Low-emission operation of an NH 3 -diesel dual-fuel 4-stroke engine targeting EU Stage V performance via combined combustion control and exhaust gas aftertreatment

Abstract

Internal combustion engines operated in combined heat and power generation applications can achieve high overall efficiencies by utilizing both electrical output and exhaust heat, while also offering rapid load flexibility for grid support. However, engines fueled with conventional diesel or natural gas contribute significantly to greenhouse gas emissions. To mitigate CO₂ emissions, engines operated with hydrogen admixture to natural gas or in pure hydrogen mode are under development. Nevertheless, the high reactivity of hydrogen and the associated risk of knocking combustion posing challenges for engine operation and can limit thermal efficiency. In addition, the low volumetric energy density of hydrogen complicates fuel transport and storage, particularly in applications lacking direct access to gas or hydrogen infrastructure. In this context, ammonia (NH₃) represents a promising alternative due to its higher volumetric energy density and established transport and storage infrastructure. Direct combustion of ammonia in internal combustion engines, however, introduces several technical challenges. While ammonia combustion is CO₂-free, the formation of nitrous oxide (N₂O), a potent greenhouse gas, can occur. Additionally, compliance with stringent emission regulations requires effective control of nitrogen oxides (NO_x) emissions and unburned ammonia (NH₃ slip). Moreover, emissions of carbon monoxide (CO) and unburned hydrocarbons (HC), originating from small quantities of diesel pilot fuel used for ignition support, must also be minimized. To fully realize the greenhouse gas reduction potential of ammonia-fueled engines while meeting modern emission limits, an integrated optimization of combustion and exhaust gas aftertreatment (EAT) is essential. In this study, an ammonia-diesel dual-fuel combustion concept was investigated using a single-cylinder research engine (SCE). A coupled approach was applied, combining targeted adjustments of engine operating parameters to shape the raw exhaust gas composition with a systematic evaluation of two different exhaust aftertreatment catalyst configurations on the SCE.

Introduction

The use of ammonia (NH₃) as a fuel is getting increasingly relevant, particularly for large engine applications in shipping or stationary power generation. When produced using renewable electricity, ammonia offers significant greenhouse-gas reduction potential, as its combustion does not generate CO₂. Moreover, ammonia benefits from an existing global infrastructure for production, storage and transport, which may facilitate its large-scale deployment as an energy-carrier. It’s comparatively high volumetric energy density provides advantages for storage and transport, especially when compared with pure hydrogen (H₂).^1–4 With respect to its use in combustion engines, ammonia’s high knock resistance enables the use of high compression ratios, offering the potential for increased thermal efficiency. ⁵

However, ammonia also exhibits poor ignition and combustion characteristics. Its high autoignition temperature of 651°C and high minimum ignition energy of 8 mJ pose challenges for stable ignition. ⁶ Therefore, a secondary fuel is often required. In spark-ignition engines (SI), hydrogen can be added to the NH₃-air mixture to promote ignition, ⁷ while in compression-ignition engines directly injected diesel fuel can serve as an ignition promoter.^8,9 Even with the use of a secondary fuel, the flame propagation speed of NH₃-air mixtures remain relatively low.^6,10,11 This can lead to significant concentrations of unburned NH₃ in the exhaust gas (NH₃ slip), particularly for premixed combustion strategies.^11,12 Tyrewala et al. ¹³ carried out investigations on a single-cylinder Cummins engine with NH₃ port-fuel injection (PFI) and diesel direct injection and observed NH₃ slip levels of up to 10,000 ppm. However, they demonstrated that optimization of the air-fuel equivalence ratio (λ) and diesel injection strategy can reduce NH₃ slip to approximately 6000 ppm.

Nitrogen oxide (NO_x) emissions can be substantially higher than in conventional engines, due to fuel-derived NO_x originating from the nitrogen contained in the NH₃ molecule. ¹⁴ Reported NO_x concentrations typically range between 1000 and 6000 ppm, with higher ammonia fuel shares further increasing NO_x emissions.^3,15–19 Moreover, NO_x formation strongly depends on the air-fuel equivalence ratio with a maximum observed under slightly lean combustion conditions.^3,17,18 The fuel-bound nitrogen content can also facilitate nitrous-oxide (N₂O) formation, which can diminish ammonia’s greenhouse-gas benefits, as the 100-year global warming potential (GWP₁₀₀) of N₂O is 273 times higher than that of CO₂.^20–22 Engine-out N₂O concentrations between 10 and 50 ppm are typically reported,^13,18 while some studies report values up to 90 ppm.^16,23 Although complete elimination of engine-out N₂O in 4-stroke NH₃ engines has not yet been demonstrated, strategies have been proposed to effectively control its formation. Operation at near-stoichiometric conditions has been shown to be effective in reducing N₂O.^11,24

As recently reported by WinGD, it has been demonstrated that 2-stroke, low-speed NH₃-diesel marine engines can be operated at high-load with engine-out NH₃ slip below 10 ppm, N₂O emissions below 3 ppm and NO_x emissions lower than those observed during diesel-only operation. ²⁵

However, to achieve compliance with modern emission regulations for 4-stroke NH₃ engines, it is essential to consider both combustion optimization and exhaust gas aftertreatment (EAT) solutions. Recent studies have therefore examined different aftertreatment concepts for their use in NH₃-fueled combustion engines. Selective catalytic reduction (SCR) is a well-established technology for catalytic NO_x conversion (DeNO_x). Lampkowski et al. ¹⁸ investigated copper- (Cu) and vanadium-tungsten-titanium- (VWT) SCR catalysts under synthetic gas conditions based on engine-out emission data derived from an NH₃-fueled passive pre-chamber SI engine. They demonstrated that the Cu-SCR exhibits advantages in terms of DeNO_x performance compared to the VWT-SCR. However, both technologies showed limited N₂O reduction activity (DeN₂O). Due to non-selective NH₃ oxidation, even increased N₂O concentrations at the SCR outlet were observed, particularly for the VWT-SCR.

Cano Blanco et al. ²⁶ investigated iron-based (Fe-zeolite) SCR catalysts, which also represent a state-of-the-art technology for catalytic DeNO_x. These catalysts enable simultaneous DeNO_x and DeN₂O. However, efficient DeN₂O requires high temperatures, typically above 450°C. High H₂O concentrations were found to further increase the required temperature levels, whereas the presence of NH₃ in the gas composition can promote N₂O conversion by acting as a reducing agent. At the same time, excessive NH₃ should be avoided, as it can also lead to inhibition effects and thus lower DeN₂O activity.

Besides Fe-based catalysts, rhodium-containing (Rh) catalysts can also provide high N₂O conversion efficiency with light-off temperatures below 300°C. ²⁷ However, due to rhodium being one of the scarcest platinum group metals (PGMs), its use in EAT applications for NH₃ combustion engines is economically challenging.

Ammonia slip catalysts (ASC) are a well-established approach to oxidize excess NH₃, primarily originating from overdosing of urea in conventional SCR applications for diesel engines. For NH₃-fueled engines, an ASC can also be applied downstream of the SCR to treat remaining NH₃ slip. However, when using such catalysts, especially those with PGM-loading, it must be considered that both NO and N₂O can be formed during NH₃ oxidation as a result of inherent selectivity of the catalyst.^17,28–30

Recent studies have already investigated different aftertreatment configurations under real NH₃ engine operating conditions. Xiang et al. ³¹ evaluated the emission characteristics of a four-cylinder NH₃-fueled diesel-pilot-ignited engine combined with a platinum-palladium (Pt/Pd) diesel oxidation catalyst (DOC) and a Cu-based SCR aftertreatment setup. While NH₃ conversion up to 100% was achieved at engine loads above 50%, significant N₂O formation across the DOC up to 2500 ppm was observed within this load-range. At low engine load, exhaust temperatures below 200°C limited NH₃ conversion to approximately 10%. At 50% engine load, with exhaust temperatures around 270°C, NO formation was additionally observed. The SCR system provided high DeNO_x performance at low and medium engine loads, whereas NO_x conversion efficiency at high load was reduced due to elevated NO_x inlet concentrations. DeN₂O activity was limited to a maximum conversion efficiency of 45%.

Kuta et al. ³² conducted similar experiments on an NH₃-fueled diesel-pilot-ignited single-cylinder engine with NH₃-PFI in combination with a V-based SCR catalyst. Increasing engine load raised exhaust gas temperatures, thereby improving DeNO_x. However, NH₃ conversion performance was limited due to high engine-out NH₃ concentrations.

Tyrewala et al. ³³ investigated a Cu-SCR + ASC system under high engine load conditions in a single-fired cylinder of a multi-cylinder NH₃-fueled diesel-pilot-ignited engine, focusing on the influence of the air-fuel equivalence ratio (λ) on engine-out emissions and EAT performance. NO_x was effectively removed across a wide λ range, with conversion efficiencies of up to 100%. Catalytic NH₃ conversion increased under lean combustion conditions due to higher oxygen availability for NH₃ oxidation across the ASC. At λ = 1.4, an ammonia-to-NO_x-ratio (ANR) of approximately 1.5 was achieved, enabling favorable simultaneous NO_x and NH₃ conversion using the SCR. N₂O formation across the SCR was also observed and attributed to oxidation reactions consuming NH₃ with a high selectivity toward N₂O.

This work investigates two different aftertreatment concepts on an NH₃-fueled diesel-pilot-ignited single-cylinder engine over a wide range of engine loads and under ammonia energy shares of up to 95%. A particular focus is placed on the combination of combustion management and aftertreatment strategy in order to achieve EU Stage V-compliant emission levels. Furthermore, the influence of additional NH₃ dosing into the exhaust stream on EAT performance is examined.

Experimental setup

Test bench and engine setup

The experimental investigations in this study were conducted on a medium-duty engine test bench equipped with a single-cylinder CI engine (SCE) operating in NH₃-diesel dual-fuel mode with NH₃ PFI and direct-injected diesel pilot injection (Figure 1). The engine is coupled to an AVL 220 kW dynamometer, which absorbs the engine’s power output. To enable charged operation, the test bench is supplied with externally compressed air from a screw compressor, allowing charge-air pressures up to 4 bar absolute. Furthermore, the intake air can be temperature-conditioned between 20°C and 100°C. Both, the cooling water and lubricating oil systems are also capable of full temperature conditioning.

Figure 1.

Schematic representation of the single cylinder engine test bench setup.

The diesel fuel system consists of an AVL 733S fuel meter and AVL 753C fuel conditioning system, enabling the measurement of fuel consumption and variation in fuel temperature. Additionally, the high-pressure diesel fuel system is equipped with a Bosch CP4 common rail high-pressure pump and a Bosch CRIN2-16 common rail injector for commercial vehicles.

The ammonia fuel module is installed in a safety cabinet that is continuously ventilated to safely remove any ammonia in case of leakage. Liquid ammonia is stored in two pressurized gas cylinders at approximately 8 bar absolute. To ensure that the ammonia is in gaseous state for port fuel injection, it passes through an evaporator after leaving the gas cylinders. A dome pressure regulator is used to control the injection pressure, allowing NH₃ pressures between 1 and 10 bar absolute. Ammonia fuel consumption is monitored using a Siemens Sitrans Mass 2100 Coriolis measurement system. Finally, the gaseous ammonia is sequentially injected into the engine’s air system with a Bosch large engine gas valve (LEGV) during intake valve opening period.

To maintain stable and reproducible engine conditions, all quasi-stationary values such as pressure and temperature of the various mediums are monitored by a measurement data acquisition and controlled by a LabView-based programmable logic controlling system. To enable direct intervention in the combustion process, an open-access research engine control unit is installed, allowing for the adjustment of engine parameters such as diesel injection timing, injection pressure, and number of injections per cycle.

For the analysis and monitoring of the combustion process a Kistler flush-mounted piezoelectric pressure sensor (type 6041C) and a KiBox2 combustion analysis system were used. The in-cylinder pressure is measured and correlated with the crank angle, achieving a resolution of 0.1 °CA. Using different calculation tools, implemented in the indication system, parameters such as indicated mean effective pressure (IMEP), peak pressure as well as combustion phase and ignition delay can be determined. The calculation of the mean values is based on 250 consecutive cycles.

The SCE is based on the DEUTZ TCD 5.2 diesel engine, which is primarily used for non-road mobile machinery (NRMM) applications. Relevant engine parameters of the SCE are shown in Table 1.

Table 1.

Research engine parameters.

1 Cylinder, 4 stroke diesel
Rated speed	2200 rpm
Rated power	45 kW
Bore	110 mm
Stroke	136 mm
Engine displacement	1.29 l
Compression ratio	16.4:1
Intake air system	External charging

Aftertreatment (EAT) testing environment and exhaust gas analysis

To evaluate different catalyst configurations on the SCE, a comprehensive aftertreatment testing environment was integrated into the test bench. A schematic overview of the setup is shown in Figure 2. As is typical for single-cylinder test benches, the SCE exhaust is routed into a damping tank, which reduces pressure pulsations, enabling the downstream exhaust flap to generate stable backpressure, similar to the turbine of the turbocharger on full-cylinder engines.

Figure 2.

Schematic representation of the aftertreatment testing environment installed on the SCE test bench.

The testing environment is designed for the investigation of low-pressure aftertreatment concepts, with the EAT system located downstream of the exhaust flap. From this point, the system is divided into a bypass line and the dedicated aftertreatment path. The bypass is implemented to allow exclusion of the EAT system from operation when required, for example during specific combustion development under uncommon exhaust conditions, in order to avoid exposing the EAT system to potentially critical conditions. During normal operation with the EAT system active, the bypass is fully closed and the entire exhaust gas flow is directed through the aftertreatment system.

The aftertreatment path consists of two identical sections. Section 1 includes the first dosing position for introducing a reducing agent into the exhaust stream, when required for SCR operation. Instead of using AdBlue, gaseous NH₃ from the fuel system is injected, as it is readily available and avoids the need for an additional urea-water solution system. The NH₃ dosing is controlled and monitored using a mass flow controller (MFC), which was factory-calibrated for NH₃. A static mixer ensures homogenization of ammonia and exhaust gas prior to entering Reactor 1. This reactor contains two identical catalyst slots, each designed for 150 × 150 mm substrates with a length of 300 mm.

Section 2 mirrors the design of Section 1 and enables the implementation of dual-reactor and dual-dosing concepts.

To achieve various temperature gradients between the engine’s exhaust port and the first catalyst, the whole system can be externally heated.

For comprehensive emission analysis, five exhaust sampling locations are provided: one in the raw exhaust upstream of the EAT system and one downstream of each of the four catalyst slots. These sampling points are labeled I-V in Figure 2. At all five locations, the exhaust gas composition can be measured using an MKS MultiGas 2030G Fourier-transform-infrared (FT-IR) analyzer, an ABB Fidas24 flame-ionization detector (FID) for unburned hydrocarbons, an ABB Magnos28 paramagnetic oxygen analyzer (PMA) and an V&F Hsense electron ionization mass spectrometer (EI-MS) for determining the H₂ concentration.

Methodology

The aim of this study was to develop a holistic operating strategy for ammonia-diesel dual-fuel engines that integrates combustion management and aftertreatment to achieve stable and efficient operation while simultaneously minimizing end-of-pipe emissions. The EU Stage V emission standard for stationary engines with a rated power of ≥560 kW was selected as the emission target, with a focus on gaseous pollutants. Particulate emissions were not considered in this work, but will be addressed in future publications. The targeted emission limits are listed in Table 2.

Table 2.

Targeted EU Stage V (≥560 kW) emission limits (if no limit: as low as possible).

Pollutant	Emission limit	Targeted
NOx/g/kWh	0.67	Yes
CO/g/kWh	3.5	Yes
HC/g/kWh	0.19	Yes
NH3/ppm	10	Yes
PM/g/kWh	0.035	No
PN/-	1 × 1012	No
N2O/ppm	No limit	-
H2/ppm	No limit	-
HCN/ppm	No limit	-

To identify suitable engine settings for the aftertreatment concept evaluation, the combustion process was investigated first. Three specific parameter variations for one representative load point are discussed in detail before a broader overview of the entire engine load map is provided. Based on the developed combustion strategy, the performance of two aftertreatment concepts, consisting of SCR and ASC technologies, is evaluated. Particular attention is given to the influence of additional NH₃ dosing on the effectiveness of the aftertreatment concept, as well as its impact on the overall ammonia consumption and thus system efficiency. All catalyst samples were provided by Hug Engineering AG, Elsau, Switzerland.

Combustion strategy

During the early stages of combustion investigations, parameters were identified to characterize the combustion strategy with respect to achieving optimized conditions for the EAT system. A key parameter for assessing the suitability of a given engine setting was the engine-out ammonia-NO_x-ratio (ANR_engine-out), defined as the molar (flow) ratio of NH₃ to NO_x in the raw exhaust gas prior to the EAT system:

$${\mathrm{ANR}}_{\mathrm{engine}-\mathrm{out}}=\frac{{\stackrel{·}{n}}_{{\mathrm{NH}}_3,\mathrm{raw}}}{{\stackrel{·}{n}}_{{\mathrm{NO}}_x,\mathrm{raw}}}$$

(1)

By managing the combustion such that ANR_engine-out remains within a window around 1, favorable conditions for the EAT system can be expected. This approach reduces the risk of excessive NH₃ slip reaching the ASC, while still providing a sufficient NH₃ fraction to limit the required additional NH₃ dosing for DeNO_x. Based on this consideration, two engine operating modes were defined:

• Engine Mode A: ANR_engine-out > 1

• Engine Mode B: ANR_engine-out < 1

Figure 3 provides an overview of the basic approach applied for the combustion process in this study, illustrated by two representative operating points corresponding to Engine Mode A and Engine Mode B. Cylinder pressure traces are shown on the left, while the rate of heat release is presented on the right. In addition, injector energization profiles for both the NH₃ PFI injector and the diesel direct injector are visualized.

Figure 3.

Cylinder pressure traces and rates of heat release for Engine Mode A and Engine Mode B representing the basic combustion strategy approach.

Ammonia is injected into the air system during the intake valve opening phase, with the start of energization (−230 °CAaTDC) kept constant throughout all experiments. The diesel injection strategy consists of a single pilot injection, with the start of diesel injection (SODI) ranging between −42 and −30 °CAaTDC. SODI values outside this range were only applied at the boundaries of specific parameter sweeps.

Engine Mode A and B differ in terms of diesel pilot fuel share and SODI. Engine Mode B operates with a slightly higher diesel share, which requires earlier SODI to maintain a constant MFB₅₀ (50% mass fraction burned, i.e. center of combustion). The corresponding diesel shares required for a specific operating condition are discussed in the Results section.

Besides ANR_engine-out control, attention was also given to minimizing N₂O formation in the raw exhaust gas, thereby reducing the demand on the DeN₂O aftertreatment process. Control of the exhaust gas temperature was also integrated into the combustion strategy to maintain sufficiently high temperatures to support the remaining DeN₂O reactions, while keeping them below critical levels to avoid catalyst degradation. Furthermore, engine efficiency was considered and stable combustion was ensured. Combustion stability was assessed using the coefficient of variation of IMEP (COV_IMEP).

It was observed that variations in the air-fuel equivalence ratio Lambda (λ) as well as the ammonia energy share (AES) have significant impact on ANR_engine-out, N₂O formation, exhaust temperature, engine efficiency, and COV_IMEP. These parameter variations are therefore discussed in detail in this paper.

The air-fuel equivalence ratio λ was calculated using the mass flow rates of air ${\stackrel{·}{m}}_{\mathrm{Air}}$ , NH₃ ${\stackrel{·}{m}}_{{\mathrm{NH}}_3}$ and diesel ${\stackrel{·}{m}}_{\mathrm{Diesel}}$ as well as the fuel’s stoichiometric air requirements $L_{\mathrm{St},{\mathrm{NH}}_3}=6.1$ and $L_{\mathrm{St},\mathrm{Diesel}}=14.595$ :

$$\lambda =\frac{{\stackrel{·}{m}}_{\mathrm{Air}}}{{\stackrel{·}{m}}_{{\mathrm{NH}}_3}·L_{\mathrm{st},{\mathrm{NH}}_3}+{\stackrel{·}{m}}_{\mathrm{Diesel}}·L_{\mathrm{st},\mathrm{Diesel}}}$$

(2)

The ammonia energy share is defined as:

$$\mathrm{AES}=\frac{{\stackrel{·}{m}}_{{\mathrm{NH}}_3}·H_{u,{\mathrm{NH}}_3}}{{\stackrel{·}{m}}_{{\mathrm{NH}}_3}·H_{u,{\mathrm{NH}}_3}+{\stackrel{·}{m}}_{\mathrm{Diesel}}·H_{u,\mathrm{Diesel}}}$$

(3)

with ${\stackrel{·}{m}}_{{\mathrm{NH}}_3}$ , ${\stackrel{·}{m}}_{\mathrm{Diesel}}$ as the respective fuel mass flow rates and $H_{u,{\mathrm{NH}}_3}$ , $H_{u,\mathrm{Diesel}}$ as the respective lower heating values of ammonia (6.1 MJ/kg) and diesel (42.785 MJ/kg).

Both λ and AES were varied at an engine load of 15 bar IMEP and an engine speed of 1500 rpm while maintaining MFB₅₀ constant. This was achieved by adjusting SODI, shifting from later SODI at low λ to earlier SODI at higher λ. Additional details of the engine settings used in these experiments are listed in Table 3.

Table 3.

Engine settings for the three investigated parameter variations.

Speed/rpm	IMEP/bar	AES/%	MFB50/°CAaTDC	SODI/°CAaTDC	pDiesel/bara	pNH3/bara	Lambda λ/-	TCharge-air/°C
1500	15	85	9.0	−30.0 to −45.2	900	5	Varied	50
1500	15	87.5	9.0	−25.3 to −42.9	900	5	Varied	50
1500	15	90	9.0	−24.0 to −37.6	900	5	Varied	50

Aftertreatment strategy

The aftertreatment strategy must account for all EU Stage V-regulated gaseous components as well as N₂O, H₂ and HCN. Under stoichiometric ANR conditions, both NH₃ and NO_x can be effectively converted over an SCR catalyst via the standard SCR reaction:

$$4{\mathrm{NH}}_3+4\mathrm{NO}+O_2\to 4N_2+6H_{2}O$$

(4)

Using an appropriate SCR catalyst, simultaneous DeNO_x and DeN₂O can also be achieved, with NH₃ acting as a reducing agent:

$$2{\mathrm{NH}}_3+2\mathrm{NO}+N_{2}O\to 3N_2+3H_{2}O$$

(5)

In addition, N₂O can be reduced via thermal decomposition of N₂O into N₂ and O₂:

$$2N_{2}O\to 2N_2+O_2$$

(6)

An ASC can be installed downstream of the SCR to oxidize residual NH₃:

$$4{\mathrm{NH}}_3+3O_2\to 2N_2+6H_{2}O$$

(7)

In addition to NH₃ oxidation, CO, HC, and HCN can be converted using an ASC or other oxidation catalysts:

$$\mathrm{CO}+4O_2\to 2{\mathrm{CO}}_2$$

(8)

$$\mathrm{HC}+O_2\to {\mathrm{CO}}_2+H_{2}O$$

(9)

$$4\mathrm{HCN}+5O_2\to 4{\mathrm{CO}}_2+2H_{2}O+2N_2$$

(10)

As hydrogen also contributes to greenhouse gas effects, their release should be limited. ³⁴ H₂ can be oxidized over an oxidation catalyst:

$$2H_2+O_2\to 2H_{2}O$$

(11)

It must be considered that most of these reaction pathways require a certain amount of oxygen, indicating that these mechanisms are only applicable to engine operation under over-stoichiometric combustion conditions.

For this study, two low-pressure aftertreatment concepts installed downstream of the exhaust gas flap were evaluated (see Figures 2 and 4). Both concepts utilize two reactors, with NH₃ dosing applied at position 1 upstream of Reactor 1. The second dosing position was not utilized in this study.

Figure 4.

Schematic representation of the two tested aftertreatment concepts on the SCE.

Concept 1 features two similar sized N₂O-SCR substrates installed in Slot 1, followed by a standard SCR catalyst in Slot 2. An ASC is mounted in Slot 3, while Slot 4 remains empty.

Concept 2 employs the same catalyst technologies but omits the standard SCR catalyst in Slot 2 reducing system cost, complexity and backpressure. Details of the installed catalysts are summarized in Table 4.

Table 4.

Properties of the catalyst samples used for the aftertreatment concepts.

Type	Length/mm	Cross section/mm × mm	Cell density/cpsi
N2O-SCR	2 × 100	150 × 150	200
SCR	300	150 × 150	150
ASC	200	150 × 150	150

The NH₃ dosing at position 1 was adjusted according to the NH₃ requirements for the DeNO_x and DeN₂O processes. To assess the system performance, the conditions upstream of Reactor 1 must account for the additionally dosed NH₃. Accordingly, the dosed NH₃ molar flow rate can be calculated from the NH₃ mass flow rate provided by the MFC ${\stackrel{·}{m}}_{{\mathrm{NH}}_3,\mathrm{dos}}$ and the molar mass of NH₃ $M_{{\mathrm{NH}}_3}$ :

$${\stackrel{·}{n}}_{{\mathrm{NH}}_3,\mathrm{dos}}=\frac{{\stackrel{·}{m}}_{{\mathrm{NH}}_3,\mathrm{dos}}}{M_{{\mathrm{NH}}_3}}$$

(12)

Furthermore, the effective ammonia-NO_x-ratio considering the additionally dosed NH₃ is calculated as:

$${\mathrm{ANR}}_{\mathrm{EAT}}=\frac{{\stackrel{·}{n}}_{{\mathrm{NH}}_3,\mathrm{raw}}+{\stackrel{·}{n}}_{{\mathrm{NH}}_3,\mathrm{dos}}}{{\stackrel{·}{n}}_{{\mathrm{NO}}_x,\mathrm{raw}}}$$

(13)

Moreover, the amount of NH₃ required for the aftertreatment can be considered part of the fuel consumption, as it is directly drawn from the fuel system, thereby affecting the overall efficiency. By combining the NH₃ consumption for engine operation with the dosing amount, a total indicated efficiency can be determined:

$${\eta }_{i,\mathrm{total}}=\frac{P_i}{{\stackrel{·}{E}}_{{\mathrm{NH}}_3,\mathrm{fuel}}+{\stackrel{·}{E}}_{{\mathrm{NH}}_3,\mathrm{dos}}+{\stackrel{·}{E}}_{\mathrm{Diesel}}}$$

(14)

Here, $P_i$ denotes the indicated power and ${\stackrel{·}{E}}_{{\mathrm{NH}}_3,\mathrm{fuel}}$ , ${\stackrel{·}{E}}_{{\mathrm{NH}}_3,\mathrm{dos}}$ , ${\stackrel{·}{E}}_{\mathrm{Diesel}}$ represent the respective energy flow rates, calculated from the product of the corresponding mass flow rates and lower heating values.

To investigate the EAT performance under realistic temperature conditions, an attempt was made to reproduce the expected temperature drop between the engine exhaust outlet and the EAT inlet. Due to the significantly longer exhaust piping system on an SCE test bench compared to a full-cylinder engine, external heating of the entire system was required.

The target temperature at the EAT inlet was defined based on the assumption that, in a full-cylinder engine, exhaust heat losses occur primarily across the turbine of the turbocharger (TC). In stationary engine applications, this represents a reasonable approximation, as the TC is typically located close to the engine, while the piping system is often insulated and, in some case, encapsulated. ³⁵

The temperature drop across the hypothetical turbine was calculated assuming an isentropic expansion with a heat capacity ratio of κ = 1.35:

$$T_{\mathrm{ex},{\mathrm{EAT}}_{,}s}=T_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}·{\left(\frac{p_{\mathrm{ex},\mathrm{EAT}}}{p_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}}\right)}^{\frac{\kappa -1}{\kappa }}$$

(15)

With $T_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}$ and $p_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}$ as the temperature and pressure conditions at the engine exhaust outlet and therefore upstream of the exhaust flap. $T_{\mathrm{ex},{\mathrm{EAT}}_{,}s}$ and $p_{\mathrm{ex},\mathrm{EAT}}$ represent the isentropic exhaust temperature and the exhaust pressure upstream of the EAT system.

The calculated isentropic exhaust gas temperature was then corrected using the isentropic turbine efficiency ${\eta }_{s,\mathrm{TC}}$ of the TC corresponding to the full-cylinder equivalent of the SCE (DEUTZ TCD 5.2):

$$\begin{array}{c}\multicolumn{1}{c}{T_{\mathrm{ex},{\mathrm{EAT}}_{,}r}=T_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}}\\ \multicolumn{1}{c}{-{\eta }_{s,\mathrm{TC}}·(T_{\mathrm{ex},\mathrm{engine}-\mathrm{out}}-T_{\mathrm{ex},\mathrm{EAT},s})}\end{array}$$

(16)

Results and discussion

Based on the methodology described in the previous section, the experimental results are presented in four parts. First, the effects of varying λ and AES on engine performance and raw engine-out emissions are analyzed in detail, followed by an overview of how these findings translate to other engine load conditions. Third, the performance of aftertreatment Concept 1 is evaluated across all presented load conditions. Finally, aftertreatment Concept 2 is assessed for one specific load point, with particular focus on the NH₃ dosing strategy.

Effects of varying λ and AES

Figure 5 shows the effect of varying λ and AES on engine-out NH₃ and NO_x emissions, including the resulting ANR_engine-out. N₂O emissions are depicted below.

Figure 5.

Effect of varying λ and AES on NH₃, NO_x, the resulting ANR_engine-out and N₂O emissions at 15 bar IMEP and 1500 rpm.

The observed NO_x emission levels, are higher compared to conventional combustion engines. This is primarily due to the nitrogen contained in the fuel, which increases the potential for nitrogen oxide formation. NO_x emissions exhibit a λ-dependent trend that is qualitatively similar to that observed in conventional engines. However, the decrease in NO_x with increasing λ is less pronounced compared to conventional fuel combustion, indicating a reduced contribution of thermal NO_x and a stronger dominance of fuel-derived NO_x. However, despite this fuel-NO_x dominance, the influence of λ on NO_x remains visible. At very lean conditions, NO_x emissions decrease due to the dilution effect of excess air, which lowers combustion temperatures and suppresses thermal NO_x formation. ³⁶ At stoichiometric and rich λ conditions, NO_x formation is generally decreased due to limited oxygen availability. ³⁷ The NO_x peak occurs between λ = 1.3 and λ = 1.4. Varying AES within this specific range has no significant effect on NO_x.

Within this NO_x-peak window, NH₃ slip is at a comparable level, resulting in approximately stoichiometric ANR_engine-out. Reducing λ leads to an increase in unburned NH₃, likely due to locally insufficient oxygen for complete combustion. It should be noted that ammonia-diesel combustion is not a fully homogeneous process. Local fuel-rich regions around the diesel spray require a globally sufficient λ to achieve high combustion efficiency of both NH₃ and diesel. Under leaner conditions, NH₃ also increases, probably due to flame quenching effects associated with lower combustion temperatures. Additionally, adjustments of SODI, which are required to maintain a constant MFB₅₀ when varying λ, influence ignition delay, combustion duration and consequently NH₃ slip. For a detailed discussion of the impact of varying SODI during a λ variation, it is referred to a previous study conducted on the same test bench. ¹⁷

By varying λ, NO_x and NH₃ and therefore ANR_engine-out can be effectively controlled. As reported by Kröcher, ³⁸ the optimal ANR for high SCR performance depends on the catalyst temperature. At higher temperatures, NH₃ oxidation can occur, which necessitates a certain NH₃ excess to achieve full DeNO_x. Based on these considerations, three distinct ANR_engine-out operating regions with respect to aftertreatment suitability were defined for this study:

• ANR < 1: 0.0 < ANR_engine-out < 1.0

• ANR > 1: 1.0 < ANR_engine-out < 1.5

• ANR >> 1: ANR_engine-out > 1.5

Within the ANR < 1 window, high aftertreatment performance can be expected if the NH₃ required for DeNO_x is supplied by additional dosing into the exhaust gas stream.

In the ANR > 1 window, excess NH₃ may already be critical, depending on the oxidation activity of the SCR catalyst. If a significant portion of NH₃ is oxidized within the SCR, the amount entering the ASC may remain manageable, meaning that this operating condition may still be tolerable. When using SCR catalysts with high oxidation activity, small amounts of additional NH₃ dosing may even be required.

The ANR >> 1 window is likely unfavorable for combined engine and aftertreatment operation. At ANR_engine-out > 1.5, the ASC is expected to experience severe NH₃ excess, with untreated NH₃ leaving the system. Moreover, substantial formation of additional NO or N₂O due to unselective NH₃ oxidation is anticipated.

Varying AES has also a major impact on unburned NH₃ emissions. Compared to 90% AES, the additional 2.5%–5% ignition energy provided by diesel at 87.5% and 85% AES enables more complete ammonia combustion. Consequently, ANR_engine-out can also be effectively controlled by adjusting AES. By modifying the ammonia fraction, the engine operating mode within the defined ANR_engine-out windows can be selected. For this specific engine load point, 90% AES was selected as Engine Mode A (ANR_engine-out > 1), whereas 85% AES corresponds to Engine Mode B (ANR_engine-out < 1).

When considering N₂O emissions, it was observed that lower λ levels are beneficial. The higher combustion temperatures at low λ suppress the N₂O formation, shifting nitrogen oxidation pathways toward increased NO formation, or promote the decomposition of N₂O that has already formed. ³⁹ However, λ < 1.1 is not usable due to the resulting unsuitable ANR_engine-out >> 1 conditions. Within the ANR < 1 and ANR > 1 windows, N₂O emissions remain at acceptable levels. Higher AES lead to slightly increased N₂O emissions.

In Figure 6 the effect of varying λ and AES on carbon-based emissions (CO, HC, HCN) as well as unburned H₂ is shown. CO is primarily influenced by λ at stoichiometric and sub-stoichiometric conditions due to limited oxygen availability in the diesel pilot combustion. A similar trend is observed for H₂ emissions, which shows a significant increase under richer conditions. Li et al. ⁴⁰ investigated H₂ formation under NH₃-rich combustion conditions and attributed it to intermediate reactions of nitrogen radicals generated during NH₃ decomposition at elevated temperatures. The produced H₂ cannot undergo further oxidation due to the limited oxygen available at λ values around or below 1.

Figure 6.

Effect of varying λ and AES on CO, HC, HCN & H₂ emissions at 15 bar IMEP and 1500 rpm.

Unburned HC emissions tend to increase slightly under lean conditions, likely due to incomplete reaction progress at lower combustion temperatures and thus extinguishing effects. It can also be observed, that unburned HC emissions can be decreased by increasing AES, presumably because of the reduced diesel fraction.

HCN emissions are on a low level and show no clear trend with respect to λ. Higher AES appear to be beneficial, as HCN levels are generally lower at increased AES, likely due to the reduced diesel share. Formation of HCN during engine combustion has received limited attention to date, as HCN concentrations are typically very low in conventional engines. However, a study by Croner et al. ⁴¹ has shown that HCN can be formed during SCR reactions in premixed methanol (MeOH) engines, particularly when VWT SCR catalysts are used. Formaldehyde (HCHO) can be part of the engine’s raw exhaust gas composition and can react with NH₃ dosed as a reducing agent to form HCN. This effect may be further enhanced when MeOH slip occurs, which is characteristic for premixed combustion strategies. In the exhaust gas stream, MeOH can be partially oxidized to HCHO, thereby increasing the potential for HCN formation.^41,42

Since exhaust gas temperature, indicated engine efficiency and COV_IMEP were also elements of the operating strategy development, their dependance on λ and AES is shown in Figure 7. Exhaust temperatures increase as λ decreases due to the reduced cooling effect of air dilution. Sub-stoichiometric λ operation, however, leads to a temperature drop, which is attributable to poor combustion efficiency, evidenced by high NH₃ slip as well as CO and H₂ emissions (Figures 5 and 6). This behavior is also reflected in the indicated engine efficiency, which is significantly lower below λ = 1. Under very lean conditions, efficiency decreases slightly as well, driven by the lower reactivity of the ammonia-air mixture. This is also reflected in COV_IMEP, which increases under leaner combustion conditions. Consequently, the preferred λ range lies between λ = 1.2–1.5, where engine efficiency is highest, COV_IMEP remains below 2% and exhaust gas temperatures are sufficiently high to ensure good aftertreatment performance, yet not excessively high to risk catalyst degradation. This λ range also corresponds closely to the optimal ANR_engine-out window.

Figure 7.

Effect of varying λ and AES on exhaust gas temperature, indicated engine efficiency and COV_IMEP at 15 bar IMEP and 1500 rpm.

AES shows no significant influence on exhaust gas temperature and its impact on engine efficiency and COV remains limited within the preferred λ range.

Transfer to other engine load conditions

By analyzing the effects of varying λ and AES, optimal operating regions for ANR_engine-out, raw N₂O emissions, exhaust gas temperature, indicated engine efficiency and COV_IMEP were identified. The λ-range between λ = 1.2 and λ = 1.5 provides favorable conditions for all these parameters and is expected to enable an effective coupling between combustion and aftertreatment. By adjusting the fuel share, the desired engine operating mode (ANR_engine-out < 1 or ANR_engine-out > 1) can be selected.

The same investigation approach was applied to additional IMEP levels, resulting in a comprehensive operating strategy that covers a wide range of engine loads. To realize the conditions defined by this strategy, engine parameters were adjusted accordingly for each engine load point. Figure 8 illustrates the selected AES and λ levels over IMEP, along with the corresponding boost and exhaust back pressure settings. In addition, the COV_IMEP was again used as an indicator of combustion stability.

Figure 8.

Engine settings over IMEP at 1500 rpm according to the defined operating strategy.

At low engine loads, AES is limited due to lower combustion temperatures, which can lead to unstable combustion and elevated NH₃ slip. Under these conditions, a higher diesel share is required to stabilize combustion and maintain a suitable ANR_engine-out, as defined in the combustion strategy. At higher engine loads, AES of up to 95% are feasible.

For low-load operation, increasing λ was found to be beneficial to achieve the desired combustion strategy conditions. This is likely related to the higher diesel fraction, which creates locally fuel-rich zones around the diesel spray and thus requires a higher global oxygen excess. Boost and thus back pressure were adjusted accordingly to achieve the desired λ levels. Although at 5 bar IMEP the highest λ is used compared to the other engine loads, the target λ (λ = 1.49) could only be achieved in throttled engine operation due to the low absolute fuel demand at this load.

COV_IMEP slightly increases at lower engine loads but remains below 2.5%, which can be attributed to the stabilizing effect of the higher diesel fuel share.

At 5 bar IMEP, Engine Mode B was not implemented due to injection limitations, particularly with respect to SODI. Operation of Engine Mode B at low load would require a comparatively high diesel share combined with a very early SODI. During initial phases of combustion development, it was observed that such injection settings can lead to diesel spray impingements on the cylinder liner and consequently to combustion instabilities and lubrication oil dilution. Therefore, these engine settings were excluded from further testing.

In Figure 9 the engine-out emission levels for both engine modes are depicted as a function of IMEP. Both NH₃ and NO_x emissions generally increase with engine load. The rise in NO_x is primarily driven by higher heat release and associated increase in combustion temperature. An exception occurs at 20 bar IMEP with 95% AES, where NO_x slightly decrease, likely because λ has already shifted below the NO_x peak region (compare Figure 5).

Figure 9.

Engine-out emissions over IMEP at 1500 rpm according to the defined operating strategy.

Although higher engine loads generally facilitate more complete ammonia combustion, NH₃ slip still tends to increase with IMEP. This is mainly attributable to the lower diesel fuel share used at higher IMEP, which appears to counteract the effects of higher engine load. Nevertheless, suitable ANR_engine-out levels, slightly above or below 1 depending on the selected engine mode, can be achieved across all engine load conditions.

N₂O, H₂ and all carbon-based emissions are higher at low engine load. The reduced combustion temperatures at these conditions can promote N₂O formation and suppress complete oxidation of hydrocarbons and hydrogen. In addition, a comparison of Engine Mode A and Engine Mode B shows that N₂O emissions are higher in Mode A due to higher AES. Conversely, CO, HC, HCN emissions are elevated in Mode B, which is consistent with the higher share of diesel fuel. H₂ Emissions are also higher in Engine Mode B. The underlying mehchanism is not yet fully understood. However, it may be hypothesized that the increased diesel amount leads to higher local oxygen consumption. Under such conditions, temperatures may be sufficient for NH₃ decomposition but limited local oxygen availability inhibits subsequent H₂ oxidation.

Aftertreatment concept 1

Aftertreatment Concept 1 was tested under all defined load conditions in both engine modes. The boundary conditions of these experiments, including exhaust gas temperatures, total system GHSV, the H₂O content of the exhaust gas and the ANR_EAT considering additional NH₃ dosing, are shown in Figure 10.

Figure 10.

Boundary conditions for testing aftertreatment concept 1 over IMEP at 1500 rpm.

Exhaust gas temperatures increase with engine load because of the larger amount of heat released during combustion. Even at 5 bar IMEP, engine-out exhaust temperatures exceed 450°C, indicating favorable conditions, particularly for catalytic N₂O conversion in the downstream EAT system.

The pre-EAT temperature conditions are governed by two factors: the engine-out temperature and the temperature drop across the hypothetical turbine, which was calculated using equations (15) and (16). The temperature delta between engine-out and pre-EAT conditions decreases at lower engine load due to the reduced boost pressure demand and thus lower exhaust back pressure (compare Figure 8), resulting in a smaller enthalpy drop across the turbine. At 5 bar IMEP, the engine was operated in throttled mode, resulting in no temperature drop across the turbine according to the applied calculation approach. Due to the low exhaust mass flow at this operating point, it was not possible to further increase the pre-EAT temperature with the external heating, resulting in a maximum level of approximately 400°C.

GHSV depends on the catalyst substrate volume and increases at higher engine load due to higher exhaust gas mass flow rates. The H₂O content in the exhaust gas is higher than in conventional engines and becomes even more relevant at high load, where high AES is used.

Additional NH₃ dosing was manually adjusted according to the requirements for DeNO_x and DeN₂O. For Engine Mode B, substantial dosing amounts were necessary whereas Engine Mode A required only minor dosing at 5 and 10 bar IMEP. At 15 and 20 bar IMEP, Engine Mode A was directly operable without any additional NH₃ supply. As shown in Figure 10, an ANR_EAT in the range of approximately 1.0 to 1.5 was used, already indicating a significant oxidation activity of the SCR catalysts, as a certain NH₃ excess was necessary to achieve high NO_x and N₂O conversion performance. The used ANR_EAT range directly corresponds to the previously defined ANR > 1 window (see Figure 5).

In Figure 11, the measured end-of-pipe (EoP) emissions (downstream of the ASC) are shown as a function of IMEP and directly compared to the targeted EU Stage V emission limits. For all emission limits specified in g/kWh, the value was converted into ppm, resulting in a limit curve that indicates the maximum permissible pollutant concentration $c_{x,\max ,\mathrm{StageV}}$ . Equation (17) describes this procedure, with x representing the respective pollutant:

$$c_{x,\max ,\mathrm{StageV}}=\frac{{\mathrm{Limit}}_{x,\mathrm{Stage}V}}{\frac{M_x}{{\overline{M}}_E}+\frac{{\stackrel{·}{m}}_E}{P_i}}·0.001$$

(17)

With $M_x$ and ${\overline{M}}_E$ as the molar mass of the pollutant and the mean molar mass of the exhaust gas. The exhaust gas mass flow is described as ${\stackrel{·}{m}}_E$ and $P_i$ corresponds to the indicated engine power.

Figure 11.

End-of-pipe emissions over IMEP at 1500 rpm according to the defined operating strategy using Aftertreatment Concept 1.

For NO_x, it was calculated with the molar mass of NO₂ (46 g/mol) and for unburned HC a mean molar mass of 13.875389 g/mol was used (EU Stage V Annex VII 3.3.1). The limit curves of Engine Mode A and B may differ because each mode operates at a slightly different λ, which results in small variations in the exhaust mass flow rates.

Aftertreatment Concept 1 demonstrates a high overall exhaust gas cleaning performance. NH₃ slip is effectively controlled under nearly all engine loads and in both engine modes. Measured NH₃ concentrations were below 3 ppm for most operating points. Only at 10 bar IMEP in Engine Mode B did NH₃ exceed the 10 ppm limit, reaching approximately 15 ppm. This condition also represents the only case with noticeably elevated NO_x emissions compared to the other operating points, although NO_x still remains within the EU Stage V limit (0.67 g/kWh). For all other engine loads and both engine modes, NO_x concentrations are below 15 ppm (<0.1 g/kWh). For comparison, the proposed EURO VII Heavy-Duty NO_x limit for 2028 is set at 0.2 g/kWh ((EU) 2024/1257), highlighting the high DeNO_x performance of this aftertreatment concept.

DeN₂O performance is also high, especially considering the relatively moderate temperatures at the aftertreatment inlet and the high H₂O content in the engine’s exhaust gas. It is known, that temperatures of up to 500°C are typically required for nearly complete N₂O conversion, especially with high H₂O concentrations. ²⁶ The high conversion efficiency observed here can be attributed to the exothermic nature of the SCR reactions, which can induce substantial temperature increases within the catalyst, particularly under higher NO_x and NH₃ concentrations present in the raw exhaust gas. Furthermore, it can be observed, that both very low and very high engine loads show a slight increase in end-of-pipe N₂O emissions. At low load, this is likely due to the higher N₂O levels in the raw exhaust gas combined with lower exhaust temperatures and thus decreased DeN₂O activity. At high load, the increased N₂O is assumed to result primarily from formation mechanisms associated with the ASC, where N₂O is generated during NH₃ oxidation due to the catalyst’s intrinsic selectivity ²⁹ :

$$4{\mathrm{NH}}_3+4O_2\to 2N_{2}O+6H_{2}O$$

(18)

These effects may be further amplified at high load because of the higher GHSV. At 10 and 15 bar IMEP N₂O concentrations below 1 ppm were achieved. H₂ emissions were effectively reduced to below 10 ppm across all tested engine loads.

Carbon-based emissions can also be effectively treated using this aftertreatment concept. As expected, low engine loads and Engine Mode B exhibit higher CO and HC concentrations, consistent with the trends observed in the raw exhaust gas. Nevertheless, EU Stage V emissions levels are met under all operating conditions. Although, HCN is not regulated under EU Stage V, its high toxicity requires that end-of-pipe emissions remain as low as possible. HCN was reduced to below 1 ppm or fully eliminated.

Aftertreatment Concept 2

After demonstrating the high aftertreatment performance of Concept 1, the next step was to evaluate whether emissions levels below EU Stage V limits can still be achieved when minimizing the EAT system complexity. For this purpose, the standard SCR catalyst was removed. To challenge the reduced system under demanding conditions in terms of GHSV, it was tested at high engine load, where exhaust gas mass flow is elevated.

Figure 12 compares both aftertreatment concepts at 20 bar IMEP in Engine Mode A with 95% AES. The top diagram shows the total GHSV for each concept, highlighting the substantial increase resulting from the removal of the standard SCR catalyst. The lower two diagrams present the measured end-of-pipe emissions and the corresponding conversion rates, calculated relative to the concentrations in the raw engine exhaust gas.

Figure 12.

Comparison of overall GHSV and end-of-pipe emissions at 20 bar IMEP & 1500 rpm with 95% AES (Engine Mode A) for Aftertreatment Concept 1 and Concept 2.

While Concept 1 achieved nearly 100% NH₃ conversion with NH₃ levels below 1 ppm, Concept 2 showed a slight increase in NH₃, but still remained well below the EU Stage V limit. A similar trend is observed for DeNO_x performance, which is slightly reduced in Concept 2 but remains within regulatory limits. This moderate decrease in NH₃ and NO_x conversion is expected, as one complete SCR catalyst was removed, but it demonstrates that the SCR activity of the N₂O-SCR is sufficient to effectively remove NH₃ and NO_x. A comparable behavior is observed for the H₂ emissions.

Notably, end-of-pipe N₂O emissions for Concept 2 are lower than for Concept 1. Since the N₂O-SCR catalyst remained unchanged, the DeN₂O activity of the system is maintained. The reduced N₂O emissions can rather be attributed to the absence of the standard SCR catalyst, which apparently exhibits a significant selectivity toward N₂O formation (compare equation (18)).

CO concentrations at end-of-pipe are low for both concepts. It should be noted that raw CO emissions are already very low under these engine load conditions (see Figure 9), resulting in only minimal change downstream of the EAT system. Full conversion was observed for both HC and HCN in both concepts. It should be noted, that no FID data was available for this data set, and the sum of HC species derived from the FT-IR system was used instead.

After it was shown that emission levels remained below EU Stage V limits at high engine load and high AES even if the standard SCR catalyst is omitted, the influence of additional NH₃ dosing on overall aftertreatment performance was further investigated. To span a wider range of ANR_EAT, the engine was operated in Engine Mode B at 20 bar IMEP with 90% AES, resulting in an initial ANR_engine-out of approximately 0.9. This operation point was selected as the baseline for an NH₃ dosing variation at injection position 1, located upstream of the N₂O-SCR (compare Figures 2 and 4). The results of this variation are presented in Figure 13, where end-of-pipe emissions and overall efficiency are shown as a function of ANR_EAT.

Figure 13.

Effect of varying NH₃ dosing and therefore ANR_EAT on end-of-pipe emissions and total system efficiency at 20 bar IMEP & 1500 rpm with 90% AES (Engine Mode B) using Aftertreatment Concept 2.

The green windows highlight the optimal ANR_EAT operating range. For NH₃ and NO_x, this window is located between ANR_EAT values of approximately 1.2 and 1.3, where both emissions are within the EU Stage V limits. At lower ANR_EAT, a NO_x breakthrough occurs due to underdosing and thus insufficient NH₃ availability, whereas values above 1.3 lead to untreated NH₃ slip and additional NO formation, especially originating from unselective NH₃ oxidation over the ASC:

$$4{\mathrm{NH}}_3+5O_2\to 4\mathrm{NO}+6H_{2}O$$

(19)

When relating these findings to the previously defined ANR_engine-out operating windows (Figures 5 and 9), it becomes evident that these windows remain generally valid but are subject to operating-point-specific constraints. ANR values of up to 1.5 were previously considered tolerable, which was successfully demonstrated at other engine load conditions (see Figures 10 and 11). However, the present high-load operating point exhibits a narrower feasible ANR_EAT range, which can be attributed to the high GHSV, causing NH₃-slip and NO formation to occur already at lower ANR_EAT levels.

Since N₂O is not directly regulated, the feasible ANR_EAT range is wider compared to that of NH₃ and NO_x. Nevertheless, the lowest N₂O emissions are observed at ANR_EAT values between 1.1 and 1.3. As NH₃ acts as a reducing agent for N₂O conversion, lower ANR_EAT result in increasing N₂O breakthrough. Conversely NH₃ overdosing with ANR_EAT above 1.3 can lead to N₂O formation during NH₃ oxidation over the ASC (compare equation (18)). H₂ emissions show only a negligible dependance across ANR_EAT, indicating that oxidation reactions are the dominant pathway for H₂ abatement.

No clear trend with respect to the NH₃ dosing amount was observed for the carbon-based emissions. Unburned HC concentrations show an increase under NH₃ overdosing conditions. It should be noted, that for this specific parameter variation no FID data were available. Therefore, the sum of the FT-IR measured HC molecules had to be used instead.

Finally, the overall indicated system efficiency is compared to the indicated engine efficiency as a function of ANR_EAT. Increasing NH₃ dosing leads to higher fuel consumption, which corresponds to a reduced overall efficiency. However, within the ANR_EAT range preferred for low end-of-pipe emissions for this operating point, the mean absolute efficiency penalty remains below 0.5%.

After evaluating Aftertreatment Concept 1 over a wide range of engine loads and further investigating Aftertreatment Concept 2 with a particular focus on the NH₃ dosing strategy, it became evident that achieving high aftertreatment performance and meeting modern emission regulations requires a certain NH₃ excess upstream of the EAT system. This is primarily due to partial NH₃ oxidation occurring over the SCR catalysts.

When ANR_EAT is controlled within a range of approximately 1.0 to 1.5 both, NH₃ and NO_x can be effectively converted via SCR reactions. In addition, N₂O is efficiently removed, with overall conversion rates between 95% and 100% being achievable for most operating points. Since the presence of NH₃ in the exhaust gas is required in any case, Engine Mode A (ANR_engine-out > 1) represents the preferred engine operating strategy. Allowing a limited amount of NH₃ slip at the engine outlet, engine operation with higher AES is possible, thereby reducing overall GHG emissions. This approach requires, that the ANR_engine-out can be reliably maintained. As demonstrated during the combustion process investigations, such control can be achieved in stationary engine operation through appropriate λ and AES management.

Nevertheless, Engine Mode B (ANR_engine-out < 1) remains relevant, particularly in operating regimes where tighter ANR control might be required (e.g. engine-load transitions). In such cases, controlling ANR via additional NH₃ dosing (ANR_EAT) may be more feasible than relying only on combustion-based control of engine-out emissions. However, the overall GHG reduction potential is limited, as operation at ANR_engine-out < 1 requires higher diesel pilot fuel shares.

Conclusion

In this study, a comprehensive low-emission operating strategy for premixed NH₃-diesel engines was developed by coupling combustion management with exhaust gas aftertreatment (EAT). The goal was to achieve emission levels below the levels defined in the EU Stage V emission standard for stationary engines with power outputs ≥ 560 kW. For this purpose, two different aftertreatment concepts were investigated on a single-cylinder engine test bench. All experiments were conducted at a fixed engine speed of 1500 rpm, which is representative of stationary engine operation in 50 Hz power generation applications.

To identify a suitable combustion strategy compatible with aftertreatment, different engine parameter variations were first conducted on the SCE. Subsequently, both aftertreatment concepts were tested under different engine loads, raw exhaust gas conditions and additional NH₃ dosing levels.

Key-findings

• - λ and AES management enables flexible control of the engine’s raw exhaust gas, providing suitable ANR_engine-out levels, reduced engine-out N₂O emissions, optimized exhaust gas temperatures and high engine efficiency

• - Engine operation is feasible for both ANR_engine-out < 1 and ANR_engine-out > 1

• - The combustion strategy can be applied across a wide range of engine loads, with low-load operation limited to 60% AES and high-load operation allowing AES up to 95%

• - Optimizing ANR is essential for high DeNO_x performance; depending on ANR_engine-out, additional NH₃ dosing may be required

• - NH₃-slip can be controlled with appropriate ANR management (ANR_engine-out + ANR_EAT)

• - Near-complete catalytic N₂O conversion can be realized, when ANR and exhaust gas temperature are controlled accordingly

• - Emissions of NO_x, NH₃, CO, HC were successfully reduced below EU Stage V limits across all tested engine loads

• - ANR_engine-out > 1 is the preferred engine operating strategy, as the optimal ANR for high aftertreatment performance was found to lie between 1.0 and 1.5

This study demonstrates that ammonia engines can meet modern emission regulations when combustion and aftertreatment are effectively coupled, therefore supporting ammonia’s potential as a carbon-free fuel option for decarbonization of the stationary or maritime sector. For full EU Stage V compliance, dynamic cycle-based engine testing and particulate emissions levels must also be considered. In particular, the formation of particulate matter (PM) and particle number (PN) emissions under NH₃-diesel operation requires further investigation. Corresponding experimental investigations have already been conducted on the same test bench and are currently under evaluation. Future work will therefore focus on a detailed analysis of these results.