EXPERIMENTAL AND CFD STUDIES OF FAT FOULING IN A NOVEL SPINNING DISC SYSTEM

Proceedings o International Conerence on Heat Exchanger Fouling and Cleaning VIII - 29 (Peer-reviewed) June 14-19, 29, Schladming, Austria Editors: H. Müller-Steinhagen, M.R. Malayeri and A.P. Watkinson EXPERIMENTAL AND CFD STUDIES OF FAT FOULING IN A NOVEL SPINNING DISC SYSTEM R.Y. Nigo 1, Y.M.J. Chew 1 *, N.E. Houghton 2, W.R. Paterson 1 and D.I. Wilson 1 1 Department o Chemical Engineering and Biotechnology, University o Camridge, New Museums Site, Pemroke Street, Camridge CB2 3RA, UK 2 Department o Engineering, University o Camridge, Trumpington Street, Camridge, CB2 1PZ, UK *corresponding author: ymjc2@cam.ac.uk ABSTRACT Fats, like waxes, can cause reezing ouling when sujected to temperatures elow their cloud point, oth in heat exchangers and during transport o mixtures along pipelines in actories where it is termed coring. This paper reports the use o a novel spinning disc apparatus (SDA) to study reezing ouling rom at mixtures, here a model solution o tripalmitin in a non-crystallising parain oil. The SDA employs smaller volumes o solution than conventional low cell loops, is simple to operate, allows the ouled surace to e recovered, and eatures well-deined low conditions. For this application the device operates in the laminar regime, allowing computational luid dynamics (CFD) simulations to elucidate the heat transer and low ehaviour in the system, with particular ocus on the heat lux and the shear stresses imposed on the surace. The CFD results showed good agreement with experimental heat transer measurements. The scope o the device is demonstrated with a short experimental study o PPP deposition rom 1 wt% solutions on smooth stainless steel suraces. INTRODUCTION Fouling o heat transer and other process equipment suraces is a prolem in many industries, and can e particularly severe in the ood sector where the materials eing processed contain components such as proteins, ats and mineral salts that are precursors or the uild-up o ouling layers. Such deposits reduce the eiciency o process units and incur costs via extra cleaning to avoid cross-contamination among products, or to maintain hygiene and microial security (Fryer et al., 1997). Epstein (1983) classiied ouling according to the mechanisms o deposit ormation, and identiied two variants o crystallisation ouling, determined y the soluility ehaviour: scaling associated with inverse soluility salts such as calcium caronate and phosphate in heating aqueous systems, and reezing ouling where cooling the luid induces solidiication. Most o the work on reezing ouling has concentrated on petroleum lends where cooling induces solidiication o waxes and is indeed exploited in the manuacture o luricants. Examples o recent work in wax ouling include those y Akarzadeh and Zougari (28), Parthasarathi and Mehrotra (25) and Singh et al. (21). Signiicant advances in the understanding o kinetics o wax ormation and ageing have een achieved and models developed or scaling up experimental results and predicting operating scenarios. Fouling phenomena analogous to wax deposition are experienced in the ood sector, where liquid and semicrystallised mixtures o ats are used in large quantities in aking and iscuit manuacture. Large quantities o at mixtures are prepared in a central acility and transported to the point o use, e.g. mixers. Food ats are mixtures o triglycerides and smaller quantities o diglycerides and, like waxes, can cause reezing ouling when sujected to temperatures elow their cloud point, T c, so that deposits can uild up on pipe walls. This coring occurs via crystallisation, and yields a viscous gel which can harden to give a solid deposit over time. The impact o coring includes impairing the thermal and hydraulic eiciencies o the equipment. Relatively little work has een reported on ood at ouling: Fernandez-Torres et al. (21) reported a modelling approach including a ouling regime map using concepts taken rom wax deposition in crude oil pipelines. Fitzgerald et al. (24) studied ouling utilising a model at solution prepared out o a single crystallising component, tripalmitin (PPP), in a non-crystallising parain solvent using a lat plate heat exchanger. PPP is oten used as a model at ecause it arises in many vegetale and ood at lends, and the melting point o pure PPP, at approximately 63 o C, means that deposition can e studied with coolants operating near amient temperature. This paper extends the experimental investigations o Fitzgerald et al. using similar model solutions ut using a novel test coniguration, the spinning disc apparatus (SDA). The two most common techniques reported in the literature or studying the ouling ehaviour o waxes in crude oil are the low cell loop (e.g. Ghedamu et al., 1997) and the cold inger (e.g. Jennings and Weispennig, 25). In the ormer, warm oil (aove its cloud point) lows along a long duct oten a pipe with cooled walls so that the wax solidiies on the wall. The volumes o luid and size o apparatus are 263

Nigo et al. / Experimental and CFD Studies o Fat Fouling usually sizeale and therey limit its use as a routine assessment method. The cold inger test employs smaller volumes and is simple in operation. Its asic principle is that warm oil (with ulk temperature aove its T c ) lows over an upright cylinder whose surace is held elow T c. Deposit orms on the surace o the inger and is collected and analyzed. Their principal shortcoming lies in the complexity o the low ield (turulent low) and therey extrapolation o the results to operating systems, although CFD simulations o these devices have een reported (Jennings and Weispennig, 25). Spinning disc devices oer well-deined low conditions which have prompted their use in mass transer studies (see Rashaida et al., 26) and cleaning (Grant et al., 1996). Heated spinning discs have een used in ouling studies (e.g. Rosmaninho and Melo, 26), where heat is either supplied y circulating hot oil or y electrical heating via slip-ring connections. Chilled spinning discs are, to the authors knowledge, rarely used, principally owing to the challenges involved in supplying coolant to the rotating assemly. The advantages o spinning discs over conventional low cell loops and cold ingers are that they simultaneously (i) use smaller volumes o solution; (ii) are simple to operate, (iii) allow the ouling surace to e recovered or analysis; and (iv) eature well-deined laminar low conditions. The design and operation o an SDA eaturing cooled, removale heat transer suraces is reported here. Surace temperature and heat transer rates are key parameters in reezing ouling so these have een investigated experimentally and y CFD simulations. Simulation is easile here ecause the device is operated in the laminar low regime. The SDA device is employed in a study o reezing ouling or a model at solution similar to that employed y Fitzgerald et al. (24), augmenting the results otained therein with a larger low loop system. central tue and impinges on the ase o the can, and leaves via the annular gap o the 3 mm etween the inner and outer tues. The coaxial tues are stationary and constitute the shat aout which the can rotates. The inner tue extends to within a ew millimetres o the ase o the can. This arrangement aords the resh coolant rapid contact with the test plate and promotes good mixing. T cw2 T cw3 15 1 T 1 T 2 T cw1 Fig. 1: Schematic o SDA unit. Dimensions are in mm (not to scale). 4 4 5 EXPERIMENTAL Spinning disc apparatus The main eature o SDA device is a vertical cylinder whose lower ase rotates in a warm solution, as shown in Fig. 1. The apparatus consists o a jacketed vessel holding the warm ulk solution, the rotating can and a magnetic stirrer to aid mixing and maintain temperature uniormity in the ulk solution. Deposition occurs only on the cold, exposed surace at the ase o the rotating cylinder as the side wall o the cylinder is insulated y water- and greaseproo ruer oams. The ulk reservoir is an insulated, 3 L orosilicate glass vessel. The jacket is connected to a recirculating water ath and temperatures measured y T-type thermocouples. The ulk liquid is mixed y a 5 cm long PTFE coated magnetic ar stirrer, which rotated at 2 rad/s (in the opposite direction to the can) in all studies reported here. Rotation o the can is provided y a stepper motor. Coolant, here a water/glycol mixture, is supplied y a second recirculating water ath through a pair o coaxial tues. The incoming coolant is channelled through the Fig. 2: Construction o the rotating can ase. Dimensions in mm (not to scale). Fig. 2 shows a schematic o the can ase. The detachale 4 mm 316 stainless steel disc was separated rom the coolant y a rass lock, in which a micro-oil heat lux sensor was mounted. The sensor housing was lined with heat sink gel and the components screwed together tightly to exclude air and other contact resistances. Temperatures were measured using T-type thermocouples at the locations marked on Fig. 1, where T cw1 (inside the can, in contact with the surace o SS 316 disc), T cw2 (inlet coolant) and T cw3 (outlet coolant) are coolant temperatures while T 1 (~ 5 cm elow the ase o the disc) and T 2 (~ 5 cm aove the ase o the reservoir) are ulk temperatures. All except T cw1, were connected to a multi-channel temperature data logger: T cw1 was monitored using a T-type thermocouple connected to a attery-powered stand-alone data logger located on the can roo. A similar device was used to record the heat lux sensor signal and eliminated the need or slip rings. T cw1 was www.heatexchanger-ouling.com 264

Heat Exchanger Fouling and Cleaning VIII 29 ound to e similar to T cw2 and T cw3, i.e. ~ ±.5 K, so the coolant temperature, T cw, was taken to e T cw1. The values o T 1 and T 2 were similar, i.e. ~ ±.5 K, and their arithmetic mean was used as the average ulk temperature, T. The warm solution is held at a temperature, T, aove its cloud point, and is in contact with an initially clean cool outer wall surace at temperature T ss,out < T c. Solution at the wall will e locally saturated and orm crystals: deposition generates an insulating ouling layer, with solid-liquid interace temperature, T s, initially close to T ss,out ut gradually increasing and approaching T as deposit uilds up. I T s reaches T c the solution at the interace will e too warm or crystallisation. The solid-liquid interace temperature T s can e calculated rom measurements o the heat lux (explained later). The local heat lux through the rotating disc, q, is given y Newton s law o cooling: ( ) ( ) q = U T T = h T T (1) cw s where h is the ilm heat transer coeicient on the solution side and U is the overall heat transer coeicient, calculated rom: 1 1 1 R R δ δ = + + + = R + + U h h cw w else λ λ Here, δ and λ are the thickness and thermal conductivity o the ouling deposit; R cw and R w are the resistance to heat transer on the coolant side and through the ase plate(s), respectively. Both o the latter terms are expected to remain constant during a ouling experiment, while R cw is expected to e weakly related to rotational speed owing to the strong inluence o the jet on the low pattern within the can. Tale 1 summarises the thermal resistance o the ixed components in the heat transer coniguration. Tale 1: Heat transer properties o components. Material λ δ R (W/m K) (m) (m 2 K/W) Brass 19.9.83 1-4 Stainless steel 16.4 2.45 1-4 Heat lux sensor n/a negligile 5. 1-4 Coolant.58.64-5. 1-4 PPP deposit/ ulk parain.15 varies varies The value o R w, at approximately 8.3 1-4 m 2 K/W, corresponds to heat conduction through a parain layer o thickness 6 µm: R w is not, thereore, expected to e a controlling actor in heat transer. Model solutions Heat transer experiments utilised liquid parain (density at 2ºC, 855 kg/m 3 ), which was used as the solvent in the model solutions. Tripalmitin, PPP, was otained as 9% pure and dissolved in parain to give 1 wt% solutions. The apparent viscosity was measured using a Bohlin CV12 controlled stress rheometer with 5 mm parallel plates and ound to e independent o PPP (2) concentration aove the cloud point. The data were ound to ollow a temperature dependency o the orm µ = 375.74 exp(.31t) (3) where T is in Kelvin. The cloud point o solutions was measured using a test apparatus similar to that reported y the European Oleochemicals and Allied Products Group (1987) and yielded a T c value o 37ºC or the 1 wt% solution used in these ouling tests. The reezing point o the PPP was measured using a Pyris 1 DSC (Perkin Elmer, UK) itted with a rerigeration intercooler. The value otained, o 61.8ºC, compared avouraly with the trend in data reported y Fitzgerald et al. (24), o 63ºC or 95 wt% PPP and 65 ºC or 99 wt% PPP. The melting points o the solutions were also measured using DSC. The melting points o solutions across the range 2-3 wt% PPP were consistently higher (y c. 12 K) than their corresponding T c values, and could e descried y the solid-liquid equilirium relationship expected or an ideal solution and pure solid as descried y Atkins (1997). The deposits ormed during ouling could e recovered and analysed. For instance, the rheology was characterised using a Bohlin CV12 controlled stress rheometer and the particle distriution and size y scanning electron microscopy and laser scattering. Details o the characterisation methods are given in Nigo (28). Experimental methods The reservoir was charged with 2 L o the test solution and heated to the desired ulk temperature y circulating hot water through the heating jacket and mixed y a magnetic stirrer. The cooled can was initially isolated rom the reservoir and rought to the required temperature y circulation o coolant. A support rame was constructed to hold the can and motor assemly horizontal eore and ater immersion. Once temperatures had equilirated, condensate was removed rom the can assemly, the disc cleaned with hexane and dried. The can was then immersed in the solution and rotation started. It was important at this point to inspect the disc surace visually or air ules, as these can aect heat transer and deposition. Air ules could usually e displaced y increasing the rotation speed. Two sets o experiment were perormed, termed heat transer tests and ouling studies. Heat transer tests. The main purpose here was to test the heat transer perormance and operaility o the unit. The results were compared with CFD simulations. These experiments were perormed using liquid parain with temperature driving orces, T = T T cw, ranging rom 17-52 K, and rotational speeds, ω d, rom 3-6 rpm. Heat lux and temperatures were recorded over 3 minutes to ensure that any transients had een eliminated. The eect o T was investigated with T held constant, at 6 o C, while T cw was varied etween 8 and 43 o C at a can rotation speed o 6 rpm. The eect o ω d was studied at T = 28 K with T = 5ºC and T cw = 22ºC. www.heatexchanger-ouling.com 265

Nigo et al. / Experimental and CFD Studies o Fat Fouling Fouling tests. Fouling tests reported here were conducted with 1 PPP wt% solutions at T = 5ºC. The coolant temperatures used were 2 o C, (T c 5) K, and (T c 15) K. The lowest value, 2 o C, relects winter conditions in the UK and can e readily generated in a laoratory chiller over extended periods, whilst the latter values represent dierent degrees o sucooling. Rotational speeds used were 3, 33 and 6 rpm, corresponding to Reynolds numer, Re r, values o 11, 118 and 215, respectively. The Reynolds numer is deined as 2 4ρω r d d Re = (4) r 6 µ where r d is the radius o the disc and the physical properties are evaluated at the ilm temperature. The experimental conditions used in the tests are summarised in Tale 2. Tale 2: Summary o experimental conditions. Heat transer Fouling Coolant temperature, T cw 17 52 o C 2 o C, (T c 5) (T c 15) Bulk temperature, T ( o C) 6 5 PPP concentration, (wt%) 1 Rotational speed, ω d (rpm) 3-6 3, 33, 6 Temperatures and heat luxes were monitored over a ouling test. At the end o the test, the rotation was stopped and the can assemly lited o the main unit, placed on the support rame and let standing or aout 2 minutes to allow excess solution to drip o the test plate. The gel ormed on the test plate, including residual solution held y surace tension, was then careully scraped o using a plastic spatula, weighed and stored or analysis. The amount o residual solution could e signiicant so a lank run was perormed ater each ouling test to determine how much liquid remains on the ouling cell plate as a result o surace tension. The test plate was cleaned thoroughly, the can lowered into the reservoir and rotated at the experimental conditions or one minute eore withdrawing it and resting it on the support rame or 1-2 minutes. Liquid adhering to the test plate was removed and weighed. This amount was sutracted rom the measured ouled mass to give the true deposit mass. It should e noted that ouling tests could last 24 h or longer and a small numer o tests were repeated in order to gauge the reproduciility o the approach. These displayed good agreement so tests were thereater only repeated when the results were inconsistent with oserved trends. NUMERICAL SIMULATIONS Laminar low aout a rotating disc immersed in a large ody o quiescent luid was irst studied y von Kármán (1921). Surace temperature and the shear stress acting on the disc surace are key parameters in reezing ouling. The commercial inite element method (FEM) sotware COMSOL MULTIPHYSICS TM (version 3.5, Chemical Engineering Module), was used or simulating the luid low and heat transer ehaviour o pure parain liquid in the SDA, i.e. simulating the heat transer experiments, and are compared with experimental measurements o heat lux. Simulations o ouling experiments were not attempted. The low-ield is simulated y solving the continuity equation and the axisymmetric, incompressile, steady state Navier-Stokes (NS) equation or a Newtonian liquid. All lows are laminar. The steady state energy equation with no heat source or heat sink can e written as: ( ) ( ) ρ C v T = λ T (5) p, where T is the temperature, C p, the ulk speciic heat capacity and λ the ulk thermal conductivity. Physical properties such as density, thermal conductivity and speciic heat did not change signiicantly with temperature and are assumed constant. The temperature dependence o the dynamic viscosity is incorporated and was modelled y Eqn. (3). The physical and thermal properties used in the simulations are summarised in Tale 3. Tale 3: Summary o parameters used in CFD simulations. Parameters Value Radius o disc, r d.4 m Bulk temperature, T 5 o C Coolant temperature, T cw 22 o C Rotational speed o can, ω d 3-6 rpm Apparent viscosity o ulk, µ (2ºC):.4 kg/m s (5ºC):.16 kg/ms Density o ulk, ρ 855 kg/m 3 Thermal conductivity o ulk, λ.15 W/m K Speciic heat capacity o ulk, C p, 217 J/kg K Rotational speed o stirrer, ω mag 2. rad/s The physical coniguration is cylindrically symmetric and the geometry o the model is illustrated in Fig. 3. Axisymmetry allows considering the computational domain as hal o the system to e modelled. The mesh contains approximately 5 triangular elements, with a higher concentration o elements at the oundary etween the disc and the liquid (approximately ive times greater than the other oundaries). The numer o elements was optimized y perorming a series o simulations with dierent mesh sizes, starting rom a coarse mesh and reining it until the results were mesh-independent. A converged solution took approximately 15 min on a desktop PC with a 3.16 GHz dual core processor and 3.33 GB RAM. (A) ase o disc (B) axis o symmetry (C, D) magnetic stirrer (H) side surace o can (G) liquid surace (F) wall - heated jacket (E) wall - heated jacket Fig. 3: FEM mesh o the simulation domain showing oundary laels (A-H). The darkness o the areas in the igure indicates the density o the mesh. z, v z v r, v r www.heatexchanger-ouling.com 266

Heat Exchanger Fouling and Cleaning VIII 29 Convergence was assessed y comparing the values o velocity and temperature rom successive iterations; tolerances were set at 1 5 m/s (versus a lowest mean tangential velocity o the cooling can o 1.3 1-2 m/s) and 1 5 K (versus a lowest coolant temperature o 2 o C), or the velocity and temperature, respectively. The tolerance dictates the error in each iteration. The quantitative inormation speciied or each simulation is the rotational speed o the can, ω d, that o the magnetic stirrer, ω mag, and the temperatures o the coolant, T cw, and the ulk warm solution, T. The outputs o the CFD calculation are the velocity ield and temperature proile. The latter allows the heat lux across the region o the rotating disc eneath the heat lux sensor in the experimental apparatus (Fig. 2) to e calculated and compared with experimental data. The oundaries are laelled (A-H) on Fig. 3 and are suject to the ollowing conditions: (A) Base o disc: Uniorm temperature: the surace temperature, T ss,out, is assumed to e the coolant temperature, T cw. It is shown, later, in Fig. 7 that the resistance or the ase plates, R w, is small compared to the thermal resistances o the ulk, R, and coolant, R cw. Thereore, it is reasonale to assume that the temperature o the disc is uniorm. The oundary is impermeale and the rotational speed is speciied via v = ω r (6) θ d where v θ is the velocity component in the azimuthal direction and r the radial coordinate. (B) Axis o symmetry: There is no luid or thermal energy low across the line o symmetry, so it is adiaatic. (C, D) Magnetic stirrer: This oundary is adiaatic and impermeale. The rotational speed is speciied, at v = ω r (7) θ mag (E, F) Wall-heated jacket: The inner wall temperature is speciied, at T, the temperature o the solution. The oundary is impermeale and there is no slip at the wall. (G) Liquid surace: There is little heat loss rom the liquid surace, so is treated as adiaatic. This ree surace is modelled with slip conditions: vz = and v = θ (8) where v z is the velocity component in the axial direction. (H) Side surace o can: The wall is insulated so is treated as adiaatic, with rotational speed given y Eqn. (6). RESULTS AND DISCUSSION Heat transer Fig. 4 shows a sample set o experimental data rom the heat transer experiments. The heat lux is linearly T = T T, proportional to the temperature driving orce ( ) as expected, and the overall heat transer coeicient, U exp, can e extracted rom the regression line. The eect o dimensionless disc speed, Re r, on heat lux at ixed T (and thereore U) is presented in Fig. 5. The itted trend line shows that the heat lux varies with Re r.51, cw indicating that the overall heat transer coeicient, U exp, is roughly proportional to ω. qexp [W/m 2 ] 35 3 25 2 15 1 5 1 2 d y = 65.28x R 2 =.998 1 2 3 4 5 6 T [K] Fig. 4: Eect o temperature driving orce, ( T T T ) =, on measured heat lux. Conditions: liquid parain at 5ºC, ω d = 5.2 rad/s (5 rpm) and ω mag = 2. rad/s. Symol size relects experimental uncertainty. Solid line shows regression it. qexp [W/m 2 ] 1 2 3 4 5 6 7 25 2 15 1 5 y = 132.24x.51 R 2 =.99 Rotational speed [rpm] 5 1 15 2 25 Fig. 5: Eect o Reynolds numer, Re r, on heat lux. Locus shows line o est it or simple power law model. Conditions: liquid parain, T cw = 22 o C, T = 5 o C and ω mag = 2. rad/s. A similar relationship was otained y Sparrow and Gregg (1959), in their investigations o the heat transer characteristics o rotating discs located in a large pool o quiescent liquid. These results indicate that the SDA is operating in the laminar regime and that the conditions employed in the experiments did not exceed the sensor sensitivity. CFD simulations The CFD simulation predicts the velocity and temperatures distriutions in the liquid in the heat transer experiments. Mass transer, which can also e involved in limiting ouling, is not considered ut could e readily included. Fig. 6 shows the temperature proiles (coloured map) and low patterns (contour lines) or a set o simulations at can rotational speeds employed in the experiments reported in Fig 5. Two vortices are evident in the ulk liquid: an upper one driven y the rotation o the can and a lower one induced y the magnetic stirrer acting in the opposite direction. As the magnetic stirrer speed is Re r cw www.heatexchanger-ouling.com 267

Nigo et al. / Experimental and CFD Studies o Fat Fouling kept constant, increasing the can speed increases the size o the upper recirculation zone, as expected. It is also evident that the rotational speed has an eect on the low patterns and thus the temperature proiles. At high can speeds, i.e. > 1 rpm, the temperature o almost the entire domain approaches that o the ulk. Note that CFD validation is not presented in this work as there is no single parameter that allows direct comparison using the current setup. However, the use o particle imaging velocimetry to study the low patterns o the ulk solution is planned and this will allow us to conirm the low ield predictions. Fig. 6: Flow patterns and temperature proiles in the SDA or can rotational speeds o 3 rpm (let) and 6 rpm (right). Black arrows are velocity vectors. Shading indicates temperature, with 5 o C dark red, 22 o C dark lue. The ilm heat transer coeicient on the ulk side, h,sim, can e calculated rom the temperature proiles and these are compared with the overall heat transer coeicient, U exp, otained rom experiments. The h,sim values were consistently larger than the U exp values, which is expected as the latter includes the resistances across the can and coolant. R = R + R, can e estimated The latter resistance, ( ) else cw w rom (1/U exp 1/h,sim ), according to Eqn. (1), and the results are plotted in Fig. 7. Thermal resistances [m 2 K/W] Cooling can Magnetic stirrer 1 2 3 4 5 6 7.5.4.3.2.1 Cooling can Magnetic stirrer Rotational speed [rpm] R R cw R w indicating that the dominant resistance to heat transer lies on the solution side. R else varies rom.2 m 2 K/W to.6 m 2 K/W, which is noticealy greater than the estimated value o R w, o.82 m 2 K/W (Tale 1), suggesting that the coolant side resistance, R cw, is signiicant. This also implies that wall resistances play a minor part and the assumption that the wall is at uniorm temperature is reasonale. Rotation speed has a larger eect on R than R cw, which is expected as the coolant low is also determined y the internal circulation in the can. The low ield within the can was not simulated as initial estimates o Reynolds numers indicated that the low lay in the turulent regime, requiring extensive urther computational eort. The surace temperature o the disc in contact with the warm solution in ouling experiments (eore ouling occurs) will e near, ut not at, T cw. Fig. 7 suggests that a working estimate o surace temperatures could e made 1 using R ~ R and assuming R w eing negligile, giving: cw 2 ( T T ) ( T T ) ss,out cw cw q = = R R R ( + ) cw cw R T = T + T T = T + T 1 2 2 1 ss,out cw ( cw ) 1 3 cw 3 R 2 + R (9) (1) The shear rate and the shear stress imposed on the surace disc can also e calculated rom the simulation velocity ield. The shear stress distriutions show that the maximum shear stress is ound at the outer edge o the disc. The same trend was oserved or shear rates and or all other temperature and rotational speeds investigated. Figure 8 shows the shear stress values at r =.35 m (the radius o the disc r d is.4 m). It is also evident that the eect o rotational speed is greater than the eect o surace temperature. The shear stresses imposed on the surace in the SDA device can e compared with those imposed y an oil in turulent low. For a ulk velocity o 1 m/s, an oil density o 8 kg/m 3 and a Fanning riction actor o c..5, this gives τ = ½ C ρ u 2, ~ 2 Pa. This estimate suggests that inormation on ouling ehaviour can e otained at the laoratory scale in the SDA using relatively simple measurements under conditions relevant to industrial operation.. 5 1 15 2 25 Re r Fig. 7: Thermal resistances in the SDA apparatus: = R R. R ( 1 h,sim ) =, R w and R cw ( ) Both resistances decrease with increasing Re r, and R is consistently larger than R else at all rotational speeds, else w Surace temperature Fig. 8: Shear stress values on disc surace or dierent values temperatures and rotational speed at radial location r =.35 m. www.heatexchanger-ouling.com 268

Heat Exchanger Fouling and Cleaning VIII 29 Fouling experiments Results presented here are primarily the mass o deposit ormed, m, ouling resistance, R, and the inerred deposit thickness, δ. The reproduciility o ouling tests was conirmed y repeated tests with 1 wt% PPP solutions at T cw = 2 C and ω d = 3 and 6 rpm (Re r = 11, 215). The data in Fig. 9 show agreement within the ounds o experimental error, as well as noticealy dierent deposit mass-time proiles. At higher ω d, i.e. 6 rpm, there is a short induction period ollowed y rapid growth up to 4 h, ater which deposition was slow. At lower speed, 3 rpm, no induction period was oserved, with 6 g o deposit ormed ater 1 h; deposit growth thereater is slow, reaching a slightly larger inal value than at 6 rpm ater 24 h. m [g] 14 12 1 8 6 4 2 4 8 12 16 2 24 28 Time [hr] 3 rpm 6 rpm Fig. 9: Reproduciility o ouling runs. Conditions: T cw = 2 o C, T = 5 o C: circles - 3 rpm, triangles - 6 rpm. The dierent symols indicate separate runs. The asymptotic or ouling rate ehaviour oserved is expected as the tests are perormed under conditions o constant overall temperature driving orce: as deposit accumulates, the deposit-solution interace temperature, T s, will increase. Estimates o T s or the proiles in Fig. 9 conirmed that T s approached T c at the end o the test. The dierence in ehaviour etween the high and low rotational speeds is elucidated y the heat transer proiles in Fig. 1(a) (c), which were otained under similar conditions. The ouling resistance, R, shown in Fig. 1 (), is calculated rom 1 1 R = U U (1) o where U o is the initial, clean, overall heat transer coeicient. This is most readily estimated y extrapolating the q-t data ack to t =, as the early values contain transients associated with the start o rotation. The heat luxes otained at the lower speed, i.e. 3 rpm, are 3-4 times smaller than those otained at higher rotational speeds so contain more measurement scatter, ut the data clearly show a sharp initial increase in R, mirroring that seen in the mass deposition measurements. This can e attriuted to the ormation o a weak gel on the surace due to the low temperature in the liquid which is ale to resist removal as the shear induced y the rotation is low. This is not oserved at higher rotational speeds ecause the shear stress is large enough to shear o the weak gel ormed at this temperature. The deposit thickness proiles in Fig. 1(c) were estimated using δ = R λ (11) where the deposit thermal conductivity, λ, was taken to e.15 W/m K, as the thermal conductivity o solid PPP is conveniently close to that o the parain. The plots show a steady increase to a inal thickness o 2-3 mm, which is consistent with visual oservations and deposit volume. (a) () (c) qexp [W/m2] R [m 2 K/W] δ [mm] 25 2 15 1 5.4.3.2.1 1 2 3 4 5 6 7 Time [min] 6 rpm 33 rpm 3 rpm. 1 2 3 4 5 6 7 6. 5. 4. 3. 2. 1. Time [min] 3 rpm 33 rpm 6 rpm. 1 2 3 4 5 6 7 Time [min] 3 rpm 33 rpm 6 rpm Fig. 1: Fouling o 1 wt% PPP solution: (a) heat lux, () ouling resistance and (c) estimated deposit thickness. Conditions: T cw = 22 o C, T = 5 o C, ω mag = 2. rad/s. Electron microscopy and X-ray analysis o the deposits conirmed that the PPP was crystallizing as pure needles o the β polymorph. The composition o the oulant was ound to e strongly inluenced y the surace conditions, particularly temperature and shear stress: at 3 rpm, the deposit was approximately 3 wt% PPP solids whereas at 6 www.heatexchanger-ouling.com 269

Nigo et al. / Experimental and CFD Studies o Fat Fouling rpm the solids content was closer to 6 wt%. These values indicate that the structure o the deposit varies and is determined y the physics o gelation. Separate rheometrical analysis o PPP-parain gels at 2 ºC indicated that they exhiited yield-stress ehaviour, with a critical stress around 2-3 Pa, which is consistent with the shear stress values in Fig. 8. Further analyses and experimental studies are reported in Nigo et al. (29). CONCLUSIONS 1. A novel ouling apparatus, the SDA, has een developed or studying reezing ouling using moderate volumes o liquid. Fouling can e monitored in situ and samples readily recovered or analysis. 2. The laminar lows in the SDA in this reezing ouling application can e simulated using CFD techniques, yielding good agreement with heat transer measurements and providing reliale estimates o surace conditions. 3. A preliminary investigation o reezing ouling using model solutions o PPP in parain highlighted the importance o gel ormation conditions on ouling ehaviour. ACKNOWLEDGEMENTS Financial support rom the Association o Commonwealth Universities (UK) and the Papua New Guinea University o Technology or RYN, and unding or YMJC rom the Royal Academy o Engineering are all grateully acknowledged. Construction o the SDA was supported y a Food Processing Faraday Fast Track grant. NOMENCLATURE C Fanning riction actor C p speciic heat capacity, J/kg K g gravitational constant, m/s 2 h heat transer coeicient, W/m 2 K m mass o deposit, kg q heat lux, W/m 2 r radial coordinate, m R thermal resistance, m 2 K/W Re r Reynolds numer ased on radius o rotating disc T temperature, o C or K u mean velocity, m/s U overall heat transer coeicient, W/m 2 K v velocity vector, m/s z axial coordinate, m δ thickness, m λ thermal conductivity, W/m K µ dynamic viscosity, kg/m s θ azimuthal coordinate, o ρ density, kg/m 3 τ shear stress, Pa ω rotational speed, rad/s Suscript ulk c cloud point cw coolant d disc else everything else apart rom ulk exp o mag w s sim ss,out experiment ouling deposit initial magnet comined setup o rass, heat lux sensor and stainless steel surace o deposit in contact with ulk solution simulation outer surace o stainless steel disc REFERENCES Akarzadeh, K., and Zougari, M., 28, Introduction to a novel approach or modeling wax deposition in luid lows. 1 Taylor-Coutte system. Ind. Eng. Chem. Res., Vol. 47, pp. 953-963. P.W. Atkins, 1997, Physical Chemistry (6 th ed.), Oxord University Press, Oxord, UK. Epstein, N., 1983, Thinking aout heat transer ouling: A 5 5 matrix. Heat Transer Eng., Vol. 4(1), pp. 43-56. Fernandez-Torres, J.M., Fitzgerald, A.M., Paterson, W.R., and Wilson, D.I., 21, A Theoretical study o reeze ouling: limiting ehavior ased on a heat and mass transer analysis. Chem. Eng. Proc., Vol. 4, pp. 335-344. Fitzgerald, A.M., Barnes, J., Smart, I., and Wilson, D.I., 24, A model experimental study o coring y palm oil ats in distriution lines. Food Bioprod. Process., Vol. 82, pp. 27-212. J.P. Fryer, D.L. Pyle and C.D. Rielly, 1997, Chemical Engineering or the Food Industry, Blackie Academic and Proessional, London, UK. Ghedamu, M., Watkinson, A.P. and Epstein, N., 1997, Mitigation o wax oil uild-up on cooled suraces, in Fouling Mitigation o Industrial Heat Exchange Equipment, eds. Panchal, C.B., Begell House, NY, pp. 473-489. Grant, C.S., Perka, A.T., Thomas, W.D., and Caton, R., 1996, Cleaning o ehenic acid residues rom stainless steel. AIChE J., Vol. 42, pp. 1465-1476. Jennings, D.W. and Weispennig, K., 25, Eects o shear and temperature on wax deposition: coldinger investigation with a Gul o Mexico crude oil. Energ. Fuel., Vol. 19, pp. 1376-1386. Kármán, Th.Von, 1921, Uer laminare and turulente. Zeitschrit ür Angewandte Mathematik und Mechanik, Vol. 1, pp. 233-252. Nigo, R.Y., 28, Fouling ehaviour o ood ats, PhD Dissertation, University o Camridge, UK. Nigo, R.Y., Chew, Y.M.J., Houghton, N.E. Paterson, W.R. and Wilson, D.I. (29) Experimental studies o reezing ouling o model ood at solutions using a novel spinning disc apparatus, in preparation. Parthasarathi, P., and Mehrotra, A.K., 25, Solids deposition rom multicomponent wax-solvent mixtures in a enchscale low-loop apparatus with heat transer. Energ. Fuel., Vol. 19, pp. 1387-1398. Rashaida, A.A., Bergstrom, D.J., and Sumner, J.R., 26, Mass transer rom a rotating disk to a Bingham luid. J. Appl. Mech., Vol. 19, pp. 18-111. Rosmaninho, R., and Melo, L.F., 26, The eect o citrate on calcium phosphate deposition rom simulated milk www.heatexchanger-ouling.com 27

Heat Exchanger Fouling and Cleaning VIII 29 ultrailtrate (SMUF) solution, J. Food Eng., Vol. 73(4), pp. 379-387. Singh, P., Venkatesan, R., and Fogler, H.S., 21, Morphological evaluation o thick wax Deposits during aging. AIChE J., Vol. 47(1), pp. 1-18. Sparrow, E.M. and Gregg, J.L., 1959, Heat transer rom a rotating disk to luids o any Prandtl numer. J. Heat Trans., Vol. 81, pp. 249-251. www.heatexchanger-ouling.com 271