Experimental Ignition Delay Time Measurements and Chemical Kinetics Modeling of Hydrogen/Ammonia/Natural Gas Fuels

The United States Energy Information Administration has projected that worldwide carbon emissions will increase 0.6% per year through 2050 [1]. In order to minimize greenhouse gases and other toxic gas emissions and promote sustainability and efficient energy production, there has been interest in utilizing hydrogen (H₂) and hydrogen-carrying compounds as a replacement fuel. Through movements such as the Paris Climate Agreement as well as the executive order in the U.S. to reach net-zero emissions by 2050 (with the intent to restrict government agencies to use 100% clean electricity by 2030), the idea of carbon-neutral cycles has accrued interest from multiple governments around the world as well as private companies looking to minimize their carbon footprint and decrease costs. Accordingly, there have been new initiatives to utilize H₂ and H₂-carriers (e.g., ammonia, NH₃) for zero-emission power generation. Hydrogen functions as a zero-emission fuel and, as a result, is highly desirable for use in various power systems as, when burned with oxygen, water is the only product. One of the key impediments to the widespread use of hydrogen is the high cost. Industry users have little incentive to transition to hydrogen fuels over the already-existing infrastructure for propane and natural gas, due to the increased cost for transportation and storage. For example, for large-scale power generation such as gas turbines, it would be preferable to use existing technologies and infrastructure for burning hydrogen fuels with natural gas. NH₃ is a hydrogen-carrier that has thermal properties similar to that of propane and could thereby utilize the already existing storage and transport facilities. NH₃ has gained international interest as a fuel due to this already existing infrastructure and comparable energy density [2]. Such natural gas blends are essential to the future of clean energy.

Ammonia's use as a fuel was previously studied in the 1960s, but early shock tube studies showed low accuracy and repeatability [3–9]. More recently, ammonia oxidation has accrued increased interest and has had its ignition characteristics studied under a wide set of temperatures and pressures. He et al. studied ammonia oxidation mixtures at 20 bar and 60 bar in a rapid compression machine while Song et al. investigated ammonia oxidation reactions at pressures of 30 bar and 100 bar in a laminar flow reactor; also, Iki et al. investigated the use of ammonia as a fuel in a microgas turbine [10–13]. Additionally, a study by Mathieu and Petersen looked into ammonia oxidation reactions at pressures of 1.4 atm, 11 atm, and 30 atm and high dilution percentages (98%/99% dilution) [14].

Some studies focus on ammonia/air and ammonia/syngas mixtures with NO_x formation carried out with techniques ranging from modeling studies, laminar burning velocity studies, and shock tube studies [15–17]. For example, Mathieu et al. measured ignition delay times (IDTs) using syngas mixtures doped with parts per million levels of ammonia using a shock tube. This study found that ammonia had little effect on ignition delay times throughout their pressure ranges of 1.5 atm, 12 atm, and 30 atm [18]. Shu et al. also carried out a shock tube study focusing on ammonia/air mixtures at pressures of 20 bar and 40 bar, investigating auto-ignition characteristics at these relatively high pressures [19]. Furthermore, da Rocha et al. computationally investigated chemical kinetic modeling of ammonia/hydrogen/air mixtures by studying auto-ignition and flame speed and NO formation and found that hydrogen addition to ammonia could improve its combustion behavior [20].

Ammonia/hydrogen mixtures have also been studied more in recent years. Chen et al., for example, investigated the ignition delay times of ammonia/hydrogen blends in a shock tube at temperatures from 1020 K to 1945 K and pressures from 1.2 atm to 10 atm with hydrogen fractions of 0%, 5%, 30%, and 70% while Valera-Medina et al. studied lean premixed combustion of ammonia/hydrogen in swirling gas turbines combustors, finding high instability in equivalent NH₃/H₂ blends [21,22]. There have also been multiple numerical and simulation-based studies. One carried out by Pochet et al. found that hydrogen addition to ammonia in a rapid compression machine promoted ignition delay times to a significant degree in the low temperature range (1000–1100 K) [23]. Li et al. also carried out a numerical study focusing on hydrogen addition to ammonia flames, finding increased reactivity of ammonia combustion with hydrogen addition [24].

Very little literature exists focused on ammonia/natural gas mixtures, especially the ignition characteristics. Most recently, Oh et al. investigated the laminar flame speeds and carbon monoxide (CO) production of natural gas/ammonia mixtures in a spark-ignition engine and found that emissions increased with increasing air–fuel ratio and drastic reduction of CO₂ emissions with the introduction of ammonia as a replacement fuel at temperatures of 300 K and 600 K [25]. Additionally, Ishaq and Dincer investigated ammonia/natural gas blends in a wind-energy based system and saw increased combustion efficiency with ammonia addition while Ito and Onishi have investigated ammonia/natural gas cofiring in a gas turbine, though these studies focus primarily on emissions in the combustor with no model comparisons of ignition characteristics [26–28]. While in the past, ammonia has been studied with biodiesel/diesel fuels in engines (resulting in similar energy replacement as an engine fuel), there are few present studies at these baseline conditions [29]. With such limited available literature, there is a strong need for experimental IDTs in order to properly understand the behavior of ammonia/natural gas combustion.

There is currently a gap in the literature for natural gas/ammonia ignition studies as well as natural gas/hydrogen studies with equivalent ammonia/hydrogen fractions. The current study offers to improve the understanding of ignition and combustion behavior of hydrogen-carrier fuels with natural gas. The ignition delay times and CO concentration time histories are reported using a shock tube and laser absorption spectroscopy. To the best of the authors' knowledge, this study provides novel shock tube measurements of NH₃/natural gas and H₂/natural gas mixtures at these relevant conditions, essential points to act as validation targets for further development of hydrogen and natural gas mechanisms. While this study focuses on natural gas + NH₃ and natural gas + H₂, mixtures of H₂/NH₃, neat H₂, and neat natural gas were also studied to provide a comparison at similar conditions.

Experiments for the combustion of NH₃/H₂ and natural gas/NH₃ were carried out in the University of Central Florida's (UCF) facility containing a high-purity, stainless-steel shock tube having an inner diameter of 14.17 cm. Specific details of the facility can be found in earlier publications [30–32]. The velocity of the incident shock wave was measured using five piezo-electric pressure transducers (PCB 113B26; 500 kHz frequency response) equally spaced along the last 1.4 m of the shock tube linked to four time-interval counters (Agilent53220A; 0.1 ns time resolution). These incident shock velocity values were then linearly extrapolated to obtain the reflected shock velocity at the end wall location. The pressure and temperature after the reflected shock wave were calculated based on the extrapolated velocity, initial pressure, initial temperature, and normal one-dimensional shock relations. The test section consists of eight equally spaced optical ports located 2 cm from the end wall, with one port being equipped with a pressure transducer (Kistler 603B1-piezo-electric). All measurements for this experimental campaign were taken at this 2 cm location.

Anhydrous ammonia and a research-grade ammonia/nitrogen mixture (>98% pure; Air Liquide, Orlando, FL), natural gas mixture (>99.99% purity; nexAir), oxygen (O₂; >99.99% purity; nexAir), and balance gases (nitrogen (N₂), argon (Ar); >99.99% purity; nexAir) were used to prepare the test mixtures, where the natural gas mixture is made up of 93% CH₄, 4.75% C₂H₆, 1.5% C₃H₈, and 0.75% C₄H₁₀. The mixtures were prepared manometrically in the 33 L Teflon coated, stainless steel mixing tank fitted with a magnetically driven stirrer to ensure homogeneity of the mixture, measuring the partial pressures with 100-Torr (MKS Instruments/Baratron E27D), and 10,000 Torr (MKS Instruments/Baratron 628D) full-scale range capacitance manometers during mixture preparation and the filling process. Helium was used as the driver gas for all experiments. Test mixtures were allowed to homogenize for 4+ h before experiments were carried out. The components of the test mixtures in this campaign can be seen in Table 1, all having an equivalence ratio of 1 except for mixture 4 which has an equivalence ratio of 0.767. Mixtures will henceforth be referred to by mixture number or M#.

Since ammonia has been known to adsorb on contact with stainless steel, all mixtures containing ammonia were prepared using a passivation technique during mixture preparation on the internal surface of the mixing tank as well as before each experiment on the internal surface of the shock tube, as suggested by Mathieu and Petersen [14]. During passivation of the mixing tank, ammonia was introduced to the system at an equivalent pressure to the desired partial pressure of ammonia in the mixture. The ammonia was allowed to adsorb for approximately 15 min before the tank was vacuumed down to <1 Torr and the final mixtures could be prepared. Passivation of the shock tube took place before each experiment in a similar process. Ammonia was introduced to the shock tube at a pressure equal to the desired P1, allowed to adsorb for approximately 10 min, and was then vacuumed out to a pressure of 5 × 10⁻⁵ Torr before refilling to the final P1.

Ignition delay times can be determined in multiple different ways and, depending on the definition, can result in differences of over 100 μs [33]. The definition selected for use for the current study is the time between the arrival of the reflected shockwave (determined from the pressure spike) to the time coinciding with the peak of the normalized emission signal at the sidewall location. A representative pressure and emission profile can be seen in Fig. 1. Ignition delay time measurements were carried out via a photomultiplier tube (model 2032) with a 310 ± 2 nm bandpass filter to isolate the OH* emission also at the 2 cm location from the end wall. The overall uncertainty in shock tube ignition delay time measurements is less than ±10%, resulting largely from uncertainty in mixture composition, thermodynamic parameters, and reflected shock temperatures. The uncertainty for the CO mole fraction is estimated via a time-varying root-mean-square quantity where the uncertainties from the parameters of Beer's law (Eq. (1)) are taken into account (these parameters include absorbance, pressure, temperature, and absorption cross section). More information about this method can be found in Ref. [34].

where $αv$ is the absorbance at frequency ν; R (J/mol K) is the universal gas constant; T (K) is the temperature of the gas; σ (m²/molecule) is the absorption cross section of the absorbing species; P (Pa) is the pressure of the gas; L (m) is the path length of the shock tube; and in Eq. (2), I and I_o are the measured intensities of laser power in test gas mixture and in vacuum, respectively, at frequency ν. The absorption cross section values were obtained from HITRAN 2016 database at each P5 and T5 condition where absorbance from potential interfering species was negligible [35].

Numerical simulations of shock tube experiments were performed using chemkin-pro [36] with a closed (0D) homogeneous batch reactor under adiabatic conditions with constant UV assumption. When the shock wave passes through the reaction mixture, the step change in temperature and pressure justifies the former assumption while the adiabatic assumption can be justified as the whole process takes place in less than 5 ms.

To simulate auto-ignition delay times and CO time histories for natural gas/hydrogen/ammonia mixtures, a kinetic model was developed using the base mechanism from Ref. [37]. The Rahman et al. model is well validated for ignition delay times of methane oxidation for pressures up to 300 bar. It is also validated with experimental data for oxidation of smaller hydrocarbons in the presence of nitrogen oxides. However, this mechanism lacked reactions of C₃ and C₄ hydrocarbons which were added from the Aramco 3.0 mechanism [38]. Since this work focuses on oxidation studies with a nitrogen bath gas, the rates for the reactions N₂H₂ = NNH + H and N₂H₂ + H = NNH + H₂ were replaced with the rates from study of Dean and Bozzelli that are relevant to a nitrogen bath gas [39]. The new reaction mechanism consists of 121 species and 1040 reactions and will be referred to as the “UCF 2022” mechanism hereafter. The reaction mechanism is available from authors upon request. Model results reported here include simulations using the UCF 2022 and GRI 3.0 mechanisms. The hydrocarbon C₄H₁₀ from the natural gas mixture was added to the next highest species component (C₃H₈) for the GRI 3.0 mechanism as it does not contain this higher hydrocarbon.

Shock tube ignition delay times of neat natural gas are offered as a comparison to observe the effect that hydrogen-carrying compounds have on the baseline neat mixture. The experimental data with best fit lines can be seen in Fig. 3 with neat natural gas shown as the baseline. It is apparent that hydrogen addition to natural gas promotes ignition on a large scale, while ammonia addition slows down the ignition. For 1% natural gas + 1% ammonia, the ignition delay time decreases by almost a factor of two while for 1% natural gas, 0.1% ammonia, the ignition delay time increases, more so at higher temperatures than compared with the low temperature conditions. The latter can be attributed to the leaner equivalence ratio of mixture 4.

Shock tube ignition delay times of neat hydrogen are offered as a comparison to observe the effect of the addition of hydrogen-carrying compounds to the baseline neat mixture. The experimental data with best fit lines can be seen in Fig. 4 with neat hydrogen as the baseline. It can be seen that with the addition of a hydrogen-carrying compound, the ignition delay time of the mixture increases as compared to the neat hydrogen case. For mixture 5, the slope mirrors that of the neat H₂ case while mixture 2 varies in slope. Both mixtures 2 and 5 show a strong increase in the trend of ignition delay times, with the ignition delay time of the mixtures doubling that of the neat H₂ case.

Figure 5 shows a comparison between CO production during combustion of natural gas + hydrogen-carrier compound at low and high temperature conditions. In general, each mixture shows the expected trend of the elevated temperature condition resulting in faster CO formation. The elevated temperature and pressure due to the combustion process causes natural gas to oxidize rapidly, leading to the steep increase in CO mole fraction seen in the high temperature conditions.

When comparing H₂ addition to the neat natural gas case, it is found that at the low temperature conditions, mixture 2 (Fig. 6) shows increased CO production when compared to the baseline mixture 1 (Fig. 5) at much lower temperature. This is because the hydrogen added to mixture 2 enhances the reaction progress. Mixture 3 (Fig. 7), which contains 1% ammonia needs a higher temperature (∼1509 K) for CO to form in significant amounts compared to the baseline mixture 1. At similar temperatures (∼1390 K), mixture 4 produces less CO (Fig. 8) compared to mixture 1 at the end of test time. This is due to the combined effect of lower natural gas concentration (1% natural gas) and ammonia addition slowing the reaction progress. For the high temperature condition, mixtures 2, 3, and 4 (Figs. 6–8) show a large drop in CO production as compared to mixture 1 (Fig. 5). This is attributed to the higher concentration of natural gas in mixture 1 (2%) compared to other mixtures. For the high temperature condition, the peak concentration of mixture 4 (which has 0.1% NH₃ addition) is now much closer in scale to the CO production of mixtures 2 and 3 than at the low temperature condition, indicating dependence on temperature.

It can be noted that the low and high temperatures between mixtures are varying. This is due to the fact that some mixtures have faster ignition and form CO at lower temperatures while others were slow to ignite. The low temperature condition for each mixture was selected as the temperature close to where detectable amounts of CO was formed within the test time available during experiments (∼0.8–1.5 ms). The high temperature condition for each mixture was chosen so that the important features of the CO time history (steep increase, peak, and consumption) were clearly visible. The difference in temperature between the mixtures is primarily due to the combustion properties of individual components in each mixture. For example, mixture 2 (1% natural gas + 1% hydrogen) requires only 1275 K to produce CO within test times (Fig. 6), due to the presence of hydrogen. However, mixture 3 (1% natural gas + 1% ammonia) does not produce detectable amounts of CO at this temperature and requires higher temperatures (∼1500 K, Fig. 7) to form CO within the test time.

Figure 9 shows the model comparison with neat mixture 1, mixture 2, and neat mixture 6, as a baseline validation for the UCF 2022 model. The in-house model is compared with the industry standard GRI 3.0 mechanism in order to further confirm model functionality. It can be seen that both the UCF 2022 and GRI 3.0 models closely align with the experimental data, indicating a good baseline.

Figure 10 shows the model comparison for all mixtures containing ammonia as the additive for the hydrogen-carrying compound as a baseline before adjustments were made to the model. The UCF 2022 mechanism is shown along with the GRI 3.0 mechanism in order to compare model functionality. It can be seen that neither model is fully able to accurately capture the combustion chemistry of ammonia when added to baseline H₂ or natural gas. For mixture 3, as well as mixture 4, both the GRI 3.0 mechanism and the UCF 2022 mechanism are within the margin of error. For mixture 3, there is more variation from the model at low temperatures, where the UCF 2022 model trends closer to the experimental data. However, for the mixture 5 case, the GRI 3.0 model greatly overpredicts the IDTs while the UCF 2022 model aligns well at lower temperatures and is still within the margin of error for the experimental points at higher temperatures.

The experimental data obtained for CO time histories was used to validate the UCF 2022 model. The results are shown in Fig. 11. For the neat mixture 1 case, both models underpredict CO formation at 1393 K. However, for mixtures 2 and 3, both models were able to capture CO time histories with reasonable accuracy with GRI 3.0 being closer to experimental results.

The mechanism selected for use during the sensitivity analysis is the UCF 2022 mechanism. This was selected because, while both mechanisms match at certain conditions, the UCF mechanism is closer to the experimental data more often. OH* sensitivity analyses were carried out in order to identify key reactions that affect the ignition delay times of ammonia/natural gas and ammonia/hydrogen mixtures. IDT sensitivity analyses were done for critical reactions, and the specific reactions of interest were isolated, results of which can be seen in Fig. 12.

A positive sensitivity indicates increased ignition delay times (lower reactivity) while a negative sensitivity value indicates lower ignition delay times and higher reactivity. The specific reactions of interest for this study include reactions due to the interaction of the hydrogen-carrier and the ammonia compound. As such, these specific reactions are the only ones shown in the sensitivity studies.

Between the two conditions for mixture 3 (both low and high temperatures), only two reactions, though they vary in sensitivity coefficient, are shared indicating a high dependence of the combustion chemistry on temperature. However, for the two mixture 5 studies (once again, low and high temperature), four reactions are shared between the two cases with the similar reactions being of closer scale than in the mixture 3 studies.

The reaction path analysis was carried out for high temperature oxidation of ammonia in the presence of hydrogen and natural gas to understand important reaction pathways involved in NO formation and ammonia decomposition. The UCF 2022 reaction mechanism was used for reaction path analysis as it predicted IDTs with a good degree of accuracy.

NH also reacts with NO to form N₂O and then undergoes H-abstraction to form nitrogen atoms, both of which eventually form a stable nitrogen molecule as shown in Fig. 13.

Figure 14 shows the reaction pathway diagram for ammonia oxidation in the presence of natural gas. Similar to the case with hydrogen in the ammonia + hydrogen mixture, a significant amount of H, OH, and O radicals are created by the initial reactions of larger hydrocarbons (C₃H₈, C₄H₁₀, etc.) with oxygen. These radicals help in the decomposition of ammonia via the H-abstraction reactions shown in reactions (R1)–(R3). Since the concentration of C₃–C₄ species in natural gas is very low (<0.01%), the pathways of decomposition of these species are not significant for the mixture under consideration except for their contribution to the initial radical pool.

In this work, the authors offered the needed experimental data for natural gas + NH₃ and natural gas + H₂ mixtures to better understand the combustion phenomena at engine-relevant conditions. It was found that hydrogen addition to natural gas decreases the ignition delay times during combustion to a significant degree while ammonia addition was shown to have varying effects depending on the fuel composition. 1% natural gas + 1% ammonia resulted in a large increase in the ignition delay times, while 1% natural gas + 0.1% ammonia showed a slight decrease in ignition delay times, indicating the importance of the ammonia fraction in the mixture and equivalence ratio on the combustion characteristics. For the neat hydrogen comparison, it was found that ammonia addition delayed ignition. IDTs increased at both low temperatures and high temperatures with large variance at both conditions. A chemical kinetic model “UCF 2022” was compiled and the predictions were found to be in good agreement with experimental results. A sensitivity analysis and a reaction path analysis have been conducted and important reactions involved in ammonia oxidation in the presence of hydrogen and natural gas have been identified.

Disclaimer: This report was prepared as an account of work sponsored by an agency of the U.S. Government. Neither the U.S. Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U.S. Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the U.S. Government or any agency thereof.

• NASA (Grant No. 80NSSC21K2056; Funder ID: 10.13039/100000104).

• The U. S. Department of Energy (DE-FE0032072; Funder ID: 10.13039/100000015).

• The NASA Florida Space Grant Consortium.

• UCF.