EFFECTS OF NON-UNIFORM AIRFLOW DISTRIBUTION ON GRAIN MOISTURE CONTENTS DURING AERATION Desmnd Essien, Richard Bani and Edward Sabi Department f Agricultural Engineering, University f Ghana, Legn, Ghana E-Mail: desmndessien@gmail.cm ABSTRACT Maintaining the quality f grains in strage ver lng perids f time is dependent n several factrs, including the prvisin f a functinal aeratin system and an adequate management strategy. The grwth and activities f insects in stred grains is a functin f time, grain misture cntent and grain temperature, but this can be cntrlled with effective aeratin. T ensure effective aeratin the cnditins f the grains in strage must be mnitred. Thugh several mdels exist fr the predictin f grain misture cntents during aeratin, mst f these mdels assume airflw t be unifrm during aeratin. Thmpsn s mdel which is cmmnly used fr predicting grain misture cntent assumes unifrm airflw during aeratin. Cnversely, airflw is nn-unifrm in hpper bttm sils and partially perfrated flrs. Therefre the bjective f this study is t mdify Thmpsn s mdel used fr natural drying s it culd be used t predict the misture cntents f grains during aeratin when nn-unifrm flw f air is cnsidered. A hpper bttm sil f 3m radius filled t a grain height f 1.8m was used fr this study. Fur tests, each test under different testing cnditins and lasting 120 hurs, were carried ut t investigate the nn-unifrm mvement f air. The investigatins revealed that the mdified mdel presented in this study is useful fr predicting misture cntent t within 0.5% f measured misture cntent when nn-unifrm airflw is cnsidered. The cefficient f determinatin (R 2 ) between the measured and predicted values btained fr the misture tests ranged frm 0.97 t 0.98. Keywrds: grain, aeratin, nn-unifrm, hpper bttm. INTRODUCTION Airflw distributins in mst grain structures is nn-unifrm due t the variatins in material prperties f the grain mass, the gemetry f the strage structure and/r the design f the aeratin system. Airflw is generally assumed t be unifrm in sils with fully perfrated flrs and nn-unifrm in sils with aeratin ducts, pads r partially perfrated flrs. The airflw distributin can als be nn unifrm in a sil with hpper bttm, peaked grain frm verfilling, inverted grain frm partial uplading and high fine material cncentratin in the cre f the grain mass (Bartsik and Maier, 2006). Accrding t Garg and Maier (2006), ne f the primary causes f nn unifrm airflw distributin is variatin in the material prperties f the grain mass. Airflw resistance is a functin f particle size and prsity f the grain. Therefre a number f material prperties like the distributin f fine material, the lading methd, misture cntent and level cmpactin cause nn unifrm airflw distributins. Several investigatrs have investigated the nnunifrmity f airflw and resistance in a grain mass; hwever the investigatrs fcused n develping mdels t calculate pressure drp and cnsequently pwer cnsumptin by aeratin fans. Little has been reprted n the effects f nn-unifrmity f airflw n misture cntent. Ergun (1952) presented an equatin t calculate the pressure drp f fluid flw based n Reynlds Thery. Accrding t this equatin, the ttal energy lss in a packed bed is the sum f the viscus and kinetic energy lsses. Ergun s equatin has been mdified by several investigatrs t mdel nn-unifrm airflw. Lai (1980) investigated three-dimensinal axisymmetric (3D) nnunifrm airflw in cylindrical grain beds. Accrding t Lai (1980), the prsity f grains differs at different lcatins within the bin and therefre he divided the grain mass int tw regins with different prsities (0.4 at the centre and 0.6 at the periphery). He then used Ergun s equatin t calculate the resistance t airflw. Smith (1996) mdified Ergun s (1952) nn nnlinear mmentum equatin int curvilinear crdinates and used it t predict the pressure drp thrugh a grain mass. Garg and Maier (2006) mdified Ergun s (1952) equatin and inserted it int FLUENT (a Cmputatinal Fluid Dynamics sftware) t mdel nn-unifrm airflw distributin in large sils. The mdified Ergun s equatin was applied t three scenaris: a peaked grain mass, a grain mass with a high fines cncentratin cre, and a grain mass aerated frm a ring duct arund the bttm f the sil wall. They reprted that nn-unifrm airflw resulted in the reductin f air velcity and vlumetric airflw rate within grain layers during aeratin. Hwever, the mdel was nt validated with measured data. Hukill and Shedd (1955) presented a mdel t predict the pressure drp ver an airflw range between 0.01-0.20 m 3 /s m 2. The American Sciety f Agricultural and Bilgical Engineers (ASABE) mdified Shedd's equatin and apprved it fr determining the static pressure drp f airflw thrugh a grain bed (ASABE Standards, 2006) Kay et al. (1989) investigated airflw resistance in crn grains and reprted that the hrizntal airflw resistances thrugh crn grains were abut 58 and 45% f the vertical airflw resistance when airflw rates were abve and belw 0.lm 3 /s m 2, respectively. Garg (2005) investigated the heat and mass transfer due t nn-unifrm airflw distributin in a grain 250
mass in tw dimensins. He used Lai s (1980) variable cncept t develp a nn-unifrm airflw mdel fr stred grain. Tw different prsities, ne fr the centre cre vlume and the secnd fr the periphery were used. He then used FLUENT and 2D PHAST-FEM (Pst-Harvest Aeratin and Strage Simulatin Tl develped at Purdue University) t simulate the heat and mass transfer. Bartsik and Maier (2006) studied the nnunifrm airflw distributin in cred, peaked, and leveled grain mass. They measured airflw at the periphery and centre fr each grain mass within a bin. A nn-unifrmity factr (NUF) which was defined as a functin f the difference in average airflw rate at the centre versus periphery was used t mdel nn-unifrm airflw. Lawrence and Maier (2011) used a 3D airflw mdel t investigate the nn-unifrm airflw in leveled, peaked and inverted cne grain mass. They used tw cnstant prsities (0.38 and 0.40) and three variable prsity ranges (0.34-0.38, 0.36-0.38, and 0.38-0.40) fr crn t validate the airflw distributin. They reprted that fr the variable prsity f 0.34-0.38 the mdel predictins clsely fllwed the experimental results. As air mves upstream in bin during aeratin, initial vlumetric airflw rate f the air is reduced. The reductin in vlumetric airflw rate is largely influenced by the gemetry f the grain bin and the prsity f the grains. The vlumetric airflw rate at the flr entry f the bin differs frm the vlumetric flw rate at ther psitins within the bin especially, when hpper bttm sils and sils with aeratin ducts abve the flr are used. The vlumetric flw rate therefre becmes nn-unifrm at different layers within the grain mass. This study seeks t investigate the effect f reductin in the initial vlumetric airflw rate n grain misture cntents due t nn-unifrm air flw. MATERIALS AND METHODS Data cllectin Fur aeratin tests were cnducted t cllect data fr the validatin f the mdified mdel (Equatins 8). Each test lasted fr a cntinuus perid f 120 hurs. Fr each test, apprximately 7059 Kg (7.059 tnnes) f crn stred in a steel hpper bttm sil lcated at Anlga (in the Vlta Regin f Ghana) was used. The sil was equipped with 0.37kW fan and als with perfrated air distributin ducts f 0.125 m (5 inches) in diameter, arranged in a ring type t a ttal circumferential length f apprximately 6m. This arrangement gives a ttal duct surface area f 1.89 m 2. Perfratins n the ducts were apprximately 40% f the ttal ducts surface area. This gives the perfrated flr area (F A ) f 0.756m 2. The bin was filled t a height f 1.8m f the grain level and the tpmst surface f the grain leveled t avid peaking f grains at the surface. A ttal grain depth f 1.5m was prbed t determine misture cntent and temperature changes. Prbing was carried ut at grain intervals (depth) f 0.5m. The bttm layer was taken as Layer ne, the middle layer as Layer tw and the tp layer as Layer three. After every 24 hurs, grains were sampled fr misture cntent determinatin. The grains were sampled frm five sample pints in each layer accrding t standards presented by Lewer et al., 1994. Initial airflw rates, ranging frm 2 m 3 min -1 m -2 t 5 m 3 min -1 m -2 (0.2 m 3 min -1 tnne -1 t 0.5 m 3 min -1 tnne -1 ) were used fr the fur tests (Table-1). Test Cnditin Nvember 12-16, 2012 Average initial grain misture cntent (%WB) Initial airflw rate (m 3 min -1 m- 2 ) Average initial ambient air temperature ( C) Initial grain temperature ( C) Average initial relative humidity (%) Initial abslute humidity (kg water/kg f dry air) Table-1. Test cnditins under which aeratin was cnducted. Test 1 Test 2 Test 3 Test 4 Nvember 19-23, 2012 Nvember 26-30, 2012 February 7-11, 2013 17 18 19 20 2 3 4 5 27.7 28.4 28 28.6 28.2 29.2 28.9 29.4 77.8 77.7 73.3 78.9 0.0154 0.0154 0.0154 0.0154 Thmpsn s mdel Thmpsn 1972 presented Equatin (1) fr predicting the average misture cntent f grain within a grain layer. Thmpsn s mdel has a lng successful histry f being used t predict misture cntents f grains accurately within a unifrm airflw bin Hwever. Thmpsn s mdel can be mdified t predict the misture cntents f grains when airflw is nn-unifrm. H f ( M Mf) R H = 100 Kg Water/Kg Dry air (1) 251
Frm Equatin (1) M f = M f 100 H = H H H R M = Initial misture cntent (%) M f = Final misture cntent (%) H = Initial abslute humidity (Kg/Kg f dry air) H f = Final abslute humidity (Kg/Kg f dry air) H = Change in abslute humidity (Kg/kg f dry air) R = Dry matter t air rati (Kgm -2 /Kgm -2 ) Mdificatins made t Thmpsn s mdels T calculate the final misture cntent, Thmpsn (1972), used the dry matter t air rati (R) as an input parameter (Equatin 6). This input parameter can be expressed mathematically and mdified t calculate the cntributin f nn-unifrm mvement f air t changes in misture cntent. Mathematically, the dry matter rati can be derived as; ρ X R = dm 60 t V V s Accrding t Al-Yahya (1996), V = 0.0252 ( T+ 460) (1 + 1.6055 H ) 0.063 (5) s Where ρ dm = Bulk density f grain (kg/m 3 ) X = Grain layer thickness (depth) (m) Vs = Specific vlume f air (m 3 /kg) t = Time interval (hurs) V = Vlumetric flw rate m 3 min -1 m -2 T = Initial air temperature ( C) 60 is a cnversin factr fr t, frm hurs t minutes. Nn-Unifrm mvement f air The mathematical relatinship between the initial vlumetric flw rate U a (m 3 min -1 m -2 ) and the effective vlumetric flw rate V (m 3 min -1 m -2 ) within the layers f the grain can be used t calculate the reductin in the initial vlumetric flw rate. The effect f the reductin in the initial vlumetric flw rate n misture cntent f grains during aeratin can then be determined. The mathematical relatinship fr this actin can be derived as: V = Where FA 1 ε. L A U a (2) (3) (4) (6) U a = Initial vlumetric flw rate at the flr/duct entry (m 3 min -1 m -2 ) V = Vlumetric flw rate at a specified pint within the sil ther than entry (m 3 min -1 m -2 ) L A = Area ccupied by grain layer (m 2 ) F A = Perfrated flr area (m 2 ) ε = Grain Prsity (Dimensinless number factr) Each layer f the grain is cnsidered t ccupy a specific area (L A ). Therefre the area available fr the air t mve within the grain in each layer can be btained by multiplying grain layer area (L A ) by the grain prsity (ε ). The Perfrated flr area (F A ) is the ttal pen r perfrated area available n the surface area f the ducts fr the air t emerge frm the ducts int the sil. Duct surface area can be btained by using the Table prvided by Nyes (1967). The perfrated flr area (F A ) can then be calculated by multiplying the percentage f duct perfratins n the duct surface area with the ttal duct surface area. Substituting Equatin (6) int Equatin (4) gives Equatin (7) R fn ρdm X = F A 60 t 1 Ua ε. LA VS Where R fn = Dry matter t air rati in a given layer. Inserting Equatin (7) int Equatin (2) gives M fn (7) 100 H = M ρdm X F A 60 t 1 Ua ε. LA VS (8) The mdified mdel (Equatin 8) can be used t predict the misture cntents within a grain layer when hpper bttm sils and sils with aeratin ducts abve the flr are used. The principles f deep bed simulatin were emplyed t develp a cmputer simulatin prgram fr predicting misture cntents. The simulatin prgram was develped using Visual Basic Editr in Micrsft Excel. Grain thickness (depth) f 0.05m, time increment f 8 hurs, bulk density f 720kg/m 3 accrding t ASABE standards (ASAE, 1998), whiles an average grain mass per simulated layer f 2000kg, prsity f 0.38 and perfrated flr surface area f 0.756m 2 were used as fixed values in the cmputer prgram. During the simulatin prcess, the grain misture cntent was calculated fr each layer in an iterative way. The term which describes the prperties f the aeratin air (the secnd term at the right hand f Equatin 8) was prgrammed t run iteratively. The value frm the iteratin at a particular layer was then subtracted frm the 252
initial misture cntent f the grain at that layer t give the final misture cntent at that layer. The exhaust air frm the previus layer was prgrammed t be the input air fr the subsequent layer. This prcess was repeated until the iteratin reached the grain height f 1.5m. RESULTS AND DISCUSSIONS T determine the crrelatin between predicted values and measured values, a scatter plt was develped fr each test and the cefficient f determinatin (R 2 ) determined. Accrding t Lewer et al. (1994), a very gd mdel has an R 2 value equal t r greater than 0. 95. A single factr analysis f variance (ANOVA) was made fr each test t determine the significance f variatins between the predicted and the measured results. At the cnfidence interval f 95%, the null hypthesis emplyed was that, significant variatins exist between the predicted and measured value If the P value cmputed is less than r equal t 0.05 r if the F value value cmputed is greater than r equal t F critical value (P value 0.05 and F Fcit, respectively) then there is a premise t reject the crrelatin between predicted and measured values must be rejected. Accuracy f predictin was determined by bserving the errr difference. Fr each test, average abslute errr and standard errr was calculated t determine errrs between predicted and measured misture cntents. The average abslute errr was calculated using Equatin (9) while standard errr was calculated using Equatin (10). ( Y Y') AAVR = (9) bs ( Y Y' ) 2 SE. = (10) df Where AAVR = average f the abslute value f the residuals. Y = measured value Y = value predicted by the mdel. bs = number f bservatins. S.E. = standard errr f the residuals. Y = measured value Y = value predicted by the mdel. df = degree f freedm (n-1). Figure-1 thrugh t Figure-4 shw the misture cntent results fr the fur tests. The results frm the predictin equatin are shwn with P attached t their labels whiles the measured results have M attached t their labels. Figure-1. Predicted and Measured Misture Cntents recrded at different grain depths and aeratin hurs during Test 1. 253
Figure-2. Predicted and Measured Misture Cntents recrded at different grain depths and aeratin hurs during Test 2. Figure-3. Predicted and Measured Misture Cntents recrded at different grain depths and aeratin hurs during Test 3. Figure-4. Predicted and Measured Misture Cntents recrded at different grain depths and aeratin hurs during Test 4. 254
The predicted values were numerically lwer than measured values fr all tests at each grain depth and fr all time intervals (Figures 1 t 4). This implies that the mdified mdel (Equatin 18) presented in this wrk remves a greater amunt f misture than the actual aeratin prcess. Hwever, the amunt f misture remved by the mdified mdel is less than that remved by the riginal Thmpsn mdel (Equatin 6). The term (Equatin 6) which was inserted int the riginal Thmpsn (1972) mdel reduced the vlumetric flw rate used fr aeratin at each layer. A reductin in the vlumetric flw rate increased the dry matter t air rati (Equatin 7). Cnsequently, an increase in the dry matter t air rati resulted in lesser amunt f misture being remved. The lesser amunt f misture remved minimized under predictin. The mdified mdel presented in this wrk therefre reduced the extent f under predictin as cmpared t the riginal Thmpsn (1972) mdel. Average abslute errr and the standard errr f the residuals fr all tests were within 0.5%. The analysis f variance made fr each test shws that at the 95% cnfidence interval, there were n significant variatins between the predicted and measured misture cntent. The results btained frm this study agree with the hypthesis made by Garg and Maier (2006) that nn unifrm mtin f air results in reductin f vlumetric airflw rates amng grain layers. CONCLUSIONS AND RECOMMENDATION The mdified Thmpsn mdel (Equatin 8) presented in this wrk is useful fr predicting the misture cntent amng grain layers t within 0.5% f actual misture cntents f grains in a sil when a reductin in the initial airflw rate due t nn-unifrm mtin f air is cnsidered. The mdified mdel presented in this wrk under predicted misture cntent values because rewetting f the grains may have ccurred. It is therefre recmmended that a mdified a term that calculates fr rewetting can be develped in subsequent studies. REFERENCES Al-Yahya A. S. 1996. Mathematical Simulatin Mdel t Predict Strage Time f Grain. Jurnal f King Saud University (Agric Science). 1( 8): 71-78. Ergun S. 1952. Fluid flw thrugh Packed Clumns. Chemical Engineerng Prgress. (48): 89-94. Garg D. 2005. Mdeling Nn-Unifrm Airflw and its applicatin fr Partial Chilled Aeratin using PHAST- FEM. (Unpublished Master s thesis). Purdue University, West Lafayette, Ind. USA Garg D. and Maier D. E. Octber, 2006. Mdeling Nn- Unifrm Airflw Distributin in Large grain sils using Fluent. Paper presented at the 9 th Internatinal Wrking Cnference n Stred Prtectin, Sa Paul, Brazil. Retrieved frm www. spiru.cgahr.ksu.edu/prj/iwcspp/pdf2/9/6159.pdf n 21 st Nvember, 2012. Hukill W. V. and Shedd C. K. 1955. Nnlinear Airflw in Grain Drying. Agricultural Engineering. 36: 462-466. Kay R. L., Bern. C. J. and Hurburgh C. R. Jr. 1989. Hrizntal and Vertical Airflw Resistance f Shelled Crn at Varius Bulk Densities. Transactins f the ASAE. 32(2): 733-736. Lai F. S. 1980. Three-Dimensinal Flw f Air thrugh Nn-Unifrm Grain Beds. Trans. ASAE. 23(3): 729-734. Lawrence J. and Maier D. E. 2011. Three-dimensinal Airflw Distributins in Peaked, Leveled and Cred Grain mass cnfiguratins. Bisystems Eng. 110(3): 321-329. Lewer J., Bridges T. and Bucklin R. 1994. On Farm Drying and Strage Systems. St. Jseph, MI: ASAE. Nyes R. T. 1967. Aeratin fr Safe Grain Strage. West Lafayette: Purdue University-Cperative Extensin Service Publicatins. Smith E. A. 1996. Pressure and Velcity f Air during Drying and Strage f Cereal Grains. Transprt in Prus Media. 23: 197-218 Thmpsn T. L. 1972. Temprary Strage f High Misture Shelled Crn Using Cntinuus Aeratin. Transactins f the ASAE. 15(2): 333-337. ASABE Standards. 2006. Resistance t Airflw f Grains, Seeds, Other Agricultural Prducts and Perfrated Metal Sheets. St. Jseph, Michigan: ASABE. ASAE Standards. 1998. Density, Specific Gravity, and Mass-Misture Relatinships f Grain fr Strage. St. Jseph, Michigan: ASAE. Bartsik R. E and Maier D. E. 2006. Effect f Airflw Distributin n the Perfrmance f NA/LT in-bin Drying f Crn. Transactins f the ASABE. 49(4): 1095-1104. 255