Improvement of heat transfer during low temperature pasteurization processes

Journal of Food Eqineeting 27 (1996) 409-422 Copyright 0 1996 Elsevier Science Limited Printed in Great Britain. All rights reserved 0260-8774/96 $15.00 +O.OO 0260-8774(95)00023-2

Improvement of Heat lkansfer During Low Temperature Pasteurization Processes A. Le Bail & G. Cornier ENITIAA, Laboratoire de GCnie des Pro&d& Alimentaires chemin de la GCraudihe, 44072 Nantes, France (Received 2 February 1994; revised version received 1 August 1994; accepted 2 April 1995)

ABSTRACT Heat transfer control remains important during pasteurization of a foodstuff or ready-to-eat meal. This paper presents a new stirring system which can be used in a water cooking bath. This system, besides increasing the heat transfer rate, decreases the temperature gradient in the cooking bath. Air bubbles are introduced at the bottom of the bath and their ascent promotes stirring and uniformity of the bath temperature during cooking. Stirring eficiency has been estimated experimentally. The factors involved are: position of the foodstuff (horizontal or vertical), air mass flow rate and volume occupation ratio of the foodstuff in the cooking bath. The heating rate was estimated from the temperature rise at the centre of a slab of methylcellulose gel which was used as a model food. Analysis of the results indicates that bubbles significantly increase the heating rate of the heat transfer phenomenon, It also shows that the effect of bubble stirring is much more efficient when the slabs are in a horizontal than in a vertical position.

NOTATION

V” 2 ; j

Vat volume (m”) Experimental volume (m”) Heating rate factor (min) Relative humidity of air Heat transfer coefficient (W rnp2 K- ‘) Lag-factor 409

410 Mda

0,

P

:

Rrn t T

A. Le Bail, G. Cornier

Molar weight of dry air (29 x 10W3 kg M-‘) Occupation ratio fraction Pressure (Pa) Volumetric flow rate of humid air (m” s-l) Massic flow rate (kg s-‘) Gas constant, 8.32 J K- M-’ Time (s) Temperature (“C)

Subscripts

atm bath da (N) i Pw sa *

Atmospheric conditions Bath conditions Dry air Normal condition (P= 1.0013 x lo5 x Pa, T=O”C) Initial conditions Partial pressure of water Static pressure of air Normalized value

INTRODUCTION Heat treatment remains the most common process used for the stabilization of foodstuffs. This study is devoted to low temperature in-pack pasteurization applied to packaged ready-to-eat meals. Although the improvements in the processing equipment have been minimal, the consumption of these products continues to increase. Several technologies may be used to apply the heat treatment, depending on the goal, either pasteurization or sterilization. Two main technologies are used for the pasteurization of packaged products: steam retorts and water baths. Steam is already used in the canning industry. Steam or a mixture of air and steam can be used (Peyremorte, 1988). Steam can be combined with hot water spray to increase the heat transfer (Pinot, 1988). The water bath technology is much simpler (Hallstriim et al., 1988). The cooking bath is generally linked to a refrigerating bath to cool the foodstuff after cooking. Dreano (1988) presents equipment with a single bath in which hot or refrigerated water is introduced for cooking or refrigerating. The tested equipment has two baths: one for cooking (heated by a gas burner and a heat exchanger); and one for cooling of the packed food. Indeed, French legislation (Journal Officiel de la Rdpublique Frangaise, juillet 1974) requires cooling of ready-toeat meals at a controlled refrigeration rate (centre temperature must be under 10°C within 2 h after cooking). MATERIALS

AND METHODS

During the experiments, slabs of methylcellulose gel were used to serve as model ready-to-eat meals. The slabs were fixed between two stainless steel grids which were clipped together. Three slabs were fitted with a thermocouple at the centre for each experiment. The temperature increase

Heat transfer during low temperature pasteurization processes

411

was used to estimate the heating rate and then the efficiency of the stirring. Agitation was by air injected at the bottom of the cooking bath (AfremBretagne patent). The bubble plenum chamber was made of seven tubes which had been drilled (27 holes per tube; hole diameter=1 mm) and placed under the heat exchanger (Fig. 1). The initial temperature of the samples was 15°C. The temperature of the bath was 80°C and was controlled by a thermostat with an accuracy of k2”C. Experimental plan The efficiency of the bubbles in promoting stirring was studied using an experimental plan (Sado & Sado, 1988). A 23 factorial design with a linear model was chosen. Three factors were involved: air mass flow rate, position, occupation ratio. Codes were assigned to the different levels of each variable chosen (Table 1). (a) Air mass flow rate

The air flow rate was measured as the air velocity. A measuring duct (Fig. 2) was placed in the air distribution network between the fan and the bubble

water

level

1

,=_

mean tti;g

grid + prod;

b

as burner

heat

exch

ubble plenum

chamber

Fig. 1. Side view of the cooking bath; grids are horizontal. TABLE 1 Factors of the Experimental Factor Air mass flow rate p($&Occupation

‘) ratio

Plan

-1

Level 0

+1

0

3x10m3

6 x 10 3

(Impossible) 0.15

Vertical 0.2

Horizontal 0.11

N.B. The maximum level is equivalent and a pressure of 1.0013 x lo5 Pa.

to 5.03 x lo--’ m3 s- ’ of humid air at 20°C

A. Le Bail, G. Cornier

412

plenum chamber. The measuring duct (circular cross-section) contained a honeycomb air distribution grid (1). The static pressure was measured with a U-tube manometer (2) (accuracy f0.5 mm Hg). The mean flow velocity of air was measured with a rotary vane anemometer (3) (Testoterm 4510: accuracy O-1 m/s). The relative humidity of the ambient air was measured with a psychrometer (accuracy + 1%). Measuring duct dimensions were 80 mm ID and 15 m length. The volumetric flow rate QV of ambient humid air was given by the product of the measured velocity and duct crosssectional area. The partial pressure of water vapour in the air, Ppw, was estimated from the relative humidity measurements and the vapour pressure of water at the measured temperature. By assuming humid air as a perfect gas, the mass flow rate Qm of dry air (da) was given by expression (1). The levels of the air mass flow rates which were used are quoted in Table 1. Q

=

m

Fig. 2.

QvMia

(Pa,,+Ps,-

ppw)

R(T,,+273)

(1): Equalization grid; (2): static pressure sensor; (3): air velocity sensor.

15c Z

.92m

15c

IF

0.92m Fig. 3.

I

Top view of the cooking bath; grids are horizontal.

Heat transfer during low temperature pasteurizationprocesses

413

with the following mean values: H,,,=OG3 T,,, = Tda= 20°C P,,, z 1.013 x lo5 Pa. (b) Position of the slabs of methylcellulose gel in the bath

The cooking bath can accommodate both horizontal or vertical placements of the prepared meals. The concept of air stirring is protected by a French patent. It appears that bubble stirring is more efficient when products are placed vertically. For this reason, vertical positioning was chosen as a variable even though this is not convenient for many products, especially when using soft packaging of products containing such as meat and sauce. The levels of the position factor are quoted in Table 1. (c) Occupation ratio The occupation ratio of the foodstuff in the cooking bath is an important factor. The design of the cooking bath is such that total occupation of the available volume of the vat for both horizontal and vertical positioning is impossible. Therefore, an experimental volume (EY) is defined which is such that grids containing the foodstuff can be positioned both vertically or horizontally in the E, without changing its dimensions. The vat dimensions are 0.92 m length, O-92 m width and 0.80 m depth so that the total vat volume is (1/,)=0.68 m3. The experimental volume in which samples were located is E,=O-62 x 0.83 x 0.62 m=0*32 m3. Two kinds of gel were used as model foodstuffs: a liquid gel (95% water, 3% sodium chloride, 2% alginate) packed in plastic bags and slabs of solid gel made of methylcellulose (75% humidity, made up to DIN 8953, DIN 8954, ISO/TC/86/SC 5 and DIN 5155.2 standards) which were vacuum packed. The packed gel samples were fixed between two stainless steel grids which were clipped together. The mass of each metal grid is 1.35 kg (two grids are clipped together to fix the gel samples). The grids holding liquid gel samples contained 2.6 kg of gel (four bags). The fraction of grid covered by the bags was 0.87 (Fig. 4). The grids with methylcellulose slabs contained eight slabs. The average dimensions of a slab were 10 x 20 x 2.5 cm. The average mass of eight slabs was 4.4 kg and

grid.

4 bags of gel 7 of the grid surface)

Fig. 4.

Location of the bags of liquid gel on a grid.

A. Le Bail, G. Cornier

414 grid _

labs of the grid surface)

Fig. 5.

Location of the bags containing slabs of methylcellulose gel on a grid.

the fraction of grid surface covered by the packed slabs was 0.90 (Fig. 5). The occupation ratio is defined by eqn (2) assuming that the mass of water in the experimental volume is 320 kg (density of water equal to 1000 kg m-‘). The volume and the heat capacity of the grids have not been taken into account. O,=

Mass of foodstuff Mass of water in the experimental

volume

(2)

The maximum occupation ratio (0,=0.2, coded+ 1) was obtained with 22 grids (three grids with gel slabs interspaced among 19 grids with liquid gel) which represent 62.6 kg of foodstuff. The free space between the grids was 2.6 cm. The median occupation ratio (0,=0.15, coded 0) was obtained with 16 grids (three grids with gel slabs interspaced among 13 grids with liquid gel) which represent 47 kg of foodstuff. The free space between the grids was 54 cm. The minimum occupation ratio (O,=O*ll, coded-l) was obtained with 12 grids (three grids with gel slabs interspaced among nine grids with liquid gel) which represent 36.6 kg of foodstuff. The free space between the grids was 8.8 cm. An important point is that tilting of the grids from the horizontal position plays an important role in bubble circulation; indeed, bubbles must have an effect on the foodstuff located at the bottom of the bath as well as on the foodstuff located at the top of the bath. In our case, mean tilting of the foodstuff was +5X from the horizontal position as shown in Fig. 1. Measurements and data analysis A total of 10 thermocouples (K type) were located inside and outside some of the methylcellulose slabs. Three sensors were located at the centre of three slabs (one instrumented slab on each of the three grids containing slabs of gel). These instrumented grids were located at the top, at the middle and at the bottom of the experimental volume. Then six sensors were placed in the water close to the two sides of each instrumented slab; one thermocouple measured the bath temperature outside the experimental volume. Measurement acquisition was performed by a PC data acquisition card. All temperatures were measured at time zero and each minute

Heat transfer during low temperature pasteurization processes

41.5

thereafter. At time zero, products were immersed in the bath. Measured temperatures were normalized using the following expression: T-T,

T*=

T bath

(3) -

T

so that T*=O at initial time and T*= 1 at the end of the heating. This normalization allowed us to overcome fluctuations of the bath temperature or of the initial temperature and to make a direct comparison between the experiments. Normalized temperature was plotted against time represented by a curve shape as shown in Fig. 6. The following assumptions were made: the instrumented slab is considered as an infinite slab, the initial temperature of the slab is homogeneous, the temperature of the bath is constant during the heating and the convective coefficient h is constant. These assumptions permit the use of the Ball model (Ball, 1923). Normalized temperatures were plotted versus time with eqn (8). -~+log~.[i(T~~th-Ti~,

b3o[Tih-T*(t)]=

h

When the time function is large, the curve plotted with (8) becomes coincident with the straight line asymptote which is described by two parameters: a direction function fh which represents the rate of the heating and an intercept function j. The linear section temperature calculated with expression (8) is within 5% of the exact temperature calculated with the expressions (5) if (T * - TF)/( T&h - T*) i >0.3 (Pflug et al., 1965). The fh factor of expression (8) was then calculated from the slope using a least squares regression. The fact that the fh factor is not dependent on the thermocouple location minimizes the uncertainty of the thermocouple positioning. As the fh factor is a slope, it is less sensitive to the zero time of the experiment than an intercept value. Therefore the fh factor has been chosen as the response of

0.8 0.6 ;

0.4

0

100

200

300

400

500 Time

-

T’(1)

-

I

Fig. 6.

Normalized

600

700

800

900

1000

(I) T’(2)

-

T’(3) I

temperature

versus time for experiment

2.

416

A. Le Bail, G. Cornier

our experimental plan; it represents a mean value of the heating rate in the foodstuff. The experimental matrix was designed using the STATGRAPHICS software. It contained 23 experimental points plus four central point experiments. The experimental error was calculated from the four central points (two experiments in the horizontal position with median values of air mass flow and occupation ratio and two experiments in the vertical position with median values of air mass flow rate and occupation ratio). Experimental results are presented in Table 2. For each experiment, three values of the fi, factor were measured. Unfortunately, during certain experiments the packing of some instrumented slabs was pierced, leading to an inflow of water. The flow of water between the plastic bag and the instrumented slab adds a conductive thermal resistance to the convective thermal resistance. This delayed the heating (this happened during experiments No. 1, 2, 3 and 4 forfh factor No. 2). During other experiments, the geometry of some instrumented slabs was damaged (this happened during the experiments of the central points of the experimental matrix No. 9, 10, 11 and 12 forfh No. 3 which were done as follows). Figure 6 shows temperature versus time for experiment No. 2. The temperature coded (2) was measured in a pierced bag and has been rejected. Figure 7 shows the logarithm of [Tcath- T*(t)]. The fh constant was calculated from the slope of these curves. One can observe in Table 2 that the experimental values of fi, are scattered at f5% from the average value for each experiment (except for experiment No. 2 for which the disparity is + 10%). In the aggregate, we can assume that experimental error comes from (in order of decreasing magnitude): the influence of the location of the tylose slab in the bath (influencing the heat transfer TABLE 2 Experimental Results

f,,=heating rate

Parameters Exp. no.

: : 5

6 ;: 9 10 11 12

Air mass

Position

-1 +1 -1 +l -1 +l -1 +l

-1 -1 +l +l -1 -1 +l +l -1 -1 +l +l

flowrate

8 0 0

Occup. ratio

fhNo. 1

r: -1 -1 +1 fl +l +l

18.8 13.9 13.8 13.4 18.3 12.8 195 16.0

x 0 0

15.8 14.8 15.6 16.2

fhNo.

18.2 13.8 195 16.8 16.5 16.4 16.6 16.9

2

fhNo. 20.1 17.1 14.8 15.0 16.4 13.4 18.1 16.1

3

Mean

fh

19.4 15.5 14.3 14.2 17.6 13.3 19.0 16.3 16.1 15.6 16.1 165

N.B. Values between brackets have not been used for calculating the mean value of

f h.

Heat transfer during low temperature pasteurization processes

417

coefficient and then influencing the f,, factor); the hypothesis that the heat transfer coefficient is the same on both faces of the slab; disparity in the thickness of the tylose slabs (furthermore, the slabs were gently compressed between the metallic grids), and disparity in the thermal properties of the tylose samples. These experimental inaccuracies have been taken into account together by using the ‘central points’ of our experimental matrix.

RESULTS

AND DISCUSSION

Results are presented in Table 3. As a first analysis, we can say that the position, B, the occupation ratio, C, and the interaction of air flow rate and occupation ratio (4-C) have little influence on fh. However, the air flow rate, A, and interaction between and occupation ratio (B-C) have a large influence on fh. Bubble presence reduces fr, (a negative effect of A means that the average fh constant is reduced by 2.75 min when the maximum air flow rate is set). This effect is larger in the horizontal position (positive

0 -0.2 -0.4

f

:;:

B

-’ -1.2 -1.4 I 0

10

5

15

25

20

TwElwNl

IFig. 7.

~T~bathT~llll

-

bg[T’batbT*(2ll

T*) versus time for experiment 2. T* (1) and T* correct; T* (2): pierced bag during experiment (rejected).

~&O(~&ath-

TABLE 3 Mean Effect of the Factors of the Experimental

(3) are

Plan

Factor

Mean effect of the factor

Mean value of fh A: air mass flow rate B: position C: occupation ratio Interaction (A-B) Interaction (A-C) Interaction (B-C)

16.2 min -2.75 min -0.5 min 0.7 min 1.35 min -0.75 min 2.7 min

A. Le Bail, G. Cornier

418

effect of interaction A-B). When the slab is horizontal (code -1) the average value of fi, is reduced by 1.35 min when the maximum bubble rate is set. To confirm the validity of these effects, we need to know if they are significant. This is performed by the Fisher analysis. STATGRAPHICS creates the analysis of variance table (Table 4). In the last column of Table 4, the P-value is calculated according to the Fischer analysis. It gives information on the significance of the effects; P represents the probability of incorrectly rejecting the null hypothesis (i.e. the factor’s effect is nonsignificant). We assume that effects for which the P value is less than O-05 are accepted as significant. The validity of the hypothesis of the linear response to an effect is given by the P-value of the ‘lack of fit’ factor (Table 4). In our case the P-value is 0.24 which means that there is no lack of fit between the data and the model (with a significance level of l-0*24=0*76). We can conclude that the assumption of the linear model is suitable for our study. However, a better fit of data should be obtained with a quadratic model. The R2 value indicates the level of correlation between the predicted and the observed (measured) values; R2=0.97 means that 97% of the observed variance on data are represented by the linear model. The 3% left comes from other causes (i.e. experimental errors). Figure 8 shows a plot of observed versus

TABLE 4 Analysis of Variance

Table

ss

DF

MS

A: air flow

15.1

1

15.1

B: position

050

1

050

0.15

C: occup. ratio

0.98

1

O-98

0.075

AB

3.64

3.64

o-014

AC

1.12

1.12

0.064

Factor

BC

14.6

1

14.6

Lack of fit

0.647

2

0.32

Pure error Total R==0.97

0.407 37.0

3 11

0.136

Notation: SS=sum of squares; SS/DF; P-value = Fischer analysis.

DF=degrees

of freedom;

P-value 0*0018

0.0019 0.24

MS=mean

Significant Nonsignificant Nonsignificant Significant Nonsignificant Significant Nonsignificant

square=

Heat transfer during low temperature pasteurizationprocesses

419

J 19 observed (min) 17

,/

b

15 4

/ b

+ / .,

13

, 12

Fig. 8.

14

16

19

predicted

(min)

;

3

1

20

Observed versus predicted f,, factor.

predicted values; a reasonable fit is obtained with the linear model hypothesis. As one of our variables is not continuous (position is vertical or horizontal), two response surface plots have been plotted (Figs 9 and 10) corresponding to each position. For both plots, when the air flow rate increases, the f,, factor decreases. On the other hand, the effect of the occupation ratio is opposite in both the vertical and horizontal positions. In the vertical position, the fh factor increases when the occupation ratio increases regardless of air flow rate. In the horizontal position, the fi, factor decreases when the occupation ratio increases whatever the air flow rate. This is confirmed by the significance of the interaction (A-B) (Table 4). For a better understanding of the bubble effect, we have plotted in Fig. 11 fh versus air flow rate (with maximum occupation ratio). It shows that air bubble stirring increases the heat transfer rate in the horizontal position and that air bubbles have a lower effect in the vertical position than in the horizontal position. This phenomenon can be explained by the fact that air bubbles are more efficient in breaking the thermal boundary layer in the horizontal position than in the vertical position. The thermal boundary layer controls the heat transfer rate. A plausible explanation is that in the vertical position, stirring affects only the centre of the water volume between each food grid. In the horizontal position, stirring affects both the water volume

420

A, Le Bail, G. Cornier

ratio

Air mass flow rate (coded)

Fig. 9.

Response surface plot number 1: f,, factor in the vertical position versus air mass flow rate and occupation ratio.

and the thermal boundary layer. However, stirring affects only the lower side of the foodstuff. has also been checked using seven Temperature homogeneity thermocouples located in the bath. When the air flow rate is zero, mean fluctuation of the bath temperature is within plus or minus 2.5”C. When the bubble rate is maximum, mean fluctuation of the bath temperature is reduced to plus or minus 1°C.

CONCLUSION We can conclude by saying that bubble stirring is more efficient in the horizontal position than in the vertical position with respect to our experimental conditions. When the maximum air flow rate is set, changing the foodstuff from the vertical to the horizontal position with the occupation ratio at its maximum level reduces the fr, factor of the heating phenomenon by 3 min (difference betweenf,, factor of experiments No. 6 and No. 8). This difference represents 18% of the mean value of fr, (16.2 min). Furthermore, the vertical position is not so attractive because the food may move in its container during cooking and refrigerating. These results have been obtained on a given range of the occupation ratio. We can assume that if the foodstuffs were closer to each other in the stirring would affect the thermal boundary layer. vertical position,

421

Heat transfer during low temperature pasteurization processes

:----_._. --..._ -----__ ----._ ----...____ A-

10 m

Alr mass flow rate (coded)

Fig. 10.

Response

surface plot number 2: J, factor in the horizontal air mass flow rate and occupation ratio.

position versus

20

18

12

I

10 0.5

0

-0.5

-1

1

Air mass flow rate (coded) Fig. 11.

: vertical position

---c---

: horizontal position

f,, factor versus air mass flow rate (occupation

ratio is maximum).

422

A. Le Bail, G. Cornier

Unfortunately the positioning system does not permit us to increase the occupation ratio. We can also suppose that the ‘critical’ occupation ratio (limit of the stirring efficiency) might be greater in the vertical than in the horizontal position.

ACKNOWLEDGEMENTS Financial support for our work was obtained from the ‘Minis&e de 1’Enseignement Superieur et de la Recherche-Republique Francaise’ (grant No. 89GO505) and from the AFREM companies. The experiments were performed on an AFREM cooking bath.

REFERENCES Afrem-Bretagne (1989). French Patent No. 89-08-232. Arr&tC du 26 juin 1974. Journal ofjiciel de la Ripublique Francaise du 16 juillet 1974 p. 7389, article No. 26. Ball, C. (1923). Thermal process time for canned food. Bull. 7-1, (37) National Research Council, Washington, DC. Dreano, M. (1988). Cuisson des plats cuisines sous-vide: pro&de de cuissonrefroidissement Thermix. Industries Agricoles et Alimentaires, 105-4, 277-9. Hallstrom, B., Skjoldebrand, C. & Tragardh, C. (1988). Heat Transfer and Food Products. Elsevier Applied Science, London. Peyremorte, J. P. (1988). De la sterilisation traditionnelle a la cuisson de plats cuisines sous vide. Industries Agricoles et Alimentaires, 105-4, 301-2. Pflug, I. J., Blaidsell, J. L. & Kopelman, J. (1965). Developing temperature-time curves for objects that can be approximated by a sphere, infinite plate, or infinite cylinder. Paper 3340 of the Proceedings of the ASHRAE Semiannual Meeting, Chicago, 2.5-28 January. Pinot, J. P. (1988). Cuisson sous vide par le systeme Barriquand. Industries Agricoles et Alimentaires, 105-4, 305-7. Sado, G. & Sado, M. C. (1988). Les Plans d’Exp&ences. Ed. Dunod, Paris.