Experimental and numerical study of NO_x formation in a domestic H₂/air coaxial burner at low Reynolds number

Highlights

• Hydrogen is a potential fuel for domestic burners.

• NO_x reduction can be achieved by flame splitting method.

• Laminar and turbulent flames show opposite NO_x trends with increasing inlet powers.

• In low Reynolds number flames lower equivalence ratios imply higher NO_x concentrations.

• Competing factors describing local NO_x production intensity can be defined.

Abstract

Thermal NO_x formation in ${\text{H}}_2$/air jet flames from a coaxial burner is studied experimentally and numerically. The aim is to study possible NO_x reduction strategies for domestic gas boiler burners. Following a flame splitting method strategy, a single burner is studied at different inlet powers (from 0.2 to 1.0 kW). The effect of three different fuel-air ratios (or equivalence ratio φ) is considered by varying the coaxial air stream, with fuel-air ratios corresponding to values of $\phi <1$, relevant for domestic boiler applications (here $\phi =0.77$, $\phi =0.83$ and $\phi =0.91$). NO_x concentrations increase with increasing inlet power between 0.2 and 0.6 kW and numerical results are in good correspondence with available experimental data. The opposite trend is observed above 0.6 kW and no numerical results are obtained, indicating a transition from laminar to turbulent flames. On the other hand, in contrast to the observations made in turbulent non-premixed flames, reducing the equivalence ratio implies higher NO_x concentrations in the low Reynolds number flames considered. The numerical results in the laminar regime are used to highlight and quantify three competing main factors concerning NO_x production in order to interpret the experimental observations: the volume of the region where NO_x is produced, and within this region, the competition between residence time and NO_x reaction rate. Based on this analysis, different design strategies for low NO_x hydrogen diffusion burners are finally discussed.

Introduction

Hydrogen is a potential alternative to fossil fuels in transportation, industrial, commercial and residential sectors [[1], [2], [3]]. Since hydrogen can be generated through different renewable energy sources [[4], [5], [6]], the so-called hydrogen economy could be an integral part of a renewable energy system, allowing the substitution of pipeline natural gas with renewable hydrogen to reduce carbon emissions as well as the dependence on fossil fuels [2,7,8]. The end use of this renewable hydrogen has been a research topic during the last decades. For instance, different studies focused on the use of generated renewable hydrogen as fuel in combustion devices such as internal combustion engines or burners [[9], [10], [11], [12]]. An extensive research literature on hydrogen flames can be found, since its combustion behaviour presents several particularities compared to other conventional gaseous fuels, due to its physical properties. Among others, hydrogen presents a wide flammability limit in air (4–75 vol$\text{\%}$), low ignition energy in air (0.019 mJ), low density (0.0899 kg/m³) and high adiabatic flame temperature (2380 K) [[13], [14], [15], [16]]. Hence, the variety of combustion devices in different sectors still requires an in-depth research work to re-adapt or re-design them so as to ensure safe and efficient operating conditions when using hydrogen as inlet fuel. In this context, the present study focuses on the use of gaseous hydrogen as fuel in domestic gas boiler burners. As discussed by de Vries et al. [17], while fundamental changes in combustion phenomena and exhaustive analysis of individual combustion devices has been widely studied, there is no clear conclusion concerning the maximum fraction of hydrogen that maintains the performance of installed domestic combustion appliances without adverse consequences.

The safety risk when burning hydrogen is closely related to the flame flashback phenomena due to the high flame front velocities in premixed and partially premixed flames, mainly caused by an imbalance in the local flame and flow speeds, when the former exceeds the latter [[18], [19], [20], [21]]. Recently, Zhao et al. [22] studied the influence of hydrogen addition to pipeline natural gas on the combustion performance of a partially premixed cooktop burner. They concluded that the burner could operate safely and efficiently with a maximum hydrogen concentration of 20$\text{\%}$. In addition, the measured NO_x emissions were over 100 ppm, without meaningful variations when adding hydrogen. Similarly, Choudhury et al. [23] studied the admissible hydrogen percentage in natural gas in low-NO_x and conventional water heaters, working with partially premixed and lean premixed conditions. They analyzed characteristics such as flashback, ignition delay, flame structure and emissions. The most interesting conclusion was that the maximum hydrogen tolerance in both low-NO_x and conventional water heater was below 10$\text{\%}$ by volume.

On the other hand, many combustors operate in the non-premixed mode since they ensure safer operating conditions, avoiding flashback and explosion risks [15,24,25]. However, in hydrogen flames around atmospheric pressure as in the present study, where NO_x formation can be correctly described by the well known Zeldovich mechanism or thermal mechanism [26,27], the high adiabatic flame temperature of hydrogen diffusion flames (peak temperatures above 2300 K in this study) can yield to high levels of NO_x [[28], [29], [30]]. Therefore, the main technological challenge in the design of hydrogen burners is to maintain both a high hydrogen fuel content and acceptable levels of NO_x emissions. This clearly requires new or modified existing combustion equipments for hydrogen. Non-premixed combustion can be considered to be the best option to design domestic hydrogen burners since they completely eliminate the flashback phenomena, and voluminous research has been published on turbulent non-premixed hydrogen burners. However, laminar non-premixed hydrogen burners have been hardly studied in the literature. We summarize here some relevant studies.

Igawa et al. [31] examined the effect of splitting hydrogen diffusion flames on NO_x emissions for stove gas burners. The experimental setup had no confinement or combustion chamber for the flames, and hydrogen was mixed with the ambient air through diffusion. They recorded measurements at different nozzle diameters for individual and multi-flame burners (0.6, 1.2 and 2.6 mm) and at different nozzle intervals for multi-flame measurements. They also experimentally concluded that splitting the flame was an effective method to decrease NO_x emissions in hydrogen diffusion flames, since they measured lower NO_x emissions when dividing a single flame burner into three flames for the same total input power.

Chen and Driscoll [33] experimentally studied the effects of coaxial air and fuel jet Reynolds number on NO_x formation in non-premixed jet flames. They aimed at deriving general scaling laws for NO_x production (in methane/air and hydrogen/air flames, both with and without coaxial air, and both in laminar and turbulent flames). The results showed that coaxial air reduced NO_x formation in hydrogen flames and had a negligible effect in methane flames at Re = 5000. In addition, the results for simple hydrogen jet diffusion flames without coaxial air showed a reduction of NO_x levels when reducing the hydrogen inlet diameter from 0.37 to 0.16 cm in both laminar and turbulent regimes. Their attempt to derive general scaling laws is interesting in order to higlight what we will consider here as the three main physical aspects that play a key role, and may compete, in NO_x production:

• Following the Eulerian approach of Peters et al. [34], they related the production of NO_x to a reaction volume ${\mathcal{V}}_{\text{reac}}$ (proportional to $L_f^2.\delta$, where $L_f$ is the flame length and δ the flame thickness);

• From a Lagrangian point of view, they pointed out how the relevant dimension is a residence time ${\tau }_{\text{res}}$ (within ${\mathcal{V}}_{\text{reac}}$);

• They also indirectly considered the chemical time of NO_x production (i.e. the inverse of a reaction rate ${\dot{\omega }}_{\text{NO}x}$).

Keeping this in mind, the following references are interesting since they give some insight on these factors in laminar diffusion flames.

Zhao et al. [35] studied the OH formation in a hydrogen diffusion co-flow burner through chemiluminescence images and CFD simulations. The flow conditions were fixed with and without co-flow air with a central hydrogen jet of D = 15 mm and low thermal powers ($\sim$0.2 kW). The numerical results showed a flame height of $\sim$ 20 mm and were in good agreement with the experimental measurements. The numerical results showed slightly longer flames for the case with co-flow air as well as higher maximum temperature. This could be attributed to the higher inlet thermal power used in the flame with co-flow air.

Li et al. [36] experimentally and numerically investigated the combustion and heat release characteristics of a hydrogen laminar diffusion flame in a non-premixed micro-jet burner (1 mm diameter for H₂ inlet and 3.16 mm for air) together with OH distribution measurements at various inlet powers (from 0.1 kW to 1.0 kW) and equivalence ratios. The flame structure was obtained by flame image detection and spectroscopic systems. However, NO_x emissions were not measured in this work. Laminar flames were studied varying the inlet hydrogen Reynolds number between 14 and 144, while the Reynolds number of the inlet air flows varied between 2 and 52. Their experimental and numerical results showed longer flames for higher inlet powers, whereas different equivalence ratios would lead to similar flame volumes at a given inlet power.

Khan and Raghavan [37] investigated non-premixed laminar jet flames using carbon monoxide-hydrogen mixtures. Varying the inlet composition, they carried out a numerical and experimental analysis of the flame length, thermal plume width, temperature and flame luminosity. They concluded that the flame became shorter as hydrogen content was increased and that thermal plume width got wider. In addition, when adding hydrogen to the inlet mixture, the maximum temperature isotherm moved closer to the burner exit (moving the reaction volume to a zone of shorter residence times ${\tau }_{\text{res}}$), but also showing an increasing trend in the maximum temperature (and therefore possibly higher ${\dot{\omega }}_{\text{NO}x}$).

In this paper, we present the results of experiments in a coaxial H₂/air burner at low Reynolds number in order to study possible flame splitting strategies for NO_x reduction in domestic boiler burners. The different flames are modelled numerically in order to further analyze the results observed experimentally and better understand the different trends in thermal NO_x concentrations depending on inlet power and equivalence ratio. Following the ideas suggested by Peters et al. [34], the numerical results are used to extract the thermal NO_x reaction volume ${\mathcal{V}}_{\text{reac}}$ and a weighted average $\overline{{\dot{\omega }}_{\text{NO}}}$ indicating the competition between residence time ${\tau }_{\text{res}}$ and chemical reaction time $1/{\dot{\omega }}_{\text{NO}}$ within ${\mathcal{V}}_{\text{reac}}$. In the light of this discussion of the measured NO_x trends, different strategies are discussed for the design of low NO_x hydrogen diffusion burners. Finally, the effect of coaxial air on NO_x reduction compared to a simple jet flame is illustrated in the appendix.