Pilot Ignition of Ammonia Spray Using Dimethyl Ether Spray at Elevated Temperature: A Numerical Study

Abstract

Ammonia (NH₃) is a promising zero-carbon fuel to eliminate carbon footprint while the high autoignition temperature and low combustion rate of NH₃ remain challenging for practical implementation. Using dimethyl ether (DME) as pilot ignition fuel can substantially promote the reactivity of NH₃, thus paving the way for a widespread application of NH₃. In this study, the ignition process and nitrogen oxides (NO_x) emissions of the NH₃ liquid spray ignited by liquid DME spray were numerically investigated using Converge software. The ambient temperatures (T_amb) ranging from 900 K to 1100 K were used to mimic the in-cylinder temperature typically encountered in turbocharger engines. The effect of ammonia energy ratio (AER) and fuel injection timing was examined as well. It is found that only half of NH₃ is consumed at T_amb = 900 K while 97.4% of NH₃ is burned at T_amb = 1100 K. Nitric oxide (NO) and nitrogen dioxide (NO₂) formation also have strong correlation with T_amb and NO₂ is usually formed around the periphery of NO through these two channels HO₂ + NO = NO₂ + OH and NO + O(+M) = NO₂(+M). Extremely high nitrous oxide (N₂O, formed by NH + NO = H + N₂O) and carbon monoxide (CO) are produced with the presence of abundant unburned NH₃ at T_amb = 900 K. Additionally, increasing AER from 60% to 90% results in slightly declined combustion efficiency of NH₃ from 98.7% to 94%. NO emission has a non-monotonical relationship with AER owing to the ‘trade-off’ relationship between HNO concentration and radical pool at varying AERs. A higher AER of 95% leads to failed ignition of NH₃. Advancing DME injection not only increases combustion efficiency, but also reduces NO_x and CO emissions.

1. Introduction

Ammonia has been widely considered as a promising zero-carbon fuel to replace traditional hydrocarbon fuels in combustion engines to tackle carbon emissions [1,2,3]. It can be synthesized from renewable hydrogen (H₂) [4,5] and easily liquefied at room temperature with compression pressure of ~10 bar, which remarkably reduces its storage and transportation cost compared to hydrogen [6]. However, using NH₃ alone in conventional engines [7,8] remains challenging because of its high ignition energy [9] and slow flame speed [10]. The ignition temperature of NH₃ is approximately 400 K higher than diesel [11], which necessitates the usage of reactive additives as pilot ignition fuel. Following this strategy, Reiter and Kong [12,13] modified a diesel engine to supply NH₃ through the intake manifold while directly injecting diesel or biodiesel into the cylinder as a pilot fuel, achieving a maximum ammonia energy ratio (AER) of approximately 60%. Soot emissions could be remarkably reduced with high AER; however, high unburned ammonia concentrations were observed due to poor combustion efficiency in low ammonia flame temperatures. According to the lower heating values of these two fuels, AER is defined using the following equation:

$$\begin{array}{c}AER={\displaystyle \frac{m_{{NH}_3}·{LHV}_{{NH}_3}}{m_{{NH}_3}·{LHV}_{{NH}_3}+m_{DME}·{LHV}_{DME}}}\times 100\%\end{array}$$

(1)

where $m_{{NH}_3}$ and $m_{DME}$ represent the injection masses of NH₃ and DME, respectively. ${LHV}_{{NH}_3}$ and ${LHV}_{DME}$ denote their lower heating values. Here, the combustion efficiency of NH₃ is defined as the ratio of burned NH₃ to total NH₃ injected into the combustion chamber. Wang et al. [14] explored the reduction in greenhouse gas and unburned ammonia by manipulating the pilot diesel injection strategy with a maximum AER of ~80%. Zhang et al. [15] applied NH₃/diesel high-pressure dual-fuel direct injection (HPDI) method in a two-stroke low speed engine and achieved acceptable NO_x emission using reasonable arrangement of injection times of diesel and ammonia. It has been recognized that dual-fuel high-pressure direct injection has superior performance compared to port injection ammonia engines in emission control and maintaining power performance owing to higher charge efficiency and flexible fuel injection strategy [15,16]. Li et al. [17] reported that the dual-fuel high-pressure direct injection strategy increased the AER by 17% compared to the low-pressure injection mode; meanwhile, unburned ammonia and NO_x were substantially reduced. Liu et al. [18] reported that increasing the diesel injection pressure from 80 MPa to 120 MPa leads to improved combustion efficiency to 82.64% in the ammonia/diesel dual-fuel engine.

Aiming at accelerating the ignition of ammonia, DME is an attractive candidate, having a higher cetane number (~55) and less tendency of soot formation compared to diesel. DME can also be synthesized from renewable energy sources [19] and has been widely used as a solvent [20], suggesting that its storage and transportation technologies are as mature as those for ammonia. Our previous measurements [9] show that 5% DME addition can lower the autoignition temperature of NH₃ by ~250 K. As a result, using DME as pilot fuel and simultaneously adopting HPDI method can pave the way for the application of ammonia in CI engine with high efficiency and low emissions.

Gross et al. [21] used direct injection of premixed liquid NH₃/DME fuel in compression ignition engines and discovered low soot emissions but increased carbon monoxide (CO) and unburned hydrocarbons (UHCs) compared to pure DME engines due to the lower flame temperature of NH₃. AER was set lower than 40%, which is far from the goal of using ammonia as primary fuel. Ryu et al. [22] suggested that 60% DME addition is necessary to achieve a stable combustion using the same fuel injection strategy. They also pointed out that the combustion of high concentration of ammonia exhibits HCCI characteristics and deteriorated engine performance was reported at a higher AER [22]. According to these studies, direct injection of premixed NH₃/DME liquid is not a promising method in CI engines due to a limited AER and difficulties of flexible fuel injection control, which is extremely important to match engine load. In our previous study, the NH₃/DME dual-fuel direct injection (HPDI) method was applied in a constant volume combustion chamber (CVCC) and the ignition process of NH₃ spray, ignited by pilot DME spray, was examined at a high AER (up to 80%) [23]. It was found that the heat absorption of NH₃ spray evaporation plays a significant role; thus, properly adjusting NH₃/DME injection timing to allow for mature flame kernel formation of DME spray is a key factor affecting the ignition of NH₃ spray. The maximum combustion efficiency of NH₃ was obtained at ~93% [20], which remains challenging for unburned NH₃ emissions. The ambient temperature was set at 900 K in [20], which might be a barrier to achieving higher combustion efficiency and AER. Furthermore, turbocharged engines enable higher compression ratios and elevated ambient temperatures at top dead center. As a result, investigating the pilot ignition process of NH₃ spray using DME spray towards higher temperatures can help to bridge the gap.

In this study, the ignition process and NO_x emissions of NH₃ spray ignited by DME spray were investigated through CFD simulation using Converge 2.4 software. To accelerate the simulation, a compact NH₃/DME mechanism is used in the simulation, which is reduced from a full NH₃/DME mechanism developed in our previous study [24]. The reduced mechanism is validated against the measured ignition delay times (IDTs) [9] and laminar burning velocities (LBVs) [9]. The performance of predicting NO using the reduced mechanism was verified against the measured NO concentration from the jet stirred reactor (JSR) measurements [25]. The effect of ambient temperature, the AER and injection timing of NH₃ and DME spray were studied to optimize the ignition process and mitigate NO_x emissions. The constant-pressure assumption was intentionally chosen to isolate and clarify the chemical and physical mechanisms governing DME-assisted NH₃ ignition under a well-controlled environment, avoiding the additional complexities associated with dynamically varying pressure and piston motion.

2. Numerical Approach

2.1. Schematic and Simulation Schemes

The schematic of the NH₃/DME dual-fuel high-pressure injection model in a constant volume combustion chamber (CVCC) is shown in Figure 1. The distance between two injectors is set at 7 cm and the angle ($\theta$) between two sprays is set at 120°. The start of injection (SOI) of NH₃ is set at time = 0 ms. The physical properties of NH₃ and DME are presented in

Table 1. Comparison of properties between liquid ammonia [17], DME [26] and diesel [27].

Parameter	Ammonia	DME	Diesel
Boiling point/K	239.8	358	450–643 K
Autoignition temperature/K	924	623	226–233
Octane number	130	-	-
Cetane number	-	55–60	40–55
Lower heating value/(MJ/kg)	18.8	28.43	43.5
Latent heat of vaporization/(KJ/kg)	1.37	460	270
Laminar burning velocity/(m/s)	0.07	0.54	-

. The simulation schemes are summarized in

Table 2. Combustion chamber parameters and simulation schemes.

Parameter	Value
Chamber size/mm	100 × 100 × 100
Ambient temperature/K	900, 1000, 1100
Ambient gas	Air
Injector distance/cm	7.0
DME injection pressure/MPa	75
Ammonia Nozzle diameter/mm	0.22
DME Nozzle diameter/mm	0.18
Ammonia injection mass/mg	18.15
DME injection mass/mg	3
Nozzle angle/°	120
NH3 injection pressure/MPa	75
AER/%	60, 80, 90, 95
SOI of DME/ms	−1, 0.5, 0, 0.5, 1
Ambient pressure/MPa	3.8

. The commercial software Converge was used to run CFD simulations. The computational domain was defined as a rectangular box with dimensions of 48 mm × 48 mm × 97 mm. Mesh independent analysis has been performed in our previous study [23], and a base grid size of 4 mm was proved to maintain simulation accuracy with high computational efficiency.

2.2. Mechanism Reduction and Validation

An accurate NH₃/DME combustion mechanism was proposed in our previous study [24], which was validated rigorously against the measured IDTs, LBVs and NO concentrations. This mechanism is reduced using the DRGEP method [27,28] in the Chemkin Pro 19.2 software and a few key reactions are updated to achieve better prediction in this study. In particular, the interaction between DME and amine radical, CH₃OCH₃ + NH₂ = CH₃OCH₂ + NH₃, is very important for IDT prediction. The A-factor of the Arrhenius equation was reduced by a factor of 2.5 and the activation energy was increased by 2 kcal/mol from the calculated ab initio rate constant in our previous study [9] to achieve better agreement between calculated and measured IDTs. In this reduced mechanism, the rate constant was changed back to the original calculated values, which are fairly accurate. Eventually, a reduced NH₃/DME mechanism is obtained with 94 species and 562 reactions and further used for CFD simulation.

The measured and calculated LBVs of NH₃, with 20% DME addition as a function of $\phi$ using different O₂ content at 298 K/1 bar and 373 K/473 K/5 bar using an air oxidizer, are shown in Figure 2. Clearly, the reduced mechanism accurately predicts the measured LBVs and shows even better performance when compared to the original full mechanism under some conditions. More validations are provided in the Supplementary Material. The measured and calculated IDTs at $\phi$ = 1.0 and compression pressure (P_c) of 60 bar are shown in Figure 3. The reduced mechanism reproduces the measured IDTs well at condition studies. More validations at $\phi$ = 0.5 and 2.0 are provided in the Supplementary Material as well. The measured NO concentration in a jet stirred reactor (JSR) from Yin et al. [25] was used for mechanism validation. The calculated NO concentration of NH₃ with 50% DME, using the reduced mechanism from this study and measured NO as a function of temperature, is shown in Figure 4. The reduced mechanism accurately predicts the measured NO concentrations well, although slight underprediction is observed at around 1150 K, $\phi$ = 1.0.

2.3. Spray Model Verification

The accuracy of spray ignition modeling heavily depends on the accuracy of physicochemical models of NH₃ and DME. The Kelvin–Helmholtz and Rayleigh–Taylor (KH-RT) model [29] was employed to simulate droplet breakup and atomization processes [30]. The KH (Kelvin–Helmholtz) component primarily describes the primary and secondary droplet breakup induced by the instability of cylindrical liquid jets and viscous effects. In this model, the breakup time (τ_KH) and the droplet radius after breakup (r_KH) are defined as follows:

$$\begin{array}{c}{\tau }_{\mathrm{K}\mathrm{H}}={\displaystyle \frac{3.726B_1{r}_0}{{\mathit{\Lambda }}_{\mathrm{K}\mathrm{H}}{\mathit{\Omega }}_{\mathrm{K}\mathrm{H}}}}, r_{\mathrm{K}\mathrm{H}}=B_0{\mathit{\Lambda }}_{\mathrm{K}\mathrm{H}} \end{array}$$

(2)

where r₀ means the initial droplet radius, Λ_KH denotes the wavelength of the perturbation wave and Ω_KH represents the maximum growth rate of the perturbation wave. The RT (Rayleigh–Taylor) model describes droplet breakup induced by the effect of drag forces, proposing that RT wave instability induces liquid breakup at a specific distance from the nozzle. The breakup time (τ_RT) and droplet radius after breakup (r_RT) are expressed as follows:

$$\begin{array}{c}{\tau }_{\mathrm{R}\mathrm{T}}={\displaystyle \frac{C_{\mathrm{\tau }}}{{\mathit{\Omega }}_{RT}}}, r_{\mathrm{R}\mathrm{T}}={\displaystyle \frac{C_{\mathrm{R}\mathrm{T}}{\mathit{\Lambda }}_{\mathrm{R}\mathrm{T}}}{{\mathit{\Omega }}_{RT}}} \end{array}$$

(3)

where Λ_RT denotes the wavelength of the most unstable perturbation wave and Ω_RT represents its frequency. Because liquid ammonia (NH₃) and dimethyl ether (DME) differ notably in physical properties such as viscosity and density, separate parameter settings are required for each in the KH–RT spray breakup model.

In addition to the KH-RT model, the Redlich–Kwong equation [31] was employed to characterize gas states, while the Reynolds-Averaged Navier–Stokes (RANS) RNG k-ε model [29] was used to describe the complex turbulent flow within the combustion chamber. The method proposed by O’Rourke and Amsden was used to model droplet collisions and wall interactions [32]. The No Time Counter (NTC) method and Frossling correlation [33] were used to simulate droplet collisions and evaporation, respectively. The SAGE model [34] was used to compute the reaction rate coupling the reduced mechanism. Detailed spray model validations of NH₃ and DME were performed in our previous study [23], which show good agreement between the simulations and experiments from the literature. As a result, the same spray models are used in this study, and the detailed coefficients of spray models are presented in

Table 3. Parameter settings for DME and liquid ammonia KH-RT models.

Parameter	Value
B1 of DME	11
B1 of NH3	8
B0	0.61
Cτ	1
CRT	0.1

.

3. Results and Discussion

3.1. Effect of Ambient Temperature on the Ignition Process

The ambient temperature is the initial temperature inside the combustion chamber before fuel injection, which has a profound effect on the ignition process. The calculated temperature contour and mass fractions of NH₃ and DME at different ambient temperatures are shown in Figure 5. The DME and NH₃ sprays are injected simultaneously and AER is set at 80%. The dark particles represent fuel droplets which are not evaporated due to either low temperatures or short evaporation time. As can be observed, the tip of the DME spray ignites at time = 0.9 ms and the flame front starts to ignite the ammonia spray at the head of the NH₃ spray. Considering the significant cooling effect of NH₃ evaporation, the temperature of NH₃ vapors remain lower than the ambient temperature even after coming into contact with the flame front of DME spray at time = 0.9 ms. The NH₃ flame grows to ~1700 K after DME is totally burned at time = 2.5 ms and remarkably cold NH₃ spray and droplets are surrounded by NH₃ flames, which are continuously consumed at time = 3.0 ms. When the ambient temperature is increased to 1000 K, the flame kernel area of DME spray is larger than that at T_amb = 900 K and time = 0.9 ms, which results in higher heat transfer to heat up NH₃ spray. Furthermore, the reactivity of NH₃ is increased at higher ambient temperatures; as a result, the flame kernel temperature of NH₃ spray is much higher that at T_amb = 1000 K and time = 2.5 ms. Increasing the ambient temperature to 1100 K leads to a dramatic acceleration of DME autoignition and NH₃ combustion. Specifically, DME autoignition is accelerated and stable flame kernel of DME sprays is observed at 0.66 ms as shown in Figure 5c. The flame kernel area is twice as large than it was at T_amb = 900 K and time = 0.9 ms, which wraps up the cold NH₃ spray facilitating heat transfer from DME flame to cold NH₃ spray. Consequently, much higher NH₃ flame temperature is observed in Figure 5c. Focusing on the DME and NH₃ mass fractions at these three different ambient temperatures, the evaporations of DME and NH₃ are both slightly prompted at higher T_amb. However, compared to more pronounced changes in flame kernels area and temperatures, it can be found that increased ambient temperatures enhance autoignition of NH₃/DME sprays primarily via enhanced fuel reactivity rather than faster evaporation rate. The calculated IDTs of pure NH₃ and NH₃ with 20% DME at 3.8 MPa are shown in Figure 6. Clearly, increasing temperature from 900 K to 1100 K leads to ~2 orders of magnitude decrease in IDT of pure NH₃. The IDTs of NH₃ with 20% DME are reduced from 2.1 ms to 0.53 ms, which is in line with the accelerated autoignition observed in Figure 5.

The higher ambient temperature not only accelerates the autoignition process, but also improves the combustion efficiency. As shown in Figure 7, the combustion efficiency of NH₃ at T_amb = 900 K is 52.1%, which indicates that only half of NH₃ is burned at the end. The combustion efficiency of NH₃ is increased to 93.8% and 97.4%, respectively, when T_amb is increased to 1000 K and 1100 K. Considering the significant heat absorption of NH₃ evaporation and low reactivity, using DME as pilot fuel and enhancing the ambient temperature simultaneously are realistic solutions to promote both NH₃ autoignition and combustion efficiency.

To further understand the pollution formation during the ignition and flame propagation process, the NO, NO₂, N₂O and CO mass fractions at the same condition as that in Figure 5 are presented in Figure 8. The NO formation location strongly overlaps with the flame front where high flame temperature and OH concentration at the flame front facilitate the NO formation compared to the flame contour and NO mass fraction at time = 0.9 ms and T_amb = 900 K in Figure 5 and Figure 8. That is because NO formation is prompted at high flame temperatures; such strong correlations between NO and temperature have been discussed in a previous study [35]. NO mass fraction after 0.9 ms is dramatically reduced in Figure 8a; it can be attributed to the lower flame temperature (see Figure 5a) and combustion efficiency of NH₃ at T_amb = 900 K, as shown in Figure 7. Nearly half of NH₃ is not burned and generates de-NO_x agent NH₂ radicals. The abundant NH₂ radical consumes NO via

NH₂ + NO = N₂ + H₂O (R1).

R1 is a famous selective non-catalyst reduction (SNCR) reaction, which works efficiently at the temperature range of 1150~1350 K. The flame temperature at time = 2.5 and 3.0 ms in Figure 5a is close to this temperature range and thus benefits the NO reduction through R1.

NO₂ is primarily formed around the periphery of NO, and the concentration is nearly one magnitude lower than NO. The NO₂ is mainly generated from NO via

HO₂ + NO = NO₂ + OH (R2)

NO + O(+M) = NO₂(+M) (R3).

The forward direction of R3 favors low temperature, which is why NO₂ is formed at medium-flame temperatures or a post-flame zone in a practical combustion device. Additionally, R3 is a pressure-dependent reaction, which becomes very important at a high ambient pressure of 3.8 MPa. However, due to the presence of DME and its oxidation product, CH₂O and CH₃, the produced NO₂ is partially consumed by these species via

CH₂O + NO₂ = HCO + HONO (R4)

CH₃ + NO₂ = CH₃O + NO (R5)

NH₂ and HNO play an important role in converting NO₂ back to NO, via

NH₂ + NO₂ = H₂NO + NO (R6)

HNO + NO₂ = HONO + NO (R7).

In conclusion, NO formation and NO₂ formation is strongly correlated to high-flame temperature and medium-flame temperature, respectively. The competition and conversion chemistry between NO and NO₂ is responsible for the distinct boundary between NO and NO₂ observed in Figure 8a.

N₂O exists in the interlayer between NO and NO₂ at time = 0.9 ms and T_amb = 900 K as shown in Figure 8a. N₂O formation depends on concentration of NH and NO, i.e.,

NH + NO = H + N₂O (R8).

Meanwhile, a NH radical is formed through a sequence of consecutive reactions of NH₃, i.e., NH₃ → NH₂ → NH, which is abundant in the intermediate flame temperature at time = 0.9 ms. Both R2 and R8 favor intermediate flame temperature; however, the HO₂ is usually produced in more O₂ enriched condition. As a result, NO₂ is produced in a region further outward than N₂O.

The produced N₂O can be destroyed by

H + N₂O = N₂ + OH (R9)

and thermal decomposition as follows:

N₂O(+M) = N₂ + O(+M) (R10)

The reactions between N₂O and O/OH radicals also consume N₂O whereas the rates of these reactions are much lower than those of R8–R10.

Similar distinct boundary between NO, NO₂ and N₂O can also be observed at T_amb = 1000 K and 1100 K at time = 0.9 ms as shown in Figure 8b,c. However, at time = 2.5 ms and 3 ms, T_amb = 1000 K, and a significant amount of NO is formed at the head of NH₃ diffusion flame compared to that at T_amb = 900 K, because the flame temperatures at time = 2.5 ms and 3.0 ms are much higher and NO formation is prompted at a higher temperature via the following key steps:

HNO(+M) = H+ NO(+M) (R11)

HNO + H = H₂ + NO (R12)

HNO + OH = H₂O + NO (R13) and

NH + O = H + NO (R14).

Moreover, the higher flame temperature also enhances the thermal NO formation through

N + O₂ = O + NO (R15) and

N + OH = H + NO (R16).

These two thermal NO formation channels compete with NO consumption reaction,

N + NO = N₂ + O (R17).

NO₂ and N₂O formation are both enhanced as well at T_amb = 1000 K and 1100 K due to increased NO formation, which further facilitates NO₂ and N₂O formation channels, i.e., R2 and R8.

In the DME-assisted NH₃ jet flame, CO is mainly generated from the partial oxidation of DME and accumulates in the intermediate-temperature zone ahead of the main reaction front. The CO concentration at T_amb = 900 K is much higher than that at T_amb = 1000 K and 1100 K because the intermediate flame temperature hinders the further oxidation of CO to CO₂.

The net NO, NO₂, N₂O and CO emission profile as a function of time is presented in Figure 9. At T_amb = 900 K, the NO is formed quickly at the ignition moment and then decreases due to the de-NO_x reaction R1 at low NH₃ combustion efficiency. It is partially converted to NO₂ at the post-flame zone, which is in line with the NO and NO₂ contour in Figure 8. At T_amb = 1000 K and 1100 K, the NO emissions also increase sharply then decrease before time = 2.5 ms; however, a slight increase is found after that moment. This is because, unlike the low NH₃ combustion efficiency (~52.1%) at T_amb = 900 K, the NH₃ diffusion flame becomes self-sustained after time = 2.5 ms at T_amb = 1000 K and 1100 K. It can be regarded as a second stage of combustion of NH₃, which further increases NO by burning more NH₃. Consequently, the total NO emission is increased from 0.18 mg to 0.3 mg and 0.35 mg, respectively, when T_amb is increased from 900 K to 1000 K and 1100 K. Similarly, the two-stage NO₂ formation at T_amb = 1000 K and 1100 K can also be attributed to the two-stage combustion of NH₃. NO₂ emission exhibits similar trends to those of NO with increased ambient temperature. However, an opposite trend is observed for N₂O, i.e., increasing T_amb leads to a sharp reduction in N₂O. Such a phenomenon can be anticipated because the destruction reaction of N₂O, i.e., R9 and R10, are both promoted at higher temperatures. The formed one is promoted due to increased H concentration and the later one is enhanced in a way that it is a decomposition reaction. At T_amb = 900 K, due to low ammonia combustion efficiency, intermediate flame temperature leads to very high CO emissions, which are almost five times higher than NO. Similarly to N₂O, higher T_amb results in higher flame temperatures, thus promoting CO oxidation to CO₂ at T_amb = 1000 K and 1100 K.

3.2. Effect of AER at T_amb = 1100 K

The aforementioned combustion efficiency of NH₃ at T_amb = 1100 K reaches 97.4% in Figure 7, which implies that using the elevated ambient temperature is beneficial for reducing unburned NH₃. As a result, the autoignition process at T_amb = 1100 K is calculated at different AERs, as shown in Figure 10. The SOI of DME and SOI of NH₃ occur at the same time. At AER = 60%, 40% DME is used as pilot ignition fuel, which hinders the evaporation rate of DME. Thus, the autoignition zone of DME spray is smaller at 0.66 ms compared to that at AER = 80% (see Figure 5c) and AER = 90%, as shown in Figure 10b. However, given enough time for DME evaporation, the flame kernel of DME at 0.9 ms becomes the largest and thus transfers more heat and reactive radicals (H, O and OH) to NH₃ spray. Moreover, due to reduced NH₃ amounts, the cooling effect of NH₃ evaporation is reduced at a higher AER, facilitating NH₃ combustion. The NO formation again shows a strong correlation with flame temperature. However, the highest temperature is observed at time = 0.9 ms in Figure 10a, generated by the DME flame, and the highest NO formation is observed at time = 2.5 ms, generated by the NH₃ flame. Such a phenomenon indicates that NO formation is not only dominated by temperature, but that the presence of NH₃ and its subsequent oxidation products also plays a critical role. Increasing AER from 60% to 80% and 90% leads to less DME injection, which benefits DME evaporation and leads to advanced DME autoignition and a larger flame kernel at 0.66 ms, as shown in Figure 5c and Figure 10b. However, increased AER deteriorates the ignition of NH₃. For instance, the NH₃ spray penetrates the flame zone at time = 2.5 ms in Figure 10b and more NH₃ liquid still can be observed at time = 3 ms. For all three different AERs, the NO₂ is always formed around the periphery of NO. For instance, at AER = 90% and time = 3 ms, NO₂ can be found upstream of NH₃ spray and DME spray, where the flame temperature is lower than the flame front due to the cooling effect of NH₃ evaporation. N₂O is again formed in the interlayer between NO and NO₂ as shown in Figure 10.

The combustion efficiency of NH₃ at three different AERs is shown in Figure 11. The combustion efficiency of NH₃ decreases from 98.3% to 97% and 94%, respectively, when AER is increased from 60% to 80% and 90%. For practical engines with higher AER, such decreases of NH₃ combustion efficiency seem to be acceptable and other strategies can be utilized to compensate for such slight decreases in NH₃ combustion efficiency. Further increasing AER to 95% leads to a failed ignition in the present configuration.

The net NO, NO₂, N₂O and CO emissions as a function of time are depicted in Figure 12. For all three cases, NO emissions present similar profiles to those in Figure 9. For AER = 90%, significant two-stage formation can be observed due to two-stage combustion of NH₃, i.e., the first stage NH₃ combustion induced by pilot DME and the second stage NH₃ combustion due to ammonia’s own diffusion flame. Such two-stage formation of NO is not significant at lower AER. Generally, the highest NO emission is found at AER = 60%, followed by AER = 90% and AER = 80%. Such a non-monotonical relationship between NO formation and AER has been discussed in our previous study [24]. The radical pool (H, O and OH) increases almost linearly with higher DME fraction [24]; however, HNO is one of the most important NO formation precursors, which generates NO by decomposition or reacting with the radicals, H and OH, through R11, R12 and R13. HNO has a non-monotonical relationship with a growing DME fraction [24]. Thus, the ‘trade-off’ between HNO concentration and the radical pool concentration induced by DME addition is responsible for the non-monotonical relationship between NO formation and AER in Figure 12. Such a ‘trade-off’ relationship can be affirmed by the HNO spatial distribution shown in Figure 10c, where the highest HNO concentration is observed at AER = 80% rather than AER= 60% or 90%. The emissions of NO₂ and CO vary little at different AERs because flame temperatures are high enough to promote the consumption of these two species as discussed in Section 3.2. However, N₂O at AER = 90% is nearly two and three times higher than those at AER= 80% and 60%, respectively, which can be attributed to the increased NH concentration at higher AER.

3.3. Effect of DME Injection Timing at T_amb = 1100 K and AER = 80%

The autoignition process of NH₃/DME spray with varying injection strategies at T_amb = 1100 K and AER= 80% is shown in Figure 13. Compared to the case of simultaneous SOI of NH₃ and DME shown in Figure 13c, when SOI of DME is 1 ms and 0.5 ms earlier, it can be observed that DME autoignition is less affected by the cooling effect of NH₃ spray; thus, more heat and radicals are generated at the early stage. Consequently, higher flame temperatures are achieved at time = 0.9 ms in Figure 13a,b. Such a hot atmosphere is beneficial for NH₃ combustion; for instance, more intensified NH₃ diffusion flames are observed at 2.5 ms in Figure 13a,b than that in Figure 13c. When SOI of DME is postponed after SOI of NH₃ by 0.5 ms and 1 ms, NH₃ sprays have more time to evaporate before contacting DME spray and the autoignition of DME is inhibited by the heat absorption of NH₃ spray. As a result, advancing DME injection time leads to a more intensified NH₃ diffusion flame, which is favored from the perspective of enhancing the autoignition of NH₃ spray. Regarding NO and NO₂ formation, pure DME combustion also generates NO and NO₂ at the head of DME diffusion flame at time = 0.9 ms as shown in Figure 13a,b. The NO_x is produced via thermal NO_x as explained previously. NO₂ is found to be significant at the upstream of DME diffusion at time = 2.5 ms shown in Figure 13a,b, and it can be attributed to the produced HO₂ radicals from the DME promoting reaction

HO₂ + NO = NO₂ + OH (R2).

The NO formation is intensified when SOI of DME is postponed, albeit NH₃ diffusion flame temperature is lowered simultaneously. It seems to contradict the observations in Figure 9 that higher ambient temperatures lead to higher flame temperature at fixed AER and SOI. It is necessary to point out that NO formation is not only dependent on flame temperature, but it also heavily depends on the equivalence ratio. Advanced SOI of DME results in more and faster consumption of O₂ which substantially reduces the O₂ concentration for NH₃ diffusion flame. Richer NH₃ flames have higher NH₂ concentration which reacts with NO through

NH₂ + NO = N₂ + H₂O (R1).

Much less NO formation in richer NH₃ flames have also been reported in [36] and discussed in Section 3.1.

In terms of N₂O, it is not observed at SOI = −1 ms and −0.5 ms, though NO and NO₂ are produced at times before 2.5 ms, as shown in Figure 13a,b,d,e. As explained previously, NH₃ flame should exist to produce NH, which is necessary to react with NO to generate N₂O via R8. Similarly to Figure 8, N₂O is generally formed in the interlayer between NO and NO₂. The majority of CO is generated at the core and periphery of DME jet before time = 0.9 ms and almost disappears at the end of combustion due to less interference by NH₃ jet as shown in Figure 13a,b. Delaying the SOI of DME leads to more unburned CO at the end of combustion, owing to lowered flame temperature and lower O₂ concentration which prohibit further oxidation of CO.

Net NO, NO₂, N₂O and CO emissions as a function of time is calculated accordingly at different SOIs of DME. As shown in Figure 14, the NO emission at SOI = −1 ms shows an obvious two-stage formation; the first stage is produced by thermal NO from DME flame and the second stage is generated by NH₃ diffusion flame. The net NO emissions at the combustion end are generally reduced by advancing DME injection due to an O₂-deficient environment caused by faster DME combustion, which is in line with the discovery in Figure 13. The NO₂ emission is also reduced by advancing DME injection. At SOI = 0.5 ms and 1 ms, the NO₂ emission rate is much higher than other cases. It can be attributed to intermediate flame temperatures benefiting NO₂ formation as discussed in Section 3.1. Advancing SOI of DME substantially reduces the N₂O and CO emissions as well, which can be attributed to the increased flame temperature at earlier DME injection as depicted in the temperature contour in Figure 13. When DME is injected first, the ignition occurs earlier, forming hot atmosphere to benefit NH₃ combustion. Higher flame temperature facilitates complete oxidation of CO while suppressing N₂O pathways. Conversely, when NH₃ is injected earlier than DME, the subsequent DME injection leads to a staged combustion process where NH₃ undergoes low-temperature oxidation first, forming NH and HNO intermediates, followed by delayed high-temperature oxidation of DME. This results in a prolonged low temperature, which promotes the formation and accumulation of N₂O and CO.

The effect of SOIs of DME on combustion efficiency is presented in Figure 15. Clearly, a slight increase of NH₃ combustion efficiency from 96.7% to 97.5% is achieved when SOI of DME is advanced from time = 0.5 ms to −1 ms. Delaying DME injection by 1 ms shows 4% decrease. It can be anticipated that further delaying DME injection could lead to lower combustion efficiency, and even failed ignition.

4. Conclusions

This study conducted a numerical investigation of NH₃ spray ignition by DME spray in a constant volume chamber. A new reduced reaction mechanism (94 species and 562 reactions) was developed and validated for CFD simulations. The effects of ambient temperature, AER and SOI of DME were investigated. The main findings are as follows:

• Increasing the ambient temperature from 900 K to 1100 K significantly accelerates the autoignition of both DME and NH₃, reducing IDT by two orders of magnitude. Consequently, NH₃ combustion efficiency increases from 52.1% to 97.4%. N₂O and CO are significantly reduced at higher ambient temperature owing to higher flame temperature. However, a substantial increase in NO and NO₂ emission is observed. NO formation has a strong correlation with flame temperature. NO₂ is usually formed around the periphery of NO because the two primary NO₂ formation channels HO₂ + NO = NO₂ + OH (R2) and NO + O(+M) = NO₂(+M) (R3) favor intermediate flame temperature.
• Raising the AER from 60% to 90% results in a slight decline in combustion efficiency of NH₃ from 98.7% to 94% due to enhanced evaporative cooling and reduced pilot fuel energy. NO emission has a non-monotonical relationship with AER, which can be attributed to the ‘trade-off’ relationship between NO formation precursor (HNO) and the radical pool at varying AERs. N₂O formation is promoted due to increased NH concentration at high AER.
• Advancing DME injection not only benefits combustion efficiency but also reduces NO, NO₂, N₂O and CO emissions. The reduced NO can be attributed to increased equivalence ratio caused by faster DME combustion, thus facilitating de-NO_x reactions (NH₂ + NO = N₂ + H₂O). Meanwhile, reduced NO₂, N₂O and CO are caused by increased flame temperature with earlier DME injection.

Generally, this study offers insight into the autoignition process and NO_x formation of NH₃/DME dual-fuel direction injection combustion, which helps to bridge the gap between fundamental ammonia combustion theory and practical combustion system design. More calculations at a wider range of conditions will be performed in our future work to further optimize the ignition process and reduce NO_x emissions.