See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357133245

Spray Breakup Modelling for Internal Combustion Engines

Chapter · January 2022

DOI: 10.1007/978-981-16-8618-4_4

CITATION READS

1 283

2 authors:

Utkarsha Sonawane Avinash Kumar Agarwal

Indian Institute of Technology Kanpur Indian Institute of Technology Kanpur
17 PUBLICATIONS   63 CITATIONS    551 PUBLICATIONS   19,689 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

I am working on HCCI/ PHCCI and Mode Swiching for PHCCI Automotive engines; laser ignition of HCNG mixtures, GDI engine development and particulate
characterisation. View project

Technology Developement to tackle COVID-19 View project

All content following this page was uploaded by Utkarsha Sonawane on 22 March 2022.

The user has requested enhancement of the downloaded file.

 1 SPRAY BREAKUP MODELLING FOR INTERNAL
2 COMBUSTION ENGINES
3
4 Utkarsha Sonawane, Avinash Kumar Agarwal*
5 Engine Research Laboratory, Department of Mechanical Engineering
6 Indian Institute of Technology Kanpur, Kanpur-208016, India
7 *Corresponding Author’s email: akag@iitk.ac.in
8 Abstract
9 Rising concerns about emissions have led to a significant tightening of pollution
10 norms for internal combustion (IC) engines. High-pressure direct injection (HPDI)
11 technologies have been adopted for most on-road and off-road engines to meet the
12 global demand for clean and efficient powertrains. Higher fuel efficiency, superior
13 combustion, and lower pollutant formation are the characteristic features of the
14 HPDI. The introduction of alternative fuels, modified combustion geometry, and
15 novel combustion concepts demand continuous improvement in fuel injection
16 equipment (FIE). The complicated physics of HPDI and its modelling is an active
17 area of research among researchers and engine developers. Fuel-injected in the
18 combustion chamber breaks up into a spray of fine droplets, evaporating, mixing
19 with ambient air, and forming a fuel-air mixture, greatly affecting the engine
20 combustion and emission characteristics. Therefore, it is necessary to study the fuel
21 breakup phenomenon under different engine conditions comprehensively. Detailed
22 understanding of the spray breakup phenomenon is unavailable due to difficulties
23 in optical access, highly dense sprays, complex processes, etc. However, recent
24 advances in measurement technologies and computational tools have made it
25 feasible for researchers.
26 This chapter attempts to capture widely used spray breakup models and research
27 studies involving IC engines. Fundamental spray breakup and atomization have
28 been discussed at the beginning of the chapter. Subsequently, the basis and
29 fundamentals of popular spray models have been discussed. Finally, the authors
30 have comprehensively discussed the key contributions in sprays to provide an
31 overall idea about the spray models and their application for IC engine studies.
32 Various spray breakup models such as Blob Model, Linear Instability Sheet
33 Atomization (LISA) Model, Kelvin-Helmholtz (KH) Model, Kelvin-Helmholtz-
34 Aerodynamics Cavitation Turbulence (KH-ACT) Model, RT (Rayleigh-Taylor)
35 Model, Hybrid/ Modified Kelvin-Helmholtz Rayleigh-Taylor (KH-RT) Model,
36 Taylor Analogy Breakup (TAB) Model, Enhanced TAB breakup model (ETAB) are
37 discussed briefly in this chapter. Towards the end, a summary of the contents of the
38 chapter is provided, which covers highlights and significant observations.
39
40 Keywords: Spray Breakup Model, Eulerian-Lagrangian method, Atomization,
41 Engine
42
43 1. Introduction
44 Increased goods transport to cater to ever-increasing population leads to harmful
45 emissions from the transport sector. A major portion of transport is based on IC

1
46 engines. IC engines emit harmful pollutants such as NOx, CO, CO2, HC, and PM.
47 These pollutants degrade the ambient air quality, leading to serious health issues
48 among the people exposed to these pollutants. Therefore, stringent emission
49 regulations have been implemented on the IC engines globally. The emission norms
50 are becoming increasingly stricter with time. Complying with these emission norms
51 is the driving force for the automotive industry and researchers to develop cleaner
52 and more efficient combustion systems. The fuel injection equipment (FIE) is an
53 integral part of engine combustion architecture. Therefore, fundamental
54 understating and knowledge of the fuel injection process would help to achieve
55 cleaner combustion. Spray formation and atomization are crucial processes
56 controlling the IC engine emissions. The fuel-injected in the combustion chamber
57 forms sheet/ jet/ ligaments [1] [2]. Interactions between the surrounding ambient air
58 and fuel jet spray lead to interfacial instabilities and momentum diffusion. The
59 mechanisms involved in these phenomena are complex and influenced by ambient
60 conditions in which fuel is injected. The fuel spray is dense near the nozzle, which
61 dilutes out in the far-field, resulting in multiphase, multiscale, turbulent physical
62 processes in the engine cylinder. Aerodynamic interactions affect the liquid core
63 column. Liquid ligaments form when the liquid surface becomes unstable, creating
64 parent droplets followed by children droplets. Therefore, modelling of spray and jet
65 breakup is one of the most critical steps in the simulation of IC engines [3]. The
66 study of droplet deformation is vital as it significantly affects droplet drag and hence
67 trajectory. The instantaneous drag force, which leads to droplet deformation, is
68 given as:
1 𝜋𝑑𝑐𝑟𝑜 2
69 𝐹𝐷 = 𝜌𝑎 𝑈0 2 𝐶𝐷
2 4
70 Here, 𝐶𝐷 is the instantaneous drag coefficient, and 𝑑𝑐𝑟𝑜 is deformed droplet
71 diameter. The equation requires information of 𝐶𝐷 as a function of deformation and
72 𝑑𝑐𝑟𝑜 as a function of time.
73 Spray modelling includes macroscopic and microscopic events, including the
74 spatial distribution of fuel vapour and droplets. The total surface area of fuel
75 droplets contributes to the fuel evaporation rate, which further influences the
76 combustion reactions. Misinterpretation of droplet size distribution may lead to
77 inaccurate anticipation of fuel evaporation rate, fuel vapour distribution, and
78 mixture formation, resulting in an incorrect prediction of combustion reactions. The
79 complexity in understanding spray breakup is due to the interdependency of
80 multiple processes and mechanisms [4]. These processes include spray jet breakup,
81 heat and mass transfer, evaporation, droplet to droplet interactions, ambient gas
82 forces, fluid properties, droplet deformation, collision, coalescence, etc. Failure to
83 predict any of the above processes may lead to misinterpretation of successive
84 processes. CFD simulation is a powerful tool for investigating such complex
85 processes and mechanisms. The simulation assumes boundary conditions based on
86 the experiments and simplifies the actual physics behind these phenomena. CFD
87 evaluates complex equations and solves them numerically to capture complex flow
88 phenomena. Also, improvement in computation power and capability opens up
89 doors for new research and technology development [5]. CFD analysis of spray is

2
 90 an integral part of combustion research and improves the interpretation of
91 complicated in-cylinder processes. Practical and effective application of modelling
92 is related to the present understanding of the physical process involved.
93 Numerical modelling and simulations can be used to explore the grey and less
94 understood areas of engine research. This chapter aims to provide an overview of
95 CFD spray breakup models used in IC engines. An attempt has been made to present
96 the fundamental concepts and assumptions in various spray breakup models in a
97 simplified manner. This chapter is mainly divided into three parts. First, detailed
98 information of primary and secondary breakup phenomena has been presented.
99 Then the authors briefly discussed primary and secondary breakup models in the
100 second part. The third part is dedicated to spray breakup modelling literature,
101 covering the application and accuracy of the spray models for various applications.
102 Finally, a summary is presented towards the end of this chapter.
103
104 2. Fundamentals of Liquid Spray Atomization
105 Spray atomization is categorized into two: primary and secondary spray breakup.
106 The combined effect of both breakups leads to the formation of fine droplets from
107 the liquid jets emanating from the injector nozzle. Both types of breakups happen
108 in different ways, depending on the ambient conditions and droplet properties.
109 These properties, along with geometrical parameters, are grouped in the form of
110 non-dimensional numbers such as Reynold Number (Re), Weber number (We), and
111 Ohnesorge number (Oh). The critical points in these breakup mechanisms are
112 governed by the specific value of these non-dimensional numbers. Figure 1 shows
113 a classification of primary and secondary breakup models.
𝐼𝑛𝑒𝑟𝑡𝑖𝑎 𝐹𝑜𝑟𝑐𝑒 𝜌𝑙 𝑢𝐷
114 𝑅𝑒𝑙 = =
𝑉𝑖𝑠𝑐𝑜𝑢𝑠 𝐹𝑜𝑟𝑐𝑒 𝜇𝑙
𝐼𝑛𝑒𝑟𝑡𝑖𝑎 𝐹𝑜𝑟𝑐𝑒 𝜌𝑙 𝑢2 𝐷
115 𝑊𝑒𝑙 = =
𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝑡𝑒𝑛𝑠𝑖𝑜𝑛 𝐹𝑜𝑟𝑐𝑒 𝜎
𝑉𝑖𝑠𝑐𝑜𝑢𝑠 𝐹𝑜𝑟𝑐𝑒 𝜇𝑙
116 𝑂ℎ = =
𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝑇𝑒𝑛𝑠𝑖𝑜𝑛 𝐹𝑜𝑟𝑐𝑒 √𝜎𝜌𝑙 𝐷
117 Here, 𝜌𝑙 = liquid density, 𝜎 = surface tension, 𝜇𝑙 = dynamic viscosity, 𝐷 = nozzle
118 diameter, 𝑢 = jet velocity. ‘l’’ represents the liquid phase.
119

120
121 Figure 1: Spray breakup type (Adapted from [6])

3
122 A higher weber number generates a higher number of fine droplets through the
123 primary and secondary breakups. Therefore, spray with a higher weber number is
124 of specific interest for IC engines since they have fast and superior atomization and
125 faster fuel evaporation.
126
127 2.1 Primary Spray Breakup
128 The primary breakup of sprays is related to the breakup of the intact liquid core.
129 Therefore, it is a crucial breakup process since it determines the sizes of droplets
130 formed from the liquid core.

131
132 Figure 2: Differences in the actual and simulated breakup of the sprays (Adapted from [5])
133
134 The cylindrical liquid column emerging from the injector nozzle breaks into discrete
135 ligaments/ blobs/ droplets in the primary spray breakup, primarily because of
136 surface instabilities. The surface instabilities depend upon the interactions of
137 competing surface forces (surface tension, aerodynamic drag) and initial
138 perturbations from the internal nozzle surfaces. Surface instabilities form on the
139 liquid sheet surface and interact with the ambient high-speed gas. These instabilities
140 grow in magnitude downstream, leading to a breakup. When a liquid sheet is
141 injected from the nozzle, it may encounter three instabilities: a) Kelvin-Helmholtz
142 (K–H) instability, b) Rayleigh–Taylor (R–T) instability, and c) capillary (Rayleigh)
143 instability. During atomization of the liquid sheet, shear-induced capillary waves
144 grow at the vapour-liquid interface, known as KH instabilities, resulting in ligament
145 growth and breakup. RT instabilities are due to the density difference between the
146 vapour and liquid interface, which improves the disintegration of droplets [7].
147 Figure 2 shows the actual and simulated spray liquid core differences close to the
148 nozzle [5]. Near nozzle visualization of spray remains an active area of research.
149 The complicated physics near the nozzle is not well understood and is often
150 modelled using simplified assumptions, matching subsequent processes
151 (downstream droplet dispersion) with reasonable accuracy.

4
 (a) [8] [9]

(b)
152 Figure 3: Different primary breakup regimes [6]
153
154 Figure 3 (a) shows different primary breakup regimes as a function of Re and Oh
155 numbers. Rayleigh breakup happens at low spray jet velocity because of
156 axisymmetric oscillations due to inertial and surface tension forces. The breakup
157 length is higher in this case, which decreases with increasing spray jet velocity.
158 Here, droplets of size equal to jet diameter are formed. Aerodynamic forces increase
159 the inertial and surface tension forces in the first wind-induced breakup (Figure 3).
160 In contrast, short unstable surface waves initiate in the second wind-induced
161 breakup, further amplified by the aerodynamic forces. Atomization takes place at
162 the nozzle; therefore, intact liquid core length is zero. Here, droplets of much
163 smaller size compared to nozzle jet diameter are formed [10]. This is a very relevant
164 zone for direct fuel injection diesel engines.
165
166 2.2 Secondary Spray Breakup

5
167 The secondary breakup process disintegrates the primary spray droplets to form
168 smaller droplets due to aerodynamic forces. The aerodynamic forces are
169 proportional to the relative velocity of droplets w.r.t. the ambient air. Therefore, the
170 size of droplets is a characteristic parameter for the secondary spray breakup. Higher
171 We facilitate smaller droplets because of increased aerodynamic forces and faster
172 breakup with increasing We [11]. As We increase, various kinds of breakup
173 morphologies begin to happen, such as vibrational, bag, bag-and-stamen,
174 multimode, sheet-thinning, and catastrophic breakup modes, as shown in figures 4
175 and 5 [12].

176
177 Figure 4: Secondary spray breakup morphology [6]
178
179 Vibrational breakup breaks down the liquid column into coarse droplets. Bag
180 breakup forms a single bag-like shape, whereas multimode breakup leads to the
181 formation of more than one bag-like shape. The striping and catastrophic breakup
182 occur when the outer edges of the liquid column accelerate and break up before the
183 center. Each morphological breakup forms secondary droplets of different sizes. All
184 droplet breakups happen at the same time for the high-velocity sprays in the IC
185 engines. The breakup is most intense near the nozzle due to higher We, which
186 decays with We downstream of the spray. The lower We downstream of the nozzle
187 is due to the smaller droplets and lower relative velocities [14]. Depending on the
188 ambient and initial conditions and injected fluid properties, the droplets reach a
189 minimum diameter, which can not be disintegrated further. This diameter is called
190 the ‘critical diameter’ of the droplets. Conditions favourable for lower critical
191 diameters are desirable for rapid evaporation of sprays in IC engine applications.
192 However, a rigorous breakup may lead to slower and lower spray penetration,
193 leading to inferior air utilization.
194
195 3. Spray Breakup Models
196 Unlike real spray ligaments, the spray is treated as a collection of discrete droplet
197 parcels and tracked using the Lagrangian approach in simulations. Generally, spray

6
198 droplets are modelled using primary breakup models until droplets achieve model
199 switching criteria. Once the switching criteria are met, the secondary breakup
200 modelling is initiated for the particular droplet/ droplet group. Droplet groups are
201 formed by taking droplets of a certain range to save computation time. The primary
202 breakup phenomenon determines subsequent air-fuel mixture formation in the
203 combustion chamber. Therefore, it is essential to perform detailed modelling of
204 dense sprays [15]. Primary spray breakup modelling is meant for predicting the
205 starting conditions of spray droplets, like initial droplet radius and velocity
206 parameters (spray angle), which depend on the flow characteristics in the nozzle
207 and near nozzle instabilities. The relative velocity between the liquid film and
208 ambient air generates instability, which causes the break up of liquid film to
209 ligaments. These ligaments create droplets that further disintegrate into secondary
210 droplets [16]. A linear instability theory is implemented to determine the most
211 unstable waves acting on the bulk liquid before the breakup. Any variation in
212 instabilities near the nozzle exit will create variations in dominant unstable
213 wavelength, hence the resulting droplet diameter. The flow conditions near nozzle
214 exit fluctuate under practical conditions, influencing the droplet sizes. Therefore,
215 discrete probability function (DPF) and maximum entropy principle (MEP) were
216 applied by Sovani et al. [17] and Jaynes [18] to model droplet size in the spray.
217
218 3.1 Blob Model
219 This model introduces a set of computational parcels into the multidimensional
220 domain, presented as ‘blobs’ or portions of the injected liquid column. The blob is
221 defined by a sphere and characterized by an initial diameter same as the effective
222 nozzle diameter. The blob model is the simplest and most widely accepted approach
223 for defining initial conditions for the first spray droplets at the nozzle exit. However,
224 this approach assumes that spray atomization and droplet breakup in the dense spray
225 near the nozzle are not separate phenomena. Therefore, an exhaustive simulation
226 can be replaced by a simplified approach injecting coarse droplets with the same
227 diameter as the nozzle. These droplets are further exposed to secondary
228 aerodynamic induced breakup (Figure 5) [19].

229
230 Figure 5: Blob Model
231
232 Spray droplet diameter is equal to the nozzle diameter (D), and droplet generation
233 rate is calculated from the mass flow rate of the fuel. The blob breaks up after
234 interacting with the surrounding gas, but a domain with large discrete particles near

7
235 the nozzle represents the dense spray core. It is assumed that flow inside a nozzle is
236 slug and injection velocity Uinj of the blobs is given by using conservation of mass
237 as given below:
𝑚𝑖𝑛𝑗 (𝑡)
238 𝑈𝑖𝑛𝑗 (𝑡) =
𝐴𝑛𝑜𝑧𝑧𝑙𝑒 × 𝜌𝑙
239 Here, Anozzle is the nozzle cross-section area, ρl is the liquid density, and minj (t) is
240 the mass flow rate.
241 The flow is generally turbulent, and the blob with nozzle diameter (D) size is
242 injected with injection velocity (Uinj) at the start of injection. The cavitation is
243 observed in the flow during the main injection of fluid, as shown in figure 6. Due
244 to cavitation, an effective cross-sectional area (𝐴𝑒𝑓𝑓 ) of the nozzle decreases,
245 reducing the blob diameter. Therefore, the highest injection velocity and the lowest
246 blob diameter are anticipated during the main injection event.
4𝐴𝑒𝑓𝑓
247 𝐷𝑒𝑓𝑓 = √
𝜋

248
249 Figure 6: Cavitation Inside the nozzle hole
250
251 A more practical value of initial injection velocity and blob diameter can be
252 obtained from the dynamic calculation of both during the whole injection event,
253 including the influence of cavitation. However, only a reduction in cross-sectional
254 area due to cavitation is taken into consideration. Rise in turbulence and break up
255 energy because of the bursting of cavitation bubble is not involved. Blob is a
256 fundamental and widely used approach for investigating the primary breakup in
257 simulation. It is the most suitable method for defining initial conditions of liquid
258 injected in the combustion chamber if there is a lack of knowledge about the
259 composition of primary spray; however, the value of spray angle is obtainable [14].
260 This model does not characterize comprehensive physical and modelling of the
261 related phenomenon during the primary break up. Also, mapping spatial spray angle
262 and droplet size distribution from the three-dimensional nozzle flow cannot be done.
263 Another disadvantage of this approach is excluding the evolution of primary
264 breakup due to turbulence and bursting of bubbles downstream of the nozzle.
265
266 3.2 Linear Instability Sheet Atomization (LISA) Model
267 LISA is used to model the liquid sheet breakup. This model requires the spray cone
268 and dispersion angles as input. The spray cone angle represents plume development,
269 and the dispersion angle represents liquid sheet fluctuations from the mean cone
270 angle. Dispersion angle is a critical parameter since it leads to the radial distribution

8
271 of spray droplets from the mean cone angle [19]. This model has a general liquid
272 sheet breakup and a liquid injection methodology, especially pressure swirl
273 atomizers. Hollow cone sprays produce fine droplets, leading to superior fuel-air
274 mixture formation, lower penetration, and improved spray atomization. The
275 conversion from injector flow to fully evolved spray is modelled in three stages:
276 film formation, sheet breakup, and droplet disintegration (Figure 7).
277

278
279 Figure 7: LISA breakup model [20] [21]
280
281 Figure 7 shows that initial perturbations result in the onset of optimal wave mode
282 (λoptimal). These are unstable waves governed by linear stability theory. The
283 unbroken liquid sheet is assumed to fragment into segments (λoptimal/2). These
284 broken segments further roll up to form cylindrical ligaments due to surface tension,
285 followed by the breakup of these ligaments into droplets because of Plateau–
286 Rayleigh instability [20]. Uniform velocity at the injector exit is assumed. Total
287 sheet velocity (U) is given by:
. 2𝛥𝑝
288 𝑈 = 𝑘𝑣 √
𝜌𝑙
289 Here, 𝜟𝒑 is the injection velocity, 𝒌𝒗 is the coefficient of velocity and is given by:
4𝑚̇ 𝜌𝑙
290 𝑘𝑣 = 𝑚𝑎𝑥 [0.7, √ ]
𝜋𝑑02 𝜌𝑙 𝑐𝑜𝑠 𝜃 2𝛥𝑝
291 Here, 𝒎̇ is the mass flow rate, 𝒅𝒐 is the orifice diameter, 𝝆𝒍 is the liquid density.
292 The axial (u) and angular (v) components of the injector exit velocity can be
293 calculated from
294 u = U𝑐𝑜𝑠 𝜃, v = U𝑠𝑖𝑛 𝜃
295 The mass flow rate 𝒎̇ is given by
296 𝑚̇= 𝜋𝜌𝑙 𝑢ℎ𝑜 (𝑑𝑜 − ℎ𝑜 )
297 Here, u is injector exit axial velocity component, 𝒉𝒐 is initial liquid sheet thickness.
298 It is assumed that there is no drag on the liquid sheet and no air displacement by the
299 sheet volume due to no direct relationship between the sheet and the gas phase.

9
300 Hence, there will be no droplet breakup, collision, and drag until the droplet reaches
301 breakup length Lb [22] given by:
𝑈 𝜂𝑏
302 𝐿𝑏 = 𝑙𝑛 [ ]
Ω 𝜂0
𝜼
303 Here, 𝒍𝒏 [ 𝒃 ] is proposed by Dombrowski et al. [23]. 𝛀 is the maximum growth rate.
𝜼𝟎
304 𝜼𝒃 and 𝜼𝒐 are critical and original amplitudes. The new diameter after the breakup
305 at the breakup length 𝐿𝑏 is calculated by:
3𝜋 𝑑𝐿2
306 𝑑𝐷3 =
𝐾𝐿
307 Here, 𝒅𝑳 is the ligament diameter and 𝑲𝑳 is the most unstable wave number.
308
309 3.3 Kelvin-Helmholtz (KH) Model
310 Kelvin-Helmholtz (KH) and Rayleigh-Taylor (RT) breakup models are extensively
311 used. It includes two kinds of instabilities, which arise in multiphase flows and can
312 lead to perturbation growth resulting in disruption of spray droplets (Figure 8).

313
314 Figure 8: Secondary spray breakup: (a) KH, (b) RT [13]
315
316 The growth rate of perturbation depends on the conditions of ambient gas and
317 droplet itself. It is assumed that the growth of KH instabilities on the surface of a
318 cylindrical liquid jet causes its disintegration into droplets. These instabilities are
319 due to high shear at the interface of two fluids [24]. This cylindrical jet penetrates
320 the gaseous environment with a relative velocity of urel. The liquid and gas are
321 assumed as incompressible. It is assumed that gas has negligible viscosity. The
322 diameter of newly formed droplets is directly proportional to the wavelength of
323 unstable KH wave, which has a fast growth rate on the parent droplet s [25].
324 𝑑𝑛𝑒𝑤 = 2𝐵0 𝛬
325 Here, 𝑩𝟎 = constant. 𝜦= wavelength of the fastest-growing unstable wave having
326 growth rate (𝛺𝐾𝐻 ).

10
 0.34 + 0.38 𝑊𝑒𝑔1.5 𝜎
327 𝛺𝐾𝐻 = √
(1 + 𝑂ℎ)(1 + 1.4𝑇 0.6 ) 𝜌𝑙 𝑟 3
9.02𝑟 (1 + 0.45√𝑂ℎ)(1 + 0.4𝑇 0.7 )
328 𝛬𝐾𝐻 = 0.6
(1 + 0.865 𝑊𝑒𝑔 1.67 )
329 Here, 𝝈 = surface tension, T is the Taylor number = 𝑂ℎ √𝑊𝑒𝑔 , r is the parent
330 droplet radius, which reduces continuously during the breakup. The rate at which
ⅆ
331 the parent droplet radius reduces until it reaches a stable radius (𝑟𝐾𝐻 = 𝑛𝑒𝑤 ) is
2
332 given by:
𝑑𝑟 𝑟 − 𝑟𝐾𝐻
333 = , 𝑟𝐾𝐻 ≤ 𝑟
𝑑𝑡 𝜏𝐾𝐻
3.276 𝐵1 𝑟
334 𝜏𝐾𝐻 =
𝛺𝐾𝐻 𝛬𝐾𝐻
335 Here, 𝝉𝑲𝑯 = breakup time, 𝑩𝟏 = KH constant. 𝑩𝟎 and 𝑩𝟏 are the tuning constants
336 that affect the rate at which parent droplet contracts define the child droplet size
337 [26]. Mass is collected from the parent droplet till the shed mass is 5% of the initial
338 parcel mass. Here, a new parcel is formed having a diameter given by 𝑑𝑛𝑒𝑤 =
339 2𝐵0 𝛬. The new parcel provided similar properties as the parent parcel, excluding
340 radius and velocity. The secondary breakup of child droplets happens because of
341 the competing effects of KH and RT models [27]. Cavitation and turbulence inside
342 the nozzle greatly influence primary breakup; hence primary breakup should
343 incorporate these effects.
344
345 3.4 Kelvin-Helmholtz-Aerodynamics Cavitation Turbulence (KH-ACT)
346 Model
347 This model is a remodelled conventional KH model with aerodynamics, cavitation,
348 and turbulence effects in the primary breakup. Turbulence inside the nozzle strongly
349 influences the spray breakup, liquid penetration length, and the diameter of child
350 droplets [28]. KH-ACT is quasi dynamically linked to the internal nozzle flow
351 simulations and produces cavitation and turbulence transients at the nozzle exit,
352 acting as starting and boundary conditions [25].
353
354 Turbulence Induced Breakup Model: It calculates turbulent time and length
355 scales by assuming isotropic internal flows and using decay laws:
𝐾(𝑡)1.5 𝐾(𝑡)
356 𝐿𝑡 = 𝐶µ ( ) ; 𝜏𝑡 = 𝐶𝜖 ( )
𝜖(𝑡) 𝜖(𝑡)
357 Here, 𝑲(𝒕) and 𝝐(𝒕) are instantaneous kinetic energy and turbulent dissipation rate,
358 𝑪µ and 𝑪𝝐 are model constants. Assuming isotropic turbulence and no diffusion,
359 convection and production terms in 𝐾- 𝜖 equation, 𝐾(𝑡) and 𝜖(𝑡) can be estimated
360 as below:
1/(1−𝐶𝜖 )
(𝐾0 )𝐶𝜖
361 𝐾(𝑡) = [ ]
𝐾0 (1 + 𝐶µ − 𝐶µ 𝐶𝜖 ) + 𝜖𝑡(𝐶𝜖 − 1)

11
 𝐾(𝑡) 𝐶𝜖
362 𝜖(𝑡) = 𝜖0 [ ]
𝐾0
363 Here, 𝑲𝟎 and 𝝐𝟎 are the starting value of kinetic energy and turbulent dissipation
364 rate at nozzle exit [25].
365
366 Cavitation-Induced Breakup Model: This model considers the cavitation formed
367 inside the nozzle, which may move downstream and reach the nozzle exit. The
368 implosion of bubble improves spray atomization and reduces characteristics
369 breakup timescale. The timescale is decided by the fastest of the two physical
370 phenomena, i.e., the burst of cavitation bubble at the spray periphery or collapse
371 before arriving at:
372 𝜏𝐶𝐴𝑉 = minimum (𝜏𝑐𝑜𝑙𝑙𝑎𝑝𝑠𝑒 , 𝜏𝑏𝑢𝑟𝑠𝑡 )
373 The bubble collapse time is estimated as the time taken to reduce the bubble’s radius
374 from ‘r’ to ‘zero.’
𝜌𝑙
375 𝜏𝑐𝑜𝑙𝑙𝑎𝑝𝑠𝑒 = 0.9145 𝑅𝐶𝐴𝑉 √
𝑝𝑣
376 Here, 𝝆𝒍 and 𝒑𝒗 are the density of fuel and fuel vapour pressure, respectively, 𝑹𝑪𝑨𝑽
377 is the effective radius of an equivalent bubble from the nozzle and is given by,
378 𝑅𝐶𝐴𝑉 = 𝑟ℎ𝑜𝑙𝑒 √1 − 𝐶𝑎 ; 𝑟ℎ𝑜𝑙𝑒 is the exit radius and 𝑪𝒂 is the coefficient of area
379 reduction.
380 The time required for the bubble to reach the jet periphery is given by:
𝑟ℎ𝑜𝑙𝑒 − 𝑅𝐶𝐴𝑉
381 𝜏𝑏𝑢𝑟𝑠𝑡 =
𝑢′𝑡𝑢𝑟𝑏
2 𝐾(𝑡)
382 Here, 𝒖′𝒕𝒖𝒓𝒃 is the turbulent velocity and is given by 𝑢′𝑡𝑢𝑟𝑏 = √
3
383 The length scale for this breakup model is given as 𝐿𝐶𝐴𝑉 = 𝑅𝐶𝐴𝑉
384
385 Aerodynamically Induced Breakup Model: This model calculates the
386 instantaneous length and timescales for each parcel:
387 𝐿𝐾𝐻 = 𝑟 − 𝑟𝐾𝐻
3.276 𝐵1 𝑟
388 𝜏𝐾𝐻 =
𝛺𝐾𝐻 𝛬𝐾𝐻
389 The dominant breakup process is decided by the largest value of the ratio of length
390 and the timescale for each process and is presented mathematically as below:
𝐿𝐴 𝐿𝐾𝐻 (𝑡) 𝐿𝐶𝐴𝑉 (𝑡) 𝐿𝑡 (𝑡)
391 = 𝑚𝑎𝑥 [ ; ; ]
𝜏𝐴 𝜏𝐾𝐻 (𝑡) 𝜏𝐶𝐴𝑉 (𝑡) 𝜏𝑡 (𝑡)
392 If aerodynamically-induced breakup dominates, then the KH model from the
393 previous section is applied for the primary atomization. However, for cavitation or
394 turbulence-induced breakup, below mentioned breakup law is used:
𝑑𝑟 𝐿𝐴
395 = −𝐶𝑇,𝐶𝐴𝑉
𝑑𝑡 𝜏𝐴
396 Here, 𝑪𝑻,𝑪𝑨𝑽 is the model constant
397

12
398 3.5 RT (Rayleigh-Taylor) Model
399 KH instabilities occur when velocity gradients are normal to the interface. However,
400 RT instabilities are initiated due to the density difference of the two fluids and can
401 arise at the circumference. These instability waves originate due to the rapid
402 deceleration of droplets because of the drag force (Figure 8). It is based on
403 investigating the stability of the liquid-ambient air interface when liquid accelerates
404 in a direction normal to the plane [29]. Generally, RT instabilities form when
405 acceleration is in the opposite direction to the density gradient. When the droplet
406 decelerates due to ambient gas drag, it may build up instability at the trailing edge
407 of the droplet. The mathematical equation for the acceleration of droplet governed
408 by the gas drag is given as:
2
3 𝜌𝑔 𝑣𝑟𝑒𝑙
409 |𝐹⃗ | = 𝐶𝐷
8 𝜌𝑙 𝑟
410 Here, 𝒗𝒓𝒆𝒍 is the relative velocity of the droplet w.r.t. the ambient gas and r is the
411 droplet radius.
412 The frequency and wavelength of the fastest growing wave are given below by
413 assuming a linear disturbance growth rate and negligible viscosity.
1/4
2 |𝐹⃗ | |𝐹⃗ | (𝜌𝑙 − 𝜌𝑔 )
414 𝛺𝑅𝑇 = √ [ ]
3 3𝜎
415
3𝜎
416 𝛬𝑅𝑇 = 2𝜋√
|𝐹⃗ | (𝜌𝑙 − 𝜌𝑔 )
417 It was found that acceleration creates the fastest growth of RT instabilities; however,
418 the surface tension of liquid opposes the breakup. Standard RT model defines the
419 growth rate as below:
(𝜌𝑙 − 𝜌𝑔 ) 𝑘𝑅𝑇 3 𝜎
420 𝜔𝑅𝑇 = √𝑘𝑅𝑇 𝑎−
(𝜌𝑙 + 𝜌𝑔 ) 𝜌𝑙 + 𝜌𝑔
421 And, the breakup time (𝑡𝑏𝑢 ) is given by:
1
422 𝑡𝑏𝑢 =
𝛺𝑅𝑇
423 The diameter of child droplet is evaluated from the RT wavelength (𝛬), and breakup
424 occurs when 𝛬 < 𝑑𝑝𝑎𝑟𝑒𝑛𝑡 . Generally, the RT model is used with the KH model and
425 is known as the hybrid KH-RT model.
426
427 3.6 Hybrid/ Modified Kelvin-Helmholtz Rayleigh-Taylor (KH-RT) Model
428 The single breakup model cannot completely capture the spray breakup
429 phenomenon as various types of instabilities act on liquid jets simultaneously.
430 Therefore, a hybrid KH-RT model is used, in which KH and RT models are used to
431 capture breakup near the nozzle and at a particular distance downstream of the
432 nozzle, respectively [30]. KH instabilities occur when velocity gradients are normal
433 to the interface. However, RT instabilities are initiated due to the density difference
434 of the two fluids and can arise at the circumference. Taking account of both

13
435 instabilities provides a breakup phenomenon approaching to practical. In this
436 model, KH-RT instability theory is applied on the droplet surface to resemble a
437 breakup. Here, spray droplets are injected using the ‘blob’ model, where the liquid
438 core is represented by defining droplet size equal to the nozzle hole diameter (Figure
439 9).

440
441 Figure 9: KH-RT model [24]
442
443 In this model, RT and KH breakup models are used in competing modes. The
444 breakup time for both models is calculated for the given set of conditions. The
445 shorter breakup time models are used for further simulation of the breakup [31]. RT
446 model is used at the nozzle region as droplet velocities are higher; whereas, the KH
447 model is applied further downstream. These droplets are traced and applied with the
448 KH instability. Calculations of the diameter of parent droplets at each time interval
449 and diameter reduction rate are done in this model. Primary breakup is modelled
450 using KH instabilities, while secondary breakup is modelled using KH and RT
451 models [32].
452
453 3.7 Taylor Analogy Breakup (TAB) Model
454 TAB model represents an analogy between a spring-mass system with forced
455 oscillations and an oscillating droplet passing through a gaseous atmosphere with a
456 relative velocity of urel (Figure 10). As per the analogy, the force (F) producing
457 oscillations of mass M, which are like aerodynamic forces, deform the droplet. On
458 the other hand, the restoring spring force (Fspring = -kx) is similar to a surface tension
459 force, which resists the droplet deformation and maintains its shape [33]. Thus, it is
460 a model to measure the drop distortion and droplet breakup [34].

14
461
462 Figure 10: TAB model (Adapted from [34])
463
464 Damping force (Fdamper = c𝑥̈ ) represents friction force due to liquid dynamic
465 viscosity. Therefore, the 2nd order differential equation of motion for the spring-
466 mass system is given as:
𝐹 𝑘 𝑐
467 𝑥̈ = − 𝑥 − 𝑥̇
𝑀 𝑀 𝑀
468 Where x = displacement of droplet equator from equilibrium, k is the spring
469 constant, c is the damping coefficient.
470
471 Table 1: TAB Breakup Model Analogy
TAB Breakup Model Analogy
Spring Mass System Oscillating Droplet
𝐹 𝜌 𝑢2
External Force =
𝑀 Droplet Drag = 𝑎
𝜌𝑙 ⅆ
𝑘 𝜎
Restoring force = Surface Tension =
𝑀 𝜌𝑙 ⅆ 3
𝑐 𝑢𝑙
Damping Force = Viscous Force =
𝑀 𝜌𝑙 ⅆ 2
472
473 TAB model can track one oscillation mode, whereas there are several such modes.
474 Several investigators have shown that the TAB model is most appropriate for spray
475 under high fuel injection pressures, where aerodynamic effects are predominant,
476 like CI engines.
477
478 3.8 Enhanced TAB breakup model (ETAB)
479 This model includes the droplet deformation dynamics in the TAB model with a
480 new approach for defining the droplet breakup process. In the TAB model, the
481 radius of product droplets is calculated by the energy balance equation, which
482 generally results in an underprediction of droplet diameter under CI engine
483 conditions. However, the ETAB model utilizes the breakup conditions of the
484 standard TAB model but manages droplet breakup differently. In ETAB, the
485 product droplet formation rate is assumed to be proportional to the number of
486 product droplets, with a proportionality constant depend on the breakup regime.
487 This combines the mass conservation principle and exponential decay law, which
488 relates the mass of product droplets and the breakup time of the parent droplet.

15
489 The rate of change of droplet size is calculated from
𝑟
490 = 𝑒 𝐾𝑏𝑟𝑡
𝑎
491 Here, a and r are the radius of parents and product droplets. Kbr is the breakup
492 constant, which depends on the breakup regime as per We as follows,
493
𝑘1 ω, 𝑖𝑓 We ≤ 𝑊𝑒𝑡
494 𝑘𝑏𝑟 = {
𝑘2 ω √We , 𝑖𝑓 We > 𝑊𝑒𝑡
495 Here, k1 and k2 are constant, ω is droplet oscillation angular velocity, Wet= 80,
496 known as transitional Weber number.
497 The deformation of the droplet surface is calculated as follows:
𝑊𝑒
𝑊𝑒 −𝑡 𝑊𝑒 𝑦̇ (0) 𝑦 (0) − 12
498 𝑦(𝑡) = + 𝑒 𝑡𝑑 {[ 𝑦(0) − ] 𝑐𝑜𝑠ωt + [ + ] sinωt}
12 12 ω ω 𝑡ⅆ
499 The breakup happens when the normalized drop distortion, y(t), exceeds the critical
500 value 1. Details of each parameter are given in reference [34]. The product droplets
501 are initially supplied with velocity components perpendicular to the path of the
502 parent droplet with a value 𝜈⊥ = 𝐴 𝑥̇ , Here, 𝑥̇ is the radial velocity of droplet
503 surface, and A is a constant, which can be calculated from the energy conservation
504 criterion and given as follows:
𝑎
505 𝐴2 = 5𝐶𝐷 /4 + 18 [1 − ] / 𝑊𝑒
𝑟32
506 Here, 𝐶𝐷 = aerodynamic drag coefficient, 𝑟32 = Sauter mean radius. In the standard
507 TAB model, A=1; whereas, for high fuel injection conditions, A = 0.7. To take into
508 account droplet surface stripping at the nozzle exit initially injected parcels have
509 been equipped with a power-law size distribution as per the following equation:
𝑛 + 4 𝑟 𝑛+3
510 𝑔(𝑟) = ( )
𝑟0 𝑟0
511 Here, 𝑟0 = radius of nozzle, n = 0.5 [35]
512
513 4. Overview of Research Studies
514 Spray atomization depends on many interdependent parameters such as fuel
515 injection pressure, spray jet velocity, ambient pressure, fuel temperature, ambient
516 temperature, nozzle geometry, and liquid properties. Many researchers investigated
517 the spray phenomena by simulating engine-like conditions in a constant volume
518 spray combustion chamber [36] [37] [38]. The engine combustion network (ECN)
519 provides high fidelity experimental results of engine-like conditions for diesel and
520 gasoline sprays. For this purpose, ECN has collaborated and worked with many
521 leading universities globally. The database has been generated from the experiments
522 conducted under precisely controlled experimental conditions as per ECN
523 guidelines. This experimental database can validate spray models, the most crucial
524 and mandatory step in the simulations. The accuracy of different spray models
525 matching these experimental results indicates the suitability of models for a
526 particular application. The controlled boundary conditions are extremely helpful for

16
527 model initialization and setup. The uncertainty in the simulations reduces with the
528 increasing quality of validated results. Some widely accepted CFD tools are
529 CONVERGE, KIVA, OpenFoam, AVL FIRE, and AVBP. In addition, numerous
530 researchers have conducted research studies on spray breakup, which are available
531 in the open literature.
532 The droplet formation from a primary breakup happens due to shear instabilities
533 like KH instabilities, whereas secondary breakup happens due to RT instabilities.
534 The hybrid KH-RT has been a preferred model used for fuel injection simulations
535 [39]. Som et al. [25] examined the role of primary breakup modelling on the
536 characterization of compression ignition (CI) engines using the Eulerian-
537 Lagrangian framework. KH and KH-ACT models were used for three-dimension
538 simulations of CI engine-like conditions. Simulated results were compared with X-
539 ray radiography results. They reported finer droplets, reduced liquid penetration
540 length, and more radially dispersed spray with the KH-ACT model due to improved
541 primary breakup resulting from nozzle cavitation and turbulence effects. Simulation
542 results showed good agreement with the experimental results of evaporating and
543 non-evaporating sprays. KH-ACT model has predicted experimental results with
544 minimum deviation for liquid and vapour penetration lengths, spray cone angle,
545 axial velocity, droplet dispersion, and flame lift-off location. Improved primary
546 breakup modelling helped superior prediction in the case of the KH-ACT model.
547 However, most diesel engine simulations employ KH-RT atomization models.
548 Many simulation studies employed KH and KH-RT models for primary and
549 secondary breakup [27] [40]. This modelling methodology is computationally cost-
550 effective and replicates global spray characterization closely. However, studies
551 have shown that cavitation and turbulence inside the nozzle greatly influence near
552 nozzle spray breakup [3].
553

554
555 Figure 11: KH and KH-ACT model effect on (a) liquid penetration and (b) SMD [41]
556
557 Figure 11 exhibited liquid penetration at a fixed condition. It can be observed that
558 internal nozzle turbulence has a major influence on spray atomization. However,
559 Sauter mean diameter (SMD) was less sensitive [41]. A similar conclusion can be
560 drawn from figure 12. This experimental study was conducted in a constant volume

17
561 combustion chamber, where operating conditions were maintained constant: fuel
562 injection pressure (FIP) and fuel temperatures were 1420 bar and 438 K,
563 respectively. The KH and KH-ACT models predicted the overall nature of spray;
564 however, the KH-ACT model predicted higher degree atomization with reduced
565 liquid penetration and SMD [41]. KH model showed higher penetration as it has not
566 included aerodynamic, cavitation, and turbulence effect on the breakup
567 phenomenon, which is included in KH-ACT.

568
569 Figure 12: Comparison of hybrid KH model and KH-ACT model for (a) liquid penetration, (b)
570 vapour penetration length [41]
571
572 Rostami and Moghaddam [42] predicted primary spray breakup length, droplet
573 diameter, and ligaments for diesel, biodiesel, and water using a linear instability
574 model. They reported faster atomization due to increased disturbance by lowering
575 the fuel viscosity and increasing the axial and swirl velocity. Such numerical studies
576 provide a better understanding of spray atomization and how it behaves for various
577 fluids. Generally, the secondary spray breakup modelling assumes that the droplet
578 is moving with the laminar flow. However, the presence of turbulence affects the
579 critical We. Therefore, it results in an inaccurate prediction of the breakup
580 phenomenon. Khaleghi et al. [43] investigated the influence of turbulence on the
581 secondary breakup using a modified model and found good agreement between
582 simulation results with the experimental results. They concluded that turbulence
583 resulted in an earlier breakup, higher evaporation rate, and lower penetration.
584 However, the difference between the models with and without the turbulence was
585 negligible as the ambient gas pressure increased. This could be a probable reason
586 for the acceptable performance of the KH-RT model in diesel engine simulations
587 where the fuel is injected under high-density ambient.
588 However, simplified models (turbulence neglected) may lead to inaccuracies in
589 low-density ambient, as commonly seen in SI engines. Chen et al. [44] numerically
590 investigated the effect of injection angle on the spray structure and droplet gas
591 mixing using the Eularian-Lagrangian framework. The KH-RT model is used for
592 the secondary breakup of droplets and helps understand the mixing processes. Niu
593 et al. [45] simulated droplet atomization using the KH-RT breakup model.
594 Simulation results showed good agreement with the experimental results. Zhao et

18
595 al. [46] performed experiments and simulations to investigate the spray, flow
596 structure and droplet size distribution. Eulerian-Lagrangian approach with KH-RT
597 breakup model was used for the simulations. The simulated results of spray and
598 flow showed good agreement with the experiments. Badra et al. [47] developed a
599 predictive ability for gasoline compression engines with accurate spray
600 characterization using the Eulerian-Lagrangian framework. For this, they used
601 LISA and KH-RT breakup models, which were found to be most appropriate. They
602 studied the effect of different spray parameters by comparing various available
603 spray models. Results showed a comparison of modified KH-RT and LISA models
604 for various chamber pressure (5, 10 bar) and temperature (300, 600 K) conditions.
605 TAB was used as a secondary breakup model with LISA. The liquid penetration
606 length predicted by both models matched very well with the experimental values.
607 However, the modified KH-RT model showed a longer liquid penetration length
608 than the LISA model. However, trends matched qualitatively well with the
609 experimental results. For better understating of spray modelling, further effects of
610 secondary breakup models of KH-RT and TAB have to be examined.
611 Liu et al. [48] studied diesel spray impingement characteristics in small-bore IC
612 engines. LISA was used as a spray breakup model, which resulted in an error of 3%
613 between the experimental and simulation results. Park et al. [49] analyzed gasoline
614 and bioethanol sprays for swirl-type GDI injectors using the LISA model, used
615 extensively for hollow cone spray simulations [50] [51]. The numerical studies
616 predicted spray development, spray morphology and spray breakup approximately.
617 Similar observations were also reported by Gao et al. [52]. Liu et al. [53] simulated
618 spray atomization with a pressure swirl injector using LISA with TAB as a
619 secondary breakup model. They concluded that these models could predict the
620 physical phenomenon of atomization quite well. In addition, results showed the
621 ability of the model to capture local SMD and global spray shapes well. Wang et al.
622 [54] evaluated the effect of deposit on spray behaviour in the Euler Lagrangian
623 framework. They performed simulation using CONVERGE. The blob method was
624 implemented to inject spray into a constant volume chamber. At the same time,
625 spray breakup was predicted by the KH-ACT-RT model. Spray simulation results
626 showed longer penetration, smaller spray cone angle, and larger droplet size on the
627 formation of deposits. Also, simulation results were in good agreement with the
628 experiments.
629 Generally, the TAB model is applied for the low-velocity sprays. It is mainly
630 grounded on the physical mechanisms that occur during the spray droplet breakup.
631 However, it has few shortcomings, which many researchers attempted to overcome.
632 Beatrice et al. [55] concluded that TAB underpredicted the droplet lifetime,
633 reducing its accuracy for high-pressure sprays. However, Bianchi and Pelloni [56]
634 optimized the model constant and validated it with a correlation of Pilch and
635 Erdman [11]. There were also disagreements in the prediction of liquid penetration
636 length. Tanner [57] and Park et al. [58] tried to improve this discrepancy in their
637 study. Matysiak [59] concluded that the TAB model underestimates the disruptive
638 drag force effect. After taking motivation from Matysiak [59], Sula and
639 Papalexandris [33] proposed a modified TAB model, including the influence of drag
640 forces. The model showed the potential to predict global macroscopic spray

19
641 parameters such as liquid and vapour penetrations better. However, vapour
642 penetration was overpredicted compared to the experimental values.
643 Pischke et al. [60] modelled gasoline direct injection using a combined high
644 Eulerian-Lagrangian spray model. Primary and secondary atomization was
645 modelled with a combined LISA-KH-TAB approach. This combined model was
646 validated with the experimental data, and it showed good agreement with the
647 macroscopic and microscopic spray characteristics. Zamani et al. [61] studied the
648 simulation of a spray of GDI multi-hole injector. They concluded that model
649 coefficient validated for single plume could not be used for multi plume spray
650 simulation without modification due to plume-to-plume interactions and pressure
651 drag. They also presented new coefficients for multi-hole injectors. Huh and
652 Gosman [62] developed a model to consider internal turbulent stress on the primary
653 breakup. As per this model, initial disturbance on the liquid-gas interface arises
654 from turbulent flow inside the liquid jet, further growing as per the KH instability.
655 It was found that the Huh Gosman-KH-RT model predicted the spray morphology
656 of multi-hole GDI injectors in a better way than the KH-RT model under various
657 ambient and fuel injection pressures [63].
658

659
660 Figure 13: Comparison of spray morphology for experimental, KHRT, and WF-KH RT
661 simulations [64]
662
663 Li et al. [64] coupled Wu-Faeth (WF) turbulent primary breakup theory with KH-
664 RT instability theory to investigate the effect of nozzle-produced turbulence and
665 aerodynamic force on GDI spray atomization. This study provided insights into the
666 relationship between various mechanisms influencing spray atomization and the
667 method of calibrating the proposed atomization model for the GDI engines. Figure
668 13 shows simulation results of WF-KH-RT and KH-RT with two different sets of
669 model coefficients. It was found that WF KH-RT showed highly resembling
670 morphology against experimental, where both shape and penetration are predicted
671 well.
672 Table 2: Hybrid atomization model [65]

20
 Hybrid Atomization Model

Primary Breakup
WAVE
Low – Medium injection pressure
Turbulence-cavitation-aerodynamic
High injection pressure induced model

Secondary Breakup
TAB
12 < We <16 (Vibrational)
DDB
16 < We <45 (Bag)
DDB+WAVE
45 < We <100 (Chaotic)
WAVE
100 < We <1000 (Stripping)
WAVE+RT
We > 1000 (Catastrophic)
673
674 Table2 represents a hybrid atomization model for different We [65]. The secondary
675 breakup model selection criteria for the vibrational regime was TAB [57]. For bag
676 breakup, the deformation and breakup (DDB) approach was applied [66]. For
677 chaotic breakup, DDB was used with the Wave breakup model. The competition
678 between these two models determines the breakup model that could be used. At a
679 higher We, the RT model was implemented in competition with the KH model [67].
680
681 5. Summary
682 Researchers have done substantial work to model the spray breakup to understand
683 the spray breakup phenomenon in IC engine combustion chambers. There are
684 numerous publications and articles on spray modelling, which is evidence of the
685 success of this approach for many applications. Modelling and simulations have
686 opened doors for developing innovative technologies and engine optimization.
687 However, literature shows that model calibration/ validation with the experimental
688 data by tuning model constants and parameters is essential. The validated models
689 capture the actual spray breakup process quite well. This chapter covers important
690 research studies available in the field related to IC engine sprays. Advanced spray
691 development may result in a superior and improved understanding of the in-cylinder
692 combustion process, which would help reduce the brake-specific fuel consumption
693 and tailpipe emissions.
694
695 6. Future Directions
696 Despite advancements, spray breakup models have provided opportunities to
697 improve modelling aspects to address its drawbacks. There are grey areas of spray
698 breakup, which require attention. A few of them are: (i) the capability to capture the
699 breakup of the entire spray area from dense to dilute, (ii) the ability to predict nozzle
700 turbulence and cavitation and their interdependent effect, (iii) modelling of needle

21
701 deformation and its impact. Spray modelling for applications such as electrostatic
702 sprays, acoustic induced spray breakup could also be explored. Another limitation
703 is difficulty in the inclusion of internal surface roughness of the nozzle in numerical
704 modelling. The capacity of spray modelling to resolve flow field with LES/ DNS
705 models and capturing the changes in fuel properties without any a priori requirement
706 of recalibration. Development of cavitation models, which considers multiple
707 phases, nucleation, and condensation, is required. The sole purpose of spray
708 modelling is to predict the combustion phenomena accurately. Therefore, more
709 research effort is required to evaluate the spray model for combustion studies. The
710 existing spray models often over-predict the effect of cross-flow under swirl
711 supported diesel engine simulations. Furthermore, the impact of large-scale
712 convection on spray dispersion has been scarcely studied. As mentioned in this
713 section, combined experimental and simulation efforts are required to investigate
714 many of these challenging problems.
715
716 References
717 1. Smirnov, N. N., Nikitin, V. F., & Tyurenkova, V. V. (2012). Nonequilibrium
718 diffusion combustion of liquid fuel droplets and sprays modelling. Heat
719 Transfer Research, 43(1). DOI: 10.1615/HeatTransRes.v43.i1.10
720 2. Magnotti, G. M., & Genzale, C. L. (2017). Exploration of turbulent atomization
721 mechanisms for diesel spray simulations. SAE Technical Paper No. 2017-01-
722 0829. DOI: https://doi.org/10.4271/2017-01-0829.
723 3. Arcoumanis, C., Gavaises, M., & French, B. (1997). Effect of Fuel Injection
724 Processes on the Structure of Diesel Sprays. SAE Transactions, 106, 1025-
725 1064. Retrieved June 25, 2021, from http://www.jstor.org/stable/44730737.
726 4. Stiesch, G. (2003). Modelling engine spray and combustion processes. Springer
727 Berlin Heidelberg.
728 5. Von Helldorff, H., & Micklow, G. J. (2019). Primary and Secondary Spray
729 Breakup Modelling for Internal Combustion Engine Applications. Journal of
730 Multidisciplinary Engineering Science and Technology (JMEST), ISSN: 2458-
731 9403, 6(4).
732 6. Web source: https://cfdflowengineering.com/turbulent-multiphase-
733 combustion/. Accessed on 25th May 2021.
734 7. Rao, D. C. K., Karmakar, S., & Basu, S. (2017). Atomization characteristics
735 and instabilities in the combustion of multi-component fuel droplets with high
736 volatility differential. Scientific reports, 7(1), 1-15. DOI:
737 https://doi.org/10.1038/s41598-017-09663-7
738 8. Shao, C., Luo, K., Chai, M., & Fan, J. (2018). Sheet, ligament, and droplet
739 formation in swirling primary atomization. AIP Advances, 8(4), 045211. DOI:
740 https://doi.org/10.1063/1.5017162.
741 9. Shimasaki, S. I., & Taniguchi, S. (2011). Formation of uniformly sized metal
742 droplets from a capillary jet by electromagnetic force. Applied Mathematical
743 Modelling, 35(4), 1571-1580. DOI: https://doi.org/10.1016/j.apm.2010.09.033
744 10. Shao, C., Luo, K., Chai, M., & Fan, J. (2018). Sheet, ligament and droplet
745 formation in swirling primary atomization. AIP Advances, 8(4), 045211. DOI:
746 https://doi.org/10.1063/1.5017162

22
747 11. Pilch, M., & Erdman, C. A. (1987). Use of breakup time data and velocity
748 history data to predict the maximum size of stable fragments for acceleration-
749 induced breakup of a liquid drop. International journal of multiphase flow,
750 13(6), 741-757. DOI: https://doi.org/10.1016/0301-9322(87)90063-2.
751 12. Yang, W., Jia, M., Sun, K., & Wang, T. (2016). Influence of density ratio on
752 the secondary atomization of liquid droplets under highly unstable conditions.
753 Fuel, 174, 25-35. DOI: https://doi.org/10.1016/j.fuel.2016.01.078.
754 13. Baumgarten, C. (2006). Mixture formation in internal combustion engines.
755 Springer Berlin Heidelberg.
756 14. Bekdemir, C., Somers, L. M. T., & De Goey, L. P. H. (2008). Numerical
757 modelling of diesel spray formation and combustion. MS thesis, Eindhoven
758 University of Technology.
759 15. Duronio, F., De Vita, A., Allocca, L., & Anatone, M. (2020). Gasoline direct
760 injection engines–A review of latest technologies and trends. Part 1: Spray
761 breakup process. Fuel, 265, 116948. DOI:
762 https://doi.org/10.1016/j.fuel.2019.116948.
763 16. Davanlou, A., Lee, J. D., Basu, S., & Kumar, R. (2015). Effect of viscosity and
764 surface tension on breakup and coalescence of bicomponent sprays. Chemical
765 engineering science, 131, 243-255. DOI:
766 https://doi.org/10.1016/j.ces.2015.03.057
767 17. Sovani, S. D., Sojka, P. E., & Sivathanu, Y. R. (1999). Prediction of drop size
768 distributions from first principles: the influence of fluctuations in relative
769 velocity and liquid physical properties. Atomization and Sprays, 9(2). DOI:
770 10.1615/AtomizSpr.v9.i2.20
771 18. Jaynes, E. T. (1957). Information theory and statistical mechanics. Physical
772 Review, 106(4), 620. DOI:https://doi.org/10.1103/PhysRev.106.620
773 19. Shrestha, K., Van Strien, J., Singh, N., & Inthavong, K. (2020). Primary
774 breakup and atomization characteristics of a nasal spray. PloS one, 15(8),
775 e0236063. DOI: https://doi.org/10.1371/journal.pone.0236063.
776 20. Deshpande, S. S., Gurjar, S. R., & Trujillo, M. F. (2015). A computational
777 study of an atomizing liquid sheet. Physics of Fluids, 27(8), 082108. DOI:
778 https://doi.org/10.1063/1.4929393.
779 21. Asgarian, A., Heinrich, M., Schwarze, R., Bussmann, M., & Chattopadhyay,
780 K. (2020). Experiments and modelling of the breakup mechanisms of an
781 attenuating liquid sheet. International Journal of Multiphase Flow, 130,
782 103347. DOI: https://doi.org/10.1016/j.ijmultiphaseflow.2020.103347
783 22. Schmidt, D., Nouar, I., Senecal, P., Rutland, J., Martin, J., Reitz, R., &
784 Hoffman, J. (1999). Pressure-Swirl Atomization in the Near Field. SAE
785 Transactions, 108, 471-484. Retrieved June 25, 2021, from
786 http://www.jstor.org/stable/44743386.
787 23. Dombrowski, N., & Johns, W. R. (1963). The aerodynamic instability and
788 disintegration of viscous liquid sheets. Chemical Engineering Science, 18(3),
789 203-214. DOI: https://doi.org/10.1016/0009-2509(63)85005-8.
790 24. Nowruzi, H. (2019). Review on Simulation of Non-reacting Fuel Spray.
791 American Journal of Mechanical Engineering, 7(1), 1-8. DOI:10.12691/ajme-
792 7-1-1.

23
793 25. Som, S., & Aggarwal, S. K. (2010). Effects of primary breakup modelling on
794 spray and combustion characteristics of compression ignition engines.
795 Combustion and Flame, 157(6), 1179-1193. DOI:
796 https://doi.org/10.1016/j.combustflame.2010.02.018.
797 26. Nsikane, D., Mustafa, K., Ward, A., Morgan, R., Mason, D., & Heikal, M.
798 (2017). Statistical Approach on Visualizing Multi-Variable Interactions in a
799 Hybrid Breakup Model under ECN Spray Conditions. SAE International
800 Journal of Engines, 10(5), 2461-2477. Retrieved June 25, 2021, from
801 https://www.jstor.org/stable/26422627.
802 27. Beale, J. C., & Reitz, R. D. (1999). Modelling spray atomization with the
803 Kelvin-Helmholtz/Rayleigh-Taylor hybrid model. Atomization and Sprays,
804 9(6). DOI: 10.1615/AtomizSpr.v9.i6.40.
805 28. Kawaharada, N., Thimm, L., Dageförde, T., Gröger, K., Hansen, H., &
806 Dinkelacker, F. (2020). Approaches for Detailed Investigations on Transient
807 Flow and Spray Characteristics during High-Pressure Fuel Injection. Applied
808 Sciences, 10(12), 4410. DOI: https://doi.org/10.3390/app10124410.
809 29. Hwang, S. S., Liu, Z., & Reitz, R. D. (1996). Breakup mechanisms and drag
810 coefficients of high-speed vaporizing liquid drops. Atomization and Sprays,
811 6(3). DOI: 10.1615/AtomizSpr.v6.i3.60.
812 30. Hossainpour, S., & Binesh, A. R. (2009). Investigation of fuel spray
813 atomization in a DI heavy-duty diesel engine and comparison of various spray
814 breakup models. Fuel, 88(5), 799-805. DOI:
815 https://doi.org/10.1016/j.fuel.2008.10.036.
816 31. Beale J.C. (1999), Modelling Fuel Injection Using the Kelvin-Helmholtz/
817 Rayleigh-Taylor Hybrid Atomization Model In KIVA-3V, MS Thesis in
818 Mechanical Engineering, University of Wisconsin, Madison.
819 32. Senecal, P. K., Richards, K. J., Pomraning, E., Yang, T., Dai, M. Z., McDavid,
820 R. M., ... & Cartesian, A. N. P. C. C. (2007). A New Parallel Cut-Cell Cartesian
821 CFD Code for Rapid Grid Generation Applied to In-cylinder Diesel Engine
822 Simulations, SAE Technical Papers 2007-01-0159. DOI:
823 https://doi.org/10.4271/2007-01-0159.
824 33. Sula, C., Grosshans, H., & Papalexandris, M. V. (2020). Assessment of Droplet
825 Breakup Models for Spray Flow Simulations. Flow, Turbulence and
826 Combustion, 105, 889-914. DOI: https://doi.org/10.1007/s10494-020-00139-9.
827 34. O’Rourke, P. J., & Amsden, A. A. (1987). The TAB method for numerical
828 calculation of spray droplet breakup. SAE Technical Paper 872089. DOI:
829 https://doi.org/10.4271/872089.
830 35. Tanner, F. X., & Weisser, G. (1998). Simulation of liquid jet atomization for
831 fuel sprays by means of a cascade drop breakup model, SAE Technical Paper
832 No. 980808. DOI: https://doi.org/10.4271/980808
833 36. Sonawane, U., Kalwar, A., & Agarwal, A. K. (2020). Microscopic and
834 Macroscopic Spray Characteristics of Gasohols Using a Port Fuel Injection
835 System. SAE Technical Paper No. 2020-01-0324. DOI:
836 https://doi.org/10.4271/2020-01-0324.
837 37. Bardi, M., Payri, R., Malbec, L. M. C., Bruneaux, G., Pickett, L. M., Manin, J.,
838 ... & Genzale, C. L. (2012). Engine combustion network: comparison of spray

24
839 development, vaporization, and combustion in different combustion vessels.
840 Atomization and Sprays, 22(10). DOI: 10.1615/AtomizSpr.2013005837.
841 38. Pickett L. (2008). Engine Modeling User’s Group Meeting at the SAE
842 Congress, Detroit, MI.
843 39. Duret, B., Reveillon, J., Menard, T., & Demoulin, F. X. (2013). Improving
844 primary atomization modelling through DNS of two-phase flows. International
845 Journal of Multiphase Flow, 55, 130-137. DOI:
846 https://doi.org/10.1016/j.ijmultiphaseflow.2013.05.004.
847 40. Reitz, R. D., & Bracco, F. V. (1982). Mechanism of atomization of a liquid jet.
848 The physics of Fluids, 25(10), 1730-1742. DOI:
849 https://doi.org/10.1063/1.863650.
850 41. Bravo, L., & Kweon, C. B. (2014). A review on liquid spray models for diesel
851 engine computational analysis. Army Research Lab Aberdeen Proving Ground
852 MD. Accession Number: ADA603658.
853 42. Rostami, E., & Moghaddam, M.H. (2020). The velocity and viscosity impact
854 on the annular spray atomization of different fuels. Combustion Theory and
855 Modelling, 1-35. DOI: https://doi.org/10.1080/13647830.2020.1845399.
856 43. Khaleghi, H., Farani Sani, H., Ahmadi, M., & Mohammadzadeh, F. (2021).
857 Effects of turbulence on the secondary breakup of droplets in diesel fuel sprays.
858 Proceedings of the Institution of Mechanical Engineers, Part D: Journal of
859 Automobile Engineering, 235(2-3), 387-399. DOI:
860 https://doi.org/10.1177/0954407020958581.
861 44. Chen, S., Xing, Y., Liu, X., & Zhao, L. (2020). Numerical investigation of the
862 effect of the injection angle on the spray structures of an air-blast atomizer.
863 Engineering Computations. DOI: https://doi.org/10.1108/EC-03-2020-0175.
864 45. Niu, Y. Y., Wu, C. H., Huang, Y. H., Chou, Y. J., & Kong, S. C. (2020).
865 Evaluation of breakup models for liquid side jets in supersonic cross flows.
866 Numerical Heat Transfer, Part A: Applications, 79(5), 353-369. DOI:
867 https://doi.org/10.1080/10407782.2020.1847513.
868 46. Zhao, J., Ren, Y., Tong, Y., Lin, W., & Nie, W. (2021). Atomization of a liquid
869 jet in supersonic cross-flow in a combustion chamber with an expanded section.
870 Acta Astronautica, 180, 35-45. DOI:
871 https://doi.org/10.1016/j.actaastro.2020.11.051.
872 47. Badra, J. A., Sim, J., Elwardany, A., Jaasim, M., Viollet, Y., Chang, J., ... &
873 Im, H. G. (2016). Numerical simulations of hollow-cone injection and gasoline
874 compression ignition combustion with naphtha fuels. Journal of Energy
875 Resources Technology, 138(5). DOI: https://doi.org/10.1115/1.4032622.
876 48. Liu, Y., Xiang, Q., Li, Z., Yao, S., Liang, X., & Wang, F. (2018). Experiment
877 and simulation investigation on the characteristics of diesel spray impingement
878 based on droplet impact phenomenon. Applied Sciences, 8(3), 384. DOI:
879 https://doi.org/10.3390/app8030384
880 49. Park, S. H., Kim, H. J., Suh, H. K., & Lee, C. S. (2009). Atomization and spray
881 characteristics of bioethanol and bioethanol blended gasoline fuel injected
882 through a direct injection gasoline injector. International journal of heat and
883 fluid flow, 30(6), 1183-1192. DOI:
884 https://doi.org/10.1016/j.ijheatfluidflow.2009.07.002.

25
885 50. Lee, C. S., Kim, H. J., & Park, S. W. (2004). Atomization characteristics and
886 prediction accuracies of hybrid breakup models for a gasoline direct injection
887 spray. Proceedings of the Institution of Mechanical Engineers, Part D: Journal
888 of Automobile Engineering, 218(9), 1041-1053. DOI:
889 https://doi.org/10.1243/0954407041856746.
890 51. Lucchini, T., D'Errico, G., & Nordin, N. (2005). CFD modelling of gasoline
891 sprays, SAE Technical Paper 2005-24-086. DOI: https://doi.org/10.4271/2005-
892 24-086
893 52. Gao, J., Jiang, D., Huang, Z., & Wang, X. (2005). Experimental and numerical
894 study of high-pressure-swirl injector sprays in a direct injection gasoline
895 engine. Proceedings of the Institution of Mechanical Engineers, Part A: Journal
896 of Power and Energy, 219(8), 617-629. DOI:
897 https://doi.org/10.1243/095765005X31333.
898 53. Liu, R. C., Le, J. L., Song, W. Y., & Yang, S. H. (2016). LISA model for
899 simulation of liquid sheet breakup in swirl injection. In Material Science And
900 Environmental Engineering: The Proceedings of 2016 International Workshop
901 on Material Science and Environmental Engineering (IWMSEE2016) (pp. 653-
902 662). DOI: https://doi.org/10.1142/9789813143401_0071.
903 54. Wang, B., Jiang, Y., Hutchins, P., Badawy, T., Xu, H., Zhang, X., ... &
904 Tafforeau, P. (2017). Numerical analysis of deposit effect on nozzle flow and
905 spray characteristics of GDI injectors. Applied Energy, 204, 1215-1224. DOI:
906 https://doi.org/10.1016/j.apenergy.2017.03.094
907 55. Beatrice, C., Belardini, P., Berteli, C., Camerotti, M., & Cirillo, N. (1995). Fuel
908 Jet Models for Multidimensional Diesel Combustion Calculation: An Update.
909 SAE Transactions, 104, 194-204. Retrieved June 25, 2021, from
910 http://www.jstor.org/stable/44633211.
911 56. Bianchi, G. M., & Pelloni, P. (1999). Modelling the diesel fuel spray breakup
912 by using a hybrid model. SAE Technical Paper 1999-01-0226. DOI:
913 https://doi.org/10.4271/1999-01-0226.
914 57. Tanner, F. (1997). Liquid Jet Atomization and Droplet Breakup Modeling of
915 Non-Evaporating Diesel Fuel Sprays. SAE Transactions, 106, 127-140.
916 Retrieved June 25, 2021, from http://www.jstor.org/stable/44730666.
917 58. Park, J. H., Yoon, Y., & Hwang, S. S. (2002). Improved TAB model for
918 prediction of spray droplet deformation and breakup. Atomization and Sprays,
919 12(4). DOI: 10.1615/AtomizSpr.v12.i4.20.
920 59. Matysiak A. EV zur S tropfenbeladener S in einem V, (2007). Euler-Lagrange
921 Verfahren zur Simulation tropfenbeladener Strömung in einem Verdichtergitte
922 Ph.D. Thesis, Department of Mechanical Engineering, Helmut Schmidt
923 Universität - Universität der Bundeswehr Hamburg.
924 60. Pischke, P., Martin, D., & Kneer, R. (2010). Combined spray model for
925 gasoline direct injection hollow-cone sprays. Atomization and Sprays, 20(4).
926 DOI: 10.1615/AtomizSpr.v20.i4.60
927 61. Zamani, H., Hosseini, V., Afshin, H., Allocca, L., & Baloo, M. (2016). Large
928 Eddy Simulation of GDI Single-hole and Multi-hole Injector Sprays with
929 Comparison of Numerical Break-up Models and Coefficients. Journal of
930 Applied Fluid Mechanics, 9(2).

26
 931 62. Huh, K. Y. (1991). A phenomenological model of diesel spray atomization. In
932 Proc. of The International Conf. on Multiphase Flows’ 91-Tsukuba.
933 63. Li, Z. H., He, B. Q., & Zhao, H. (2014). Application of a hybrid breakup model
934 for the spray simulation of a multi-hole injector used for a DISI gasoline engine.
935 Applied Thermal Engineering, 65(1-2), 282-292. DOI:
936 https://doi.org/10.1016/j.applthermaleng.2013.12.063
937 64. Li, Y., Huang, Y., Luo, K., Liang, M., & Lei, B. (2021). Development and
938 validation of an improved atomization model for GDI spray simulations:
939 Coupling effects of nozzle-generated turbulence and aerodynamic force. Fuel,
940 299, 120871. DOI: https://doi.org/10.1016/j.fuel.2021.120871
941 65. Bella, G., Rocco, V., & Ubertini, S. (2002). Combustion and Spray Simulation
942 of a DI Turbocharged Diesel Engine. SAE Transactions, 111, 2549-2565.
943 Retrieved June 25, 2021, from http://www.jstor.org/stable/44743269.
944 66. Ibrahim, E. A., Yang, H. Q., & Przekwas, A. J. (1993). modelling of spray
945 droplets deformation and breakup. Journal of Propulsion and Power, 9(4), 651-
946 654. DOI: https://doi.org/10.2514/3.23672.
947 67. Patterson, M., & Reitz, R. (1998). Modelling the Effects of Fuel Spray
948 Characteristics on Diesel Engine Combustion and Emission. SAE
949 Transactions, 107, 27-43. Retrieved June 25, 2021, from
950 http://www.jstor.org/stable/44736506.

27

View publication stats