SINGLE DROPLET EXPERIMENTATION ON SPRAY DRYING: EVAPORATION OF SESSILE DROPLETS DEPOSITED ON A FLAT SURFACE

Transcription

1 17th International Drying Symposium (IDS 1) Mageburg, Germany, 3-6 October 1 SINGLE DROPLET EXPERIMENTATION ON SPRAY DRYING: EVAPORATION OF SESSILE DROPLETS DEPOSITED ON A FLAT SURFACE J. Perana 1, M. Fox, M.A.I Schutyser 1, R.M. Boom 1 1 Foo Process Engineering Group, Wageningen UR P.O. Box 819, 67 EV Wageningen, The Netherlans Tel.: , jimmy.perana@wur.nl NIZO Foo Research P.O. Box, 671 BA Ee, The Netherlans Tel.: , martijn.fox@nizo.nl Abstract: Iniviually ispense roplets were rie on a flat surface to mimic the rying of single roplets uring spray rying. A robust ispensing process is presente that generates small roplets ( p >15 μm). A preictive moel base on Bernoulli s law accurately escribes roplet size with varying liquis an ispensing parameters. Shrinkage of the roplets, monitore with a camera, was escribe using mass balance equations. Finally, a Sherwoo correlation was erive to escribe the mass transfer coefficient for sessile roplets. This work forms the basis for the evelopment of a platform for high throughput experimentation on spray rying. Keywors: spray rying, sessile roplet, flui ispensing, Sherwoo correlation INTRODUCTION Spray rying has a long history in foo processing an other areas of inustrial prouction. For more than 3 years, spray rying is the most common metho to ehyrate for example milk proucts (Pierre ). Nowaays, spray rying is wiely use for manufacturing e.g. milk power, whey power, baby foo, lactose power, an maltoextrin ue to its avantages as state by Filkova (1995): 1. Properties of the power can be effectively controlle.. Spray rying allows high prouction capacity in continuous operation. 3. Heat-sensitive foos, biological proucts an pharmaceuticals can be ehyrate uner relatively mil rying conitions. Although spray rying is well establishe, it is ifficult to fin the optimal rying conitions for a given prouct. Numerous process parameters an fee properties uring spray rying have an influence on final prouct quality an process efficiency. To avoi upscaling issues, it is necessary to carry out trials in representative pilot-scale spray ryers. These trials are costly an increase the time to market. Other isavantages of existing spray ryers are the given esign an operational parameters, e.g. operating temperatures, resience time, an liqui fee viscosity (have to be within certain limits), which makes it ifficult to iscover new winows of operation in combination with the evelopment of new proucts. Recently, researchers have attempte to scale own the experimental conitions from pilot-scale to as small as a single roplet (Brask et al. 7; Schiffter an Lee 7). Major avantage of this approach is the reuction of the number of pilot-scale tests. Drying of single roplets was for example assesse by using acoustic levitation to suspen a free roplet in the air. From imaging the shrinking roplets, it was possible to etermine the rying kinetics at ifferent rying conitions (Brask et al. 7). In the current stuy we aim at the rying of single roplet ispense on a flat surface. Avantages of the propose approach is that multiple roplets can be rie simultaneous, resience time can be varie, an particles can be collecte for further analysis (e.g. to monitor survival of bacteria after rying). In this stuy a robust ispensing metho is investigate to prouce a small initial roplet size of various liquis. The properties of the surface (on which the roplets are eposite) shoul be selecte such that the roplets will retain their spherical shape, but o not move by the applie air flow. In this way the ifference in rying behavior between a sessile an a free falling roplet is minimize. Finally, the rying behavior of the sessile roplets shoul be mappe to compare their rying behavior to that of roplets uring spray rying. 1449

2 AIM The aim of this project is the evelopment of a single roplet rying approach (base on the rying of a roplet eposite on a flat surface) that mimics rying of a roplet uring spray rying. MATERIALS AND METHODS The experimental set-up is schematically shown in Fig. an consists of the following units: 1. Micro-ispenser. Drying chamber 3. Camera system The micro-ispenser was a Microot 741MD-SS Dispense Valve [EFD]. Nitrogen (6 barg) was use to operate the micro-ispenser. As shown in Figure, a static pressure is applie to the flui in the cylinrical flui reservoir. The neele valve or piston opening time is controlle by the actuating pressure. When the valve is opene, liqui is ispense through the neele tip. The stanar neele iameter use in this stuy was.1 mm. An Ultra TT positioning system (EFD) was use to automate the ispensing process an to position the ispense roplets. air was carrie out with two heat exchangers an a packe bubble column. The ry air stream was then split into two streams of which the flow rates were measure an controlle inepenently. The first stream was saturate with water in a packe bubble column (A). The secon stream was passe through a heat exchanger to increase the air temperature to a set value. Both streams were then mixe an le through the secon heat exchanger (B). Thus, the first heat exchanger was use to control the humiity of the air an the secon heat exchanger was use to control the outlet air temperature. Drying air was fe to the rying chamber via an insulate channel. In this channel, the air flow was evelope while the temperature was maintaine constant. The roplet was ispense on the soli surface with the ai of the Ultra TT system. The position of the roplet sample was exactly in the center of the outlet of the channel. The surface on which the roplets were eposite was a hyrophic membrane, i.e. Accurel type PP E HF (Akzo Nobel Faser Ag.). For this membrane the water contact angle was etermine to be 13 O. Fig. 1. Pneumatically riven micro-ispenser To evaluate the micro-ispensing process, three ifferent fluis were ispense; monoethylene glycol (MEG), iethylene glycol (DEG), an polyethylene glycol 4 (PEG). Analytical grae solutions manufacture by Merck were use. Distille water was ispense for the rying experiments. As shown in Fig. 3, the preconitioning of the rying Fig. 3. Set-up for preconitioning of the rying air; consisting of a packe bubble column an a heat exchanger The camera system consiste of a CCD camera (μeye 148ME) with a magnification lens. The roplet was illuminate from the back using a Droplet sample Micro-ispenser Droplet sample Camera Insulate air-fee tunnel Sample holer Insulate air-fee tunnel Air flow irection Fig.. Spray rying screening tool setup, sie view (S) an top view (T) Sample holer Back light 145

3 iffuse light (Lumimax SQ5-W). Micro-ispensing THEORY The micro-ispensing process can be consiere as a flui flowing through a series of channels. To escribe the micro-ispensing process, a moel base on Bernoulli s law is propose. Several assumptions an consierations were checke an foun to hol: Newtonian flow behaviour of the ispense fluis. The ispensing resistance is (>99%) ominate by the neele tip, which is the narrowest channel in the ispensing system (Darby 1). The small increase in temperature (< 3 o C) has no significant effect on flui viscosity. The Laplace pressure is less than 5% of the applie pressure for ispensing the smallest roplets an thus assume negligible For non-compressible fluis, Bernoulli s equation (Steffe an Singh 1997; Deplanque an Rangel 1998) can be written as: ( v v ) + w + e = Po Pi 1 + g( z o zi ) + α (1) o i f ρ which gives the energy per unit mass ue to the applie pressure ifference (first term), the potential energy ifference (secon term), the kinetic energy ifference (thir term), the shaft work (fourth term), an the friction in pipes an appenages (fifth term). The value of the kinetic energy correction factor (α) is taken, which is vali for laminar flow (Darby 1). In the micro-ispenser, the neele piston movement to open an close the valve can be regare as shaft work an etermines the minimum ispense flui that can be achieve. The total friction loss of the system is the summation of all frictions ue to the length of the pipes, valves, fittings, contractions or expansions: f ivi K e f = i () 99% of the resistance to liqui flow is calculate to be present in the neele an the contraction between neele an neele chamber. The fanning friction factor (f) is etermine for laminar flow, which epens on the Reynols number, regarless the wall roughness. The friction coefficient (K f ) is escribe by: L K f = 4f ; h 16 f = (3) Re To preict the resistance to flow ue to contraction between the neele an the neele chamber the following friction coefficient was propose by Sylvester an Rosen (197): K' K f = K + (4) Re with K =.4 an K = 95. This correlation is vali for a contraction factor ( c,out / c,in ) of.16 an 6<Re<, which is vali to the ispensing conitions in this stuy. Equation 1 is numerically solve an yiels the outlet flui velocity (v o ) of the ispense flui. The roplet volume can then be calculate as follows: V = v t (5) Drying of a single sessile roplet The evaporation of a single sessile roplet cannot be consiere similar to that of an ieal spherical boy because of the evolution in geometry uring the rying. In Fig. 4 the evolution of a rying sessile roplet is sketche. It is foun that uring our rying experiments the wette roplet surface ( l ) remaine constant. Thermoynamically, one woul expect a constant contact angle. However, ue to surface roughness a constant base iameter is often observe as well. This phenomenon is also known as contact angle hysteresis (Picknett an Bexon 1976). It is note that uring the rying of actual foo suspensions (e.g. concentrate milk) the roplet height will only be reuce with approximately 3% an thus the final particle shape will not be completely flat x 1-4 Fig. 4. Droplet evolution ue to evaporation of a pure liqui; sessile roplet (left) an spherical boy (right) Drying of single roplets has been investigate in numerous stuies, e.g. the work of Ranz-Marshall (195). The mass transfer coefficient can be obtaine from Sherwoo correlations that escribe the mass transfer coefficient as a function of hyroynamic conitions an the geometry of the rying object. The general Sherwoo correlation is the following (Oliveira an Oliveira 3): k evlc P3 P4 Sh = = P1 + P Re Sc (6) D In which, the Reynols an Schmit numbers are efine as: o is 1451

4 ρa pva Re = (7) μ gas μa Sc = (8) ρ D In the fiel of spray rying the Sherwoo correlation for a free falling spherical boy is mostly applie. The Sherwoo an Nusselt relations for heat an mass transfer for this case are: a k 1 1 evl Sh = =. +.6 Re Sc 3 (9) D h c p 1 1 Nu = =. +.6 Re Pr 3 (1) k a in which the Prantl number is equal to: cpaμa Pr = (11) k In this stuy we focus on the rying of a roplet eposite on a flat surface. Baines an James (1994) inicate that the Sherwoo correlation for a eposite roplet is comparable to that for flow across a flat plate: 1/ Sh =. 664Re Sc ; Sc>.6 (1) Evaporation of water from a sessile roplet is a couple heat an mass transfer process (Sloth et al. 6; Mezhericher et al. 8). For a sessile roplet heat is partially transferre via the rying air an via the soli surface that is heate by the rying air. Therefore, the total heat balance is: m c pl a 1/ 3 T = UconvAlaΔTla + t U con m A lsδ Tls ΔH vap (13) t in which ΔT la is the temperature ifference between the roplet an the air; an ΔT ls is the temperature ifference between the roplet an the soli surface. The exact temperature of the roplet was not etermine uring this stuy. For the calculations on mass transfer it was assume that the rying occurre near the wet bulb temperature (Pisecký 1995). The change in roplet mass uring evaporation can be escribe as follows: m V M l P = ρ = s Pba l k eva la (14) t t R Ts Tba in which P s is the saturate vapour pressure at 6 o C (the correlate wet bulb temperature of the rying air with temperature 8 O C an relative humiity %) an P ba is the vapour pressure of the rying air, which is taken zero. The change in roplet volume can be relate to the change in height: h V = πrc h (15) in which r c is the raius curvature of the roplet. r c l h + = (16) h The contact area between the sessile roplet an the rying air (A la ) is given by: la ( h l ) A = π + (17) Finally, the change in roplet height can then be escribe: ρ l 1 π ( h + l ) t h = M l ( ) Ps Pba h + l R Ts Tba k evπ (18) The equation above can be simplifie to: h t k = ρ ev l M l P s R Ts P T ba ba (19) The mass transfer coefficient (k ev ) can be obtaine by fitting Equation 19 to the experimental ata. RESULTS AND DISCUSSION To systematically evaluate the micro-ispensing an the evaporation, the results are iscusse separately. Dispensing process A pneumatic micro-ispenser is use to prouce roplets of a size similar to the roplet size uring spray rying. A robust ispenser was selecte that can prouce small roplets an at the same time ispense viscous liquis, which is a prerequisite for the ispensing of concentrate liqui foos. The micro-ispensing process is influence by process parameters (valve opening time an applie pressure) an prouct properties (viscosity an rheological behavior). The top view of the eposite roplets was visualize with a microscope. From the top view an the contact angle the roplet volume was calculate. The roplet volume is shown as a function of applie ispensing pressure an valve opening time in Fig. 5: 145

5 3 valve opening time (ms): 4 5 neele tip size (mm) Droplet size (nl) Droplet size (nl) Pressure (mbar) Fig. 5. Effect of ispensing pressure an valve opening time on MEG roplet volume. The symbols represent the experimental values. The lines represent the values preicte by the moel base on Bernoulli s law Fig. 5 shows that roplet volume increases with pressure an valve opening time. In practice, the smallest roplet that can be generate has a iameter of 15 μm (~ nl). This minimum roplet size is influence by the moving neele valve (in the microispenser) uring opening an closing. The experimental results are compare to the preicte roplet volume by a moel base on Bernoulli s law (Equation 1). As shown in Fig. 5, the preicte roplet size is in close agreement with the experimental ata. Droplet size (nl) 3 1 MEG DEG PEG Pressure (mbar) Fig. 7. Effect of neele tip size on MEG ispense volume (valve opening time = 81.5 ms) Three ifferent sizes of neele tips were use; 7 gauge (. mm), 3 gauge (.16 mm), an 3 gauge (.1 mm). As shown in Fig. 7, the moel was in agreement with the experimental ata. From these results, it is conclue that Bernoulli s law can account for the effects of applie pressure, valve opening time, neele tip size, an flui viscosity on roplet size. Thus, if the liqui viscosity is known an an appropriate neele tip size is chosen, the pressure an valve opening time can be controlle to eposit the require roplet size. Drying process The evaporation of a single ispense roplet was investigate as a function of the rying air conitions. The effect of the slip velocity between the roplet an rying air on the rying of a sessile water roplet is investigate here. In Fig. 8, snapshots of a roplet are shown uring the rying process Pressure (mbar) Fig. 6. Effect of flui properties (viscosity) on roplet size (valve opening time = ms) Fig. 8 Snapshots of a shrinking roplet, rying at a constant rying air velocity of.3 m/s (T = 8 O C, RH ~ %) The viscosities of MEG, DEG, an PEG-4 at 3 O C are 13, 1, an 71 mpa.s, respectively. Rheology experiments show that the fluis exhibit Newtonian behavior in the range of shear rates applie. Fig. 6 shows that roplet size ecreases with increasing flui viscosity. Bernoulli s law also inclues the effect of viscosity on ispense roplet size. As shown in Fig. 6, the preictions are in goo agreement with the experimental ata. 1453

6 roplet height (μm) 6 4 slip air velocity (m/s): A πh l c = = () P θ The mass transfer coefficient (k ev ) for a free spherical roplet is calculate using the Sherwoo number from the Ranz-Marshall correlation (Equation 9). The mass transfer coefficient is then use to escribe the height as a function of the rying time using Equation 19. The results are shown in Fig rying time (s) Fig. 9. Droplet height as a function of rying time at ifferent slip air velocities roplet base iameter (μm) contact angle ( O ) rying time slip air velocity (m/s): Fig. 1. Droplet base iameter as function of the contact angle uring the rying process Fig. 8, Fig. 9, an Fig. 1 show that the roplet geometry evolves mainly in height, while the contact area between roplet an the flat surface (base iameter) remains constant. This fining is in agreement with the observations of Hu an Larson () who observe that the contact angle of water roplets eposite on a glass surface change from 4 O to a minimum contact angle of 4 O, while the base iameter remaine constant. In the current stuy similar observations were obtaine; the roplet base iameter tens to be constant uring rying (Fig. 1). It was observe that uring the rying experiment with the air slip velocity of.17 m/s, the roplet base iameter suenly change. Possibly, this is ue to an irregularity in the microscopic surface of the membrane. After the suen change in the roplet base iameter, it remains constant again with changing contact angle. The experimental ata on the change in height may be compare to Equation 19. A characteristic length is use for the calculations of the Sherwoo an Reynols number. The characteristic length (l c ) is efine as: l c = object surface area perpenicular object perimeter For sessile roplets, the characteristic length is: normalize roplet height ( ) slip air velocity (m/s): rying time (s) Fig. 11. Normalize roplet height as a function of the rying time at ifferent slip air velocities; the soli line represents the moel in which the particle is a free falling sphere (Ranz-Marshall correlation) Fig. 11 shows that the height of a free roplet ecreases faster than that of a sessile roplet. It is conclue that the rying of the sessile roplet, even when with a high contact angle (13 o ), cannot be approximate by irectly using the mass transfer coefficient for a free spherical roplet. Therefore, a specific Sherwoo correlation was etermine for the rying of a sessile roplet by fitting the parameters of the general Sherwoo correlation (Equation 6) to the experimental ata. Vapour-air iffusivity an air viscosity were taken at the wet bulb temperature of 6 o C an assume constant in the experiments. The Schmit number is thus constant in this stuy, i.e..6. Therefore, it was not possible to estimate the power of the Schmit number. The estimate Sherwoo correlation are shown in Table 1 an compare to the values from literature for several cases. The parameter values for the Sherwoo correlation (Equation 6) in this stuy were obtaine by fitting the preictive moel (Equation 19) to the experimental height ata as shown in Fig. 1. Table 1 Estimate parameter values for the general Sherwoo correlation for ifferent cases P 1 Sh = P + P4 P3 1 PSc Re P P4 Sc P 3 Case 1454

7 normalize roplet height ( ) Sessile roplet, initial contact angle 13 o This work Free sphere Ranz-Marshall (195) Laminar flow across flat plate (Oliveira an Oliveira 3); Sc>.6 slip air velocity (m/s): rying time (s) Fig. 1. Comparison between the estimate Sherwoo correlation (soli line) an the experimental ata (points) Table 1 shows that the estimate parameter values in this stuy are ifferent compare to those of the flat plate an the falling sphere. Specifically, the parameter value P 1 is foun to be in between the two cases. The value of P 1 reflects the limiting Sherwoo value where iffusional mass transfer occurs preominantly. The value of P 1 is. for a spherical object an for a flat object. The sessile roplet changes from a nearly spherical to a flat object, which is reflecte in a ecreasing contact angle. The value of P 1 (.4) in this stuy is only vali for a sessile roplet with an initial contact angle of 13 o that ecreases to nearly o. More information about the effect of contact angle on the Sherwoo number for the iffusional regime can be foun in literature (Baines an James, (1994). It is expecte that the P 1 value for rying of a roplet from a suspension will be higher if the roplet will only partially evaporate. Furthermore, to better mimic the rying behavior of a free falling roplet it is suggeste to for example micro-fabricate a hairy surface structure that is better able to retain the spherical shape of the rying roplet (i.e. making use of the Lotus effect). The estimate value of P for this work is.6, if it is assume that P 4 is 1/3 (a common value for the power of the Schmit number). For a free falling sphere P is foun equal to.6 an for a flat surface equal to.67. The value of the estimate parameter P 3 in this stuy (.51) is comparable to the P 3 values in the Sherwoo correlations for flow across a flat surface an a falling sphere. CONCLUSIONS The ispensing of liqui roplets was carrie out with a pneumatic ispenser an coul be controlle accurately by varying applie pressure an valve opening time. A preictive moel base on Bernoulli s law was evelope an compare to experimental ata. Goo agreement was foun between ispense an preicte roplet volume. Drying of single sessile water roplets was experimentally monitore with a camera set-up. Drying was assume to occur at the wet bulb temperature. Experimental ata of the shrinking roplets were compare with a preictive moel base on mass transfer for a single free falling sphere. It was foun that the Sherwoo correlation of a free falling sphere i not preict the appropriate mass transfer coefficient for the rying sessile roplet. Therefore, a Sherwoo correlation was erive for the rying of sessile water roplets:.51 1/3 Sh = Re Sc for Sc=.6. This correlation was foun almost similar to the Sherwoo correlation for flow across a flat surface an a free falling sphere except for the value of P 1. It is expecte that this value will increase when rying foo suspensions or when a micro-fabricate hairy surface structure is use. It is planne to further investigate the rying of complex solutions, i.e. foo proucts, which are highly viscous or contain specific heat sensitive ingreients. During rying of these proucts, the temperature of the prouct will start eviating from the wet bulb temperature. Furthermore, a challenge will be the subsequent analysis of very small particles. Therefore, it is intene to evelop the platform into a high throughput experimentation system to ry multiple roplets at the same time. NOMENCLATURE A surface area m c p specific heat capacity at J.kg -1.K -1 constant pressure iameter m D iffusion coefficient m s -1 e f friction loss per unit mass N.m.kg -1 f fanning friction factor g acceleration gravity constant m.s - h c convective heat transfer W.m - K -1 coefficient h Droplet height m ΔH vap enthalpy of vaporization J.kg -1 k thermal conuctivity W.m -1 K -1 k ev Convective mass transfer m.s -1 coefficient K Hagenbach-Couette correction factor for narrow gap contraction K Hagenbach-Couette 1455

8 correction factor for narrow gap contraction K f friction loss coefficient l c characteristic length in m Sherwoo number l Droplet base raius m L Channel length m m mass kg M Molecular weight kg.mol -1 Nu Nusselt number P pressure Pa Pr Pranlt number r c raius curvature m Re Reynols number Sc Schmit number Sh Sherwoo number t time s T temperature O C or K U Overall heat transfer W.m - K -1 coefficient v velocity m.s -1 V volume m 3 w shaft work in microispenser N.m.kg -1 per unit mass x length coorinate m z height coorinate m Greek letters α Kinetic energy correction factor Δ Difference μ ynamic viscosity kg.m -1.s -1 π pi value; 3.14 θ Contact angle ra ρ ensity kg.m -3 Subscripts a Air ba Bulk air conv Convection con Conuction Droplet is Dispensing i In l Liqui, i.e. water o Out p Particle s Soli ACKNOWLEDGEMENTS We thank Avance Chemical Technologies for Sustainability (ACTS) Netherlans Organization for Scientific Research (NWO) for financial ai through Process on a Chip (PoaC) program. REFERENCES Baines, W. D. an D. F. James (1994). "Evaporation of a roplet on a surface." Inustrial & Engineering Chemistry Research 33(): Brask, A., et al. (7). "High-Temperature Ultrasonic Levitator for Investigating Drying Kinetics of Single Droplets." Niro A/S. Darby, R. (1). Chemical Engineering Flui Mechanics. New York, Marcel Dekker, Inc. Deplanque, J. P. an R. H. Rangel (1998). "A comparison of moels, numerical simulation, an experimental results in roplet eposition processes." Acta Materialia 46(14): Filkova, I. an A. S. Mujumar (1995). Inustrial Spray Drying System. Hanbook of Inustrual Drying. A. S. Mujumar. New York Marcel Dekker, Inc. Hu, H. an R. G. Larson (). "Evaporation of a Sessile Droplet on a Substrate." The Journal of Physical Chemistry B 16(6): Mezhericher, M., et al. (8). "Heat an mass transfer of single roplet/wet particle rying." Chemical Engineering Science 63(1): 1-3. Oliveira, F. A. R. an J. C. Oliveira (3). Sherwoo Number, Taylor & Francis: Picknett, R. G. an R. Bexon (1976). "The Evaporation of Sessile or Penant Drops in Still Air." Journal of Colloi an Interface Science 61(): Pierre, S. (). "Spray rying of airy proucts: state of the art." Le Lait 8(4): Pisecký, J. (1995). Evaporation an Spray Drying in the Dairy Inustry Hanbook of Inustrial Drying: Volume 1. A. S. Mujumar. Boca Raton, CRC Press. Ranz, W. E. an W. R. Marshall (195). "Evaporation from Drops." Chemical Engineering Progress 48: ; Schiffter, H. an G. Lee (7). "Single-roplet evaporation kinetics an particle formation in an acoustic levitator. Part 1: Evaporation of water microroplets assesse using bounary-layer an acoustic levitation theories." Journal of Pharmaceutical Sciences 96(9): Sloth, J., et al. (6). "Moel base analysis of the rying of a single solution roplet in an ultrasonic levitator." Chemical Engineering Science 61(8): Steffe, J. F. an R. P. Singh (1997). Pipeline Design Calculations for Newtonian an Non- 1456

9 Newtonian Fluis. Hanbook of Foo Engineering Practice, CRC Press. Sylvester, N. D. an S. L. Rosen (197). "Laminar flow in the entrance region of a cylinrical tube: Part I. Newtonian fluis." AIChE Journal 16(6):