IMA Journal of Mathematics Applied in Business & Industry (1995) 5, 303-324

Heat and mass transfer in food processing

W. LEwis

AND

K. N.

SEETHARAMU

Institute of Numerical Methods in Engineering, University College of Swansea, Swansea SA2 8PP

1. Introduction Historically, the food industry and its associated technology has been developed on a relatively qualitative and empirical level. It is only in recent years that management and engineers have come to recognize that an advancement in knowledge would require a more fundamental and quantitative understanding of the underlying mechanisms of their process. The food industry is also under continuing pressure to provide food that is more natural and less processed, and also to provide food with even higher levels of safety. There is also the issue of reducing energy usage and effluent production to meet both environmental and economic goals (Johns 1992). Hence, the engineering of food processes is now undergoing a transition towards a more technical approach. The speed of product innovation in the fast moving consumer-goods business, to which the majority of the food business belongs, is accelerating. The half-life of product development times has decreased from 10 years in 1970 to what will be an estimated 2 to 3 years in the year 2000. The speed of the product/process development cycle is therefore of paramount importance (Bruin 1992). Meat and meat products worth in excess of 6000 million per year are sold in the United Kingdom, and it is estimated that the use of existing technology in the field of refrigeration could reduce the combined drip and evaporative loss by at least one percent. This would result in a minimum saving to the meat industry of 60,000,000 per annum. To achieve further savings, a more fundamental understanding of the factors influencing heat and mass transfer within meat is required (Malton & James 1984). Research has shown that the rates of reducing and subsequently maintaining the temperature of the meat has important consequences in terms of microbiological safety, eating quality, appearance, weight loss, and overall economies of the processing chain. However, there have been no legislative requirements, for internal trade within the United Kingdom, that defined specific meat temperatures. The new Food Hygiene Regulations, 1990, were first implemented on 1 April 1991. As part of the process to harmonize legislation throughout the European Community, regulations that currently only cover the export or intervention purchase of foods are being modified, and mostly will apply to internal trade by 1993. Among these regulations are some that specify the maximum internal temperature to which carcass meat must be chilled before cutting and transportation, maximum internal temperature after the freezing process, and maximum surface temperatures during thawing. Others specifically define the chilling rates for minced meat and poultry, temperatures during the storage
303

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

304

ROLAND W. LEWIS AND K. N. SEETHARAMU

of offal, and display temperatures for quick-frozen meat. The legal requirements to attain and maintain specific temperature criteria will create an increased demand for data on the relationship between environmental conditions and the temperature history of meat in carcass, joints, and many other shapes. The penalty for not complying with these requirements is 20,000. Problems caused by biological variability make it very time-consuming and expensive to produce all the data required by practical experimentation. Increasing predictive modelling techniques, usually in combination with limited experiments, are being used to generate the data and optimize the process (James 1992). Many of the problems associated with the manufacture and storage of food are similar to those found in a number of other process industries where computational fluid dynamics (CFD) is already a useful tool. The time is now right to adapt as much of the CFD experience as appropriate. This will bring significant benefits for a relatively low investment. CFD can provide benefit to the food-processing industry in the areas of drying, mixing, cooking (baking), refrigeration, and clean-room conditions. Important examples would be in the drying area, the performance of spray driers, and also the ability to solve problems of heat and mass transfer in food, along with conjugate problems where it is important to avoid the specification of unknown surface transfer coefficients, in batch-mixing vessels, in modelling of the phase-change problem of baking, and in the simulation of a cold display cabin for refrigeration (Quarinip 1992). Today, consumers are adventurous in their purchases of frozen food, and the trend is for the manufacturer to develop convenient products of high added value. This trend has been brought about by significant changes in society such as the increasing proportion of women going out to work and the greater social demand for leisure-time activities. These factors have brought about a general trend towards less formal meal times, with the replacement of traditionally prepared meals by foods that are easily stored and quick and convenient to prepare. These evolving social patterns have in some measure been made possible by the corresponding improvements in freezing, storage, and distribution technology (Summers 1984). Figure 1 shows the worldwide production of processed food (King 1984). As another example of the food-processing industry, we consider confectioneryin particular, the manufacture of chocolates. The processes required for the production of chocolates are very complex. One commonly used technique is moulding, where conditioned liquid chocolate is cast in moulds carried through a cooler by a chain system. The blocks are then demoulded and transferred, at rates up to 2000 per minute, to a battery of wrapping and cartoning machines. Many products consist of centres contained within an envelope of chocolate, these centres being a variety of confectionery materials such as creams, caramels, or fudges, together with other ingredients such as nuts, raisins, cereals or biscuits. Each type of centre requires its own special process (Greeves & Knott 1984). As a further example, beverages, fruits, and vegetables are often stored in cans. A computerized mathematical model has been developed to evaluate the thermal processing of low-acid foods (e.g. some vegetables) in cylindrical plastic cans. The sterilizing values are found to be more affected by the wall thickness than the thermal diffusivity of the wall (Shin Seonggyun & Bownik 1990). Natural convection heating

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

HEAT AND MASS TRANSFER IN FOOD PROCESSING

305

- 100 - 10

1
Million 01 tonnes per year 0.01 0.001 1530 1940 1950 1960 1970 1980 1990 2000
FIG. 1. World-wide production of processed food.
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

Year

of a canned liquid food during sterilization has also been simulated by the finite-element method (Kumar & Bhattacharya 1991). Ohmic heating is the term used to describe the electrical resistance heating of food by the passage of electric current, usually through a continuous flow of food. The major benefit of this process is that heating takes place volumetrically, and the product does not experience a large temperature gradient within itself as it heats, even when particulates are present. This obviates the need to process the liquid phase excessively to ensure heating of the core of large particulates, as happens when conventional heat-transfer equipment is used, such as in can heating by pressure cooker. This results in considerably less heat damage to the fluid and the elimination of overheating and excessive softening of the outside of the particulates. The finished product is therefore of higher quality. Skudder (1991) describes the development and commercial application of ohmic heater systems, and includes information on the extensive experiments for process validation to ensure that the consumer is provided with safe and wholesome food. There has been a growing interest in the use of microwave energy for pasteurization and sterilization of food. Since its discovery by Fleming during the second world war, distinct advantages have been found for the use of microwave heating as compared with conventional heating methods. These have been exploited to a limited extent in food processing. Rapid product heating by immediate internal heat generation and a more uniform heating of homogeneous products may be obtained, in contrast to the overheating of the surface layers or underheating of the centre region common in conventional heating methods with slow heat penetration. Centre temperatures necessary for microbial destruction are more quickly generated. This has led to the development of microwave food processes with greater retention of heat-labile nutrients (e.g. vitamins) and flavour constituents and with processing times as little as 5-10% of those obtained in conventional heating. As the differential in capital and energy cost between microwave and conventional heating equipment continues to

306

ROLAND W. LEWIS AND K. N. SEETHARAMU

narrow, the economic incentives for microwave process development will increase (Mudgett & Schwartzberg 1982; Dutta & Liu 1992). A new concept for the concentration of juice, called diffusion concentration, has been developed in which the juice percolates downwards through a packed column with dehumified air flowing counter-currently upwards. This technology avoids the thermal degradation of nutrient compounds and the loss of valuable aromatic volatiles that occurs at high temperatures. The operating conditions, including suitable hydrodynamics modelling the two phases of juice and dehumidified air, have been mathematically investigated (Xu et al. 1990). The cryoflow process is a new process for the production of frozen free-flowing droplets of liquid food products which are otherwise frozen in plastic pouches, pots, or similar containers having freezing times of the order of 10-20 minutes for products in quarter-litre plastic pouches. The process involves the injection of food products (fruit juices, yoghurts, dairy cream, etc.) into streams of liquid nitrogen. Thereafter, the food forms pea-sized droplets, which are crust-frozen in less than 10 seconds. Freezing of the product is completed in the gas phase of the nitrogen, resulting in a closely controlled product output temperature and efficient usage of nitrogen. The cryoflow process has considerable potential both for producing products of high quality and value and for adding value to some of the byproducts of the food industry. In the baking and confectionery industries, there are many applications for cryoflow products. Quality, flexibility, and low capital have always been three of the great advantages of liquid-nitrogen freezing systems (Taylor 1984; Rosen 1990). To achieve considerable savings and to optimize the thermal processing of solid foods, it is essential to have a more fundamental understanding of the factors influencing heat and mass transfer in porous bodies. Luikov's coupled partial differential equations for heat and mass transfer can be used to describe the multiphase distribution in porous media, for both the freezing and drying processes. The formulation of these coupled equations is based on the principles of the conservation of heat and mass transfer and irreversible thermodynamics. A detailed description and derivation of the equations involved has been presented by Luikov (1975) and Thomas et al. (1980). Due to the complexity of these equations, any analytical solution is mostly restricted to simple one-dimensional forms (Mikhaelov 1976; Dural & Hiens 1990; Salvador & Mascharoni 1991). The effective speed of electronic computation has approximately doubled every year over the past 30 years, and the trend is expected to continue over the coming decade. As this rapid increase in computational capabilities enables solutions to be obtained for increasingly complex problems, numerical methods, especially the finite-element technique, are becoming evermore widely accepted for solving partial differential equations. Comini & Lewis (1976) successfully applied the finite-element method to solve the linear form of the Luikov equations for heat and mass transfer in two dimensions. Some typical examples of the freezing and drying problems are used to demonstrate the feasibility of their solution scheme. Later the fully (and partially) nonlinear forms of these coupled equations were solved by Thomas et al. (1980), with examples of stress induced in the drying process of timber. Volume changes also occur during food processing, and a number of diffusional models take this change of volume into account (Fusco et al. 1991; Hawlder 1991; Rohman & Lai 1990). An implicit

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

HEAT AND MASS TRANSFER IN FOOD PROCESSING

307

finite-difference method was used to integrate the partial differential equations (Sereno 1990) in order to obtain both temperature and moisture profiles. Sereno's simulation predicted the experimental results for the drying of apple slices and carrot cubes to within 1.1% of moisture content and 12% of the drying rate. Many solid foods are first chilled and then freeze-dried for storage purposes. In the above two processes, convective heat exchange between the surrounding medium and the food has been the sole method of heat transfer. However, in 'vacuum' thawing, very high heat transfer coefficients are produced by the condensation of low-pressure steam on the surface of the meat being thawed in an air-free chamber. A mathematical model, including volumetric changes for food freezing or thawing, is obtained by modifying the general heat-conduction equation. The shape of the food is assumed to be a body of revolution (Sheen & Hayakawa 1991). A number of procedures are used to predict the freezing time of food, and experimental measurements can be made to test the accuracy, adequacy, and applicability of the various mathematical models, e.g. the freezing time of a small cuboidal food productviz. french friessee LeBlanc et al. (1990). In part 2 of the same paper, nineteen mathematical models are compared with experimental results. Three empirical models and one approximate model yielded estimates within 10% of the experimental freezing time. Other empirical and appoximate models gave answers greater than 10%. Neither of the two exact models examined were applicable to the freezing of a small food product when the surface heat-transfer coefficient was finite. The effective average heat-transfer coefficient is a complex function of many physical and experimental variables: flow conditions; velocity profiles; food shape and orientation; air temperature, velocity, and humidity; radiation; water evaporation; respiration; and phase changeall these have an effect upon the transient heat loss from the food product, and hence affect the effective heat-transfer coefficient. The local values vary over a wide range, and thus the average value is only value that is practical from a design consideration. The value of the heat-transfer coefficient is based on experimental methods of measuring both the surface and centre temperatures of the food product, e.g. apples or potatoes (Stewart et al. 1990; Ibrahim 1991). Heat-transfer coefficients are determined in the case of fixed-bed and fluized-bed freezers with the aid of computer models (Kairullah & Singh 1991). A surface-boiling boundary condition is encountered in the freezing of foods that are immersed in boiling freezants, such as R12. This phenomenon may be incorporated in a mathematical model of the freezing process as a surface-temperature-dependent convective boundary condition (Evans et al. 1991). The determination of accurate thermophysical data for moist food materials under various conditions is an immediate necessity for a better prediction of food processing (Narayana & Murthy 1981). Most foods spent the greatest part of their lives in some form of 'package'bag, box, sleeve, folded paper/plastic/foil, sealed paper/plastic/foil, can, bottle, and so on, often in multiple layers in single or multiple applications. The interrelationship between the processing and packaging is often ignored, or given little thought, although some major companies are now realizing that the whole system, from the input of raw material to the point of consumption is, in fact, one long process, requiring a total-integration approach by engineers and scientists of all disciplines

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

308

ROLAND W. LEWIS AND K. N. SEETHARAMU

(Batterly 1984). Emulsive edible film on food products is often used to control mass transfer, moisture movement, oxygen transfer, and preservative migration. The microvoid model and micropathway model are used for such mass transfer (Krochta 1990). Chemical and colour changes in tomato paste of 26% total solids, as influenced by storage temperature duration and type of packaging material, have also been investigated (Luch & Sharpe 1982). Mathematical models for the prediction of the behaviour of food products stored inflexiblepackaging materials have been presented for the case of moisture diffusion and adsorption. These models can be used for a prediction of the shelf life of various food products and for the optimization of flexible packaging materials (Khanna & Pappas 1982). In the sections to follow, the Luikov system of partial differential equations, which calculates the distribution of temperature, moisture potential, and pressure within a capillary porous body such as food, is given. Followed by this, some examples of the food processing are presented. 2. Analysis
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

2.1

Governing system of equations

The governing system of equations which describe the variation of temperature, moisture potential and pressure within a capillary-porous body, as used by Lewis & Ferguson (1990) and Ferguson & Lewis (1991) to solve practical engineering problems are * dt Kl2V2U + Kl3V2P
l2

(1) Cp = K3lV2T + K32V2U + K33V2P where C, = pocq ^ i i = (kq + eXkm6') K21 = M ' Cm = p o c m Kl2 = ekkm K22 = km Cp = pocp K13 = e-i/cp K23 = kp ^23 = p

(see Table 1 for nomenclature). 2.2 Boundary conditions

The boundary conditions associated with these equations are T=TW dT ~ -c/ f l ) = 0 onf2 (3) onr\ (2)

HEAT AND MASS TRANSFER IN FOOD PROCESSING TABLE 1

309

Nomenclature P T U V" cm cp cq j, j , dif j , fil jq km kp kq au at, 6 e p0 /. total pressure of humid air inside the body, kN/m2 temperature, ; C moisture potential, M element deformation velocity (m s " ' ) moisture capacity, kg/kg M coefficient of humid air capacity, kg/kg n M 2 heat capacity, J / h r m k density of mass transfer flow density of diffusion mass transfer flow density of filtration mass transfer flow heat flux coefficient of moisture conductivity, m / h r 2 moisture filtration coefficient, k g m / s k N coefficient of thermal conductivity, W / m - ' C convective mass transfer coefficient, kg/rrr-s- c M convective heat transfer coefficient, W/"C-m 2 thertnogradient coefficient, "M K ratio of vapour diffusion to total diffusion dry density, kg/m 3 latent heat of vaporisation of water, J/kg

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

U = UW km ^-+jm en P = PW. + kj ^ + a u (t/ - 14) = 0 en

on T3 on T4 on T5.

(4) (5) (6)

Equations (2), (4), and (6) represent the portion of the material boundary where a constant temperature, moisture content and pressure, respectively are applied. Equations (3) and (5) represent the portion of the boundary to which a specified heat flux or moisture flux is applied. This set of boundary conditions can be written in the more generalized form

r=T

on F,

(7)
2

dT

, = o on F *

(8) (9) (10) (11)

u = IT
dU
K22

on F 3 * = 0 on F4 o n T5

~dn~ +

P = PW where

310

ROLAND W. LEWIS AND K. N. SEETHARAMU

A,-

K2I5'*r 2I5*
p

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

J*. =

K? > I

2.3 Sublimation front tracing scheme Freeze-drying can be closed as a Stefan problem, where instead of melting, sublimation takes place and the moving boundary is the interface between the frozen material and the dried material. Initially, when all the material to be freeze-dried is frozen, the moving boundary is situated at the sample surface and as the drying process progresses with time, the frozen/dried region interface advances through the material sample. Once the interface has advanced through the sample, when no further sublimation takes place and only a dried region remains, the sample is further dried until the desired residual moisture content is obtained. Therefore, in order to be able to model the freeze-drying process the exact location of the sublimation interface must be calculated. This is done by studying the equilibrium of heat energy existing at the advancing sublimation interface. On assuming that there are no heat sources along the interface, then the heat balance equation may be expressed as dT,ried d dx
~~ / '-Pfrozen V int x=b

dx

(12)

The first term in equation (12) describes the heat flux into the interface from the dried region whilst the second term describes the heat flux from the interface into the frozen region. The third expression describes the heat given oft" at the interface due to the phase change caused by sublimation. The mesh deformation is accounted for by introducing the finite-element shape function as an implicit function of time. Hence, the node movement A^T), is given

HEAT AND MASS TRANSFER IN FOOD PROCESSING

311

by the following expression:

The approximation function, T(x, x), and its derivative with respect to time are and dx [_dr dx

The derivative of the shape function with respect to time, described in equation (15), is used to detect the distortion of the element due to the moving front. The deformation of the element in the global coordinates can be mapped into the local coordinates by using the isoparametric transformation, such as I i f ) . The element deformation velocity, V, is given by (16)
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

The mesh deformation in the global axes, described by equation (17), can be expressed in terms of the element deformation velocity, Vc, and gives: ^
OX

(18)

By substituting equation (18) into equation (15), the time derivative of the approximated function will be

dx

^ dT

'

In the absence of mesh deformation, equation (19) will become the conventional finite-element equation with the shape function depending on the spatial domain. The application of the finite element discretization will derive a total differential equation of the unknown function, T, with respect to time. When replacing the traditional shape function defined in equation (13), the resulting approximated function will have an additional term, equation (19), which can be viewed as the convective effects due to mesh deformation. If the finite-element technique and a suitable weighting procedure are applied to evaluate the set of parabolic equations, the resulting total differential equations will contain a convective term in the stiffness matrix, which is similar to that derived from the convection-diffusion equation. One of the advantages of using the space-time shape function is that the existing finite-element program can be easily altered to accommodate the mesh deformation. The moving interface is described by the Stefan boundary condition, equation (12);

312

ROLAND W. LEWIS AND K. N. SEETHARAMU

its location can be obtained at each timestep by any finite-difference approximation. Hence (20)
AT

If the front location at the nth iteration is given, the new position can be derived by iteration according to the following equations:
J

/n+l _ fn + ,
J

AT

\U

/.Pol

"^frozen >zen

frozen

dx

K n

<jdried ' "", 1

^Sried]

ex

\L' >

,.,.,

with
'--'dried

dx
?T
v

= a

__ j f

dx
frozen

( 1 - -a)-

^dried

dx 1.
'frozen

(22)
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

* frozen

dx

dx

This method was used by Lynch & O'Neill (1981) to solve a one-dimensional Stefan problem where both one-phase and two-phase numerical results were found to be quite accurate in comparison with analytical solutions. 2.4 Finite-element formulation

The variation of the temperature, pressure, and moisture potential throughout the domain of interest, Q, is approximated in terms of the nodal values, Ts, Us, and Ps. If the approximation given by equations (19) and the approximations for the working variables, temperature, moisture potential, and pressure, are substituted into equations (1) a residual is obtained, which is then minimized using the Galerkin method. This requires that the integral of the weighted errors over the domain, 2, must be zero, with the shape functions, Nr, being used as the weighting functions.

A V-(K,,Vr) + V(Kl2VU) + V(Kl3VP)-Cq?f U Nr\ V-(X21Vr) + V-(K22VU) + V(K13VP)-Cm

dfi = 0 dQ = 0 dft = 0. ,
(23)

r Nr\V(K3lVT) Jn L

+ V(K32VU) + V(K33VP)-Cp

3P~\ ct J

The application of Green's theorem (integration by parts) and the introduction of the generalized boundary conditions to equations (23) produces a system of differential equations which may be written in matrix form as

dt

= 0

(24)

HEAT AND MASS TRANSFER IN FOOD PROCESSING

313

where K(<t>). C(<D). and C'(<D) are solution-dependent matrices.

K =

K. ) [
^32

0 C = 0 0

0 0 (

cm
0 0

o "

C' =

0 0 T~ V p

0 0 0 0.
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

J =

Typical matrix elements are

c rs = t f c/vr/
'1=1 J n *

C'/VrK"V/VsdQ
r.s=l J o

[
>-=i J r

4=1 [
r=\ J f

314

ROLAND W. LEWIS AND K. N. SEETHARAMU

2.5

Timesteppiny algorithm

The numerical solution of equation (24) is achieved by using the Lees three-level timestepping scheme (Lees 1966), which employs the finite-difference technique in time. ($" +' 4. d>" + fly - ') 3
+

(fl>"+1 _ d)""1) 2Ar

((D"+1 -(- cp" + O""1) 3

(25) The superscript ;i refers to the time level whilst Ax refers to the timestep. The Lees three-level timestepping scheme has the advantage of solving for the time level ;i+ 1, by evaluating the coefficient matrices at time level n, which avoids the necessity for an iterative solver. Wood (1978) and Wood & Lewis (1975) showed that this scheme was stable although, when a convective boundary condition was used, oscillations appeared in the solution. The noise can be dampened to an acceptable level by introducing a maximum permissible time step.
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

3. Application 3.1 Freeze-dryiny of coffee solution

The results presented in this section are a comparison of experimental and computational results for the freeze-drying of a 3.6% coffee solution (Ferguson et al. to appear). The coffee solution was made up by dissolving coffee granules in water until the desired concentration was attained. For a 1 kg coffee solution sample, 36 g (336 cm3) of coffee granules were dissolved into 964 g (964 cm3) of water. The coffee sample to be freeze-dried was placed into the material holding tray within the drying chamber to a uniform depth of 10 mm. The solution within the tray is idealized as
TABLE 2

Material properties for the freeze-drying of a 3.6% coffee solution Material property Po

Dried S.I. units kg/m3 J/hrmK kg/hr-m-N-m"2 m/hr2 J/hrmK kg/kgn-M" 2

granular 34.42 39.96 1.7* 10" 6 3.72* 10" 3 2600.0 8.0* 10" 2 0.1 0.01 0.865 8.78* 1 0 ' 3 Frozen 962.0 720.0 2.0* 10 " 8 5.0* 10" 6 1900.0 8.0* 10" 2 1.0 0.001

km
c

i
P

MK W/m 2o C kg/m2s-M

HEAT AND MASS TRANSFER IN FOOD PROCESSING

315

43

59

75

91

107

123

B
FIG. 2. Finite-element discretization of the model for the 3.6"; coffee solution.

an infinite sheet, hence, the numerical simulation of the sublimation front becomes a uni-dimensional problem, since edge effects are ignored. The material properties employed during the numerical simulation of the freezedrying problem were obtained experimentally and assumed to be constant throughout the freeze-drying process. The material properties used are shown in Table 2. The initial conditions of the frozen coffee sample, before the freeze-drying operation commences, are temperatures 35 3 C, moisture content 96.4% and pressure 0.5 kN/m 2 . In order to ensure that sublimation, and not melting, takes place the pressure within the drying chamber must be held below the triple point pressure of water, such that the pressure/temperature path crosses the sublimation/ablimation. The heater, in this example, was placed 15 mm away from the upper surface of the frozen coffee sample and held at a constant temperature of 300cC. A radiative heating type boundary condition was applied to temperature and a convective type moisture boundary condition was applied, with a steady-state equilibrium moisture content of 4%. The finite-element mesh used in the numerical modelling of this freeze-drying example is shown in Fig. 2. The boundary conditions were applied along the face AB, whilst all other faces, BC, CD, and DA were assumed to be insulated non-conducting boundaries. Thermocouples were placed at locations within the coffee sample to measure the temperature, and coincided with nodes 3, 43, 59. 75, 91, 107, and 123 of the finite-element mesh. Figure 5 shows the variation of total moisture content of the coffee sample with time. Again for the first 25 to 30 hours of the freeze-drying operation the total moisture content within the body decreases linearly with time, indicating that the sublimation interface advances with a constant velocity through the coffee sample. It is not until after sublimation has ceased, and drying continues, that the moisture content within the sample approaches its residual value. The numerical simulation of the problem compares favourably with the experimental results, although, the numerical model predicts a slightly faster drying time than was actually achieved experimentally. Figures 3 and 4 show the variation of temperature with time at nodes 75 and 123 of the finite element mesh. Again, in proceeding through the sample the temperature takes longer to reach steady state equilibrium values, thus giving an indication of the position of the sublimation front with time. The numerical solutions of temperature against time compare well with the observed experimental values, which on inspection

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

316

ROLAND W. LEWIS AND K. N. SEETHARAMU

60 50 40
O30 20

Experimental Computational

10

10

15 20 25 30 35 40 Time (hours)

45 50

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

FIG. 3. Temperature versus time at node 75.

60 50 40

8 30

1
| 10

Experimental Computational

oo

I o
-10 -20 -30 -40' 5 10 15 20 25 30 35 40 Time (hours) 45 50

FIG. 4. Temperature versus time at node 123.

HEAT AND MASS TRANSFER IN FOOD PROCESSING

317

Experimental Computational

10

15

20 25 30 Time (hours)

35

40

45 50

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

FIG. 5. Total moisture content versus time.

of the figures, tend to be scattered. However, the numerical results compare very favourably with the overlying trend of the experimental solution.
3.2 Seafood processing

Seafood items such as fish, shrimps, prawns are shipped from India to most of the European countries. During transit, these items are to be preserved by passing cold air at a temperature of around 5C and at high relative humidity over the seafood, to maintain the quality. This involves both the heat and mass transfer between the seafood and surroundings. Before processing the fish, the head and tail are removed and its trunk is gutted. In order to simulate the actual conditions one such specimen is taken and its

FIG. 6. Discretization of the model.

318

ROLAND W. LEWIS AND K. N. SEETHARAMU

longitudinal section investigated. Uniform boundary conditions were assumed along all the boundary surfaces with constant temperature and moisture potentials. The thickness of the fish was assumed to be constant. The longitudinal section of the specimen was then discretized into finite elements and is shown in Fig. 6. The nodes are numbered in such a way that the bandwidth in the global matrix was a minimum. The values of the thermophysical properties (Khumbar 1984) of fish, which were assumed to be constant, are given below. k= 1.00498; Am = 8.90 x 10 - 3 . : cm = 4.1598 x 106. c, = 7.429 x 106;

The conditions under which the fish was processed are, Initial conditions Temperature = 35 JC Moisture potential = 45C. Environmental conditions
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

Case 1 Case 2 Temperature Moisture potential 5C 125M 0.5C 125.0M.

The temperature and moisture potentials corresponding to selected nodes on the surface and interior or the body are plotted as a function of time in Figures 7 and 8. Figure 7 shows the temperature and moisture potential variation for an environment of 5C and 120M. Figure 8 shows the temperature variation for nodes 3, 16

40 0 10 100 1000 Time (seconds) 10000

FIG. 7. Temperature and moisture variations at nodes 3, 16, 10. and 12.

HEAT AND MASS TRANSFER IN FOOD PROCESSING

319

35
t =5"C

170

30

V*^ Temperature without ^ / ^

1150 otent ial

Temjlerature <

25 "moisture potential ^ c ^ v 20 15 10 5 0
3. 16
X

\ Temperature with V\ moisture potential

130110 90 70 50 30

Si 10^< |

.2

Moisture p o t e n t M ^ ^

i
10

OT
10000

100 1000 Time (seconds)

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

Flu. 8. Temperature (dashed curves), neglecting moisture migration, compared with temperature curves accounting for moisture migration.

(on the surface) and 10, 12 (interior) as a function of time. It is seen from Fig. 7 that for a period of 100 sees the temperature drop is small for the surface nodes 3, 16 whereas the temperature at the interior nodes 10, 12 remain constant up to 400 sec. After these time limits, the temperature drop is considerable. The temperature values at the surface approach the environment temperature after 7000 sec whereas the interior values remain higher than the surface temperatures by 3CC. Figure 7 shows the moisture potential variation for the same set of nodes. It can be observed that the moisture potential remains almost constant for all nodes up to a period of 100 sec. After this, the moisture potential for the surface nodes increases rapidly with time as expected and approaches the environmental value in about 4000 sec. The moisture potential at the interior node remains (node 12) constant for about 4000 sec and then starts increasing. Similar trends are observed in Fig. 8 for the temperature and moisture potentials when the environment of the processing is maintained at value of OS'C and 125M, respectively. When moisture migration is neglected in the analysis the results are shown in Fig. 8, indicates that the temperature was lower (after 10 sec). The time required was also less compared to the case when moisture migration was accounted for. 3.3 Hamburger processing

As a third example, the processing of a hamburger was considered. Because of the symmetry only one-fourth section analysed. The initial conditions in the hamburger were assumed to be lS^C and 100cM. The results obtained for the temperature and moisture potentials in the hamburger are shown in Figs 9 and 10 for two instants

320

ROLAND W. LEWIS AND K. N. SEETHARAMU

Temperature at O.Olhrs
x x x x x x x x x x x x
X x x x xx xx xx xx xx xx xx x X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X K X X X X X X X X X X x x x x x x x x x x x x x x x x x x x x x x x X X X X X X x x x x x x x x x x x X X X X X X X x x x x x x x x x X X X X X X X X X X X X X )

x x x x x x x x x x

x x x x x x x x x
X X X X X X X X X X X X X X X X X X X X X X X X X X X 3 X X X X X X X X X X X X X X Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

X X X X X X X X X X X X K X X X X X X X X X X X X X X X X X X X X X X X X

x x x x x x x x x x x x x
X X X X X X

x x x x x x x x x x x x x x x x x x x x x
X X X X X X X X X X X X

x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x X X X X X X X X X X X X X C X X X X x x x x x x x x X X X X X X X X X X X X X X x x x x x x x x x x x x x x x
X X X X X X X X X X X X K X X X X X X X X X X X X X X X X X X X X X X X X X X X X

x x x x x x x x x x x X X X X X x xX xX xX xX xX x X x X x X x X X X X X X X

X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X x x x x x x x x X X X X X X X X X X X X X X X X > X X > X

X X X X X C X X X X X X X X X

Temperature at 0.0341us
X X X X X X X X X X X X

20-C 19'C

Initial temperature - - 18'C

9.

of time during the processing. Thus, any specified condition, either at the surface or at the centre can be achieved by suitable processing environments. 4. Conclusions The importance of the coupled phenomena of heat and mass transfer in food processing is highlighted along with the ensuing commercial aspects. A numerical model is presented to calculate the distribution of temperature, moisture, and pressure within a capillary-porous body, according to the system of partial differential equations as defined by Luikov. Examples of the freeze drying of a coffee solution

HEAT AND MASS TRANSFER IN FOOD PROCESSING

321

Moisture potential at 80.25hrs

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

Moisture potential at 205.25his Initial moisture potential - 100'M

FIG. 10.

and the processing of both seafood and reconstituted meat products are given to illustrate the application of the theory. A versatile three-dimensional heat and mass transfer model of food processing with

322

ROLAND W. LEWIS AND K. N. SEETHARAMU

freeze drying, thawing and shrinkage prediction capabilities is an urgent need for achieving a better simulation of such phenomena. The surface heat and mass transfer coefficients can be generated by the application of CFD software for better and economic simulation of food processing. There is a need to evaluate the properties of foods under various conditions with a limited number of experiments but gainfully using techniques such as inverse methods.

REFERENCES

BATTERLY, P.. 1984. High speed packagingthe process engineer's contribution. Profitability of food processing. IChemE. Symp. Series No. 84. Pp. 231-40. BRUIN. S.. 1991. Integrated process design: issues and opportunities. Food and Bioproducts Processing 70(3), 126-30. COMINI, C . & LEWIS. R. W.. 1976. A numerical solution of two-dimensional problems involving heat and mass transfer 19, 1387-92. DUTTA, A. K... & Liu. J., 1992. Thermal time distribution for microwave and conventional heating of food. Food and Bioproducts Processing 70(2), 83-90.
EVANS. K. E.. MANNAPERUMA, J. D., & SINGH, R. P.. 1991. Freezing of foods under surface
Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

boiling conditions. Int. J. Refriy. 14(4). 2409. FERGUSON, W. J., & LEWIS. R. W., 1991. A comparison of a fully nonlinear and a partially nonlinear heat and mass transfer analysis of a timber drying problem. Proceedings of the 7th International Conference on Numerical Methods in Thermal Problems. Pp. 973-84.
FERGUSON, W. J.. LEWIS. R. W.. & TOMOSY, L.. A finite element analysis of freeze-drying

of a coffee sample. Computational Methods in Appl. Mech. Engy. (to appear). Fusco, A. J. ET AL., 1991. Diffusional model for drying with volume change. Drying Technology 9, 397-417. GREEVES, T. W.. & KNOTT. C. M., 1984. The role of integrated control and information systems in manufacturing productivity. Profitability of food processing. IChemE. Svmp. Series No. 84. Pp. 121-8. HAWLDER, M. N. A.. 1991. Drying charactacteristics of tomatoes. J. Food Enyg. 14(4), 259-68. IBRAHIM, D., 1991. Simple model for estimation of film coefficient during cooling of certain spherical food stuffs with water. Int. Comm. Heat & Mass Transfer 18, 431-43. JAMES, S. J.. 1992. A review of predictions of heat and mass transfer in meat refrigeration. Food engineering in a computer climate. IChemE Symp. Series No. 126. Pp. 4810. JOHNS, W. B., 1992. Simulation of food processes with uncertain data. Foof and Bioproducts Processing 70(2). 59-68. KAIRULLAH, A., & SINGH, R. P., 1991. Optimisation of fixed and fulidised bed freeze processes. Int. J. Refrig. 14, 240-5. KHANNA, R.. & PAPPAS. N. A., 1982. Mathematical analysis of transport properties of polymer films for food packaging: moisture and oxygen diffusion. Food process engineering (H. G. Schwartzberg, D. Land, & J. L. Bomber, Eds). AIChE. Symp. Series No. 78. Pp. 185-91. KING, R., 1984. Carbohydrate foods for the nineties. Profitability offood processing. IChemE. Symp. Series No. 84. Pp. 359-72. KUMBHAR, B. K., 1984. Studies on transient heat transfer and thermal properties of food during freezing. Ph.D. thesis, Indian Institute of Technology, Delhi. KROCHTA, J. M., 1990. Emulsive films on food products to control mass transfer. Food emulsions and foam, theory and practice. A.I.Ch.E. Symp. Series No. 86. Pp. 57-61. KUMAR, S.. & BHATTACHARYA, M.. 1991. Transient temperature and velocity profiles in a canned non-Newtonian liquid food during sterilisation in a still cook retort. Int. J. Heat & Mass Transfer 34, 1083-96.
LEBLANC, D. I., ROBERTS, K., & TIMBERS. G. E.. 1990. Freezing of a parallelpiped food

product: part 1, experimental determination; part 2, comparison of experimental and calculated results. Int. J. Refrig. 13, 371-8, 379-92.

HEAT AND MASS TRANSFER IN FOOD PROCESSING

323

LEES, M., 1966. A linear three-level difference scheme for quasilinear parabolic equations. - Maths Comp. 20, 516-22. LEWIS. R. W.. & FERGUSON, W. J.. 1990. The effect of temperature and total gas pressure on the moisture content in a capillary porous body. Int. J. Numer. Meth. Engg. 29, 357-69. LUH. B. S., & SHARPE, R., 1982. Stability of tomato pastes packaged in plastic laminated pouches. Food processing engineering (H. G. Schwartzberg, D. Land, & J. L. Bomber, Eds). AIChE. Symp. Series No. 78. Pp. 178-84. LUIKOV, A. V.. 1975. Systems of differential equations of heat and mass transfer in capillary-porous bodies (review). Int. J. Heat & Mass Transfer 18, 1-14. LYNCH. D. R... & O'NEILL. K.., 1981. Continuously deforming finite elements for the solution of parabolic problems with and without phase change. Int. J. Numer. Meth. in Engg. 17, 1781-96. MALTON. R., & JAMES. S. J., 1984. Using refrigeration to reduce weight loss from meat profitability in food processing. IChemE. Symp. Series No. 84. Pp. 207-17. MIKHAILOV, M. D.. 1976. Exact solution for freezing of humid porous half-space. Int. J. Heat & Mass Transfer 19, 651-5. MUDGETT, R. E.. & SCHWARTZBERG, H. G., 1982. Microwave food processing: pasteurisation and sterilisationa review. Food process engineering (H. G. Schwartzberg et al., Eds). IChemE. Symp. Series No. 78. Pp. 1-11. NARAYANA, K. B., & MURTHY. M. V. K... 1981. Heat and mass transfer characteristics and the evaluation of thermal properties of moist food materials. Trans. ASAE 24, 789-93. NILUFAR. D., & HIENS. A. L., 1990. Fast approximate analytical solution to Luikov's equations with variable thermophysical properties. J. Food Engg. 12, 191-206. QUARINIP, G. L., 1992. Computational fluid dynamics applied to food engineering and storage. Food engineering in a computer climate. IChemE. Sym. Series No. 126. Pp. 25-35. ROHMAN. MD.. PRAKASH. L. P., 1990. Shrinkage and density of squid flesh during air drying. J. Food Engu. 12, 133-43. ROSEN, J.. 1990. Freeze concentration beats the heat. Mech. Engg. 112(12), 46-50. SALVADOR, V. O., & MASCHARONI, R. H., 1991. Prediction of freezing and thawing times of foods by means of a simplified analytical model. J. Food Engg. 13, 67-78. SEONGGYUN, S.. & BOWNIK., S. R., 1990. Computer simulation to evaluate thermal processing of food in cylindrical plastic cans. J. Food Engg. 12, 117-131. SERENO. A. M.. 1990. Simplified model for the prediction of drying rates of food. J. Food Engg. 12, 1-11. SHEEN, SH., & HAYAKAWA. K.-I., 1991. Finite difference simulation for heat conduction with phase change in an irregular food domain with volumetric change. Int. J. Heat & Mass Transfer 34, 1337-46.
SHERIFF, S. M., VEDAMOORTHY. V. N., & SEETHARAMU, K. N., 1985. Modelling of sea-food

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011

processing using finite element method. Proceedings of the Natural Heat and Mass Transfer Conference, Visakapatnam, 1985. Pp. 17-21. SKUDDER, J. P.. 1991. Industrial application of the ohmic heater for the production of high-added value food products. News in Aseptic Processing and Packaging, Espoo, Finland, Jan. 30-31. Pp. 47-52. STEWART, W. E., JR., ET AL., 1990. Experimental method of approximating effective heat transfer coefficients for food products. ASHRAE Trans. Pt 2. Pp. 142-7. SUMMERS, J. V., 1984. Cryogenic food freezing in today's market and its costs related to conventional mechanical systems. Profitability of food processing. IChemE. Symp. Series No. 84. Pp. 241-9. TAYLOR, R. I., 1984. A new cryogenic process for the food industry. Profitability of food processing. IChemE. Sym. Series No. 84. Pp. 231-40.
THOMAS, H. R., MORGAN, K., & LEWIS, R. W., 1980. A fully nonlinear analysis of heat and

mass transfer problems in porous bodies. Int. J. Numer. Meth. Engg. 15, 1381-93. WOOD, W. L., 1978. On the Zienkiewicz three and four level schemes applied to the numerical integration of parabolic equations. Int. J. Numer. Meth. Engg. 12, 1717-26.

324

ROLAND W. LEWIS AND K. N. SEETHARAMU

WOOD. W. L.. & LEWIS. R. W.. 1975. A comparison of time marching schemes for the transient hat conduction equation. Int. J. Numer. Merit. Enyg. 9, 679-89. Xu. Y. R. ET 4L., 1990. Study of diffusion concentration of juices. Proceedings ofCHEMECA 90 (Australian Chemical Engineering Conference), Aug. 27-30, 1990. Pp. 674-81.

Downloaded from imaman.oxfordjournals.org at Queensland University of Technology on September 30, 2011