Accepted Manuscript

Atomisation technologies used in spray drying in the dairy industry: A review

Jonathan J. O’Sullivan, Eve-Anne Norwood, James A. O’Mahony, Alan L. Kelly

PII: S0260-8774(18)30364-9

DOI: 10.1016/j.jfoodeng.2018.08.027

Reference: JFOE 9377

To appear in: Journal of Food Engineering

Received Date: 13 May 2018

Accepted Date: 25 August 2018

Please cite this article as: Jonathan J. O’Sullivan, Eve-Anne Norwood, James A. O’Mahony, Alan L.
Kelly, Atomisation technologies used in spray drying in the dairy industry: A review, Journal of Food
Engineering (2018), doi: 10.1016/j.jfoodeng.2018.08.027

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

1 Atomisation technologies used in spray drying in the dairy industry: A review

3 Jonathan J. O’Sullivana,b, Eve-Anne Norwooda,b, James A. O’Mahonya,b, Alan L. Kellya,b*

5 aSchool of Food and Nutritional Sciences, University College Cork, Cork, Ireland

6 bDairy Processing Technology Centre, University College Cork, Cork, Ireland

8 * Corresponding author. Email Address: [email protected]

1
 ACCEPTED MANUSCRIPT

9 Abstract

10 Atomisation is an integral element of the spray-drying process, whereby a bulk liquid

11 feed is converted to discrete droplets, greatly increasing the surface area of the feed liquid and

12 thereby increasing considerably the achievable rates of evaporation of water. These droplets,

13 through evaporation of water in the main dryer chamber, become individual powder particles

14 during the spray-drying process. This review provides a comprehensive examination of the

15 most recent developments in atomisation technology for spray-drying, with a particular focus

16 upon dairy applications (e.g., skim and whole milk powders, casein- and whey-based powders,

17 and fat-filled milk powders). As well as a review of principles of different technologies for

18 atomisation, such as rotary atomisers, pressure nozzle atomisers, pneumatic atomisers,

19 ultrasonic atomisers and electrospray atomisers, the industrial applicability and challenges in

20 use of each approach to atomisation are presented. Approaches for monitoring the atomisation

21 process and other factors that influence atomisation, such as feed composition and key process

22 parameters, are considered to provide a holistic analysis of the atomisation process.

23

24 Keywords: Atomisation, Spray-drying, Rotary atomiser, Pressure nozzle atomiser, Pneumatic

25 atomiser, Novel atomisers

2
 ACCEPTED MANUSCRIPT

26 1. Introduction

27 Spray-drying is a unique unit operation where a bulk fluid stream is converted into discrete

28 solid particles, through the dispersion of the stream, by a process known as atomisation, into

29 individual droplets. These come in intimate contact with a hot gaseous drying medium,

30 resulting in rapid and almost complete moisture evaporation, and thus drying. The application

31 of this process in the dairy industry has been a key enabler in the production of dried milk

32 ingredients (O’Connell and Flynn, 2007). Applications of spray-drying within the dairy

33 industry date from the early 1800s, yet it was not until the 1850s that it became technologically

34 feasible to implement this unit operation industrially, and it was first patented in 1872 by

35 Samuel Percy (Filkova et al., 2014; Percy, 1872).

36 A range of products are produced at large commercial scale by spray-drying in the dairy

37 industry, each with differing characteristics, based on the ratio of protein:fat:carbohydrate, and

38 the type of protein, within a given powder and, to a lesser extent, the process conditions that

39 are employed (Schuck et al., 2016). Dairy products possess an excellent nutritional profile but

40 are inherently perishable, due to their high moisture content (Muir, 1996). As a consequence,

41 a significant proportion of material derived from milk is dried to significantly decrease levels

42 of microbial activity, owing to reduced water activity, and for ease of global distribution and

43 more economic transportation due to reduced overall bulk (O’Connell and Flynn, 2007; Schuck

44 et al., 2016). There has been a progressive increase in the global production of milk products;

45 for instance, between 2012 to 2016 there was a 5.47% increase in production from 771,262 to

46 815,965 thousand tonnes, respectively (FAO, 2016). Thus, it is essential that this greater

47 volume of milk is preserved for the benefit of the global population, in particular in regions

48 where climatic conditions are not conducive for dairy farming and/or are liable to increase the

49 rate of spoilage of liquid dairy systems; for this purpose, manufacture of powder products using

50 spray drying is a widely used option.

3
 ACCEPTED MANUSCRIPT

51 Of the different stages of spray-drying (Fig. 1), one of the most critical is atomisation, as

52 this is where the conversion of the bulk fluid stream into discrete droplets takes place.

53 Atomisation typically occurs at the top of the main chamber of a spray dryer, through a device

54 known as an atomiser (Filkova et al., 2014). The atomiser transforms bulk liquid streams into

55 droplets using one of a number of different approaches, one of the most common of which is

56 pressure nozzle atomisation, whereby the liquid stream is passed through a narrow aperture,

57 generating tandem homogenisation and spray effects (Patel et al., 2009). Other approaches that

58 are used to atomise such liquid streams include rotary wheel nozzles and pneumatic nozzles,

59 and more novel approaches, including ultrasonic nozzles and electrospray technology (Gibbs

60 et al., 2009; Marshall, 1954).

61 There are numerous challenges associated with the atomisation process, ranging from

62 considerations prior to the procurement of the technology itself (e.g., available capital,

63 proposed scale of operation, intended product formats and physical properties, etc.), to its

64 successful operation, routine maintenance and replacement costs. A detailed understanding of

65 the physical properties of the material which is to be atomised and subsequently dried, and the

66 target properties of the resulting powder (e.g., bulk density, flowability and free fat) is required

67 to ensure selection of the most appropriate atomisers (Schuck et al., 2016). However, from a

68 dairy processor’s perspective, the use of a spray dryer will often not be confined to a single

69 product category, and the overall configuration of the spray dryer must be sufficiently robust

70 and flexible to produce the desired range of powders, with the required physical properties and

71 quality attributes (O’Connell and Flynn, 2007). The range of fluids being atomised within the

72 dairy sector have very different physicochemical properties (e.g., density, surface tension and

73 viscosity) and associated challenges depending on their formulations, where streams could be

74 relatively high in protein (e.g., milk protein concentrate), carbohydrate (e.g., lactose, whey

75 protein concentrate 35) or fat (e.g., whole milk powder, fat-filled milk powders) (Crowley et

4
 ACCEPTED MANUSCRIPT

76 al., 2015; Fitzpatrick et al., 2007; Kim et al., 2002; O’Sullivan et al., 2018; O’Sullivan,

77 Schmidmeier et al., 2017; O’Sullivan & O’Mahony, 2016; Roos, 2002). Furthermore,

78 depending on the scale of operation and the commercial value of the final product, the atomiser

79 must be sufficiently robust to ensure continuous production with minimal downtime for

80 maintenance (O’Callaghan and Cunningham, 2005).

81 The aim of this review is to outline the principles of atomisation as an integral component

82 of the spray-drying process, and to critically compare the more commonly-used atomisation

83 technologies. A particular focus has been placed on the industrial relevance (e.g., CapEx,

84 operational cost or OpEx, capacity, etc.) of each of the discussed atomisation technologies

85 within the dairy sector for the efficient generation of droplets of a given size, and thus powders

86 with a desired microstructure (e.g., size, surface composition, etc.), physical properties (e.g.,

87 bulk density) and functionality in use (e.g., solubility, foaming and emulsion stability

88 considerations).

89

90 2. Atomisation technology

91 The aim of atomisation is to increase substantially the surface area of the dairy

92 concentrate feed before it is dried (Filkova et al., 2014). The rate of evaporation is directly

93 proportional to droplet surface area; thus, it is highly desirable to achieve fine atomisation for

94 more efficient operation of the spray dryer, and it is also critical in determining physical

95 properties of the resultant powders, such as bulk density, wettability, dispersability, and

96 solubility (Forny et al., 2011; Hanley, Byrne, et al., 2011; Hanley, Cronin, et al., 2011; Patel

97 et al., 2009).

98 The principal properties of the spray of droplets resulting from atomisation are the mean

99 droplet size and droplet size distribution, which are reflected in the properties of the resultant

5
 ACCEPTED MANUSCRIPT

100 powder (Fig. 2a). These properties depend upon a number of factors, including the atomisation

101 technology employed, operating conditions, and the compositional and physicochemical

102 properties of the atomised liquid (e.g., total solids content, rheological behaviour, and size of

103 particles in the feed stream); key parameters include concentrate viscosity, surface tension and

104 density (Pisecky, 2012). The feed stream to a spray dryer is typically a concentrated liquid feed,

105 which possesses an elevated solids content following evaporation, ranging from 26 to 60%

106 (w/w) depending on the specific product (Schuck et al., 2016), with the achievable solids

107 content of the feed normally determined by the viscosity of, and ability to pump, the

108 concentrated feed.

109 With greater energy density applied during the atomisation process, smaller droplets in

110 the spray are achievable, often with a narrow span (Filkova et al., 2014; Hall et al., 2011;

111 O’Sullivan et al., 2015). Broadly, the particle size distribution of the resultant powder is related

112 to the initial spray in terms of initial drop size. However, there is not necessarily a direct

113 relationship between these two parameters, for a variety of reasons such as particle deflation

114 or expansion and agglomeration, which are dependent on a combination of feed composition,

115 feed physicochemical properties and process conditions (Maa et al., 1997). Fig. 2b shows the

116 different powder morphologies that can be achieved from spray-drying operations, which are

117 highly dependent upon the formulation being dried.

118

119 2.1. Rotary wheel atomisers

120 Rotary wheel atomisers consist of a horizontal disc with or without radial vanes (or

121 grooves), which can be straight or curved (Fig. 3). This type of atomiser is driven by either

122 compressed air or an electrical motor. The liquid is fed into the atomiser centrally and then

123 flows radially outward towards the edge of the atomiser in the form of a thin film of liquid,

6
 ACCEPTED MANUSCRIPT

124 which is dispersed into small individual droplets at the periphery of the wheel. The spray arises

125 from a centrifugal speed in the range 5,000 – 60,000 rpm (tip speeds from 100 to 200 m/s) and

126 forms a wide jet with an initially almost horizontal trajectory, which is referred to as a spray

127 cloud. To better control the dispersion properties, disc atomisers are often made with grooves

128 which allow the liquid feed to emerge with equal flow and pressure conditions in the form of

129 threads (Schröder and Walzel, 1998), leading to a more homogeneous spray, which elongate

130 on emerging from the periphery of the atomiser due to centrifugal acceleration and then

131 disintegrate into droplets. Moreover, this helps to optimise the drying of the droplets by

132 narrowing the droplet size distribution (Kamplade et al., 2014). The droplets produced are

133 spherical due to the influence of the imparted energy, the air turbulence effect, and the liquid

134 feed properties. Large rotary atomisers can handle flow rates of up to 200 ton/h.

135 The sprays resulting from rotary wheel atomisers are highly homogeneous. However,

136 several process factors will affect the spray characteristics and, thus, the quality of the spray-

137 dried product. Firstly, the design of the rotary wheel atomiser, in terms of the number of vanes

138 and their shapes (i.e., curved, circular, oval or rectangular), will influence the wheel capacity

139 and the droplet size and properties, respectively. For example, the use of curved vanes will give

140 rise to higher bulk density powders than those produced with straight vanes (Filkova et al.,

141 2014), due to the de-aeration of the feed liquid while travelling out through the curved vanes.

142 By controlling the flowrate of the liquid concentrate and the speed of the rotor, the

143 droplet size can be modulated. Indeed, use of a higher speed at constant wheel diameter and

144 feed flowrate can lead to smaller droplets, which are easier to dry due to increased rates of

145 evaporation (Baker, 1997). The tip speed of the atomiser is one of the most versatile processing

146 variables, from a practical point of view, as it is an established scale-up factor for a multitude

147 of processes (Hall et al., 2011). Overall, knowledge regarding the feed properties, as discussed

148 below (Section 4), is centrally important in predicting spray properties (Filkova et al., 2014).

7
 ACCEPTED MANUSCRIPT

149

150 2.2. Pressure nozzle atomisers

151 Pressure nozzle atomisers are composed of two internal features: (1) a device to create

152 turbulence within the nozzle head (swirl); and (2) an orifice (typically 0.4 to 4 mm in diameter)

153 through which the liquid feed is vertically discharged (Fig. 4). The liquid feed enters the nozzle

154 core tangentially with a decreasing diameter and leaves the orifice, under pressure, as a conical

155 spray at an angle of 40 to 160°.

156 As the spray cloud is much smaller than that for rotary wheel atomisers, it enables the

157 drying chamber to be taller and narrower, or alternatively, the use of more than one pressure

158 nozzle, with the latter allowing for increased drying capacity. The spray is created by the

159 conversion of the pressure energy applied to the liquid feed into kinetic energy as the liquid

160 passes through the nozzle aperture under pressure, usually in the range 180–220 bar

161 (Verdurmen et al., 2004). The size of the two components of the nozzle (i.e., the swirl chamber

162 and orifice), and the operating pressure, will determine the nozzle capacity and the droplet size

163 within the spray. The usual capacity of these nozzles is up to 400 L/h (Huang et al., 2006). If

164 higher feed flowrates are to be processed, several nozzles may be used in parallel in the drying

165 chamber.

166 The properties of the liquid feed and the imparted energy (i.e., pressure and kinetic

167 energies) will determine the properties of the spray and thus the droplets (size and span): the

168 higher the viscosity of the liquid feeds, the smaller the spray cone at the exit of the atomiser

169 (Chen et al., 1992), while the higher the pressure or the flowrate, the smaller the droplets

170 (Broniarz-Press et al., 2016) due to increased levels of energy density. Geometric parameters

171 also play an important role in determining the spray characteristics (Rizk and Lefebvre, 1987).

172 Rashad et al. (2016) showed that the spray angle and the Sauter mean diameter (𝐷3,2) of

8
 ACCEPTED MANUSCRIPT

𝐷𝑠
173 droplets were influenced by the ratio of swirl chamber to discharge orifice diameters ( 𝐷0),

𝐿
174 the length to diameter ratio of the swirl chamber ( 𝑠 𝐷𝑠) and the length to diameter ratio of the

𝐿
175 discharge orifice ( 0 𝐷0) using water as the test liquid. Indeed, the spray cone angle decreased

𝐷𝑠
176 in a continuous manner when 𝐷0 increased, whereas an optimal value was determined in

𝐿0
177 the range of the studied conditions for 𝐷3,2. An increase of 𝐷0 led to an increase in 𝐷3,2

178 values, while the spray cone angle reached an optimum, associated with the size distribution

179 of the resultant spray of droplets (enhancement of the rate of heat and mass transfer). Finally,

𝐿𝑠
180 careful control of 𝐷𝑠 led to optima in the spray cone angle and 𝐷3,2 values, in both cases

181 the lowest possible values.

182

183 2.3. Pneumatic nozzles

184 Pneumatic nozzles, also known as two-fluid nozzles, consist of a tube with double entry

185 flows of air, vapour and/or liquid. The liquid feed can be mixed with the air inside or outside

186 the body of the nozzle (Fig. 5). With pneumatic nozzles, the liquid feed stream is broken into

187 droplets upon contact with a second fluid, which is usually compressed air or steam, and

188 normally gives rise to a spray pattern with an angle ranging from 20 to 60° (Filkova et al.,

189 2014). The spray arises from kinetic energy of air in the range of 1-3 kPa. The air flow brings

190 high frictional forces over liquid surfaces that create atomisation arising from the disintegration

191 of the liquid feed. Sarrate et al. (2015) showed that varying the liquid feed properties modified

192 the particle size and morphology after spray drying with a pneumatic nozzle; a low-viscosity

193 feed will result in low mean droplet size and a high homogeneity, whereas a high-viscosity

194 feed will give rise to a higher mean droplet size and lower homogeneity (Filkova et al., 2014).

9
 ACCEPTED MANUSCRIPT

195 Indeed, viscosity and surface tension increase the amount of energy required to atomize the

196 spray, and an increase in either of these properties will typically increase the drop size.

197 These devices are highly flexible and enable manipulation of the drop diameter in the

198 spray by varying the gas:liquid ratio, and are consequently often employed within pilot-scale

199 spray drying systems. It is also possible to maintain the droplet size distribution when changing

200 the flow rate by adjusting the flow rate of the compressed air. The highest capacity of a

201 pneumatic nozzle is around 1,000 L/h. Three-fluid and four-fluid variants of these nozzles also

202 exist. The first is composed of two feeding atomising gases, each fed separately (Fig. 5), and

203 is able to atomise one kind of feed, while the second type of system allows the atomisation of

204 two distinct feeds. However, the latter two types are more likely to be used in the

205 pharmaceutical industry for both greater uniformity in droplet size and the capacity to

206 manufacture droplets with incompatible phases (Broadhead et al., 1992).

207

208 2.4. Critical assessment of rotary wheel, pressure nozzle and pneumatic nozzle atomisers

209 In order to precisely measure the droplet size and size distribution generated by a given

210 atomiser, a direct measuring technique should be used. Albrecht et al. (2013) proposed the use

211 of non-intrusive laser-based approaches, specifically laser Doppler and phase Doppler

212 measurement techniques for this purpose. The latter technique is based on the assumption that

213 droplets are spherical and allow the measurement of two parameters, i.e., the droplet size

214 distribution and the velocity at which the liquid feed is ejected from the atomiser. However,

215 this technique is not appropriate for rotary wheel atomisers due to the broad spray produced

216 using this type of atomiser. In this regard, several authors have developed alternative methods

217 based on imaging and allowing wide size distributions, with large and non-spherical droplets

218 to be measured in scattered spray systems (Glover et al., 1995; Zhou et al., 2015). Other

10
 ACCEPTED MANUSCRIPT

219 mechanical or optical-based techniques can also be used (Dodge et al., 1987; Dodge, 1987;

220 Lee Black et al., 1996). Table 1 shows a comparison of the equations employed to determine

221 the D3,2 of the resultant spray, in addition to the principal advantages and disadvantages, for

222 rotary, pressure and pneumatic nozzles.

223

224 2.5. Novel atomisation approaches

225 There are numerous liquids which are challenging to atomise using the aforementioned

226 conventional atomiser configurations (i.e., rotary wheel, pressure nozzles and pneumatic

227 nozzles) and, to achieve atomisation of such liquids, novel approaches may be required. Such

228 liquids may be, for example, highly viscous, contain long-chain high molecular weight

229 macromolecules, which may yield a more cohesive fluid, or demonstrate highly non-

230 Newtonian rheological behaviour (e.g., shear-banding, an inconsistency in the response to a

231 constant shear rate); such fluids may, rather than forming discrete droplets, form filaments.

232 Moreover, another desirable feature is the high degree of monodispersity achievable using

233 certain novel atomisation approaches.

234 The most recent developments in this field of atomisation are focused upon ultrasonic

235 atomisation and high voltage-induced sprays, as well as single-drop drying studies (Chen,

236 2004; Fu et al., 2012; Sadek, Schuck, et al., 2015; Schutyser et al., 2012). However, it should

237 be noted that investigations of these atomisation approaches are limited predominately at this

238 time to fundamental research, rather than industrial implementation, owing primarily to issues

239 of scale, perceived complexity of the setup and the inherent complexity of liquid dairy systems,

240 as well as high costs and lack of available equipment.

241

242 2.5.1. Ultrasonic nozzle atomisation

11
 ACCEPTED MANUSCRIPT

243 The effect of atomisation from ultrasonic waves was first reported by Wood and Loomis

244 in 1928, where the phenomenon was originally referred to as ultrasonic fogging, due to the

245 formation of mists and sprays of the liquids subjected to ultrasonic processors (Antonevich,

246 1959; Söllner, 1936; Wood and Loomis, 1928). Ultrasound involves generation of acoustic

247 longitudinal waves of higher frequency than the threshold of human auditory detection (≥ 16

248 kHz), which propagate through liquids in differing manners depending on the frequency (Soria

249 and Villamiel, 2010). Ultrasound can be categorised into two distinct types, based on the

250 frequency employed, as either:

251 (1) High-frequency (> 100 kHz), low-intensity (< 1 W/cm2) ultrasound, which is typically

252 used for the physicochemical evaluation of a range of microstructures, both food and

253 non-food systems (McClements, 1995).

254 (2) Low-frequency (20 – 100 kHz), high-intensity (10 – 100 W/cm2) ultrasound, which is

255 often used for the alteration, development and generation of microstructures, again for

256 both food and non-food applications (O’Sullivan et al., 2017; O’Sullivan et al., 2018b).

257 Ultrasonic atomisers can be categorised as either gas-driven (whistle type) or

258 electrically driven, with the latter being based on the piezoelectric effect within ultrasonic horns

259 (Rajan and Pandit, 2001). The generation of droplets by ultrasound is dependent upon both the

260 properties of the liquid and the operating conditions of the ultrasonic processor. The

261 physicochemical factors which primarily determine drop formation are the surface tension at

262 the gas-liquid boundary, bulk viscosity and liquid density (Bittner & Kissel, 1999; Briceño-

263 Gutierrez et al., 2015).

264 The processing variables which effect droplet size are feed rate of liquid to the nozzle

265 and acoustic intensity, which is dependent upon the specific probe (i.e., surface area) and both

266 the ultrasonic frequency and amplitude (Rajan and Pandit, 2001). In brief, smaller droplets can

12
 ACCEPTED MANUSCRIPT

267 be achieved at lower flowrates to the nozzle, arising from increased residence times at the

268 atomiser surface, allowing for a greater exposure to acoustic energy. In addition, at higher

269 acoustic intensities, more energy is provided to the liquid, which manifests as production of

270 smaller droplets. Moreover, these effects can be combined to have a synergistic effect in

271 achieving small droplets with a narrow size distribution (O’Sullivan et al., 2017).

272 A schematic representation of a typical ultrasonic spray-head is depicted in Fig. 6

273 (Freitas et al., 2004; Rajan & Pandit, 2001). The key component of the setup is a device known

274 as a sonotrode, which has two main components, a transducer and a tip. The transducer converts

275 electrical energy into mechanical energy, which is achieved through use of piezoelectric

276 materials, such as quartz or lithium zirconate titanates, whereby these materials oscillate in

277 response to electrical energy, leading to mechanical vibrations. These mechanical vibrations

278 emanate from the tip of the sonotrode in the form of acoustic waves (Martini, 2013; O’Sullivan

279 et al., 2017). To achieve atomisation using this approach, the liquid is fed centrally to the tip

280 via a channel within the sonotrode itself, whereby it comes in direct contact with the tip, coating

281 it with a thin liquid film. The acoustic waves can then atomise the liquid into a spray of droplets

282 (Avvaru et al., 2006; Ju et al., 2008).

283 Practical applications of ultrasonic atomisation for spray drying within the food sector,

284 and more specifically within the dairy industry, have been limited, due to both the complex

285 makeup of liquid concentrates within the dairy sector and general equipment considerations

286 (i.e., comparatively high CapEx, energy utilisation, heat dissipation challenges, etc.).

287 Furthermore, this technology has been predominantly limited to laboratory-scale processing,

288 rather than pilot or industrial scales. Studies have been conducted investigating the

289 encapsulation efficiency of dried fish oil emulsions, stabilised by whey protein, prepared using

290 ultrasonic atomisation, and comparing this to other approaches; powders prepared utilising an

291 ultrasonic-based atomisation approach demonstrated a reduced encapsulation efficiency in

13
 ACCEPTED MANUSCRIPT

292 comparison to pressure nozzles (either 1 or 2 channels), yet exhibited a narrower size

293 distribution in comparison to the other atomisation approaches investigated (Klaypradit and

294 Huang, 2008; Legako and Dunford, 2010).

295 One of the distinct advantages of ultrasonic nozzles in comparison to other atomisation

296 approaches is that a larger aperture can be used, which is invariably less prone to blockages

297 than the small orifices required for pressurised flow in nozzle atomisers. However, to the

298 authors’ knowledge, this approach is not used at an industrial scale within the dairy sector,

299 owing to the smaller size of powder particles from this approach, which adversely affect

300 rehydration properties, and general issues associated with processing, in terms of lower

301 throughput levels, heat generation and perceived complexity issues (Filkova et al., 2014; Rajan

302 and Pandit, 2001).

303

304 2.5.2. High Voltage-Induced Spraying (Electrospray)

305 High voltage-induced spraying, also referred to as electrospray or electrostatic

306 atomisation, is a process which utilises electrical energy to disperse a liquid stream into a spray

307 of droplets. Droplets produced using this approach are highly charged, which enhances droplet

308 dispersion and prevents droplet coalescence (Jaworek, 2007). Furthermore, the spray of

309 droplets produced using this method demonstrates a high degree of monodispersity, and the

310 size of droplets can be precisely controlled through careful adjustment of the fluid flowrate

311 and/or the electrical conductivity of the liquid (Gañán-Calvo et al., 1997). Other factors which

312 affect drop formation through electronisation (i.e., high voltage-induction) are permittivity

313 (i.e., the measure of how an electric field affects a dielectric medium), density and rheological

314 properties of the liquid, tension at the gas-liquid surface, and charge density of the droplets

315 from a cone-jet meniscus (Tang and Gomez, 1996).

14
 ACCEPTED MANUSCRIPT

316 In order to achieve an electrospray, high voltage is applied to the liquid using an

317 apparatus known as an emitter, whereby the liquid, upon reaching the emitter, will form a

318 Taylor cone (Fig. 7a), emitting a liquid jet through its apex (Meesters et al., 1992). A Taylor

319 cone is a cone of liquid from which liquid drops emanate above a given threshold voltage for

320 a given liquid, whereby the slightly rounded tip inverts and emits a spray of liquid. This is

321 referred to as a cone-jet, and is the beginning of the electrospray process, whereby a voltage

322 greater than the threshold voltage is necessary to achieve a stable cone-jet. The magnitude of

323 the threshold voltage is dependent predominately upon the properties (i.e., conductivity) of the

324 liquid, rather than the configuration of the nozzle (Yarin et al., 2001).

325 The configuration of an electrospray atomisation approach is depicted in Fig. 7b. The

326 setup consists of the liquid to be atomised, which is fed from a storage vessel to the emitter and

327 spray head using compressed air. The emitter is connected to a high voltage power supply,

328 which induces drop formation, and in many cases it is necessary to earth the resultant spray

329 from a safety perspective (Gañán-Calvo et al., 1997).

330 Despite the capacity of these systems to convert a fluid stream into monodisperse liquid

331 droplets, to the best of the authors’ knowledge there are no examples of this technology being

332 employed for atomisation applications at laboratory, pilot or industrial scales within the dairy

333 sector. The reasons for this include (when compared to other atomisation approaches) the

334 limited throughput, the comparatively high complexity, and the limited number of fluids which

335 exhibit the capacity to be electro-sprayed, as it is not possible to electrospray complex

336 formulations incorporating several components (e.g., proteins, fats, carbohydrates, etc.), as

337 well as the challenging physical properties (e.g., non-Newtonian flow behaviour of fluids to be

338 atomised), which are commonly observed with dairy applications. Thus, the electrospray

339 atomisation approach is more commonly encountered in the production of ceramic powders,

340 aerosol standards and space thrusters (Gañán-Calvo et al., 1997; Jaworek, 2007). The few

15
 ACCEPTED MANUSCRIPT

341 examples of electrospray technology employed for food applications are limited to that of

342 coating (e.g., aluminium coated with sunflower oil using 20 µm droplets to produce a

343 hydrophobic film), rather than for use in spray-drying systems, and, moreover, are quite limited

344 due to the aforementioned issues of scale and complexity of setup (Bhushani &

345 Anandharamakrishnan, 2014; Khan et al., 2012).

346

347 2.6. Selection criteria for atomisers

348 For the majority of industrial dairy spray-drying operations, the selection of a suitable

349 atomiser is typically a choice between rotary atomisers or pressure nozzles, as there is

350 comparatively limited use of the other nozzle systems presented earlier (i.e., pneumatic,

351 ultrasonic nozzles and electrospray) (Schuck et al., 2016). Pneumatic nozzles are typically used

352 in pilot-scale spray dryers or in some cases for lecithin-dosing systems in the manufacture of

353 instant powders and for coating or wet agglomeration of powders. Nevertheless, the criteria for

354 selecting a given atomisation technology is based on a variety of considerations, which include

355 CapEx, OpEx, availability, flexibility of application, and the target particle size distribution

356 (PSD) of the final powder product. Other key factor to consider in the selection of an atomiser

357 is the practical aspects of its use, such as the run times, capability for cleaning in place (CIP),

358 and safety considerations (e.g., fire and explosion regulations). These types of considerations

359 are critically important when determining the suitability of a given system, both in terms of the

360 initial investment and the associated payback times, and the inherent running costs, in terms of

361 both energy utilisation and maintenance. However, practical considerations around processing

362 must be considered to ensure a robust and flexible system, which is capable of atomising a

363 wide range of dairy liquid feed streams, and its overall plant usability.

16
 ACCEPTED MANUSCRIPT

364 The predominant rationale for use of rotary wheel and pressure nozzle atomisers within

365 the dairy sector is that they are robust, reliable and well-established systems, and demonstrate

366 most of the aforementioned characteristics, in terms of capacity, ease of use, capability of CIP,

367 etc. On the other hand, for the alternative atomisation technologies, there is typically a higher

368 associated cost, and lack of familiarity, which in turn could potentially lead to capacity

369 constraints during powder manufacture. Thus, for these systems to be considered commercially

370 and industrially feasible by dairy processors, further research is required to assess their

371 practical implementation for powder manufacture.

372

373 3. Approaches for monitoring atomisation processes

374 Atomisation plays a critically important, integral role in the spray-drying process, and

375 thus its consistent operation is essential. Consequently, it is necessary to monitor the

376 atomisation operation through a variety of means to mitigate against production downtime and

377 also to facilitate optimal control of critical quality attributes of the finished product properties.

378 When considering methods that can be employed for monitoring the atomisation process, three

379 different approaches can broadly be employed: (1) characterisation of process properties (e.g.,

380 feed and air flowrate, air relative humidity, stream pressure, etc.) of streams entering and

381 exiting the spray dryer (Anderson, 2001), (2) control of parameters (e.g., air temperature)

382 within the spray dryer (Nath and Satpathy, 1998); and (3) use of computational fluid dynamics

383 (CFD) to gain a better understanding of material flow through the process (Fletcher et al.,

384 2006).

385

386 3.1. Inlet and outlet stream process parameters

17
 ACCEPTED MANUSCRIPT

387 Process variables of the inlet and outlet streams to a spray dryer can indicate whether

388 the process is performing as designed. One of the key processing parameters which is

389 monitored industrially to give an indication of atomisation performance is pressure at a number

390 of points in the system. This parameter should be maintained within specified ranges during

391 the drying process to ensure consistent product quality and maximise overall efficiency of the

392 drying process. Variations in pressure, either for inlet or outlet streams, give indications of

393 performance of a spray dryer system, in particular atomiser operation; for example, a spike in

394 pressure on the inlet to the dryer could indicate potential blockages within the system; pressure

395 sensors on the feed line could indicate a build-up in pressure due a blockage in the system (Fig.

396 1).

397 Inline viscosity measurement is another means of controlling the inlet liquid stream to

398 be atomised. The atomisation step requires droplets with high surface area:mass ratio (i.e.,

399 minimisation of droplet size), which can be achieved by variation of the pressure, which is

400 influenced by both the liquid feed viscosity and volumetric flowrate. Indeed, the droplet size

401 varies directly with the viscosity at a factor ranging between 0.17-0.25 (Masters, 2002). There

402 are two ways to achieve in-line viscosity measurements:

403 (1) Direct measurement through use of an inline viscometer, of which there are numerous

404 examples, such as the Couette type, whereby the main challenges associated with inline

405 viscometers are the response times for measurements for a given volume of fluid, and the ability

406 to measure viscosity within pressurised (150-250 bar) liquid feed lines to the spray-dryer.

407 (2) Indirect measurement of viscosity based on measurements of either the electrical current of

408 the concentrate pump or calculation of the theoretical viscosity from pressure drop data using

409 the Hagen-Poiseuille equation (O’Sullivan et al., 2017, 2018,). Schuck et al. (2005) proposed

18
 ACCEPTED MANUSCRIPT

410 that the electrical current recorded from the feed pump could be used as a method to monitor

411 the viscosity of dairy concentrates before spray drying.

412 Electrical capacitance tomography (ECT) is a novel approach which demonstrates the

413 capacity for assessing the performance of drying processes in real-time, such as spray-drying,

414 fluidised beds and pneumatic conveying of powders (Jaworski & Dyakowski, 2001; Yang,

415 1996). The principles of ECT rely upon measurement of variations in the dielectric permittivity

416 distribution (i.e., a measure of resistance when forming an electric field in a given medium)

417 across a planar section of electrodes within a given process, whereby each electrode takes

418 measurements with respect to all other planar electrodes (Yang, 2010; Yang & Peng, 2003).

419 Variation in the dielectric permittivity from a given baseline (e.g., that of air) across a given

420 planar section indicate the presence of material passing through that section (Yang & Peng,

421 2003). Thus, for the case of spray drying, the measured material would be droplets produced

422 by atomisation. Furthermore, several planes of electrodes could be employed to give a

423 composite, proxy three-dimensional view within a given section of the process, which allows

424 for the tracking of a given droplet and provides a detailed understanding of both the mechanics

425 of motion (i.e., trajectory) and drying (i.e., change in dielectric permittivity as a function of

426 moisture loss on drying) for a given droplet. ECT has been successfully used to understand the

427 fluid mechanics with pneumatic conveying and circulating fluidised bed systems (Liu et al.,

428 2005; Zhu et al., 2003).

429 One of the key considerations of these process control systems is their positioning in

430 the process with respect to the inlet to the spray dryer, and thus proximity to the atomiser nozzle

431 and the outlet. The further these sensors are from these points, the greater the response times

432 will be and, in addition, the more difficult it will be to assess the root cause of the problem.

433 Furthermore, in an ideal integrated system control, the inline sensors on the inlet feed would

434 work in parallel with some of the approaches discussed in the next section regarding monitoring

19
 ACCEPTED MANUSCRIPT

435 systems within the dryer, whereby, if a fault were detected through use of a given sensor, the

436 specific reason could be corrected using integrated advanced process control (APC) systems.

437

438 3.2. Monitoring of atomisers within the spray-drying unit

439 There are generally two types of monitoring approaches which can be used for the

440 atomisation process within the spray-drier or chamber, i.e., (1) visual approaches, which are

441 relatively labour-intensive, and (2) automated approaches utilising data acquisition from within

442 the dryer chamber.

443 Visual approaches that can be used to monitor the atomisation process within a spray

444 drying chamber include the use of sight glasses and/or closed circuit television (CCTV)

445 systems (Listiohadi et al., 2005). For best use, these are located near the top of the chamber in

446 view of the nozzles. However, both techniques are limited as sight glasses may foul during

447 drying operations and CCTV cannot trigger a corrective intervention (Fig. 1).

448 The infra-red (IR) camera adds an extra level of safety and process control through

449 temperature mapping within close proximity to the nozzle and electronic alarm points, and

450 actions can be triggered in advance of the development of processing challenges (e.g., a

451 blocked nozzle), thereby maintaining the quality of the product and reducing further the

452 possibility of overheating (Fig. 1). GEA Niro has developed a visual monitoring system for

453 spray dryers called the SprayEye. The SprayEye system links IR cameras to a pre-warning

454 system that detects high temperature increases in the powder within the spray dryer before one

455 would notice it using a normal camera. The SprayEye was first made of conventional digital

456 cameras to monitor the performance of spray nozzles. These cameras show a continuous picture

457 of the spray zone, allowing for the checking of leak occurrences or nozzle blockages.

20
 ACCEPTED MANUSCRIPT

458 Therefore, the spray pattern of the nozzles into the drying chamber can be maintained at an

459 optimum level through careful monitoring.

460 Automated approaches can be more accurate and user-friendly than visual systems;

461 however, they are normally associated with higher cost, in terms of CapEx, which play a key

462 role when determining procurement of a specific dryer configuration and associated control

463 measures (O’Callaghan and Cunningham, 2005; Schuck et al., 2016).

464

465 3.3. Computational fluid dynamics (CFD) assessments of spray dryers

466 Computational fluid dynamics (CFD) is a powerful technique which provides

467 information in terms of both material and energy flow through a given process, and over the

468 past decade has garnered particular interest as a means to transition to an in silico (i.e., studies

469 conducted solely through use of computers) methodology for process development. For the

470 specific application of spray drying, CFD can be used to understand the heat distribution within

471 a dryer, the material flow through the atomiser, and identify where potential issues could arise,

472 such as dead-zones in terms of both airflow and temperature (Fig. 8) (Fletcher et al., 2006).

473 Jaskulski et al., (2015) developed a three-dimensional (3D) CFD model of the agglomeration

474 of droplets and particles in a counter-current spray-drying process. Their model takes into

475 account droplet coalescence and shrinkage and the hydrodynamic segregation of particles in

476 order to accurately calculate the mass balance of the discrete phase. Two agglomeration areas

477 were observed in the chamber; the first comprised of wet particle agglomeration in the

478 atomization zone, while the second involved dry agglomeration above the air inlets, due to the

479 intensive mixing of particle streams. The elaborated model also allows prediction of the final

480 PSD of the powder in the spray dryer.

21
 ACCEPTED MANUSCRIPT

481 Despite this technique providing significant information about the process, there are

482 several limitations, which include the high computer processing power required to run

483 simulations, the time-consuming nature of these runs, taking in some cases several weeks, and

484 the necessity for both highly experienced CFD modellers to develop representative simulations

485 and developed mathematical models that accurately reflect physicochemical and biochemical

486 processes. For the specific case of the monitoring of atomisation, CFD cannot be used as a

487 standalone technique, but rather provide information as to where issues could arise, and assist

488 with the troubleshooting process in expediting resolution of issues.

489

490 4. Other atomisation-related considerations

491 Aside from the aforementioned process conditions, which can be varied, depending

492 upon the specific nozzle, other factors can affect atomisation. Broadly, these can be categorised

493 as (1) process parameters for the inlet stream to the dryer, and (2) the feed composition.

494 Variation in these will alter how the atomiser performs and, thus, the physical properties (e.g.,

495 size distribution) of the resultant powder.

496

497 4.1. Effect of process and feed parameters on atomisation

498 As discussed in Section 3.1, there are several process parameters which affect

499 atomisation and can provide information regarding the process during a given production cycle.

500 The two main process parameters of interest are temperature in general terms and feed flow-

501 rate, as these will both alter the physical properties of the liquid stream to the atomiser to

502 differing degrees.

22
 ACCEPTED MANUSCRIPT

503 For the production of dairy powders, typical preheating temperatures for the feed are

504 ~75-85°C, which have the effect of reducing concentrate viscosity (Gharsallaoui et al., 2007).

505 The temperature of the inlet stream may also need to be maintained below a certain level,

506 depending on the product, as certain dairy solutions are prone to thermal aggregation (e.g.,

507 heat-induced whey protein aggregation), which could result in blockages. These include whey-

508 protein-dominant systems, whereby thermal denaturation of the protein occurs at temperatures

509 in excess of 68oC (Damodaran, 1997). However, denaturation in these systems is not an

510 instantaneous process (Moakes et al., 2015), allowing for sufficient time for evaporation

511 operations to increase the solids content of the inlet stream, to maximise efficiency of the

512 dryer, and the drying itself within the spray drying chamber, leading to minimal whey protein

513 denaturation with adequate control of process parameters (Haque, 2015).

514 The feed flow rate is considered to be the primary parameter to adjust in order to balance

515 the throughput and the optimal drying of a liquid concentrate considering the drying capacity

516 of the respective spray dryer set-up (Miller and Gil, 2012). The feed flowrate governs the

517 drying history of the feed by influencing the residence time of a particle according to the

518 relative humidity and the outlet air temperature (Paudel et al., 2013). In this regard, the feed

519 flow rate for a given solution is fixed to enable sufficient drying of the particles before they

520 make contact with the spray dryer wall, and is additionally related to other factors such as

521 atomiser type, dryer chamber geometry, etc (Vehring et al., 2007). It has also been shown that,

522 for the case of drying of mannitol solutions, the product was almost completely dried in the

523 upper part of the chamber at low feed flow rates, whereas the breaking strength (the greatest

524 stress that a material is capable of withstanding without rupture) was decreased at low rates

525 (Littringer et al., 2012).

526

23
 ACCEPTED MANUSCRIPT

527 4.2. Effect of feed properties on atomisation

528 4.2.1. Feed composition

529 The specific formulation of different components within a feed stream has a significant

530 effect on the performance of the atomiser. The four main components within dairy-based

531 products which play a role in drying are protein, fat, carbohydrate (i.e., lactose) and water. The

532 properties of the particles resulting from atomisation are strongly dependent on the respective

533 intrinsic material properties of each component. For example, Elversson & Millqvist-Fureby,

534 (2005) showed that atomisation of amorphous or crystalline carbohydrates leads to hollow or

535 porous particles, respectively. Furthermore, the presence of high concentrations of lactose

536 within an inlet stream to the dryer can potentially result in the blockage of nozzles due to large

537 lactose crystals. Surface activity and mechanical properties of components are also known to

538 affect particle geometry and morphology (Kawakami et al., 2010; Sadek et al., 2013).

539 Drying of a whey protein isolate (WPI) droplet leads to a smooth semi-spherical particle

540 shape, whereas drying of a native phosphocaseinate droplet results in a twisted and wrinkled

541 particle without fractures (Sadek et al., 2015a). Surface activity of the components of the feed

542 also influence their distribution into the particle during the drying process (i.e., migration of

543 certain components from the core to the surface of the droplet/powder particle), resulting in

544 differences between the particle surface and bulk concentrations (Briggs, 1994; Kim et al.,

545 2006). Both protein and fat accumulate at the particle surface during spraying and drying,

546 whereas lactose content is reported to be enriched in the core of the particle (Drapala et al.,

547 2017; Kelly et al., 2015; Nijdam and Langrish, 2006). As a further example, whole milk

548 powder (WMP) particles are mainly covered with fat followed by lactose and then protein,

549 whereas particles in SMP are mainly covered with proteins, with interstitial lactose crystals

550 (Murrieta-Pazos et al., 2012). Furthermore, the type of product being processed will strongly

551 influence the moisture content, and thus the efficiency of the dryer, as the ability of components

24
 ACCEPTED MANUSCRIPT

552 to bind to water governs the drying kinetics and thus influences particle characteristics (Ameri

553 and Maa, 2006).

554

555 4.2.2. Feed concentration

556 The concentration and viscosity of the feed are known to affect the geometric mean

557 diameter of spray-dried particles. Viscous concentrated solutions induce large droplets, that

558 form larger powder particles (Elversson and Millqvist-Fureby, 2005; Walton and Mumford,

559 1999) due to increasing breakup length of the liquid jets during atomisation. In addition, the

560 feed concentration may affect the solvent removal (Paudel et al., 2013), which is water in the

561 case of dairy products. The rate of evaporation from the droplet surface decreases with

562 increasing feed viscosity, due to higher fraction and/or molecular weight of polymers present

563 in solution (Wu et al., 2011), such as proteins.

564

565 5. Conclusions and future trends

566 While several of the aforementioned atomisation technologies are well established and

567 have been in active industrial use for over a century, numerous advances have been made in

568 recent decades in understanding the fundamentals responsible for the generation of a spray of

569 discrete droplets from a bulk liquid stream and the factors associated with their formation, such

570 as formulation and processing conditions.

571 Many technological approaches are capable of achieving atomisation, such as rotary

572 wheel atomisers, pressure nozzles, pneumatic nozzles and ultrasonic nozzles, and there are

573 benefits and limitations associated with each, which will influence their selection in an

574 industrial setting, broadly categorised in terms of the functionality (i.e., efficient droplet

575 formation) and cost considerations (i.e., CapEx and OpEx) of a given atomiser.

25
 ACCEPTED MANUSCRIPT

576 Given that atomisation plays a central role in spray drying, continual monitoring of the

577 system is of paramount importance. This monitoring may be achieved using a number of

578 different approaches, such as inline characterisation of the process streams, or direct

579 observation of the nozzle, where these approaches are, at times, used in conjunction with CFD

580 modelling of the process to highlight potential challenges during operation. In addition to

581 monitoring of the process, a detailed understanding of the properties of the material to be dried

582 is essential, as these will affect atomisation. Such properties include rheological flow

583 behaviour, and the potential for concentrated dairy streams at elevated temperatures to undergo

584 protein denaturation, or possess lactose crystals, depending on the specific formulation and

585 processing history of the liquid being dried.

586 Although numerous advances have been made in understanding the droplet formation

587 and spray behaviour mechanisms of a variety of atomisation technologies, this understanding

588 is predominately at a laboratory scale. Although studies are being conducted at pilot-scale,

589 further work is required to fully understand the relationship between nozzle selection, design

590 criteria, process parameters and product formulation to allow for the effective utilisation of

591 atomisation technologies in a variety of product categories within the dairy sector.

592 Furthermore, greater investigation of novel atomiser designs could allow for unique powder

593 morphologies and consequently variation in resultant powder properties.

594

595

596

597 Acknowledgements

598 The authors would like to acknowledge the Dairy Processing Technology Centre

599 (DPTC), an Enterprise Ireland initiative, for financial support and permission to publish this

26
 ACCEPTED MANUSCRIPT

600 work. This work was supported by the Irish State through funding from the Technology Centres

601 programme (Grant Number TC/2014/0016).

602 References

603 Albrecht, H.-E., Damaschke, N., Borys, M., Tropea, C., 2013. Laser Doppler and Phase

604 Doppler Measurement Techniques. Springer Science & Business Media.

605 Ameri, M., Maa, Y.-F., 2006. Spray Drying of Biopharmaceuticals: Stability and Process

606 Considerations. Dry. Technol. 24, 763–768.

607 https://doi.org/10.1080/03602550600685275

608 Anderson, N.G., 2001. Practical Use of Continuous Processing in Developing and Scaling Up

609 Laboratory Processes. Org. Process Res. Dev. 5, 613–621.

610 https://doi.org/10.1021/op0100605

611 Antonevich, J.N., 1959. Ultrasonic Atomization of Liquids. Trans. IRE Prof. Gr. Ultrason.

612 Eng. 6, 6–15. https://doi.org/10.1109/T-PGUE.1959.29249

613 Anu Bhushani, J., Anandharamakrishnan, C., 2014. Electrospinning and electrospraying

614 techniques: Potential food based applications. Trends Food Sci. Technol. 38, 21–33.

615 https://doi.org/10.1016/j.tifs.2014.03.004

616 Avvaru, B., Patil, M.N., Gogate, P.R., Pandit, A.B., 2006. Ultrasonic atomization: Effect of

617 liquid phase properties. Ultrasonics 44, 146–158.

618 https://doi.org/10.1016/j.ultras.2005.09.003

619 Baker, C.G.J., 1997. Industrial Drying of Foods. Springer Science & Business Media.

620 Bittner, B., Kissel, T., 1999. Ultrasonic atomization for spray drying: a versatile technique for

621 the preparation of protein loaded biodegradable microspheres. J. Microencapsul. 16,

622 325–341. https://doi.org/10.1080/026520499289059

623 Briceño-Gutierrez, D., Salinas-Barrera, V., Vargas-Hernández, Y., Gaete-Garretón, L.,

624 Zanelli-Iglesias, C., 2015. On the Ultrasonic Atomization of Liquids. Phys. Procedia 63,

27
 ACCEPTED MANUSCRIPT

625 37–41. https://doi.org/10.1016/j.phpro.2015.03.006

626 Broadhead, J., Edmond Rouan, S.K., Rhodes, C.T., 1992. The spray drying of

627 pharmaceuticals. Drug Dev. Ind. Pharm. 18, 1169–1206.

628 https://doi.org/10.3109/03639049209046327

629 Broniarz-Press, L., Włodarczak, S., Matuszak, M., Ochowiak, M., Idziak, R., Sobiech, Ł.,

630 Szulc, T., Skrzypczak, G., 2016. The effect of orifice shape and the injection pressure on

631 enhancement of the atomization process for pressure-swirl atomizers. Crop Prot. 82, 65–

632 74. https://doi.org/10.1016/j.cropro.2016.01.005

633 Chen, S.K., Lefebvre, A.H., Rollbuhler, J., 1992. Factors Influencing the Effective Spray

634 Cone Angle of Pressure-Swirl Atomizers. J. Eng. Gas Turbines Power 114, 97–103.

635 https://doi.org/10.1115/1.2906313

636 Chen, X.D., 2004. Heat-Mass Transfer and Structure Formation During Drying of Single

637 Food Droplets. Dry. Technol. 22, 179–190.

638 Crowley, S. V., Desautel, B., Gazi, I., Kelly, A.L., Huppertz, T., O’Mahony, J.A., 2015.

639 Rehydration characteristics of milk protein concentrate powders. J. Food Eng. 149, 105–

640 113. https://doi.org/10.1016/j.jfoodeng.2014.09.033

641 Damodaran, S., 1997. Food proteins: an overview, in: Damodaran, S., Paraf, A. (Eds.), Food

642 Proteins and Their Applications. Marcel Dekker, New York, pp. 1–24.

643 Dodge, L.G., 1987. Comparison of performance of drop-sizing instruments. Appl. Opt. 26,

644 1328–1341. https://doi.org/10.1364/AO.26.001328

645 Dodge, L.G., Rhodes, D.J., Reitz, R.D., 1987. Drop-size measurement techniques for sprays:

646 comparison of Malvern laser-diffraction and Aerometrics phase/Doppler. Appl. Opt. 26,

647 2144–2154.

648 Drapala, K.P., Auty, M.A.E., Mulvihill, D.M., O’Mahony, J.A., 2017. Influence of emulsifier

649 type on the spray-drying properties of model infant formula emulsions. Food Hydrocoll.

28
 ACCEPTED MANUSCRIPT

650 69, 56–66. https://doi.org/10.1016/J.FOODHYD.2016.12.024

651 Elversson, J., Millqvist-Fureby, A., 2005. Particle size and density in spray drying-effects of

652 carbohydrate properties. J. Pharm. Sci. 94, 2049–2060. https://doi.org/10.1002/jps.20418

653 FAO, 2016. Milk and Milk Products.

654 Filkova, I., Huang, L.X., Mujumdar, A.S., 2014. Industrial Spray Drying Systems, in:

655 Mujumdar, A.S. (Ed.), Handbook of Industrial Drying. pp. 215–256.

656 Fitzpatrick, J.J., Barry, K., Cerqueira, P.S.M., Iqbal, T., O’Neill, J., Roos, Y.H., 2007. Effect

657 of composition and storage conditions on the flowability of dairy powders. Int. Dairy J.

658 17, 383–392. https://doi.org/10.1016/J.IDAIRYJ.2006.04.010

659 Fletcher, D.F., Guo, B., Harvie, D.J.E., Langrish, T.A.G., Nijdam, J.J., Williams, J., 2006.

660 What is important in the simulation of spray dryer performance and how do current CFD

661 models perform? Appl. Math. Model. 30, 1281–1292.

662 https://doi.org/10.1016/j.apm.2006.03.006

663 Forny, L., Marabi, A., Palzer, S., 2011. Wetting, disintegration and dissolution of

664 agglomerated water soluble powders. Powder Technol. 206, 72–78.

665 https://doi.org/10.1016/j.powtec.2010.07.022

666 Freitas, S., Merkle, H.P., Gander, B., 2004. Ultrasonic atomisation into reduced pressure

667 atmosphere—envisaging aseptic spray-drying for microencapsulation. J. Control.

668 Release 95, 185–195. https://doi.org/10.1016/j.jconrel.2003.11.005

669 Fu, N., Woo, M.W., Chen, X.D., 2012. Single Droplet Drying Technique to Study Drying

670 Kinetics Measurement and Particle Functionality: A Review. Dry. Technol. 30, 1771–

671 1785.

672 Gañán-Calvo, A.M., Dávila, J., Barrero, A., 1997. Current and droplet size in the

673 electrospraying of liquids. Scaling laws. J. Aerosol Sci. 28, 249–275.

674 https://doi.org/10.1016/S0021-8502(96)00433-8

29
 ACCEPTED MANUSCRIPT

675 Gharsallaoui, A., Roudaut, G., Chambin, O., Voilley, A., Saurel, R., 2007. Applications of

676 spray-drying in microencapsulation of food ingredients: An overview. Food Res. Int. 40,

677 1107–1121. https://doi.org/10.1016/J.FOODRES.2007.07.004

678 Gibbs, B., Kermasha, S., Alli, I., Mulligan, C., 2009. Encapsulation in the food industry: a

679 review. Int. J. Food Sci. Nutr. 50, 213–224.

680 Glover, A.R., Skippon, S.M., Boyle, R.D., 1995. Interferometric laser imaging for droplet

681 sizing: a method for droplet-size measurement in sparse spray systems. Appl. Opt. 34,

682 8409–8421. https://doi.org/10.1364/AO.34.008409

683 Hall, S., Cooke, M., El-Hamouz, A., Kowalski, A.J., 2011. Droplet break-up by in-line

684 Silverson rotor–stator mixer. Chem. Eng. Sci. 66, 2068–2079.

685 Hanley, K.J., Byrne, E.P., Cronin, K., Oliveira, J.C., O’Mahony, J.A., Fenelon, M.A., 2011a.

686 Effect of pneumatic conveying parameters on physical quality characteristics of infant

687 formula. J. Food Eng. 106, 236–244. https://doi.org/10.1016/j.jfoodeng.2011.04.029

688 Hanley, K.J., Cronin, K., O’Sullivan, C., Fenelon, M.A., O’Mahony, J.A., Byrne, E.P.,

689 2011b. Effect of composition on the mechanical response of agglomerates of infant

690 formulae. J. Food Eng. 107, 71–79. https://doi.org/10.1016/j.jfoodeng.2011.05.042

691 Haque, M., 2015. Study on denaturation of whey protein isolate and its control in convective

692 and spray drying processes. RMIT.

693 Huang, L.X., Kumar, K., Mujumdar, A.S., 2006. A comparative study of a spray dryer with

694 rotary disc atomizer and pressure nozzle using computational fluid dynamic simulations.

695 Chem. Eng. Process. Process Intensif. 45, 461–470.

696 https://doi.org/10.1016/j.cep.2005.11.004

697 Jaskulski, M., Wawrzyniak, P., Zbiciński, I., 2015. CFD Model of Particle Agglomeration in

698 Spray Drying. Dry. Technol. 33, 1971–1980.

699 https://doi.org/10.1080/07373937.2015.1081605

30
 ACCEPTED MANUSCRIPT

700 Jaworek, A., 2007. Micro- and nanoparticle production by electrospraying. Powder Technol.

701 176, 18–35. https://doi.org/10.1016/j.powtec.2007.01.035

702 Jaworski, A.J., Dyakowski, T., 2001. Application of electrical capacitance tomography for

703 measurement of gas-solids flow characteristics in a pneumatic conveying system. Meas.

704 Sci. Technol. 12, 1109–1119. https://doi.org/10.1088/0957-0233/12/8/317

705 Ju, J., Yamagata, Y., Ohmori, H., Higuchi, T., 2008. High-frequency surface acoustic wave

706 atomizer. Sensors Actuators A Phys. 145, 437–441.

707 https://doi.org/10.1016/j.sna.2008.01.001

708 Kamplade, J., Mack, T., Küsters, A., Walzel, P., 2014. Break-Up of Threads From Laminar

709 Open Channel Flow Influenced by Cross-Wind Gas Flow V002T20A001.

710 https://doi.org/10.1115/FEDSM2014-21243

711 Kawakami, K., Sumitani, C., Yoshihashi, Y., Yonemochi, E., Terada, K., 2010. Investigation

712 of the dynamic process during spray-drying to improve aerodynamic performance of

713 inhalation particles. Int. J. Pharm. 390, 250–259.

714 https://doi.org/10.1016/j.ijpharm.2010.02.018

715 Kelly, G.M., O’Mahony, J.A., Kelly, A.L., Huppertz, T., Kennedy, D., O’Callaghan, D.J.,

716 2015. Influence of protein concentration on surface composition and physico-chemical

717 properties of spray-dried milk protein concentrate powders. Int. Dairy J. 51, 34–40.

718 https://doi.org/10.1016/j.idairyj.2015.07.001

719 Khan, M.K.I., Schutyser, M.A.I., Schroën, K., Boom, R., 2012. The potential of

720 electrospraying for hydrophobic film coating on foods. J. Food Eng. 108, 410–416.

721 https://doi.org/10.1016/j.jfoodeng.2011.09.005

722 Kim, E.H.-J., Chen, X.D., Pearce, D., 2002. Surface characterization of four industrial spray-

723 dried dairy powders in relation to chemical composition, structure and wetting property.

724 Colloids Surfaces B Biointerfaces 26, 197–212.

31
 ACCEPTED MANUSCRIPT

725 Klaypradit, W., Huang, Y.-W., 2008. Fish oil encapsulation with chitosan using ultrasonic

726 atomizer. LWT - Food Sci. Technol. 41, 1133–1139.

727 https://doi.org/10.1016/j.lwt.2007.06.014

728 Lee Black, D., McQuay, M.Q., Bonin, M.P., 1996. Laser-based techniques for particle-size

729 measurement: A review of sizing methods and their industrial applications. Prog. Energy

730 Combust. Sci. 22, 267–306. https://doi.org/10.1016/S0360-1285(96)00008-1

731 Legako, J., Dunford, N.T., 2010. Effect of Spray Nozzle Design on Fish Oil-Whey Protein

732 Microcapsule Properties. J. Food Sci. 75, E394–E400. https://doi.org/10.1111/j.1750-

733 3841.2010.01708.x

734 Listiohadi, Y.D., Hourigan, J.A., Sleigh, R.W., Steele, R.J., 2005. An exploration of the

735 caking of lactose in whey and skim milk powders. Aust. J. Dairy Technol. 60, 207–213.

736 Littringer, E.M., Mescher, A., Eckhard, S., Schröttner, H., Langes, C., Fries, M., Griesser, U.,

737 Walzel, P., Urbanetz, N.A., 2012. Spray Drying of Mannitol as a Drug Carrier—The

738 Impact of Process Parameters on Product Properties. Dry. Technol. 30, 114–124.

739 https://doi.org/10.1080/07373937.2011.620726

740 Liu, S., Chen, Q., Wang, H.G., Jiang, F., Ismail, I., Yang, W.Q., 2005. Electrical capacitance

741 tomography for gas–solids flow measurement for circulating fluidized beds. Flow Meas.

742 Instrum. 16, 135–144. https://doi.org/10.1016/j.flowmeasinst.2005.02.013

743 Maa, Y.-F., Costantino, H.R., Nguyen, P.-A., Hsu, C.C., 1997. The Effect of Operating and

744 Formulation Variables on the Morphology of Spray-Dried Protein Particles. Pharm.

745 Dev. Technol. 2, 213–223. https://doi.org/10.3109/10837459709031441

746 Marshall, W.R., 1954. Atomization and spray drying. American Institute of Chemical

747 Engineers.

748 Martini, S., 2013. Sonocrystallization of Fats, 1st ed. Springer US, New York.

749 McClements, D.J., 1995. Advances in the application of ultrasound in food analysis and

32
 ACCEPTED MANUSCRIPT

750 processing. Trends Food Sci. Technol. 6, 293–299. https://doi.org/10.1016/S0924-

751 2244(00)89139-6

752 Meesters, G.M.H., Vercoulen, P.H.W., Marijnissen, J.C.M., Scarlett, B., 1992. Generation of

753 micron-sized droplets from the Taylor cone. J. Aerosol Sci. 23, 37–49.

754 https://doi.org/10.1016/0021-8502(92)90316-N

755 Miller, D.A., Gil, M., 2012. Spray-Drying Technology, in: III, R.O.W., Watts, A.B., Miller,

756 D.A. (Eds.), Formulating Poorly Water Soluble Drugs, AAPS Advances in the

757 Pharmaceutical Sciences Series. Springer New York, pp. 363–442.

758 Moakes, R.J.A., Sullo, A., Norton, I.T., 2015. Preparation and characterisation of whey

759 protein fluid gels: The effects of shear and thermal history. Food Hydrocoll. 45, 227–

760 235. https://doi.org/10.1016/j.foodhyd.2014.11.024

761 Muir, D.D., 1996. The shelf-life of dairy products: 1. Factors influencing raw milk and fresh

762 products. Int. J. Dairy Technol. 49, 24–32. https://doi.org/10.1111/j.1471-

763 0307.1996.tb02616.x

764 Murrieta-Pazos, I., Gaiani, C., Galet, L., Scher, J., 2012. Composition gradient from surface

765 to core in dairy powders: Agglomeration effect. Food Hydrocoll. 26, 149–158.

766 https://doi.org/10.1016/j.foodhyd.2011.05.003

767 Nath, S., Satpathy, G.R., 1998. A systematic approach for investigation of spray drying

768 processes. Dry. Technol. 16, 1173–1193. https://doi.org/10.1080/07373939808917459

769 Nijdam, J.J., Langrish, T.A.G., 2006. The effect of surface composition on the functional

770 properties of milk powders. J. Food Eng. 77, 919–925.

771 https://doi.org/10.1016/j.jfoodeng.2005.08.020

772 O’Callaghan, D., Cunningham, P., 2005. Modern process control techniques in the

773 production of dried milk products – a review. Lait 85, 335–342.

774 https://doi.org/10.1051/lait:2005021

33
 ACCEPTED MANUSCRIPT

775 O’Connell, J.E., Flynn, C., 2007. The manufacture and application of casein-derived

776 ingredients, in: Hui, Y.H. (Ed.), Handbook of Food Products Manufacturing. John Wiley

777 & Sons, New Jersey, pp. 557–593.

778 O’Sullivan, J., Murray, B., Flynn, C., Norton, I., 2015. Comparison of batch and continuous

779 ultrasonic emulsification processes. J. Food Eng. 167(B), 141–121.

780 O’Sullivan, J., Park, M., Beevers, J., Greenwood, R., Norton, I., 2017. Applications of

781 ultrasound for the functional modification of proteins and nanoemulsion formation: A

782 review. Food Hydrocoll.

783 O’Sullivan, J.J., Drapala, K.P., Kelly, A.L., O’Mahony, J.A., 2018a. The use of inline high-

784 shear rotor-stator mixing for preparation of high-solids milk protein-stabilised oil-in-

785 water emulsions with different protein:fat ratios. J. Food Eng. 222, 218–225.

786 O’Sullivan, J.J., Espinoza, C.J.U., Mihailova, O., Alberini, F., 2018b. Characterisation of

787 flow behaviour and velocity induced by ultrasound using particle image velocimetry

788 (PIV): Effect of fluid rheology, acoustic intensity and transducer tip size. Ultrason.

789 Sonochem. 48, 218–230.

790 O’Sullivan, J.J., O’Mahony, J.A., 2016. Food Ingredients, in: Reference Module in Food

791 Science. Elsevier, Amsterdam, Netherlands, pp. 1–3.

792 O’Sullivan, J.J., Schmidmeier, C., Drapala, K.P., O’Mahony, J.A., Kelly, A.L., 2017.

793 Monitoring of pilot-scale induction processes for dairy powders using inline and offline

794 approaches. J. Food Eng. 197, 9–16.

795 Palzar, S., Dubois, C., Gianfrancesco, A., 2012. Generation of Product Structures During

796 Drying of Food Products. Dry. Tech. 30, 97-105.

797 Patel, R.P., Patel, M.P., Suthar, A.M., 2009. Spray Drying Technology: an overview. Indian

798 J. Sci. Technol. 2, 44–47. https://doi.org/10.17485/IJST/2009/V2I10/30719

799 Paudel, A., Worku, Z.A., Meeus, J., Guns, S., Van den Mooter, G., 2013. Manufacturing of

34
 ACCEPTED MANUSCRIPT

800 solid dispersions of poorly water soluble drugs by spray drying: Formulation and

801 process considerations. Int. J. Pharm., Poorly Soluble Drugs 453, 253–284.

802 https://doi.org/10.1016/j.ijpharm.2012.07.015

803 Percy, S., 1872. Improvement in drying and concentrating liquid substances by atomizing.

804 US125406.

805 Pisecky, J., 2012. Handbook of Milk Powder Manufacture. GEA Niro, Copenhagen.

806 Rajan, R., Pandit, A.B., 2001. Correlations to predict droplet size in ultrasonic atomisation.

807 Ultrasonics 39, 235–255. https://doi.org/10.1016/S0041-624X(01)00054-3

808 Rashad, M., Yong, H., Zekun, Z., 2016. Effect of geometric parameters on spray

809 characteristics of pressure swirl atomizers. Int. J. Hydrogen Energy, 4th International

810 Conference on Energy Engineering and Environment Engineering (ICEEEE2016), 15-

811 16 April 2016, Hong Kong, China 41, 15790–15799.

812 https://doi.org/10.1016/j.ijhydene.2016.04.037

813 Rizk, N.K., Lefebvre, A.H., 1987. Prediction of Velocity Coefficient and Spray Cone Angle

814 for Simplex Swirl Atomizers. Int. J. Turbo Jet Engines 4, 65–74.

815 https://doi.org/10.1515/TJJ.1987.4.1-2.65

816 Roos, Y.H., 2002. Importance of glass transition and water activity to spray drying and

817 stability of dairy powders. Lait 82, 475–484.

818 Sadek, C., Pauchard, L., Schuck, P., Fallourd, Y., Pradeau, N., Le Floch-Fouéré, C., Jeantet,

819 R., 2015a. Mechanical properties of milk protein skin layers after drying: Understanding

820 the mechanisms of particle formation from whey protein isolate and native

821 phosphocaseinate. Food Hydrocoll. 48, 8–16.

822 https://doi.org/10.1016/j.foodhyd.2015.01.014

823 Sadek, C., Schuck, P., Fallourd, Y., Pradeau, N., Le Floch-Fouéré, C., Jeantet, R., 2015b.

824 Drying of a single droplet to investigate process–structure–function relationships: a

35
 ACCEPTED MANUSCRIPT

825 review. Dairy Sci. Technol. 95, 771–794.

826 Sadek, C., Tabuteau, H., Schuck, P., Fallourd, Y., Pradeau, N., Le Floch-Fouéré, C., Jeantet,

827 R., 2013. Shape, Shell, and Vacuole Formation during the Drying of a Single

828 Concentrated Whey Protein Droplet. Langmuir 29, 15606–15613.

829 https://doi.org/10.1021/la404108v

830 Sarrate, R., Ticó, J.R., Miñarro, M., Carrillo, C., Fàbregas, A., García-Montoya, E., Pérez-

831 Lozano, P., Suñé-Negre, J.M., 2015. Modification of the morphology and particle size of

832 pharmaceutical excipients by spray drying technique. Powder Technol. 270, Part, 244–

833 255. https://doi.org/10.1016/j.powtec.2014.08.021

834 Schröder, T., Walzel, P., 1998. Design of Laminar Operating Rotary Atomizers under

835 Consideration of the Detachment Geometry. Chem. Eng. Technol. 21, 349–354.

836 https://doi.org/10.1002/(SICI)1521-4125(199804)21:4<349::AID-CEAT349>3.0.CO;2-

837 9

838 Schuck, P., Jeantet, R., Bhandari, B., Chen, X.D., Perrone, Í.T., de Carvalho, A.F., Fenelon,

839 M., Kelly, P., 2016. Recent advances in spray drying relevant to the dairy industry: A

840 comprehensive critical review. Dry. Technol. 07373937.2016.1233114.

841 https://doi.org/10.1080/07373937.2016.1233114

842 Schuck, P., Méjean, S., Dolivet, A., Beaucher, E., Famelart, M.-H., 2005. Pump amperage: a

843 new method for monitoring viscosity of dairy concentrates before spray drying. Lait 85,

844 361–367. https://doi.org/10.1051/lait:2005014

845 Schutyser, M.A.I., Perdana, J., Boom, R.M., 2012. Single droplet drying for optimal spray

846 drying of enzymes and probiotics. Trends Food Sci. Technol. 27, 73–82.

847 Söllner, K., 1936. The mechanism of the formation of fogs by ultrasonic waves. Trans.

848 Faraday Soc. 32, 1532–1536. https://doi.org/10.1039/TF9363201532

849 Soria, A.C., Villamiel, M., 2010. Effect of ultrasound on the technological properties and

36
 ACCEPTED MANUSCRIPT

850 bioactivity of food: a review. Trends Food Sci. Technol. 21, 323–331.

851 https://doi.org/http://dx.doi.org/10.1016/j.tifs.2010.04.003

852 Tang, K., Gomez, A., 1996. Monodisperse Electrosprays of Low Electric Conductivity

853 Liquids in the Cone-Jet Mode. J. Colloid Interface Sci. 184, 500–511.

854 https://doi.org/10.1006/jcis.1996.0645

855 Vehring, R., Foss, W.R., Lechuga-Ballesteros, D., 2007. Particle formation in spray drying. J.

856 Aerosol Sci. 38, 728–746. https://doi.org/10.1016/j.jaerosci.2007.04.005

857 Verdurmen, R.E.M., Menn, P., Ritzert, J., Blei, S., Nhumaio, G.C.S., Sonne Sørensen, T.,

858 Gunsing, M., Straatsma, J., Verschueren, M., Sibeijn, M., Schulte, G., Fritsching, U.,

859 Bauckhage, K., Tropea, C., Sommerfeld, M., Watkins, A.P., Yule, A.J., Schønfeldt, H.,

860 2004. Simulation of Agglomeration in Spray Drying Installations: The EDECAD

861 Project. Dry. Technol. 22, 1403–1461. https://doi.org/10.1081/DRT-120038735

862 Walton, D.E., Mumford, C.J., 1999. The Morphology of Spray-Dried Particles: The Effect of

863 Process Variables upon the Morphology of Spray-Dried Particles. Chem. Eng. Res. Des.

864 77, 442–460. https://doi.org/10.1205/026387699526296

865 Wood, W.R., Loomis, A.L., 1928. The physical and biological effects of high-frequency

866 sound-waves of great intensity. J. Franklin Inst. 205, 151–153.

867 https://doi.org/10.1016/S0016-0032(28)92379-0

868 Wu, J.X., Yang, M., Berg, F. van den, Pajander, J., Rades, T., Rantanen, J., 2011. Influence

869 of solvent evaporation rate and formulation factors on solid dispersion physical stability.

870 Eur. J. Pharm. Sci. 44, 610–620. https://doi.org/10.1016/j.ejps.2011.10.008

871 Yang, W., 2010. Design of electrical capacitance tomography sensors. Meas. Sci. Technol.

872 21, 042001. https://doi.org/10.1088/0957-0233/21/4/042001

873 Yang, W.Q., 1996. Hardware design of electrical capacitance tomography systems. Meas.

874 Sci. Technol. 7, 225–232. https://doi.org/10.1088/0957-0233/7/3/003

37
 ACCEPTED MANUSCRIPT

875 Yang, W.Q., Peng, L., 2003. Image reconstruction algorithms for electrical capacitance

876 tomography. Meas. Sci. Technol. 14, R1–R13. https://doi.org/10.1088/0957-

877 0233/14/1/201

878 Yarin, A.L., Koombhongse, S., Reneker, D.H., 2001. Taylor cone and jetting from liquid

879 droplets in electrospinning of nanofibers. J. Appl. Phys. 90, 4836–4846.

880 https://doi.org/10.1063/1.1408260

881 Zhou, W., Hu, J., Feng, M., Yang, B., Cai, X., 2015. Study on imaging method for measuring

882 droplet size in large sprays. Particuology 22, 100–106.

883 https://doi.org/10.1016/j.partic.2014.12.003

884 Zhu, K., Madhusudana Rao, S., Wang, C.-H., Sundaresan, S., 2003. Electrical capacitance

885 tomography measurements on vertical and inclined pneumatic conveying of granular

886 solids. Chem. Eng. Sci. 58, 4225–4245. https://doi.org/10.1016/S0009-2509(03)00306-3

887

38
 ACCEPTED MANUSCRIPT

1 Figure captions

2 Fig. 1. Overview of the spray-drying unit operation and its associated process control systems

3 for monitoring the performance of the atomisation process.

4 Fig. 2. (a) Factors effecting powder manufacture for spray drying applications, including

5 properties of the inlet stream, atomisation conditions, and process parameters associated with

6 the drying air, and (b) potential powder morphologies yielded from spray-drying.

7 Fig. 3. Rotary atomizing wheels with (A) straight radial vanes, (B) curved vanes and (C)

8 bushing (Pisecky, 2012, reproduced with permission from GEA Niro).

9 Fig. 4. Design of a pressure nozzle with (A) a grooved core and (B) a swirl chamber (Pisecky,

10 2012, reproduced with permission from GEA Niro).

11 Fig. 5. Pneumatic nozzle with (a) internal air / liquid mixing and (b) external mixing (Pisecky,

12 2012, reproduced with permission from GEA Niro).

13 Fig. 6. A detailed schematic of an ultrasonic atomiser (Freitas et al. (2004), reproduced with

14 permission from Elsevier).

15 Fig. 7. (a) Photograph of a liquid cone and (b) depiction of setup for electrospray atomisation.

16 Images taken from (Meesters et al., 1992) and (Gañán-Calvo et al., 1997), reproduced with

17 permission from Elsevier.

18 Fig. 8. Computational fluid dynamics (CFD) simulations of (a) trajectory of flow from the

19 atomiser, and (b) swirling flow within the spray dryer main chamber. Images taken from

20 Fletcher et al., (2006), reproduced with permission from Elsevier.

1
 ACCEPTED MANUSCRIPT

21 Figures

22 Fig. 1.

23

24

2
 ACCEPTED MANUSCRIPT

25 Fig. 2.

26

27

3
 ACCEPTED MANUSCRIPT

28 Fig. 3.

29

30

4
 ACCEPTED MANUSCRIPT

31 Fig. 4.

32

33

5
 ACCEPTED MANUSCRIPT

34 Fig. 5.

35

36

6
 ACCEPTED MANUSCRIPT

37 Fig. 6.

38

39

7
 ACCEPTED MANUSCRIPT

40 Fig. 7.

41

42

8
 ACCEPTED MANUSCRIPT

43 Fig. 8.

44

45

9
 ACCEPTED MANUSCRIPT

Highlights

 The principles of atomisation within spray-drying applications are outlined.

 Conventional and novel atomisers are critically discussed.

 Developments in approaches for monitoring atomisation are described.

 Effects of feed composition and process variables on atomisation are reviewed.

1 Tables

2 Table 1. Comparison of the equations used for the determination of Sauter mean diameter, and advantages, and disadvantages for use of rotary,

3 pressure and pneumatic nozzles in spray-drying systems.

Sauter mean diameter

Atomisers Advantages Disadvantages
(𝑫𝟑,𝟐 ; Maters 1979)

- variation of the liquid flow rate without - production of wall deposits arising
having consequences on the droplet size from broad spray (Huang et al., 2006),
𝑚1 0.6 𝑏.𝜇.𝑁 0.2 𝜎.𝜌.𝑏3.𝑁 3 0.1
𝑣 𝑣 thus requiring wider body drying
0.4.( ) .( ) .( ) .𝑟 - simple droplet size control (wheel
2 𝑚1 2 chambers
𝑏.𝜌.𝑁.𝑟 .𝑁𝑣 𝑚1 revolution speed)
- not recommended for the drying of high
- can operate at high liquid feed viscosity
added value products such as protein
where 𝑚 (kg/s) is mass feed rate; 𝑁 (rps) is - can handle a high capacity isolate powders (too much waste)
rotational speed; 𝑟 (m) is wheel diameter; 𝑏 (m) is
- suitable for liquids that contain - the diameter of the chamber is a
Rotary vane height; 𝑁𝑣 is number of vanes; 𝜌 (kg/m3) is limiting factor when different powder
abrasive solids
atomisers fluid density; 𝜎 (N/m) is fluid surface tension and 𝜇 particle sizes are needed
(P.s) is fluid viscosity - the use of curved vanes can help to
reduce air incorporation (Hui, 2008) - incorporation of air into the droplets
due to air aspiration by the rotating
- reduction of wall deposit using wheel, resulting in high levels of
suitable system composed of a short occluded air in the final powder particles
chamber with a large diameter (i.e.,
newer generation wide-body spray- - air incorporation leads to modified
dryer configurations); this allows the properties of the final powder, such as a
particles to dry in a horizontal direction lower powder density and higher powder
before making contact with the walls volume
 - amount of occluded air depending on
the feed properties (high protein contents
will lead to a stronger effect due to the
well-established good surface activity of
dairy protein)

- simple design, compact, easy to clean - rapid wear and tear

39
and easy maintenance
[ ‒ (3.13·10 ‒ 3)·𝑣1] - occurrence of partial or full blockage
𝑣𝑎𝑥
286·[(2.54·10 ‒ 2)·𝑑𝑜𝑟 + 0.17]·𝑒 - low costs, and low energy consumption and/or clogging or excess build-up of
compared to rotary atomisers heat on the nozzle head due to the small
size of the nozzle orifice (Baker, 1997;
- production of particles in a narrow
Filkova et al., 2014) leading to :
where 𝑑𝑜𝑟 (m) is orifice diameter; 𝑣𝑎𝑥 (m/s) is axial range of sizes dependent on the pressure
velocity; 𝑣1 (m/s) is inlet air velocity nozzle type and the producer’s  viscosity limits (blockage)
expectations making rotary atomizers a better
choice for higher feed rate (Baker,
The range of droplet diameters produced by nozzles - air free droplets 1997)
Pressure is narrower than those produced by rotary  or non-uniform spray, which
nozzle atomizers. Moreover, the sprays resulting from - Possible orientation of the spray cloud
comes with consequences for the
atomisers pressure nozzles are composed of a more (useful in a system that contains several
final powder properties (e.g.
homogeneous drop size distribution than from pressure nozzles); the nozzles placed
particle size distribution)
rotary atomizers. close to each other will favour - in the case of protein-rich powders,
agglomeration or, if placed apart from excess heating could induce denaturation
each other, will avoid contact between and/or acceleration of Maillard
the different sprays and increase the reactions, which may already be
capacity observed during drying (Haque, 2015;
Palzer et al., 2012)
- limited use of pressure nozzles with
liquid feeds containing hard particles
(e.g., lactose crystals)
 - occurrence of corrosion and erosion
causing enlargement of orifices and
alterations to spray characteristics
- the feed rate is the only parameter that
can be modulated
- limited maximum capacity per nozzle

- suitable for the drying of viscous, thick - highest energy consumption of all three
1.5
and/or abrasive liquid feed products devices due to the cost of compressed
𝜎 𝜇 0.45 𝑉𝑓
air/steam
585000· + 597[ ] ·[1000· ] - two-fluid nozzles are mainly used in
𝑣𝑟𝑒𝑙· 𝜌 𝜎·𝜌 𝑉𝑎𝑖𝑟
pilot-scale spray-dryers due to - production of fine particles
Pneumatic limitations of pressure nozzle or wheel
- may need extra air supply
atomisers atomizers at this scale
where 𝜎 (N/m) is fluid surface tension; 𝜌 (kg/m3) is
density; 𝜇 (Pa.s) is viscosity; 𝑣𝑟𝑒𝑙 (m/s) is relative
velocity; and 𝑉𝑓 and 𝑉𝑎𝑖𝑟 (m3/s) are volumetric flow
rate of the fluid and air, respectively