 0.2	β = 0.3	β = 0.4	β = 0.5	Trend
Single press., η/%	48.57	48.32	48.05	47.75	47.40	46.90	Monotone ↓
Single press., N/MW	231.8	243.1	254.5	265.7	276.2	285.5	↑ +23%
Single press. + RH, η/%	50.94	50.71	50.94	51.22	51.50	51.12	Non-monotone
Dual press., η/%	50.78	50.44	50.12	49.78	49.34	48.82	Monotone ↓
Trinary ORC, η/%	51.57	51.31	51.08	50.84	50.54	50.18	Monotone ↓
Trinary ORC + RH, η/%	52.57	52.70	52.84	52.98	53.08	52.88	↑ then ↓

All values at T₀ = 515 °C with a fixed temperature difference across the superheater; ORC working fluid: R236ea; ↑ = increase; ↓ = decrease. The thermodynamic impact of the supplementary firing ratio beta on the net efficiency of the selected CCPP configurations is graphically illustrated in Figure 5.

Table 12. Net efficiency (%) and net output (MW) of selected CCPP configurations versus supplementary firing ratio β at T₀ = 515 °C.

Configuration	β = 0	β = 0.1	β = 0.2	β = 0.3	β = 0.4	β = 0.5	Trend
Single press., η/%	48.57	48.32	48.05	47.75	47.40	46.90	Monotone ↓
Single press., N/MW	231.8	243.1	254.5	265.7	276.2	285.5	↑ +23%
Single press. + RH, η/%	50.94	50.71	50.94	51.22	51.50	51.12	Non-monotone
Dual press., η/%	50.78	50.44	50.12	49.78	49.34	48.82	Monotone ↓
Trinary ORC, η/%	51.57	51.31	51.08	50.84	50.54	50.18	Monotone ↓
Trinary ORC + RH, η/%	52.57	52.70	52.84	52.98	53.08	52.88	↑ then ↓

All values at T₀ = 515 °C with a fixed temperature difference across the superheater; ORC working fluid: R236ea; ↑ = increase; ↓ = decrease. The thermodynamic impact of the supplementary firing ratio beta on the net efficiency of the selected CCPP configurations is graphically illustrated in Figure 5.

Figure 5. Effect of supplementary firing ratio β on net efficiency for selected CCPP configurations (GTE-170, T₀ = 515 °C, R236ea). The trinary ORC + RH configuration shows an efficiency peak near β = 0.4, while most other configurations exhibit monotone efficiency decline.

Figure 5. Effect of supplementary firing ratio β on net efficiency for selected CCPP configurations (GTE-170, T₀ = 515 °C, R236ea). The trinary ORC + RH configuration shows an efficiency peak near β = 0.4, while most other configurations exhibit monotone efficiency decline.

Net output rises 20–24% across all configurations from β = 0 to β = 0.5, regardless of efficiency behavior. This peaking-power capability is the primary practical justification for supplementary firing. When H₂–CH₄ blends are used in supplementary burners, the carbon intensity of the incremental firing decreases in proportion to the hydrogen fraction, adding a marginal environmental benefit [18].

4.4.4. ORC Working Fluid Selection

R1336mzz(Z) is the strongest candidate among those assessed, combining high thermodynamic efficiency (estimated ORC efficiency: 13.5%), minimal environmental impact (GWP = 2, ODP = 0), and A1 safety classification. Compared to R236ea (which was selected for this study on availability grounds), R1336mzz(Z) would add approximately 1.2 percentage points in ORC efficiency—equivalent to roughly 1.5 MW of additional output at GTE-170 scale. Deployment of R1336mzz(Z) in new CCPPs in regions under the EU F-Gas Regulation 2024/573 represents a practical near-term application [18]. The thermophysical properties, environmental metrics (ODP, GWP), and estimated cycle efficiencies of the investigated ORC working fluid candidates for low-grade heat recovery are comparatively assessed in

Table 13. Comparative assessment of ORC working fluid candidates for low-grade heat recovery in the HRSG tail.

Fluid	${\mathbf{T}}_{\mathbf{c}\mathbf{r}}$, °C	${\mathbf{P}}_{\mathbf{b}\mathbf{a}\mathbf{r}}$,	ODP	GWP (100 yr)	ASHRAE Class	Est. ORC Eff., %	Notes
R236ea	139.3	34.2	0	1370	A1	12.3	Used in this study
R1233zd (E)	166.5	36.2	0	1	A1	13.1	Promising
R134a	101.1	40.7	0	1430	A1	10.8	Lower Tcr
R1336mzz (Z)	171.3	29.0	0	2	A1	13.5	Best candidate
R124	122.3	36.3	0.022	527	A1	11.6	Minor ODP
R227ea	101.7	29.3	0	3220	A1	10.5	High GWP
R744 (CO2)	31.0	73.8	0	1	A1	8.2	Transcritical cycle
R41	44.1	59.0	0	92	A1	8.8	High pressure

ODP = ozone depletion potential; GWP = global warming potential; R1336mzz(Z) compatibility note: thermal decomposition onset ≈ 185 °C [30]); operating margin of 25–85 °C in the HRSG tail under clean conditions. H₂ combustion at γ = 30% raises exhaust H₂O to 18–25% v/v, increasing cold-end corrosion risk; 316 L stainless steel or titanium heat exchanger tubes are recommended. H₂ permeation through austenitic steel at HRSG conditions is negligible. Full material compatibility testing under hydrogen-enriched conditions has not been reported and is identified as a research priority (Section 5.2, gap 5). $T_{cr}$ = critical temperature; estimated ORC efficiency at $T_{source}$ = 160 °C, $T_{cond}$ = 30 °C; data from NIST REFPROP DATABASE (V10.0; NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY, GAITHERSBURG, MD, USA) and ASHRAE [18,26].

.

5. Comparative Analysis

5.1. Comparison with Published Data

Results from this study agree with comparable swirl burner configurations (Emre et al. [19], Ji et al. [9], Nemitallah et al. [14]) within ±13%, which validates the experimental methodology. The ~10–13% systematic NOx excess in Nemitallah et al. [14] traces to a multi-point probe positioned close to the burner exit, which samples hotter gases from the inner recirculation zone. The structured quantitative comparison of NOx and CO results obtained in this study with comparable published datasets is compiled in

Table 14. Quantitative comparison of NOx and CO results with published studies under comparable conditions.

Reference	Burner Type	Fuel	γ, %	φ	SW	NOx, ppm	CO, ppm	vs. Present (NOx)	vs. Present (CO)
Present study (T2, 45°)	Swirl, premixed	H2 + LPG	30	0.7	0.8	29.7	187	Baseline	Baseline
Dostiyarov [13]	Swirl, diffusion	H2 + LPG	30	0.7	0.8	32.1	201	+8.1%	+7.5%
Emre et al. [19]	Swirl, premixed	NG + H2	30	0.7	0.8	26.5	175	−10.8%	−6.4%
Wang et al. [22]	Flat, premixed	H2 + LPG	30	0.7	—	18.2	—	−38.7%	—
Ji et al. [9]	Swirl, premixed	CH4 + H2	30	0.7	0.75	31.0	195	+4.4%	+4.3%
Nemitallah [14]	Swirl, premixed	CH4 + H2	30	0.7	0.8	33.5	210	+12.8%	+12.3%
Elbaz [21] (staged)	Double swirl	LPG	0	0.7	var	12.5	195	−57.9%	+4.3%
Park [26] (1 atm)	Flat, premixed	CH4 + H2	30	0.7	—	22.0	—	−25.9%	—

+ = above present study value; − = below. Comparisons adjusted for fuel composition differences, normalized by LHV. Inclusion criteria for Table 14: (a) H₂–CH₄ or H₂–LPG fuel blend; (b) swirl-stabilized burner; (c) φ = 0.7 ± 0.05; (d) γ = 30 ± 5%; (e) NOx or CO at 15% O₂ reference. Excluded: Kindra et al. [16,18,31]—CCPP model, no burner NOx/CO data; Ahmadi & Dincer [32]—thermoeconomic model; Abdollahian & Ameri [17]—supplementary firing only; Gaucherand [20]—H₂ + NH₃ fuel; Aravindan [25]—IC engine; Xu [6]—pure H₂, closed volume.

.

As illustrated in Figure 6, the flat-flame configuration of Wang et al. [22], without swirl, recorded NOx about 38% below the present study at comparable gamma and phi.

This quantifies the direct contribution of the recirculation zone to NOx: roughly one-third of total NOx in the present study comes from increased gas residence time in the recirculation zone, not from combustion chemistry per se. This result provides strong motivation for developing gas turbine combustors with minimal recirculation zones for hydrogen-enriched fuels.

Elbaz et al.’s [21] staged double-swirl configuration reached 12.5 ppm NOx for pure LPG versus 14.2 ppm in the present study at comparable conditions (Type 1, 30°, γ = 0%). The 12% gap suggests that staged combustion with optimized air distribution could deliver additional NOx reduction beyond what has been achieved here, and combining staged combustion with hydrogen enrichment is an as-yet-unexplored direction.

5.2. Knowledge Gaps and Research Priorities

The critical analysis identifies five specific gaps that current research has not addressed:

(1)

Pressure effects. The present study and most of the cited experiments were conducted at atmospheric pressure. Systematic investigation of NOx and CO formation in swirl burners with H₂-enriched fuels at 5–30 bar (representative of gas turbine combustion chambers) is needed for reliable scale-up. Park [26] established superlinear NOx growth with pressure, but the pressure–hydrogen fraction interaction for swirl configurations has not been mapped.

(2)

High hydrogen fractions (γ > 50%). Hydrogen roadmaps envision eventual 100% H₂ operation. Swirl burner behavior at γ = 50–100%—covering flashback, thermoacoustic stability, and NOx at turbine pressures—requires systematic investigation.

(3)

Transient operation. Gas turbines frequently operate during startup, load ramps, and shut down. Flame dynamics and emission transients under changing φ and SW with hydrogen-blended fuels have not been systematically studied.

(4)

Integrated CCPP–combustion optimization. This study treats combustion behavior and CCPP thermodynamics as separate analyses. A unified framework that simultaneously optimizes combustion chamber emissions, cycle efficiency, and economic performance with carbon pricing is missing from the open literature.

(5)

Low-GWP ORC fluid compatibility. Material compatibility and potential decomposition of R1233zd and R1336mzz at elevated temperatures in heat exchangers exposed to H₂–CH₄ combustion products remain an unresolved engineering problem.

6. Engineering Recommendations

Optimal Operating Parameters by Application

SCR feasibility: Standard V₂O₅/WO₃/TiO₂ catalysts operate at 280–420 °C and are compatible with H₂–CH₄ exhaust at γ ≤ 40% (no significant catalyst poisoning reported). At gas turbine pressure (30 bar) with scaled NOx of 114–184 ppm (Supplementary Table S1), achieving the EU IED target of <25 ppm requires 85–87% SCR efficiency. For a 250 MW CCPP: estimated capital cost €4–6 million (1.6–2.4% of plant capital cost, current European market data); catalyst volume approximately 800–1200 m³; ammonia consumption ~40–60 t/yr urea solution (≈60–90 k €/yr operating cost). These costs are within the range accepted for regulatory compliance in commercial CCPPs. The comprehensive engineering guidance and optimal fuel-combustion parameters categorized by specific application contexts are systematically summarized in

Table 15. Engineering guidance on H₂–CH₄ combustion parameters by application context.

Application	H2 Fraction γ	Equiv. Ratio φ	Vane Angle	Injection	NOx Target	Rationale
Base load, low emissions	20–30%	0.5–0.6	45° (SW = 0.8)	Type 2	<25 ppm	Balance NOx, stability, CO
Peak load (max power)	10–20%	0.7–0.85	60° (SW = 1.3)	Type 1	<50 ppm	Stability at high load
Part load (stability)	30–40%	0.4–0.5	45–60°	Type 1	<35 ppm	LBO margin at lean φ
EU IED compliance	0–20%	0.5–0.7	30–45°	Type 2	<25 ppm	NOx minimization
HRSG supplementary firing	10–30%	0.3–0.5	N/A (forced)	Staged	<40 ppm	HRSG mode control
Trinary CCPP with ORC	20–30%	0.5–0.7	45°	Type 2	<30 ppm	CO2 reduction + ORC

All NOx values at 15% O₂, atmospheric pressure. Scaling to gas turbine pressure (15–45 bar) is expected to increase NOx 2–4×; SCR (selective catalytic reduction) will be required for γ > 20% in most regulatory jurisdictions. CO limit: <150 ppm in all applications.

.

Table 16. Performance comparison: pure CH₄ vs. hydrogen-enriched blends in a trinary CCPP (GTE-170, R236ea, T₀ = 515 °C, β = 0).

Parameter	CH4 (Baseline)	20% H2	30% H2	Unit	Δ (20% H2)	Δ (30% H2)
Net efficiency	51.57	51.72	51.78	%	+0.15 pp	+0.21 pp
Net electrical output	249.1	249.8	250.2	MW	+0.7	+1.1
Specific CO2 emissions	0.397	0.360	0.339	kg/kWh	−9.3%	−14.6%
Burner NOx (SW = 0.8, φ = 0.7)	17.5	27.8	34.6	ppm	+58.9%	+97.7%
Burner CO (SW = 0.8, φ = 0.7)	248	190	168	ppm	−23.4%	−32.3%
Adiab. flame temp. (stoich.)	2230	2310	2360	K	+80 K	+130 K
φLBO (45°, Type 1)	0.52	0.44	0.41	—	−15.4%	−21.2%
ORC output (R236ea)	18.7	18.9	19.0	MW	+1.1%	+1.6%
Fuel cost (relative, LHV basis)	1.00	1.18	1.31	—	+18%	+31%
Required H2 infrastructure	None	~14.8 MWth	~23.7 MWth	—	New infra.	New infra.

pp = percentage points; NOx at atmospheric pressure; at gas turbine conditions, NOx will be 2–4× higher; fuel cost based on current H₂ production cost ~3× natural gas by LHV; CO₂ reduction reflects fuel carbon composition only.

lays out the core trade-off in hydrogen blending. The environmental case is clear: 14.6% specific CO₂ reduction (thermodynamic model prediction; Table 16) and better flame stability at 30% H₂. The problems are also clear: a 97.7% increase in burner NOx and a 31% fuel cost premium at current hydrogen prices. Carbon pricing mechanisms are probably required to make hydrogen blending economically viable in the near term. On the technical side, advanced combustion chamber designs—staged combustion, swirl optimization, and active NOx control—are necessary prerequisites for regulatory compliance, not optional upgrades.

7. Conclusions

Gas turbines around the world run on natural gas—and they account for a significant share of carbon emissions in the power sector. One way to cut those emissions without replacing equipment is to add hydrogen directly into the pipeline, blending it with conventional methane. While conceptually straightforward, the introduction of hydrogen into the combustion chamber gives rise to several coupled effects: the burning velocity increases, flame stability changes, and NOx emissions rise. The present study quantifies these effects experimentally, rather than through calculation alone, in a physical burner across 240 operating conditions.

Flame stabilization. Hydrogen enrichment extended lean blowout stability across all swirler configurations tested. With Type 1 injection, every 10% H₂ added improved stability by approximately 32%; with Type 2, by 46%. Maximum stability (φLBO = 0.18) was recorded at 60°, γ = 40%, v = 5 m/s, Type 1. The critical blowout window narrows from Δφ ≈ 0.05 at γ = 0% to ≈ 0.02 at γ = 40%.

NOx emissions. Each 10% increase in γ raised NOx by 23–24%. The vane angle effect is comparable in magnitude: moving from SW = 0.4 to SW = 1.3 raised NOx by ~65% independently of γ. An empirical regression model explained 97% of NOx variance. Type 2 injections produced 11% lower NOx than Type 1. At atmospheric pressure, EU NOx limits (<25 ppm) are achievable at γ ≤ 30% with a 45° vane angle and Type 2 injection. All NOx values reported are atmospheric-pressure measurements and represent lower bounds for gas turbine conditions. Based on NOx(P) = NOx(P₀) × (P/P₀)ⁿ with n = 0.5–0.7 [7,26,30], values at 15–30 bar are estimated 2–5× higher; SCR is required for EU IED compliance at all tested configurations at turbine pressures.

CO emissions. CO dropped ~8–10% per 10% H₂, driven by elevated OH-radical concentrations and higher combustion temperature. Minimum recorded CO was 101 ppm (γ = 30%, Type 2, 45°, φ = 0.3). Simultaneous EU compliance (NOx < 25 ppm, CO < 90 ppm) was not achievable within the tested parameter space at φ ≥ 0.5 without additional abatement.

Optimal operating window. The recommended parameter set is γ = 20–30%, φ = 0.5–0.7, 45° vane angle (SW = 0.8), Type 2 injection, yielding NOx = 24–30 ppm (atmospheric), CO = 180–210 ppm, and φLBO = 0.37–0.43—a viable operating range for lean-premixed gas turbine combustors.

Thermodynamic integration. The trinary cycle (ORC with R236ea + reheat, GTE-170) achieved 52.57% net efficiency—4.0 percentage points above the single-pressure baseline and 1.9 pp above the dual-pressure reheat scheme. Specific CO₂ emissions fell 10.5%. Supplementary firing at β = 0.1–0.5 increased net output 20–24%; efficiency fell in most configurations except reheat schemes at moderate β. Using H₂–CH₄ blends in supplementary burners reduced CO₂ by up to 14.6% at γ = 30%.

Knowledge gaps. Five research priorities emerge: (1) pressure-dependent NOx and CO behavior at γ > 0% for swirl configurations; (2) combustion behavior at γ > 40%; (3) transient flame dynamics under load change; (4) integrated combustion chamber–CCPP optimization frameworks; (5) material compatibility of low-GWP ORC fluids (R1233zd, R1336mzz) with hydrogen–methane combustion products. These are the remaining obstacles to safe, efficient, and regulation-compliant hydrogen blending in gas turbines.