Insights on the morphology of air-assisted breakup of urea-water-solution sprays for varying surface tension

The efficacy of NOx reduction in diesel engines is mainly dependent on how uniformly urea-water solutions (UWS) are dispersed onto the catalyst surface of the Selective Catalytic Reduction (SCR) systems. The urea-based SCR systems also suffer drawbacks due to the formation of urea deposits onto the walls of after-treatment devices due to poor atomization characteristics of UWS. In this work, the impact of lowering the surface tension of UWS on the morphology of UWS sprays was explored using high-speed shadowgraph imaging techniques. The surface tension of UWS was lowered by adding surfactants; two surfactants viz., Sodium Dodecyl Sulfate (SDS) and Dodecyl-Dimethyl-Amine-oxide (DDA) were considered in this investigation. The surface tension of UWS was reduced to a maximum from 73.7 to 30.2 mN/m and 39.8 mN/m with the addition of DDA and SDS respectively at 75% of its respective Critical Micelle Concentration (CMC) in UWS. Even at a very low-pressure difference of 500 mbar of co-flowing air, the surfactant-added UWS tends to break-up relatively closer to the nozzle tip due to flapping-induced bag breakup, which improved its drop-size distribution. Under a relatively higher pressure difference of 2000 mbar of co-flow atomizing air, the liquid breakup was mostly due to surface stripping in surfactant-added UWS sprays that generated a large number of fine droplets. The image anal∗ Corresponding author Email address: aniket.kulkarni@brunel.ac.uk (Aniket P. Kulkarni) Preprint submitted to International Journal of Multiphase Flow August 11, 2020 yses of sprays were performed at far downstream locations from the nozzle to quantify the variations of their droplet-sizes caused by varying the surface tension of UWS. The surfactants added UWS sprays revealed a considerably narrower drop-size distribution by up to 43% compared to UWS sprays under high-pressure conditions, and this was due to a combination of flapping-induced bag breakup, surface stripping and secondary atomization of big droplets caused by reducing the surface tension of UWS. Reducing the surface tension of UWS has the potential to improve NOx reduction in SCR systems due to the reduction in droplet sizes of UWS sprays and also to reduce the formation of urea deposits.


Introduction
The nature of fuel to energy conversion process in various combustion systems generate many harmful emissions including soot and nitric oxides (NO x ).
After-treatment devices are commonly used to achieve the simultaneous reduction of both soot and NO x emissions (Guan et al., 2014). The abatement of 5 NO x is effectively achieved by using selective catalytic reduction (SCR) systems (Guan et al., 2014;Birkhold et al., 2006). This method relies on the injection of urea water solution (UWS, 32.5% urea by weight) into the exhaust gases (Birkhold et al., 2006;Ebrahimian et al., 2012). The injected UWS evaporates due to heat available in the exhaust gases and undergoes a series of 10 chemical processes such as hydrolysis and thermolysis to generate ammonia vapour which acts as a reducing agent to decompose NO x in the presence of a catalyst (Guan et al., 2014;Spiteri et al., 2015;Varna et al., 2015). Many efforts have been made to improve the performance of SCR systems to achieve better NO x conversion efficiency and to minimize impingement of UWS droplets 15 on the walls of the SCR systems. These efforts include use of mixer configurations (Oh and Lee, 2014;Lecompte et al., 2014), which enhances the mixing of UWS with exhaust gases and this improves the NO x conversion efficiency of the SCR system (Guan et al., 2014). The mixing process is mainly governed by atomization characteristics of UWS sprays (Guan et al., 2014;Varna et al., 20 2015; Wong et al., 2004;Payri et al., 2019a). Moreover, poorly atomized and non-vaporized droplets from UWS spray may impinge on the internal surfaces of the SCR system which leads to urea residues Liao et al., 2017;Huang et al., 2020;Koebel et al., 2000;Dörnhöfer et al., 2020;Eggers and Villermaux, 2008). Thus, it is important to study atomization characteristics 25 of UWS sprays to achieve improved NO x conversion efficiency with minimum wall impingement.
The spray characteristics of UWS have been widely studied under nonevaporative and evaporative cross-flow conditions (Varna et al., 2015;Payri et al., 2019a;Liao et al., 2017;Payri et al., 2019b;Shi et al., 2013;Kapusta 30 et al., 2019;Kapusta, 2017). Varna et al. (2015) studied UWS sprays from a pressure-driven atomizer at 9 bar injection pressure for different velocities of cross-flow; wall-hitting was observed under low cross-flow velocities. The effect of injection angle on the mixing length in an SCR system was studied for a pressure-driven atomizer (Shi et al., 2013). It was found that orthog-35 onal injection to exhaust flow leads to a shorter mixing length. Payri et al. (2019a,b) studied the effect of injection pressure and temperature of cross-flow using a pressure-driven atomizer on the size and velocity distributions of UWS droplets. They found that droplet velocity in cross flow increases with high gas temperatures due to change in density of gas flow. They concluded that smaller 40 droplets with higher droplet velocities can be produced at high temperatures of cross-flow (Payri et al., 2019a,b), these observations are consistent with those reported in (Postrioti et al., 2015;Needham et al., 2012). Spiteri et al. (2015) compared spray-air interactions for air-assisted and pressure-driven atomizers under cross-flow conditions, the UWS sprays from pressure-driven atomizer are 45 least affected by cross-flow velocities and this leads to wall-impingement. On the other hand, air-assisted atomizers produced droplets of smaller Sauter Mean Diameter (SMD), resulting in better mixing of UWS and exhaust gases. Thus, air-assisted atomization strategy has gained attraction for atomization of UWS in SCR systems Zheng, 2017). Recently, it was demon-50 strated that atomization of UWS can also be improved with the help of electrostatic force (Pratama et al., 2019). Many studies have employed Phase Doppler Interferometry (Varna et al., 2015;Liao et al., 2017;Pratama et al., 2019;Shi et al., 2013) or back-light illumination measurements (Payri et al., 2019a,b;Lieber et al., 2019) for characterization of UWS sprays. Most of these stud-55 ies have focused on droplet SMD and distributions of droplet size and velocity.
It was reported that UWS droplets below 20 µm can completely entrain or evaporate in engine-exhaust like situations Liao, 2017). The impingement rate of UWS droplets was higher with large droplets and, droplets larger than 90 µm droplet diameter are likely to impinge on walls of the SCR 60 system Liao, 2017). These larger droplets may also affect NO x conversion efficiency. Thus, it important to reduce the presence of bigger size droplets (i.e. droplets larger than 90 µm droplet diameter). This observation highlights the need of narrower drop-size distributions in UWS sprays for effective NO x conversion in SCR systems with minimum wall residues. This can be achieved by varying the physical properties of UWS, which might improve its atomization characteristics. The physical properties of UWS can be altered by adding surfactants to improve atomization (Lecompte et al., 2014;Ayoub et al., 2011;Wasow and Strutz, 2015;Ayyappan et al., 2015;Tareq et al., 2020;Kooij et al., 2018;Sijs et al., 2019;Sijs and Bonn, 2020). Ayoub et al. (2011) 70 studied the effect of using surfactants as additives to improve NO x conversion efficiency in SCR systems. Various surfactants viz., dodecyldimethylamine oxide (DDA), Stearyl trimethyl ammonium chloride, Sodium lauryl ether sulfate (SLES) and Sodium dodecyl sulfate (SDS) were considered. It was reported that the addition of surfactants to UWS improved NO x conversion efficiency, 75 and better reduction of NO x was achieved with anionic surfactants such as SDS and SLES in engines (Ayoub et al., 2011). However, spray characteristics were not explored in their work. The alcohol ethoxylates-based surfactants were also studied to improve drop-sizes in UWS sprays (Wasow and Strutz, 2015). It was reported that surface tension of UWS reduced from 67 to 28.87 mN/m while, droplet sizes were reduced upto 75% using these surfactants. Surfactants have also been used to reduce wall hitting in SCR systems due to improved atomization characteristics of UWS sprays. Ayyappan et al. (2015) used Formaldehydebased additive to reduce urea wall deposit and, achieved up to a maximum of 77.1% reduction. These observations corroborate with the findings of Lecompte  drop-size distribution helps to achieve better mixing due to smaller droplet diameters, and the absence of big droplets reduces wall-residues. The near-nozzle breakup also influences the resultant drop size distributions in the far-field region of the UWS sprays. Thus, it is important to understand the breakup processes occurring in the near-nozzle region of the air-assisted UWS sprays, 95 particularly for varying surface tension values of UWS as not much work has been done to explore the mechanisms that influence near-nozzle breakup.
In this work, the effect of adding surfactants on the breakup processes of airassisted UWS spray have been explored at various pressures of the atomizing air. The breakup processes occurring in the near-nozzle region, global spray 100 structure and drop-size distributions of UWS and surfactant-added UWS sprays were studied using the high-speed shadowgraphy method.

Experimental setup and back-light imaging
The schematic of the experimental setup is shown in Fig. 1. Air-assisted 105 spray of UWS was injected into a non-reactive, quiescent glass chamber under ambient conditions (i.e. ambient gas temperature of 20 • C and atmospheric pressure conditions). The spray was back-illuminated using an LED light source and  The field of view of the measurements (width×height) are: A -5.3 mm × 3.5 mm; B -15 mm × 30 mm; and C -20 mm × 3.5 mm.
shadowgraph images were acquired using a high-speed camera (Photron, SA X-2) equipped with Tokina macro 100, F2.8 objective lens of 100 mm focal length.

110
The imaging system was moved to different positions to study global spray structure, near-nozzle and drop-sizing as shown in Fig. 2. The breakup processes occurring in the near-nozzle region A were captured in a window of 5.3 mm × 3.5 mm (width×height) just below the injector tip at 168,000 frames per second (fps) and a shutter speed of 0.29 µs. The adopted high frame rate and small were performed at 50 mm below the injector tip to ensure that most of the atomization processes have occurred and most of the droplets were spherical.
The shadow images of the droplets were analyzed using an in-house developed MATLAB code for drop-size measurements. The images were preprocessed using the median filter and image subtraction. A binary image was then obtained 125 using the subtracted image and global thresholding. Image segmentation was performed to extract area of the droplets. Out-of-focus droplets and droplets with sphericity less than 0.8 were neglected for reliable drop-sizing (Blaisot and Yon, 2005;Kulkarni and Deshmukh, 2017). Same droplets that appeared in a sequence of frames due to high-speed imaging were filtered using residence 130 time calculated for every operating condition. Statistically sufficient number of droplets (more than 6000 droplets) are ensured in drop-size distribution calculations. Uncertainty in drop-size measurements was estimated to be less than 9.7%. The resolution of drop-size images was 22.2 µm per pixel, thus droplets with diameter less than 80 µm were neglected to avoid diffraction limit of the op-135 tical system and for reliable drop-size measurements. This also ensured that the droplets larger than 90 µm are captured in the drop-size distributions, which may cause wall-hitting. Figure 3 shows an example of the images obtained and droplets considered for determining the drop-size distribution. Further details

Preparation of test liquids and measurements of physical properties
In this subsection, methods adopted to prepare test liquids and details on measurements of physical properties of the test liquids (surface tension, dynamic viscosity and density) have been provided.

Physical properties of the test liquids
The physical properties of the test liquids were measured at room temperature of 20 • C. The surface tension measurements were carried out using 165 pendent-drop method (First Ten Angstroms, FTA100), the measurements were calibrated using deionized water as a standard liquid. Minimum ten measurements were performed at each test condition, and standard deviation and mean Physical properties of the test liquids are summarized in Table 3.

Test conditions 180
The experiments were carried out using UWS and six solutions of surfactantadded UWS. Three concentrations were used to form SDS25, SDS50 and SDS75 using 25, 50 and 75% of cmc value of SDS surfactant in UWS, respectively.  swirling coflow of atomizing air jet at high ∆P conditions (Lasheras et al., 1998;Guildenbecher et al., 2009;Varga et al., 2003;Zhao et al., 2018). Even at high ∆P=2000 mbar condition, bag formation was observed which resulted in few liquid lumps and big droplets, originated from the rim of bags for UWS sprays, as observed in Fig. 6(f). These big droplets could not undergo further secondary 230 breakup due to high surface tension of UWS. Addition of surfactants reduced surface tension of UWS, this favoured the secondary breakup of big droplets and liquid lumps as can be seen in Figs. 6(g) and 6(h). Hence, surfactant-added UWS sprays showed more uniformly distributed droplets along with very less number of big droplets at high gas pressure conditions.

235
Thus, better atomization was achieved through the addition of the surfac-tant in UWS and this was due to the combination of droplet stripping and secondary atomization of big droplets at high air pressures in surfactant-added UWS sprays.

Effect of surfactants on breakup processes in near-nozzle region 240
This section presents the breakup processes occurring in near-nozzle region of air-assisted UWS sprays. Primary breakup processes were studied to reveal the effect of addition of surfactants and atomizing air pressures (∆P  (Guildenbecher et al., 2009;Zhao et al., 2018). In general, the bag breakup mode was observed over a range of We rep between 678.43 and 1356.85 which corresponds to relatively higher  (Lasheras et al., 1998;Guildenbecher et al., 2009).  The sheet was further expanded by the co-flowing atomizing air to form a baglike structure due to competition between surface tension force of the liquid and aerodynamic force of the atomizing air as can be observed in Figs. 8(b) to 8(d).
When the surface tension force was overcome by the aerodynamic force of the 280 atomizing air, the bag-like structure fragments into droplets and ligaments from the rim of the bag structure became visible as seen in Figs. 8(e) and 8(f). Relatively larger size liquid droplets and ligaments were formed from the breakup of rim of the bag. These big droplets contain relatively more mass of liquid and may not evaporate completely in the SCR system due to its lower surface area 285 available for evaporation.
Addition of surfactants increased We rep due to the reduction in surface tension values as shown in Figure 7. The length of intact liquid core was relatively short for surfactant added UWS sprays compared to that of UWS sprays at 500 mbar condition. Contribution to liquid jet breakup due to flapping instability 290 was observed for We rep more than 1459.43 and this has been presented in Fig 9 for DDA50 sprays at ∆P=500 mbar. In flapping-induced bag breakup mode, a bag was initially formed, and then expanded by the atomizing air as shown breakup as can be shown in Fig.9(d).
At higher values of We rep (more than 3488.37), contribution of surface strip- locity of atomizing air leads to stripping of droplets and ligaments from liquid surface due to strong shearing forces, thus fine mist-like droplets were observed at higher We rep (Lasheras et al., 1998;Guildenbecher et al., 2009;Varga et al., 2003;Zhao et al., 2018). However, few big droplets could also be identified as can be seen in Figs. 10(a) and 10(b). Further at higher We rep (more than 315 4966.89), secondary breakup of these bigger droplets were observed. This could be attributed to the combined effect of reduction in surface tension of UWS and higher velocities of atomizing gas that might have helped with breakup of these big droplets.
The observed breakup processes of UWS sprays have been summarized in 320 a regime map for the range of We rep and GLR as shown in Fig. 11. Distinct breakup processes (such as flapping-induced bag breakup, surface stripping and secondary atomization of big droplets) were observed in the near-nozzle region Dropsize distributions improved even at low ∆P conditions with addition of surfactants. Narrow drop-size distribution with most of the droplets smaller than 120 µm was observed with surfactant added UWS at high ∆P conditions. Fig. 12(b). A similar observation was made at high ∆P conditions. The size of D max decreased from 980 µm in UWS sprays to 660 µm in DDA75 sprays at 500 355 mbar condition showing a reduction up to 33% as shown in Fig. 12(b). The dropsize distributions became slightly narrower for surfactant-added UWS at low ∆P conditions as can be seen in Figs. 12(a) and 12(b)). This could be attributed to flapping-induced bag breakup observed in the near-nozzle spray structures at these conditions that might have produced narrower drop-size distributions. The findings summarize that atomization characteristics in air-assisted UWS sprays can be improved with the addition of the surfactants to UWS. In this work, addition of surfactants reduced surface tension of UWS considerably and 375 improved breakup of air-assisted UWS spray through the combination of various breakup mechanisms. This resulted in narrower drop-size distributions with the higher number of smaller size droplets with larger surface to volume ratio. The associated increase in the droplet surface area due to lowering the surface tension of UWS will result in enhanced convective transfer of heat from surrounding hot 380 exhaust gases to the droplets. This helps in faster evaporation of UWS droplets and hence, better mixing with the hot exhaust gases and better NO x conversion efficiency. Further, reduction in droplet diameter will also help to minimize spray-wall interaction and formation of urea-residues in SCR systems. Overall, narrower drop-size distributions and thus, improved atomization of surfactantadded UWS sprays might help to enhance the performance of SCR systems with better NO x conversion efficiency and lower urea-wall residues.

Conclusions
An experimental study was undertaken to study the effect of addition of surfactants in UWS on breakup morphology and atomization characteristics