Rotor-Stator Devices: The Role of Shear and the Stator

T. L. Rodgersa,, M. Cookea
a

School of Chemical Engineering & Analytical Science, The University of Manchester, Oxford Road, Manchester, M13 9PL

Abstract High shear rotor-stator mixers are widely used in process industries including the manufacture of many food, cosmetic, pharmaceutical, and health care products. Many of these products involve emulsication where the drop size distribution aects the processing and the product properties. Therefore, an understanding of the mechanisms that breaks the drops is key for any design process. In rotor-stator devices there are two main mechanisms that can break drops, one due to the rotor and one due to the stator. For the invisid systems studied, this article shows that when a rotor-stator device is used in a recycle loop the eective equilibrium drop size is largely unaected by the presence of the stator and is mainly dependant on the rotor. The article also goes on to show that the eective equilibrium drop size data can be correlated on the agitator shear rate. Keywords: Rotor-stator devices, Shear rate, Equilibrium drop size

1. Introduction High shear rotor-stator mixers are widely used in process industries including the manufacture of many food, cosmetic, pharmaceutical, and health care products. Rotor-stator devices provide a focused delivery of energy, power and shear to accelerate physical processes such as mixing, dissolution, emulsication, and de-agglomeration. To reliably scale-up these devices we need to understand the relationship between rotor speed, ow rate, shear rate, and the energy dissipated by these devices.

2 3 4 5 6 7 8

Corresponding Author Email address: T.Rodgers@postgrad.manchester.ac.uk (M. Cooke)

Preprint submitted to Chemical Engineering Research and Design

June 12, 2013

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

For emulsication, the drop size distribution aects the processing and the product properties. In a two phase process, the mass transfer rate between the phases is proportional to the interfacial area. This interfacial area changes with the drop size distribution which varies with the conditions inside the vessel and time. Hence successful process design depends on developing a mechanistic understanding of drop break-up in these reactors. There are two competing theories on drop break-up mechanisms. These are break-up due to turbulent eddies, i.e. energy dissipation rate, and break-up due to the agitator shear rate. Break-up due to turbulent eddies is generally based on the work of [7] and [6] which utilises the concept of eddy turbulence to dene a limiting drop size. It is usually assumed that drop break-up occurs due to the interactions of drops with the turbulent eddies of sucient energy to break the drop [9]. Therefore, for a given uid system the eective equilibrium drop size (this is the drop size after a sensible processing time, when the drop size reduction with time is very small and almost unmeasurable) is dependent on the energy per unit mass and thus should scale-up with this value when using geometrically similar vessels. For low viscosity dispersed phase dilute liquid-liquid systems, the drops are inviscid since the internal viscous stresses are negligible and only the interfacial tension surface force contributes to stability. The maximum stable equilibrium drop size, dmax can be related to the maximum local energy dissipation rate, max , by equation 1 for isotropic turbulence [8, 3]. dmax = C1
3/5 2/5 max

(1)

30 31 32 33 34 35 36 37 38 39 40 41 42

Figure 1 presents drop size data from previously published literature for a silicone oil and water system as a function of the energy dissipation rate, ; the gradient of the line has been set to 2/5 in agreement with equation 1. The lack of correlation between the eective equilibrium drop size and energy dissipation rate seems to point to the fact that this may not be the correct mechanism. This is not surprising since this theory applies to isotropic turbulence in the universal equilibrium regime, whereas it is known that breakup occurs close to the agitator where the turbulence is both non-isotropic and intermittent. Break-up due to the agitator shear rate is based on a balance between the external viscous stresses and the surface tension forces [9]. If the break-up is due to the agitator shear rate then the eective equilibrium drop size is related to the maximum shear rate. This would mean that lower power number agitators can produce smaller drops than higher power number agitators, as low power number 2

43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72

agitators may have a higher shear rate. This has been seen experimentally by [14]. [1] states when scale-up is performed on a constant energy dissipation rate, smaller drops are observed at larger scales. This is likely due to the shear rate increasing at larger scales when the energy dissipation rate is kept constant. The maximum shear rate is proportional to the agitator tip speed [11]; however, this maximum shear rate constant is dicult to measure for all systems. It makes physical sense that this maximum shear rate constant is proportional to an average shear rate constant. Although, it is strictly only applicable in the laminar regime [4], the Metzner-Otto constant, KS , is a good measure of the average shear rate near the impeller. It has been claimed that for power law uids Ks varies with the power law index although for practical considerations this aect has found to be small. Tanguy et al. (1996) [12] concluded that for practical considerations a constant value of KS can be considered for shear thinning and shear thickening uids. KS has been found to vary linearly with the agitator ow number which is a function of Reynolds number [13]. However, in the turbulent regime the ow number is constant, so again we have a constant (though higher) value of KS . This means the correlative shear rate used will be KS ND, i.e. the proportional constant multiplied by the tip speed. It should be noted here that this shear rate technically has units of m s1 instead of s1 , but as previously mentioned, this is just a representative value as the true value is proportional to KS ND, which means that this proportionality constant must have units of m1 . Figure 2 presents drop size data for a silicone oil and water system as a function of the agitator blade shear rate, KS ND, the best t line gradient is equal to 1.2. The fact that all the values lie on the same line points towards shear rate being the dominant break-up mechanism. Neither of these two mechanisms will predict the correct values for the drop size if the system undergoes coalescence. If there is coalescence within the system the bulk ow from the agitator is important as well, as this eects the circulation time, thus the time away from breakage. The systems studied within this paper are non-coalescing systems, which was checked over a period of several days. 2. Methodology The experimental rig (Figure 3) consists of an agitated mixing tank with an in-line Silverson 150/250 MS high shear rotor-stator mixer (Silverson Machines Ltd., Chesham, UK). The mixing tank has a 60 litre capacity with a diameter of 0.420 m. To allow both analysis of equlibrium drop sizes and single pass drop size data the mixing tank was connected to the rotor-stator mixer via a 38.1 mm 3

73

74 75 76 77 78

79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95

pipeline, with a second line used to feed the rotor-stator outlet to the drain for the single pass experiments. The inlet and outlet lines include temperature probes so the temperature could be controlled to 25 C for all the experiments. The Silverson rotor-stator mixer has a double concentric rotor with a diameter of 0.0635 m which sits within close tting screens. The tted screens are standard double emulsier screens [2]. Two experimental systems were used; a 1 wt% 350 cSt silicone oil and 0.5 wt% sodium laureth sulfate in water solution, which was passed through the Silverson both a single pass and equlibrium conguration, both with and without the screens as a range of rotor speeds and ow rates; and a 0.13 wt% 5 cSt silicon oil in water system as used by [10] which was passed through the Silverson in the equlibrium conguration both with and without the screens at a range of rotor speeds at a ow rate of 0.167 kg s1 .. Samples of the solutions were collected just after the Silverson in-line mixer and analysed using a Mastersizer X long bed laser diraction particle analyser (Malvern Instruments, Malvern, UK) to determine the drop size distribution, full details are given by [5]. 3. Eect of the Screens on Drop Size A 1 wt% 350 cSt silicone oil and 0.5 wt% sodium laureth sulfate in water solution has been passed through a Silverson 150/250 in-line rotor stator mixer. This is carried out under a range of agitation rates and ow rates. First as a single pass and then as a recycle arrangement for both the standard double screen arrangement and with no screens. Figure 4 shows the results for these experiments. When the system is used for a single pass the screens reduce the drop size compared to a single pass with no screens. Whereas when the system is used in a recycle arrangement there is very little dierence in drop size when there are screens or no screens. With no screens there is a large variation of the single pass drop size with ow rate, especially at low agitation rates. This is due to by-pass of some of the drops around the rotor. This is shown by the reduction in drop size for a single pass with the screens, and the more consistent drop size produced with variation in ow rate. The screens also cause some re-circulation to the agitator which is why the drop size is slightly reduced. There is evidence that the stator has some eect on the break-up, especially at the higher ow rates. This is because the high ow rate d3,2 s are the smallest drop sizes for the single pass data with the screens, see Figure 4, and the largest 4

96

97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114

Setup Po(turbulent) PoRe(laminar) Standard dual screens 0.24 573.6 No screens 0.21 276.2

KS k1 46.2 8.02 6.6 7.05

Table 1: Summary of power and shear constants for the Silverson 150/250 [2].

115 116 117 118 119 120 121 122 123 124 125 126 127 128

size without the screens. This is due to the fact that as the ow rate increases the pressure drop through the screens is increased which increases the shear through the holes. As the system approaches the eective equilibrium this bypass and recirculation does not matter as material that my have bypassed on a single pass will eventually get broken by the agitator on another pass. The drop size for recycle loop are smaller than for the single pass, this is because for the recycle loop the drops are approaching their eective equilibrium drop size for that shear rate whereas those after a single pass are not. This is because under the recycle arrangement the drops pass through the agitator more than once increasing the probability that the small drops produced will become stable. The power number of an in-line rotor-stator device can be given by equation 2 [2]. Table 1 provides the power and shear constants for the Silverson mixer from [2]. Po = Poz + k1 NQ (2) This means that the power number for the above experiments vary from about 0.25 for the lowest ow rate to 0.66 for the highest ow rate, over a 2.5 times increase in power. However, the equlibrium drop sizes are mostly the same for dierent ow rates. This provides evidence that the energy dissipation rate may not be a suitable correlator for drop sizes, especially for rotor-stator devices. 4. Rotor-Stator Shear For comparison to the data in Figures 1 and 2, the 0.13 wt% 5 cSt silicon oil in water system used by [10] was used in the Silverson 150/250 inline mixer to an eective equilibrium drop size for both the standard screens and no screens. No real dierence between the values for the two setups was discovered, again showing the small impact of the screens on the equilibrium drop size. Figure 5 presents the same data as Figure 1 but includes the authors data for the Silverson system both with and without the screens. The best t line is given a slope of 0.4; however, it can be seen that this does not t the data well. It is clear 5

129 130 131 132 133 134

135

136 137 138 139 140 141 142 143

144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177

that although the exponents for all the agitators are around 0.4, the multiplier for each is dierent. In considering equilibrium drop size, it is typical to use the maximum local value of the rate of energy dissipation rather than the average value (as used in Figures 1 and 5). The best order of magnitude estimate of the maximum dissipation rate is based on using the swept volume of the impeller instead of tank volume [11]. Therefore, Figure 5 has been modied to produce Figure 6. Figure 6 has the same best t line slope as Figure 5. It can be seen that Figure 6 does better at correlating agitators of the same type together, e.g. all the Rushton turbine data is now almost collapsed onto one trend; however, it is still not able to correlate the drop size data from dierent agitator types. As the screens seem have little eect on the eective equilibrium drop size it can be surmised that the drops are broken mainly by the agitator. This means that the shear rate of importance is that from the agitator only, not the full rotorstator shear rate. Therefore, for the data to line up with the data from Figure 2 the Metzner-Otto constant that should be used is the one just for the agitator, i.e. 6.6. As previously mentioned, the agitation rates can be converted to a representative shear rate by multiplying by the agitator Metzner-Otto constant and its diameter. This means that the data for the Silverson can be plotted on Figure 2 producing Figure 7. It can be seen that using the agitator shear rate the data ts much better than with the energy dissipation rate. If the total Metzner-Otto constant was used for the standard screens system the drop size would be under predicted. The fact that the screens provide very little advantage for eective equilibrium drop size and that the agitator is the limiting factor, then a system like this can be attributed an apparent drop size Metzner-Otto constant, which is equal to the that of just the agitator. It should be noted that although the screens provide very little help for equilibrium drop break up they play a part in other processes, such as deagglomeration. It is also noted that since the Metzner-Otto constant is much higher when the screens are present, the apparent viscosity will be lower for shear thinning uids. This will change the viscosity ratio which may impact on the equilibrium drop size. There is also some evidence that at very high ow rates the increased velocity through the holes help to break up material that bypasses the agitation, resulting in a shorter time to the eective equilibrium.

178

5. Discussion and Conclusions The agitator shear rate correlates the drop size for a range of agitator types and scales for a particular uid system, For rotor-stator devices it is the rotor shear rate that is important for drop break-up not that from the stator, especially at low ow rates, The stator helps single pass design as it reduces bypass around the agitator, Design for drop break-up should focus on improving the rotor as this is dominant in drop break-up. 6. Acknowledgments Tom Rodgers would like to thank The University of Manchesters EPSRC CTA (Collaborative Training Account) and Unilever for nancial support during his PhD. The authors would also like to thank the SCEAS workshop sta who helped with equipment modications and construction. Nomenclature D Agitator diameter d3,2 Volume to surface average diameter dmax Maximum stable diameter k1 Power ow constant KS Metzner-Otto constant N Agitation rate P Power Q Flow rate energy dissipation rate Density NQ Flow number Po Power number Poz Power number at zero ow rate Re Reynolds number m m m s1 W kg s1 W kg1 kg m3 =Q/ND3 =P/(N 3 D5 ) =P/(N 3 D5 ) =ND2 /

179 180

181 182

183

184 185

186

187 188 189 190

191

192 193 194 195 196 197 198 199 200 201 202 203 204 205

206

References [1] J. Badyga, J.R. Bourne, A.W. Pacek, A. Amanullah, A.W. Nienow, Chemical Engineering Science 56 (2001) 33773385. [2] M. Cooke, T.L. Rodgers, A.J. Kowalski, Journal of Chemical Engineering in press (2010). [3] J.T. Davies, Chemical Engineering Science 42 (1987) 16711676. [4] D. Doraiswamy, R.K. Grenville, A.W. Etchells III, Industrial and Engineering Chemisrty Research 33 (1994) 22532258. [5] S. Hall, M. Cooke, A. El-Hamouz, A.J. Kowalski, Chemical Engineering Science in press (2011). [6] J.O. Hinze, Journal of the American Institute of Chemical Engineers 1 (1955) 289295. [7] A.M. Kolmogorov, Doklady Akademii Nauk 66 (1949) 825828. [8] D.E. Leng, R.V. Calabrese, Handbook of Industrial Mixing: Science and Practice, Wiley-Interscience, 2004, pp. 639753. [9] Y. Liao, D. Lucas, Chemical Engineering Science 64 (2009) 33893406. [10] M. Musgrove, S. Ruszkowski, Proceedings of the 10th European Conference on Mixing, Delft, Netherlands July 2-5 (2000) 165172. [11] E.L. Paul, V.A. Atiemo-Obeng, S.M. Kresta (Eds.), Handbook of Industrial Mixing: Science and Practice, John Wiley & Sons Inc., New Jersey, USA, 2004. [12] D.E. Tanguy, E. Thibault, E. Brito De la Fuente, Canadian Journal of Chemical Engineering 74 (1996) 222228. [13] J. Wu, L.J. Graham, N.N. Mehidi, AIChE Journal 52 (2006) 23232332. [14] G. Zhou, S.M. Kresta, Chemical Engineering Science 53 (1998) 20632079.

207 208

209 210

211

212 213

214 215

216 217

218

219 220

221

222 223

224 225 226

227 228

229

230

231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260

Fig. 1. Variation of mean eective equilibrium drop size for a 5 cSt silicon oil and water system against the power per unit mass. Data taken from [14] and [10]. Dotted lines are 20% from the best t. Fig. 2. Variation of mean eective equilibrium drop size for a 5 cSt silicon oil and water system against the agitator blade shear rate. Data taken from [14] and [10]. Dotted lines are 20% from the best t. Fig. 3. Schematic of the equipment used for the experiments. The Silverson is used in either a single or multiple pass arrangement. Fig. 4. Variation of the single pass and eective equilibrium drop size for a 1 wt% 350 cSt silicone oil and 0.5 wt% sodium laureth sulfate in water. OP and EQ refer to a one pass and eective equilibrium run respectively. WS and NS refer to with screens and without screens respectively. Fig. 5. Variation of mean eective equilibrium drop size for a 5 cSt silicon oil and water system against the power per unit mass. Data taken from [14] and [10]. *Data from current work in Silverson both with and without screens. Dotted lines are 20% from the best t. Fig. 6. Variation of mean eective equilibrium drop size for a 5 cSt silicon oil and water system against the maximum energy dissipation rate (power per unit agitator swept mass). Data taken from [14] and [10]. *Data from current work in Silverson both with and without screens. Dotted lines are 20% from the best t. Fig. 7. Variation of mean eective equilibrium drop size for a 5 cSt silicon oil and water system against the agitator blade shear rate. Data taken from [14] and [10]. *Data from current work in Silverson both with and without screens. Dotted lines are 20% from the best t.

d3,2 / m

261

10

-3
A310 T=0.24 m, D/T=0.35 A310 T=0.24 m, D/T=0.55 HE3 T=0.24 m, D/T = 0.25 PBT T=0.24 m, D/T=0.25 RT T=0.24 m, D/T=0.25 RT T=0.29 m, D/T=0.5 RT T=0.26 m, D/T=0.33 RT T=0.17 m, D/T=0.5 RT T=0.17 m, D/T=0.33

10

-4

10

10

-5

10

-2

10

-1

10 / W kg
-1

10

d3,2 / m

262

10

-3
A310 T=0.24 m, D/T=0.35 A310 T=0.24 m, D/T=0.55 HE3 T=0.24 m, D/T=0.25 PBT T=0.24 m, D/T=0.24 RT T=0.24 m, D/T=0.25 RT T=0.29 m, D/T=0.5 RT T=0.26 m, D/T=0.33 RT T=0.17 m, D/T=0.5 RT T=0.17 m, D/T=0.33

10

-4

11

10 1

-5

10 -1 KSND / m s

100

To Drain (for single pass) (for multiple pass)

Sample point

263

Silverson

12

d3,2 / m

264

40 35 30 25 20 15 10
13

OP NS 0.083 kg s OP NS 0.278 kg s OP NS 0.472 kg s OP NS 0.667 kg s

-1 -1 -1 -1 -1 -1 -1 -1

OP WS 0.083 kg s OP WS 0.278 kg s OP WS 0.472 kg s OP WS 0.667 kg s EQ NS 0.083 kg s EQ NS 0.278 kg s EQ NS 0.472 kg s EQ NS 0.667 kg s

-1 -1 -1 -1 -1 -1 -1 -1

EQ WS 0.083 kg s EQ WS 0.278 kg s EQ WS 0.472 kg s

EQ WS 0.667 kg s

50

100

150

N/ s

-1

d3,2 / m

265

10

-3
A310 T=0.24 m, D/T=0.35 A310 T=0.24 m, D/T=0.55 HE3 T=0.24 m, D/T = 0.25 PBT T=0.24 m, D/T=0.25 RT T=0.24 m, D/T=0.25 RT T=0.29 m, D/T=0.5 RT T=0.26 m, D/T=0.33 RT T=0.17 m, D/T=0.5 RT T=0.17 m, D/T=0.33 Silverson 150/250 inline mixer*

10
14

-4

10

-5

10

-6

10

-2

10 -1 / W kg

10

d3,2 / m

266

10

-3

10
15

-4

10

-5

10

-6

A310 T=0.24 m, D/T=0.35 A310 T=0.24 m, D/T=0.55 HE3 T=0.24 m, D/T = 0.25 PBT T=0.24 m, D/T=0.25 RT T=0.24 m, D/T=0.25 RT T=0.29 m, D/T=0.5 RT T=0.26 m, D/T=0.33 RT T=0.17 m, D/T=0.5 RT T=0.17 m, D/T=0.33 Silverson 150/250 inline mixer*

10

-2

10

-1

10 -1 max / W kg

10

10

d3,2 / m

267

10

-3
A310 T=0.24 m, D/T=0.35 A310 T=0.24 m, D/T=0.55 HE3 T=0.24 m, D/T=0.25 PBT T=0.24 m, D/T=0.24 RT T=0.24 m, D/T=0.25 RT T=0.29 m, D/T=0.5 RT T=0.26 m, D/T=0.33 RT T=0.17 m, D/T=0.5 RT T=0.17 m, D/T=0.33 Silverson 150/250 inline mixer*

10
16

-4

10

-5

10 1

-6

10 -1 KSND / m s

100