modelling and simulation of direct steam injection for tomato concentrate sterilization ·...

MODELLING AND SIMULATION OF DIRECT STEAM INJECTION FOR TOMATO CONCENTRATE STERILIZATION

Paolo Casoli(a), Gabriele Copelli(b)

(a) Department of Industrial Engineering, University of Parma, Viale G.P. Usberti 181/A, 43124 Parma Italy

(b) Interdepartmental Center SITEIA.Parma, University of Parma, Viale G.P. Usberti 181/A, 43124

(a)[email protected], (b)[email protected]

ABSTRACT Direct steam injection (DSI) is a sterilization technique which is often used for high viscosity fluid food when the preservation of the quality characteristics and energy efficiency are the priority.

In this work an apparatus for the sterilization of tomato concentrate has been analyzed by means of multidimensional CFD (Computational Fluid Dynamics) models, in order to optimize the quality and safety of the treated food.

A multidimensional two-phase model of steam injection inside a non-newtonian pseudoplastic fluid was adopted to evaluate the thermal history of the product and the steam consumption during the target process.

Subsequently CFD analysis has been extended to examine the effects of the different process parameters (sterilization temperature, steam flow rate, radial and axial temperature profiles, nozzle geometry) on the resulting product.

Result obtained are in agreement with available data acquired in industrial plant.

Keywords: Steam injection, non-Newtonian flow, CFD, Multiphase flow, Design optimization 1. INTRODUCTION Fluid food products in the agri-food industry are commonly subjected to thermal treatments to ensure safety and improve quality. Process parameters need to be accurately selected and monitored in order to effectively sterilize the product while at the same time avoiding over-processing, which would negatively affect product quality (Abdul Ghani et al 2001). A compromise is therefore required between safety, taste preservation and energy efficiency.

Among the different thermal treatments available, continuous direct steam injection is widely used to quickly raise the temperature of a process media, either for pure heating or for sterilization of the product. The

basic idea is to heat the liquid flow by injection of superheated steam from several nozzles, in order to reach homogeneous heating. The main benefit of the direct contact condensation process is the high heat transfer rates and the low product fouling compared with other methods such as heat exchangers. For these reasons, steam injection is used in various applications across in the food industry, such as the sterilization of milk, fruit juices and puree.

In this work, an apparatus for the sterilization of the tomato concentrate has been analyzed by means of multidimensional CFD (Computational Fluid Dynamics) models. CFD has been applied in recent years to model the sterilization problem in order to gain a better understanding of the process and optimize quality while at the same time guaranteeing the safety of the product (Debonis and Ruocco 2009; Marsh 2006).

While the overall energy balance for the process can be easily calculated from the process parameters (steam properties and product flow rate), a numerical model is required for the evaluation of temperature history and distribution Moreover, the available literature on direct steam injection is mostly limited to the case of steam injection in a stagnant liquid, typically water (Sagar 2006; Sachin 2010).

Modeling and simulation of fluid process allow: (a) the optimization of heat transfer in terms of energy efficiency, equipment design, product safety and quality retention; (b) better understanding of the heat transfer process, which helps to control the process and manage deviations, thus reducing production costs and improving quality and safety of the product (Norton 2006).

The rheological behavior of tomato concentrate is non-Newtonian. Due to the high viscosity of the product, enhanced heat transfer surface such as embossed or corrugated pipes are usually employed to ensure a good overall heat transfer coefficient; in this

Proceedings of the International Conference on Modeling and Applied Simulation, 2012978-88-97999-10-2; Affenzeller, Bruzzone, De Felice, Del Rio, Frydman, Massei, Merkuryev, Eds. 229

μ

R

case CFD seffects of pipe 2. MODEL

2.1. GeomA 3D CAD mgeometry conequally-spacesection (injec(mixing sectiowhose functieffect and thestratification.

Figu

The innelength of the e

2.2. RheolTomato concflow behavior2007):

μ=Kγn-1,

where is consistency,

Flow of generalized R

Re'=ρW2-nDn

K'8n-1

where ′ =

Note thanumber. Re’ t

An expoassumed for temperature rTrifirò (2001) = ⋅ ⋅

imulations me shape on tem

L SET-UP

metrical modemodel of the

nsisted of a hoed radially injction sectionon) presents sion is suppoe heat transfe

ure 1: Pipe and

er diameter ofexchanger is 2

logical Modelentrate has a r that follows

the dynamic the shear str

pseudoplasticReynolds numb

n

.

at Re’ is a getends to Re whonential depeK, while for

reciprocal has), as follows: ⋅ 1000⁄1000

may help to mperature dist

l exchanger w

orizontal pipe ectors placed

n). The seconseveral hemissed to impro

er coefficient

d injectors geo

f the pipe is 52.1 m.

l pseudoplastic the power-la

c viscosity, rain rate and nc fluids is charber

eneralization ohen n is close

endence on ter n a linear s been adopte

understand ttribution.

was defined: twith a series in the first h

nd half sectispherical bossove the mixiand avoid flo

ometry

51 mm; the to

c non-newtoniw model (Roz

(

the viscosn the flow inderacterized by t

(2

(3

of the Reynolto 1.

emperature wdependence

ed, according

(4

the

the of

half ion ses, ing ow

otal

ian zzi

1)

sity ex. the

2)

3)

lds

was on to

4)

wpran

rety T

padew

Fipa

arte

AwstbewEucois co

phramlorepabe

= + ⋅ 1where n0 and

roduct, while nd the flow ind

The valueeported in Taypical Hot-bre

able 1: RheolKKnn

In Figurearameters witescribes a sig

with temperatur

igure 2: Temparameter for to

Rheologicre assumed mperature and

2.3. NumeriAn Eulerian-Euwas used to des

eam entering een modeled

water vapor waulerian multipontinuous con

present in onnected.

The Eulerhase is treatedange of volum

move at differocal equilibriuesulting in aarticle model etween the tw

1000⁄ ,

nT are parak0 and n0 are

dex at the refes used for theable 1; they ak tomato con

ogical constanK0 KT n0 nT

e 2 the varh temperature

gnificant decrre.

perature variatomato concen

al and physicafrom tabulatd pressure.

ical model ulerian homoscribe the flowthe exchangeas a continuo

as modeled aphase model thnnected region

discrete reg

rian-Eulerian d separately an

me fractions (Frent velocitieum over sh

strong coupwas chosen

wo phases; th

ameters depere the viscositerence tempere rheological have been o

ncentrate.

nt of Hot-Brea0.3314 7.8814 1.114

-0.3152

riation of the is reported;rease of dyna

tion of rheologntrate

al properties oted values d

ogeneous twow of tomato coer. Tomato co

uous phase, was a dispersedhe continuousn, while the dgions which

model was cand it is applicFrank 2007). Ts, but the m

hort spatial lupling betwee

to describe the dispersed p

(5)

nding on thety consistencyrature. parameter are

obtained from

ak concentrate

he rheologica; Equation (1amic viscosity

gical

of water steamdepending on

o-phase modeoncentrate andoncentrate ha

while saturatedd phase. In thes fluid forms adispersed fluid

can be no

hosen as eachcable for wide

The phases canmodel assumelength scalesen phases. Athe interactionphase droplet

)

e y

e m

e

al ) y

m n

el d

as d e a d

ot

h e n s s, A n ts


are assumed to be spherical. All phases are assumed incompressible, accordingly with low kinetic energy; physical properties of all phases are assumed to be temperature dependent.

The flow of tomato concentrate, due to its high viscosity, is laminar; generalized Reynolds number Re’ is always lower than 20.

The phases are not generally in thermal equilibrium, due to temperature differences across phase boundaries, therefore heat is transferred across phase interfaces via interphase transfer terms (Brennen, 2005). Heat transfer across a phase boundary is described in terms of a heat transfer coefficient. Only heat transfer between phases is taken into account, walls are assumed adiabatic. Two resistance heat transfer model was used to describe combined heat and mass transfer due to steam condensation. Ranz-Marshall correlation was used to evaluate the heat transfer coefficient of the continuous phase (Pechenko 2010); Nusselt number is evaluated through the following equation: = 2 + 0.6 . . (6)

Inter-phase mass transfer has been tracked by using the thermal-phase change model. Latent heat is not directly specified, but is obtained indirectly as the difference between the enthalpies of the two phases.

The governing equations are the unsteady Navier-Stokes equations in their conservation form. For a multicomponent fluid, scalar transport equations are solved with respect to velocity, pressure, temperature and other quantities of the fluid. An additional equation must be solved to determine how the components of the fluids are transported within the fluid. Each component has its own equation for mass conservation:

+ ∇ ∙ = Γ , (7)

which is solved for each phase i; Uij is the mass-averaged velocity of fluid component i: = 1

(8)

and = ∑ , (9) where αi is the volume fraction of phase i. The term Γi in Equation (7) represent the mass source per unit volume into phase i due to interphase mass transfer.

The following general form is used to model interphase drag force acting on phase i due to phase j:

= − , (10) where cij is the drag coefficient. For the particle model (spherical particle), the drag coefficient can be obtained

in terms of dimensionless drag coefficient CD as follows: = − . (11)

Energy equation is also modified adding the

contribution SE due to steam condensation inside the continuous phase, depending on Γi in Eq. (7) = Δℎ (12)

2.4. Initial and boundary conditions

Temperature and flow rate have been set for tomato concentrate at the exchanger inlet. A constant temperature of 75°C was set for tomato concentrate at the inlet of all simulations.

During normal operation the exchanger is under a constant head of 5 bar. The absolute level of pressure at the exchanger outlet was assumed to be 5 bar. Steam pressure and temperature have been fixed at the injectors inlet accordingly to normal operation settings of the exchanger. All walls are assumed adiabatic, with imposed no slip condition.

2.5. Simulation Details An unstructured tetrahedral grid of 1.3x106 nodes was employed; meshes were created with ICEM CFD Software. Grid density has been increased near walls and at injector-pipe junctions. To improve stability and reduce the overall number of elements the hexa-core option was activated.

The commercial code Ansys CFX© was employed for all simulations. A coupled solver was used to solve governing equations. A high resolution discretization scheme was used for the continuity, energy and momentum equations, while the upwind discretization scheme was employed for the volume fraction equations. Tolerance was set for all variables at 1x10-3. 3. RESULTS AND DISCUSSION

3.1. Injector detail simulations A first set of simulations were carried out to estimate the flow characteristic of the injectors using a simplified computational domain. As shown in Figure 3, the computational domain consist of a 200 mm pipe section with a single radial injector. The junction between the main pipe and the injector is obtained with six radial slots, through which steam enters in the exchanger.

Different simulations were carried out with imposed tomato flow rates ranging from 1500 l/h to 20000 l/h; all simulations were performed with vapor mass flow from 0.002 kg/s to 0.04 kg/s. Vapor penetration inside tomato concentrate is heavily dependent on both tomato and vapor flow rate (Figures 4 and 5).


Figure

Figure 4: Dispcontours at di1500 l/h)

Figure 5: Tomsurface (volumflow rate (tom

Steam pextremely lowdistributed raover-processi

Simulaticharacteristic rate, as shocharacteristic high values o0.02 kg/s.

e 3: Simulatio

persed phase (ifferent steam

mato volume fme fraction =

mato flow rate

penetration inw for high

adial pattern oing of fluid neons allowedof the injecto

own on figof tomato fl

of steam mas

on grid around

(steam) superflow rates (to

fraction contou0.5, green) at

e 20000 l/h)

nside tomato product flow

of injector is ear the walls. d to compor depending

gure 6 influlow rate is ress flow rates

d injector

ficial velocityomato flow rat

urs and iso-t different stea

concentrate w rates. A w

critical to avo

ute the floon tomato flo

uence on floelevant only f, typically ov

y te

am

is well oid

ow ow ow for ver

Fiflo

coex

Sitotefloalefindiflostteavcoth

ofbecoaxnecasepr

anraadincopr

infloouapexusto

di

igure 6: Injectow rate) at dif

Knowledgontrol of thexchanger durin

3.2. Injectioimulations of

o investigate mperature hiow rates at thl the injector

ffects of diffnvestigated. Fifferent activeow rate of 0.0eam are remperature is

verage exchanoncentrate volhe simulations

Due to higf steam flow efore reachingontour represexial cut plane.ear the end ofan assure a gection reduceroduct exit the

In the indnd expensive ate. Steam fldjusting thenjectors. Due onstant steam roduce differe

Injection nlets was simuow rate on thutlet. Constapplied to eigxchanger; the sing a wall coo exchanger ou

Figure 9ifferent steam

tor flow charafferent produc

ge of the injectamount of

ng the steriliza

n simulationsthe complete the effects ostory of prod

he injector inles are active d

ferent numberigure 7 showinjectors wor

02 kg/s. Inlet spectively 7

s referred tonger absolutelumetric flow at 15000 liter

gh temperaturerate, the steam

g the exchangentation of tom. A complete f the injection

gradual tempees temperatue exchanger. dustrial practic

to measure alow rate is pressure in to high presspressure at t

nt steam flowwith constanulated to invee temperaturent pressure ht equally- sremaining fo

ndition. The vutlet absolute pshows the t

m relative pres

acteristic (presct flow rate

ctor characteristeam inject

ation process.

s e geometry weof steam flowduct. Constanlet were impoduring the ster of active i

ws the simulatrking with samtemperature

75°C and 1o saturated e pressure (5

rate at inlet irs per hour. e difference anm is complet

ger outlet. Figmato volume condensation

n section of terature rise anure differenc

ice it is extreand adjust thusually set

n the manifosure losses withe injectors

w rate in the exnt pressure aestigate the ee distribution

boundary cspaced injectour injectors values imposepressure.

tomato volumssure at the in

ssure vs mass

stic allows theted inside the.

ere carried ouw rate on thent steam massed. Since noerilization, theinjectors weretion results ome steam masof tomato and58°C. Steamconditions abar). Tomato

is fixed for al

nd a low valueely condensed

gure 8 shows afraction on an

n which occurthe exchangernd the mixingce when the

emely difficulhe steam flow

indirectly byld before theithin the pipeinlet normally

xchanger. at the injectoffect of steamat the produc

condition waors along thewere modeleded are referred

me fraction anjectors. In al

e e

ut e s

ot e e

of s d

m at o ll

e d a n rs r, g e

lt w y e e, y

or m ct as e d d

at ll


the simulationof the first injthe exchangersolver; due

ns the pressurjectors is lower; an automatito high pres

Fi

Fig

e imposed at ter than tomatoic wall conditssure differen

igure 7: Toma

gure 8: Tomato

the injector ino pressure insition is set by tnce on the l

ato Temperatu

o volume fract

nlet ide the last

inco

ure contours w

tion contours

njectors, steamondensed steam

with 2, 4, 8, 10

with 2, 4, 8, 1

m flow rates arm exits throug

0 active injecto

10 active injec

are extremely gh the exchan

ors

ctors

high and nonnger outlet.

n-


Figu

The matomato on thminimum temcomputed. Fieach simulatio

Figure 10: Ma(tomato) on thand minimum

ure 9: Tomato

ass flow ratehe exchanger mperature on igure 10 showon with differ

ass flow rate ahe exchanger

m values on the

o volume fract

e averaged outlet, with the same sec

ws the temperrent injection p

averaged tempoutlet. Lines se same section

Figure 1

tion contours

temperature maximum a

ction have berature values fpressure.

perature show maximun.

1: Steam supe

with pressure

of and een for

um

exinteth

hico

exavav

anFoth

anco

erficial velocit

e value set to 1

Temperatuxtremely highn degradation mperature aro

he typical set pAs shown

igher than 2 onstant at arou

Considerinxchanger, valverage outletvailable data a

A last setnalyze differenor confidentia

he injectors areThe simul

nd 12. Despitondensation is

ty contour, ax

1, 2, 2.5, 3 bar

ure difference; maximum va

of the tasteound 105°C point for steriln in Figure

bar the aveund 105°C. ng same opeues obtained t temperatureacquired in indt of simulationt pressure dial reasons thee not reportedation results e high global

s reached befo

ial plane .

r on injectors

es on outletvalues over 12e of the prodon the exchalization proces10, for injec

erage temper

eration paramwith CFD s

e are in agdustrial plant. on has been istributions one pressure di

d. . 1 are shown

l steam flow ore the exchan

inlet

sections are5°C can resulduct. Averageanger outlet isses. ction pressureature remain

meters of thesimulations o

greement with

performed ton the injectorsistributions on

in Figures 1rate, complete

nger outlet.

e lt e s

e s

e of h

o s. n

1 e


Figure 1

three differeninlets is alway

Figure 13: (tomato) on thand minimum

Compar

temperature dis more thanpressure distrtemperature d

As prtemperature vavoids degrapressure valuaverage prodcomparable tthe sterilizattemperature a 4. CONCLIn this work aconcentrate wCFD models,the treated foo

A flow geometry andANSYS CFapproach watomato concen

First simflow characteof the injecto

3 shows that tnt pressure patys lower than

Mass flow he exchanger

m values on the

red to Figurdifference on n 4 times lributions on thdifference at areviously revalues ensureadation of foues on injectduct temperato constant ption is achi

at outlet.

LUSIONS an apparatus f

was analyzed b, in order to ood. domain was d a flow mo

FX® commeas adopted antrate was cre

mulations wereeristics of steaor characterist

Figure 12: To

the temperatutterns imposed5°C.

rate averagoutlet. Lines e same section

re 10, the dthe exchange

lower. Therehe injectors ca maximum vaemarked, aes better prodood taste. Detors inlet allature on exressure simulieved with

for the sterilizby means of moptimize quali

created fromodel was builercial code. and a rheologeated. e carried out toam injectors. tic allows to p

omato concent

re difference fd on the inject

ed temperatushow maximun.

decrease in ter outlet sectifore a suitab

could reduce talue of 5°C. avoiding hiduct quality aecreasing stealowed to reaxchanger outlation; as resmore unifor

zation of tomamultidimensionity and safety

m the exchanglt by using t

A multiphagical model

o investigate tThe knowled

precisely contr

trate volume f

for tor

ure um

the ion ble the

igh and am ach tlet sult rm

ato nal of

ger the ase of

the dge trol

thdu

coefprinditefoin

prfroutodi AThsuIn RBr

M

S.

Cu

A

Sa

fraction conto

he amount ofuring sterilizat

In the secoomplete geomffects of steamroduct. Differnlets were usifferent procemperature of

ound with conlets.

Better resuressure distribom temperatuutlet section oo identify betifferences low

ACKNOWLEDhe authors wupport of thindustriali S.r.l

REFERENCErennen, E. B

flow. CamMaria Valeria D

and massflow procinjectionand Mass

. Rozzi, R. MRainieri, Afoods in ComparisocorrugatedEngineeri

urtis Marsh, direct conConferencCSIRO, M

Abdul Ghani, AP., 2001. T3-D poucJournal of

agar S. GulS.Shah, C2006. CF

ur, axial plane

f steam injection process. ond part of th

metry were carm flow rate onrent boundarysed to inves

ess parameterf the product anstant pressu

ults were reacbutions on thure differenceof the exchangtter injection

wer than 5°C w

DGEMENTSwould like tos research b.

S B., 2005. Fo

mbridge UniverDe Bonis, Gias transfer mocessing of fInternational Transfer, 37 (Massini, G. A. Trifiro 200

a shell anon between d wall tubng 79 (2007) 2Denis Wither

ntact steam ince on CFD

Melbourne, AusA.G., Farid, MThermal sterilh using comf Food Engineawani, Jyesh

Chaganti S. RaFD analysis o

e

cted inside t

he work, simuarried out to in the temperay conditions astigate the ers. High diffeat the exchangure applied to

ched by applyhe injector ies higher thanger, the simu

n settings andwere obtained.

S o acknowledg

by Ing. A. R

oundamental ersity Press. anpaolo Ruocodeling durinfluid food byl Communica(2010), 239-2Paciello, G. 06. Heat treand tube hea

n smooth aubes Journa

249–254. rs 2006. CFDnjection. Fifthin the Procestralia.

M.M., Chen, Xlization of can

mputational flueering 48 (2), htharaj B. JamaPrasad, Dof flow patt

the exchange

ulations of theinvestigate theture history oat the injectoeffects of theerences in theger outlet wereo the injecto

ying a suitablenlets: starting

n 25°C on thelation allowedd temperature

ge the activeRossi Impiant

of multiphase

cco 2009. Heang continuouydirect steamtions in Hea

244. Pagliarini, S

tment of fluidat exchangerand helicallyal of Food

D modeling oh Internationaess Industries

X.D., Richardsnned food in auid dynamics147-156.

Joshi, ManishDaya S. Shuklatern and hea

er

e e

of or e e e

or

e g e d e

e ti

e

at us m at

S. d r: y d

of al s

s, a s.

h a

at


transfer in direct contact steam condensation. Chemical Engineering Science 61 (2006) 5204 – 5220.

Sachin K. Dahikar, Mayur J. Sathe, Jyeshtharaj B. Joshi, 2010. Investigation of flow and temperature patterns in direct contact condensation using PIV, PLIF and CFD. Chemical Engineering Science 65 (2010) 4606-4620.

Pecenko A. 2010. Numerical simulation methods for phase-transitional flow. Thesis (PhD). Eindhoven University.

Frank T., 2007. Simulation of flashing and steam condensation in subcooled liquid using ANSYS CFX. 5th FZD & ANSYS MPF Workshop, April 2007.

Trifirò, A., Gherardi, S., Castaldo, D., 1991. Use of rheological parameters for pressure losses calculation in continuous heat exchangers for tomato paste. Journal of Food Engineering 23 (1991) 233.

Norton, T., Sun Da-Wen 2005 Computational Fluid Dynamics (CFD) – an effective and efficient design and analysis tool for the food industry: a review Trends in Food Science & Technology 17 (2006) 600-620.

ANSYS CFX-Solver Theory Guide. ANSYS Ltd, 2010. AUTHORS BIOGRAPHY Paolo Casoli is Associate Professor of fluid machinery for food industries at the Faculty of Engineering of the University of Parma

Gabriele Copelli is a Research Fellow of the Centro Interdipartimentale SITEIA.PARMA


modelling and simulation of direct steam injection for tomato concentrate sterilization ·...

Documents