Minimization of Thermal Impact by Application of Electrode Cooling in a Co-linear PEF Treatment Chamber

E:FoodEngineering&PhysicalProperties

Minimization of Thermal Impact byApplication of Electrode Cooling in aCo-linear PEF Treatment ChamberNicolas Meneses, Henry Jaeger, and Dietrich Knorr

Abstract: A co-linear pulsed electric field (PEF) treatment chamber was analyzed and optimized considering electricalprocess conditions, temperature, and retention of heat-sensitive compounds during a continuous PEF treatment of peachjuice. The applicability of a jacket heat-exchanger device surrounding the ground electrode was studied in order toprovide active cooling and to avoid temperature peaks within the treatment chamber thus reducing the total thermal loadto which the product is exposed. Simulation of the PEF process was performed using a finite element method prior toexperimental verification. Inactivation of polyphenoloxydase (PPO) and peroxidase (POD) as well as the degradation ofascorbic acid (AA) in peach juice was quantified and used as indirect indicators for the temperature distribution. Peaks ofproduct temperature within the treatment chamber were reduced, that is, from 98 to 75 ◦C and retention of the indicatorsPPO, POD, and AA increased by more than 10% after application of the active electrode cooling device.

Keywords: active cooling, enzyme inactivation, heat exchanger, kinetic model, numerical simulation, pulsed electricfield, thermal effects

Practical Application: The co-linear PEF treatment chamber is widely used for continuous PEF treatment of liquidproducts and also suitable for industrial scale application; however, Joule heating in combination with nonuniformelectric field distribution may lead to unwanted thermal effects. The proposed design showed potential to reduce thethermal load, to which the food is exposed, allowing the retention of heat-sensitive components. The design is applicableat laboratory or industrial scale to perform PEF trials avoiding temperature peaks, which is also the basis for obtaininginactivation kinetic models with minimized thermal impact on the kinetic variables.

IntroductionPulsed electric field (PEF) treatment of liquid foodstuff is con-

sidered as a nonthermal alternative for the inactivation of microor-ganisms. This preservation technology consists of the applicationof short electric pulses (1 to 100 μs) at high electric field intensi-ties (10 to 50 kV/cm) and at moderate temperatures to affect theintegrity of cell membranes by electroporation (Knorr and others1994; Zhang and others 1995).

The PEF application leads to an electrical current flow throughthe liquid medium which, depending on media electrical conduc-tivity, initial temperature, and treatment intensity, could result inan increase in product temperature reaching values above 70 ◦C(Gerlach and others 2008). PEF and temperature are thus stronglycorrelated, and can affect heat-sensitive food compounds simul-taneously (Jaeger and others 2010). One of the main treatmentchambers used is the co-linear one, which is used as a singlechamber or as several units connected in series. This treatmentchamber configuration has been widely used to evaluate the im-pact of PEF on heat-sensitive compounds; however, the thermaleffects which may occur during the application of PEF have been

MS 20110287 Submitted 3/5/2011, Accepted 7/16/2011. Authors are with Dept.of Food Biotechnology and Food Process Engineering, Berlin Univ. of Technology,Koenigin-Luise-St. 22, D-14195 Berlin, Germany. Direct inquiries to author Meneses(E-mail: [email protected]).

scarcely excluded from those analyses (Jaeger and others 2010).The temperature rise during the PEF treatment can be minimizedor intensified depending on the treatment chamber configura-tion. Hence, the treatment chamber design and its optimizationas well as the optimization of process conditions are of consider-able importance. The design and optimization of PEF treatmentchambers requires the consideration of an average electric fieldstrength with a low standard deviation as well as a homogeneousflow profile in order to control the temperature increase (Fiala andothers 2001; Lindgren and others 2002; Jaeger and others 2009;Meneses and others 2011). In addition to that, the applicationof an active electrode cooling provides the possibility of avoidingtemperature peaks in the treatment chamber and of reducing thetotal thermal load during the PEF treatment. The main purposeof the present investigation was the evaluation of the potentialof a treatment chamber design and optimization using an activeelectrode cooling to increase the retention of heat-sensitive com-pounds. This alternative would allow cooling down the liquidproduct during the PEF treatment within the treatment zone andcould be implemented in any kind of co-linear treatment system.

Materials and Methods

Juice preparation: enzyme activity and ascorbic acidcontent

Fresh juice was prepared by pressing peaches (variety CrimsonRocket) after grinding using a hand-lever press. The electrical

C© 2011 Institute of Food Technologists R©E536 Journal of Food Science � Vol. 76, Nr. 8, 2011 doi: 10.1111/j.1750-3841.2011.02368.x

Further reproduction without permission is prohibited

E:Fo

odEn

ginee

ring&

Phys

icalP

ropert

ies

Minimization of thermal impact. . .

conductivity (conductometer LF95, WTW, Weilheim, Germany)and the content of total soluble solids were 1.82 mS/cm and8 ◦Brix (digital refractometer RFM 80, Winopal Forschung,Hannover, Germany), respectively, at 25 ◦C. The juice wasPEF-treated and samples were analyzed.

Peroxidase (POD) activity was determined by the reduction ofhydrogen peroxide (Merck, Darmstadt, Germany), and the oxi-dation of pyrogallol (Sigma-Aldrich Chemie GmbH, 34 Munich,Germany) to pyrogallin. One hundred microliters of sample wasmixed with 320 μL of 50 mM potassium phosphate buffer, 160 μLof hydrogen peroxide (5%), 320 μL of 0.1 M pyrogallol solution,and 2100 μL of distilled water. The POD activity is proportional tothe generation of pyrogallin which was measured in a spectropho-tometer (HITACHI, San Jose, Calif., U.S.A.) at 420 nm as theaverage increase of absorbance per minute during 3 min at 20 ◦C.

Polyphenoloxydase (PPO) activity was determined by mixing100 μL of sample solution with 1900 μL of 50 mM potassiumphosphate buffer and 1000 μL of 0.1 M catechol in cuvettes witha path length of 10 mm (ratiolab GmbH, Dreieich-Buchschlag,Germany). The PPO activity correlates with the oxidation ofcatechol to benzoquinone which was measured in a spectropho-tometer at 420 nm during 1.5 min at 20 ◦C.

Ascorbic acid (AA) was determined by high-performance liq-uid chromatography (HPLC) according to Rueckemann (1980).After extraction from the juice sample with 6% methaphosphoricacid, the sample was centrifuged at 12535 g for 10 min, filtrated(22-μm pores) and analyzed by HPLC (differential-refractometer,HPLC pump 64, EuroChrom Analytical HPLC Software V 3.05,Knauer, Berlin, Germany) at 20 ◦C. Twenty microliters of theextracted sample was injected at a flow rate at 1 mL/min into aHypersil ODS column with 5-μm particle size, using as a mo-bile phase a mixture of tetrabutylammonium hydrogen sulfate,methanol, and water (2.5:945:55). The AA peaks appeared after4 min and were detected at a wavelength of 250.7 nm (Variablewavelength monitor, Knauer, Berlin, Germany).

PEF system and preheating deviceContinuous PEF treatment was performed using a 7 kW pulse

modulator (ScandiNova Systems AB, Uppsala, Sweden) providingrectangular pulses in the range of 3 to 8 μs with a maximum voltageof 50 kV and a maximum repetition rate of 400 Hz. The co-lineartreatment chamber (Berlin University of Technology) was fed witha flow rate of 4.1 L/h using a peristaltic pump 323Du (Watson

Marlow, Wilmington, N.C., U.S.A.). The adjustment of the inlettemperature (25 ◦C) was made by a stainless steel tempering coil(Berlin University of Technology) immersed in a VWR 1160Scirculating water bath (VWR, Darmstadt, Germany). The coolingcoil also served to compensate the peristaltic movement of theproduct resulting from the pump. The tempering coil consists ofa stainless steel tube of 2340 mm of total length with an innerdiameter of 2 mm and a wall thickness of 1 mm. The temperingcoil itself has a diameter of 86 mm and a height of 135 mmleading to 9 windings with a distance of 15 mm to each other(Jaeger and others 2010). A Takaoka fiber optic thermometerwith sensors of 0.8-mm diameter FT1110 (Chiyoda Corporation,Tokyo, Japan) were used for temperature measurements duringthe PEF treatment at the measurement points T1, T2, T3, and T4according to Figure 1.

Treatment chamber and implementationof jacket heat exchanger

The treatment chamber consists of one central high-voltageelectrode and two outer grounded electrodes (all stainless steel withan inner diameter of 6 mm and a wall thickness of 4 mm) separatedby a distance of 4 mm using 2 polyoxymethylene insulators withan inner diameter varying from 4 up to 6 mm. Details aboutdimensions of insulator geometry can be found in Meneses andothers (2011). The average electric field strength was estimated bymultiplying the applied voltage (V 0) [V] by a factor (g) of 1.75 ±0.33 [m] (Meneses and others 2011).

In order to cool down the liquid product after and during thePEF treatment, the co-linear treatment chamber was connectedto a jacket heat exchanger surrounding the 2nd grounded elec-trode (Figure 1). The water was circulated through the jacket heatexchanger using a VWR 1160S circulating water bath adjusted at2 ◦C (VWR) and a connected peristaltic pump 323 Du (WatsonMarlow) with a capacity of 50 L/h.

PEF treatmentThe process parameters used for PEF treatments are summarized

in Table 1.The total specific energy input (W spec in kJ/kg) was calculated

by multiplying the energy per pulse (W pulse) by the pulse frequency(f ) divided by the mass flow rate m of the treated product (Jaegerand others 2010). The pulse energy was obtained by integrationof the voltage V (t) and electrical current I(t) profiles based on

Figure 1−Co-linear PEF treatmentchamber with a jacket heat exchangerlocated around the 2nd ground electrode.Gray arrows represent the flow direction,the dashed line shows the symmetry axis,HV refers to the high-voltage electrode, Grefers to the ground electrode, and T1, T2,T3, and T4 denote the temperaturemeasurement points. Dimensions areindicated in millimeter.

Vol. 76, Nr. 8, 2011 � Journal of Food Science E537



the measurement with a TDS 220 oscilloscope (Tektronix Inc.,Beaverton, Oreg., U.S.A.). The average pulse number was calcu-lated by multiplying the total residence time (tres) of the product inthe treatment zone (0.13 s at a flow rate of 4.1 L/h) with the pulsefrequency (f ). The total treatment time (ttreat) was obtained bymultiplying the number of pulses with the pulse width (τ in μs).Based on the numerical simulation of the electric field strengthdistribution, electric field intensities able to achieve microbial orenzyme inactivation (higher than approximately 15 kV/cm) weredetected in the zone 1 mm in front and behind of each insula-tor and was therefore considered to belong to the treatment zone(Meneses and others 2011).

Thermal enzyme inactivationThermal inactivation kinetics of POD was performed using

the glass capillary method (Haas and others 1996a, 1996b). Thetemperature–time combinations were in a range of 0 to 35 s and 45to 80 ◦C. The POD kinetic was fitted to a 1st-order kinetic basedon Eq. 1, which was differentiated and included into the numericalcode. The heat inactivation kinetic was used for the evaluation ofthermal inactivation within the whole PEF treatment chamber,

RA(T,t ) = exp[−k(T) · t ] (1)

k(T) = exp (a + b · T) . (2)

k(T ) denotes a temperature-dependent factor describing the inac-tivation rate. This factor was determined experimentally based onthe known residual relative activity (RA) at a specific temperaturedepending on treatment time.

Numerical simulation of electric field strength,temperature, flow velocity, and enzyme inactivationprofiles

A numerical simulation of PEF processing was performed,which consisted of coupling a set of governing equations ableto describe properly the multiphysics phenomena: an electrostaticmodel for electric-field simulation (Eq. 3), the Navier–Stokesequations to consider fluid flow (Eq. 6 and 7), and thermal bal-ance (Eq. 4) to consider the increase of temperature due to Jouleheating effects (Eq. 5) as well as the heat exchange between liq-uid product and cooling device (Fiala and others 2001; Lindgrenand others 2002; van den Bosch and others 2003; Gerlach andothers 2008; Jaeger and others 2009; Buckow and others 2010;Meneses and others 2011). According to the aforementioned au-thors, a steady-state simulation can be assumed. Instead of simu-lating the pulsation of the electric field, the source term of theenergy equation (Eq. 4) can be multiplied by the factor f·τ . As

a consequence, the heat dissipated in the chamber in this time-independent model is equivalent to the pulsating case (Gerlachand others 2008). In addition to this set of equations, the use of atransport equation (Eq. 8) was necessary for simulating biochemi-cal reactions, such as enzyme inactivation (Hartmann and Delgado2002, 2003; Hartmann and others 2004; Rauh and others 2009).Equation 8 was included in the numerical code to describe andsimulate the thermal enzyme inactivation. The governing equa-tions are shown as follows:

E = −∇V (3)

ρCp∂T∂ t

+ ∇(−λ · ∇T + ρCp Tu) = Q (4)

Q = |σ E|2 (5)

∂ρ

∂ t+ ∇ · (ρu) = 0 (6)

ρ∂u∂ t

− ∇ · η · (∇u + (∇u)T) + ρ · (u · ∇)u + ∇ p = 0 (7)

∂c i

∂ t+ (u · ∇) c i − Di · c i = S. (8)

E is the electric field strength (V/m), V the voltage applied(V ), ∇ is the Nabla operator, Cp denotes the specific heat capacity(m2/s2/K), T is temperature (K), λ is the thermal conductivity(kg m/s3/T), ρ is the density (kg/m3), u is the velocity vector(m/s), σ is the electrical conductivity (S/m), Q is a sink or sourceterm (kg/m/s3), η is the dynamic viscosity (kg/m/s), p is the pres-sure (Pa), ci denotes species concentration, and S is a source, whichis related to chemical reactions of species ci. According to Hart-mann and others (2004) and Rauh and others (2009), the relativeactivity (RA) is directly related to the enzyme concentration (ci)and the diffusion coefficient can be neglected; hence, Eq. 8 turns to

∂ RA∂ t

+ (u · ∇) RA = S. (9)

RA is the relative enzyme activity and S is related to the enzymaticinactivation due to thermal and time effects as shown in Eq. 1 and2. In this case, the RA is not only related to temperature and time,but also related to space, resulting in the necessity of including

Table 1–Parameters used to study the impact of active cooling on temperature distribution, PPO, POD, and AA retention in PEF treatedpeach juice. All treatments were performed with a pulse width of 3 μs, a flow rate of 4.1 L/h and at an initial temperature of 25 ◦C. Thestandard deviation given for the electric field strength results from an inhomogeneous electric field distribution.

Cooling No cooling Cooling No cooling

V (kV) 18.5 21E (kV/cm) 32.38 ± 6.11 36.75 ± 6.93

f (Hz) 35 45 55 35 45 55 35 45 55 35 45 55I (A) 28.2 30.4 32.2 27.6 30 33.11 32.2 36.4 38.3 32.2 36.4 38.3W pulse (J/pulse) 2.1 2.2 2.32 2.2 2.2 2.36 2.73 2.83 2.90 2.73 2.83 2.90W spec (kJ/kg) 64.5 86.9 112.0 67.6 86.9 114.0 84 112 140 84 112 140pulse number 5 6 7 5 6 7 5 6 7 5 6 7ttreat (μs) 15 18 21 15 18 21 15 18 21 15 18 21

E538 Journal of Food Science � Vol. 76, Nr. 8, 2011

E:Fo

odEn

ginee

ring&

Phys

icalP

ropert

ies


a mass transport equation. This set of equations was discretizedon time using the finite element method commercial Softwarepackage COMSOL Multiphysics (COMSOL 2010). Boundaryconditions are shown in Table 2.

Numerical mesh, boundary conditions, and thermophysicalproperties

The numerical simulation considers the inner part of thetreatment chamber (ground electrode–1st insulator–high-voltageelectrode–2nd insulator–ground electrode) and an axis of symme-try according to Figure 1.

The geometric domain was discretized using an unstructuredmesh, which consists of triangular and quadrilateral elements. The2nd insulator and a part of the high voltage and ground electrodearea are shown as an example in Figure 2.

The double layer mesh used at the walls is necessary to computethe high rate of heat exchange expected in this zone. The num-bers noted in Figure 2 refer to boundary conditions according toTable 2. Boundary conditions for the thermal model at the wall(numbers 4, 5, and 6) were solved using an experimentally cal-culated heat transfer coefficient, which differs between differentwalls (electrode and insulator walls). For wall 4 (high-voltage elec-trode), the calculated heat transfer coefficient was approximately9 W/m2/K (Gerlach and others 2008), wall 5 (insulator) was as-sumed to be thermally insulating (α = 0) and the heat transfercoefficient for wall 6 (ground electrode connected to jacket heatexchanger) differs between experiments, which are summarizedin Table 3.

The heat transfer coefficient (α) for the jacket heat exchangerwas determined experimentally and calculated according to Kessler(2002) using Eq. (10), wherem is the mass flow rate, Cp is thespecific heat capacity, T in and Tout are the inlet and outlet tem-peratures of the 2nd ground electrode, A is the inner groundedelectrode area for the heat transfer (calculated based on the innerpipe diameter of 6 mm and electrode length of 85 mm giving aneffective area of approximately 1.6 × 10–2 m2), and T log is thelogarithmic mean temperature difference between the inside (T in

and Tout) and outside (T ext) of the ground electrode (according

Table 2 –Boundary conditions according to COMSOL (2010). Theboundary location is shown in Figure 2.

Boundary and number Value

Electrostatic modelHV electrode (4) V = Vin

Ground electrode (6) V = 0.Insulator (1,3,5) n σ ∇ V = 0Axial symmetry (2) r = 0

Thermal modelInflow (1) T = Tin

Outflow (3) n · q = 0, q = −λ · ∇TWall (4,5,6) −n · (−λ · ∇T) = α · (Text − T)Axial symmetry (2) r = 0

Flow modelInflow (1) u = u in

Outflow (3) u = 0Wall (4,5,6) n u = 0Axial symmetry (2) r = 0

Mass transfer modelInflow (1) RA = RAin

Outflow (3) RA = 0Wall (4,5,6) n · RA = 0Axial symmetry (2) RA = 0

to Eq. 11),

m · Wspec = m · CP · (Tin − Tout) = α · A · Tlog (10)

Tlog = (Tin − Text) − (Tout − Text)

ln(

Tin − Text

Tout − Text

) . (11)

T ext is the temperature of the circulating water (2 ◦C), Tout wasmeasured at the outlet of the treatment chamber (T3), and T in wascalculated from the total thermal load, which was derived fromthe total specific energy input W spec according to

Tin = T1 + Wspec

CP. (12)

Even though the heat-exchanger device consists of circulatingwater, the inlet and outlet temperature, T4 and T5, respectively(Figure 1) were identical. This was because of the high mass flowof the cooling water in comparison with the treated liquid media(approximately 10-fold higher than the treated liquid media). Thecalculated heat transfer coefficient was between 1679 W/m2/K forthe lowest treatment intensity and 2215 W/m2/K for the high-est treatment intensity (Table 4). Implementation of the knowncharacteristics for the heat transfer in COMSOL allowed the sim-ulation of the temperature during the Joule heating and coolingprocesses.

Peach juice electrical conductivity (mS/cm) was measured as afunction of the temperature, fitted to a linear equation (Eq. 13)and added within the numerical code (Buckow and others 2010;Meneses and others 2011). The density and specific heat vary onlyslightly as a function of the temperature. In contrast, the viscositydecreases significantly with the raising temperature and can alterthe flow field considerably (Gerlach and others, 2008). Since thepeach juice used in the numerical simulations contains a very lowconcentration of salts and sugar, all thermophysical properties forpeach juice were assumed to be similar to water and set up astemperature depending in the numerical code. This assumptioncan lead to deviations on the predicted temperature, mainly dueto differences in viscosity changes which are assumed to be thesame as in water. The predicted temperature was compared toexperimental results, and the root mean square error (RMSE) wascalculated as an indicator of the numerical method accuracy,

σpeach(T) = 0.0136 · T + 1.48. (13)

In order to validate the numerical simulations, temperaturewas measured at different process parameter settings and at 2

Table 3–Comparison between lowest POD residual activity (occur-ring in the treatment chamber) for PEF treatments with and withoutactive cooling as estimated by numerical simulation. The total PODresidual activity at the outlet of the treatment chamber is also shown.

Lowest POD residual Total POD residualactivity (%) activity (%)

E F W spec Cooling No cooling Cooling No cooling

32 35 65 99 88.5 99 9945 87 98 87 99 9955 112 97.5 87 99 98

36 35 84 97.5 87 99 9945 112 97 86 99 9855 140 96 81.9 99 97




different locations within the treatment chamber: 5 and 85 mmdownstream from the 2nd insulator using a fiber optic ther-mometer (points T2 and T3, respectively, in Figure 1). Thesmall diameter of the electrodes (6 mm) does not allow tem-perature measurements at different positions along the radialcoordinate inside the treatment chamber without affecting theflow and temperature distributions. Hence, an average tem-perature value was estimated from numerical simulations alongthe radial coordinate and then compared to the experimentaltemperature.

AA, PPO, and POD retention results were also used as indirectindicators of the treatment effectiveness and temperature distri-bution. Furthermore, an experimental POD inactivation kineticmodel was obtained and coupled to the multiphysics simulationof PEF processing. The simulated POD activity was integratedat the treatment chamber outlet (85 mm downstream from the2nd insulator) in order to quantify the inactivation proportiondue to heat effects. It was expected to observe higher experi-mental POD inactivation results than the simulated one, sincethe numerical code only includes the POD thermal inactivationkinetics and does not include inactivation kinetics as a functionof the electric field strength or any combination of this withtemperature.

Results

Simulation of temperature and POD thermal inactivationIn order to describe the thermal inactivation of POD in peach

juice and its dependence on treatment temperature and time,

Eq. 1 was used to fit the POD residual activity (Eq. (14)) thathas been determined experimentally by the capillary tube methodfrom a total of 4 replicates. This equation was included into thenumerical code as a partial derivative according to Eq. 15, which

Figure 3–Surface plot of POD relative activity after thermal treatmentaccording to Eq. 13. D60-value: 33 s, D75-value: 2.7 s, Z-value: 13.8 ◦C.

Figure 2−Geometric domain and numericalmesh for FEM simulation. The mesh consists of37504 triangular elements and 17280quadrilateral elements. For numbers 1 to 6 seeTable 2.

Table 4 –Comparison between simulated and experimental temperature for treatment chamber with and without jacket heat exchanger.PEF process parameters applied are listed in Table 1. Temperature measurement points at 0.5 and 8.5 cm downstream from the 2nd insulator(T2 and T3) are given in Figure 1. Standard deviation for experimental temperature (exp) is given by the fiber optic accuracy, which is0.5 ◦C.

Cooling No cooling

T2 T3 T2 T3E f W spec α exp sim exp sim exp sim exp sim

32 35 65 1679 27.1 26.2 ± 3.6 23.3 22.9 ± 4.1 41.9 42.3 ± 1.1 38.3 37.1 ± 0.545 87 1884 40 39.5 ± 2.6 24.9 25.6 ± 3.9 46.7 47.2 ± 1.3 43 40.1 ± 0.855 112 1902 43.9 42.9 ± 3.2 27.5 28.4 ± 4.0 52.2 53.2 ± 1.4 43.6 41.9 ± 0.8

36 35 84 1964 41.2 39.4 ± 2.6 24.6 25.5 ± 3.9 47.1 48.2 ± 1.2 40 38.9 ± 0.745 112 1914 44.4 43.9 ± 3.4 27.5 28.9 ± 3.9 53.5 54.9 ± 1.4 44.1 42.8 ± 0.855 140 2215 51.1 48.3 ± 4.2 28 30.8 ± 4.5 59.3 60.9 ± 1.8 48.3 46.1 ± 0.9

RMSE 3.37% 5.17% 2.05% 4.23%


E:Fo

odEn

ginee

ring&

Phys

icalP

ropert

ies


was then coupled as source term (S) in Eq. 9,

RAT (T, t ) = exp[− (exp[−12.84 + 0.169 · T]) · t ] (R2= 0.981)(14)

S = dRAT

dt= − exp [−12.84 + 0.169 · T] · RAT . (15)

Figure 3 shows the POD residual activity at different temper-ature and time combinations in a range of 0 to 35 s and 45 to80 ◦C. As observed, the thermal instability of peach juice POD ishigh and 1-log inactivation can be reached in 2.7 s at 75 ◦C.

During the PEF treatments, peach juice samples are taken85 cm downstream from the 2nd insulator and immediately cooleddown in ice water. At a flow rate of 4.1 L/h and an electrodediameter of 6 mm, the average flow velocity is approximately0.04 m/s, which leads to a residence time of 2.11 s within the2nd ground electrode section.

In order to determine the temperature occurring in this sec-tion, a simulation of the temperature profile was performed forthe 2 different cases with and without the application of the ac-tive electrode cooling (Figure 4). In addition, the integration ofEq. 15 along the electrode allowed us to couple the occurringtemperature profile with the heat inactivation data of POD result-ing in a thermal inactivation profile of POD within the section ofthe ground electrode (also shown in Figure 4).

Numerical results of POD inactivation profiles within the treat-ment chamber are shown in Table 3.

As shown in Figure 4, the temperature downstream of the in-sulator zone can reach high values when no active cooling is used.The temperature hot spots were reduced from 98 to 75 ◦C (exper-imental measurements were performed at 0.5 and 8.5 cm down-stream of the 2nd insulator, indicating a difference below 5.17%between measurement and simulation). At higher PEF intensities,it would be expected to observe boiling temperatures, which canbe avoided by controlling the pressure. The right column of Fig-ure 4 also shows a comparison of the simulated POD inactivationwithout and with active electrode cooling. POD retention was in-tegrated at the outlet of the treatment chamber. As a result, averagePOD residual activity integrated at the treatment chamber outlet

increased up to 2% (see also Table 4) and POD retention in zonesclose to the insulator increased up to 14.1% due to the tempera-ture decrease and the avoidance of temperature peaks. The highestPOD thermal inactivation was always found immediately down-stream the 2nd insulator where temperature peaks occur. Thismodel does not take into account the POD inactivation due toPEF effects. However, it is applicable for design optimization pur-poses, especially when the retention of heat-sensitive compoundsand reduction of thermal effects is desired.

Experimental validation by temperature measurementA comparison between the measured temperatures behind the

2nd insulator (T2) and after the active cooling section (T3) is givenin Table 4.

For each of the 2 simulation points, this value is an average alongthe whole radial coordinate (the standard deviation is also givenfor each value). The given error is the relative error between thesimulated and the experimental temperature.

A RMSE between 2.05% and 5.17% was calculated showing anadequate prediction of the temperature by using the experimen-tally determined heat transfer coefficients.

Impact of electrode cooling on PPO, POD, and AA retentionThe implementation of a jacket heat exchanger contributes

to a considerable reduction of the thermal load to which thefood is exposed during PEF treatment. Consequently, this phe-nomenon affects heat-sensitive compounds and their retention isimproved. Peach juice POD, POD, and AA were identified asthermal-sensitive compounds. The impact of implementing theactive electrode cooling on these compounds is shown in Figure 5.

Differences between simulated and experimental POD inacti-vation (Table 3 and Figure 5) are higher at 45 and 55 Hz. Sincethe temperature is higher in the treatment zone at these treat-ment intensities, a synergetic effect between temperature and PEFinactivation of enzymes can be assumed as reported in the litera-ture earlier (Riener and others 2008; Schilling and others 2008).In addition, the numerical code does not include POD inacti-vation kinetics due to PEF effects. It is shown that enzyme andAA inactivation and retention is more affected by temperature ef-fects than by the PEF treatment itself. Even at a moderate outlet

Figure 4–Top row: treatment chamber with jacket heat exchanger. Bottom row: treatment chamber without jacket heat exchanger. Left column:temperature profile comparison. Right column: POD inactivation profile. PEF treatment performed at 36.75 kV/cm 140 kJ/kg, 55 Hz, 31 s pulse width.




Figure 5–PPO, POD, and AA retention after PEFtreatment with and without implementation ofthe jacket heat exchanger (electrode coolingand no cooling, respectively). Indicated also thegain of relative retention (percentage inside thelight bars) when implementing the electrodecooling device. PEF process inlet temperaturewas 25 ◦C, mass flow 4 L/h, electric fieldstrength 32 and 36 kV/cm, and frequency 35,45, and 55 Hz. In all cases standard deviationwas lower than 5%.

temperature of 55 ◦C, peaks of temperature above 90 ◦C can occurin the treatment chamber (Figure 4), which can contribute to theinactivation of enzymes and a reduction of AA content. Further-more, the low flow velocity in zones close to the walls resultsin higher residence times, which further promotes the thermalinactivation effect in combination with temperature peaks.

ConclusionsThe occurrence of a temperature increase during PEF preser-

vation of liquids is unavoidable due to ohmic heating phenomena.The suggested active electrode cooling was shown to provide acapable tool in order to reduce thermal effects and to improve theretention of heat-sensitive compounds.

The heat inactivation characteristics of POD were success-fully implemented into numerical simulations of a PEF treatment,which allows the process- and product-specific design and opti-mization of treatment chambers. Based on experimental and sim-ulated data, temperature measured at the outlet of the treatmentchamber (8.5 cm downstream from the treatment zone) can beup to 11 ◦C lower than that achieved temperature immediatelydownstream from the 2nd treatment zone (0.5 cm downstreamfrom the treatment zone) and PPO, POD, and AA retention in-creased more than 10% after the PEF treatment. These data suggestthat the temperature should be measured as close as possible fromthe treatment zone. Even a distance from 8 cm would lead towrong temperature interpretations and also to lower retention ofheat-sensitive compounds, which is not necessarily related to thePEF impact itself.

The presented investigation contributes to the fulfillment ofarising requirements for a successful industrial implementation ofthe PEF technology such as avoidance of temperature peaks, re-tention of heat-sensitive compounds, and the selective inactivationor retention of enzyme activity in liquid food systems; however,further studies are necessary to analyze the implementation ofjacket heat exchangers within different treatment chamber con-figurations. It was also shown that enzymes can be used as inte-grators of temperature-time domains under dynamic conditions,as coupled to the multiphysics simulation of PEF processing. Re-ducing thermal effects and minimizing the thermal load to whicha product is exposed during PEF treatment also provides the ba-sis to study PEF inactivation kinetics in continuous systems thatare less affected by superimposing temperature effects or to in-vestigate synergetic effects of PEF and temperature under definedconditions.

List of symbolsGreek variables

∇ nabla operator (gradient operator)∂∂ t partial derivativeα heat transfer coefficient (W/m−2/K)η dynamic viscosity (kg/m/s)λ thermal conductivity (kg/m/s3/T)ρ density (kg/m3)σ electric conductivity (S/m)τ pulse width (μs)


E:Fo

odEn

ginee

ring&

Phys

icalP

ropert

ies


Latin variablesA inner grounded electrode area (m)a kinetic constant (–)b kinetic constant (◦C1)c i substance or specie iCp specific heat capacity (m2/s2/K)Di diffusion coefficient of specie i (m2/s)_E electric field strength (V/m)f pulse frequency or repetition rate (Hz)g conversion factor (m)I electrical current (A)k model kinetic constant (s)m mass flow rate (kg/s)n normal vectorp local pressure or pressure (Pa)Q sink or source term (kg/m/s3)q heat flux (W/m2)r radius (m)RA relative activityRAin initial relative activityS source term of the inactivationt time (s)tres residence time (s)ttreat treatment time (μs)T temperature (◦C)Text external temperature (◦C)Tin inlet temperature (◦C)Tout outlet temperature (◦C)Tlog logarithmic mean temperature difference (◦C)u velocity or velocity vector (m/s)uin initial velocity (m/s)V0 initial voltage (V)V voltage (V)Wpulse energy per pulse (J/pulse)Wspec energy input (J/kg)

AbbreviationsAA ascorbic acidexp experimentalG ground electrodeHPLC high-performance liquid chromatographyHV high-voltage electrodePEF pulsed electric fieldPOD polyphenol oxidasePPO polyphenol peroxidaseRMSE root mean square errorsim simulated

AcknowledgmentsThis research project was supported by the Commission of the

European Communities, Framework 6, Priority 5 ‘Food Qualityand Safety’, Integrated Project NovelQ FP6-CT-2006-015710.

ReferencesBuckow R, Schroeder S, Berres P, Baumann P, Knoerzer K. 2010. Simulation and evaluation of

pilot-scale pulsed electric field (PEF) processing. J Food Eng 101:67–77.COMSOL 2010. COMSOL user guide. Burlington, Vt.: COMSOL Inc.Fiala A, Wouters PC, van den Bosch E, Creyghton YLM. 2001. Coupled electrical-fluid model

of pulsed electric field treatment in a model food system. Inno Food Sci Emerg Technol2:229–38.

Gerlach D, Alleborn N, Baars A, Delgado A, Moritz J, Knorr D. 2008. Numerical simulationsof pulsed electric fields for food preservation: a review. Inno Food Science Emerg Technol9(4): 408–17.

Haas J, Behsnilian D Schubert H. 1996a. Determination of the heat resistance of bac-terial spores by the capillary tube method: I. Calculation of two borderline cases de-scribing quasi-isothermal conditions. Lebensmittel-Wissenschaft und- Technologie 29:197–202.

Haas J, Behsnilian D. Schubert H. 1996b. Determination of the heat resistance of bacterialspores by the capillary tube method: II. Parameters of Bacillus stearothermophilus spores.Lebensmittel-Wissenschaft und-Technologie 29:299–303.

Hartmann C, Delgado A. 2002. Numerical simulation of thermofluiddynamics and enzymeinactivation in a fluid food system under high hydrostatic pressure. In R. Hayashi, editor.Trends in high pressure bioscience and biotechnology, Amsterdam: Elsevier. p 533–40.

Hartmann C, Delgado A. 2003. The influence of transport phenomena during high-pressureprocessing of packed food on the uniformity of enzyme inactivation. Biotechnol Bioeng82(6):725–35.

Hartmann C, Schuhholz J-P, Kitsubun P, Chapleau N, Le Bail A, Delgado A. 2004. Experi-mental and numerical analysis of the thermofluiddynamics in a high-pressure autoclave. InnoFood Sci Emerg Technol 5:399–411.

Jaeger H, Meneses N, Knorr D. 2009. Impact of PEF treatment inhomogeneity such as electricfield distribution, flow characteristics and temperature effects on the inactivation of E. coli andmilk alkaline phosphatase. Inno Food Sci Emerg Technol 10:470–80.

Jaeger H. Meneses N, Knorr D. 2010. Model for the differentiation of temperature and electricfield effects during thermal assisted PEF processing. J Food Eng 100:109–18.

Kessler HG. 2002. Food and bio process engineering—dairy technology. Munich, Germany:Verlag A. Kessler.

Knorr D, Geulen W, Grahl T, Sitzmann W. 1994. Food application of high electric field pulses.Trends Food Sci Technol. 5:71–5.

Lindgren M, Aronsson K, Galt S, Ohlsson T. 2002. Simulation of the temperature increasein pulsed electric field (PEF) continuous flow treatment chambers. Inno Food Sci EmergTechnol 3:233–45.

Meneses N, Jaeger H, Moritz J, Knorr D. 2011. Impact of insulator shape, flow rate and electricalparameters on inactivation of E. coli using a continuous co-linear PEF system. Inno Food SciEmerg Technol 12:6–12.

Rauh C, Baars A Delgado A. 2009. Uniformity of enzyme inactivation in a short-time high-pressure process. J Food Eng 91:154–63.

Riener J, Noci F, Cronin DA, Morgan DJ Lyng JG. 2008. Combined effect of temperatureand pulsed electric fields on apple juice peroxidase and polyphenoloxidase inactivation. FoodChem 109(2):402–7.

Rueckemann H. 1980. Methoden zur Bestimmung von L-Ascorbinsaure mittels Hochleistungs-Flussigchromatographie (HPLC). Zeitschrift fur Lebensmittel-Untersuchung und –Forschung171, 357–9.

Schilling S, Schmid S, Jaeger H, Ludwig M, Dietrich H, Toepfl S, Knorr D, Neidhart S, SchieberA Carle R. 2008. Comparative study of pulsed electric field and thermal processing of applejuice with particular consideration of juice quality and enzyme deactivation. J Agric FoodChem 56:4545–54.

van den Bosch H, Morshuis P, Smit J. 2003. Temperature distribution in continuous fluid flowstreated by PEF. XIII Internat. Symp. on High Voltage (ISH), Delft, the Netherlands.

Zhang Q, Barbosa-Canovas G, Swanson B. 1995. Engineering aspects of pulsed electric fieldpasteurization. J Food Eng 25:261–81.


Documents

Minimization of Thermal Impact by Application of Electrode Cooling in a Co-linear PEF Treatment Chamber