roo

an

utio

784bUS Army, Engineer Research and Development Center, Construction Engineering Research Laboratory, 2902 Newmark, Drive,

Champaign, IL 61822-1076, United States

ating agent type and concentration is required to prevent the supercooling phenomenon. Pressure drop and convective heat transfer datawere measured using a heat transfer loop operated at dierent ow rates and heat ux values. Results indicate that the phase change

fourfold [1]. However, wide implementation of PCM incooling systems has been limited by several operational fac-tors including clogging of pipes and limited heat transfer

in recent years. However, MPCMs also host a series ofimplementation concerns including supercooling, durabil-ity, increased pressure drop and limited heat transfer. Inthe end, the heat transfer eectiveness of any MPCM slurrymust be fully characterized and understood to become afully viable heat transfer uid.

* Corresponding author. Tel.: +1 979 458 1900; fax: +1 979 862 7969.E-mail address: alvarado@entc.tamu.edu (J.L. Alvarado).

International Journal of Heat and Mass Tprocess and slurry mass fraction aect the heat transfer process. The paper also examines the impact of using enhanced surface tubingin combination with MPCM slurry under constant heat ux and turbulent conditions. 2006 Elsevier Ltd. All rights reserved.

Keywords: Microencapsulated phase change material slurry; Turbulence; Supercooling; Enhanced surface

1. Introduction

Recently, researchers and engineers have investigatedthe eects, advantages and disadvantages of incorporatingphase change materials in secondary heat transfer uidswith the objective of increasing thermal capacity in districtcooling applications. More recently, research has shownthat heat capacity in those systems can be increased by

rates in heat exchangers [2,3]. Recent developments havedemonstrated that microencapsulation of phase changematerial (MPCM) can be applied to circumvent the prob-lems associated with PCMs. By microencapsulating andisolating the phase change material from its surroundingsand the carrier uid, the phase change material is less likelyto hamper the heat transfer process. MPCM slurries haveindeed become a viable option for heat transfer processescUS Army, Engineer Research and Development Center, Cold Regions Research and Engineering Laboratory, 72 Lyme Road,

Hanover, NH 03755-1290, United StatesdMechanical and Industrial Engineering Department, University of Illinois at Urbana-Champaign, 140 Mechanical Engineering Building,

206 West Green Street, Urbana, IL 61801, United States

Received 15 May 2006; received in revised form 29 September 2006Available online 30 November 2006

Abstract

Current cooling and heating distribution systems that use water as secondary uid exhibit limited thermal capacity which can only beovercome by high ow rates and large (volume) capacity. A successful way to enhance the thermal capacity of secondary uid systems isby incorporating microencapsulated phase change material (MPCM) slurry. However, a full understanding of the physical properties andheat transfer characteristics of MPCM slurry in the 28 C range (35.546.5 F) still is lacking. In the paper, latent heat of fusion, meltingand freezing points, and temperature- and concentration-dependent viscosity data, are presented. Results indicate that selection of nucle-Thermal performance of micmaterial slurry in turbulent

Jorge L. Alvarado a,*, Charles Marsh b, ChaDepartment of Engineering Technology and Industrial Distrib

College Station, TX 70017-9310/$ - see front matter 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.ijheatmasstransfer.2006.09.026encapsulated phase changew under constant heat ux

g Sohn b, Gary Phetteplace c, Ty Newell d

n, Texas A&M University, 117 Thompson Hall, 3367 TAMU,

3-3367, United States

www.elsevier.com/locate/ijhmt

ransfer 50 (2007) 19381952

LHF latent heat of fusion, kJ/kg

k latent heat of fusion, kJ/kg

He_m mass ow rate, kg/sMF mass fractionNu Nusselt number, hL/kfPr Prandtl number, cpl/kPe Peclet number, RePrq00w wall heat ux, W/m

2

q00 heat ux, W/m2Nomenclature

C heat capacity, kJ/kgcp specic heat, kJ/kg Cdp particle diameter, lm or md pipe or tubing diameter, m or mmf Darcy friction factorg gravity constant, m/s2

h convective heat transfer coecient, W/m2 C orkW/m2 C

hf head or pipe loss, mk thermal conductivity, W/m Ck+ dimensionless roughnessk roughness, mL tubing length, mLe characteristic turbulence length scale, m

J.L. Alvarado et al. / International Journal ofIn this paper, thermophysical properties data of MPCMslurry, including latent heat of fusion, apparent specicheat, melting and crystallization temperature points, andapparent viscosity, are presented. The paper also presentshow enhanced surface copper tubing combined withMPCM slurry impact the heat transfer process under con-stant heat ux and turbulent conditions.

2. Background

2.1. Phase change materials

For several years, research activities have shown thatPCMs can enhance the heat capacity of secondary uids.Researchers have tested parans and ice slurries as heatcapacity enhancers. Bellas et al. [4] used ice slurries toincrease the thermal capacity of an existing system. Choi[3] and others [510] have used parans because their melt-ing points are at least few degrees greater than the carrieruid (water). Full implementation of plain PCM in second-ary uid applications has been limited due to the tendencyof PCM to form clumps, which exhibit lower thermal con-ductivity during the solidication process. Also, PCM stud-ies have revealed that higher mass fractions yield highereective specic heat [3]. However, higher mass fractionsalso result in higher apparent viscosity, which translatesinto greater pumping power [3,4,7] and lower thermal con-

R pipe radius, mRe Reynolds number, Vdq/lRep particle Reynolds number, Vdpq/le roughness, m/ percent of particles migrating to before-melting

regionq uid density, kg/m3

m kinematic viscosity, m2/sl, g dynamic viscosity, cP or mPa ss uid shear stress, PaDP pressure drop, kPa

Subscripts

B bulkSte Stefan numberthp particle thickness, m or lmTbulk bulk uid temperature, CTwall wall temperature, CU* friction velocity, m/sV uid velocity, m/sy particle size, m

Greek symbols

a heat diusivity, k/qcp

at and Mass Transfer 50 (2007) 19381952 1939ductivity near or on the boundary layer. Choi [3] observedthat melted PCM tends to remain near the surface of theheat-conducting pipe while still-solid emulsions remainnear the uid core. Since liquid parans have lowerthermal conductivity than water, the overall heat transferprocess is critically limited which outweighs any coincidentenhancements in heat capacity. Despite the use of emulsi-ers [3], accumulation of frozen PCM particles on heatexchanger surfaces has been observed, which limits heattransfer. Choi [3] also observed signicant changes in pres-sure drop once the PCM emulsion had melted.

2.2. Microencapsulated phase change material

Few journal articles concerning MPCM physical prop-erties have been published. Roy and Sengupta [11] con-ducted experimental studies to evaluate the properties ofMPCMs with two characteristic thicknesses. Dierentialscanning calorimetery (DSC) was used to determine thethermal properties. Yamagishi et al. [12] presented experi-mental results indicating that MPCM particle size doesnot aect melting temperature and latent heat of fusion.However, the degree of supercooling or the dierencebetween crystallization and melting temperature pointsincreased when particle size, dp, was less than 100 lm. Asa result, Yamagishi et al. [12] also used and tested nucleat-ing agents with molecular structure similar to the PCM

MPCM microencapsulated phase change materialP particleW wall

yields higher viscosity and suppressed turbulence. In lami-

Hemolecules and were able to suppress the supercooling eectconsiderably. Alvarado [13] conducted a series of experi-ments and evaluated the thermal properties of tetradecanemicrocapsules with an average size of 4.4 lm. Yamagishiet al. [2] presented empirical data for microencapsulatedoctadecane obtained by using DSC equipment, a Couetteviscometer [12], and a heat transfer loop [2]. Their resultsindicate that anionic surfactants can decrease the apparentviscosity of MPCM slurry and turn it into Newtonian-likeuid at high mass fractions (2030%). The results also indi-cate that the relative viscosity of the MPCM slurry wasindependent of temperature. As a result, Yamagishi et al.[2] used the Vand model [14] to predict the relative viscosityof the MPCM slurry at dierent volume fractions.

Ohtsubo et al. [15] presented experimental data thatexplained why microcapsules fail structurally. Severalexperiments show that as dp/thp increases, the percentageof broken capsules increases, where dp and thp are micro-capsule diameter and thickness, respectively.

A limited number of journal articles present and discussheat transfer and pressure drop data and characteristics forMPCMs. Goel et al. [16] describe experiments of MPCMslled with n-eicosane where the experimental conditionswere limited to laminar ow and constant heat ux. Theirresults indicate that the Stefan number (Ste) was the mostdominant parameter, especially for a Ste less than 1.0. Royand Avanic [17] showed that laminar forced convectionheat transfer characteristics for PCM are similar toMPCMs. Based on observations and past research activi-ties, MPCM walls seem not to have a signicant impacton the heat transfer process. For MPCM microcapsules(less than 20 lm) lled with n-tetradecane, the calculatedBiot number is less than 0.1. Results also indicate thatthe Reynolds number plays a signicant role in the heattransfer process. Yamagishi et al. [12] experimental datashows that MPCM slurry approximately follows theBlasius equation. They also conducted several heat transferexperiments with uniform heat ux and turbulent condi-tions. In their research, MPCM particles were made ofoctadecane (C18H38) which varied in size between 2 and10 lm. Results show that as mass fraction increases, turbu-lent ow changes to laminar, which consequently changesthe heat transfer characteristics of the MPCM slurry.Unlike the work by Choi [5], no evidence of pressure dropuctuations can be found when PCM melts within theMPCM particles, and a constant relative viscosity isobserved when the slurry temperature increases [12,13].At high mass fraction, pressure drop decreased, indicatingthat the slurry became laminarized. One important obser-vation was the eect of particle size on the convective heattransfer coecient. It is known that particle size [18,19] caneither enhance or suppress turbulence which aectsmomentum and heat transfers. In the case of small MPCM,turbulence is signicantly suppressed [2,13]. One signicantassumption is that all microcapsules are small, hence, they

1940 J.L. Alvarado et al. / International Journal ofmelt and solidify instantaneously. Experimental data [2,13]suggest that at an identical heat ux, water has a higher hnar ow, MPCM benets are limited because solid andmelted microencapsulated PCM segregates around the coreuid [2]. However, recent publications [20,21] based onnumerical simulations show that laminar convective heattransfer of MPCM slurries can be enhanced by tweakingseveral variables including Reynolds number and particlesize.

2.3. Enhanced surfaces

Durmus et al. [22] present experimental results wheresnail-type swirl generators were used to augment heattransfer rates. Results indicate that Nusselt number (Nu)can be increased from 80% to 200% for 1575 swirlingangles for air in a counter-ow conguration, but pressuredrop increased by 110%. Performance was better whenhigh swirling angles and low Reynolds numbers were used.Fossa and Tagliaco [23] tested and measured heat transferrates with and without smooth, nned, grooved pipeswhich indicate that polymeric additives reduce friction fac-tor but also decrease the Nusselt number. Finned andgrooved pipes show a sharper reduction in Nusselt numberthan smooth pipes for concentrations of 40 ppm and Rey-nolds number between 7000 and 10,000. Fanning frictionfactor, f, decreases with respect to pure water only in turbu-lent ows, by about 25% for nned or grooved pipes andabout 20% for smooth pipes. Results also indicate that con-ductance is greater for nned pipes than for smooth pipeswith or without additives. Results also indicate that frictionfactor reduction reaches a minimum at a specic Reynoldsnumber and additive concentration. Liao and Xin [24]than MPCM slurry. It was concluded that changes andlower values in viscosity promoted turbulence better inwater than in MPCM slurry. Yamagishi et al. [2] usedChois model [3] to predict the local Nusselt number, whichis dened as

Nux 0:00425 Re0:979bx Pr0:4bx gwxgbx

0:111

where, Nux, Rebx, Prbx, gwx, and gbx are local Nusselt, localReynolds, and local Prandtl numbers respectively, andlocal wall and bulk viscosities, respectively. They observedthat Chois model could be used to predict after-meltingbehavior (if the single-phase approximation is based onRex and Prx). However, the model poorly predicted the be-fore- and during-melting Nux because the MPCM phasechange process aects cp and h. Experimental data showthat higher heating rates yields lower h, which can beattributed to a thicker boundary layer of melted PCM.Data also show that at the same Reynolds number, h in-creases with mass fraction. It was also evident that higherReynolds number favors a higher h more than positivechanges in mass fraction because higher mass fraction

at and Mass Transfer 50 (2007) 19381952present experimental results on heat transfer and frictioncharacteristics for various liquids with turbulent, transi-

tional, and laminar ow inside a pipe with twisted-tapeinsert. Results indicate that in turbulent and transitionalows, heat transfer is increased slightly while pressure dropincreases signicantly. When VG46 turbine oil ow is lam-inar, the Stanton number is 5.8 times higher when atwisted-tape insert is used instead of a smooth pipe. Thefriction factor, f, increases 6.5 times also under the sameconditions. Stanton number and f with twisted-tape insertsincrease with lower tape twist ratio. Segmented twisted-

with a software package capable of measuring particle or

transfer section to make sure MPCM particles were uni-

tetradecane, a critical step in the characterization ofMPCMs. The DSC measures the amount of heat absorbedor released by a sample in comparison with a standard ref-erence. The resulting energy-temperature curve is used todetermine latent heat of fusion and melting point [2628].More information on DSC methods and equipment canbe found in ASTM E 1269.

A TA Instruments 2920 Modulated Dierential Scan-ning Calorimeter was used to determine the thermal prop-erties of MPCM slurry. Helium gas at a ow rate of26 cm3/min was used as purge gas. The DSC was calibratedby performing baseline, cell constant and temperature cal-ibration runs. The information collected from the calibra-tion runs were taken into consideration by the softwarebuilt into the DSC for computing thermal properties. Thebuilt-in software compared the data from the all the cali-bration runs and determined baseline slope, baseline oset,cell constant values, and temperature corrections. Thecombined average cell constant value was 1.09, whichwas used as a correction factor to determine how much

J.L. Alvarado et al. / International Journal of Hemicrocapsule diameter. Magnications of 100, 400and 1000 were used to determine particle size of eachmicrograph. Over 300 microcapsules were carefully mea-sured. Fig. 1a shows the particle size distribution ofMPCM slurry with the optimal supercooling and durabil-ity characteristics.

Alvarado [13] determined that MPCM slurry behaves asa homogeneous uid based on visual observations. Analy-

Table 1aCharacteristic size range of MPCM particles

Batchnumber

Phasechangematerial

Nucleatingagent

MPCM sizerange (lm)

Averagemicrocapsulediameter (lm)

1 99.8%tetradecane

0.2% fumesilica

90150 100

2 98%tetradecane

2%tetradecanol

70260 145tape inserts can decrease f by 4144% and Stanton numberby 1518% in relation to a continuous twisted-tape insert.

As seen above, only Yamagishi et al. [2] have broughttogether some of the issues concerning MPCM. In the nextsections, experimental data for microencapsulated tetrade-cane are presented including pressure drop and heat trans-fer rates results when smooth and enhanced tubing wereused.

3. Physical characterization of microencapsulated

tetradecane

3.1. Size characterization

The MPCM used in this study were made by microen-capsulating 99% n-tetradecane with gelatin through theprocess of coacervation. The process produces cross-linkedmicrocapsules in the range of 2260 lm (diameter). Severalbatches of microencapsulated n-tetradecane were preparedto determine the best chemical composition (phase changematerial and nucleating agent) to minimize the supercool-ing phenomenon and ensure the long-term durability ofthe microcapsules, the two main challenges in the design,fabrication and implementation of MPCMs. Table 1ashows the composition and particle size range for the dier-ent MPCM slurry batches that were tested.

Each batch was examined under a microscope equipped3 94%tetradecane

6%tetradecanol

210 4.4formly distributed. Therefore, no MPCM particle segrega-tion was observed at slurry velocities equal to or greaterthan 0.6 m/s.

3.2. Thermal properties characterization of MPCM slurryby using DSC

A dierential scanning calorimeter (DSC) was used todetermine the thermal properties of microencapsulatedsis shows that the total vertical displacement of a 10 lmMPCM particle in a 1.5 m heat transfer section is less10 lm at a slurry velocity of 0.6 m/s, if gravitational andbuoyant forces are taken into account [25]. Also, turbu-lence promoters were installed and used between each heat

0

5

10

15

20

25

30

35

40

0 5 10 15Particle Size [mm]

% o

f Tot

al

Fig. 1a. MPCM particle size distribution.

at and Mass Transfer 50 (2007) 19381952 1941energy was actually delivered and received by eachspecimen. The calibration and experimental runs were

tetradecane so the supercooling phenomenon could be con-trolled eectively, bulk mixtures of tetradecane and tetra-decanol at dierent ratios were tested by using DSC.Fig. 1b shows the DSC results for dierent concentrationsof tetradecanol in tetradecane. In bulk samples, 2% oftetradecanol is sucient to suppress supercooling almostentirely.

DSC results for MPCM containing 98% tetradecane and2% tetradecanol (2nd batch) that varied in size between 70and 260 lm in diameter indicate that the degree of superco-oling is considerably less than in the previous case. How-ever, durability test results later indicated that the secondbatch was unsuitable for further testing. Detailed descrip-tion of the durability experiments and their results are pre-sented later in the article. Several thermal characterizationexperiments using DSC were conducted to determine thebest combination of phase change material, nucleatingagent and microcapsule size. The results of the experimentsare presented in Table 1b.

The results shown in Table 1b reiterate previous ndingsthat microcapsule size does have an impact on the phasechange material crystallization temperature. However, dueto durability problems of larger capsules (over 100 lm),

3

4

5

6

of F

reez

ing

[C

]

Heat and Mass Transfer 50 (2007) 19381952performed by using a heating rate of 3 C/min for all theexperiments. Based on the calibration process, the DSCresults are within a 2% relative error. Every MPCM slurrysample was placed in a vial and magnetically stirred tomake sure all the particles remained in suspension, beforeconducting the DSC experiments. Specially designed sealedDSC pans were carefully used and reopened after eachDSC test to allow evaporation of the carrier uid (water)to determine the mass fraction of each sample.

DSC experiments were carried out to measure the latentheat of fusion and melting point of bulk tetradecane withand without nucleating agent. DSC results indicate that100% bulk tetradecane heat of fusion value and meltingand crystallization points are 215 J/g, 5.5 C and 0.0 C,respectively. DSC curves and data for bulk tetradecanecan be found in Alvarado et al. [29].

To avoid any degree of supercooling (dierence betweenmelting and solidication points), a batch of MPCM con-taining 0.2% of fumed silica as nucleating agent was made.The experimental data show that an average MPCM parti-cle consists of 88.3% of n-tetradecane and 11.7% of micro-capsule wall material.

Under equilibrium conditions, the melting and crystalli-zation points are identical. However, due to the size of theMPCM particles and nite cooling rates (19 C/min),MPCM particles exhibit a certain degree of supercooling.Supercooling is dened as the dierence between freezing(i.e. crystallization) and melting points. DSC experimentsclearly show that the degree of supercooling for microen-capsulated tetradecane containing 0.2% of fumed silica stillis signicant and highly undesirable since the carrier uid iswater, whose bulk melting and crystallization points are0 C. As a result, considerable amount of eort was under-taken to understand the nature of supercooling in micro-capsules. Most knowledge about supercooling is providedby the Classical Nucleation Theory (CNT) and severalexperimental results published in the past 20 years.

Classical Nucleation Theory asserts that liquid-to-solidtransformations take place because of a homogeneous orheterogeneous nucleationmechanism. Homogeneous nucle-ation relies on the formation and growth of stable nucleiwithin the microcapsule while heterogeneous nucleation isa surface-mediated mechanism. It is known that homoge-neous nucleation has a greater nucleation barrier than heter-ogeneous nucleation and entails greater supercooling or alower temperature for stable nuclei to form and grow [30].Montenegro and Landfester [31] found that nanodroplets(125500 nm) showed a considerable degree of supercooling,indicating homogeneous nucleation as the preferred type ofnucleation mechanism. The most recent experimental dataalso show that the initiation of freezing temperature doesdepend on particle size [12,30].

Yamagishi et al. [12] encountered similar diculties insuppressing supercooling and used paran-like moleculeswith higher melting points than their homologous mole-

1942 J.L. Alvarado et al. / International Journal ofcules as nucleating agents. To determine the right amountand type of nucleating agent for microencapsulated0

1

2

0% 1% 2% 3% 4% 5%% of bulk tetradecanol in tetradecane [%]

Tem

pera

ture

Temperature error: 0.1 C

Fig. 1b. Apparent specic heat of MPCM slurry (94% tetradecane + 6%tetradecanol).

Table 1bDSC results for bulk and microencapsulated tetradecane

Mixture description Tm(C)

Tsol(C)

k(J/g)

Bulk 100% tetradecane 5.5 0 215Microencapsulated 100% tetradecane 5.2 4.2 215Bulk 96% tetradecane + 4% tetradecanol 5.5 5 206.4Microencapsulated 96% tetradecane + 4%

tetradecanol5.2 2.0 206.4

Bulk 94% tetradecane + 6%tetradecane + 6% tetradecanol

5.5 5 202.1

Microencapsulated 94% tetradecane + 6%tetradecanol

5.1 2.4 202.1Note: Tm, Tsol, k are melting temperature, crystallization temperature,latent of heat of fusion, respectively. MPCM size range: 210 lm.

turbulent ow conditions for a considerable amount of

the liquid phase (including any free tetradecane) from thesolid phase (unbroken microcapsules). If a considerableamount of particles (i.e. 10% or more) broke as a resultof durability testing, smaller capsules or capsules withgreater thickness-to-diameter ratio were then tested.Ohtsubo et al. [15] showed experimentally that the thick-ness-to-diameter ratio is the primary variable that predeter-mines the long-term durability of microcapsules. The

size range were tested, and the results are summarized in

6

8

10

12

14pe

cific

Hea

t [kJ

/kg-

C]

J.L. Alvarado et al. / International Journal of Heat and Mass Transfer 50 (2007) 19381952 1943time to be able to be used in heat transfer applications. Sev-eral batches of MPCM particles were subjected to durabil-ity tests. A durability test loop was built and used todetermine what percentage of a xed amount of MPCMslurry could survive continuous pumping and surface fric-tion losses. The durability loop consisted of a closed loopmade of copper tubing that varied in diameter frommicrocapsule size should be less than 20 lm to avoidbreakage.

To estimate the eect of using MPCM made of 94%tetradecane + 6% tetradecanol in a district cooling applica-tion, the apparent specic heat as a function of mass frac-tion was calculated for a slurry temperature rise of 5 C,Fig. 1c.

3.3. Durability of MPCM

MPCMs need to withstand continuous pumping and

0

2

4

0% 5% 10% 15% 20% 25%

Mass Fraction of MPCM [%]

App

aren

t S

Fig. 1c. Temperature of freezing of tetradecane with tetradecanol asnucleating agent.10.9 mm to 25.4 mm (3/8 to 1 in.). A Moyno progressingcavity pump was used to pump the MPCM slurry throughthe system because it has shown previously to cause theleast damage to MPCM particles and can handle viscousslurries [12]. Periodic samples were taken and analyzed.First, samples were optically examined by using a labora-tory-grade microscope with magnications varying be-tween 40 and 1000 to look for any physical damage.Second, the total amount of released or free tetradecanewas determined by using a ltration device that separated

Table 1cResults from durability experiments

Batchnumber

MPCM size range(lm)

Total time of durabilitytesta (h)

% of br(%)

1 90150 9.7 15.52 70260 5 163 210 7 0b

a Total time includes cumulative results for the same batch at low and highb No signicant amount of free or released tetradecane was detected (withiTable 1c.The results depicted in Table 1c are consistent with the

results published by Yamagishi et al. [12], which also con-clude that microcapsule size is the main factor aectingdurability. The durability experimental results clearly indi-cate that small microcapsules (210 lm) show the leastdegree of damage, and therefore are suitable for furtherexperimentation. The results also show that after 1200 cir-culation times or cycles through the progressing pump (3rdbatch), no signicant damage on the microcapsules wasdetected. Also, no perceivable physical damage or deterio-ration of MPCM thermal properties was detected after theculmination of all the pressure drop and heat transferexperiments.

3.4. Viscosity measurements of MPCM slurry

Past investigations have described and reported on theimpact of spherical particles on the apparent viscosity ofslurries. Vand [14] carefully studied the interactions amongparticles in liquids, and proposed a model, which has beenregarded as the basis for the study of slurry viscosity. Tho-mas [32] also studied in detail and proposed viscositymodels for dilute and concentrated suspensions. Yamagishiet al. [12] presented viscosity data that were acquiredfrom several experiments by using a cylindrical Couette

oken microcapsules Slurry velocity(m/s)

Accumulative circulationtimes

0.62.4 7000.6 4000.62.4 1200objective was to design and test microcapsules with a fail-ure rate of less than 2%. The ltration device used to sep-arate the solid phase from the liquid phase consisted of a0.2 lm membrane lter and membrane holder. A vacuumline was connected to the discharge side of the membraneholder to speed up the ltration process.

As part of the durability test procedure, dierent exper-imental conditions were selected, including slurry velocityand level of turbulence. The slurry velocity ranged between0.3 m/s (1 ft/s) and 2.44 m/s (8 ft/s), to determine thedegree of damage, if any. Three batches of dierent MPCMmass fraction.n a 2% margin of error).

ing the hydroscopic nature of the microcapsule wall and

drop and convective heat transfer characterization

4.1. Experimental system design

A heat transfer section was designed to measure the con-vective heat transfer and pressure drop of MPCM slurrybefore, during, and after the phase change material hasundergone a solid-to-liquid transformation. To achievefully turbulent conditions at dierent MPCM slurry massfractions, the Reynolds number was calculated taking intoaccount pipe diameter, uid velocities between 0.6 and2.4 m/s, and the measured slurry viscosity.

4.1.1. Experimental heat transfer section

A 12.2 m (40 feet) long heat transfer section was con-structed and tested to be able to determine the convectiveheat transfer coecient and pressure drop ofMPCM slurry.The copper tubing length guaranteed a full phase changetransition as well as hydrodynamic and thermal entry length

isc

Heviscometer and pressure drop measurements. In that study,the apparent viscosity of MPCM slurry clearly showed aNewtonian uid behavior when a 1% anionic surfactantwas used. A later study [2] showed that the relative viscos-ity of MPCM slurry was fairly independent of temperature,suggesting a strong correlation with changes in the viscos-ity of pure water.

For this study, absolute viscosity was measured tobe able to calculate and predict the eective Reynoldsnumber and the associated turbulence range. As a result,an experimental heat transfer test loop was designed andconstructed to measure heat transfer rates. A tempera-ture-controlled concentric viscometer was used to deter-mine the apparent viscosity of MPCM slurry. The devicemeasures the viscosity of a sample by measuring the torqueexerted by the uid against a rotating cylinder. A Brook-eld viscometer, model LVT with a microliter (UL) adap-ter was used to measure the viscosity of the MPCMslurry. For temperature control, the container or UL adap-ter was connected to a water bath that circulated water at axed temperature. A solution of laboratory-grade propyl-ene glycolwater mixture at 34.6% and known viscosityvalues was prepared to calibrate the viscometer. Theviscometer was calibrated at dierent temperatures andspindle rotational velocities after the system had reacheda constant temperature for at least 30 min. It was deter-mined that the viscometer was within a 1% margin of error,which supports the viscometer manufacturer claim.

The viscosity of MPCM samples at dierent mass frac-tions was measured at 12, 30, and 60 rpm to determine ifthe slurry behaved as a Newtonian uid. Each samplewas in the UL adapter for 30 min and stirred magneticallyto make sure all the microcapsules were in suspensionbefore measuring viscosity. Three measurements weretaken for every sample. The mass fraction of each samplewas determined by taking about three 20 ll samples ofMPCM slurry from the UL adapter. Each 20 ll samplewas dried to determine its water content and mass fraction.Statistical analysis was used to determine which set of datapoints was suitable for further analysis. Fig. 1d shows thestatistically signicant viscosity results.

Fig. 1d shows that the relative viscosity of MPCM slurry(size: 210 lm) seems to be independent of temperatureregardless of mass fraction, rearming the results pre-sented by Yamagishi et al. [2]. Relative viscosity is denedas the ratio between the apparent viscosity of MPCMslurry to that of water at a given temperature. Thomas[31] analyzed several experimental results reported earlierand noted lower relative viscosity for the same mass frac-tion shown in Fig. 1d. However, Yamagishi et al. [2] alsofound that MPCM slurry has a higher relative viscositythan indicated in the data collected and analyzed by Tho-mas [32] and comparable to the data depicted in Fig. 1d,suggesting that microcapsule shape and rigidity may beplaying a role in increasing relative viscosity [2]. However,

1944 J.L. Alvarado et al. / International Journal ofit is not clear if particle rigidity or shape can explain theincreased relative viscosity values. Further studies are nec-the tendency to attract surrounding water. The MPCM vis-cosity results also indicate that MPCM slurry behaves as aNewtonian uid at least up to a mass fraction of 17.7%.

The impact of higher viscosity at lower temperature andhigher mass fraction should also be taken into consider-ation when selecting operating conditions or sizing equip-ment, because higher viscosity represents higher pumpingpower, lower turbulence, and lower overall thermal con-ductivity, which could reduce local heat transfer rates inthe absence of a phase change process. The data depictedin Fig. 1d can also be used for thermal system simulationsand to estimate level of turbulence in the slurry.

4. Microencapsulated phase change material slurry pressureessary to determine what causes increased viscosity includ-

1.0

1.5

2.0

2.5

0 2 4 6 8 10 12Temperature [C]

Rela

tive

V

Fig. 1d. Relative viscosity of MPCM slurry as function of mass fractionand temperature.3.0

3.5

4.0

osity

MPCM slurry at 4.6%MPCM slurry at 8.3%MPCM slurry at 13.4%MPCM slurry at 16.1%MPCM slurry at 17.7%

at and Mass Transfer 50 (2007) 19381952requirements. The section consisted of eight 1.52-m (5 foot)long subsections made of 10.9 mm (3/8 in.) copper tubing.

Each subsection had a total of ve thermocouples type Tsoldered to the outer surface at 30.5, 61, 76.2, 91.4 and121.9 cm (12, 24, 30, 36, and 48 in.) from either end of thesubsection. Each tubing subsection was then coated withplastic paint to minimize electrical conduction. A 24 gaugeinsulated nichrome wire was coiled around the copper tub-ing to provide constant heat ux. Each tubing subsectionhad three independent nichrome wire sections all connectedin parallel. The entire heat transfer section consisted of 24independent nichrome wire sections all connected in parallelto ensure constant heat ux. Additional nichrome wire wasadded as external resistance as needed to balance the electricload (voltage-squared/resistance or current-squared * resis-tance) so each section could provide the same overall heatux. Additional nichrome wire was connected by usingceramic terminal blocks that were not in direct contact with

used to measure total power consumption as well as cur-rent and voltage delivered by each transformer.

The last subsection of the heat transfer loop was inter-changeable, which allowed for changes in pipe diameteror tubing type. In addition to using 10.9 mm (3/8-in.) cop-per tubing, soft copper and enhanced copper tubing of8 mm was also used to determine the impact of dpipe/dparticleratio and tubing roughness (or enhancement) on the heattransfer process. The enhanced tubing section consisted ofhelical microns of 200 lm in height and width, with about60 microns around the circumference. The orientation ofthe microns with respect to the longitudinal axis was 18.

A water-ice bath was used as the cooling medium toforce the microencapsulated phase change material toundergo a liquid-to-solid transformation. Two compact

ecti

J.L. Alvarado et al. / International Journal of Heat and Mass Transfer 50 (2007) 19381952 1945the tubing sections. A thick layer of berglass insulationwas used to thermally insulate each subsection.

The subsections were connected by using PVC couplingand plastic tubing to minimize axial heat conduction. Ther-mocouples were placed between each subsection to mea-sure the uid temperature. A strip of twisted copper sheetwas inserted in the PVC coupling to ensure mixing cupor well mixed temperature conditions. All the sections werealigned horizontally by using a laser level. A Moyno Pro-gressing cavity pump was used to pump the slurry throughthe system. An Omega magnetic owmeter was used tomeasure uid velocity to an accuracy of 0.5%. The perfor-mance of the magnetic owmeter was validated by per-forming a simple bucket-watch experiment with water.Three Cole Parmer pressure dierential transducers wereused to measure pressure drop. Each pressure transducercould detect up to 17.2 kPa (2.5 psig) with 0.25% accuracyof the full range. Two alternating current variable voltagetransformers (Variacs) were used as power supplies. Eachcould provide from 0 to 208 V. For data logging, an Agi-lent Data Logger (34970A Data Acquisition/Switch Unit)with three multiplexer cards was used to record tempera-ture and pressure drop. An independent power meter was

Pump

Heat Transfer S

Flowmeter PowerSupplyFig. 2a. Schematic ofheat exchangers were used to provide enough heat transferarea to ensure full phase change. A simple schematic of theentire system is shown in Fig. 2a.

Calibration of the thermocouples was performed in situ.Based on the recorded and scanned temperature values,appropriate and acceptable oset and gain values weredetermined for each thermocouple, all within 10% of thenon-calibrated settings. Several heat transfer experimentswith plain water were conducted to validate the calibrationprocedures. The following sections explain in details thevalidation procedure and results.

4.2. Pressure drop measurement water test

Pressure drop data was analyzed and compared with theDarcy-Weisbach Eq. (2), the Colebrook correlation forfriction factor (3), and the steady-ow energy Eq. (4):

Darcy-Weisbach:

hf f LdV 2

2g; m 2

where f, L, d, V, and g are the friction factor, pipe length,pipe diameter, uid velocity, and gravitational constant,respectively.

on

Ice-water Bath

HeatExchangers

DataLoggerheat transfer loop.

smooth tubing section. Fig. 2c also reveals that enhancedand smooth tubing sections display similar pressure dropbehavior and magnitude. The plot also shows that theexperimental pressure drop for water is greater than forthe MPCM slurry at 5.9% mass fraction when enhancedtubing was tested. The results may suggest that a smallamount of microcapsules ruptured and released phasechange material into the carrier uid, creating favorableconditions for a drag-reducing eect [34]. Samples weretaken and examined to look for broken capsules or releasedtetradecane within a 2% margin of error. No signicantamount of tetradecane could be detected and micrographs

0.0

0.5

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0Fluid Velocity [m/sec]

Pr

Reynolds No. Range: 3600 - 9800Pressure error: 0.2 kPa

Fig. 2b. Pressure drop of MPCM slurry at 5.9% mass fraction, 10.9 mmregular tubing.

0.0

2.0

4.0

6.0

8.0

10.0

12.0

1.00 1.50 2.00 2.50 3.00Slurry Velocity [m/sec]

Pres

sure

Dro

p [kP

a/m]

Enhanced 8 mm tubing @ 5.9% Mass Fraction, kPa/m

Regular 8 mm tubing @ 5.9% Mass Fraction, kPa/m

Water, 8 mm regular tubing and based on Colebrook at1.3 C, kPa/mEnhanced 8 mm tubing with water only, kPa/m

Reynolds No. Range: 8000 - 11300

Pressure error: 0.2 kPa

Fig. 2c. Pressure drop of MPCM slurry at 5.9% mass fraction, 8 mmregular and 8 mm enhanced tubing.

HeColebrook equation [33]:

1f

p 1:14 2 log de

2 log 1 9:3

Re ed

fp

" #;

for Re > 3000 3where f, e/d, and Re are the friction factor, relative rough-ness, and Reynolds number, respectively.

Steady-ow energy equation (for horizontal pipes):

hf DPqg 4

where hf, DP, q, and g are friction head loss, pressure drop,uid density, gravitational constant, respectively.

Three wet-to-wet dierential pressure transducers fromCole-Parmer were used to measure pressure drop at severalvelocities and slurry concentrations. The pressure transduc-ers were calibrated using the specications provided by themanufacturer. The calibration results were validated byusing plain water. The pressure transducers generated acurrent signal which was converted to pressure units bythe data acquisition unit (Agilent unit).

The validation results indicate good agreement with thecalculated results for the same ow conditions assumingsmooth tubing surface (within 8% relative error). Theresults from the pressure drop test using enhanced tubingwere compared with simulated and calculated results basedon the Colebrook equation. Simulated results were basedon the standard roughness factor (e) for copper tubingand 0.2 mm for the enhanced tubing to take into accountthe micron enhancement size.

Pressure drop experiments [13] suggest that a roughnessfactor value for an 8 mm enhanced tubing section fallsbetween the characteristic value for smooth copper tubingand a roughness value of 0.2 mm. As per the Colebrookequations, a roughness factor of 0.1 mm seems to reason-ably match the pressure drop prole of the enhanced tub-ing [13].

4.3. MPCM slurry pressure drop test

Several pressure tests were conducted to determine pres-sure drop of MPCM slurry at two mass fractions. Pressuredrop data were collected when the slurry reached steady-state temperature and ow conditions. Fig. 2b shows theresults from a pressure drop test at low mass fraction(5.9 0.4%). Fig. 2b also shows a simulated pressure dropprole for water based on the Colebrook equation at thesame ow conditions as for the MPCM slurry. The resultsindicate that pressure drop increases slightly when MPCMparticles are used, but not signicantly enough to aectpumping power.

Fig. 2c shows pressure drop results at 5.9% (0.4%)mass fraction when 8 mm smooth, and 8 mm enhancedtubing sections were tested. Pressure drop for MPCM

1946 J.L. Alvarado et al. / International Journal ofslurry is lower than a simulated pressure drop prole forwater based on the Colebrook equation and an 8 mm1.0

1.5

2.0

2.5

3.0

3.5

4.0

essu

re D

rop

[ kPa

/m]

Water, 10.9 mm regular tubing andbased on Colebrook at 1.3 C, kPa/mRegular 10.9 mm tubing @ 5.9% MassFraction at 1.3 C, kPa/m

at and Mass Transfer 50 (2007) 19381952could not reveal the presence of broken capsules. It is plau-sible that the microencapsulation process produced a small

3.0

/m

Heamount of aggregates (not fully formed capsules) that wereprone to early breakage. Visual inspection of the entirebatch did not reveal any signicant amount of free tetrade-cane at the surface of a 10-litre container, so only a smallamount of free tetradecane might have remained in solu-tion. It is plausible that a small amount of free tetradecanecould have cancelled any drag eect associated with theenhanced surface [23]. Fossa and Tagliaco [23] observedsignicant drag reduction or friction coecient reduction(525%) when long-chain molecules in concentrations from5 to 40 ppm were tested in combination with tubing thathad helical groove enhancements.

Fig. 3a shows pressure drop results at 13.4% and 15.2%(0.4%) mass fractions for a 10.9 mm smooth tubing sec-tion. Pressure drop for MPCM slurry at 15.2% mass frac-tion is relatively higher than a simulated pressure drop

1.0

1.5

2.0

2.5

0.80 1.00 1.20 1.40Fluid Velocity [m/sec]

Pres

sure

Dro

p [kP

a

Reynolds No. Range: 6000 - 8700Pressure error: 0.2 kPa

Fig. 3a. Pressure drop of MPCM slurry at 13.4% and 15.2% massfraction, 10.9 mm smooth tubing.3.5

]Regular 10.9mm tubing @ 15.2% MassFraction, kPa/mRegular 10.9mm tubing @ 13.4% MassFraction, kPa/mWater, 10.9 mm regular tubing andbased on Colebrook, 1.3 C , kPa/m

J.L. Alvarado et al. / International Journal ofprole for water based on the Colebrook equation whena 10.9 mm tubing section was used. The results are consis-tent with the pressure drop results seen at low mass frac-tions, suggesting that the same mechanisms are also inplay at higher mass fractions.

As seen in all cases, pressure drop does not increase withparticle loading, suggesting that a drag-reducing eectcould play a major role in reducing friction and pressuredrop. As noted earlier, small microcapsules (210 lm) sup-press turbulence [35] and have a considerable impact onpressure drop too. The Reynolds number range is shownin each gure and is based on the apparent viscosity ofthe MPCM slurry at the specied mass fraction.

4.4. Convective heat transfer coecient measurement

water test

Water heat transfer tests were conducted and tempera-ture data were collected by using thermocouples and a sen-sitive data acquisition system as described above. Theamount of power for each section was measured to deter-mine each sections associated heat ux. Each sectionexperimental heat transfer coecient was calculated basedon the following relation:

h q00

T wall T bulk 5

where q00, Twall, and Tbulk are the wall heat ux, wall tem-perature, and bulk uid temperatures respectively. Theexperimental convective heat transfer coecient was ana-lyzed and compared with the Gnielinski correlation:

Nud f8

Red 1000Pr1 12:7 f

8

1=2Pr2=3 1 6where Nud, f, Red, and Pr are the Nusselt number, frictionfactor, Reynolds, and Prandtl numbers, respectively. TheGnielinski correlation is valid for 0.5 < Pr < 2000 and3,000 < Re < 5 106, and takes into account the frictionfactor. Other correlations, including the Dittus-Boelterand Sieder and Tate correlations, are valid when the Rey-nolds number is greater than 10,000.

The following equation is used to determine the convec-tive heat transfer coecient based on the Nusselt number:

h Nud kd 7

where h, Nud, k, and d are the convective heat transfer coef-cient, Nusselt number, uid thermal conductivity deter-mined at the bulk uid temperature, and pipe diameter,respectively.

The validation heat transfer experiments indicate goodagreement with the Gnielinski correlation evaluated atthe same conditions and uid properties. Uncertainty anal-ysis revealed that the compounded error was less than8.0%. Using the Gnielinski correlation for smooth tubingunder identical conditions, uid properties, and tubingcharacteristics, the enhanced tubing enhancement factoris 1.17. The enhancement factor is dened as follows:

Enhancement Factor henhancedhsmooth

8

where henhanced, and hsmooth are the heat transfer coecientsfor enhanced tubing and smooth tubing, respectively. Theexperimental heat transfer coecient for the enhanced tub-ing is based on a diameter of 8 mm (smooth). However, thedened enhancement factor does not reect the enhancedtubing actual heat transfer area. The enhanced tubing hasapproximately 1.6 times the amount of surface area ofsmooth tubing for the same diameter.

After taking into account the enhanced tubing addi-tional surface area, it can be concluded that that the micro-ns or enhancements on the tubing inner surface curtail theheat transfer process but make up the dierence by provid-ing more surface area for heat transfer. This suggests thatenhancements aect the momentum transfer from the wallto the bulk uid. Zukauskas [36] studied the impact of

at and Mass Transfer 50 (2007) 19381952 1947enhancements in detail and concluded that the height ofthe enhancement should only be high enough to disturb

the viscous sublayer in liquids. Zukauskas [36] dened adimensionless roughness height, k+, which is dened asfollows:

k kum

9

where k+, k, u*, and m are the dimensionless roughness,roughness height, friction velocity, and kinematic viscosity,respectively.

Previous experiments [35] showed that k+ should fallbetween 5 and 70 for improving the heat transfer rate;otherwise the enhancement could be ineective. In the caseof water at 1.76 m/s, k+ is about 0.25, which suggests that ahigher enhancement is advisable as long as the pressuredrop does not increase drastically.

Three distinct temperature regions are shown in Fig. 3b

including before-melting (BM), during-melting (DM), andafter-melting (AM). In the DM region, where most of theMPCM melting process takes places, the slurry tempera-ture increases slightly even with signicant constant heatux. The melting process of tetradecane helps maintain alower temperature in the carrier uid (water), which in turnallows heat transfer to occur at lower temperatures thannormal.

Fig. 3b also shows that the MPCM slurry temperatureproles display three slope changes at approximately5.5 C. Under ideal circumstances, the temperature gradi-ent for the DM region should be close to zero during thephase change process. However, the experimental datareveal a slope greater than zero because not all the micro-capsules could undergo phase change at the same time.Microcapsules closer to the tube wall undergo phasechange rst, aecting the slurry bulk temperature as they

1948 J.L. Alvarado et al. / International Journal of Heat and Mass Transfer 50 (2007) 193819520.02.04.06.08.0

10.0

0.0 2.0 4.0 6.0 8.0 10.0 12.0Distance [m]

Tem

pera

ture

[C]

8.5 kW/m2, 0.64 m/sec8.5 kW/m2, 0.83 m/sec8.3 kW/m2, 0.92 m/sec8.5 kW/m2, 1.08 m/sec14.1 kW/m2, 1.08 m/sec

4.1=

WATERWATER

MPCMMPCM

C

C

m

m

Temperature error: 0.1 C

Reynolds No. Range: 3200 5400BM DM AM4.5. MPCM slurry heat transfer test

Several heat transfer experiments were conducted todetermine the convective heat transfer coecient ofMPCM slurry at high and low mass fractions. Temperatureand power readings necessary for the calculation of theheat transfer coecient were taken when the slurry inlettemperature reached a steady state value of approximately1.8 C or less. When the inlet temperature reached 1.8 Cor less, a full liquid-to-solid transformation of the phasechange material was assumed. Simple energy balance calcu-lations based on temperature, power, and ow-meter read-ings were performed to verify that the MPCM slurrysensible and latent heat capacities match the energy inputwithin a margin of error of 10% or less.

Fig. 3b shows the experimental temperature proles(temperature vs. tubing length) for several heat transfertests of MPCM slurry at 7.0% (0.4%) when the slurryvelocity was varied between 0.64 and 1.08 m/s. Fig. 3b alsoshows that a 40% experimental heat capacity enhancementcan be obtained when assuming identical ow conditions(initial temperature and ow rate) when using plain water.Fig. 3b. Heat transfer coecient of MPCM slurry at 7.0%, 10.9 mmregular tubing.exchange energy with the surrounding uid and othermicrocapsules that have undergone only partial phasechange or none at all.

Figs. 3c and 3d show the experimental heat transfercoecient for several heat transfer tests of MPCM slurryat 7.0% (0.4%) within a 7% margin of error. Fig. 3c showsthat the convective heat transfer coecient (h) peaks atapproximately 5.0 m. By comparing Figs. 3b and 3c, it isevident that h reaches its maximum value at approximately5.5 C or near the melting point of tetradecane. Also, theadditional heat capacity available during the phase changeprocess (DM region) enhances the heat transfer coecientas shown in Figs. 3c and 3d and is in line with the notionthat at a higher heat capacity or Prandtl number, the con-vective heat transfer coecient increases accordingly [10].It is also evident that h increases signicantly withuid velocity because of greater momentum transfer athigh velocities. However, the heat transfer coecient forMPCM slurry is lower than that for pure water at the samevelocities, suggesting that MPCM particles attenuate tur-bulence and momentum transfer. It is known that particle

1.01.52.02.53.03.54.04.5

0.0 2.0 4.0 6.0 8.0 10.0 12.0Distance [m]

Hea

t Tra

nsfe

r Coe

fficie

nt

[kW/m

2 -C

]

8.5 kW/m2, 0.64 m/sec8.5 kW/m2, 0.83 m/sec8.3 kW/m2, 0.92 m/sec8.5 kW/m2, 1.00 m/sec8.5 kW/m2, 1.08 m/sec14.1 kW/m2, 1.08 m/secWater based on Gnielinski at 1.08m/sec

4.1=

WATERWATER

MPCMMPCM

C

C

m

m

Reynolds No. Range: 3200 - 5400

Heat transfer coefficient error: 7%Fig. 3c. Temperature prole for MPCM slurry at 7.0%, 10.9 mm smoothtubing.

Fig. 4c shows the heat transfer coecient of MPCMslurry at 7.0% mass fraction when smooth and enhanced8 mm tubing sections were used. The experimental datasuggest that enhanced tubing does improve the heat trans-fer process at low mass fraction, as indicated by theenhancement factor shown in Fig. 4c. However, plainwater still has a greater heat transfer coecient under

0.0

1.0

2.0

3.0

4.0

5.0

6.0

0.0 2.0 4.0 6.0 8.0 10.0 12.0Distance [m]

Hea

t Tra

nsfe

r Coe

fficie

nt [k

W/m

2 -C

]

Water based on Gnielinski at 1.5m/sec14.1 kW/m2, 1.36 m/sec14.1 kW/m2, 1.49 m/sec18.3 kW/m2, 1.50 m/sec

4.1=

WATERWATER

MPCMMPCM

C

C

m

m

Heat transfer coefficient error: 6%

Reynolds No. Range: 5800 7500

Fig. 3d. Heat transfer coecient of MPCM slurry at 7.0%, 10.9 mmregular tubing.

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Hea

t Tra

nsfe

r Coe

fficie

nt [k

W/m

2 -C

]

16.5%, 16.3 kW/m2, 1.21 m/sec

Water based on Gnielinski at1.21m/sec

7.1=

WATERWATER

MPCMMPCM

C

C

m

m

Heat transfer coefficient error: 6%

Reynolds No: ~ 4000

40.h

h_Average

GNIELINSKI_WATER

REGULAR_MPCM=

J.L. Alvarado et al. / International Journal of Heat and Mass Transfer 50 (2007) 19381952 1949size can either enhance or suppress turbulence. Experimen-tal data [35] suggest that small particles such as the onesused here (210 lm), and whose relation to the characteris-tic uid length scale (dp/Le) is less than 0.1, have shownconsiderable turbulence suppression and therefore, lowermomentum transfer.

Fig. 4a shows the MPCM slurry temperature prole at amass fraction of 16.5% in a 10.9 mm heat transfer section.Fig. 4a shows that at a higher heat ux (16.3 kW/m2) andhigher velocity (1.21 m/s), the MPCM slurry exhibitsnoticeable changes in temperature gradient. Fig. 4b showsthe heat transfer coecient for MPCM slurry under sameconditions as in Fig. 4a, which is lower than that for plainwater under the same conditions. This indicates thatMPCM slurry at higher mass fractions (higher apparentviscosity) results in lower turbulence, which curtails heattransfer from the wall to the bulk uid. On the other hand,the experimental heat capacity (70%) is considerably largerthan that of water under the same conditions. To take full0.01.02.03.04.05.06.07.08.09.0

10.0

0.0 5.0 10.0 15.0Distance [m]

Tem

pera

ture

[C]

16.5%, 16.3 kW/m2, 1.21 m/sec

7.1=

WATERWATER

MPCMMPCM

C

C

m

m

Temperature error: 0.1 C

Reynolds No: ~ 4000

Fig. 4a. Temperature prole for MPCM slurry at 16.5%, 10.9 mm smoothtubing.advantage of the MPCM slurry heat capacity, enhancedtubing that provides 60% more heat transfer area was alsotested.

0.00.00 4.00 8.00 12.00

Distance [m]Fig. 4b. Heat transfer coecient of MPCM slurry at 16.5%, 10.9 mmsmooth tubing.0.01.02.03.04.05.06.07.08.0

1.0 3.0 5.0 7.0Temperature [ C]He

at T

rans

fer C

oeffi

cient

[kW

/m2 -

C] 8 mm Regular Tubing

8 mm Enhanced TubingWater based on Gnielinski at 1.9 m/sec

Heat transfer coefficient error: 8%. Reynolds No: ~ 6900

Average hAverage h

MPCM ENHANCED

MPCM REGULAR

_

_

.

_

_

= 13

Average hAverage h

MPCM REGULAR

WATER GNIELINSKI

_

_

.

_

_

= 0 6

Average hAverage h

MPCM ENHANCED

WATER GNIELINSKI

_

_

.

_

_

= 0 7

Fig. 4c. Heat transfer coecient of MPCM slurry at 7.0%, 8 mm regularand 8 mm enhanced tubing.

identical conditions. This suggests two courses of action forfuture studies: greater tubing enhancement or thermal con-ductivity enhancement of the heat transfer uid [13]. Thelatter is currently being considered by the authors.

Fig. 4d shows the heat transfer coecient of MPCMslurry at 12.0% mass fraction when a 8 mm smooth tubingsection was used. As in the case of the 10.9 mm tubing sec-tion, the heat transfer enhancement is 0.4, which can beattributed to lower turbulence at higher mass fractionsand smaller tubing size. Fig. 5 shows that h decreases con-siderably at higher mass fractions in enhanced tubing [13].This may suggest that a larger pipe diameter and a dierenttype of enhancement could still enhance the MPCM slurryheat transfer performance at high mass fractions.

Another important goal of the heat transfer experimentswas to determine the percentage (/) of particles that under-went phase change before the bulk uid reached the melt-ing point of tetradecane. Energy balance calculationswere performed using Eq. (10) and taking into accountslurry temperature rise (DT) and heat transfer section

observed when both heat ux and slurry velocities areincreased considerably [13]. This observation can beexplained by the increased level of turbulence and greaterthermal gradient near the tube wall.

5. Conclusions

The experimental data presented in this paper show thatMPCM slurry can provide considerable heat capacity inheat transfer applications. Other important observationsand conclusions are as follows:

Thermal characterization of MPCM slurry by usingDSC reveals that supercooling of the phase changematerial can be suppressed signicantly by incorporat-ing the right amount and type of nucleating agent.

MPCM slurries exhibit a Newtonian-like behavior atmass fractions below 17.7%, and the relative viscosityis independent of temperature.

Microcapsules become durable and impact-resistantwhen smaller than 10 lm.

Pressure drop experiments revealed a possible drag-reducing eect, which should be investigated further.

Heat transfer experiments showed that the heat capacity

1950 J.L. Alvarado et al. / International Journal of Helength (L) to determine /. The left and right sides of Eq.(10) represent the heat transfer through the tube wall andMPCM slurry heat rate, respectively.

q00wpdL 1 / _mDT 1MF cp water MFcp MPCM / _mMFkMPCM 10

By also taking into account MPCM slurry mass owrate, heat ux, and mass fraction, the value of / for eachheat transfer experiment was calculated [13]. In thebefore-melting (BM) region, / represents the average per-centage of MPCM particles transferring from the core uidto the viscous layer region, where the wall layer uid is at asuciently high temperature to cause phase change of theMPCM particles entering that region. Alvarado [13]

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

4.0 4.5 5.0 5.5 6.0 6.5 7.0Temperature, C

Hea

t Tra

nsfe

r Coe

fficie

nt [k

W/m

2 -C

]

17.8 kW/m2, 1.93 m/sec

28.3 kW/m2, 2.24 m/sec

Water based on Gnielinskiat 1.93 m/secWater based on Gnielinskiat 2.24 m/sec

Average hAverage h

MPCM REGULAR

WATER GNIELINSKI

_

_

.

_

_

= 0 4

Heat transfer coefficient error: 8% Reynolds No. Range: 4600 - 5400Fig. 4d. Heat transfer coecient ofMPCM slurry at 12.0%, 8 mm regulartubing.observed that / increases with heat ux (or a greater ther-mal gradient near the surface) and ow rate (or highermomentum transfer from the wall to the core uid). How-ever, the heat transfer experiments revealed that slurryvelocity (momentum transfer) has a greater impact on /than heat ux (thermal gradient near the tube wall) [13].Also, signicant MPCM particle migration can be

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

4.0 4.5 5.0 5.5 6.0 6.5 7.0Temperature [C]

Hea

t Tra

nsfe

r Coe

fficie

nt [k

W/m

2 -C

]

28.4 kW/m2, 2.23 m/sec17.9 kW/m2, 1.94 m/secWater based on Gnielinski at 2.23 m/secWater based on Gnielinski at 1.94 m/sec

7.1=

WATERWATER

MPCMMPCM

C

C

m

mAverage hAverage h

MPCM ENHANCED

WATER GNIELINSKI

_

_

.

_

_

= 0 3

Heat transfer coefficient error: 5.5%

Reynolds No. Range: 3900 4500

Fig. 5. Heat transfer coecient of MPCM slurry at 15.2%, 8 mmenhanced tubing.

at and Mass Transfer 50 (2007) 19381952enhancements are considerable, even at low massfractions.

MPCM slurry heat transfer coecient can be increased at

Hehigher mass fraction. Also, ways to increase the thermalconductivity of heat transfer uids should be identiedand studied carefully.

Acknowledgements

The authors would like to express their deepest thanksto Professor Shelly Schmidt and Dr. Dawn Bohn of theFood Science Department at the University of Illinois;and Mr. Vince Hock, Dr. Don Cropek, Pat Kemme, JoseParrilla, Eric Crowley, Angelo DeGuzman, Bill Gordon,and Moeliere Vilceus of the US Army Corps of EngineersConstruction Engineering Research Laboratory. I am alsograteful for the technical support and expertise providedby Professor Curt Thies of Thies Technology, whoseknow-how was indispensable during the execution of theproject.

References

[1] H. Bo, M. Gustafsson, F. Setterwall, Tetradecane and hexadecanebinary mixtures as phase change materials for cool storage in districtcooling systems, Energy 24 (1999) 10151028.

[2] Y. Yamagishi, H. Takeuchi, A.T. Pyatenko, N. Kayukawa, Charac-teristics of MPCM slurry as a heat transfer uid, AIChE J. 45 (1999)696707.

[3] E. Choi, Forced convection heat transfer with water and phase-change material slurries: turbulent ow in circular tube, Ph.D. Thesis,Drexel University, Philadelphia, PA, 1993.

[4] J. Bellas, I. Chaer, S.A. Tassou, Heat transfer and pressure drop of iceslurries in plate heat exchangers, Appl. Thermal Eng. 22 (2002) 721732.

[5] E. Choi, Y.I. Cho, H.G. Lorsch, Forced convection heat transfer withphase-change-material slurries: turbulent ow in a circular tube, Int.J. Heat Mass Transfer 37 (2) (1994) 207215.

[6] K. Chen, M.M. Chen, Analytical and experimental investigation ofthe convective heat transfer of phase-change slurry ows, J. HeatTransfer (1984) 496501.

[7] E. Choi, Y.I. Cho, H.G. Lorsch, Eects of emulsier on particle size MPCM slurry heat transfer coecients are typicallylower than that of water at the same velocities. Use ofa dierent type of enhanced tubing or enhancement ofMPCM slurry thermal conductivity could improve heattransfer beyond the current levels.

The heat transfer coecient increases considerably dur-ing the phase change process.

Enhanced surface tubing is more advantageous at lowmass fraction than at a high mass fraction.

Particle migration toward the near-wall region before,during and after the bulk uid reaches the phase changematerial melting point is aected by slurry velocity moresignicantly than by heat ux.

Future studies should focus on the identication ofdurable capsule materials able to resist microbial actionand harsh environments. Greater and other types ofenhanced tubing should be considered to determine if the

J.L. Alvarado et al. / International Journal ofof a phase change material in a mixture with water, Int. Commun.Heat Mass Transfer 18 (1991) 759766.[8] J.P. Dumas, M. Krichi, M. Strub, Y. Zeraouli, Models for the heattransfer during the transformations inside an emulsion I. Crystalli-zations of the undercooled droplets, Int. J. Heat Mass Transfer 37 (5)(1994) 737746.

[9] B. He, F. Setterwall, Technical grade paran waxes as phase changematerials for cool thermal storage and cool storage systems capitalcost estimation, Energy Convers. Manage. 43 (2002) 17091723.

[10] K.E. Kasza, M.M. Chen, Improvement of the performance of solarenergy waste heat utilization systems by using phase-change slurry asan enhanced heat transfer storage uid, J. Solar Energy Eng. 107(1985) 229236.

[11] S.K. Roy, S. Sengupta, An evaluation of phase change microcapsulesfor use in enhanced heat transfer uids, Int. Commun. Heat MassTransfer 18 (1991) 495507.

[12] Y. Yamagishi, T. Sugeno, T. Ishige, H. Takeuchi, A.T. Pyatenko, Anevaluation of microencapsulated PCM for use in cold energytransportation medium, in: Proceedings of the Intersociety of EnergyConversion Engineering Conference, IEEE, 1996, pp. 20772083.

[13] J.L. Alvarado, Thermal performance of microencapsulated phasechange material slurry, Ph.D. Thesis, University of Illinois at Urbana-Champaign, Urbana, IL, 2004.

[14] V. Vand, Viscosity of solutions and suspensions, J. Phys. ColloidChem. 52 (1948) 277321.

[15] T. Ohtsubo, S. Tsuda, K. Tsuji, A study of the physical strength offenitrothion microcapsules, Polymer 32 (13) (1991) 23952399.

[16] M.S. Goel, S.K. Roy, S. Sengupta, Laminar forced convection heattransfer in microcapsulated phase change material suspensions, Int. J.Heat Mass Transfer 37 (4) (1994) 593604.

[17] S.K. Roy, B.L. Avanic, Laminar forced convection heat transfer withphase change material emulsions, Int. Commun. Heat Mass Transfer24 (5) (1997) 653662.

[18] R.A. Gore, C.T. Crowe, Turbulence in slurry pipe ow. Cavitationand multiphase ow forum, Proc. ASME Fluids Eng. Div. 98 (1990)175178.

[19] G. Hetsroni, Particles-turbulence interaction, Int. J. Multiphase Flow15 (5) (1989) 735746.

[20] X. Hu, Y. Zhang, Novel insight and numerical analysis of convectiveheat transfer enhancement with microencapsulated phase changematerial slurries: laminar ow in a circular tube with constant heatux, Int. J. Heat Mass Transfer 45 (2002) 31633172.

[21] Y. Zhang, X. Xu, X. Wang, Theoretical analysis of convective heattransfer enhancement of microencapsulated phase change materialslurries, Heat Mass Transfer 40 (2003) 5966.

[22] A. Durmus, A. Durmus, M. Esen, Investigation of heat transfer andpressure drop in a concentric heat exchanger with snail entrance,Appl. Thermal Eng. 22 (2002) 321332.

[23] M. Fossa, L.A. Tagliaco, Experimental heat transfer of drag-reducing polymer solutions in enhanced surface heat exchangers, Exp.Thermal Fluid Sci. 10 (1995) 221228.

[24] Q. Liao, M.D. Xin, Augmentation of convective heat transfer insidetubes with three-dimensional internal extended surfaces and twisted-tape inserts, Chem. Eng. J. 78 (2000) 95105.

[25] M.M. Clark, Transport Modeling for Environmental Engineers andScientists, John Wiley and Sons, New York, 1996.

[26] E.L. Charsley, S.B. Warrington, Thermal Analysis: Techniques andApplications, Thomas Graham House, Cambridge, 1992.

[27] T. Hatakeyama, F.X. Quinn, Thermal Analysis: Fundamentals andApplications to Polymer Science, John Wiley & Sons, New York,NY, 1994.

[28] R.F. Speyer, Thermal Analysis of Materials, Marcel Dekker Inc.,New York, NY, 1994.

[29] J.L. Alvarado, C. Marsh, C. Sohn, V. Hock, G. Phetteplace, T.Newell, Characterization of supercooling suppression of microencap-sulated phase change material by using DSC, J. Thermal Anal.Calorimetry 86 (2006) 505509.

[30] R. Montenegro, M. Antonietti, Y. Mastai, K. Landfester, Crystal-

at and Mass Transfer 50 (2007) 19381952 1951lization in miniemulsion droplets, J. Phys. Chem. 107 (2003) 50885094.

[31] R. Montenegro, K. Landfester, Metastable and stable morphologiesduring crystallization of alkanes in miniemulsion droplets, Langmuir19 (2003) 59966003.

[32] D.G. Thomas, Transport characteristics of suspension: VIII. A noteon the viscosity of Newtonian suspensions of uniform sphericalparticles, J. Colloid Sci. 20 (1965) 267277.

[33] American Society of Heating, Refrigeration and Air ConditioningEngineers (ASHRAE), ASHRAE Handbook Fundamentals, Chapter38, 2001.

[34] M. Kostic, On turbulent drag and heat transfer reduction phenomenaand laminar heat transfer enhancement in non-circular duct ow ofcertain non-Newtonian uids, Int. J. Heat Mass Transfer 37 (suppl. 1)(1994) 133147.

[35] C. Crowe, M. Sommerfeld, Y. Tsuji, Multiphase Flows with Dropletsand Particles, CRC Press, Boca Raton, FL, 1998.

[36] A. Zukauskas, Enhancement of forced convection heat transfer inviscous uid ows, Int. J. Heat Mass Transfer 37 (suppl.1) (1994)207212.

1952 J.L. Alvarado et al. / International Journal of Heat and Mass Transfer 50 (2007) 19381952

Thermal performance of microencapsulated phase change material slurry in turbulent flow under constant heat fluxIntroductionBackgroundPhase change materialsMicroencapsulated phase change materialEnhanced surfaces

Physical characterization of microencapsulated tetradecaneSize characterizationThermal properties characterization of MPCM slurry by using DSCDurability of MPCMViscosity measurements of MPCM slurry

Microencapsulated phase change material slurry pressure drop and convective heat transfer characterizationExperimental system designExperimental heat transfer section

Pressure drop measurement - water testMPCM slurry pressure drop testConvective heat transfer coefficient measurement - water testMPCM slurry heat transfer test

ConclusionsAcknowledgementsReferences