An Experimental and Modeling Investigation on Drying of Ragi (Eleusine corocana) in Fluidized Bed

Transcription

1 Drying Technology, 24: , 2006 Copyright # 2006 Taylor & Francis Group, LLC ISSN: print/ online DOI: / An Experimental and Modeling Investigation on Drying of Ragi (Eleusine corocana) in Fluidized Bed C. Srinivasakannan 1 and N. Balasubramaniam 2 1 Curtin University of Technology, Sarawak Campus Malaysia, Miri, Sarawak, Malaysia 2 AC College of Technology, Anna University, Chennai, India Experimental investigation on drying of ragi (Eleusine corocana) in a fluidized bed has been attempted covering the operating parameters such as temperature, flow rate of the drying medium, and solids holdup. The drying rate was found to increase significantly with increase in temperature and marginally with flow rate of the heating medium and to decrease with increase in solids holdup. The duration of constant rate period was found to be insignificant, considering the total duration of drying and the entire drying period was considered to follow falling rate period. The drying rate was compared with various simple exponential time decay models and the model parameters were evaluated. The Page model was found to match the experimental data very closely with the maximum root mean square of error (RMSE) less than 2.5%. The experimental data were also modeled using Fick s diffusion equation and the effective diffusivity coefficients were estimated. The effective diffusion coefficient was found to be within 5.7 to m 2 /s for the range of experimental data covered in the present study with RMSE less than 5%. Keywords Drying kinetics; Fluidized bed; Food grain drying INTRODUCTION Ragi is one of the principal cereal crops for many peoples in India, Sri Lanka, and East Africa, in India, over 2.5 million hectares are cultivated annually. Although it does not enter international markets, it is a very important cereal grain in areas of adaptation. Grain is higher in protein, fat, and minerals than rice, corn, or sorghum. [1] It is usually converted into flour and made into cakes, puddings, or porridge. When consumed as food it provides a sustainable diet, especially for people doing hard work. Grain may also be malted and flour of the malted grain is used as a nourishing food for infants. Ragi is considered an especially wholesome food for diabetics. Earheads are gathered when they ripen; three or four pickings are usually required to collect all earheads from a field. The practice among farmers is to stack the just Correspondence: C. Srinivasakannan, Curtin University of Technology, Sarawak Campus Malaysia, CDT 250, Miri, Sarawak, Malaysia. chandrasekar.s@curtin.edu.my cut ragi until the cold weather gives way to sunshine. By the time ragi is threshed, it starts developing mold. Grains that are not sun-dried thoroughly have high moisture content and are prone to accumulate mold and fungus. The grain so produced suffers from microbial contamination, which affects the health of the poor, who consume such grains. A properly dried ragi grain can be stored up to 50 years, which points to the importance of drying. Fluidized beds are widely used for the drying of granular solids such as grains, fertilizers, chemicals, pharmaceuticals, and minerals. This technique offers advantages such as the high heat capacity of the bed, improved rates of heat and mass transfer between the phases, and ease in handling and transport of fluidized solids. The drying rates in fluidized beds are strongly influenced by the characteristics of the material and the conditions of fluidization. Drying of solids is generally understood to follow two distinct drying zones known as constant rate period and falling rate period demarcated with critical moisture content. The critical moisture content is reported to vary with operating parameters and with the type of drying equipment. The constant rate period is understood to have maximum drying rate, which remains constant until the critical moisture content with the resistance for moisture transfer in the gas phase. The rate of diffusion of moisture to surface of solids becomes the limiting factor for moisture transfer as far as the falling rate period is concerned. The extent of drying zones are decided based on the type of material, with materials like sand, ion exchange resin, glass beads, etc., reported to have larger duration of constant rate period and short linear falling rate period while the fibrous grains such as mustard, pepper, ragi, poppy seeds, etc., are reported to have a very short duration constant rate period and a longer curvilinear falling rate period. [2] Knowledge of drying kinetics is essential for sizing the dryer as well as for choosing the optimal drying conditions. The complex hydrodynamics and process calculations are material and dryer specific and hence numerous mathematical 1683

2 1684 SRINIVASAKANNAN AND BALASUBRAMANIAM TABLE 1 List of various simple models tested with the drying data of the present study Name of model Model equation Newton model (NM) MR ¼ exp ( kt) Page model (PM) MR ¼ exp ( kt n ) Henderson and Pabis (HPB) MR ¼ a exp ( kt) Two-term exponential model (TEM) MR ¼ a exp ( kt) þ (1 a) exp ( kat) Approximate diffusion model (ADM) MR ¼ a exp ( kt) þ (1 a) exp ( kbt) models have been developed to estimate the drying kinetics. These range from analytical models solved with a variety of simplified assumptions to purely empirical models, often built by regression of experimental data. In general, the drying rate in constant rate period in fluidized bed drying is modeled using (a) simple empirical correlation relating drying rate to the influencing parameters or utilizing heat=mass transfer coefficient between solids and gas in fluidized bed [3 7] or (b) using mass transfer models, assuming the bed to made of bubble phase, emulsion phase, and a dense phase with the exchange of mass and energy between these phases. [8 14] Similarly, drying kinetics in falling rate period is modeled with complex models, which serve the purpose of improving the fundamental understanding. However, these models may not serve for practical applications in a straightforward manner, due to their complexity. [15] Simple models that can be used to design drying system are much sought after to provide an optimum solution to different aspects of drying operation, with a minimum number of parameters. A series of simple models based on exponential time decay were developed in the past and are being continuously revised=improved; these are popularly known as the Newton model, the Page model, the Henderson and Pabis model, the two-term exponential model and the approximate diffusion model. These simple models are recently utilized for drying applications by Mujumdar, [16] Diamante and Munro, [17] Zhang and Litchfield, [18] Henderson, [19] Yaldeiz and Ertekin, [20] and Sharaf-Eldeen et al. [21] to represent the drying kinetics. Different approaches were reported for complex models, most common among them is based on second-order partial differential equation, commonly known as Fick s diffusion equation. The solution of partial differential equation differs based on the boundary conditions, often requiring numerical computations to estimate the drying rate. [7,14,22 25] The objective of the present study is to experimentally investigate the drying kinetics of ragi grain in a fluidized bed with respect to the operating parameters such as the temperature, flow rate of the drying medium, and the solids holdup. Although the effect of operating parameters on the drying rates are well known and one expects the influencing parameters to respond in similar fashion qualitatively, the drying kinetics can vary quantitatively depending on the nature of the material and the drying conditions. It is further attempted to verify the compatibility of experimental drying kinetics with various simple models reported in the literature (Table 1), and with complex models such as Fick s diffusion equation. The model parameters are estimated by minimizing the root mean square of error (RMSE) between the experimental drying rate and the model prediction (Tables 2 and 3). EXPERIMENTAL SECTION Drying experiments were conducted using fluidized columns of m in internal diameter with a height of 1.2 m. The gas distributor was 2 mm thick with 2 mm perforations having 13% free area. A fine wire mesh was spot welded over the distributor plate to arrest the flow of solids from the fluidized bed in to the air chamber. Air from the blower was heated and fed to the fluidization column through the air chamber. The electrical heater consisted of a multiple heating element each of 2 KW rating. A temperature controller, provided to the air chamber, facilitated control of air temperature within 5 C of the set temperature for the entire operating range of 30 to 110 C. Air flow was measured using a calibrated orifice meter. Table 4 shows the physical characters of ragi grain as well as the experimental conditions covered in the present study. A good fluidization behavior in terms of perfect mixing of the bed material was observed visibly. This was substantiated with low fluctuation in the bed pressure drop, which is an indication for smooth fluidization without formation of slugs. The minimum fluidization velocity was not found to vary with the temperature within the range of temperatures covered in the present study. A known quantity of ragi with known initial moisture content was taken in the batch fluidized bed, and air at the desired rate was introduced into the column. As fluidization continued, ragi samples were scooped out of the bed at regular intervals of time for moisture content estimation. The sample collector that had an approximate capacity of five grams was utilized to scoop out the samples from the fluidized bed. However, after transferring about 1 g to the sample holder for moisture analysis, the rest of it was transferred back to the fluidized bed immediately. The ragi

3 DRYING OF RAGI IN A FLUIDIZED BED 1685 TABLE 2 Evaluated model parameters at various operating conditions NM PM HPB TEM ADM T ( C) W (kg) U (m=s) k 10 3 RMSE k 10 3 n 10 RMSE a 10 k 10 3 RMSE a 10 k 10 3 RMSE a 10 k 10 b 10 3 RMSE moisture content was determined by drying the samples to constant weight in an air oven at 105 C. The moisture contents were expressed on dry basis as kilograms of moisture per kilogram of dry solid. Ragi as received from the farm, with 41.2% moisture content, was used for all drying experiments. The experimental data was checked for reproducibility and were found to deviate within 3%. The equilibrium moisture content was estimated by keeping the samples in an air oven at the desired temperatures until no further weight change. The difference in weight between a bone-dry sample and the weight of the sample obtained at the desired temperature was utilized to calculate the equilibrium moisture content. RESULTS AND DISCUSSION Experimental data showing the effect of temperature, flow rate of the heating medium and solids holdup are shown as plots of C=C i versus time, in Figs. 1 to 4. Figures 1 through 4 show the drying rate decreasing, from the starting (t ¼ 0) until the end of drying, indicating the absence of constant drying rate period or presence of constant rate period for an insignificant period of time compared to the total drying time. The rate of drying is higher at the early stage of drying while the moisture content was high and reduces as the moisture content decreases. Figures 1 and 2 show the effect of temperature of the heating medium at two different solids holdup. An increase in temperature of the heating medium increases the drying rate and it can be attributed to the higher bed temperature of particles in the bed, which increases the intra particle moisture diffusion to the surface of the solid resulting in a higher drying rate. Figure 3 show a marginal increase in drying rate with air flow rate and it may be attributed to a reduction in external mass transfer resistance during early stages of drying while the drying rate and the moisture content is high. Looking at the drying curve, as the rate of drying reduces from start until the end of drying period one would expect the entire operation to be an internal mass transfer controlled and would expect a negligible effect of air flow rate on the drying rate. However, a continuous recording of the bed temperature indicated that the effective bed temperature increased with flow rate of the heating medium, which increases the moisture diffusion rate and thereby results in higher drying rate. Repeat experiments were conducted to eliminate the effect of experimental error on assessing the effect of air flow rate on the drying rate and the conclusion of a marginal effect on drying rate was based on the consistent observation on two sets of a experiments conducted with two different solids holdups. An increase in the solids holdup is found to decrease the drying rate (Fig. 4) and it can again be attributed to the lower effective bed temperature at higher solids holdup. The lower effective temperature may be due to higher water content in the bed due to increased solids holdup. This indirectly

4 1686 SRINIVASAKANNAN AND BALASUBRAMANIAM TABLE 3 Evaluated effective diffusivity coefficient at various operating conditions T ( C) W (kg) U (m=s) C e (kg=kg) D eff (m 2 =s) RMSE reduces the rate of moisture diffusion from inside solid to the surface of the solid, resulting in a reduced drying rate. All three observations are in qualitative agreement with most of the earlier observations reported in the literature. [7,14,26,27] However, Topuz et al. [14] reported a reduction in drying rate with increase in flow rate of the heating medium, which was attributed to poor contact between the solid and gas phase, due to spouting of the bed at higher flow rate. The experimental drying data were converted to dimensionless moisture ratio, MR ¼ C C e =C i C e, for the sake of comparison with the various models. The list of simple exponential time decay models, popularly known as the Newton model, the Page model, the Henderson and Pabis model, the approximation of diffusion model, the two term exponential model, as listed in Table 4, were compared with the experimental data. The experimental drying rates were fitted with various model equations, by minimizing the root mean square error (RMSE) between the experimental drying rate and the model equation. The RMSE is defined as It is also attempted to model the experimental data with a fundamental diffusion equation for moisture distribution within the solid particle with appropriate boundary conditions. The model assumes that the moisture diffuses from inside the particle to the surface of the particle and evaporates at the surface and that all the particles are uniform in size and spherical in shape. The fluidized beds are perfectly mixed beds, and the solids at any point in the bed are exposed to same drying conditions. The general form of the diffusion equation known as Fick s diffusion equation is the boundary conditions are " # dc dt ¼ D d 2 C eff dr 2 þ 2 dc r dr ð2þ " RMSE ¼ 1 X # n 0:5 ðmr pre;i MR exp;i Þ 2 N 100 ð1þ i¼1 The evaluated model parameters along with the RMSE values are listed in Table 2. It can be seen from Table 2 that among all the models, the most simple among all, the Page model is found to match experimental data very closely, with the RMSE error less than 2.5%. The standard deviation between the experimental data and the model prediction using the page model parameters is less than 6%. Although three parameters are used in approximate diffusion model, the RMSE values are much higher than the Page model. The model parameters can be utilized to estimate the drying time as well as for designing and scale up of the drying process. FIG. 1. Effect of temperature of the heating medium (U: 1.6 m/s, W: 0.3 kg, ^ T: 60 C, ~ T: 80 C, & T: 100 C).

5 DRYING OF RAGI IN A FLUIDIZED BED 1687 FIG. 2. Effect of temperature of the heating medium (U: 1.6 m/s, W: 0.6 kg, ^ T: 60 C, & T: 80 C, ~ T: 100 C). at t ¼ 0; 0 < r < R s ; C ¼ C i at t > 0; r ¼ 0; dc=dr ¼ 0 at t > 0; r ¼ R s ; D eff ðdc=drþ ¼KðC sj C be Þ where C sj is the moisture concentration just within the sphere and C be is the concentration required to maintain equilibrium with the surrounding atmosphere. Analytical solution to Eq. (2) for the above boundary conditions was provided by Crank [28] as given below: C C e C i C e ¼ X1 n¼1 6Bi 2 m expð b2 n D efft=r 2 s Þ b 2 n ðb2 n þ Bi mðbi m 1ÞÞ FIG. 3. Effect of flow rate of the heating medium (T: 80 C, W: 0.6 kg,. U: 1.6 m/s, ~ U: 1.2 m/s). ð3þ FIG. 4. Effect of solids holdup (T: 80 C, U: 1.6 m/s, ~ W: 0.60 kg, ^ W: 0.30 kg, & W: 0.15 kg). where b n are the roots of the equation b n cot b n þ Bi m 1 ¼ 0 The mass Biot number (Bi m ) is defined as KR s =D eff and the mass transfer coefficient (K) is calculated based on the equation due to Richardson and Szekely [29] as given below: Sh ¼ Kd p D 0:374 Re1:16 for 0:1 < Re < 15 ¼ 2:01 Re 0:5 for 15 < Re < 250 Sherwood number is the ratio of external mass transfer resistance to the molecular diffusivity, while Biot number is the ratio of external mass transfer resistance to the overall mass transfer resistance. The evaluated effective diffusivities are reported in Table 3 along with the RMSE values. The effective diffusivity is found to significantly increase with increase in temperature of the heating medium, while a marginal increase is registered with increase in the flow rate of the heating medium. Since the effect of flow rate on effective diffusivity being marginal, with significant RMSE values, statistically the optimized effective diffusivity will have an overlapping range, varying with the degree of confidence. Although the variations are statistically not significant, the authors would like to emphasize that the air flow rate contributes to a marginal increase in the drying rate and similar observations have been reported in the literature. The increase or decrease in effective diffusivity coefficient is according to the increase or decrease in drying rate. An increase in the solids holdup is found to decrease the effective diffusivity coefficient. The increase in drying rate with flow rate and decreased solids holdup has been attributed to the increase in effective bed temperature. The effective diffusivity is found to vary within 5.7 to m 2 =swith RMSE less than 5%. Although the errors are higher, these kinetic parameters are very essential in the design and scale ð4þ ð5þ

6 1688 SRINIVASAKANNAN AND BALASUBRAMANIAM TABLE 4 Characteristics of the material and the range of experimental parameters covered in the present study Name of material Shape of material Ragi (Eleusine corocana) Spherical Size, d p 10 3 (m) 1.48 Particle density (kg=m 3 ) 1200 Minimum fluidization velocity, U mf (m=s) 0.47 Terminal velocity, U t (m=s) 6.9 Temperature of fluidizing air ( C) 60, 80, 100 Fluidizing air velocity (m=s) 1.2, 1.6 Solid holdup (kg) 0.15, 0.30, 0.60 up of drying process with certain order of magnitude. Uckan and Ulku [24] have reported an effective diffusivity of 2.1 to m 2 =s for drying of corn in fluidized bed and Kundu et al. [23], have summarized the effective diffusion coefficient estimated for different types of grains using a modified Fick s diffusion equation applied mostly to stationary beds reported a diffusion coefficient of 11 to m 2 =s for corn; for wheat, 5.5 to m 2 =s, and for parboiled rice, 4.1 to m 2 =s. The estimated effective diffusion coefficient in the present study is on the same order of magnitude reported in the literature. Although the simple models could closely match the experimental data much better than the complex model, they are more empirical in nature and lack scientific background, restricting their applicability to within the experimental range covered in the present study. However, the fundamental models, although they have a higher error rate, can be extended even beyond the experimental range in the present study with a certain degree of confidence. CONCLUSION The drying characteristics of ragi, one of the popular cereal crops in India and Srilanka, have been assessed in a fluidized bed dryer with respect to the various operating variables. The drying rate was found to increase significantly with increase in temperature and marginally with flow rate of the heating medium, and to decrease with increase in solids holdup. The duration of the constant rate period was found to be insignificant, considering the total duration of drying and the entire drying period was considered to follow falling rate period. The kinetics of drying was tested with various simple exponential decay models and the Page model was found to match the experimental drying rate closely with the RMSE value less than 2.5%. The experimental data were also modeled using more fundamental Fick s diffusion equation and the effective diffusivity coefficient was estimated to be within 5.7 to m 2 =s for the range of experimental data covered in the present study with RMSE less than 5%. The estimated effective diffusion coefficient is compared with the literature reported effective diffusion coefficient for other grains and found to be within the same order of magnitude. NOMENCLATURE Bi m Biot number (K R s =D eff ) C Moisture content of ragi grain at any time (kg of moisture=kg of dry solid) C e Equilibrium moisture content of ragi grain (kg of moisture=kg of solid) C i Initial moisture content of ragi grain (kg of moisture=kg of dry solid) D Molecular diffusivity of moisture in air (m 2 =s) D eff Effective diffusion coefficient (m 2 =s) d p Particle diameter (m) K Mass transfer coefficient across particle surface (m=s) MR Moisture ratio C C e C i C e Re Reynolds number (d p Uq=m) R s Particle radius (m) r Radial coordinate (m) Sh Sherwood number (K d p =D) T Temperature of heating medium ( C) t Time (s) U Superficial velocity of heating medium (m=s) W Solids holdup (kg) Greek Symbols l Viscosity of air (kg=ms) q Density of air (kg=m 3 ) REFERENCES 1. Bogdan, A.V. Tropical Pasture and Fodder Plants; Longman: London, Strumillo, C.; Kudra, T. Drying: Principles, Applications and Design; Gordon and Breach: New York, Syromyatnikov, N.I.; Vasanova, P.K.; Shimanskii, D.N. Heat and Mass Transfer in Fluidised Bed; Khimiya: Moscow, Kunni, D.; Levenspiel, O. Fluidization Engineering; Wiley: New York, 1969.

7 DRYING OF RAGI IN A FLUIDIZED BED Anantharaman, N.; Ibrahim, S.H. Fluidised bed drying of pulses and cereals. In Recent Advances in Particulate Science and Technology, Ibraham, S. H., Ed.; Madras, Chandran, A.N.; Subbarao, S.; Varma, Y.B.G. Fluidised bed drying of solids. AIChE Journal 1990, 36 (1), Thomas, P.P. Drying of solids in beach and continuous fluidized beds, Ph.D thesis, Indian Institute of Technology, Madras, Hoebink, J.H.B.J.; Rietema, K. Drying of granular solids in a fluidised bed I. Chemical Engineering Science 1980, 35, Hoebink, J.H.B.J.; Rietema, K. Drying of granular solids in a fluidised bed II. Chemical Engineering Science 1980, 35, Alebregtse, J.B. Fluidised bed drying: A mathematical model for hydrodynamics and mass transfer. In Heat and Mass Transfer in Fixed and Fluidised Bed; Van Swaajj, W.P.M., Afgan, N.H., Eds.; Hemisphere Publishing Corp.: New York, 1986; Palancz, B.A. Mathematical model for continuous fluidised bed drying. Chemical Engineering Science 1983, 38 (7), Parti, M. Evaluation of selected mathematical model for grain drying. In Drying 91; Mujumdar, A.S., Filkova, I., Eds.; Elsevier: Amsterdam, 1991; Kafarov, V.V.; Dorokhov, I.N. Modeling and optimisation of drying process. International Chemical Engineering 1992, 32 (3), Topuz, A.; Gur, M.; Gul, M.Z. An experimental and numerical study of fluidized bed drying of hazelnut. Applied Thermal Engineering 2004, 24, Keey, R.B. Drying of Loose and Particulate Materials; Hemisphere Publishing Corporation: New York, Mujumdar, A.S. Handbook of Industrial drying; Marcel Dekker: New York, Diamante, L.M.; Munro, P.A. Mathematical modeling of the thin layer solar drying of sweet potato slices. Solar Energy 1993, 51, Zhang, Q.; Litchfield, J.B. An optimization of intermittent corn drying in a laboratory scale thin layer dryer. Drying Technology 1991, 9, Henderson, S.M. Progress in developing the thin layer drying equation. Transactions of the ASAE 1974, 17, Yaldiz, O.; Ertekin, C.; Uzun, H.I. Mathematical modeling of thin layer solar drying of sultana grapes. Energy 2001, 26, Sharaf-Eldeen, Y.I.; Blaisdell, J.L.; Hamdy, M.Y. A model for ear corn drying. Transactions of the ASME 1980, 23, Hajidavalloo, E.; Hamdullahpur, F. Mathematical modeling of simultaneous heat and mass transfer in fluidized bed drying of large particles. Transaction of the CSME, 1999, 23, Kundu, K.M.; Das, R.; Datta, A.B.; Chatterjee, P.K. On analysis of drying process. Drying Technology 2005, 23, Ukan, G.; Ulku, S. Drying of corn grains in batch fluidized bed. In Drying of Solids; Mujumdar, A.S., Ed.; Wiley Eastern Limited: New Delhi, 1986; Feng, H. Analysis of microwave assisted fluidized bed drying of particulate product with simplified heat and mass transfer model. International Communications in Heat and Mass Transfer 2002, 29 (8), Syahrul, S.; Hamdullahpur, F.; Dicer, I. Energy analysis of fluidized bed drying of moist particles. Energy, an International Journal 2002, 2, Kudras, T.; Efremov, G.I. A quasi-stationary approach to drying kinetics in fluidized particulate materials. Drying Technology 2003, 21 (6), Crank, J. The Mathematics of Diffusion; Oxford University Press: Oxford, Richardson, J.F.; Szekely, J. Mass transfer in a fluidized bed. Transactions of the Institution of Chemical Engineering 1961, 39,