Journal of Food Engineering 79 (2007) 445–452 www.elsevier.com/locate/jfoodeng

Investigation of far infrared radiation heating as an alternative technique for surface decontamination of strawberry F. Tanaka a, P. Verboven b

b,*

, N. Scheerlinck b, K. Morita a, K. Iwasaki a, B. Nicolaı¨

b

a Environmental Science & Technology Department, Kagoshima University, Korimoto 1-21-24, Kagoshima 890-0065, Japan Biosystems Department, Laboratory/Centre of Postharvest Technology, Katholieke Universiteit Leuven, de Croylaan 42, Leuven B-3001, Belgium

Received 11 June 2005; accepted 4 February 2006 Available online 29 March 2006

Abstract Postharvest heat treatments have recently received attention as a means to prevent fungal spoilage of strawberry during shelf life. Far infrared radiation (FIR) heating technology may be an alternative to conventional methods because it can achieve rapid and contactless heating. Monte Carlo FIR radiation simulations combined with convection–diffusion air flow and heat transfer simulations were performed in the CFD code ANSYS CFX5.7 to investigate the suitability of the method for surface decontamination. Computations were validated against measurements with a thermographic camera. FIR heating achieved more uniform surface heating than air convection heating, with a maximum temperature well below the critical limit of about 50 C at the same average temperature. In this configuration, the resulting surface FIR heating rate was, however, smaller or only equal to the air convection heating (at 0.2 m s1), depending on the heater temperature used. A better configuration consisted of FIR heaters on four sides combined with a cyclic heating operation.  2006 Elsevier Ltd. All rights reserved. Keywords: Radiation; Convection; CFD; Monte Carlo; Thermography

1. Introduction Thermal processing of foods is very important to extend shelf life. In particular, postharvest heat treatments have become increasingly popular to control insect pests, prevent fungal spoilage and affect the ripening of fruits and vegetables (Marquenie et al., 2003). Far infrared radiation (FIR) heating technology is useful for these purposes because it can achieve contactless heating. Compared to water submersion heating, the FIR heating rates were small and the FIR surface temperature uniformity was inferior. However, there is an advantage of using no water with large heat capacity and avoiding discharge wastewater. Furthermore, other organisms may grow in water at temperatures below 50 C. Today, the use and understanding of FIR heating in food applications is limited. Datta and Ni (2002) provided *

Corresponding author. Tel.: +32 16 32 14 53; fax: +32 16 32 29 55. E-mail address: [email protected] (P. Verboven).

0260-8774/$ - see front matter  2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jfoodeng.2006.02.010

an overview of advances made in this area, concerning the effects on internal heat and mass transfer. Infrared power absorption by foods has been treated by two formulations—one with zero penetration and one with finite penetration depth. For heat transfer only, Sakai, Fujii, and Hanzawa (1993) found no significant differences between the two methods. This approach was also applied by Parroufe, Dostie, Mujumder, and Poulin (1992), Fasina, Tyler, and Pickard (1997), Shilton, Mallikarjunan, and Sheridan (2002) for heat and mass transfer analysis in food dehydration applications. Ginzburg (1969) and Datta and Ni (2002) applied an exponential decay model with a finite penetration depth. The latter authors showed that the resulting internal infrared heat dissipation had considerable effects on surface moisture content of the food in combined FIR and microwave heating. To our knowledge, studies on FIR heating have only been applied to foods with a simple 1D or 2D geometry. The external infrared radiation transfer from the FIR heater to the food has not been studied in depth up to date.

446

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

This process determines the incident radiative heat flux at the surface of the food. Studies on radiation heating of foods have used either a non-linear radiation boundary condition formulated based on the Stefan–Boltzman law for pairs of diffuse grey surfaces (Shilton et al., 2002), or simply specified the value of the incident radiative heat flux (Datta & Ni, 2002). In complex geometrical configurations, these approaches are not trivial because of the difficulty of determining the view factors of the contributing surfaces (Incropera & De Witt, 1990). Monte Carlo modelling of radiation heat transfer is appropriate to compute radiation heat transfer in complex configurations, and has been widely described (Guilbert, 1989; Howell, 1988, 1998; Maltby & Burns, 1991). In Monte Carlo simulations of radiation heat transfer problems, the energy emitted from a surface is simulated by the propagation of a large number of photons, which are massless units of energy. The photon is followed as it proceeds from one interaction to another, which are described as random events. This continues until the photon is absorbed or leaves the computational domain. A large number of trajectories is required to ensure that the variation due to the random events are small. The results are used to determine the fraction of energy that has been absorbed on each surface in the geometry. To our knowledge, Monte Carlo simulation of radiation heat transfer has not been applied to food applications. The objective of this study is to develop and validate a Monte Carlo radiation heat transfer model of FIR heating of strawberry. The model considers both internal heat transfer (conduction) and external heat transfer (convection and surface radiation). The model is solved in 3D incorporating the real strawberry shape and returns the transient temperature profile of the strawberry. The surface inactivation of Monilia fructigena is calculated in FIR cyclic heat treatment. The use of the FIR method for surface decontamination purposes is discussed. 2. Experimental procedures 2.1. Material The strawberry used for this series of experiments was obtained from a retail store and stored at optimal conditions (1 C, 95%RH) for the duration of the experimental trials (10 days). The average density qs of the strawberry was 900 kg m3, the heat capacity cps and the thermal conductivity ks were 3953 J kg1 C1 and 0.55 W m1 C1, respectively (Scheerlinck et al., 2004). 2.2. Method 2.2.1. Computer imaging of strawberry The image acquisition system included a 3 CCD camera (KY-F55B, JVC, Japan), a circular lamp (100 W Philips Softone, The Netherlands), and a sample holder mounted on a rotation table with a computer controller stepper

motor (Matsushiba, Japan Servo Co. Ltd., Japan). Nine strawberries were placed on the rotation table and eight images were taken along the fruit equator for each fruit. For imaging acquisition, a PCI based Matrox Comet capture card (Matrox Inc., Canada) with RGB daughter card was used. A geometrical model was created on a PC by means of the procedure developed by Jancso´k, Nicolaı¨, Coucke, and De Baerdemaeker (1997). 2.3. IR heating measurement system Fig. 1 shows a schematic representation of the experimental apparatus. This system includes an airflow cell with flow distributors and a blower, two infrared radiation heaters (frequency range = 1.5–50 lm, emissivity = 0.95, heating surface area = 0.2 m · 0.3 m, PLR-620, Noritake Co., Japan) with a controller (PU-2060A, Noritake Co., Japan), an IR thermographic camera (ThermaCAM 3000SC, FLIR systems Inc., USA) for measuring the surface temperatures distribution of the fruit, and sample holder. Thermographic images represent the spatial temperature distribution of surfaces, based on measured infrared radiation. The infrared thermographic camera (ThermaCAM 3000SC, FLIR systems Inc., USA) was located outside the experimental cell (1.5 m · 1 m · 1 m). Calibration of the IR thermographic camera was carried out in the surface temperatures range from 20 to 50 C by using a calibrated miniature T-type thermocouple (±0.5 C) inserted just under the surface of the strawberry fruit. By comparing the observed temperatures that were collected by the IR camera and thermocouple, the following relationship was derived; ttc = 1.26tir  11.50. The correlation coefficient between ttc and tir was 0.998. As a result, the surface temperature profile of the strawberry could be determined accurately by using the IR thermographic camera. The inside walls of the airflow cell were painted black. The IR heating cell (0.2 m · 0.2 m · 0.3 m) was set in the airflow cell. Distributors are used to regulate flow from left to right in Fig. 1. Inlet air velocity was kept at 0.2 m/s and the air temperature was close to 20 C. The IR heaters were positioned parallel at a distance of 0.2 m. The IR heating experiments were carried out at three different heater tem-

IR thermo camera

Inlet

Distributors IR heaters Sample

Blower Outlet

Thermocouples Fig. 1. Schematic representation of the experimental apparatus.

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

peratures, 150, 200 and 250 C. During the experiments, the sample was positioned exactly in between the two heaters on the bottom side of the heating cell by using a sample holder. We also investigated positioning the sample at the center of heating cell.

The random sampling of the source location on a primitive surface with local curvi-linear coordinates (r1 6 r 6 r2, s1 6 s 6 s2, t1 6 t 6 t2) consisted of three random choices of the pseudo-random variable 0 6 n1 6 1, 0 6 n2 6 1 and 0 6 n3 6 1. r ¼ r1 þ n1 ðr2  r1 Þ

3. Model

S ¼ S 1 þ n2 ðS 2  S 1 Þ t ¼ t1 þ n3 ðt2  t1 Þ

3.1. Geometry The model was solved for the geometry of the IR heating cell, including the real shape of the strawberry obtained from the 3D imaging procedure (see Fig. 2). 3.2. Radiation model The surfaces in the geometry of the heating cell and the strawberry were subdivided into sets of non-overlapping primitive surfaces. These surfaces were boundary faces of volumetric zones that cover the interior computation domain. Each primitive heating cell surface j is opaque (no transmission of radiation) and treated as a separate uniform photon source with a temperature Tj and emissivity ej. The wall surfaces and strawberry surface were given an emissivity value ej = 0.95 (no penetration of radiation into the strawberry, i.e. all absorbed radiation power is dissipated at the surface). The total number of primitive surfaces is indicated by N. The total number of histories Np to be calculated was then divided up amongst the sources, according to their emission S j ej rT 4j compared to the total emission of all surfaces. Thus the number of photons nj emitted by surface j was S j ej rT 4j nj ¼ N p ¼ PN 4 x¼1 S x ex rT x

447

ð1Þ

ð2Þ

The birth of photon required not only a location but also the initial direction of travel. We generated direction cosines from pffiffiffiffiffiffiffiffiffiffiffiffiffi r0 ¼ 1  t02 cos u ð3Þ pffiffiffiffiffiffiffiffiffiffiffiffiffi s0 ¼ 1  t02 sin u where pffiffiffiffiffi t 0 ¼ n4 u ¼ pð2n5  1Þ

ð4Þ

where n4 and n5 are two choices of the pseudo-random variable 0 6 n4 6 1, 0 6 n5 6 1. The initial direction of travel can be expressed as the following unit vector: m ¼ ðr0 ; s0 ; t0 Þ

ð5Þ

Tracking photons across the geometry was the most computationally intensive task. A photon is absorbed/ reflected at a physical surface in case of the following: n6 < e ! absorption n6 P e ! reflection

ð6Þ

where n6 is the pseudo-random variable 0 6 n6 6 1. In the case of reflection on a diffuse surface, a new direction of travel was sampled according to the Eqs. (2) and (3). Finally, the amount of photons absorbed on a surface was obtained. The resulting irradiation (W m2) on surface i was Gi ¼

N nij 1 X S j ej rT 4j S i j¼1 nj

ð7Þ

with nij the number of photons that were emitted by surface j and incident on surface i. Introducing Eq. (1) into Eq. (7) leads to PN i N N X 1 1 ni X j¼1 nj Gi ¼ S x ex rT 4x ¼ S x ex rT 4x ð8Þ S i N P x¼1 S i N P x¼1 It therefore suffices to add all incident photons on surface i to the photon current ni and multiply the relative photon current with the available emissive power in the system to obtain the irradiation. The net radiation flux qi leaving the surface i then was equal to Fig. 2. Model geometry and air velocity vectors in the radiation heating cell.

qrad;i ¼ ei rT 4i  ei Gi

ð9Þ

448

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

In order to obtain good estimates of this quantity of interest, many histories of photons needed to be generated. In this work, 100,000 tracks were calculated. 3.3. Airflow model In order to solve the laminar incompressible flow in the IR heating cell, the following equations were applied: ru¼0

ð10Þ T

qr  ðu  uÞ ¼ rp þ r  ðlðru þ ðruÞ ÞÞ qcp

oT þ qcp r  ðuT Þ ¼ r  ðkrT Þ ot

ð11Þ ð12Þ

with u [m s1] the air velocity vector, p [Pa] the pressure, T [C] the temperature and t [s] time. These governing equations were solved with appropriate boundary and initial conditions. At the inlet of the heating cell, the air velocity was specified equal to a uniform measured value of 0.2 m s1. Lateral open air boundaries were assigned free slip conditions. IR heater surfaces were specified no-slip walls and the exit was a zero pressure boundary. The heater surface temperature was set to the control value, and the inlet temperature equal to the measured entering air temperature. At the strawberry surface a conservative interface heat flux condition was applied (convection + radiation = conduction). Constant air properties were chosen as ideal gas with the molecular weight specified, M = 28.96 kg kmol1; l = 1.79 · 105 kg m1 s1, cp = 1000 J kg1 C1, k = 0.0252 W m1 C1. 3.4. Conductive heat transfer model inside the strawberry It was assumed that conductive heat transfer occurs inside the strawberry, and that convective and radiation heat transfer take place at the surface (see above). The Fourier equation was used to describe the conductive heat transport through the strawberry. qs cps

oT ¼ k s r2 T ot

ð13Þ

with the properties specified earlier. The thermal properties of the fruit were kept constant in the temperature range used for this simulation and a uniform temperature distribution was chosen as initial condition for Eq. (13). The boundary condition was oT ðx; y; zÞ ¼ hðx; y; zÞðT a ðx; y; zÞ  T ðx; y; zÞÞ  qrad ðx; y; zÞ on ðx; y; zÞ 2 Cs

ks

ð14Þ where n is the outward normal to the food surface Cs, h [W m2 C1] the local surface heat transfer coefficient which results from the air flow model and qrad [W m2] the radiation flux that results from the radiation model.

3.5. Computational aspects All equations were implemented in the unstructured finite volume code ANSYS CFX5.7, which has been successfully used in the past for food heating applications (Verboven, Scheerlinck, De Baerdemaeker, & Nicolaı¨, 2000). The number of control elements used in the reported results was 91,289, including 15,000 prismatic elements in the layered mesh near the surfaces to resolve the boundary layer. A mesh refinement study showed that the use of 226,598 elements (122,000 prisms) only resulted in a 0.1 C change in average strawberry surface and body temperature, and 0.9 C change in maximum surface temperature, after 800 s of IR heating at 150 C. Similarly, changing the number of photon tracks from 50,000 to 100,000 and 200,000 revealed temperature differences smaller than 0.5 C. 100,000 photon tracks were used. Second order implicit time stepping was used with step sizes of 5 s. The simulation of 800 s of heating required 19 h of computation time on a Pentium III 933 MHz 2 GB RAM PC. Thirty iterations were performed each time step to achieve normalised residuals in the range of single precision machine accuracy. Radiation–convection–conduction coupling was achieved by performing a radiation simulation every 10 iterations to update the radiation heat flux. 3.6. Microbial inactivation model The microbial inactivation at the surface of the fruit was described using the following equations (Scheerlinck et al., 2004). dN ¼ kðT ÞN dh  kðT Þ ¼ k ref exp

ð15Þ 

Ea 1 1  R T ref T

 ð16Þ

The inactivation of conidia of important spoilage fungi, M. fructigena was targeted by integrating the equations with the computed surface temperature. The parameters for describing the microbial inactivation are Ea = 425(±28.9) kJ/mol, Tref = 316 K and kref = 0.00598 (±0.000280) s1 for M. fructigena (Scheerlinck et al., 2004). 4. Results 4.1. Comparison of simulated and measured strawberry temperature The simulated profiles at the top of strawberry fruit were compared to the actually measured profiles during radiation–convection heating at three different heater temperatures of 150, 200 and 250 C (Fig. 3). A forced airflow of 0.2 m s1 at 20 C was imposed for all cases. As shown in Fig. 3, the model predicted well the actual temperature profile at all set points with a mean error of ±2.0 C. As the heating started, the temperature of the fruit surface

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

increased rapidly and then leveled off to approach slowly a constant value.

50 Temperature, °C

Measured

Predicted

45 40

4.2. Temperature distribution during IR heating

35 30 25 20 0

200

400

600

800

1000

Time, s

50

Temperature, °C

45 40 35 30 25 20 0

100

200

300

400

Time, s 50 45

Temperature, °C

449

40 35 30

The predicted time progression of temperature contours on the surface (top view) and in a cross section of the strawberry during 450 s of heating at 200 C is presented in Fig. 4. It also presents in the bottom the measured time progression of temperature contours on the surface (top view) by using the IR thermographic camera. The same profiles can be recognized in both simulation and measurement. The strawberry heated fastest at the top surface (smallest diameter), and at the two sides of fruit closest to the heaters. These parts of the strawberry had the best exposure to the radiation. The slowest heating occurred at the bottom (at the position of the calyx leaves), which was poorly exposed to the radiation field. The strawberry was moved to the middle of the cell to avoid excessive heating of the strawberry tip. Fig. 5 clearly shows improved uniformity of heating; the calyx region remained a zone of poor heating, while overheating at the tip surface was better controlled. The temperature in the core of the strawberry increased much slower than at the surface due to internal heat resistance (Fig. 4). The aim of the IR heating is to obtain a sufficiently high surface temperature for surface microbial inactivation, without a significant increase of the internal temperature, which could cause tissue damage. After 450 s of heating, the temperature of the fruit was still lower than the heat damage point, which equals 50 C (Marquenie et al., 2003).

25

5. Discussion

20 0

50

100

150

200

250

Time, s

Fig. 3. Comparison of the predicted temperature profiles at the top of strawberry fruit with the actually measured profiles during radiation heating at three different heater temperatures of 150, 200 and 250 C.

Fig. 5 compares the surface temperature distribution of the strawberry during three different heating applications: air convection heating at 150 C, IR heating at 150 C and IR heating at 200 C. All applications used an air

Fig. 4. Predicted (top: inside, middle: top view) and measured (bottom: top view by means of the IR camera) temperature distributions as a function of heating time (0–450 s, FIR heater temperature 200 C).

450

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

Fig. 5. Strawberry surface temperature (top view) under convection (left) and FIR (middle and right) heating, at the same average surface temperature of 30 C.

50

FIR heating 150˚C

45

max surface T min surface T

40 T [°C ]

velocity of 0.2 m s1. Convection heating achieved a maximum surface temperature of 45 C, close to the critical limit above which the tissue may be damaged, after 90 s of heating. The average convection coefficient was 9 W m2 C1, comparable to the value calculated for a sphere of the same volume as the strawberry. The results in Fig. 5 show that FIR heating achieved more uniform surface heating than air convection heating, with a maximum temperature well below the critical limit of 50 C at the same average temperature. The resulting surface heating rate was, however, smaller or only equal to the air convection heating (at 0.2 m s1), depending on the heater temperature used. Corresponding time–temperature profiles are given in Fig. 6. When the maximum temperature was reached, the average surface temperature during FIR heating was still very low. Surface decontamination in this configuration could therefore not be achieved without damaging parts of the tissue. A better configuration consists of FIR heaters on four sides (strawberry at the center of the heating cell and heated from two sides, the top and bottom wall) with improved uniformity of heating, as shown in Fig. 7. Fig. 8 shows the survival curves of M. fructigena during the cyclic FIR heat treatment in Fig. 7. The effect of different heating rates at different points on the surface is translated in large variations in inactivation rates. In this operation the maximum surface temperature

Convection heating

35

max surface T min surface T

30

FIR heating 250 °C max surface T

25

min surface T

20 0

200

T [ C]

600

Time [s]

Fig. 6. Time–temperature profiles of surface temperature for the different conditions corresponding 2-sided FIR heating and the surface plots in Fig. 5.

reached 50 C after 460 s (7.6 min). Proper inactivation of M. fructigena was not achieved at this point. Continuation of FIR heating would however result in unacceptable damage of the strawberry. In order to develop a practical system for FIR surface decontamination, other periodic FIR heating cycles and other configurations should be investigated. Considering the above results, however, the feasibility of the method appears to be low. Cyclic heating in water baths has been considered before (Scheerlinck et al., 2004) and was proven successful for sur-

70

°

400

60

maximum surface temperature

50

average surface temperature

40

minimum surface temperature

30

center temperature

20 0

5

10 Time [min]

15

20

Fig. 7. Time–temperature profiles of the surface and center temperature of a strawberry during cyclic FIR heating (200 C for 200 s 20 C for 200 s), 4sided heating.

F. Tanaka et al. / Journal of Food Engineering 79 (2007) 445–452

451

100

-log (N/N0)

10 1 inact_max inact_min

0.1 0.01 0.001 0.0001 0

5

10

15

20

Time [min]

Fig. 8. Predicted inactivation curve of Monilia fructigena during cyclic FIR heating as shown in Fig. 7 at the fastest and slowest heating points on the surface.

face decontamination. Compared to water submersion heating, the FIR heating rates were more than five times smaller and the FIR surface temperature uniformity was inferior. Furthermore, fast submersion heating can be achieved at much lower temperatures (