Optimisation of industrial Food Drying Operation Henry T. Sabarez CSIRO Food and Nutrition Flagship, Werribee 3030, Victoria, Australia Abstract The process of drying food materials is extremely complex involving coupled transient mechanisms of heat, mass and momentum transfer phenomena accompanied by physical, chemical and phase change transformations. It is an energy-intensive operation that usually imparts significant alterations in product quality and functionality attributes. With increasing fuel prices, large energy consumption is a major concern in industrial drying not only due to the cost and increasing scarcity of fuel but also for the associated environmental impact as the necessity to reduce greenhouse gas emissions is critical for a sustainable future. This is coupled with an increasing consumer demand for healthy and high quality processed foods which offer more convenience. There is a significant interest to develop and optimise drying operations that will accelerate the process (increased throughput) and reduce energy consumption without compromising the quality of the dried products. This paper discusses the development and validation of a computational drying model and its application in optimising the design and process conditions of an industrial fruit drying system. Keywords: Food drying, optimisation, modelling, industrial scale Drying is one of the most important processing operations in the food manufacturing industry. It is an energyintensive process which constitutes a significant portion of the total production costs of dried-based products depending on the dryer design, operating conditions, properties of food materials, etc. In addition, the process of drying usually affects the quality attributes of the product due to exposure to high temperatures or long drying times. The main challenge is to develop and optimise a drying process in terms of increased production throughput and reduced energy consumption without compromising the quality of the dried products. In industrial drying operation, tunnel dehydrators are the most widely used method for drying fruits (e.g. plums, grapes, etc) and vegetables. The tunnel dehydrator basically consists of a tunnel (as a drying chamber) containing trays of product that are placed on mobile trolleys (also referred to as trucks) moving along the tunnel, a fan to circulate the heated air, and a heating unit to preheat the air before it is blown across the food product and then vented to the exhaust (Fig. 1). The circulated air is Corresponding author e-mail: henry.sabarez@csiro.au 1
directly heated by a gas burner and the heated air is forced into the tunnel by a fan. In tunnel drying, the raw materials are loaded onto the trolleys of trays and these trolleys are then fed into the drying tunnel at one end at regular intervals on a continuous basis and the dried product is unloaded at the other end of the tunnel. I. Materials and methods A series of laboratory drying experiments were undertaken to obtain drying kinetics data under varying operating conditions for validation of the predicted results from the proposed model. The drying experiments were performed using a computer-based drying system as illustrated in Fig. 2. The details of the drying setup are described elsewhere (Sabarez, 2012). To date, this method of drying continues to be used at industrial scale in fruit drying because it is much cheaper and easier to expand the capacity, in addition to its lower capital cost. However, the tunnel drying system is inherently less efficient than other drying systems (e.g., continuous belt dryer). There is a significant interest in the food and other processing industries to improve the efficiency in tunnel drying to maintain sustainability due to the prospect of a continuously increasing trend in fuel costs and the need for eco-friendly processes to mitigate environmental impact coupled with the rising consumer demand for high quality products. This paper presents the development and validation of a computational model accounting for the simultaneous transfer of heat, mass and momentum in air as well as within the food, capable to predict the drying behaviour of the material and changes in the conditions of the drying air over a wide range of design and operating parameters. A characteristic case study of industrial drying of plums in a tunnel dryer is presented to illustrate the application of the modelling approach to determine the optimal design and operating conditions at industrial scale. Fresh plums of d Agen variety obtained from Cavaso Farming P/L (Darlington Point, NSW, Australia) were used in the laboratory drying experiments. Fruit weighing between 10 and 20 gm, which represent the average size, were utilised. Prior to each laboratory drying experiment, fruit samples were allowed to equilibrate under room conditions. Approximately 1 kg of fruit samples for each run were spread uniformly in a single layer on the tray and then loaded into the drying chamber after the desired drying conditions had stabilised. Each drying run was carried out until the moisture content of the fruit reached about 18-20% (consistent with commercially dried prunes). All drying experiments were replicated at least twice. The initial moisture content of the fruit samples was determined using a standard AOAC method by vacuum drying over magnesium perchlorate desiccant (AOAC, 1995). 2.2 Mathematical Modeling 2
A 2-dimensional axis-symmetric model was developed to describe the simultaneous transfer of momentum (air only), heat and mass (air and food) occurring in convective air drying of fruits (e.g., plums). In this model, fruit is viewed as a composite ellipsoidal body, comprising of two materials (flesh and stone) having different properties. The governing partial differential equations (PDEs) describing the simultaneous transfer of heat, mass and momentum in two distinct sub-domains (air and food) during drying of plums were presented in previous studies (Sabarez, 2012; Sabarez, 2010). The non-isothermal turbulent flow of air in the drying chamber is described according to the standard k-ɛ model (COMSOL Multiphysics TM, 2007). The energy balance in the food material leads to the transient heat transfer equation according to Fourier s law of heat conduction (equation 1), while the energy balance in the drying air, taking into account for both convective and conductive contributions is given in equation 2: (1) (2) where T is food temperature ( C), T2 is air temperature ( C), RHO is density (kg/m 3 ), Cp is specific heat capacity (J/kg.K), k is thermal conductivity (W/m.K), is velocity (m/s), and t is time (s). The transient moisture transport within the food matrix was modelled using the basic law governing the movement of moisture according to Fick s law of diffusion (equation 3), while the water mass balance in the drying air, taking into account both convective and diffusive contributions is given in equation 4: (3) (4) where c is water concentration in food (mol/m3), c2 is water concentration in air (mol/m 3 ), and D is effective water diffusivity in food (m 2 /s). The boundary conditions used to formulate the mathematical model were described by Sabarez (2012). In particular, the boundary condition at the air/food interface (at t>0) for heat transfer, considering the mass transfer at the air/food interface, thus coupling the heat and mass transfer equations simultaneously, is given in equation 6. This means that the heat transported by convection and conduction from the drying air to the food is partly used to raise the food temperature by conduction and partly for water evaporation at the food surface: (6) where kc is mass transfer coefficient (m/s), hc is heat transfer coefficient (W/m 2.K), cs is water concentration at food surface (mol/m 3 ), Ts is food surface temperature ( 2 ), and Îż is latent heat of evaporation (J/kg). The boundary condition at the air/food interface for mass transfer is given in equation 7, which accounts for the balance between the diffusive flux of liquid water coming from the interior of the product and the flux of vapour from the food surface to the drying air: (7) The resulting systems of highly coupled non-linear PDEs in the space-time domain together with the set of initial and boundary conditions were numerically solved by the finite element method (FEM) coupled to the Arbitrary Lagrange-Eulerian (ALE) procedure to account for the shrinkage phenomenon using the commercial software package (COMSOL Multi-physicsTM, Comsol AB, Stockholm, Sweden). The details of the numerical solution are presented in previous studies (Sabarez, 2012; Sabarez, 2010). Also, the solution of the governing PDEs requires knowledge of the thermophysical and transport properties of the prod- 3
uct and air. The model parameters used in this work are given by Sabarez (2012). II. Results and discussion 3.1 Model validation The predicted results generated by the model need to be tested to verify the validity of the proposed model to represent the real systems. Fig. 3 shows a typical example of the comparison between the drying curves predicted by the model and the experimental drying tests performed at different air velocity levels (Sabarez, 2012). Generally, the measured and predicted data at different timesteps of the process banded closely around the straight line in the parity plot, which indicates the suitability of the model in describing the drying behaviour of the material under the drying conditions tested. Also, Sabarez (2012) presented further validations to verify the predictive capability of the model over a range of conditions. The results confirm the validity of the model and demonstrate that the parameters used in the model are reasonable, indicating the suitability of the model to describe the drying process of plums under various conditions. In general, the significant differences in drying kinetics between levels of the drying conditions investigated demonstrate the importance of optimising the drying conditions during industrial drying of fruits. The advantage of the proposed numerical model is that the temperature and moisture distributions across the solid food domain as well the changes of the condition of the drying air with location can be established at any time during drying. It should be noted that the modelling of the drying system comprises both material and equipment models, in which the material model describes the drying kinetics and the equipment model determines the changes of the condition of the drying medium with time and space during drying. Together, these models constitute a complete modelling tool capable of predicting the dynamic behaviour of the drying system. Thus, the prediction of the drying air stream conditions flowing across the product surface which would affect the drying behaviour of the solid product at any time and position in the dryer is of particular importance in simulating the drying process of industrial drying systems where a systematic dynamic variation in drying conditions is typical (Sabarez, 2012). 3.2 Tunnel drying simulation The industrial drying system examined here is a tunnel dryer for fruits (e.g., plums) with trucks and trays as illustrated in Fig.1. The industrial tunnel dryers are commonly operated in either counter-flow or parallel-flow mode of operation. In a counter-flow configuration, the drying air is introduced into one end of the tunnel while the trolleys of food products enter at the other end and each moves in opposite directions. This configuration is characterised by having conditions most conducive to intense heating at the end of the drying cycle when the product is nearly dry and less heating in the early stages. The operation of the parallel-flow tunnel is opposite to that of the counter-flow. The trolleys and drying air enter at the same end of the tunnel and progress through the tunnel in the same direction. This configuration is characterised by intense heating in the early stages where the product to be dried is still very wet followed by slow drying as the product approaches the cooler end. The results presented by Sabarez (2012) on the simulation of industrial drying of plums both in counter-flow and parallel-flow modes of operation have shown that the parallel-flow operation would apparently result in shorter overall 4
drying time to reach the desired final moisture content (20%) compared with the counter-flow operation. The predicted overall drying times were also found to be in close agreement with those obtained in the industrial drying trials undertaken under similar drying conditions (Sabarez, 2010). Parson (1968) reported a 33% shorter drying time with less fuel consumed per tonne of prunes in parallel-flow drying of prunes compared to the counter-flow operation. In tunnel dryers, the drying operation involves complex conditions interconnected with many factors associated in the design features and operational practices. The conditions of the drying air (i.e., airflow, temperature and relative humidity) are considered to be the main factors influencing the drying performance in tunnel dryers. In this paper, the effect of different air velocity levels was taken as an example to demonstrate the impact of this parameter on the drying kinetics and energy consumption. It is well known that the air velocity field greatly influences the heat and mass transfer rates at food/air interfaces. Therefore, the temperature and concentration of moisture in the product and the drying air are basically controlled by the level of air velocity and its distribution. A number of simulation runs were undertaken to mimic the industrial tunnel drying of plums specifically in a parallel-flow mode of operation at various inlet air velocity conditions. In these simulations, the drying inlet air temperature was kept constant at 85 C then decreased progressively with drying time towards the exit section of the tunnel. Correspondingly, the drying inlet air relative humidity was set to 15% then increased as drying progresses until the exit section of the tunnel. The selected conditions are representative for current commercial tunnel drying operation for plums. Fig. 4 depicts the simulated effect of different levels of air velocity on both drying time and energy consumption. Clearly, it shows that as the air velocity increases the energy consumption appears to increase. This is obvious as increases in air volume would result in increased energy requirement to heat the large volume of air to the desired temperature level. On the other hand, it appears that the drying time significantly decreases as the air velocity increases but only to certain point and then beyond this level the air velocity plays a proportionally decreasing role in reducing the drying time. An interesting feature that can be observed from this plot is that there appears to be an optimum level of air velocity to achieve better drying performance, which can be found at the intersection of the plots. Under these conditions, it can be seen that the optimum air velocity level is around 4 to 5 m/s. Hayashi (2007) suggested an air velocity of about 5 m/s as sufficient level in drying most products. The result indicates that further increase in air velocity beyond this level would significantly increase the energy consumption with minimal increase in throughput (i.e., slight reduction in drying time). For example, increasing the air velocity from 5 to 7 m/s would increase the energy consumption by as much as 26% without gaining any significant reduction in drying time. Consequently, there is little to be gained by using a very high air velocity. If the air velocity is too high, the costs of energy required to heat the excessive air would tend to offset the benefits of slight reduction in drying time. On the other hand, it shows that large savings in energy could be achieved if operating at lower air velocities, however with the expense of longer drying times. Apart from reduced throughput, operating at longer drying times would also increase the labour costs associated in drying and possibly affect 5
the product quality due to prolonged exposure times. Also, one of the major issues in tunnel drying is the non-uniformity in drying conditions (i.e., temperature and humidity), which is more apparent for large industrial drying systems depending on the air velocity field. If the velocity is too low, convection currents and other disturbances will cause wide variations in temperature and relative humidity across the system, resulting in uneven drying. The findings from the industrial drying trials revealed considerable non-uniformities in the conditions of the drying air (i.e., temperature and relative humidity) measured across the cross-section and along the length of the drying tunnel, particularly at low air velocity (Sabarez, 2010). III. Conclusions A numerical model describing the simultaneous transport phenomena occurring during drying of plums was developed and tested. Predicted data were found to be in good agreement with those determined experimentally. The successful validation of the simulation model under various conditions against the experimental data suggests that the predictive tool can be used with confidence for the prediction of important process design and conditions during tunnel drying of fruits. The present work demonstrated the importance of the modelling approach to identify the key operational parameters affecting the drying process in relation to the optimal design and operation of an industrial tunnel drying system for fruits. It was shown that the air velocity of about 5 m/s is the sufficient level beyond which the energy consumption would increase significantly with minimal improvement in throughput. Apart from reduced throughput and increased non-uniformities in drying conditions, operating at air velocities below this level would increase the labour costs associated in drying and possibly affect the product quality due to prolonged exposure times. IV. References [1] A.O.A.C. 1995. AOAC Official Methods of Analysis. Arlington, Virginia. [2] COMSOL Multi-physics. 2007. Chemical Engineering Module Model Library. COMSOL AB, Stockholm, Sweden. [3] Hayashi, E.M. 2007. Afghanistan fruit and vegetable dehydrator desk study. Report for the United States Agency for International Development (USAID) Accelerating Sustainable Agricultural Program (ASAP). [4] Parson, R.A. 1968. Parallel-flow drying. Agricultural Extension Service. University of California, USA. [5] Sabarez, H.T. 2012. Computational modelling of the transport phenomena occurring during convective drying of prunes. Journal of Food Engineering, 111(2): 279-288. [6] Sabarez, H.T. 2010. Improving prune dehydration cost efficiency. Unpublished report prepared by CSIRO Animal, Food & Health Sciences for Horticulture Australia Limited (HAL), Australia. 6