CFD Simulation-Based Development of a Multi-Platform SCR Aftertreatment System for Heavy-Duty Compression Ignition Engines

Abstract

Combustion processes in compression ignition engines lead to the inevitable generation of nitrogen oxides, which cannot be limited to the currently desired levels just by optimising the in-cylinder processes. Therefore, simulation-based engine development needs to include all engine-related aspects which contribute to tailpipe emissions. Among them, the SCR (selective catalytic reduction) aftertreatment-related processes, such as urea–water solution injection, urea decomposition, mixing, NOx catalytic reduction, and deposits’ formation, are the most challenging, and require as much attention as the processes taking place inside the cylinder. Over the last decade, the urea-SCR aftertreatment systems have evolved from underfloor designs to close-coupled (to the engine) architecture, characterised by the short mixing length. Therefore, they need to be tailor-made for each application. This study presents the CFD-based development of a multi-platform SCR system with a short mixing length for mobile non-road applications, compliant with Stage V NRE-v/c-5 emission standard. It combines multiphase dispersed flow, including wall wetting and urea decomposition kinetic reaction modelling to account for the critical aspects of the SCR system operation. The baseline system’s design was characterised by the severe deposit formation near the mixer’s outlet, which was attributed to the intensive cooling in the mounting area. Moreover, as the simulations suggested, the spray was not appropriately mixed with the surrounding gas in its primary zone. The proposed measures to reduce the wall film formation needed to account for the multi-platform application (ranging from 56 to 130 kW) and large-scale production capability. The performed simulations led to the system design, providing excellent UWS–exhaust gas mixing without a solid deposit formation. The developed system was designed to be manufactured and implemented in large-scale series production.

1. Introduction

The ongoing efforts in the development of combustion-based propulsion systems for automotive applications are primarily focused on pollutant emissions. They include heavy-duty and non-road applications, in which compression ignition engines are the dominant solution. The combustion process in compression ignition engines is inevitably linked with the generation of nitrogen oxides, which cannot be reduced to the desired levels solely by the optimisation of in-cylinder processes. To meet current emission standards, aftertreatment solutions need to be applied. Therefore, simulation-based engine development cannot be limited to the in-cylinder processes; it needs to take into account all engine-related processes from the inlet to the tailpipe. Within that path, SCR aftertreatment processes, taking place in the exhaust systems, are the most challenging, as they include multiphase effects in dispersed flow (urea–water injection, wall wetting), evaporation, and chemistry (urea decomposition and NOx reduction).

Urea-SCR (selective catalytic reduction) has become a principal technology to limit tailpipe NOx emissions. Although the first SCR demonstrator systems were introduced in 1999 [1,2] and, since then, have been successfully applied on a mass scale, the tightening emissions’ limits have caused conventional underfloor designs to evolve into close-coupled (to the engine) tailor-made solutions. The location close to the engine enabled higher temperature operation and faster light-off; as shown by Nishad et al. [3], the urea conversion is significantly influenced by the temperature. However, the close-to-the-engine location shortens the available space for atomisation, evaporation, and mixing. Therefore, the systems need to be designed for a specific vehicle (including engine compartment arrangement), and thus, developing a multi-platform solution is very challenging. This study, using CFD tools, aims to develop an SCR aftertreatment solution for large-scale production, capable of meeting the requirements on many different platforms in the power range from 56 to 130 kW, to meet the Stage V NRE-v/c-5 emission standard.

The crucial element necessary for achieving efficient NOx reduction in urea-SCR is to evenly distribute a urea–water solution (UWS) over the exhaust gas stream, effectively evaporating the water and converting urea into ammonia before the inlet to the catalyst. Thus, this process is strongly dependent on the injection process and spray properties, which determine the interaction between the exhaust gas and urea droplets. This interaction and the subsequent mixing of urea conversion products with the exhaust gases are also strongly dependent on the static mixer and the whole system design. All these aspects must be matched together to reach the desired mixing characteristics, which is especially challenging in multi-platform designs. Due to this reason, the development of selective catalytic reduction (SCR) systems primarily focuses on improving the mixing of urea–water solution (UWS) droplets and the resulting conversion products with the exhaust gas. As it is performed in an iterative manner, considering many geometrical versions of the mixers and the whole system, at the early stage the work is performed using CFD simulations to develop a few promising designs for prototyping and experimental testing on an engine dyno or hot gas test rigs [4,5]. Similar to in-cylinder processes, CFD tools are very important at the design phase [6] and early-stage development to limit experimental space and the number of prototypes. They can limit the extent to which the experimental work is needed. However, further validation is usually necessary to confirm the simulation’s results. The solutions elaborated in the CFD-based development presented in this study were further verified during engine dyno tests. However, those tests were outside the scope of this study.

As the NOx distribution in the exhaust gas stream is uniform, achieving a uniform spatial distribution of ammonia is considered crucial for assessing the performance of the developed SCR systems. The convenient indicator to assess the uniformity of the NH₃ at the inlet to the catalytic reactor is the uniformity index [7]. The same indicator can be used to evaluate the performance of the static mixers, which are introduced in the aftertreatment systems to improve mixing [8]. The effect of the static mixers on the ammonia distribution uniformity can be profound and much stronger than the catalyst length. As shown by Xie et al. [9], for the same mixer, the catalyst length only slightly influenced the NOx conversion efficiency. Wardana et al. [10] investigated the influence of a single-stage static mixer on ammonia uniformity and observed that a static mixer led to a more than twofold increase in the ammonia uniformity index, compared to the bare system (without the mixer). A similar effect from a static mixer on the uniformity index in a close-coupled design was observed in our previous study [11]. As simulations suggested, after removing the mixer, the uniformity index dropped from 85.8% to 60.4%. The considered application was in the same power range, and the mass flow and temperatures at the inlet to the SCR were also in the same range. Despite a different application (passenger cars) and design, the common feature of the SCR systems considered in the previous [11] and this study, i.e., close-coupled architecture with a short mixing length, allows us to expect a similar level of importance for the mixer in a close-coupled heavy-duty application from 56 to 130 kW.

There are different mechanisms for improving the mixing process with the static mixing elements. They can divide, redirect, and join the streams from different areas of the duct, enhance turbulences, or introduce a swirl to increase mixing length. They can also be used to enhance droplet break-up through the splashing mechanism. The latter mechanism can be very efficient as it acts twofold. First, it reduces droplet sizes, and second, it redirects the liquid (child droplets) from its original path in many different directions, improving spatial droplet distribution in the duct. However, the strong interaction of the liquid urea with the mixer surface can lead to the accumulation of a liquid film and deposit formation. Thus, it should be avoided, or the mixer should be designed to prevent those unwanted effects; this is usually achieved by an iterative development of a tailor-made SCR system using CFD simulations and further validation tests.

Enhancing the turbulence in the flow can also improve the mixing performance [12]. Tan et al. [13] numerically studied the effects of blade mixers on urea conversion and subsequent mixing in a tube. They reported that the mixers provided increased turbulence kinetic energy and caused droplet break-up. Together, these two factors lead to improved droplet evaporation rates (and reduced evaporation time) and a more uniform distribution of ammonia in the space. Jang et al. [14] studied the effects of mixers in marine SCR systems, observing intensified turbulence and expanded ammonia production areas. Even though the mixer improved the uniformity of the ammonia distribution at the SCR catalyst inlet, the mean NH₃ concentration did not significantly change. They also noted a temperature decrease resulting from the conversion of urea to ammonia. Park et al. [15] evaluated two types of baffle mixers. In the first design, vanes were employed to generate swirling flow; while the second mixer featured line-type vanes. The first design created a recirculation zone downstream from the mixer, the size and length of which were influenced by the angle of the vanes. All the swirl-type baffle mixers demonstrated superior mixing performance due to large-scale vortices spread over a longer region. Cho et al. [16] also noticed the positive influence of the turbulent vortices developed by a mixer on the uniformity index. They numerically analysed two different types of blade mixers with three different lengths of decomposition pipe placed either downstream or upstream from the mixer. The results indicated a negative impact on NOx conversion efficiency due to reduced retention time. Lu et al. [17] presented the development of an SCR system with baffles and outside blades that created a vortex. Both numerical and experimental tests yielded high uniformity indices of velocity and ammonia distribution at the SCR inlet (98% and 95%, respectively). On-road testing of the mixing device confirmed a low risk of deposit formation at the mixer.

Improvements in achieving a uniform ammonia distribution have also been made using unconventional mixers. Xu et al. [18] conducted experimental investigations on existing and newly developed mixing devices in a heavy-duty SCR system under various conditions. The counter-flow blade mixer provided the highest NOx conversion efficiency. In [5], Millo et al. developed and experimentally tested mixing devices that exhibited highly uniform ammonia distribution before the inlet to the SCR catalytic converter. Two different geometries were considered, both utilising a swirling flow pattern within the injection zone. One generated the swirl using guiding vanes, and the other employed a radial opening. The authors validated the numerical prediction of deposit formation in the subsequent experiments on the hot-gas flow rig. Mehdi et al. [19] performed simulations of various mixers and predicted the highest values of both NH₃ and velocity indices for a double mixer design. In that design, the first mixer utilised a baffled type, while the second mixer employed blades. Zheng et al. [20] experimented with different dual mixer configurations to enhance NOx conversion ratios. Compared to the base model featuring a single mixer, all the dual mixer options selected for the study exhibited improved mixing efficiencies. In [21], Kaźmierski and Kapusta explored the potential of perforated mixers in SCR systems. The perforated mixing tubes were equipped with symmetrical slots to increase the mass flow rate inside the tube. However, the slots could also increase turbulence effects in the primary injection zone. A similar effect was observed in another perforated system, where the cooling effect of droplets impacting the mixer’s elements was demonstrated [22]. Michelin et al. [23] designed and tested compact mixers specifically tailored for light-duty close-coupled SCR systems. These mixers were based on the generation of vortex structures. In another research study, Michelin et al. [24] employed a spiral mixer in a close-coupled configuration to extend the residence time of the UWS droplets. Using the spiral mixer, they achieved a 20% increase in residence time while enhancing the mixing of the spray and improving ammonia uniformity. The researchers identified critical areas where solid deposits tend to form due to the reduced crossflow velocity. Structural modifications were implemented to address this issue and prevent local cooling in these sections. Furthermore, Pang et al. [25] conducted numerical tests to evaluate the effect of spiral and X-type baffle mixers on the performance of a 660 MW unit’s SCR system. Both of these systems produced vortices, enhancing the mixing of NH₃ and the exhaust gas. Rajadurai [26] tested ammonia uniformity and pressure drop using a wire-mesh mixer. The UWS spray, which entered the domain, partially evaporated in the primary mixing zone, while the remaining spray underwent a further transformation into NH₃ and HNCO as it passed through the wire mesh. With this method, there is less chance of ammonia slipping due to the mesh’s high heat capacity, which ensures high uniformity of ammonia distribution and low-urea crystallisation. A 37% improvement in ammonia uniformity and 10–30% NOx reduction was observed for the tested operating conditions. However, the wire mesh led to an increased pressure drop, which could be a drawback in high-load conditions.

The essential factor reducing the performance of urea-mixing systems is solid deposit formation. One of the potential solutions to mitigate solid deposition inside the mixing system is to minimise temperature drops and increase the temperature of the mixer. Masoudi et al. [27] presented an example of an electrically heated mixer with a novel design that significantly reduced solid deposition. This design allowed for the injection of UWS even at exhaust gas temperatures as low as 130 °C without the risk of excessive deposit generation. Zhang et al. [28] conducted an analysis of the impact of blade and baffle mixers placed near the injector in the SCR system. The system was equipped with a perforated diffuser upstream from the catalyst. Baffle-type and double-vortex blade mixers exhibited the highest temperature drops downstream from the injection point, as well as the most severe solid deposition. These temperature drops were primarily caused by direct collisions of droplets with the external walls. However, dense baffle arrangements solved this problem. Kojima et al. [29] successfully mitigated solid deposition in their innovative spiral urea-mixing system. They achieved this by increasing the distance between the droplets and the walls, as well as by enhancing the heat flow around the mixer’s elements.

The importance of the angular position of the blades of a conventional UWS mixer was stressed by Venkatachalam et al. [30], based on numerical simulations. Capetillo et al. [7] presented a Taguchi analysis of the urea-mixer’s design factors: blade numbers, blade angles, and the distance between the injector and the mixer. The observed quantities were the uniformity index of ammonia at the SCR’s inlet, and the generated liquid film’s mass and pressure drops. The greatest influence on the mixer’s performance was observed for the blade angle. Moreover, the reduction in the distance between the mixer and the injector was predicted to reduce the pressure drop. Kapusta et al. [11] demonstrated that different spray properties, e.g., reduced droplet size, might require changes in the orientation of the injector to minimise wetting of the device’s walls. Zheng et al. [8] emphasised the importance of tuning a mixer to match the operating conditions, and Smith et al. [31] studied UWS interaction with low exhaust mass flow conditions. Accumulated wall film pathways on the mixer blade changes were studied for different UWS injection rates. As the injection rate increased, the amount of spray accumulated at the impingement location was reduced, resulting in less local cooling and a lower risk of deposit formation. Spray distributed over a wider area, and this would enhance the evaporation process. In addition, Dörnhöfer et al. [32] examined the interaction of UWS spray on hydrophobic and hydrophilic surfaces, aiming to demonstrate the significance of hydrophobic surface properties in mixer blade design and reducing solid deposit formation. Hydrophobic surfaces exhibited repellent behaviour towards impinged droplets, causing them to be carried away by the crossflow without forming a liquid film on the impact surface. These repelled droplets mixed with the crossflow and evaporated rapidly. In contrast, the spray impacted and accumulated on the hydrophilic surface and mixer blades. The researchers examined the primary break-up of impinged droplets on both clean and deposited mixer blades to investigate the initial formation of wall film and subsequent accumulation of solid deposits. Droplets that accumulated on the mixer blades tended to migrate towards the trailing edge and detach under the influence of local shear flow.

Additionally, urea-mixing systems contribute to the backpressure by obstructing the flow of the exhaust gas. This aspect is particularly relevant in the case of complex mixers. The counter-flow mixer tested by Xu et al. [18] exhibited the lowest pressure drop from the considered designs. However, double mixers equipped with blades are mostly reported to increase pressure drops. Fu et al. [33] presented the effect of the geometrical shape of a mixing device, observing a trade-off between mixing performance at short distances and the pressure drop. Double-blade mixers resulted in the highest pressure drop and exhibited relatively low ammonia generation. However, they demonstrated high uniformity indices of NH₃ and velocity at the catalyst’s inlet and the lowest ammonia slip. The counter-flow dual mixer presented by Zheng et al. [20] demonstrated reliable performance with only a marginal increase in pressure drop values.

The study presented here shows the early-stage CFD-based development of a multi-platform SCR system, which accounts for all these aspects while maintaining a compact design and short mixing length intended for mobile applications. The study is focused on urea–water solution injection and mixing urea conversion products with exhaust gases. These processes were primarily evaluated in terms of NH₃ distribution uniformity, wall film, and deposit formation in the SCR system. The mixer that was developed within the study was selected for implementation on a large scale.

2. Materials and Methods

The numerical flow analysis was conducted to predict the presence of liquid film or solid deposits in the mixing systems and determine the velocity fields. For this purpose, the commercial software AVL Fire™ was employed, utilising the Reynolds-averaged Navier–Stokes (RANS) approach and the k-ζ-f closure model [34]. The compound wall treatment was implemented as a wall function to mitigate the impact of cell thickness within the boundary layer [35].

2.1. SCR Mixing Devices

Effective mixing of exhaust gases with the UWS in front of the SCR catalyst is an essential factor in achieving a higher level of NOx reduction. This mixing process depends on how well the static mixer system is designed. Using an optimised mixer, injected liquid can be thoroughly mixed with exhaust gas, effectively accelerating the pyrolysis process of the UWS. This results in enhanced catalyst conversion efficiency and decreased deposit formation.

In the current analysis, the baseline model was Model 1, which is used in heavy-duty CNH tractors. Improvements were made in geometry to enhance the mixing process. Seven mixer variants were analysed to check the mixing characteristics of UWS interaction in the after-treatment systems. Model 1 consisted of a simple set of blades arranged circumferentially at the outlet of the mixing system. Improved models had two sets of blades and a guiding vane(s) that redistribute the flow along the mixer and enhance the mixing of UWS with the exhaust gases. Figure 1 shows an improved mixer model with an additional guiding vane and blade representation. The inner blades of the mixer create a swirl in the flow where the primary interaction of the spray with the exhaust gases occurs. These pre-mixed products are further mixed with secondary flow generated by the outer blades in the middle of the mixer system. Counter-flow generated by the inner and outer blades could enhance the mixing process. Vanes guide the flow along the walls of the mixer system and avoid direct crossflow influence on the near-nozzle spray.

The model’s evolution is presented in Figure 2. The number of outer blades was the same for all the new models (Models 2 to 7), although their length changed from model to model. Model 2 had twelve inner blades without any guiding vanes. Model 3 had seven inner blades and two guiding vanes. Model 4 had smaller guide vanes than Model 3 but with the same inner blades. The inner blade structure was modified for Model 5 to avoid crossflow influence on the spray at the near-nozzle section and potential wall wetting. Additionally, Model 5 was equipped with a convex plate (later referred to as a “cap”) at the location where the secondary mixing occurs in the mixer to redirect the centre stream away from the mixer’s axis and increase turbulences. Models 6 and 7 had seven inner blades and one guiding vane. The only difference between these two models was the span of the outer blades.

The number of inner blades, the width of the outer blades and the guiding vane location changed over model development (see details in

Table 1. Mixer blade and guide vane information.

Model Number	Span of the Outer Vane	No. of Inner Blades	No. of Guide Vanes
1	no vanes	6	0
2	11.95	12	0
3	12.48	7	2
4	15.26	7	2
5	5.90	6	0
6	15.11	7	1
7	12.35	7	1

).

The complete model consisted of a diesel oxidation catalyst (DOC) and a mixing system where the UWS was injected for mixing and further evaporation with the exhaust gases. Dummy volumes were created at the inlet and outlet of the geometry to increase the reliability of CFD simulations by capturing the complex flow physics near boundaries and interfaces. Details about the model are presented in Figure 3. A polyhedral mesh was generated for these models. The mesh settings are mentioned in

Table 2. Mesh properties, general and in specific areas.

Mesh Property	Value
Surface cell size: general	5 mm
Surface cell size: mixer	3 mm
Volume cell size: general	8 mm
Volume cell size: initial spray zone	1 mm
Number of boundary layers	4
Total boundary layer thickness	2.5 mm
Boundary layer compression	1.4
Minimum cell size	0.3 mm

. Mesh settings were kept constant for all the models. A refined mesh was created at the critical locations of the geometry. Refinements allowed for a more accurate representation of geometric features, gradients, and flow phenomena, leading to more accurate and reliable results in the corresponding areas of the simulation. A 1 mm cell size was considered at the near nozzle region to capture the spray droplet changes after injection. To account for spray-wall interactions and subsequent wall film formation in the primary and secondary mixing zones, a refined mesh size of 3 mm was used for the mixer. Depending on the complexity of the mixer blades, the generated volume meshes had 0.46 to 0.76 million cells.

2.2. Operating and Boundary Conditions

The composition of the gas that entered the SCR system corresponded to diesel exhaust gas, and it consisted of nitrogen, oxygen, water, carbon dioxide, and nitric oxide (NO). The mass fractions of all the species were derived from [36], and they are provided in

Table 3. Composition of exhaust gas entering SCR mixing device.

Species	Mass Fraction
N2	75.248%
O2	15%
CO2	7.1%
H2O	2.6%
NO	0.052%

.

Two sets of operating conditions were defined (see

Table 4. Considered operating conditions.

Operating Point	Inlet Mass Flow Rate	Inlet Temperature	Outlet Static Pressure *
OP1	80 kg/h	250 °C	110 kPa
OP2	350 kg/h	300 °C	137.88 kPa

* Calculated based on expected pressure drop caused by the part of the system not considered in simulations.

) to represent relatively low and moderate engine loads of a mid-size passenger car. Nitric oxide (NO) was specifically used to address the presence of nitrogen oxides.

The outlet boundary conditions were determined by the static pressure, which was defined as the sum of the reference atmospheric pressure (101.33 kPa) and the pressure drop caused by the part of the system which was not included in the model (SCR catalyst, piping, muffler). For operating point 1 (OP1), an outlet pressure of 110 kPa (equivalent to the pressure drop of 8.68 kPa) was predetermined based on internal data from similar industrial applications. In the other case (OP2), the muffler pressure drop (${∆p}_{out}$) was estimated using a second-order polynomial equation (Equation (1)):

$${∆p}_{out}=b_2{u}^2+b_{1}u+b_0$$

(1)

where $b_0$, $b_1$, and $b_2$—polynomial coefficients; $u$—average gas velocity. When the gas velocity reaches zero, no pressure drop was expected, and hence, $b_0$ was set to 0. Considering that the primary factor contributing to the overall pressure drop in the whole exhaust system is the pressure drop caused by the SCR catalytic converter, the coefficient $b_2$ was assumed to correspond with the part of the Forchheimer’s pressure drop formula [37] (Equation (2)):

$$b_2=\zeta ·{\displaystyle \frac{{\rho }_g}{2}}·l_0$$

(2)

where $\zeta$—inertial loss coefficient, ${\rho }_g$—density of the flue gas, and $l_0$—unitary length of $1 \mathrm{m}$.

The inertial loss coefficient $\zeta =500{\displaystyle \frac{1}{m}}$ was assumed according to [21]. The gas density for OP1 was evaluated using the ideal-gas law (Equation (3)), and the obtained value was used in the calculation of $b_2$ (Equation (2)).

$${\rho }_g={\displaystyle \frac{p_{out}}{R_{g}T}}$$

(3)

where $p_{out}$—outlet static pressure (110 kPa), $R_g$—gas constant of the flue gas, and $T$—temperature. The average velocity of the flue gas was then obtained using Equation (4):

$$u_g={\displaystyle \frac{{\dot{m}}_g}{{\rho }_{g}A}}$$

(4)

where ${\dot{m}}_g$—mass flow rate of the flue gas and $A$—cross-section area of the pipe. The coefficient $b_1$ was determined using Equation (5):

$$b_1={\displaystyle \frac{∆p_{out,Case1}}{u_{g,Case1}}}-b_2·u_{g,Case1}$$

(5)

Next, the outlet pressure for OP2 was determined iteratively using Equations (3) and (4) and Equation (1). The calculation resulted in an outlet pressure of approximately 137.88 kPa, corresponding to a backpressure of 36.56 kPa (see Figure 4).

The inlet boundary conditions were specified by the gas mass flow rate and turbulence properties. The integral length scale was set to 10% of the inlet diameter [38]. The turbulence kinetic energy of the inlet flow was calculated using Equation (6):

$$k={\displaystyle \frac{3}{2}}·{\left({\displaystyle \frac{{\dot{m}}_g}{{\rho }_{g}A}}·I\right)}^2$$

(6)

where $I$—turbulence intensity.

To compensate for the flow disturbances caused by duct elements upstream of the DOC (which were not part of the computational domain—see Section 2.1 SCR mixing devices), a relatively high turbulence intensity value of 0.12 was assumed for the incoming flow based on estimates presented in [39]. This value was related to the internal flow in the ducts at intermediate and high Reynolds numbers. Considering the multiple bends and diameter changes, as well as the substantial length of the actual exhaust aftertreatment piping used in cars, the selected high value from the proposed range (0.02–0.12) [36] enabled a close approximation of the flue’s gas flow turbulence parameters at the inlet of the simulated system. Consequently, the inlet turbulent kinetic energy was determined to be 0.037 m²/s² and 0.542 m²/s² for flue gas mass flow rates of 80 kg/h and 350 kg/h, respectively.

The considered SCR mixing systems were thermally insulated with a silica cover. In the numerical models, the walls were treated as adiabatic. However, the outlet duct of the mixing system consisted of an uninsulated metal tube connected to the upstream duct elements (which were not considered) via a V-clamp (Figure 5). As a result, the uninsulated duct section was the most critical area of the system, particularly susceptible to the formation of liquid film and solid deposits due to meaningful heat transfer and temperature reductions near the walls.

The uninsulated walls were characterised by the heat flux in both the thin and thick steel walls, each treated separately. The heat-flux boundary condition was applied to keep the model simple and limit the simulations only to internal flow (gas in the SCR system). However, additional calculations were required before the main simulations (Equations (7) and (8)). The thin steel walls corresponded to the section where only the metal tube was present in the actual system. In turn, the thick steel wall represented the increased diameter of the metal tube resulting from the presence of the V-clamp in the actual system.

The heat flux in the uninsulated section was estimated iteratively. The heat transfer was assumed to be affected by the external air flow with a velocity of 1 m/s, taking into account the intensified cooling effect in the practical implementation of the SCR system. The overall thermal resistance ($r_q$) was determined following Equation (7):

$$r_q=r_{pipe}+{\displaystyle \frac{1}{h_{rad}+h_{conv}}}$$

(7)

where $r_{pipe}$—thermal resistance of the pipe, $h_{rad}$—heat transfer coefficient due to radiation, and $h_{conv}$—heat transfer coefficient due to convection. Finally, the heat flux was estimated (Equation (8)):

$${\varphi }_q={\displaystyle \frac{T_g-T_a}{r_q}}$$

(8)

where $T_g$—temperature of flue gas and $T_a$—ambient temperature (25 °C). The attained heat flux values are depicted in

Table 5. Estimated heat flux in bare-duct section.

Operating Conditions	OP1	OP2
Heat flux: thin metal sheet	−5236 W/m2	−7126 W/m2
Heat flux: thick metal sheet	−5014 W/m2	−6839 W/m2

.

2.3. Urea Decomposition

The simulations utilised the urea–water solution decomposition mechanism described by Brack et al. [40,41]. A set of kinetic reactions was considered, representing the formation of ammonia (NH₃), isocyanic acid (HNCO), ammelide, biuret, cyanuric acid, and water vapour, according to the scheme presented in

Table 6. Kinetic reactions considered in urea decomposition model by Brack et al. [40], extended by urea’s change from a molten to a gaseous state [42]; g—gaseous state, l—liquid state, m—molten state, s—solid state.

No.	Reaction
I	$\mathrm{c}\mathrm{y}\mathrm{a}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{d}\left(\mathrm{s}\right)\to 3\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O} \left(\mathrm{g}\right)$
II	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)\to \mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)+\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O} \left(\mathrm{l}\right)$
III	$\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)+\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{l}\right)\to \mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t} \left(\mathrm{m}\right)$
IV	$\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)\to \mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{l}\right)+\mathrm{N}{\mathrm{H}}_3 \left(\mathrm{g}\right)$
V	$2\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)\to \mathrm{a}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{d}\mathrm{e}\left(\mathrm{s}\right)+\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{l}\right)+\mathrm{N}{\mathrm{H}}_3\left(\mathrm{g}\right)+{\mathrm{H}}_2\mathrm{O} \left(\mathrm{g}\right)$
VI	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)+\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O} \left(\mathrm{g}\right)\to \mathrm{c}\mathrm{y}\mathrm{a}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{d}\left(\mathrm{s}\right)+\mathrm{N}{\mathrm{H}}_3\left(\mathrm{g}\right)$
VII	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)+\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{g}\right)\to \mathrm{t}\mathrm{r}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t} \left(\mathrm{s}\right)$
VIII	$\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t} \left(\mathrm{s}\right)\to \mathrm{c}\mathrm{y}\mathrm{a}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{c}\mathrm{i}\mathrm{d}\left(\mathrm{s}\right)+\mathrm{N}{\mathrm{H}}_3\left(\mathrm{g}\right)$
IX	$\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)+2\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{l}\right)\to \mathrm{a}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{d}\mathrm{e}\left(\mathrm{s}\right)+{\mathrm{H}}_2\mathrm{O} \left(\mathrm{g}\right)$
X	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)\to \mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{x}\right)$
XI	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{x}\right)\to \mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\right)$
XII	$\mathrm{b}\mathrm{i}\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{t}\left(\mathrm{m}\mathrm{a}\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{x}\right)\to 2\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O} \left(\mathrm{g}\right)+\mathrm{N}{\mathrm{H}}_3\left(\mathrm{g}\right)$
XIII	$\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{s}\right)\to \mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)$
XIV	$\mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{m}\right)\to \mathrm{u}\mathrm{r}\mathrm{e}\mathrm{a}\left(\mathrm{s}\right)$
XV	$\mathrm{a}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{d}\mathrm{e}\left(\mathrm{s}\right)\to \mathrm{a}\mathrm{m}\mathrm{m}\mathrm{e}\mathrm{l}\mathrm{i}\mathrm{d}\mathrm{e}\left(\mathrm{g}\right)$
XVI	$\mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O}\left(\mathrm{l}\right)\to \mathrm{H}\mathrm{N}\mathrm{C}\mathrm{O} \left(\mathrm{g}\right)$

.

The change in amount of $i$-th species in time was evaluated using Equation (9) [40]:

$${\displaystyle \frac{d{n}_i}{dt}}=\sum _{r=I}^{XVI}{\upsilon }_{i,r}·{\displaystyle \frac{d{n}_r}{dt}}+j_i$$

(9)

where ${\upsilon }_{i,r}$—the stoichiometric factor of conversion of $i$-th species in $r$-th reaction, ${\displaystyle \frac{d{n}_r}{dt}}$—the reaction rate, and $j_i$ is a source term for evaporation.

The reaction rates (reactions II–XV) were evaluated by Equation (10) [41]:

$${\displaystyle \frac{d{n}_r}{dt}}=A_{0,r}\exp \left(-{\displaystyle \frac{E_{A,r}}{R_{u}T}}\right)·V·\prod _1^j{c}_j^{{\gamma }_j}$$

(10)

where $A_{0,r}$ is the factor of the reaction $rxn$, $E_{A,r}$—activation energy of the reaction $r$, $R_u$—the ideal gas constant, $T$—temperature, $V$—reactor’s volume, and $c_j^{{\gamma }_j}$—concentration of the educt species $j$ with a reaction order of ${\gamma }_j$. The reactor’s volume was determined by Equation (11) [42]:

$$V=\sum {\displaystyle \frac{n_i{M}_i}{{\rho }_i}}$$

(11)

where $M_i$—molar mass of $i$-th species and ${\rho }_i$—density of $i$-th species.

The temperature-dependent variations of specific heat, enthalpy, and entropy were established using NASA polynomial coefficients (Equations (12)–(14)). These coefficients were obtained from [43]. For the ammelide in the gaseous state, coefficients $a_1$–$a_5$ were assumed to be the same as those for cyanuric acid (gasous) [43]. The $a_6$ and $a_7$ were adjusted to match the ammelide’s standard enthalpy of formation (at 298.15 K) [44] and its standard entropy at 298.15 K [43].

$${\displaystyle \frac{c_{p,i}}{R_u}}=a_{1i}+a_{2i}T+a_{3i}{T}^2+a_{4i}{T}^3+a_{5i}{T}^4$$

(12)

$${\displaystyle \frac{H_i}{R_u}}=a_{1i}T+{\displaystyle \frac{a_{2i}}{2}}{T}^2+{\displaystyle \frac{a_{3i}}{3}}{T}^3+{\displaystyle \frac{a_{4i}}{4}}{T}^4+{\displaystyle \frac{a_{5i}}{5}}{T}^5+a_{6i}$$

(13)

$${\displaystyle \frac{S_i}{R_u}}=a_{1i}\mathrm{l}\mathrm{n}(T)+a_{2i}T+{\displaystyle \frac{a_{3i}}{2}}{T}^2+{\displaystyle \frac{a_{4i}}{3}}{T}^3+{\displaystyle \frac{a_{5i}}{4}}{T}^4+a_{7i}$$

(14)

where $c_{p,i}$—molar heat capacity of $i$-th species, $H_i$—molar enthalpy of $i$-th species, and $S_i$—molar entropy of $i$-th species.

2.4. Injection Properties

The calculation of the UWS mass flow rate was based on the following assumptions: the NOx concentration and the NH₃/NOx ratio ($\alpha$) remain the same in all cases; the thermolysis reaction (Equation (15)) and the hydrolysis reaction (Equation (16)) produce two moles of ammonia in total from one mole of urea; and one mole of ammonia is required to reduce one mole of the nitric oxide (Equation (17)).

$${\left(N{H}_2\right)}_{2}CO\to N{H}_3+HNCO$$

(15)

$$HNCO+H_{2}O\to N{H}_3+C{O}_2$$

(16)

$$4NO+4N{H}_3+O_2\to 4N_2+6H_{2}O$$

(17)

The NOx concentration (represented by NO in the subsequent analysis) was specified at 500 ppm [45]. The molar NH₃/NOx ratio ($\alpha$) was set at 1.2. The estimated mass flow rate of the urea–water solution required to reduce the nitrogen oxides was calculated using Equation (18).

$${\dot{m}}_{uws}={\dot{m}}_g·{\displaystyle \frac{\alpha }{2}}·{\displaystyle \frac{c_{NOx}}{{10}^6}}·{\displaystyle \frac{R_g·M_{urea}}{R_u·y_{urea}}}$$

(18)

where ${\dot{m}}_g$—mass flow rate of exhaust gas, $c_{NOx}$—NOx concentration, $M_{urea}$—molar mass of urea, and $y_{urea}$—mass ratio of urea in UWS.

The evaluated UWS mass flow rate (Equation (18)) was associated with the continuous injection. However, considering the use of a pulse injector (3-nozzle Bosch injector, 0 444 043 212), the required UWS mass for each individual injection and the corresponding opening time were determined using mass flow rate measurements obtained from separate experiments conducted at a specified injection pressure of 9 bar and an injection frequency of 4 Hz (Equations (19) and (20)):

$$m_{inj}={\displaystyle \frac{{\dot{m}}_{uws}}{f_{inj}}}$$

(19)

$$t_{inj}={\displaystyle \frac{m_{inj}}{{\dot{m}}_{inj}}}$$

(20)

where $m_{inj}$—mass of injected UWS in a single injection, $f_{inj}$—injection frequency, $t_{inj}$—time of a single injection, and ${\dot{m}}_{inj}$—mass flow rate of the injector. The summary of the injection properties is presented in

Table 7. Injection conditions in considered numerical cases.

Operating Conditions	OP1	OP2
UWS mass per injection	10.66 mg	46.63 mg
Duration of a single injection	3 ms	14 ms

.

3. Results and Discussion

The development of the SCR mixing device aimed to adapt the base variant to comply with Stage V regulations. The goal was to enhance the performance of the SCR mixing system by mitigating the intensive deposit formation close to the outlet of the mixing system and enhancing the mixing of the flue gas with droplets in the initial injection zone. The study examined various criteria, including liquid film and solid deposits in the system, mixing performance, and generated pressure drops. Moreover, the cross-sections presenting the turbulence kinetic energy (TKE) as an important mixing indicator were compared to assess the impact of geometric changes on droplet-mixing performance.

3.1. Liquid Film and Solid Deposits

The experimental tests conducted on the base version (Model 1) indicated a tendency for deposit formation at relatively high NH₃/NOx ratios. The majority of solid deposits were observed at the outlet of the mixing system, where the mixing device was connected to the downstream duct of the exhaust system. This specific area, as explained in Section 2.2, lacked thermal insulation, resulting in local drops in wall temperature, as illustrated in Figure 6.

The presence of deposits resulted in a decrease in ammonia generation efficiency, thereby reducing NOx conversion. The above process was compensated by a further increase in UWS mass flow rate; however, this practice led to the “domino effect” in terms of solid deposit generation. Consequently, one of the key design objectives in the development of the SCR mixing systems was to mitigate deposit formation near the system outlet. However, due to the design specifications of the exhaust system, thermal insulation could not be applied to this outlet section. Therefore, efforts were made to minimise deposit formation by reducing contact between UWS droplets and their by-products with the system walls.

The comparison of the total mass of the liquid film, including solid by-products like triuret, obtained from numerical simulations after four UWS injections, is shown in Figure 7.

Surprisingly, the development of the system did not reduce the liquid mass in the case of OP1, except for Model 5. For moderate engine load (OP2), the liquid film mass slightly increased for Models 2, 3, and 7, from approximately 15 mg to about 20 mg. However, the film mass for Models 4–6 was significantly higher, around 30 mg. The presence of a “cap” obstructing the gas flow inside the mixer in Model 5 resulted in a distinct response to liquid film deposition under different engine loads. This was due to droplets being trapped inside the “cap” and was influenced by the hot gas flow, promoting evaporation. However, increased UWS dosage at OP2 caused significant cooling of the “cap” (Figure 8), leading to an increase in film mass. Therefore, the occurrence of the liquid film in Model 5 was highly dependent on operating conditions and NH₃/NOx ratios.

The liquid film observed in all mixing systems consisted of approximately 95% urea (in a molten state) and 3.8–5.3% water (by mass) in both OP1 and OP2 conditions. However, small amounts of other substances were also present, mostly observed under OP2 conditions. The detailed mass compositions of the liquid film at the end of the fourth injection are presented for Models 1, 5, and 7 (OP2), for which this composition was the most diverse (see Figure 9, Figure 10 and Figure 11). It can be observed that the mass fractions of solid deposits were negligible. Among all the species other than urea and water, biuret (in a matrix or molten form) was the predominant component in all cases, which is consistent with measurements conducted by Alembath et al. [46].

The evolution of the mass fraction of the liquid film in the entire mixing system was compared between two models: Model 5 (characterised by the highest production of solid deposits; Figure 12) and Model 7 (the latest iteration in the design development; Figure 13). The time-dependent behaviour of the mass for species with relatively high concentrations followed a distinct pattern due to the cyclic injections. In the case of urea (in a molten state), the curve exhibited step-like increases in mass immediately after each injection, followed by a slight decrease before the subsequent injection. The mass of water also experienced an immediate rise after each injection, reaching a clear peak, and then rapidly decreased and remained nearly unchanged until the next injection. A similar trend in species mass over time was observed for Model 7, although the peaks in water mass were less distinct. For minor species, the mass of each species increased over time, without clear injection patterns, for both Model 5 and Model 7. Notably, the mass of biuret (in matrix form) and ammelide (in solid form) significantly increased in Model 5; while a similar trend was observed for molten biuret in Model 7. These observations suggested that the amount of UWS by-products, other than urea, water, NH₃, and HNCO, may increase considerably over time, which explains the deposition observed in the real system (Model 1).

In the next steps, the mass of the film formed on the uninsulated walls of the outlet duct (Figure 14a) and the average temperature of the uninsulated walls were compared between systems (Figure 14b) after four injections. It can be observed that the film mass decreased only for Model 5 in both OP1 and OP2. However, this resulted from the liquid film primarily developing inside the bottom “cap” of the mixer, which was additionally confirmed by the simultaneous increase in the film mass in the entire mixing system (7, OP27). The film mass in the uninsulated area increased for Models 2–4 and Models 6 and 7. However, the average wall temperature in this area also increased. Taking into account the urea melting temperature of approximately 133 °C, the minimum temperature for urea thermolysis of 152 °C [47], and the average uninsulated wall temperature of 140 °C for Model 1, even a minor temperature increase was expected, as it could significantly reduce the formation of solid deposits.

The mass of the solid UWS by-products, consisting of ammelide, biuret, cyanuric acid, triuret, and solid urea, obtained after four injections in an entire system is presented in Figure 15. The operation conditions OP1 were characterised by the relatively low temperature of the flue gas (250 °C), under which only some of the UWS by-products start to form [48]. This is the most probable explanation for the similar mass values of the solid by-products observed across different models. Interestingly, under OP1 conditions, the mass of solid deposits increased compared to Model 1. This effect was observed for all mixing devices, except for Model 5 where no deposits were predicted.

When the engine load was increased (OP2), resulting in a higher operating temperature (300 °C) and increased UWS dosing (see Section 2.4), the mass of solid deposits significantly increased for Model 1 (0.032 mg). However, the geometric changes in Models 2–4 and Models 6 and 7 demonstrated clear benefits, as the mass of solid deposits decreased, even compared to OP1 conditions with lower UWS dosing. Model 5 exhibited a drastic increase in solid deposit mass, which was again attributed to the “cap” structure partially blocking the flow inside the mixer and collecting droplets, leading to intensive cooling of its walls (see Figure 8).

Based on the results obtained under both sets of operating conditions, the most significant reduction in solid deposit occurrence was predicted for Models 2 and 7. Models 3 and 6 performed better in terms of deposit mitigation under OP1 conditions, but their performance was worse under OP2 conditions.

The injected urea–water solution was present in the system in both a liquid state (droplets and film) and vapour (referred to as an evaporated spray and evaporated film). However, a certain part of the UWS and by-products escaped the domain, and that part was referred to as the deserted mass. The mass balance is expressed by Equations (21) and (22).

$$m_{injected}=m_{droplets}+m_{film}+m_{evaporated}+m_{deserted}$$

(21)

$$m_{evaporated}=m_{evaporated droplets}+m_{evaporated film}$$

(22)

The UWS mixing system should not only reduce the mass of the film or deposits but also maximise the evaporated mass of the injected UWS and its by-products. Therefore, the total (accumulated) mass of the evaporated water, urea, and related by-products was ultimately analysed (Figure 16).

The lowest total evaporated mass was obtained for Model 1 for both OP1 and OP2 conditions. The evaporated mass was increased by about 100% for OP1 when Models 3, 4, and 6 were employed. Among the aforementioned SCR mixing systems, Models 4 and 6 also showed a significant rise in the evaporated mass under OP2 conditions (increased by about 100% compared to the base Model 1). Model 5 led to the most differentiated results between operating points as the evaporated mass was nearly identical to Model 1 under OP1 but significantly improved under OP2 (166% higher than Model 1).

3.2. Mixing of UWS Droplets

One of the main objectives of a urea-mixing device is to achieve effective mixing of UWS droplets and their by-products for a uniform distribution of ammonia and isocyanic acid before entering the SCR catalytic converter. This requires the mixing device to provide intensive mixing, which can be attained through turbulence [12] or the splashing of UWS droplets on the blades [11] causing further break-up. In the case of the present analysis, experimental tests (from outside the scope of the presented work) of the base geometry (Model 1) indicated that the mixing process of UWS droplets with the exhaust gas was hindered due to low gas velocity in the injection zone. Therefore, the development of the mixing device required the improvement of the mixing performance, particularly in the area where the initial spray was formed. The mixing performance of the presented mixers was achieved by the intensification of the swirling flow, starting from the injection zone. Hence, the blades near the outlet of the mixing system (Model 1) were replaced with a tube equipped with swirling blades (Models 2–7), accompanied by additional guiding vanes in Models 3, 4, 6, and 7.

The devices considered in the study were further analysed in terms of the turbulence kinetic energy. Although TKE cannot be considered a direct indicator of mixing capabilities, it strongly influences cross-sectional mass transfer. Cross-sections depicting the TKE distribution for Model 1 (base design), Model 3 (equipped with two large guiding vanes), Model 5 (with the partly blocked mixer), and Model 7 (with the single guiding vane) are presented in Figure 17. In the case of Model 1, the mixing efficiency, as indicated by the TKE field, was negligible. The TKE values were close to zero for both OP1 and OP2 conditions, with only slight increases observed between the injector and mixer blades. However, the geometrical modifications in the mixer design, which introduced a swirling flow right after the inlet, significantly improved the mixing in the initial spray zone. Additionally, Model 3 generated increased the turbulence kinetic energy near its outlet, potentially enhancing the spatial distribution uniformity of UWS droplets and related by-products. Model 5, with its partially blocked tubular mixer, showed a remarkable increase in TKE downstream of the “cap.” However, there was no significant improvement in TKE magnitude within the primary injection zone. Model 7 was characterised by the clear growth in TKE values, both in the initial injection zone and the outlet section, demonstrating improved mixing performance for both OP1 and OP2 conditions.

3.3. Total Pressure Drop

The final criterion considered in the analysis was the drop in total pressure (defined as the difference between the total pressure at the inlet and outlet of the mixing system). The results (see Figure 18) indicated the lowest pressure drop occurring for the base Model 1 (about 0.1 kPa and 0.12 kPa for OP1 and OP2, respectively). This was attributed to the simpler design and the smallest total size of the mixing elements. The pressure drops were the highest for Model 5, reaching 0.12 kPa for OP1 and 1.86 kPa for OP2. The increased pressure drops were expected due to the partial blockage of the exhaust flow by the “cap” downstream of the mixing element. Models 3, 4, and 6 showed similar pressure drop values. The most favourable mixing systems in terms of pressure drops were Models 2 and 7, with total pressure drops of approximately 0.04 kPa and 0.7 kPa for OP1 and OP2, respectively.

3.4. Overall Performance

The final selection of the urea-mixing device was primarily based on the reduction in the mass of solid deposits. Models 2 and 7 demonstrated the greatest decrease in solid deposits’ mass under OP2 conditions, which were considered more representative of real-world operational tests (as the low temperature in OP1 partially suppressed the formation of most solid urea by-products). Additionally, Model 7 significantly improved the mixing of the UWS both in the initial injection zone and near the outlet from the mixing system, without causing a significant increase in pressure drop.

4. Further Development

Ultimately, Model 7 was the chosen mixer from the development process, and it was selected for experimental tests on a flow rig and engine dyno. Model 1 was used as a reference. The section of the prototype system made for flow rig and engine dyno tests (Model 1) is shown in Figure 19.

After verification in the experimental tests (beyond the scope of this study), the system went through the full process of product development for series production. The production prototype was tested in the long term during thermal and vibration tests, which confirmed the robustness of the final selected design. After passing the tests, it was implemented in large-scale production for several platforms in the power range from 56 to 130 kW compliant with the Stage V NRE-v/c-5 emission standard.

5. Conclusions

This research describes the development processes of a multi-platform SCR system for mobile applications. The initial design suffered from severe deposit formation, particularly at the outlet from the mixing system. Therefore, the main objective of this study was to reduce the amount of solid deposits without a significant increase in pressure drop, and maintain the NH₃ uniformity index at a similar level as in the baseline design. Additionally, the mixing of spray droplets with exhaust gas and the pressure drops of the mixing system were numerically evaluated.

The analysis led to the following observations:

• Model 2 (with the highest number of blades) and Model 7 (with one guiding vane and seven blades) demonstrated the greatest reduction in solid deposit mass under OP2 conditions, which were considered more representative of real-world operational tests.
• Model 3 (with two large guiding vanes and seven blades) and Model 7 provided the most significant increase in turbulence kinetic energy, both in the initial injection zone and at the outlet of the mixing device.
• The newly designed mixers generated higher pressure drops, as expected, due to the increased size and complexity of the mixing elements.
• Among the newly developed designs, Models 2 and 7 exhibited the lowest pressure drops.

Considering the overall performance, with a particular focus on reducing solid deposits and improving mixing near the injector, Model 7 was selected as the final design and implemented into large-scale production.