I I I I I I I I I walk. .lll Igmumm II II 293 10064 2770 I This is to certify that the thesis entitled AMONIA SORPTION AND DRYING MODEL IN A FIXED-BED GRAIN DRYING SYSTEM presented by RONG-CHING HS I EH has been accepted towards fulfillment of the requirements for Ph. D. degreein Agricultural E . . /". f 0 OVERDUE FINES ARE 25¢ PER DAY PER ITEM Return to book drop to remove this checkout from your record. AMMONIA SORPTION AND DRYING MODEL IN A FIXED-BED GRAIN DRYING SYSTEM By Rong-Ching Hsieh A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Agricultural Engineering 1979 ABSIRACT AMI/KNIA SORPTION AND DRYING MODEL IN A FIXED-BED GRAIN DRYING SYSTEM By Bong-(hing Hsieh One of the alternative energy-saving methods in drying grain is to apply a chemical as a grain preservative when the grain is drying slowly. Recent studies have shown that various chemicals such as ' ammonia, propionic acid, acetic acid, methylene bis-propionate, formalin and sulfur dioxide can serve as fungicidal agents. A camputer simulation model has been developed for the ammonia drying system of a fixed bed of grain. The model is based on the theoretical analysis of heat and mass transfer. Mass balances are made on the air and on the corn resulting in four independent differen- tial equations. The fifth equation for the temperature was derived by cornbining the two independent heat balance equations in the air and in the grain by assuning that the grain and the air temperatures are equal. Several crucial parameters in the ammonia drying simulation are unknown. The fol lowing estimated values were obtained by analyzing closely related systems: (1) the heat of ammonia adsorption 'in corn: 35.4 KJ/mole-NHB, (2) the convective heat transfer coefficient (h,w/mk) 0.49 in a fixed-bed system: 2.3278 Ga , when Ga<2400 kg/hr/mz, (3) the mass transfer coefficient between the grain and the ammonia: 1.156xlO-6h, Rong-Ching Hsieh and (4) the amount of fixed ammonia (residual) in the com: 30 percent of the free ammonia content. The equilibrium ammonia content in corn is a function of the grain moisture content and temperature. The ammonia sorption isotherms were established by assuming the water portion in corn to be the active component in adsorbing ammonia. A semi-empirical equation has been developed for the ammonia sorpt ion isotherms at low ammonia concentra- tions (less than 20 percent, d.b.). The five-equation model was solved numerically by finite differences. The simulation model was tested for its numerical accuracy by comparing it with the predicted results at extremely small computing time incre- ment and depth interval. An acceptable combination of the carputing time increment of 1 hour and the computing depth interval of 15.24 cm (0.5 ft) was established. The selection of appropriate grid sizes was based on the analysis of the computational error compared to the true numerical solution and the central processor time required by the com- puter. An experimental bin test of ammonia grain drying was performed to verify the simulation model. The corn was dried with ambient air from 25.6 to 15.6 percent (w.b.) moisture in 72 days and 16 hours of fan operation at 1.43 (mm/m2 (l cfm/bu) at a grain depth of 1.83 m (6 ft). The final corn sample showed no unacceptable grain spoilage. The total amount of anhydrous ammonia applied was 18.3 kg (0.8 percent d.b.). Satisfactory agreement between the experimental and the predicted results was found. The energy requirement in this ammonia grain dry- ing experiment was 3696 kJ/kg-water removed which is an energy saving of over 50 percent comparing with the high-temperature drying (7500 Bong-(hing Hsieh kJ/kg—water). A series of sensitivity tests on several important parameters was performed to determine their significance on the ammonia sorption rates. A significant increase in the ammonia adsorption rate was predicted when the grain moisture content is higher and the grain is treated at a lower temperature. A higher airflow rate and, consequently, a larger mass transfer coefficient also resulted in a higher adsorption rate. This investigation has established the general grain drying simu- lation model using fungicidal chemicals. The general model can easily be made specific to simulate the treatment of a certain type of grain with a specific fungicidal chemical by specifying the relevant property parameters . .I“ /.a’/”’/" // Apprwed /;7/ / .x/J/flz/ // Jaw Major Professor ”WW? 2...ng V 'Ib : My Parents : Pin-Ho and Tsai-Lian Hsieh My family : Suewhei and Chanlee and The People of TAIWW AWS 'Ihe airtkor would like to express sincere gratitude to Dr . Fredrick W.Ba]d ecu comes: .~ .osm—.moa—..—a um ppuzuosm .— "mouconoeuz .x .o: umaoguaom .a.gua—-aa< .umuoguaom uoe:_u:— c .V p 2 8 :N ‘33:... 58:42: 8.23. use an... N e s - em oep Loaa>opo Aguczou ecu snag «Acvguaom (<5; nem— ~ mm mm. Lou->o—o sgucaou can semi Acvu—oo :gou goz a—smoa~-e_u_i Asvae_pogou guaom age: «hop ~ me n- mace—a acammouoga econ we. scuu>opu onapon :gou .ouoa choc uvu—on choc sgumavc. e=_—p=a ac: mom— —.ov N am :2 $2 533%.: 595 A522. 58 2.5. 32.3% [$3 872. 3-8 2v 2. 3:3. .aaav 3.xou poem. ougaom ecu mopaa-m muaah mu—amam :.a.sc ucvxo>aam mueou< Lou» guano—we. gap: mopasam ugougoa 8:23 no 343.: .32: 5252.3 ES 5 8532. 5.32:2 .3. as: 26 aflatoxin and zearalenone in 1973 (Stoloff et al. , 1976) of the crop corn stored on farms and in country elevators, aflatoxin contamination was most frequently encountered-in the Southeast-Appalachia area with a 34 percent incidence of marketable corn with detectable aflatoxins (Table 2.1). Aflatoxin in corn seers to be a worldwide problem. Limited surveys showed a high incidence and high level of aflatoxin contamination in corn samples from sore African and Asian countries (Stoloff, 1976). 3-B. 3 . Bright Greenish-Yellow Fluorescence and Aflatoxins Bright greenish-yellow fluorescence (BGYF) has been reported in cotton balls infected with A. flavus (Marsh et a1. , 1955). A positive correlation between aflatoxin and the occurrence of BGYF (Marsh et al. , 1969) was reported. A similar fluorescence was detected in corn by Shotwell et al. (1972) and Fennell et a1. (1973). Shotwell et al. (1972) suggested using long-wave untraviolet light (365 nm wavelength) to detect BGYF with subsequent chemical analysis for aflatoxin. Prelim- inary evidence indicated a good correlation between mYF and aflatoxin in the commercial corn (Fennell et al., 1973). The relationship between fluorescences and the types of fungus in corn during storage has been investigated (Rambo et a1. , 1975). TWO yellow dent hybrids were inoculated with A. flavus and A. parasiticus and stored in static and aerated systems. TWO types of fluorescence, bright greenish-yellow (BGYF) and blue white (BWF) fluorescence, were observed. The A. parasiticus inoculated grain had more BWF than BGYF fluorescence, whereas corn inoculated with A. flavus had similar amounts of both types. The correlation between the incidence of either the BGYF or BWF and aflatoxins in individual kernels was 74-80 percent for 27 A. parasiticus (a good aflatoxin producer), but very low (5.0—6.2 percent) for A. flavus (a poor aflatoxin producer) (Rambo et al. , 1975). Afla- toxin also occurred in 6.7 and 19.6 percent of the nonfluorescing kernels examined. Although some fungi-produced compounds in corn with other colors of fluorescence are not associated with aflatoxin, the substance responsible for the BGYF flow in corn is not aflatoxin but a substance often produced by the mold that also makes the toxin (USDA, 1977). Fluorescence, therefore, should be interpreted as a warning that more definite tests are required. 3-C. Detection and Determination of Aflatoxin in Corn A nurber of analytical methods have been developed for the detection and determination of aflatoxin. These methods vary in purpose and com— plexity. Mycotoxins do not occur uniformly in contaminated corn; only a single highly-contaminated corn kernel may show contamination in an entire bulk of clean grain. Methods for aflatoxin analysis in corn can be divided into three categories (Shotwell, 1977): (1) Rapid Presumptive Tests: to locate lots of corn that may contain mycotoxin by visual test of fluorescences under an ultraviolet lamp. Although numbers of BGY fluorescing particles are related to the level of aflatoxin, the BGY fluorescent test cannot be used to determine levels of toxin (Lillehoj et al., 1973; Shotwell et al. , 1975C). A field method for the detec— tion of aflatoxin based on this presumptive test has been proposed by Hunt et al. (1976). Samples with kernels exhibiting BGYF under high intensity black light were tested for aflatoxin content using Velasco's ( 1972) rapid florisil column method. 28 (2) Screening Procedures: to detect the presence or absence of afla— toxin at a predetermined level. A small glass colum (i.e. the minicolum) containing appropriate adsorbents (e.g. florisil or silica gel) can be used as a quick screening procedure. The test requires 0.3 to 1 hour. The screening procedure has been adopted as the official first action by the Association of Official Analytical (hemists (AOAC, 1975) for the detection of aflatoxin in corn, peanuts, peanut butter, peanut meal, cottonseed meal, mixed feeds, and pistachio nuts (Romer and Campbell, 1976). Homer (1975) described a screening method of detecting the total aflatoxins (31’ B , G1, and fruit products in samples containing as little as 5 to 15 ppb. and G2) in mixed feeds, grains, nuts A practical screening test was defined as having the following characteristics: (1) short time of analysis, about 30 minutes, (2) equipment, simple and inexpensive, (3) capability of being performed by an unskilled personnel, (4) low cost, (5) reasonable accuracy, (6) sensitivity of better than 20 ppb, and (7) suitability for analysis of mixed feeds and feed ingredients. Holaday first developed the minicolunn screening procedure for peanuts and has since published an improved procedure (Holaday and lansden, 1975). Shotwell and Stubblefield (1973) presented three screening methods for the determination of aflatoxin in com. All three methods involve minicolumn chroratography of partially purified extracts. These studies were followed by a number of research reports on screening with the aid of minicolumns (Table 2.2). Detection limits of minicolunns range from 1 to 10 ppb aflatoxin . The reasonable accuracy and short throughput 29 .ssmp ._—mzpogm an ompwaeou apmzo—>mga « L Lasagna» oumcongmu A>\>vo¢"om Amnmpv N aspecmom< pom uQPPPm eaacrpsoaacwn canaau amuezupocasuoz «usage soumcmna ouaumom A>\>VONuom Amnmpv N mavecmumoa p¢m9Lopi essav—novzowb u:.~ gouazupocaguoz Ameapo: mumapsm. A>\>Vmpumm Aesmpv F acaecoummo pampgopi cowuugcam>m Ezscoee< gmamzumcoumo< soponmgmm A>\>V amemcmgu mamuoou oNuom noun: Anew—V o_ m=_e=m0m< Pam auapwm siza.p-ea=e_a ammo um_ags.=oscu< coa ammmcugb maoepzm A>\>Vmpumm Amsmpv op m:_e=oom< _om mopppm e_=cp—-o_=c_n Ezpcoee< gmumzucouou< coccagm A>\>Vm_umm ANNaPV m a=_ecouoo ppm—go—a copuagoam>u Pom usagmi gouarucoumo< oomapo> Anaavupewp unusaopm>oe «concomom :o vangucoucou copumu.a.uwga coauuaguxm copuumo ce=p6u_:_z ass—aowcvz oucmgmmux anew . messpouac—e mc.m: avenues mcpcoogum .N.N epoch i. (3) 30 period have made the screening methods suitable for field use. A rapid field method for the detection of aflatoxin contamin- ated corn has been developed (Shotwell et a1. , 1975B). The results were compated with laboratory quantitative and thin-layer chroma.- tographic methods. No significant differences were found. Shotwell et a1. (1976) devised another rapid screening test using minicolums and chromatographic analysis for aflatoxins to be performed at the corn elevator level. Presence of aflatoxin in corn sample results in blue fluorescent band at top of florisil layer that can be identified by corparison with a standard column free of aflatoxin. The method detects total aflatoxin level of 20 ppb or higher. Quantitative Methods: to determine the type and amount of afla- toxins. These procedures are usually lengthy (about 3 hours), complex and unsuited for field use. However, they can detect the amount of aflatoxins as low as l to 3 ppb, the lowest detect- able level. The procedure recommended by the Association of Official Analytical Chemists (AOAC, 1975) and the American Assoc- iation of Cereal Chemists (AACII, 1975) is the so-called CB (Contaminants Branch) method (Sec. 26.044-26.047, AOAC Official Method of Analysis, 12th edition). Seitz and Mohr (1976) described a method extended from a simple screening test for the quantitation of aflatoxins in corn. The method is claimed to be faster, easier, and less expensive than the CB method and the results are similar. High-pressure liquid chromatography (RPM?) is used for the separation and quantitative identification of aflatoxins Bl’ B2, Gl’ and 62 31 with good resolution and high sensitivity, 1 to 2 ppb of each aflatoxin. The chemical confirmation of the identity of aflatoxin is by the derivative format ion method (AOAC, 1970) which the chemical derivatives of aflatoxins are formed directly on the thin—layer chromatographic plate. This simple rapid method for the confirma- tion of aflatoxin B was developed by Przybylski (1975). The 1 "Przybylski derivative procedure” has been shown to give good results and can be performed easily (Stack and Pohland, 1975). 3-D. Detoxification of Aflatoxin The problems associated with the proposed methods of detoxifying aflatOJdn-contaminated agricultural comodities have been reviewed by Goldblatt (1969). The toxicity and carcinogenicity of aflatoxin were investigated by Wogan (1973) and stated that aflatoxin B1 has two ftmctional groups, the dihydrofurofuran segment and the lactone— pentanone ring system, responsible for B biological activity. To 1 investigate the reaction of weak bases, such as ammonia, with aflatoxin, 1 radiolabeled—C 4 aflatoxin B were added to corn grain flour (Beckwith 1 et al. , 1975). Low levels of ammonia as ammonia hydroxide (_<_ 2gNH3/100g flour) caused irreversible binding of B1 to corn flour comments. It was further established that the binding is preferentially with the major protein fraction and water-soluble portion of the corn (Vesonder et al. , 1975). Cucullu et al. (1976) proposed a series of chemical reactions for the ammoniation of aflatoxin B1 to produce D1 (MW. 286) and a corpound with M.W. 206. In a laboratory test of corn samples inoculated with A. flavus, it was found that both amrenia (2 percent by wt.) and propionic acid 32 (1 percent by wt.) reduce mold growth and subsequent aflatoxin and ochratoxin formation. Both ammonia and propionic acid remained effective in inhibiting mold growth and aflatoxin production for 19 and 29 weeks, respectively (Vandegraft et al. , 1975). Beckwith et al. (1976) inves- tigated sore chemical methods for detoxifying aflatoxins in foods and feeds and found that ammoniation at ambient terperatures decorposes and inactivates aflatoxins. Aflatoxins in grain artificially inoculated were found to be destroyed in a continuous fermentation process and by-product isolation (Dam et al. , 1977). The results indicated a loss of aflatoxins after fermentation of slightly greater than 60 percent of the original afla- toxin level. Further alkaline extraction treatments involved in the isolation of protein concentrate led to a destruction of total afla- toxins in excess of 90 percent . The destruction of aflatoxin have hap— pened due to the alkaline (NaOH) extraction procedure during the process. Besides chemical detoxification, some physical methods have been investigated. Roasting aflatoxin-contaminated corn was f01md to reduce aflatoxin levels (Shannon and Shotwell, 1975; Peplinski et al. , 1975). Physical separation methods involving dry cleaning, wet cleaning, and hand separation of aflatoxin—contaminated grain were not successful (Brekke et a1. , 1975B) in lowering the aflatoxin content of the naturally contaminated corn . 4. Ebcperimental Results from North Regional Research laboratory (NRRL) of USDA The results were generated from various tests on ammonia detoxifi- cation, ammonia preservation, and ammonia-assisted ambient air drying. The composition analysis of the ammonia-trested samples gives very 33 important clues on the actual chemical reaction between the grain corponents and ammonia. The experimental results presented in this section were closely related to the behaviors of armenia in various ammonia treatment processes . 4-A. The Diffusion of Ammonia in Corn Lancaster et al. (1974) proposed that "since the ammonia molecule is the same order of size as the water molecule, its rate should be similar at an equivalent driving force " . However, the molecular size alone cannot be taken to compare the diffusion rates. The ammonia molecule has stronger affinity towards the water molecule than the water molecule to itself as indicated by the relative mass diffusivities (Sec. 5-C.2., Chapter 5). The diffusion of free ammonia from whole corn into water at room temperature is a slow process. The times needed for diffusion of free ammonia from the corn kernel are shown in Figure 2.1. It takes about six hours to extract 95 percent of free ammonia from the corn kernels. Results in Figure 2.1 roughly indicate the ammonia diffusion rate in the corn kernel despite the test is a soaking process. The fact of slow rates of amronia diffusion in the corn kernel is evident and a slower rate of ammonia application will lose less ammonia in the exhaust air. In another test by Lancaster et al. (1974) to investigate the time required to equilibrate ammonia treated and untreated corn, it was found that armronia took about 20 times (about four days) longer to reach equilibrium than between corn and water in a stagnant situation. The long equilibration time indicates the difficulties of losing free axmonia from the corn kernels. Nevertheless, this test does demonstrate 34 that a homogeneous sample can be obtained from mixing ammonia-treated and ammonia-free samples. In order to investigate the ammonia adsorption capability of low- moisture com, a sample of 2.7 kg of corn at 12 percent m.c. (w.b.) was placed in a glass column and 7 percent of ammonia was added slowly with no escape of armonia over a period of two days. The amount of free ammonia was 3.8 percent. Thus, the residuals constitute about 45.7 percent of the total ammonia added. It is also shown that the fixed ammonia content (or residuals) in corn increases with increasing terperature and time. In a series of ammonia detoxification tests using aqua amronia for corn samples adjusted to various moisture levels (Brekke et al. , 1975), the percentages of ammonia with respect to the total ammonia added were determined. The results indicate that at ammonia addition levels of 1.5 percent and above, the free ammonia values level off at about 2/3 of the ammonia added when samples are held 14 days at 25°C. The lower the amount of total amrenia added, the lower the percentage of free armenia is when the total ammonia level is below 1.5 percent. 4-B. The Fungicidal Effects of Ammonia High moisture corn (27 percent) was treated with 0.48 percent ammonia applied as a 22 percent aqueous solution in a gas-tight tower silo (Harvestore System) containing 52.8 m3 of corn (Bothast et al. , 1975). With the initial application of ammonia, all microorganisrs were either eliminated or reduced significantly as shown in Figure 2.2. It was suspected that the initial heating was a result of ammonia adsorption and the subsequent heating was due to the respiration of the developing microorganisms. After 30 days of storage, the free ammonia ox o I .5 O I 20- 10- Free ammonia left in corn, percentage L l l l I 1 0 l 2 3 4 5 6 7 Time of soaking, hr LA) Figure 2 . 1 - Diffusion of free ammonia from whole corn into water at room temperature (laroaster et al. ,1974) . 8.0 __.—-————"'_-_'-- -_.. //” Bacteria // \““—-- 0 ~'—-—.—-—" "60 6.0- I / Tamperature ~49 l N \l J H O) Terperature , °C Log count of microorganism 0 50 100 150 200 Days of storage Figure 2.2. The initial application of ammonia eliminated or reduced all microorganisms (Bothast et al. ,1975) . 36 disappeared and the grain moisture migrated. loss of fungicidal prop— erties of the armenia was evidenced by the increasing fungal and actinomycete populations after the first 10 days of storage. 4-C. The Closed—System Ammoniat ion In a recycle gas-phase treatment of 17 and 26 percent moisture corn (Lancaster et al. , 1975), 2.3 kg of shelled yellow dent corn was treated with ammonia in air in a glass column (6.9 on I.D. x 114 cm long). The airflow through the column was 0.56 arm/m3 (0.6 cfm/bu). The inlet ammonia concentration was 16.2 percent for the 17 percent moisture corn. The total amount of ammonia applied was equal to 2.5 percent within a three—hour period. For 26 percent moisture corn, 11.8 percent of inlet ammonia in the air was added over a l l/4—hour period. The total amount of ammonia applied is equal to 0.75 percent. Lancaster et al. (1975) stated that localized adsorption of a quantity of ammonia equal to five percent in the corn can raise the corn temperature by 39°C in a stagnant and isolated system. The heating of corn during the application of anhydrous ammonia is fast and temporary in contrast to the slow terperature rise due to microbial growth observed during the subsequent storage period. 4-D. Influence of Armonia and Moisture levels on the Detoxification of Aflatoxin B1 The influence of ammonia and moisture levels on the inactivation of aflatoxin B1 was considered by Brekke et al. (1975). The samples were held for 14 days at 25°C after the treatment of aqua ammonia at different ammonia content levels. The pronounced effect of corn moisture on the inactivation of aflatoxin B with ammonia is shown in Figure 2.3. l 50 Sample held 14 days at 25°C Initial Bl content=220 ppb 40 .. Total ammonia in corn é o 1. 0. 5%,d.b. ~ 2. l. 0% %,d.b. mfl 3.105%pdobo B 30 ... m r-l I ‘13 'H o E 20 m b g 8 10 .. o I I 0 l L n : . .12 .14 .16 .18 .20 Corn moisture ccntent,w.b. Figure 2.3. Effects of ammonia content and corn moisture level on the inactivation of aflatoxin B1 (from Brekke et a1. , 1975) . 38 The inactivation is much more effective when the corn is treated at higher moistures. At moisture contents above 17 percent, aflatoxin B1 is almost completely inactivated when treated with 1.5 percent ammonia. It can be concluded from the results in Figure 2.3 that treating lower-moisture corn requires a higher dosage of ammonia in order to reach the same degree of inactivation of aflatoxin B The reason is 1. that high—moisture corn is capable of dissolving more free ammonia which is the primary element for aflatoxin inactivation. In another ammonia detoxification test of artificially contaminated (1000 ppb) corn samples (Brekke et a1. , 1975), the effectiveness of detoxification was found to be terperature dependent as shown in Figure 2.4. At higher terperatures, shorter periods of ammonia treat- ment were needed because of higher ammonia diffusion rate and higher chemical reaction rate between the aflatoxin and ammonia. Furthermore, less ammonia was needed at higher temperatures to have the same inactiv- ation effect on aflatoxin B1. 4-E. long-Term Preservation of High-Moisture Grain Although a high dosage of amrenia (1.5 percent) is frequently used in the detoxification process, the required amount of ammonia is much lower (0.5 percent) for the purpose of preserving freshly harvested high- moisture corn free from spoilage (Nofsinger et al. , 1976). In a long-term preservation test of high—moisture grain (Peplinski et a1. , 1975), ammoniation destroyed all molds, lowered bacteria counts for up to four months, and modified some chemical and physical properties of the treated corn. During a l4—month storage test of 4036 kg (158 bu) of 23 percent (w.b.) moisture corn treated initially with 1.02 percent aqua ammonia (in 19.2 percent solution, weight basis), the moisture 39 200 Com moisture content =15.0% Initial aflatoxin Bl=1000 ppb 1. 42 days at 10°C 150 *- 2. 8 days at 25°C g 3. 2 days at 40°C L. m i «1 100 .- cm \H o o o '43 h g 50 - 8 o l l l 0.0 0.5 1.0 1.5 2.0 Ammonia concentration in corn , percentage ,d .b . Figure 2.4. Effects of temperature on the inactivation of aflatoxin B in corn (data taken from Brekke et al.,l975) . l 40 content of corn was increased to 28 percent. A 1.12-Kw (1.5-hp) centrifugal blower recycled the ammonia-air mixture. The mixture was recycled for periods of 1.5 to 25 hours at a flow rate of 1.8 orm/m3. In total, an amount of 3.04 percent of ammonia was added to the corn in aqua and anhydrous form. The relationships between the amount of ammonia applied, the average free ammonia content, the bacteria, and the mold counts are shown in Table 2.3 throughout the entire treatment. Molds were almost corpletely wiped out by ammonia after the first treatment of 1.02 percent ammonia. The bacteria counts were also reduced significantly due to the presence of ammonia. The relationship between the free ammonia content and the bacteria count did indicate a general trend of lower bacteria comt for the grain with higher free ammonia content (Peplinski et al. , 1975). A limited correlation between bacteria and free ammonia shows that a minimum level of 0.50 percent free amronia in the corn is necessary to keep bacteria below 100,000 per gram of corn. In another anmonia—srpplemented ambient drying test of high-moisture corn conducted by Nofsinger et al. (1976), 14.2 tons (560 bu) of 23.2 percent m.c. (w.b.) corn was dried with ambient air and ammonia preserv- ation over six months without corn spoilage. The fan was rated 1/3 hp (248 W) and the airflow was 1.8 omn/ton (1.6 cfm/bu). The corn was dried to 17 .7 percent in 56 days during which the fan operated 44 days. Anhydrous ammonia was applied intermittently. The ammonia application rate was much higher in the beginning of the drying process; about one- half of the total 42.9 kg ammonia was applied in the first three days and more than 75 percent of the total ammonia was applied in the first 41 Table 2.3. Effects of ammonia on inhibiting bacteria and mold grwth and the relationship between free ammonia.and the amount of ammonia added (data taken frrm Peplinski et a1.,1975) . Day' Ammcmdet .Average . Bacteria .MoLd Average applied freeeammonia count count corn.MC (%d.b.) (%d.b.) (m.c./g) (m.c./g) (%w.b.) 0 1.02 0 2,000,000 630,000 23.0 28.0* 1 0 - 3,000 ND** - 3 o - 0.69 - ' - 28.0 7 0 - 1,000 ND - 13 0 0.25 - - 27.0 26 O 0.13 - - 26.3 28 0.26 4 - — - 32 0 < 0.29 410,000 ND 26.7 45 0 0.22 22,000 ND 27.0 60 0 0.09 1,030,000 30 27.3 67 0.51 - - - - 69 0 0.42 - - 27.0 98 0 0.28 182,000 ND 27.7 111 0.51 - - - - 146 0 0.50 173,000 ND 28.0 166 0 0.37 114,000 ND 26.3 235 0 0.25 80,000 ND 27.7 242 0.26 - - - - 276 O 0.19 902,000 ND 24.7 297 0.48 - - - - 355 0 - 33,000 ND - 410 0 0.27 - - 20.3 * The moisture increase is due to the addition of aqua ammonia. ** ND = none detected 42 twelve days of ambient air drying. The air temperature varied from about 20 to -5°C and the relative humidity from 60 to 90 percent throughout the drying operation. The maximum ammonia flow rate was 0.36 g/(hr.kg). The relation- ship between the total amount of amronia actually picked up by corn and the cumulative ammonia added is shown in Figure 2.5. An appreciable amount of ammonia was lost in the exhaust air. The amount of ammonia remained in the corn consists of free and fixed ammonia during ammonia treatment. However, the portion of free ammonia will become either lost in the air or a part of fixed ammonia after an extended period of aeration. Surface-sterilized corn kernels were tested for mold and bacteria growth during ammonia application. Mold counts were reduced from 6100 to 16 propagules/g and mold—infected kernels were reduced from 92 to 14 percent. Molds increased slightly thereafter but were effectively controlled by the presence of ammonia. In contrast, bacteria counts increased from 2.4 x 104 to 1.6 x 107/g during the first seven days of storage because of favorable growing temperature and humidity. The bacteria growth then leveled off and the subsequent growth was limited since the corn moisture level was reduced and unfavorable to microbial growth. The amount of residual ammonia in the corn was found to be about 40 percent of the cumulative ammonia added in another ammonia grain drying test performed by Nofsinger et al. (1977) using typical farm—size bins. The largest retention of ammonia and lowest mold infection were detected near the bottom of the bin. 43 0.4 :5 / 3 / g / a / ~ 0.3 - / 5 / f nT'georeticallgziues of. // a / ’1‘ a .2 .. / E / / x x / 0 1 - / / X x / / x / / 0.0 .1 l l 0.0 0.1 0.2 0.3 0.4 Cumulative ammonia added, percent,d.b. Figure 2.5. The amount of ammonia residual in corn as related to the cumulative amount of ammonia added (data taken from Nofsinger et a1.,1976) . 44 4-F. The Composition Changes of Ammonia-Treated Corn The chemical changes of stored corn treated with ammonia during a l4—nonth storage test were analyzed by Peplinski et al. (1975) . The percentage changes of the major components in corn are shown in Table 2.4. The total nitrogen content in the corn increased from 1.5 to 2.1 percent which is about the same amount of increase for ammoniacal nitrogen. Among the 21.4 kg of ammonia adsorbed by the grain eventually, total ammonia (ammoniacal nitrogen) accounted for 74.3 percent (15.9 kg) of the total nitrogen increase. The rest 5.5 kg of ammonia applied reacted with corn components to form other than ammoniacal compounds which is evidenced by the increase in soluble solids from 6.5 to 9.5 percent. Analyses of the solid residues indicated that the chemical reaction products accounted for the change in soluble solids. The decrease in extracted fat content is almost entirely due to the decrease in linoleic acid (C18H3202, an unsaturated fatty acid with two double bonds) content. Four unknown compounds ranging from 0.9 to 8.0 percent of the total fatty acid composition developed during the ammonia treatment due to the degradation of linoleic acid. The reduction of non-reducing sugars from 2.9 to 1.3 percent indicates microbial growth and respiration. At the end of the 410 days storage test, the corn was removed in six separate layers. Excluding higher-moisture corn adhering to the bin walls, 90 percent of the corn was removed with moisture contents ranging from 15 to 24 percent. The chemical composition of each layer of corn is shown in Table 2.5. The free ammonia content, higher toward the top of the bin, averaged 0.2 percent while the total nitrogen content averaged 2. 2 percent . 45 . you 883038.88 mo uncommon mm 8. 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 888 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 888 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 888 8.8 8.8 u 8.88 8.88 8.8 8.8 n 888 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 888 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 88 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 88 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 88 8.8 8.8. 8.8 8.88 8.88 8.8 8.8 8.8 88 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 88 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8 8.8 8.8 8.8 8.88 8.88 8.8 8 8.8 8 ..n.888 1.82888 A.m.888 1.82888 1.89888 A.n9888 1.8.888 A.n.ewv 8838080. 88mm mummom commum «398 00908388 8.88—EB 5880.38: >3 8888808 afiBBnrnoz 088898.88 88.8 830.8. 88.98. . 18888.88 88 8088:888ch 888 8888 can 80 8888838 88828.8 .88 88888.8 . 088 888088888 88 8:888 88 .3. .3483 5.88 mfi on 98.8.8968 E800 88833898880888 men 0585.898 .8 8.88 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 8wmmmwwm 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 888888 8.8 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 8 8.88 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 e 8.88 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 8 8.88 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 8 8.88 8.8 8.8 8.88 8.88 8.8 8.8 8.8 8.88 can 1.8.888 1.8.888 1.n.888 1.8.888 1.8.888 A.n2888 1.82888 1.8.888 1.83388 888888 :98 888888 .8838 :38 . 8308888 88888: 8.8.8.8 888.80 88888 88888 888883.82 08888.3 88.8 88898. 88.8 838898 .3311? no 085308 «magma .3 098838.. 8:98.08 8.96.58 :98 mo 38:88.98 888088558 .m.m 03.8.8. 47 The general trend of variations in the chemical compositions of different layers is as follows: 1. The fat and linoleic acid increased while the soluble solids content decreased toward the bin bottomn Thus is an indication of more chemical reactions happening between ammonia and corn components toward the bin top than the bottom. This is due to the more amronia dissolved in the grain in the bin top as a result of the higher grain moisture content . 2. The non-reducing sugar (sucrose) content increased toward the bin bottomn This is a sign of more microbial activity in the top layers as was expected. 3. Starch and ash contents did not vary significantly in the grain samples through- out the bin. The following practical information has been suzmrarized from the above experimental results which have assisted this investigation: 1. The fungicidal effects of ammonia towards molds, bacteria, and aflatoxins have been verified. 2. Practical-scale tests of the ammonia treatment for high— ‘moisture corn have been performed and.were proved to be feasible, practical, and economical. 3. The total ammonia in corn consists of the free ammonia and the fixed ammonia; the fixed ammonia content was“ found to be about 30 to 50 percent of the total ammonia depending on the rapidity and the amount of ammonia application. 4. The composition analysis of the ammonia-treated corn has verified the existence and the amount of residuals f- the fixed ammonia. 5. The amount of ammonia required in a long-time preservation and drying test of corn is about 0.5 percent d.b. (based 48 on the weight of dry corn). 6. The amount of ammonia required for the detoxification of a batch of aflatoxin-contaminated corn 'is about 1.5 to 3.0 percent d.b. 7. Orly an aeration fan which delivers around 1 to 2 cmm/ton is required in the ammonia-assisted ambient air drying and the ammonia treatment permits several months to dry the corn to 15.5 percent w.b. without spoilage. 8. The trickle (intermittent injection) ammonia application process requires a high dosage in the beginning to com- pletely inactivate the initial contamination of micro- organisms . Nevertheless, it was found that some important aspects were ignored. The following information is essential in analyzing the ammonia grain drying system but is lacking in the above experimental results: 1. the ammonia sorption isotherm which is necessary in describing the steady state sorption characteristics; 2. the dynamic movement of ammonia in the grain bin, i.e. the ammonia concentration changes with time and position; 3. a systematic analysis of the influence of important parameters, such as the mass transfer coefficient, airflow rate, inlet ammonia concentration, grain moisture content, terperature, etc. , on the ammonia sorption processes. The result of a systematic analysis will not only identify the important parameters in the system but also establish the basis of a prediction model . CHAPTER 3 AMMONIA SORPTION ISOIHERMS 1 . Introduction In order to investigate the behavior of ammonia in a fixed bed of biological products, such as grains, the mass transfer processes of ammonia between the moving air stream and the fixed bed of grain must be analyzed. The armonia sorption in a fixed bed of corn is a dynamic process. The proposed model is deterministic (without random variables). In analyzing a dynamic process, a steady-state system is frequently first investigated. In the steady—state system, all the parameters are independent of time and the system is studied under equilibrium condi- tions. The curve of ammonia sorption behavior at constant terperature is called the Ammonia Sorption Isotherm. 2. Adsorption and Diffusion Models Adsorption results from the field force at the surface of the solid (adsorbent) which attracts gas molecules (adsorbate) and are in turn adhered to the solid. The force of attraction could come from physical (van der Waal's type) or chemical means. The process of adsorption has important applications in drying, purification, separation, catalysis, waste treatment and many other chemical processes. The adsorption system in the ammonia grain drying system consists of corn (the adsorbent) and ammonia gas (the adsorbate). In general, the majority of large scale commercial applications of 49 50 adsorption involves nonisothermal adsorption of multiccmponent mixtures in fixed of fluidized beds. The ammonia grain drying system is one example of an adsorption application in agriculture. Strictly speaking, in most agricultural crop drying systems, treating a fixed bed of corn with ammonia-air mixture is an example of the adsorption process. The process involves nonisothermal sorpt ions of mult icomponent mixtures of ammonia, moisture, and dry air. The rate expressions of sorption processes have been classified into three groups: (1) external diffusion, (2) surface adsorption, and (3) internal diffusion. The sorption process consists of numerous different processes; the rate at which an adsorbent takes up adsorbate can be limited by any or all of the above models. 2-A. The External Diffusion Models-Diffusion in Fluid The external diffusion model describes the concentration gradient existing within the bulk stream of the fluid phase and the transport mechanism of the ammonia to the external surface of corn. The overall rate of adsorption depends on the diffusion velocity of ammonia from the bulk stream to the vicinity of the corn surface and on the rate of adsorption of ammonia at the surface layer of the corn. The external diffusion model is usually expressed by Fick's diffusion equation: 39- = 0 vzc (3.1) where C is the concentration of adsorbate, t is the time, and D is the diffusion coefficient. Equation (3.1) is a form of Fick's second law and is analogous to the transient heat conduction equation. The solution of this equa- tion involves prescribed boundary conditions and numerical techniques, 51 such as finite differences, finite elerent methods, etc. In the external diffusion model it is assumed that the rate of surface adsorption is so rapid that there is solely diffusion control. The ammonia adsorption isotherm defines the boundary condition. Nevertheless, the mathematical model defined by the diffusion law and the boundary condition mentioned above poses a very complex problem, since the model equation (Equation 3 . 1) involves two independent variables and second order differential term(s). It is usually simplified to the "film diffusion model" by defining a simple mass transfer equation to describe the external diffusion of adsorbate from the bulk stream to the outer surface of the solid: %%= kfa(C-Cm) (3.2) where q is the adsorbate content in solid, a is the specific (external) surface area of solid particles, and kf is the fluid-phase mass transfer coefficient. The driving force for mass transfer is defined by the concentration difference across an effective "film" resistance surround- ing the particle between the fluid at the surface (C) and the bulk fluid (C00). The mass transfer coefficient kf is a fimction of fluid velocity, viscosity, density, the diffusivity of adsorbate in the fluid, and the size and shape of the solid particles. 2-B. The Surface Adsorption Models The surface adsorption model describes the mass transfer on the solid surfaces. Solid adsorbents have a limited total internal surface area on which adsorbate can be adsorbed. The theory of the adsorption rate on an active surface was first derived by Langmuir (1918) who equated the rate of capture of molecules 52 from the gas to the rate of escape of molecules from the surface under equilibrium conditions: ka p(l-6a) = kd 6a (3.3) where ka and k d are the rate constants for adsorption and desorption , 6a is the fraction of surface area covered by adsorbed molecules, and p is the partial pressure of the adsorbate molecules. Solving for 6a and letting KA = ka/kd , KA_P ea g-iTKA—P? (3.4) In the case of ammonia adsorption in corn, the equilibrium com-- a: centration of ammonia in corn (x2 ) at any concentration of ammonia in the air (x1) can be expressed using the above equation: x28 x1 * = (3.5) X2 KL + x1 where, * — x2 - ea ° X2m x1 = p/pS KI. = 1lKAps saturated ammonia pressure *6 U) II The capacity of the solid phase onto which adsorbate can be adsorbed (x2m) refers to the total coverage of active sites on the solid surface where the adsorption solely occurs. However, in the ammonia adsorption process in com, the only active component in corn absorbing ammonia is assumed to be water. The water portion is believed to be distributed evenly within the entire corn kernel. Thus, the following assumptions must be proposed in order to use the Langmuir monolayer surface adsorption 53 isotherm for the amronia—corn system: 1) The individual corn kernel is small enough that the internal ammonia concentration gradient can be neglected, or, the ammonia in corn at any moment is rapidly equilibrated and evenly distri- buted. In other words, all the active sites of corn in adsorbing ammonia are located on the surface. 2) All the other components besides water in corn absorb a negligi- ble amount of armonia through physical adsorption or chemisorption. The Iangmmir isotherm descirbes the processes of ion exchange, active carbon adsorption, etc. , since these processes obviously fall in the category of surface adsorption. Nevertheless, the sorption processes of ammonia or water in most biological products cannot be explained satisfactorily by the Langmuir monolayer isotherm alone. In dealing with the sorption of water vapor in food products, the BET equation (Brunauer, Emmett, and Teller, 1938) is frequently used. The BET equation allows multilayer adsorption on the solid surface. x2* __ “’1 X1 _ (3.6) x2111 (1 - x1) [1 + (cl - 1)xl] A simplified expression for the constant c1 is: Cl = exp (Qs/R'T) (3.7) where, Qs = site interaction energy of adsorption = Q1 - L, Q1 = total heat of adsorption of the monolayer, L = latent heat of evaporation of adsorbate at terperature T, T = absolute temperature, and R' = universal gas constant. 54 Q1 may be estimated by the heat of solution of ammonia in water, 8480 cal/mole and the latent heat of evaporation, L, of ammonia is 4889 cal/mole at 15.6°C. In certain cases when the number of layers is limited to a finite number n, the BET treatment leads to the modified equation: m+1 1 I'M-f m x2* = Cl xl l-(m+l) x1 + m x x l - x _ _ 2m 1 1+ (c1 1)):l cl x1 . (3.8) Equation 3. 8 represents the general BET equation which includes special cases such as the Langmuir isotherm (Equation 3.5) by letting m = l, and the standard BET equation (Equation 3.6) by putting m = 0°. These two extreme sorption isotherm types are shown in Figure 3.1. BET 3 isotherm a (m=°°) IS '§ 2% u. _______ __ ._}fflffl Kan Ea '5 Langmuir 33' 8 isotherm , ig LN g slope=x2"/KL 0.0 1.0 x1, relative partial pressure Figure 3. 1. Types of generalized BET adsorption isotherm. At low relative partial pressures of adsorbate in a fluid stream, the sorption behavior has negligible difference for langmuir and BET equations since they are still in the monolayer adsorption region. Therefore, it is not necessary to differentiate these models if the 55 ammonia concentration in the air is low. 2-C. The Internal Diffusion Models-Diffusion in Solid The internal diffusion process involves the transport of adsorbate to the interior of the adsorbent through one of the following mechanisms: 1) gaseous diffusion through the pore structures, e. g. pores, crevices, capillaries, cracks, and checks, in a porous adsorbent, or 2) surface diffusion along the solid surfaces of the pores in a Porous adsorbent, or 3) solid diffusion through the homogeneous, permeable, non-porous adsorbent. Adsorbents are usually classified as porous and non-porous depending on the effective radii of the pores. The effective radii of the largest variety of adsorbent pores (macropores) exceed 1000A, the effective radii of intermediate pores ranges from 18 to 1000A and those of the smallest variety of pores (micropores) ranges from 5 to 10A (Dubinin, 1972). As the pore size diminishes, the "porous" material becomes non-porous. Diffusion in porous materials may occur by one or more of the following three nechanisns (Sherwood, 1975): 1) Ordinary or bulk diffusion: prevails as the pores are large in relation to the mean free path* of the molecules of adsorbate. The process is recognized as an ordinary diffusion within the gas contained in the pores. The resistance to diffusion along the pores is due primarily to molecular collisions. Hence, the resistance * "Mean free path" is important in kinetic theory, and means the average distance a molecule travels between gollisions with each other. The mean free path of amronia is 441 at 0°C and 1 atmosphere (Moore, 1965). 56 of the pore walls may be neglected and the diffusion coefficient will be independent of the pore radius. The diffusion coefficients values range between 1.0 cmz/sec for gaseous and 10"4 5 to 10- cm2/sec for liquids (Wheeler, 1951). For gaseous, the diffusion coefficient is inversely proportional to the pressure and is proportional to 11'5 or T1'75 (Smith, 1970; Karger et al., 1973). The rate equa- tions proposed by some researchers have been thoroughly reviewed by Park (1974) . 2) Knudsen diffusion: occurs when the pore sizes are small compared to the mean free path of the gas molecules. The molecules collide much more frequently with the pore walls than with each other. A molecule travels within the pores by a series of "random flights" interrupted by collisions with the pore wall and adsorp- tion by the pore walls. 3) Surface diffusion: occurs when the diffusion species are adsorbed by the solids. The equilibrium surface concentration increases with concentration in the gas. So the surface layer tends to develop a concentration gradient in the solid in the same sign and direction as the concentration gradient of the gas in the pore. If the rate of surface diffusion is not the limiting factor or if the size of particle is small anough, the concentration gradient in the solid can be neglected. 3. Production and the Properties of Ammonia 3—A. Energy Required in Ammonia Production Anhydrous ammonia is being produced industrially from natural gas through a series of high terperature chemical reactions. The energy 57 requirement of the production of one ton of anhydrous ammonia is (Faith et al. , 1975): Input Quantity Energy Equivalent /ton 3 10 Natural Gas 736 m 2.719 x 10 J (92% methane) Fuel Gas 0.594 m3 2.321 x 1010 J (for driving compressors) Electricity 108 Kw hr 3.888 x 108 J . 10 Total. 5.079 x 10 J/ton The total energy input for the production of one ton of anhydrous armenia amounts to about 5.079 x 1010 J /ton, or 5.598 x 107 J /kg- ammonia (24067 Btu/1b-ammonia). 3—B. Physical Properties of Ammonia Ammonia is colorless and has a pungent odor. The physical properties of ammonia are listed in Table 3.1. From the solubility of ammonia in water and the density of ammonia gas, it is known that liquid water is able to absorb over 600 times its volure of ammonia gas at 0°C, one atmosphere. The toxicity limit is the concentration of ammonia to which workers may be exposed without harmful effects . The armenia concentra- tions of 20 to 50 ppm in the air can be detected by smell. At high concentrations (above 700 ppm) of amronia can cause severe irritations to the eyes, bleeding and swollen eyelids. Without immediate medical treatment, partial or complete loss of vision may occur. Moist skin is mildly irritated when a concentration of one percent of bulk air is reached in the atmosphere. Liquid anhydrous ammonia in contact with the skin will cause severe burns and frostbite due to 58 * Table 3.1. Physical prOperties of ammonia . Nblecular weight ........... . ...... . Specific volume (at 21°C, 1 atm) Boiling point (atl atm) Freezing point (atl atm) ............. Density, Gas (at 0°C, 1 atm) Liquid (at b.p.) ...... .. Viscosity, Gas (at 0°C, 1 atm) Flammability limits in air ........... . Heat of vaporization (at b.p.) ..... . . . Heat capacity, Gas (at 25°C, 1 atm) C O...OOOOOOOOOOOOOOOOOOOOO Vapor pressure (at 21°C) ........... . . . Solubity in water (at 0°C, 1 atm) ..... Thermal conductivity, Gas (at 0°C, 1 atm) ....... ........ Specific gravity, Gas (at 0°C,l atm) .. Toxicity limit (Thresrold limit value). 17.03 mumvgua6fibm) -33.4°C ('28.1°F) “77.7°C (~107.9°F) 0.00077 gxml 0.674 g/ml 0.00918 centipoise 15-28 percent (by volume) 327.4 cal/g 0.516 cal/g°C 0.407 cal/9°C 1.269 7.867 x 104 Nxm? (114.1 psig) 47.3 g/100 9 water 0. 02184 W/m K 0. 597 25 PF“! * From LINDE,Product Information , Union Carbide , specialty gases and equipment (1977) . 59 the rapid evaporation of ammonia from the skin surface. Although amronia is a non-flammable gas when contained in a tank with liquid phase, it can be ignited in air at concentrations of 15 to 28 percent by volure, when sparked. Workers handling ammonia should wear a face shield in addition to the normal safety equipment. It is also recommended that anyone working with anhydrous armenia carry a small spray bottle of water as a first aid to flush and protect the eyes . 3-C. Vapor Pressure of Armonia 3—C.l. Vapor Pressure of Liquid Ammonia The vapor pressure of liquid ammonia in a container as a function of temperature is shown in Figure 3.2. The ammonia vapor pressure is increased exponentially with temperature. At higher terperatures, the vapor pressure increases significantly. Thus, keeping the ammonia tank covered or sheltered while it is left in the field is very important in preventing overheating of the tank by direct Slmlight. 3-C.2. Partial Pressure of Ammonia in Aqueous Solution The partial pressures of ammonia over an aqueous solution of ammonia are listed in Table 3.2. The units for amronia concentration both in the air and in the solution are expressed in accordance with those of the water vapor adsorption isotherm. The amlonia concentration in the air is expressed as the volume percentage at one atmosphere based on the bulk stream total volu're. However, as a matter of consistency and simplicity in the model equations and computer implementation, the ammonia concentration in the air is expressed on the basis of voluIe percentage per dry air volure. Loglo value of vapor pressure, mm Hg 60 5.0 4.5 r- 400 _ 3.5 F 2.5 .4 1 I l l L L g -30 -20 -10 0 10 20 30 40 50 60 Temperature , °C Figure 3.2. The vapor pressure of ammonia as a function of terperature. .88888. 888osac 8:8 88888 so 8888 88 ozone 885853.80 88.883 .888 0:88 :8 5888888550888 m8: m0 88888.5 one .8. .uoumfm ooa \ 821m 688883508888 85303 :8 out c0338 98.8. 88.8 58.888888888880850 88889.58 .88 38c: . .8 61 a--- nus- n--- run: run. nun: 88.888 88.88 (1-: (ll: urn- (It; (1:. 88.888 88.88 88.88 (In- (In- 7... 88.888 88.888 88.88 88.88 88.88 a--- 88.888 88.888 88.88 88.88 88.88 88.88 88.88 88.888 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 88.88 . 88.88 88.88 88.88 88.88 88.8 88.8 88.88 88.88 88.88 88.88 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 HCOHUSHB 8.8.88 8.8.88 8.8.88 8.8.88 8.88 8.8.8 6.8 one :8 28888 883:8: 888\ 8:285 88888888088 .588 a an new on» :8 cog—8888:8200 88:98:28 888889.58 m .88888 888.826 one 88888 .888. 888988 NO 855.888 9.002% .8086 888889.58 mo monsmmmum Hmfluom .mlm 8.3.8.8. 62 Isotherms for amrenia solution at one atmosphere are plotted in Figure 3.3. Due to the characteristics of the saturation of ammonia in aqueous solution, the isotherm curves for armonia solution resemble Langmuir isotherms even though they are not certainly related. At higher temperatures, the same partial pressure of ammonia in the air will dissolve a less amount of ammonia in water under equilibrium conditions since the solubility of ammonia in water is smaller at higher terperatures. 4. Adsorption and Desorption of Ammonia in Corn 4-A. Classification of Sorption Isotherms The amount of adsorbate adsorbed (y) per unit mass of adsorbent depends on the equilibrium partial pressure of the adsorbate (p), the temperature (T), and the nature of the gas (g) and solid (5): y = f (p, T. g. S) (3.9) In general, the sorption system is investigated under constant terperature (i.e. isothermal) and under the assulption that the proper- ties and state of gas and solid do not vary significantly, Equation 3.9 simplifies to: Y = f (mtg,S (3.10) if the gas is below its critical temperature* throughout the terperature ranges considered. In other words, there is no phase change due to temperature changes alone. The following equation is most camonly referred to as the adsorption isotherm equation: * For ammonia gas ( ), the critical terperature is 405.6 K (Moore, 1965), whic is well above the temperature range considered here. y, ammonia concentration in water, g-m3/100 g-H20 50 40 30 20 10 63 L 20 x1, relative partial pressure of ammonia l 40 l 60 0°C 4.4°C 10°C 5.6°C 21.1°C 26.7°C 32.2°c l 80 100 Figure 3.3. Ammonia isotherm curves in ammonia-water system at 1 atm. 64 y=f(X)TgS where x = p/pO is the relative partial pressure and p0 is the saturation vapor pressure of the adsorbate. Most of the adsorption isotherms resulted from physical adsorp— tion and reported in the literature have been classified into five types by Brunauer, Emett, and Teller (BET) (1938) as shown in Figure 3.4. Type I is the Langmuir adsorption isotherm which is interpreted as a monolayer adsorption pheromenon. The S-shaped Type II is the well known BET isotherm for multilayer adsorption. Physical interpretations of Types II, III, IV, and V were presented by Brunauer, Deming, Deming, and Teller (1940). Types II and III resexble Types IV and V respectively except when the relative pressure approaches 1 . 0, the adsorption in Types II and III increases sharply and in some cases approaches x = 1.0 line asymptotically whereas the adsorption in Types IV and V levels off and approaches maxima. Not all the isotherm curves can be classified into the above five classical types. As a matter of fact, many of the isotherms encountered in practice also show a further upward turn (denoted by the dotted lines in Figure 3.4) as the saturation vapor pressure is approached (Gregg and Sing, 1967). 4-B. Equilibrium Ammonia Content in Aqueous Solution The equilibrium amronia content curves in aqueous solution at temperatures ranging from 0 to 322°C are shown in Figure 3.3. Although the sorption isotherm curves in a liquid solution cannot be classified in the category of the five adsorption isotherms mentioned above, the ammonia isotherm in aqueous solution can be investigated as an adsorption 65 isotherm in solids if the water portion in the solid adsorbent be considered the primary "sites" of ammonia adsorption. Judging from the shape of isotherm curves in Figure 3.3 at differ- ent terperatures, they look like isotherm curves in Type I Langmuir monolayer adsorption at lower temperatures and Type V at higher temperatures. Both types I and V possess a saturation value but differ at lower relative pressure ranges. Strictly speaking, isotherm curves in Figure 3. 3 are not purely physical adosption isotherms and they have no direct relationships with the (surface) adsorption in the solid. Nevertheless, satisfactory explanations can be stated indirectly if one would relate ammonia in aqueous solution to the ammonia in water-containing corn. An analog is existed between the saturation characteristics of gas ammonia in an aqueous solution under certain pressure and the saturation of active surface sites of adsorption for a porous medium. In the practical application of ammonia treatment for grains, the concentrations of ammonia in the air seldom exceed 20 percent of the saturated partial pressure of ammonia. Thus only the lower portion of the amrenia isotherm curves will be studied in this investigation. This part of the ammonia isotherms in aqueous solution at one atmosphere is shown in Figure 3.5. It is difficult to categorize the isotherm curves in Figure 3.5 in a specific type among the five classical isotherm types. A new model is proposed to incorporate the lower portions of curves at different temperatures in one model equation. A geometric equation has excellent correlation (Table 3.3) for lower concentrations of amronia in the air: y = A(T) me (3.11) Amount odsor bed 66 b. J I."‘ l b [11 H U «:8. Relctwe oresswe (pip?) Figure 3-4. Five types of classical adsorption isotherms classified by Braunauer, Emmett and Teller (1938). Table 3-3. values of at different temperatures. . ters in isotherm model for ammonia solution at 1 atm.: y=Ax and corresponding correlation coefficients Temperatures (°C) ,A 0.0 1.897 4.4 1.533 10.0 1.204 15.6 .964 21.1 .833 26.7 .686 32.2 .569 1.048 1.032 1.011 .993 .983 .966 .950 Correlation .9985 .9989 .9992 .9993 .9998 .9999 .9999 67 25 0°C :N 5. 2° *- 4.4° C S We 0.0°c 8 0" g 15 _ 15.6°C E '5 211° .§ ‘6’ g 10 .. 26.7°C .m 8 g 32.2°c s: 5 " o l l L 0 5 10 15 20 x,partial pressure of ammonia in the air at 1 atm, volume percentage, w.b. Figure 3.5. Ammonia isotherm curves in ammonia-water system at low concentrations . where, y = percentage ammonia concentration in aqueous solution, g-mym0€-%0 x = percentage amrenia concentration in the air, ml -— NEE/100 ml - bulk stream A = size parameter, function of terperature B = shape factor, function of terperature The value of A defines the "height" of an isotherm curve (lower terperature curves have larger A values). The shape factor B defines the concave or convex shape of the isotherm curve. The functions of A and B are solved with high correlation coefficients by plotting them against terperature: -l . 155 A(T) 108.0 (1.8 T + 32) B(T) 1.0442 - 2.974 x 10"3 T The final equation for the ammonia isotherm is: y = A(T) xB(T) for 0°C < T < 32.2°C and x _<_ 20. 4-C. Equilibrium Ammonia Content in Corn (3.12a) (3.12b) (3.13) Based on the isotherm model for aqueous ammonia solution (Equation 3.13), the isotherm equation for ammonia in corn was developed based on the following assumptions: 1) The only active component in corn responsible for ammonia: adsorp- tion is the water portion of the grain. 2) All the water, which can be reroved in the standard oven-heating moisture content determination method (103°C, 72 hours for corn), 69 has the same capability of absorbing ammonia (no matter in what form the water exists). The ammonia concentration in corn can be expressed as: 8 = x2 y Mdb (3.14) where, x2* = percentage amlonia concentration in corn, dry basis M db = moisture content of corn, dry basis, decimal. The isotherm equation for ammonia in corn at low relative pressures can then be obtained by substituting Equation 3.13 into Equation 3.14: x2* = M A(T) xB(T) db (3.15) for cereal grains at terperatures between 0 and 32.2°C. An example of the isotherm curves for corn at different moisture contents at 4.4’C is shown in Figure 3.6. The figure demonstrates the influence of moisture content and ammonia pressure on the ammonia concentration in shelled corn. Equation 3.15 is semi-erpirical. The data for ammonia concentra— tion in aqueous solutions and the ammonia partial pressures has been tabulated by Liley and Gambill (1973) from several sources. The ammonia isotherms for corn are derived based on these erpirical results and the assurptions described previously. 4-D. Hysteresis and Residuals When the multilayer adsorption on the surface of a non-porous solid (Types II and III) becomes indefinitely thick, it is a sign of condensation. If, however, the solid adsorbent is porous, the thickness of the adsorbed layer on the internal surfaces of the pores is limited by the width of the pores. The shape of the isotherm is modified x *, ammonia content in corn, percent,d.b. 7O 12 10 — .30 w.b. 8 )— .25 w.b. 6 .. .20 w.b. 4 _- .15 w.b. 2 )— ' l 1 l l 0 0 5 10 15 20 x, relative partial pressure of ammmia, percent Figure 3 . 6. Ammonia isotherms of shelled corn for different moisture content at 4.4°C. 71 correspondingly: a Type II changes to a Type IV isotherm and a Type III to a Type V isotherm. The specific surface and the pore size distribution of the porous adsorbent can thus be determined by consider- ing isotherm Type IV and Type V characteristics (Greg and Sing, 1967). Hysteresis is one phenomenon of porous adsorbents closely related to the size and configuration of pores. 4-D.l. Hysteresis The hysteresis loop happens when there are two values of concentra- tion of adsorbate in the solid (x2*) for any given value of relative partial pressure of the adsorbate (x). The value of x2* is always higher for the desorption than for the adsorption branch in portion of the sorption isotherm curves. The cause for hysteresis has been hypothesized as a capillary condensation phenomenon in "ink bottle" shaped pores by Kramer (1931) , McBain (1935) and others (880, 1941; Katz, 1949). When the desorption branch is traversed, evaporation cannot occur according to the adsorp- tion curve since the neck of the pore is blocked by a meniscus which can only be evaporated at lower pressures depending on the contact angle in the meniscus; the entire pore then empties at once. It is a common practice for most researchers to assume the adsorption, rather than the desorption, branch to represent equilibrium (Gregg and Sing, 1967 ) . The degree of departure of the desorption curve from the adsorp- tion curve under equilibrium conditions for a given terperature depends on the pore size, distribution, and characteristics. The "ink bottle" hypothesis has led to the investigation of the so—called scanning phenomenon of the hysteresis loop (Rao, 1941). It was established that the loop may be crossed by moving from the 72 adsorption to the desorption branch but not vice versa. The results in Figure 3.7a show the crossing of loops on the desorption curves when the desorptions happened at different pressures. As the pressure was increased during the desorption process, readsorption occurred and followed different paths under different pressures within the hysteresis loop. The readsorption paths do not cross each other as shown in Figure 3.7b. 4—D.2. Residuals In a purely physical adsorption model for porous biological products , the chemical reactions between the adsorbate and the components in the solid adsorbent are not considered. Hewever, in the case of amronia treatment for corn, browning of ammonia-treated corn has been reported (Lancaster et al., 1975). The browning reaction is similar to that of the browning of aged corn where it is interpreted as an example of the classical nonenzymatic browning reaction (Reynolds, 1963). The browning reaction involves the reaction between an aldose (a sugar) and an amine through a series of chemical transformations producing piglented and sometimes highly flavored compounds. Certain kinds of nitrogen containing compounds, such as guanidine and pyridine, may be formed by the browning reaction of corn with amronia (Iancaster et al., 1975). The chemical reactions and their reaction mechanisms between ammonia and certain corn components are not fully understood. It has been known that the armenia "adsorbed" by corn (referred -to as the total ammonia) contains both volatile (free) and nonvolatile (fixed) ammonia. Free ammonia is water extractable and the amount can be easily determined by a titration method. The gaseous amrenia diffuses 73 3° 1 I f I V l l l l I l l 20 25 3O 35 Pressure (ton) 3° I I r r l I Weight of water adsorbed per mm; gel 35 Pressure (ton) Figure 3.7. "Scanning" of the hysterisis loop in the adsorption of water by titania gel. The loop may be crossed by moving frcm the adsorption to the desorption branch, but not vice versa (Rao, 1941). 74 into the pores of corn surface which composes most of the free armenia and will eventually become either fixed or lost. During the desorption process of ammonia-treated corn, only the free ammonia portion is removed. The free ammonia in the pores and on the surfaces of corn is being dissolved in the water portion of corn. The free ammonia, in the pores and in the corn, reacts chemically with certain corn components continuously and becomes the fixed ammonia and is referred to as "residuals". The residuals are dynamic quantities and they are temperature and concentration dependent . 4-D.3. The Combined Effects of Hysteresis and Residuals on the Desorption Process Experimental results of the desorption of ammonia in corn are not available in the literature. Thus the formulation of a desorption isotherm has to be based on certain assumptions: 1) Corn is a porous medium and the sorption of ammonia displays hysteresis above certain ammonia partial pressures in the air. 2) The chemical reactions between corn components and ammonia consume ammonia in an irreversible manner; the reactions may take place both on the pore walls and in the solid. 3) The free armenia in the pores is considered the adsorbed ammonia which is dissolving into and reacting continuously with water and other components in the corn . 4) The fixed amrenia (or residual) is counted as part of the total ammonia adsorbed in the corn but only the free ammonia is responsible for the physical adsorption and desorption processes. 5) The amount of residual is assumed to be a known and fixed percentage of the highest free ammonia concentration (x2f*) the solid 75 ever possessed. The proposed adsorption and desorption models for ammonia in a pore on the corn surface are shown graphically in Figure 3.8. The corresponding isotherm curves are shown in Figure 3.9 where the approximate positions corresponding to the intermediate steps in adsorption and desorption are indicated. The adsorption of ammonia in an ammonia-free "ink bottle" pore (A1) at low pressures starts with an even coverage of free ammonia on the pore wall (A2). The free ammonia on the pore surface builds up and at the same time is dissolving into the water portion of solid corn as free ammonia. Meanwhile, the dynamic processes of chemical reaction between free armonia and certain corn components both on the pore wall and in the solid are progressing. As the above processes continue, the depth of the deposit of free ammonia becomes deeper (A3). More free ammonia is dissolving into the solid. The entire pore volume is finally filled with adsorbed free armouia (A4) and saturation is reached. As the ammonia pressure in the air is reduced, the originally saturated pore (D1) starts losing free ammonia. The bottle neck of the pore resists evaporation of free ammonia in the pore (D2). The solution of free ammonia from the pore to the solid is stopped and may revert its direction - the free ammonia in the solid may enter the pore again due to reduced pressure. The "ink bottle” is suddenly emptied at once (D3) when the partial pressure of ammonia is low enough. There is a very small amomt of free ammonia in the pore then; the free ammonia in the solid starts evaporating into the pore and leaving the pore. The equilibrium amount of the free ammonia in the solid depends on the partial pressure of ammonia in the air. At the end of desorption 76 Al A2 A3 A4 = D1 - Free armenia in a pore '3??? Fixed ammonia in corn _) Direction of free ammonia . I Corn solid free of movement in corn ammonia Figure 3.8. Proposed model for the ammonia adsorption and desorption process in an "ink bottle" pore. 77 x2f’ free ammonia concentration Figure 3.9. A4 (D1) --.—-—,- L xh 1.0 x, relative partial pressure of ammonia The proposed general adsorption and desorption isotherm curves of the ammonia-corn system for the entire pressure range . Approximate positions of the relative intenmediate steps are indicated in accordance with those in Figure 3 . 8 . 78 process, as the ammonia supply is completely stopped, there is no free ammonia in the entire system under equilibrium. However, the fixed ammonia (residual) is accumulated in the solid and becomes a permanent part of the solid (D4). Due to the variations of size and shape of the pores and other adsorption sites, e.g. tip of corn, cracks, etc., the sorption process is not exactly the same for all the adsorption sites. Therefore, the changes. shown in Figure 3. 9 for the adsorption and desorption pro—- cesses are not as dramatic as those in Figure 3.8. The effect of hysteresis is not seen until the ammonia concentra- tion in the air reaches a certain level xh (Figure 3.9) when some pores start being filled up. Since the residual is a dynamic quantity and cannot be indicated on the sorption isotherm curve, residuals do not play a direct role in the ammonia sorption isotherms which are purely physical adsorption and desorption processes . However, the equilibrium amount of residuals does depend on the free ammonia con- cent rat ion . 4—E. Proposed Sorption Isotherme in the Armonia-Corn System at Low Ammonia Pressures In the previous discussion, the partial pressure range of armonia in the air varied from 0 to 1 atmosphere. As was mentioned previously (Section 41—8), in the practical applications of armonia in grain drying in a fixed-bed system, the inlet concentrations of armonia in the air seldom exceed 20 percent. Equation 3.15 for the equilibrium 'ammonia content in corn may be applied to the adsorption in this pressure range. It has been shown by Ngoddy (1969) that at relative pressure of 20 percent, the adsorption of water vapor on corn is still monolayer 79 for moisture isotherms. Thus one would assume that the hysteresis loop does not exist below the almonia concentration of 20 percent in the air. Thus the effect of hysteresis can be neglected. Therefore, the adsorption and desorption isotherms for the ammonia- corn system at lower armronia pressures are identical. The adsorbed amount of ammonia in the solid for these physical adsorption processes refers to the free ammonia content only. The isotherm equation for free ammonia in corn at lower ammonia pressure x can thus be expressed as: XE (T) * = Mdb A(T) (3.16) x2f According to assumption 5 (Section 4—D.3) the equilibrium ammonia residual value (XZR) is a known and fixed fraction (<5) of the maximum free ammonia concentration in corn (x2f*) . The value of 6 is dependent on the temperature, relative partial pressure of ammonia (x) in the air, and the length of time the adsorbent is exposed to the ammonia pressures. If a rapid equilibrium state would be reached in practical dynamic problems, 5 is only a function of temperature. In the ambient air drying system, which is the case in this investigation, the temperature changes will not be too big to influence (5 significantly, 6 can be assumed to be constant. The value of 6 can be determined experimentally. The total ammonia in corn (x2*) can be expressed as the summation of free amrenia and fixed ammonia (residuals): x2* = x2f* + XZR = (1+6) x2f* (3.17) In the physical sorption processes, only the free ammonia content, X213“, is considered while the ammonia depletion due to chemical reactions is lumped in the residuals term, XZR. CHAPTER4 GENERALSORPI‘IQ‘IANDDRYINGMODEIS INAFIm-BEDSYSIEW 1. Introduction Grain drying in a fixedebed system is a batch drying process. The grain drying in a bin implies several different drying systems, among themnare natural air drying, lowetemperature drying, layer drying, solar heated air drying, dehydrated air drying, etc. Although the fixedebed drying systems are different in performance, they can be described by the same~simu1ation equations. The proposed model equations for a fixed bed grain drying system: have been part of the results of a series of grain drying simulation research efforts performed at Michigan State University (MSU) by Bakker—Arkena and Bickert (1966, 1967) and Bakker-Arkema et al. (1968, 1970, 1974, 1977, 1978). The following assumptions have been made in the development of the MSU grain dryer models (Brooker et a1. , 1974): 1) no appreciable volume shrinkage occurs during the drying process, 2) the temperature gradients within the individual particle are negligible, 3) the particle—to—particle conduction is negligible, 4) the airflow in the bin is plug-flow, 5) the BT/Bt and BH/at terms are negligible compared to the BT/az and BH/Bz terms, 6) the bin walls are adiabatic with negligible heat capacity, 80 81 7) the heat capacities of the drying air and the grain are constant during short time period, and 8) the equilibrium moisture isotherm and thin-layer drying equations are known for the grain to be dried. Based on heat and mass balances and above assuxptions, the fixed- bed drying model equations are written as (Bakker—Arkema et a1. , 1974): 92.: ~ha 32 (M) (4.1) Ga ca + Ga cv H h + c (H) 3H _a_ez_=p c +hoac M (T'e) ‘ofg +: c M Gal-52— (4'2) p p p w p p p w .313..- -feiu. 32" G at (4'3) a .335: an appropriate thin-layer drying equation (4 . 4) The appropriate thin-layer equation can be chosen for the proper grain and air temperature, humidity, and grain moisture ranges. There are four unknowns in the above model equations: air terperature (T), grain temperature (6), absolute humidity (H), and grain moisture content (M). Accordingly, the above four equations must be independent in order to solve them. The best way to solve these simultaneous differential equations with variable coefficients is by numerical methods and digital computers. According to the CRC Handbook (1976), the term ”adsorption" is defined as the condensation of gases, liquids, or dissolved substances on the surfaces of solids. Sorption, as defined by Vermeulen et a1. (1973), involves "contacting a free fluid phase (gas or liquid) with a rigid and durable particulate phase which has the property of selec- tively taking up and storing one or more solute species originally 82 contained in the fluid". Furthermore, in general sorption processes, "it is usually necessary to recover the solute or to purify and reuse the sorbent, and then conditions for desorption must also exist". Fixed-bed adsorption processes are applied in the areas of chromato— graphy, ion exchange, heterogeneous catalysis , and other separation processes in chemical engineering. To formulate the theory of fixed- bed phenomena requires the knowledge of phase equilibria and interphase mass transfer rates. The fluid-solid equilibria have been discussed in the previous chapter for ammonia-corn systems. Interphase mass transfer rates and other sorption characteristics will be the subject of this chapter. Ammonia drying systers differ from ordinary sorption processes in the following aspects (summarized from the previous chapter): 1) In an amrenia drying system, the solids (corn) not only adsorb ammonia on the surface and in the pores but also diffuse the adsorbed ammonia into the solid, and 2) the adsorbed ammonia reacts with the solid (corn) components, at relatively low temperatures (e.g. ambient terperatures), chemically to form residuals which are not regenerable. Adsorption usually refers to the physical adsorption process of a gas—solid system. Adsorption may occur through a chemical bonding between the adsorbed gas and solid on the adsorbing surface at terpera- tures usually higher than 204°C (Vermeulen et a1. , 1973); it is then termed Chemisorption. The residuals formed in the ammonia-corn system are apparently not the result of chemisorption since the corn temperature is far below 204°C (400°F) throughout the ambient air drying processes. 83 2. General FixedrBed Sorption Models 2-A. The Transport Mechanism The mechanism of transport consists of several distinctly differ- ent steps with each step contributing to the overall performance of the mass transfer process (vermeulen et a1., 1973): 1) Mass transfer from the fluid phase to the external surfaces of the solid particles. 2) Pore diffusion in the fluid phase within the particles. 3) Reaction at the phase boundary of fluid and solid. 4) Internal diffusion in the solid. In general, systems with a high total solute concentration in the fluid phase are more likely to have the mass transfer rates controlled by the internal diffusion (Step 4) while systems with a low total f1uidrphase concentration are more likely to be controlled by fluid external diffusion (Step 1) or pore diffusion (Step 2). The reaction at the phase boundary of fluid and solid (Step 3) exerts small resis- tance to the mass transfer since this process is usually very fast (Vénmeulen et a1., 1973). The boundary resistance is also independent of the solute concentrations since the concentration effects are mostly described by Steps 1, 2, and 4. The major difference between pore diffusion (Step 2) and internal diffusion (Step 4) is that they happened in different phases: pore diffusion occurs when the adsorbate is still in the fluid phase while internal diffusion occurs when the adsorbate has penetrated into the solid. The two steps are separated by a phase transition step (Step 3). For non-porous adsorbents, only mechanism 1-3-4 dominates since there is negligible pore diffusion process. Mechanisms 1-2-3 and 1-3-4 both 84 may happen in parallel in porous adsorbents with the faster one of the two controls the rate of sorption. Mechanism 1-2-3 dominates when the rate of internal diffusion in the solid is low while mechanism 1-3-4 controls when the rate of pore diffusion in the fluid phase is slow. 2—B. The Equations of Transport The fixed-bed sorption models are usually analyzed by describing the mass balances within a control volume of bed. Assume a fixed-bed system with constant cross sectional area perpendicular to the direc- tion of airflow which is plug-type. The packed material has a void volure fraction of e and the bulk density of pB. A mass balance within the control volume on the fluid and solid phases results in the follow- ing transport equation (Sherwood, 1975): is. is. a- e 8t ”1331: +ev 3t - 0 (4.5) where c and q are fluid and solid concentrations of adsorbate in moles per volume and moles per weight units, respectively. The interstitial fluid velocity is expressed as v and thus ev becomes the superficial velocity in an erpty tube. Equation 4.5 must be coupled with another equation describing the adsorption rate in the solid phase. The adsorption is assumed to be driven by a driving force function F (C,q) and the rate equation can be written as 3 ._ 033%»- hIn a F(C.q) (4.6) where hm is the overall mass transfer coefficient and a is the specific surface area. The driving force function can be expressed in several forms depending on the dominating transport mechanism. 85 As it was mentioned in Section 2-A that the reaction at the phase bomdary (Step 3) is not controlling; the mass transfer rate of mechanism 1-2-3 depends primarily on the mass transport in the fluid side. The driving force function is thus a function of the fluid concentration alone (Section 2-C). However, the pore concentration of adsorbate in the fluid is usually expressed by the equilibrium relationships with contacting solids and thus the driving force can be expressed by the adsorbate concentration in the solid phase. Pore diffusion will be discussed further in Section 2-D. The slower mass transfer rate of Steps 1 and 2 determines the overall mass transfer rate. For mechanism 1-3-4, the driving force for Step 1 is a function of fluid concentration while it is a ftmction of the adsorbate concen- tration in the solid alone for Step 4. Again, the slower rate of Steps 1 and 4 defines the mass transfer rate for mechanism 1-3—4. 2-C. The External Transport Equation The rate of mass transport in a solid can be expressed in terms of the concentration difference in the fluid phase if the transport rate is assumed to be controlled by external diffusion (Step 1) (Sherwood, 1975): 93 §%.= kf a (c-c*) (4.7) where C* is the interface concentration in the fluid in equilibrium with the outer surface of the solid. The fluid-phase mass transfer coefficient, kf, which carries the units of cm/hr, may be determined by general mass transfer correlations (Vermeulen, 1958; Wilke and Hougen, 1945) and expressed in terms of the fluid properties and the effective particle diameter. 86 The mass transfer coefficient was determined by Wilke and Hougen (1945) for packed-bed operations under linear-flow gas-solid contact conditions: k = 10.9 ev(l-e) ( Df )o.51 (Dr of )0.16 (4 9) f a dp Hp ev uf ° where ck) is the particle diameter; ev is the superficial velocity or volumetric flow rate of fluid per unit area of superficial cross sectional area; Df is the fluid-phase diffusivity; and pf and uf are the density and viscosity of the fluid respectively. 2-D. The Pore Diffusion Equations The diffusion rate of adsorbate in the fluid—filled pores inside the particles for spherical-shaped pores with an internal radius r can beexpressed as: %%S._Dpore(1-e) (267“? 27..) (4.10) where the pore concentration C* is determined by the equilibrium rela- tionship with the contacting solid. The pore diffusivity, Dpore(cm2/hr), has been solved and summarized for various gases and liquids by Satter- field (1970) as a function of the internal porosity of the particle, the average pore radius, the tortuosity (or, winding, twisting) and the terperature in the pore. The major disadvantage of the above pore diffusion equation in the difficulties in the numerical integration of second order differen- tial equation with a cormonly-encountered solution stability* problem. Vermeulen and Quilia (1970) expressed the pore diffusion equation by a * The stability problem in the solution of differential equations using numerical integration has been discussed in detail by Nogotov (1978). 87 modified driving force approximation involving the overall particle concentrations: dq = t k a q*'q (4 11) '3? pore pore [1+(R-l)q/q']l/T with 60 Dpo re kpore a =T-(l-e) (4.12) P where R is the separation factor defined by Vermeulen and Hiester (1954) %}:3% in describing the shape of the isotherm curves and X, Y are dimensionless concentrations referred to some reference concentrations C’ and q’, i.e. X = C/C' , Y = q/q' . The separation factor for the Langmuir adsorption is a constant while other isotherm curves are expressed by variable separation factors. For the pore diffusion case, R is equal to (l-Y*)/Y* at X = 0.5. The term up is a correction factor and equal to 0.548/(1-0452121') for R < 1. 2-E. The Internal Diffusion Equations The solid-phase diffusion equation expresses the concentration gradient of adsorbate existed within each individual solid particle. The solid phase is assured to be homogeneous, isotropic, and permeable. For spherical particles the rate of internal diffusion can be expressed by the Fick's second law of diffusion as follows: _gg_ azr gar ~ at-Dp ( -——2+r5?) (4.13) 3r where D p is the diffusivity of solid particles and Y is the dimensionless solid-phase concentration at an internal radius r and time t.. The theoretical Equation 4.13 is similar to Equation 4.10 and is difficult to solve. Solution of Equation 4.13 not only solves the concentration of adsorbate in the solid as a function of time but 88 also shows the concentration gradient within the particle which is soretimes trivial for small particles or for a macroscopic system. Thus, Equation 4. 13 is often approximated by the linear-driving-force equation of Glueckauf (1955): §1= w k a (q*-q) (4.14) t P P where q is the average concentration for the. entire particle. The specific area a+ can be expressed as 6(1-e)/dp. The mass transfer coefficient k p has units of cm/hr and was recommended by Vermeulen et a1. (1973) as: 60 D kp a =72?- , or (4.15) p k " 1o 4 16 p (1-6) dp ( ' ) . Equation 4.16 is obtained by substituting a = 6(1-e)/dp into Equation 4. 15. The correction factor up which partly corrects the linear driving force approximation is obtained from the slope of curves of the particle uptake vs. time with Equation 4.14 at Y = 0.5 (Hall, 1966). The value of up was estimated from: 1P l p 0.894/(1-0.106 R /4) , for Rl P Vermeulen (1953) proposed a quadratic-driving-force equation which better approximates the behavior of the theoretical equation (Equation 4.13) than the linear-driving—force equation (Equation 4.14). The quadratic equation is: 2 area = de = 6(l-e) volume 1 3 d E "dp / ( 1-6) p +a= 89 2 ch! _ Y* - Y2 --dt ‘Pq kp a (4.18) where l 2 u = 0.590 (1-0.410 R / q ) , when Rl (4.19) and R = X*/(1-X*) at Y* = 0.5 3. The Local-Equilibrium Theory The equations for the different stages within the sorption trans- port mechanism have been described in the previous section. The solu- tion of these transport equations requires the knowledge of the isotherm equation and the dynamic behavior of a number of parameters such as the mass transfer coefficient and the product properties. Perhaps the most difficult parameter to determine is the mass transfer coefficient which has to be calibrated for each experiment under a set of specified conditions in order to be consistent to the results from the experimental determinations which are sometimes tedious and time—consuming. A good estimation of the mass transfer coefficient under a particular set of experimental conditions can be obtained from a carefully designed and performed experiment . 3-A. Local-Equilibrium Transport Equations Consider the case when the mass transfer into and out of the solids have negligible resistance, i.e. mass transfer coefficient is infinite, then "local equilibriumH exists at all time and positions between the solid particles and adjacent fluid. Therefore, q may be replaced by q* in Equations 4.5 and 4.6 where q* is a frmction of the fluid concentra- tion, q* = f( C), as specified by the isotherm equations. The mass balance equation (Equation 4.5) in the gas-solid sorption system is then simplified to the following equation (Sherwood, 1975): 90 p mar—game) 1%+v§%=0 (4.20) where f' represents the first time derivative of the function. The solution of C is not straightforward since a variable coefficient f'(C) is involved. T‘I'e term x in this chapter refers to the bed position. If a solution C(x,t) exists and has been solved, an additional equation involving the partial derivatives of C can be formulated according to the ”method of characteristics” (Aris and Amundson, 1973): 3C BC __ ('5'?) (111+ (.33?) dX-dC (4.21) The solution is expressed by two equations for (BC/ 3t) and (BC/3x): 3C __ v dC '3'?" g(C) dx - v dt (4°22a) .32.- g(C) dC 3x - g(C) dx - v dt (4°22b) where, g(C) = l + f' (C) DE /e (4.22c) Certain characteristic directions in the xt plane (dx/dt) will make the denominator equal to zero. The nurerator, and accordingly dc must also be zero if the solutions are to be finite. Therefore, C is constant along the "characteristic lines" (Sherwood, 1975) which are expressed as: _d_X._.__V_.. dt g(C) (4.23) Since C is a constant, the slopes of the characteristic lines (dx/dt) are also constant. Thus, the characteristic lines are straight lines on the xt plane. An example of this graphical presentation for the solution of a fixed-bed adsorption and desorption problem is presented by Sherwood (1975) using the Langmuir isotherm. 91 The slope of characteristic lines implies an important property in a fixedebed systemn the speed of penetration. The penetration property is described by the concentration wave velocity, dx/dt, which is the velocity of the initial concentration wave front moving through a packed bed. 3—B. The Break-Through Curve In order to investigate the break-through characteristics of an adsorption wave, consider a colour adsorber (i.e. a deep bed of adsorbent) and a solution containing a strongly adsorbed solute at concentration Co as shown in Figure 4.1. As the adsorbent is at first contact with the solution, the solid adsorbs solute very rapidly; very little solute is left in the solution. The effluent concentration of solute is prac- tically zero until the adsorption zone starts breaking through —- Cc is the break point. After the break point, the concentration of solute in effluent rises very rapidly until the adsorption zone almost reaches the exit of the bed (Cd). The concentration of solute in the effluent becomes saturated to the input concentration asymptotically. The break- through characteristics are usually investigated using constant input adsorption process. The steepness of the breakethrough curve depends on the sharpness of the adsorption zone in the packed bed which is in turn influenced.by the solution feed rate, packedébed.properties, mass transfer rate, etc. The adsorption and desorption curves derived fromlthe local- equilibrium theory are shown as solid line segments in Figure 4.2. Bring the time required for the wave to emerge from the bed of depth x, the total amount of solute flowing into the bed is'vaet; none flows out. The amount of accumulated solute in the bed is eCOx + quO*x. 92 (a) (a) (c) (d) Feed solution ' COflCCO'fO'IOfl - Co Ads 2009 .1. Effluent 1 concentration II C, Co Cd / Break-through curve Concentration of source In emuem C: Ca Ca __/ volume of effluent Break point Figure 4.1. The adsorption wave (Treybal, 1968). 93 = 00 '3 Local equilibrium theory / a.” m / / '5 /hmJ// /h / 2 a 8 / w / 0‘ ’//f,// tb Time (a) Adsorption c , ef fluent concentration (b) Desorption Figure 4.2. Influence of mass transfer coefficient on the adsorption and desorption concentration waves . 94 Equating the two and solving for the break-through time, tb, gives: 0 q * _ x B 0 In the case of k-wo which is implied in local equilibrium theory, the concentration of the effluent is a step output at a discontinuous point of tb' For points between A and B in the desorption curve (Figure 4.2b), the concentration in the effluent can be obtained from the following equation by solving C contained in the ftmction f' (C): 25.: t l + (OB 7,8) 9155' Actual breakthrough curves are affected by mass transfer resistance and by the longitudinal dispersion effect. As a result, sharp corners are rounded off. Corparing two sorption systers under exactly the same conditions, e.g. fluid flow rate, interstitial velocity, feed concentra- tion, isotherm curves, terperature, etc., except the mass transfer coefficient, one of the round—off curves (dotted lines) closer to the local—equilibrium theory line has a larger mass transfer coefficient. A sharper breakthrough curve is acknowledged as a result of a larger mass transfer coefficient during adsorption. Consider the same bed (adsorption) capacity for these different sorption systers, the area below each adsorption curve ought to be the same. In other words, the integral JCdt is constant if integrated over the entire adsorption process. Similar cases can be expected for desorption processes with different mass transfer coefficients. The effect of the mass transfer resistance between the fluid and solid phases on the breakthrough curve was deronstrated by Sherwood (1975). The slope of the breakthrough curve at the midpoint (where 95 C =Co/2) is: d(C/C ) = hm a x . evCo . KApo/«l (4.25) t 0/2 a v quox 1+KApo/2 using a linear function (C—C*) as the driving force and C* computed from the Langmuir isotherm. The factor (hmax/Ev) is defined as the umber of transfer units (NTU) in the bed and (quOx/c v Co) is the time required for the input flow to supply the bed with an amount of adsorbate equal to the bed capacity (TBC). A larger value of NTU and a sraller TBC value will contribute to a steeper slope of the breakthrough curve. Weyde and Wicke (1940) observed a bed of carbon particles in adsorbing carbon dioxide from nitrogen. The effect of depth of packing on adsorption and desorption is shown in Figure 4.3 using a superficial fluid velocity (CV) of 4.26 (rm/sec. The breakthrough adsorption curves showed a virtually constant pattern after a bed depth of about 18 cm. However, the slope of these curves at the midpoint C = CO/2 is decreasing as the bed depth increases which is consistent with the results from Equation 4.25. Similar arguments can be stated for desorption. 4. The Thomas Solution of Break-Through Curves Thomas (1944) derived the driving force function from the stoichionetry of the monovalent ion-exchange reaction and defined the so-called "kinetic driving force": F'(C.q) = C(l-q/qm) --§- q/qm (4.26) where K is the equilibrium reaction constant which can be calculated when the driving force is zero, or - (C -C) K “37W (4.27) The transport equation becomes 3 . 037%= K a F (C,q) - (4.23) The mass transfer coefficient hm in Equation 4.6 becomes K (kappa), the kinetic coefficient which can also be shown to be related to the mass transfer coefficient which represents the diffusional resistance in the fluid and solid phases (Sherwood, 1975). Thomas (1944) also recognized that the rate aq/at includes the specific area since the actual rate is almost always diffusion-controlled and depends on the surface area available for interphase transfer. Redefining the time scale as "the elapsed time" at a point after the fluid has arrived at the point: t = t-x/v, the transport equations (Equations 4.5, 4.28) are then simplified to: o 3q .5... 30 .. 8 3‘6 + V 56.— 0 (4.29) o -3—q-= K a F" (Qq) (4.30) B at Introducing a transformation of the dependent variables , Thoras (1944) solved these equations by reduction to a linear equation. With, the boundary conditions C(o,t) = C0 = constant and q(x,0) = 0, the concentrations are solved as: C‘,/Co = J( n/K, n Tr )/ H' (4.31a) <1/qm = [ 1-J( n Tr' n/K )1/ H' (4.311») where H' = J(n/1<, n Tr) + [1-J(n, nT' r/i<)]exp[(1—-K'1) (n-nTrH (4.32) n = Keavx = dimensionless distance or NTU (4.33). T=eant ___t (4.34) r qm pr if 97 n T =K—-a—E-9-—t—= dimensionless time (4.35) r qm pa J(d,B)= 1 - e B joe'"S I, (N's—'5 ) at (4.36) 10(2/855 is the modified Bessel function of order zero of the first kind (Ozisik, 1968). The values of the J function are shown graphically in Figure 4.4 (Vermeulen et al. , 1973). Approximation equations have been developed by Thomas ( 1944) for large 018 values: 2 t "0‘ 4?) 1 (4.37) (Hon 8) 2—[1- erf*(r -/5-)]+ 2 ((010)17‘4/8' The error is less than one percent when 018 > 36. Ledoux's ( 1948) approximation further neglects the second term on the right hand side of Equation 4.37 when 018 > 3600: 05erfc(/E-/8—) ,ifB<01 noes) = { ° (4.38) 05 [1+ erf(/' -/E‘)] . if B>a In the case that both concentrations in the fluid and solid or the change in concentration of fluid, are slall enough, the equilibrium between the gas and solid phases behaves according to Henry's law: C = (Co/gm) q = (const.) q _ (4.39) The above relationship makes the equilibrium reaction constant , K, equal to unity. The driving force becores a linear expression: 5" (C.q) = C - (C°/q'm) q (4.40) The solution of Equations 4.29 and 4.30 is thus simplified to: 2 * The error function is defined as: erf x =/?'_2 3e t dt and can be 3 5 X7 approximated by an infinite series: erf x =72: (x- 1.153- X__5_ -74- ' ') m l-e rxf The complementary error function is defined as: erfc x= 98 1.0 c/c. 0.5 Time. min Figure 4.3. Breakthrough curves for carbon dioxide on solid carbon using bulk flow rate of 4.26 (rm/s (Weyde and Wicks, 1940) . Ha. I) °'°°'1 2 5 IO 20 so 100 200 500 1000 a Figure 4.4. Emotion used in the 'I‘hotas' (1944) solution. C/c, q/qm J (n,nTr) (4.4la) l - J(Tr,n) (4.4113) If the number of transfer units (n) is large enough and the Ledoux's (1948) approximation (Equation 4.38) is used, it is found that: C/Co 2 0.5 erfc ( V6.- /fiT;) (4.42) The slope of the breakthrough curve at any position on the curve is found by differentiating Equation 4.42: 2n 5‘) 3(C/cg) __ n Tr -n(T+1) I1‘ r r The above equation can be approximated using the asymptotic series for the Bessel function when n is sufficiently large*: 1(2nfir‘)=-l—- emfi; (444) 1 r 2/1'r'mz'rr)l’4 and Equation 4.43 becomes, nT e—( /n'- fir—)2 3.191931 = _1__£. r (4.45) at 2/1? E (nzTr3) 1/4 or simply, 2 3(C/C) ___ 1 5T 1/4 e'( 5' “ET-P (4.46) 5‘: 2.471“: r The slope at the midpoint C = Co/2 on the break-through curve is when Tr = 1 (or t = qmpr/evCO) and can be expressed as: ___—4.354 > = _1 a. (4...) Tr=l 2/1? t z 3 105 * I .z) e e [1 -—- 4L2“ ————-—1 for z>lO. l( a; 5 8 Z 128 z 1024 Z3 '- If 2 is lar e enou h, I e -—£z— from Ozisik, 1968). g g 1(z) 5&— ( 100 Upon substitution of n and t, Equation 4.47 becomes, 3(C/C) ._ 1 C W—J— T =1 - —-L— EZVKa x (4.48) . “27,—ng CHAPTER 5 DESIGN OF AN AMMONIA AIBORPTION AND DRYING MERIMENT 1. The Grain - Yellow Corn The grain for the experimental phase was chosen to be yellow dent corn since Michigan is one of the corn-producing states. One hundred and twenty bushels of sample of the variety XL-12 was used. The corn was grown at a farm at Bellaire (25 miles NE of Traverse City), Michigan, and harvested on Nbvember 20, 1978. The initial moisture content of the corn was 25.6 percent wzb. ‘with a test weight of 53.0 lb/bu. The percentage of BCFM was 0.2 percent and the loo—kernel weight was 31.8 grams. A sample of 100 gram1(dried in ambient air to 15.7 percent wub.) was tested with the Stein breakage tester for 2 minutes, 2.7 gram.of the sample passed through the 12/64 sieve. The initial grain depth in the bin was 183 cm (6 ft). The mold counts for freshly-harvested.samples were very low and were assured to be negligible. The sample in the bin was assumed to be homogeneous and isotropic with respect to the heat and mass transfer processes. The initial viability of the sample was found to be 90 percent by the tetrazolium (T'Z) solution test. The tetrazolium test 'is widely recognized as an accurate means of estimating seed viability (Copeland, 1976). The TZ test distinguishes between viable and dead tissues of 101 102 the erbryo on the basis of respiration rate in the hydrated state. The test utilizes the activity of dehydrogenase enzymes as an index to the respiration rate and seed viability. The colorless tetrazolium salt solution in the soaked corn is oxidized into red formazan which is examined visually . 2. Experimental System In order to investigate the ammonia adsorption behavior in a fixed bed of corn, an experimental system (Figure 5.1) was constructed which includes the following subsysters: 2-A. The Bin Structure A 174 cm (5.7 ft) diameter bin with a capacity of 4.23 m3 (120 bu) was used. A special bin roof was constructed to confine the airflow of the exhaust stream. A regular 4-inch house dryer hose was exployed as the exhaust pipe. The bin height is 183 on (6.0 ft) and the cross sectional area of the bin is 2.38 m2 (25.6 ftz). The interior of the bin was painted with ”oleum" paint to prevent possible corrosion of the galvanized steel by ammonia. Nofsinger et al. (1976) found that ammonia at concentrations under 0. 1 percent in the air was not corrosive to unprotected galvanized bin surfaces. The entire structure was caulked to prevent leakage of ammonia from the bin. The bin is located inside a metal-structure shelter in which the auxiliary equipment is located. 2-B. The Fan A 37.3 w (1/20 hp) tubaxial fan (blower) was employed. The static pressures throughout the bin were determined using a manoreter. The pressure drop was 22.07 N/m2 per meter (0.027 inches of water per foot) 103 l 0 g H m m 59503 :1 11-1111)!!! I - . A ma . x V 2 «”92 find .I l lumfiucn mange anagrams. I III... IlllrrllL _ n _ _ _ _ _ - ‘ _ 1 I \ — \ E. r i this _ _ a mum» .2 on: I 1 q _ o u 8 I. _ a...)\ . a . ii. _ _ g =.. mm 853on 0.33m _ a... It - elm _ _ - €352 :t), . A R] _ _ refines chad - . en a: ,- 4 _ _ . was» Momma 3..-- 6 d1 _ _ r _ , oa _ _ - I: _ _ _ _ _ _ _ _ _ _ _ _ _ _ . 104 of grain which converted to 1.43 emu/m2 (4.70 cfm/ftz) from Shedd's curve (Shedd, 1953) for 12.4 percent w.b. moisture shelled corn. Thus, the total airflow rate was about 3.4 m3/min (120 cfm) or close to 1.13 cmm/ton (1.0 cfm/bu). The terperature rise due to the motor heat of the running fan was less than 1°C. 2—C. The Amronia Application System 2-C.l. Armonia Tank A 15-gallon anhydrous ammonia tank was used. The ammonia tank was placed on a scale with 4—ounce divisions which allowed a weight estimation to an accuracy of :1 ounce. The ammonia tank is a regular farm—type anhydrous ammonia tank. 2—C.2. Ammonia Flow Controlling System The most important corponent in controlling ammonia flow is a stainless steel needle valve. The desired ammonia flow rate in the experiment of 0.23 kg/hr is too low to be regulated by the crude evapor- ation valve normally attached to an ammonia tank. At a flow rate of 0.23 kg of ammonia per hour, the flow rate must be measured in gaseous form. Therefore, a miniature needle valve was used to control the gaseous ammonia flow rate. The flow was indicated by a rotometer attached to the needle valve. The needle valve is made of stainless steel to prevent ammonia corrosion. The valve has a teflon packing, is operable between ~54 °C and 232°C and can withstand a pressure up to 5000 psig (Linde, Specialty Gaseous and Equipment, Vol. IV, Union Carbide, 1977). The liquid anhydrous armenia in the tank vaporizes and escapes through the pressure valve at the top of the tank. The rotometer also has a needle valve, 105 but it proved to be too crude and not sufficiently stable to control the ammonia flow. The stainless steel needle valve constitutes the main controlling valve to limit the flow. During operation, the openings of both the pressure valve on the ammonia tank and the needle valve on the rotoreter were kept as small as possible to prevent excess leakage of ammonia. The rotoreter was calibrated at constant flow rates to indicate the true ammonia flow rate. There are ten divisions on the rotometer scale. It was found that at scale 8, the ammonia flow rate is 0.23 kg/hr. The ammonia flow rate was found to be proportional to the scale number. 2-C. 3. Terperature Recording System The temperature recording system consists of a Fsterline Angus (An Fsterline Company) digital thermocouple recorder (Model PD2064, Key programmable data system) and a paper tape puncher. The recorder can register up to 60 thermocouples with special attachments. Normally, 24 points are registered and programmed. The recorder can be used with different kinds of thermocouple wires by changing the correction factors. The results for each indivi- dual channel can be programmed to display in degrees Centigrade, Fahren- heit or in millivolts. The recording time interval and channel scanning interval can be prograrmred and an ID ntmber can be assigned to any special alteration in operation during the experiment . If the channel scanning interval is set to zero, the terperatures at all the thermo- couples are recorded at the same moment and printed accordingly on the printer . The conventional thermocouple recorder equipped with a mech- anical moving pen is unable to do this. 106 The paper tape punching machine punches the results in black paper tape in computer coding which can be transferred and stored directly in any computer system. 3. Sampling Systems 3—A. Grain Sampling A grain sampling probe with ten corpartments was used to collect samples from one to five feet levels at one foot interval as shown in Figure 5.2. Each grain sample was assigned to two compartments. Besides these five samples, the top level sample was gathered by hand and the bottom level sample was probed from side holes. This was found to be the most efficient way of collecting samples for each foot level and also minimize the disturbance and mising of the grain in bin. The major disadvantage is the closing of the 10 holes manually while the probe is in the him not only breaks corn but also needs strength to handle it. The grain samples were transported as soon as possible to the laboratory for immediate treatments. 3-B. Air Sampling To gather air samples is always a problem since it is difficult to contain the air in a container for further treatments without induc- ing experimental errors. The most direct way of analyzing air samples is gas chroratography (GC) (Lancaster et a1. , 1975). In the GC method, the air samples are collected Lsing an air-tight syringe and injected into a GC machine equipped with a specified colum (Porapak Q, N, or S, suggested by USDA Laboratory, Peoria). Helium is suggested as the carrier gas. This method is one of the most accurate methods. However, there are disadvantages: (l) the GC method requires 107 sam-l- __g/ /14{é 1* //// / 47,1? 4% Perforated floor Corn sample .83 m(6 ft) 1.52 m(5 ft) 1.22 m(4 ft) .92 m(3 ft) .61 m(2 ft) .3l m(l ft) 0 m(O ft) Figure 5.2. The position of a lO-compartment grain sampler in the grain him during sampling. Samples from 1 t0 S-ft levels are collected in the grain sampling probe while the G-ft level sample is collected from the top of the grain mass manually and the bottom level sample is collected by side-probing. 108 a reliable, accurate, and precisely calibrated GC instrument equipped with a thermoconductivity detector (most of the cormonly used GC instru- ments in biological laboratories are equipped with flame ionization detectors), (2) the column needs special packing instructions with respect to packing material, packing density, length and size of column, in order to effectively separate the corponents in air samples, (3) the operation requires constant attention, and (4) the method can only analyze one sample at a time, since the next sample has to wait until the previous one has run through the entire GC colum. The throughput time ranges from several minutes to one hour depending on the sample type and the helium flow rate. Due to the mavailability of a reliable and properly equipped GC instrurent, an alternative way of collecting air samples was developed. ' The problem to be solved was how to collect a known air sample at six different depths in the bin at the same time. A miniature pump was used to purp the air out of the bin for a fixed period of time. Each air outlet of the purp was submerged in a bottle with a measured amount of distilled water. Since ammonia is the only corpoment in which the concentration is concerend and ammonia dissolves readily in water, the ammonia concentration can be determined by direct titration using a standard acidic solution and a pH meter. The major advantages of this method are: (1) only inexpensive, easy to install, unsophisticated instruments are needed, (2) al six samples from the different levels can be collected at the same time, (3) it is realtively easy to collect samples, (4) the method is reasonably accurate if carefully calibrated, and (5) the final titration determination method is straightforward. 109 The miniature pumps used in this experiment were comercial fish- tank pumps. The pumps had to be modified since a regular fish—tank pump does not have a fixed inlet air source. The pumps were sealed externally except for the outlet air opening. An inlet hole for air was drilled on the opposite side of the plastic wall of the purp. The air inlet to the pump was connected with a piece of tycon tube placed in the bin horizontally and tied to the cable support for thermocouple wires to hold the tubes in position. The portion of tycon tubes in the bin was drilled with holes to minimize the pressure gradient during the air intake. The fish-tank pups were tested in the laboratory and found by the water displacerent method to pump an average of 1.4 liters of air per minute. A standard procedure employed throughout the experiment consisted of using 100 ml distilled water for ammonia collection and aerate for 10 minutes. Thus 14 liters of air passed through water. This procedure was calibrated with a known concentration of air samples. The correc- tion factor is assumed to be the same for all stations. 4. Weather Conditions During the Experiment The tests are split into two parts because the severe 1979 winter weather in Michigan interrupted the experiment. The first part was extended from Noverber 13, 1978, to January 10, 1979. The average air temperatures, relative humidities, and the measured average grain terperatures are shown in Figure 5.3. The measurerent of relative humidity was interrupted when the air terperature dropped below freezing. An average relative humidity was assured to be 70 percent for this period in the simulation. On January 10, the ambient air temperature decreased to below -12°C as did the average grain temperature. The fan was 110 .mevucoae zepmc teenage mmezuoaoeemu can we» con: omuasceouce mo: sueeee=e m>euo~ma so “ameunzmeoe one .acoeeeooxm so some amass we“ not moaaueaooeou cream omoeo>o eocamooe new one appose=e c>eaopoe.oc=ueamaeou new newsman omeso>o one .m.m deemed memo 2H NEE. o o o o o o o _ oo mm V 00 me _ 00 91 on Am 00 NN on ma oo_m oo flu 1.fl a, A . . U .o.\..\. x: 0 / a..a . oeauecmoeoh eseac a. .4 ,. .1 .1 gr. ) J D J)... .. . . .. 3 .. .1” a) 0 . oeaueeoeeoe acopee< e. a mu .. ,. .. .I ..s .. U l .. ‘_ rhUIl m: r1 H An Id nus: +133 (End Hn uu an 11 OH 0 ru- 1 H mm as .1 . m IA .3525... 3322... 0 r. .t no .a: nu 1... i.e....“ 00' 111 turned off and was not turned back on until April 4, 1979, when the anbient terperature had reached 4.5°C. The second part of the test was conducted between April 4 and April 24, 1979. The average ambient air temperatures, relative humidities, and the measured average grain temperatures during the experiment are shown in Figure 5.4. The average grain tenperatures follow the anbient air temperatures very closely throughout the test periods. The local amplitudes of the variation of grain temperatures are less significant than those of the anbient air temperatures as indicated in Figure 5.4. 5. Timetable of Fan Qaeration and Amnonia Application 5-A. Part I. Trial Period (ll/13/78-1/10/79) The fan was started at 8:30 a.m. , Novenber 13, 1978, inmediately after the bin had been loaded. The anhydrous ammonia was applied at the rate of 0.23 kg/hr (0.5 lb/hr) (a rate suggested by the researchers in North Regional Research Laboratory at Peoria, Illinois , Eckhoff and Bothast, 1978). Although 0.05 percent* ammonia is sufficient to kill microorganism in 27 percent moisture (w.b.) corn, it takes about 0.5 percent anmonia to control molding during long-time preservation and in-bin natural air drying (Bothast et a1. , 1975). Thus, for 4.23 m3 (120 bu) of corn, 12.9 kg (28.4 lb) (4.23 m3 x 609.3 kg—dry corn/m3 x 0.5% = 12.9 kg-NH3) of anhydrous ammonia should be adsorbed by com during the entire treatment. The amount of amnonia actually adsorbed by the corn permanently and the amount of ammonia applied through inlet air duct can 'be very different. The difference depends on the following factors: * All the amnonia concentrations referred to in this thesis are based on the dry weight of corn unless otherwise mentioned. 112 .ucoevgoaxo we when ecooom as» so; mogzwagmnEou :wmsm ommgm>m cmgzmmos as» can movupuP23; o>_uumg .mmgapmgoQEmu Lam mmmno>o och .ezm.og=mpu m>¢o 2H mzHF 85W 85m 84m 85m 8&3 85 8. 3 5 . 3 a: [I .. ‘ .4 3 \\«. 3 ogzoagoaeoh cameo meagogmasmh « J : g_< p=m_ss< . . - - - .. snaeasaz m>apa_mz am DZ'O ov~33 AlIUIHnH HAIlHWBH l 09'0 08'0 (30' I T I sz SI oo'BHfllHHHANBl SE S? 113 ( 1) the rate of amnonia application, (2) the rate of amnonia diffusion into the micropores on the surface of the corn kernels, (3) the rate of adsorption of ammonia in the pores and surface area of the corn kernels, (4) the amount of amnonia loss in the exhaust air and, (5) the rate of amnonia leakage of the tank and the hook-up jmctions. The above factors are not totally independent. The rate of anmonia supply in the inlet air determines the ammonia concentration for diffu- sion into the micropores and the rate of adsorption. The losses of amnonia in the-exhaust air increase with the increase of rate of amnonia application. Although the decrease of the ammonia application rate reduces the losses of ammonia, the effectiveness of ammonia in prevent- ing molds depends on the adsorption of a sufficient amount of anmonia 'within a limited period.of time. The deterioration of biological products is a continuous process. Thus, there is a trade—off between increasing molding risks and increasing amnonia losses. The timetable for ammonia application and the anbient air temperatures in the first part of the experiment is shown in Figure 5.5. The ammonia was applied at a rate of 0.23 kg/hr for 8 hours after which the rate was changed to 0.057 kg/hr (1/8 lb/hr). A strong amnonia odor could.be smelled at the top of the bin after about six hours of amnonia treatment at 0.23 kg/hr. The reason for the rate change was to reduce the rate of ammonia sorption and the losses to the environment. This high-rate initial injection of anhydrous ammonia to a batch of freshly harvested corn is important in reducing or eliminating the initial population of contanunants, e.g. bacteria, fungi, and other :mdcroorganisms, and in preventing further development of contaminants. This ”dying" process for the entire grain mss, especially the top 114 $2398 one no name umem on» 5 chafing—B paw ucoaafi on» 98 :oflmofimmm MEQEB How 3939.3 ma. .m .m 9.253 366955 T Mimi: .IwTIII QOQHHIITI EEhTL—TIII mQflth— ma NH NH a o m o 3 Name 3 Se o3 /4m//7//_ Eu? 2.. /—- .221_., m In I 1L1 flan-bx ’ age; uonewonuv 0 mafia; Ham ”Ewan—Em O H In H 1 3° ' annex-aims: 1T? quetqmv In I 115 portion of the grain, is very effective if it can be done before signi— ficant deterioration occurs. The rate of 0.057 kg/hr is to maintain up the ammonia concentration while ammonia is being adsorbed by the corn. The ammonia flow was stopped after five hours at the rate of 0.057 kg/hr because some mistakes were identified in the procedure of determining total ammonia. It was not until 36 hours later that the ammonia flow was resumed at the rate of 0.057 kg/hr. This ammonia flow was continued for 31.5 hours. When the ambient temperatures at night in mid-November dropped to aromd —6.7°C (20°F), the ammonia application was stopped since the adsorption behavior of ammonia in corn at temperatures below the freezing point of corn is unknown. However, the fan kept on running to dry the corn while the corn was ”protected" by a ammonia coating and low temperature. At the end of this part of the experiment, 3.89 kg of ammonia had been applied and the fan had been blowing for 56 days and 16 hours. 5-B. Part II. Constant Rate Periods (4/4/79-4/24/79) Throughout the entire period of this part of the experiment , the ammonia flow rate was fixed at .113 kg/hr as shown in Figure 5.6 in order to study the constant armonia flow sorption processes. The fan was started at 11:30 a.m. , April 4, 1979, when the ambient temperature had climbed to 4.5°C. The amronia was applied four hours later. Visible mold was seen on the surface of the corn kernels at 15-30 cm depth below the top bin surface. The molding was not serious since the grain temperatures seldom had exceeded 4. 5°C throughout the "cold storage" period when the fan was off. However, the molding is expected since the top layers received the least amount of ammonia and remained at the highest moisture content (about 25% w.b.) all the time. 116 . .955? 2.3 we puma 9508 on» :w 83mg H3 #2395 on» can sofloofimmm MES—5 How canopmfifl ooh. .m .m magma meme can mama fl}. mg. , bask 3} mg. mg. 2). x QE 0} m} , is ./ _. //. _ .. ://_ .. // // // . u u u. /mfl.5.EB mg n " cage-Em _ Hi 3 3.4 . E $.m/ . 4 8. e _ $2-9. V ._,///// m n V/ n L 3:. . . _ om ma my me maue . . e m N _ omnma omum omuaa :0 com 30 com so com mmnam we I o In W. .. ...... m can?» ham €385 m z I e x 3 m e . .4 me “a 117 After 36 hours of amnonia flow, the application of amnonia was interrupted because of renewed cold weather. Five hours later, it was determined to shut off the fan because the ambient terperature had dropped to about —6.7°C. It is assured that no moisture migration and armenia transfer in the bin occurred while the fan was turned off. The fan was not turned on until 4:30 p.m. , April 9, when the ambient terperature had reached around 0°C. The amrenia flow was resumed at 10:00 a.m., April 12. A series of adsorption tests were performed to determine the ammonia profiles in the air at different levels of the bin. The tests were performed by taking air samples oontinuosuly from six different levels in the bin. Each sarrpling process took 10 minutes; ten sets of samples (six samples per set) were collected. The ammonia was applied at the rate of 0.113 kg/hr (0.25 lb/hr) for another 46 hours immediately after these tests. The ammonia treatment was then discontinued for about three days because of illness of the technician in charge of the chemical analysis. The amnonia test was completed at 11:00 a.m. , April 19, 1979, after an additional 44 hours of ammonia application. A series of desorption tests similar to the adsorption tests were performed imnediately after the ammonia flow was shut off. The fan was turned off at 12:00 noon, April 24, after the final samples were taken. The purpose of these desorption tests was to study the rate of amnonia removal in the air from the bin. In this second part of the experiment 14.51 kg of ammonia was applied at the rate of 0.113 kg/hr; the fan was operated for 384 hours. 118 6. Procedures of Sample Analyses 6—A. Determination of Free Ammonia (Black, 1978) 1) 2) 3) 4) Place approximately 20 g of whole kernel corn in a 500-ml glass bottle containing 300 ml of distilled water. Cover the bottle, let it stand overnight. Titrate a portion (known amount) of the solution with standard sulfuric acid (0.02NI-12804) to a pH of 5.6 using a pH meter. Method of Calculation: or - LAml) (N) (MW) (100%) 300 ml 0° free m3) ‘ (1000) (g) (1-MC) b ml where, Aml (ml 0.02N H2804 needed for sample) -(ml 0.02N H280 4 needed for blank corn sample) N contration in normality of standard acid g = sample weight, grams MC = moisture content of sample, decimal, dry basis b = portion of the solution using in titration. 6—B. Determination of Total Ammonia (Black, 1978) 1) 2) 3) Put the whole kernel corn sample in a 65 ml bottle, fill to the top to reduce headspace (about 25—30 g), add 1-2 ml of concentrated hydrochloric acid. Shake the bottle, allow to stand for an hour and then spread the sample on a paper towel for air drying ovemight . The dried sample is ground; arormd 5 g is ground (or chopped) for Kjedahl distillation. . Place the ground sample in a Kjedahl distillation flask, add 200-250 ml distilled water, more than 2 g carbonate free magnesium oxide and 2 or 3 drops of defoamer (Titon X-100). 4) 5) 6) 119 Distill about 150 ml water into the distillate receiving bottle with 100 ml of 2 percent boric acid (mixed with indicator, as described in Step 5) . Titrate with 0.1 N HCl until the solution turns from blue to red. * Preparation of Boric Acid Mixed with Indicator Dissolve 2 g of Bromocresol green and 0.4 g of methyl red indicators in one liter 95% ethanol. Mix 4 ml of the above indicator for every one liter of 2% boric acid solution. The mixed solution is red colored. Method of Calculation (0.1N)[(ml 0.1 N HCl in sample)-(ml HCl Blank)]xl7x100% (1000) (sample wt. in g) (I’Mcd.b.) Total % NH3 = 6-C. Determination of Mold Count (Speck, 1976) l) 2) 3) 4) 5) Weigh 25 g (220.1 g) of a representative portion of the corn kernels aseptically into a tared sterile blender cup. Add 225 ml of sterile dilutant (phosphate buffered) and blend 2 minutes at low speed or until well blended. Plate 0.1 ml of this solution (or 10'2 g sample) on acidified potato dextrose agar (PH=3.5~:0.1). Dilute 1 ml of the blended solution to 100 ml and plate 0.1 ml of the diluted solution (or 10"4 g sample) on acidified potato dextrose agar. Repeat Step 4 and further dilute the solution to l/ 100 sample concentration, or 10.6 g sample per plate. * It takes 0.1 to 0.2 ml of 0.1 N HCl to titrate from blue to red for a blank corn sample. 120 6) Incubate for 5 days at 20°C and count colonies. CHAPTERS DEVELOPMENT OF A SIMULATION MODEL FOR AMNH‘I IA AIBORPTICN IN FIXED—BED DRYING SYSTEhB 1. Introduction The treatment of grain using anhydrous ammonia in a fixed-bed drying system is a unique problem in the area of sorption processes. According to the previous discussions , the ammonia sorption process in a packed bed of grain is not the same as any of the existing sorption systems encountered in chemical engineering. The ammonia-corn system cannot be explained by a single engineering process such as ion exchange, binary separation, chromatography, heterogeneous catalysis, or chemisorption. Biological products are usually complex in composition and are "alive” since respiration and other biological, biochemical processes are still taking place. The composition of a biological product is thus changing continuously as a result of physical and chemical reac- tions occurring within the product. These unavoidable deteriorations of biological products are usually lumped into several macroscopic phenomena by assuming a black box model for the system and Just consider- ing the inputs and outputs. The input elements for the deterioration of corn are wet corn, oxygen, and environmental conditions such as terperature, humidity, and microbiological contamination, while the output elerents involve carbon dioxide, water, and energy production which favors microbial growth. 121 122 The fixed—bed grain drying model developed by Bakker-Arkema et a1 . (1974, 1977) at Michigan State University (MSU) is a dynamic and iterative model that allows calculation of grain temperature, air terperature, absolute humidity, and grain moisture content for every specified bed location and time step employing the finite differences method for solving the simultaneous partial differential equations. The MSU numerical model can be applied to any fixed-bed grain drying system: high-temperature, natural air, low-temperature, and solar energy drying. Any type of inlet, initial, and boundary conditions can be modeled. The four-equation fixed-bed grain drying model (Bakker—Arkera et a1 . , 1974) as described in Chapter 4 predicts both air and grain temperatures along with the air absolute humidity and grain moisture content. Very small computing time steps have to be used in order to avoid instability in the numerical solution. The four-equation model was found to be very inefficient in computing slow drying processes such as natural air and low-temperature drying; it would have been advanta- geous if a larger computing time step could have been used. In order to overcome the stability problem encountered in a four-equation medel using larger time steps, Bakker-Arkema et al. (1977) reduced the system to a three—equation model by assuming no difference in temperature for air and grain throughout the entire drying processes. This assumption is justified for slow grain drying processes. The major advantage of the three-equation model is the very stable characteristics of the model with respect to numerical integration. Larger computing time steps can thus be used in the three-equation than in the four-equation model . The three-equation model has also been proved to be reasonably 123 accurate in the simulation of fixed-bed drying processes. The three- equation model described by Bakker—Arkera et al. (1977) is expressed as fol lows : pp (cp +Mcwa) —+ Ga (ca + Hcv)3 —-+ Ga [(cW -c V)(100 -T)+hf g]---- :2 =0 (6.1) 8M _3_H pp at + Ga— 32 =0 (6.2) p 3»; .m (M MM, t) <63) where rm is the proper thin-layer drying or absorption function . Never— theless, the author believes that the first term in the bracket of Equation 6.1 should be Cv(T-e) instead of (CW-CVXlOO—T). No significant difference is anticipated since both terms are very small compared to the latent heat of evaporation, h f g' 2. Assumptions Qle of the characteristics in developing a nurerical simulation model for a complicated dynamic system is that the model must be based on some assumptions which idealize the system to a more "solvable" problem. The roles of assumptions cannot be overlooked in the derivation of model equations since they link the real and carplicated world to the simplified interpretations of the numerical equations model. For the ammonia sorption and drying model in a fixed bed of grain, the following groups of assumptions are made for the different components of the system: Group A —- The Bin Structure 1. The grain bin is part of a perfect infinite hollow cylinder with a constant diameter. 2. 124 The bin walls are adiabatic with negligible heat capacities and have negligible effect on the airflow. Group B - The Grain 3. 4. The corn kernels can be represented by uniform spheres. The grain sample in the bin is homogeneous and isotropic. The volume shrinkage due to loss of moisture and settling are negligible. The dry matter change during the drying process is negligible. Accurate thin-layer drying-absorption equations and equilibrium moisture isotherms are available within the temperature and humidity ranges under investigation. The density, specific heat, and thermal conductivity do not vary significantly with terperature over the range of interest. Group C - Heat and Mass Transfer 9. 10. 11. 12. 13. 14. 15. Heat conduction due to particle-to-particle contact is negligible. Heat transfer in the radial direction is negligible. Heat conduction within the bulk stream is negligible, i.e. heat transfer by convection only in the bulk stream. The aT/Bt and aH/Bt terms are negligible compared to the aT/az and SH/az terms. No significant temperature and moisture gradients exist within an individual corn kernel. The surface heat and mass transfer coefficients are constant over the corn kernel surface. No surface condensation of liquid water or ammonia solution; any condensed liquid is restored as part of the average moisture content of the grain. 125 Group D - Airflow 16. 17. 18. 19. The bulk air stream behaves as an incompressible fluid. Dry air, water vapor, and amrenia are assured to be ideal gases and behave independently , in other words , the air bulk stream is an ideal mixture. Dry air is considered inert, i.e. dry air is not involved in any mass transfer processes except acting as a carrier. The generation and transport of gases other than water vapor and ammonia are negligible, e.g. CD2 generation from dry matter loss is neglected. Group E - Ammonia 20. 21. 22. 23. 24. The ammonia and water vapor are in constant thermal and concentration equilibrium with the dry air, i.e. no gradients exist in the bulk stream except in the z-direction. The heat of adsorption of ammonia is constant within the terp- erature and concentration ranges considered. The driving force of ammonia adsorption is linear and can be expressed as the concentration difference between the equili- brium value and the current value in the grain. The ammonia sorption isotherm is known. Part of the adsorbed amrenia retains in the grain as a residual during desorption and the amount of residuals is a fixed ratio (6) of the free amrenia content. The amrenia adsorption model in a fixed-bed drying system is com— posed of two major parts: the drying model and the amrenia sorption model . The two models are not independent since ammonia coexists with water vapor as a major transport species. Because of the assumptions, 126 the ammonia sorption model and the drying model can be simplified to a one-dimensional, spherical—particle packed-bed problem. The transport phenomena of water vapor and ammonia are similar in the following aspects: (1) both processes involved the same solid particles (corn), (2) both processes are diffusion-controlled, and (3) water (H20) and ammonia (NH3) molecules have the same molecular weight and approximately the same molecular size. The chemical proper— ties of 320 and NH3 are not very different either. Unlike water, ammonia reacts chemically with the corn components to form residuals. The diffusion rate of ammonia and water in corn can be very different. The affinity of amrenia for water is much higher than of water to itself. Hence, the ammonia isotherms do not relate to and cannot be derived from water moisture isotherms and thus must be developed separately as shown in Chapter 3. 3. Derivation of Model Equations The derivation of model equations generally involves the following sequence: 1) Identify the components and subsysters according to the structure and nature of the system. There is no unique way of doing this, all the unrelated inert components in the same phase can be lumped in a component, such as dry air and dry matter of corn. 2) Identify the state variables based upon the characteristics of the problem involved. Very frequently, an \mknown variable is also a state variable. Terperature and concentration are identified as state variables in the heat and mass transfer problems. The complexity of the problem, roughly speaking, depends on the lumber of independent state variables and their relationships to each 127 other and to other dependent and independent variables . 3) Identify the independent variables. Independent variables of a system refer to the variables which do not vary with the performance of the system. For most engineering problers, time and position are the most commonly identified independent variables. 4) Formulate the equations of conservation. For each independent state variable (i.e.unknown) one independent equation must be obtained . The equations describe the conservation of independent state variables for the bulk system or individual components. The conservation of heat and mass transfer is usually expressed by the differential change with respect to time and/or position. The formulation of model equations must be based on a series of assurpt ions as described in the previous section. The following components can be identified in the ammonia sorption and drying model: (1) dry air, (2) dry corn, (3) liquid water in corn, (4) water vapor in the air, (5) ammonia in corn, and (6) ammonia gas in the air. The model to be presented results from mass balances on ammonia both in the air and in the corn and on moisture in the corn and in the air. Also included are energy balances on the solid and fluid phases. The state variables in the ammonia sorption model and the drying model are: (l) armenia concentration in the air, x1, (2) absolute humidity in the air, H, (3) ammonia concentration in corn, x2, (4) average moisture content of the product, M, (5) air terperature, T, and (6) product terperature, e. The model equations can be derived fram a control-volume analysis by considering the heat and mass balances within a differential volume representative of the entire system. The control volume in the case 128 of a grain bin can be described by a differential height (dz) of the grain bin with constant cross sectional area (S) perpendicular to the direction of airflow as shown in Figure 6.1. The balance usually takes the following form: (flow in) + (generation) = (flow out) + (accumulation) In the case of depletion instead of generation, such as the comsmp- tion of free ammonia in corn as a result of the formation of residuals, the generation term becomes negative. The accumulation term expresses the internal energy change or concentration change in the solid or fluid phase. The heat and mass transfer between the solid and bulk stream is considered the accumulation change. 3-A. Mass Balance of Ammonia in Bulk Stream According to assurptions l7 and 20, ammonia in the bulk stream is perfectly mixed with dry air and water vapor. The mass balance on the control volume is made for a very shortperiod of time, dt. It should be noted that the definition of all the terms related to concentration, density, flow rate are expressed on a dry basis, i.e. the amount of dry matter of corn or the amount of dry air. The formulation of equations is thus simplified. The ammonia flow is expressed as the product of the dry air flow rate per unit area (G3), the ammonia concen- tration (x1), the cross sectional area of the bin (S), and time, dt. Thus, the amount of ammonia flowing into the control volume during time dt is: Ga x1 8 dt The amount of amrenia flowing out of the element in the same time period (it is: Figure 6.1. Air : xl'H'T'Ca'Ga' Fa The graphical presentation of the control volume in a grain drying bin in the simulation model. Related parameters are shown for the air and the product. 130 3x1 Ga(x1+-3-z—dz)Sdt since Ga is assumed to remain constant. The amount of armenia accumu- lated in the air within the control volume during time dt is: e pa-:-t—dt S dz where e is the void fraction and eSdz is the void volme of the cOntrol elerent. Besides the accumulation of ammonia in the air, accumulation of ammonia also occurs in the corn through interpha'se mass transfer. The accumulation in corn is expressed by the rate of the ammonia concentra- tion change in the corn: 3x __2. where pp is the dry matter density (solid, no void) or the corn, x2 is the free amrenia concentration in the corn, and 6x2 is the amount of residuals. The amount of residuals is assured to be a fixed ratio of the free ammonia content. The mass balance of ammonia in the bulk stream can then be expressed as: ammonia ammonia ,/ ammonia _ ammonia accumulation + accumulation ‘, outflow outflow in air in corn | Or simply, a so 3 p 8x ___= -.___a ___.x1 -_P___2. (1+5) (6.4) 32 G at G at a a 131 Equation 6.4 implies that the concentration gradient of the ammonia in the bulk stream decreases as the rates of ammonia accumulation in the air and in the corn increase. Comparing the absolute value of the two terms on the right hand side of Equation 6.4, it is obvious that the coefficient of the second term is much larger than that of the second term since the density of dry corn is much bigger than the bulk density of air in the void fraction. The influence of ammonia accumu- lation in the ammonia concentration gradient is thus more significant than that in the air. However, the values of coefficient do not illustrate the entire picture; the rate of concentration changes also plays an important role. In practical ammonia adsorption system it is predictable that the rate of concentration change in the air (axl/at) is faster than that in corn (axz/ 3t) during the early stages of adsorption since air has very limited capacity to hold ammonia and gets saturated sooner than the product . Although the rate of ammonia accumulation in corn decreases as time progresses under constant feed conditions, the higher rate of accumu- lation will maintain much longer than that in air due to the tremendously larger holding capacity for ammonia in the corn compared to that in air. In short, the ammonia concentration gradient in the bin depends mostly on the rate of ammonia accumulation in the corn most of the time. Nevertheless, the rate of ammonia accumulation in the bulk stream does play a significant role in the early stages of adsorption. Thus, the 3x1/ at term is remained in the equation. 3-B. Mass Balance of Water Vapor in Bulk Stream The behaviors of water vapor are analogous to those of ammonia in the air. Similar analyses on mass balance can be made using the following 132 analogies: H (absolute humidity)+—> x1; M (average grain moisture con- tent )H x2. The mass balance of water vapor in bulk stream is thus written as: i: -332. l- 6 pa fl (6 5) 32 Ga 3t Ga 3t ° Despite the similarities of Equation 6.5 to Equation 6.4, the grain drying process is basically a water desorption process which is corresponding to the ammonia desorption process . Even though both Equations 6.4 and 6.5 can represent adsorption as well as desorption processes, the significance of the 3H/ at term in Equation 6.5 is much less than theaxl/at term in Equation 6.4 for ammonia adsorption. Hence, 3H/8t term is ignored and Equation 6.5 becomes 3H '5“:- m:lmo a-g-M— (6.6) 3-C. Ammonia Sorption Rate in Corn The amount of ammonia accumulated in corn within the control volume during time dt is expressed as the product of the rate of ammonia concentration change (rate of sorption) in corn and the mass of dry matter of corn (pp S dz) in the differential element: 3x pp73?2'5 dz dt= N 5 dz dt (6.7) The interphase transport phenomena of ammonia from the gas phase to the solid phase or vice versa were discussed in detail in Chapters 3 and 4. The rate of ammonia accumulation (N) is a function of the specific surface area and the driving force (F) which is a function 133 of both amrenia concentrations in the air and in the corn. N = hm a F(xl, x2) (6.8) where hm is the mass transfer coefficient. The evaluation of hm will be presented in Section 6. The driving force function is often approxi— mated by the linear-driving-force function which is nondimensionalized as: *- x2 x2 2) = W (6.9) F(xl, x where xz° is the initial ammonia concentration in corn; x2*, the equil- ibrium concentration, is a function of x1 and can be calculated by the ammonia sorption isotherm equation (Equation 3. 16) . In Chapter 3, the ammonia concentration in the air was expressed as the partial pressure which is based on the total bulk air stream consisting of dry air, water vapor, and ammonia. Thus, the ammonia isotherm equation can be rewritten in terms of x1 which is the concen— tration based on dry air: * = A(T) M x1 Em db(l+H+xl) x2 (6.10) Combining Equations 6.7, 6.8, and 6.9, the rate of ammonia adsorption is obtained: * - 3x2 hm a x2 x2 _ = * - O at pp X2 X2 (6.11) Equations 6.10 and 6.11 describe the ammonia concentration changes in corn kernels, in other words, the rate of ammonia adsorption depends on both x1 and x2. 134 3-D. Changes of Mbisture Content in Corn Kernel - Thin-layer Drying Equations The moisture reroval rate in a deep-bed system is usually investi— gated using a thin layer of grain which is dried under specified air temperature and humidity for a certain period of time. The drying rate is then determined and the results are formulated as a thin-layer drying equation. Results of thin-layer experiment are applied to a deepébed system based on the assmpt ion that a deep bed is composed of layers of homogeneous grain and the airflow is a plug-type without concentra- tion, temperature, and flow rate gradients in the radial direction. Several thin-layer equations, both theoretical and.ampirical, were proposed for various grain products at different temperature ranges (BakkerbArkema et al., 1978). ‘Due to the difficulties in Obtaining the diffusion coefficients for different products ulder some conditions, empirical or semiempirical thin-layer grain drying equations are fre— quently used in modeling. Sabbah (1968) proposed an empirical thin-layer (one kernel deep) drying equation for corn at temperatures ranging from|2.2 to 21.1°C an 22 to 80 percent relative humidity: M - M* 0.664 M; M, = exp ( -k' t ) (6.12) where , 'v k=exp(-Ut) .(6.12a) U = [6.0142 + 1.453xlo’4(rh)2]°'5-(l.8 e + 32) - [3.353xlo'4 + 3.0xlo'8(rh)2]°'5 (6.12b) V'= 0.1245 - 2.197xlo'3(rh) + 2.3xlo'5(rh)(l.8 e +32) - 5.8xlo’5(l.8 e + 32) (6.12c) 135 A uniform airflow of approximately 15.24 cmm/m2 (50 cfm/ftz) was used. The diffusion coefficient k' is a complicated function of relative humidity (decimal), grain temperature (°C), and drying time. Constants are derived from experimental thin-layer drying results. The left term in Equation 6.12 is called the moisture ratio (MR); M* is the equilibrium moisture content (EMC), and Mo is the initial moisture content. In the case of average grain moisture. content (dry basis) is higher than M*, drying is taking place. Drying is essentially a moisture desorp- tion process. Reabsorption of water in grain would occur if M is less . than M*. This rewetting process was. described by a "thin-layer rewetting equation" (del Giudice, 1959) as follows: 0.466 rh 3.0 MR = exp[ - 0.625 (ps) (rh) t] (6.13) Tests were run over a dry bulb terperature range of 15.6 to 40.6°C and a relative humidity range of 60 to 100 percent. Air velocities were 3.05 m/min. The saturation vapor pressure at the dry bulb terperature of product is expressed as p S . The water sorpt ion processes possess hysteresis and reabsorption curves are similar to those shown in Figure 3.7b. 3—E. Enthalpy Balance of the Bulk Air Stream To analyze the energy balance in the bulk stream, the three components in the bulk stream, dry air, water vapor, and ammonia, are assumed to behave independently (assumption 1?) . The total amount of bulk stream flowing through the control volume is expressed as: 136 Ga(l+ H + x1) S dt and the enthalpy of the inflowing air is: Ga(ca + H cv + X1 cg) 'I‘ 5 dt where cg is the specific heat of gaseous ammonia. The enthalpy of the out flowing air is obtained by assuming the heat capacities of all carponents and their composition remain the same with respect to z in the control element and is expressed as: 3T Ga(ca + H cV + x1 cg) (T +-5;dz) S dt Thus, the enthalpy change with respect to bed depth z is: 8T Gama + H cV + x1 cg) Edz 8 dt The change in the amount of sensible heat stored in the fluid phase during time dt is: 3T pp(ca+HcV+x1cg) e-a-Eszdt and the total change in sensible heat is obtained by surming up the above two terms: 3T 3T (ca + H CV + x1 cg) (Ga'5E+ 6: pa???) S dz dt 0n the other hand, the convective heat transfer is taking place between the bulk stream and solid. The amount of energy transferred from the air to the product is: ha(T-6)Sdzdt 137 where h is the convective heat transfer coefficient which will be discussed later in Section 5. Since the sensible heat change is resulted from the convective heat transfer due to temperature difference, the last two terms must add up to zero according to the conservation of energy. The enthalpy balance equation in the air is then written as: 3T 3T __ - h a +€pa3?-ca+Hcv+xlcg G a '5-2- (T-G) (6.14) 3-F. Enthalpy Balance in Corn Corn is assumed to be composed of dry matter, water, and ammonia. Armenia is dissolved in water as a solution. The aqueous solution can be considered as ideal binary solution in the case of low ammonia con— centration which is held by the practical application of ammonia grain drying. The total mass of corn within the control voluze is expressed as the summation of individual components: pp(l+ M + x2) S dz and the change of enthalpy during time dt is: 36 C>p(cp + cwM + cl x2) S dz 8t dt where c1 is the specific heat of liquid ammonia. This change of enthalpy is resulted from two sources: moisture loss and ammonia adsorption. The amolmt of moisture evaporated during time dt is indicated by the change of absolute humidity in the air, or 3H Ga 32 Sdz dt 138 The latent heat required to evaporate this amount of water is expressed as: 8H hfg Ga 32 S dz dt The evaporation involves phase change as well as sensible heat change which is expressed as: 8H CV (T 6) Ga 32 S dz dt The total energy change (decrease) due to drying is: 3H [hfg '1' CV (T-GH Ga-a-E-S dz dt The heat of evaporation of water from the corn differs from free water evaporation and was fomd to be a function of grain temperature and moisture content (Rodriguez-Arias, 1956): hfg = (2502.10 - 2.386 9) [1.0 + 4.349 exp(-28.25 M” (6.15) for moisture content in the range of 0.0870-0.3514 d.b. and terperatures in the range of 5-40‘C; h = 2326 KJ/kg (1000 BTU/lb) otherwise. f8 The amount of ammonia adsorbed during time dt is equivalent to the. change of ammonia concentration in the air and is written as: 3 G fiSdz dt aaz Similar to the drying process, the total enthalpy change (increase) due to exothermic ammonia adsorption process is: axl [Had + cg (T-e)] Ga'EZ—S dz dt 139 where H ad is the heat of adsorption* and cg(T-6) is the sensible heat term. The heat*of adsorption differs frorm heat of vaporization for ammonia (1370 J/g at b.p. from Table 3.1 or 518.1 BTU/1b at 60°F) since ammonia is dissolved in water solution rather than condensing into liquid ammonia.Heat of solution for ammonia in water is given as 35.4 KJ/mole-NH:3 (or, 897 BTU/lb—NI-Ia) in very dilute (0.005 M+ solution (Lancaster et al. , 1975). Heat of ammonia adsorption, although it is not directly measured, is suggested to be approximated by the heat of solution of ammonia in water. The latent heat expression for water (Equation 6.15) indicates that the heat of evaporation increases with lower temperature and lower moisture content. It is also comprehensible to state that heat of ammonia adsorption increases with lower temperatures. However, experi- mental results for Had in the corn similar to Equation 6.15 are not available, it is assured that the heat of adsorption remains constant throughout the adsorption process. Since the solution of amrenia in water is a reversible physical process, ammonia desorption is expected to absorb the same amount of heat as in the adsorption process. Heat of desorption is thus of the same magnitude as Had but different sign. Besides the enthalpy changes due to mass transfer, convective heat transfer between the product and air stream is taking place at the same time. The amount of heat transfer is h a (T-e) S dz dt * Note that all the heat expressions are taken as absolute values, appropriate $1.ng are attached based on physical interpretation. t M, a commonly used concentration unit for solution, stands for moles of solute per liter of solution. 140 The enthalpy balance in corn can be sumrarized as: change in energy . ammonia internal = transferred - (evznpgratron) + adsorption product energy by convection rgy energy and is formulated as: ppp(c +cW M+c1x2)‘§%=ha(T-6)-[hfg +cV (T-6)]GGa-g% .331 + [Had + cg (T-e)1( 'Ga 32 ) (6.16) where h f g and Had are positive numbers. During drying, aH/az is positive (since aM/Bt < 0) which tends to reduce the internal product energy (Be/3t term); while Bxl/Bz is negative (since 33(2/31'. > 0) during ammonia adsorption which tends to add heat to the product . It is interesting to know that the effects of drying and adsorption on the heat content of the corn are oounteractive and tend to dampen the terperature change of grain in the bin. The sensible heat changes for water and ammonia during interphase mass transfer is very small compared to the latent heat terms if the air and product terperatures are not far apart. 3—G. The Model Equations In surmary , six independent model equations were developed in order to solve six unknown variables. Model equations are listed below in terms of their differential changes with respect to the independent variables: 3x 6: o a _P_3 ' 32 .. G 31: "EP— (1+6) (6.4) a a 0 iii: __2. __3M 32 G at (6.6) 141 -—2-= m -— (6.11) ) . (6.12) 36 _ _ _ __ 3H Dp(cp+cwM+clx2)-a-E-—ha(T6) [hfg+cv(Te)]Ga.5? 3x1 + [ Had + Cg (T-G) ]( - Ga-a-g-o (6.16) together with the following auxiliary equations: Equations 6.10, 6.12a, b, c, and 6.15 to evaluate parameters used in model equations and Equation 6.13 replaces Equation 6.12 during rewetting process. 4 . The Combined-Terperature Equation The six-equation model described in the previous section is similar to the four-equation fixed-bed drying model (Bakker-Arkera et a1. , 1974). It was mentioned in the beginning of this chapter that in a slow drying process, such as ambient air drying, air and product terperatures could be considered equal without significant error. This assurption not only simplifies the computation but also helps to stabilize the nurer- ical solution. The following method is expected to impose an insignifi- cant error while combining the two terperature equations. 142 Equations 6.14 and 6.16 can be written as: an Ga cpa (-§-§-+E—i 3%) =-ha (CH) (6.17) a . 36 pp opp fi= h a (T-e) - Ga L (6.18) with, Cpa = c a + Hov + x1eg = apparent specific heat of bulk air stream opp = CD + Mow + x2621 = apparent specific heat for ammonia-treated corn ax - 135 1 L - hfg 32 + Had Tz— by assuming negligible sensible heat changes compared to the latent heat terms, i.e. cV(T-6) = cg(T-9) = 0. On the other hand, the error due to neglecting the convective heat transfer term, h a(T-6), could be signi- ficant. However, Equations 6.17 and 6.18 are combined in such a way to cancel the convective heat transfer term. The combined equation is thus obtained by further assuming 36/3t = HT/Bt: ED 0 C 11'. __a. _E 22.—___}— az+( G +Gc > t- c (6.19) a a pa Equation 6.19 is therefore replacing Equations 6.14 and 6.16 and consti- tuting the five-equation model for ammonia adsorption and drying in a fixed-bed of grain. 5. Heat and Mass Transfer Coefficient for Forced Convection through Packed Beds ' 5-A. Forced Convection for a Single Sphere The forced-convection heat transfer for a single sphere in an infinite fluid was described by Ranz and Marshall (1952) by plotting 143 1/2 1/3. the average Nusselt number versus Re - Pr The relation was then determined as: 1/2 1/3 Nu = 2.0 + 0.60 Re Pr (6.20) where the Nusselt number (Nu), Reynolds number (Re), and Prandtl number (Pr) are dimensionless numbers frequently used in heat transfer problers, they are defined as follows: Nu = h D / k (6.21) Re = D Ga / u (6.22) Pr=cpu/k (6.23) where h represents the average surface heat transfer coefficient over the entire spherical surface and Nu signifies a dimensionless terperature gradient in the fluid adjacent to the wall; D is the diameter or charac— teristic length; p, u, k, CD are fluid properties and stand for density, viscosity, thermal conductivity, and specific heat respectively. Strictly speaking, all the thermal properties of fluid should be eval- uated at the film terperature: T1. = %— (Tw + Tea), which is the average value of the solid surface and bulk stream terperatures. In a slow heat transfer process (i.e. small terperature difference) T1, may be estimated by Too. Equation 6.20 indicates a minimum value for convective heat transfer coefficient when Nu = 2.0 for motionless fluid due to heat conduction. This limiting value of Nu for heat transfer from spheres at low Re or Gr (Grashof number, used in free convection) can be theoretically derived by solving the following spherical steady state heat conduction equat ion : 144 _l_.Ji_ 12$: with the boundary conditions that T = Tw on the spherical boundary r = R and T = Too as r approaches infinity. The terperature of fluid is then solved as inversely proportionally to r when r _>_ R: T-T Q T-T= woo The surface heat transfer rate is equal to the heat conducted from .3. r inside the sphere to the surface, or 8T h(Tw-Too)=-k(8r ) r=R Solving for h and substituting in Equation 6.21 gives Nu = 2. The minimum value of convective heat transfer coefficient for a single corn kernel in air is thus estimated as 1.82 x 10-4 J/(hr-m2-°C) or 0.89 mwmfi%rmmn=0nw8mmnm2fi)mrmmkmmlmdk=wn J/(hrom-°C) or 0.0143 BTU/hrft°F for air at 10°C. 5—B. Evaluation of Heat Transfer Coefficients in Packed Beds Unfortunately, heat transfer relationships for flow through a packed bed cannot be derived by extending those for a single sphere since the bed porosity is more important than the individual particle diameter in a packed-bed system. The following equation was derived experimentally by Yoshida et al. (1962), Hougen et al. (1960), and Bird et a1. (1960) for a packed bed of spherical particles: . 0.91 me'0'51 , when Re<50 J = H { 0.41 (6:24) 0.61 Re“ , when Re>50 145 where the Colburn factor, jH’ is defined by: (St) (Pr) ”3 , and (6.25) Nu h St = Stanton number = = (6.26) (Re) (Pr) c Ca In the case of a packed-bed system, the characteristic length is represented by the inverse of specific surface area (a) and Re becomes: a 11 (6.27) Solving for the convective heat transfer coefficient by substituting Equations 6.25, 6.26, and 6.27 into Equation 6.24 yields: 0.51 0.49 3 (aua) Ga Pr'z/ , when Re<50 0.91 h = 693 (6.28) -2/3 0.41 0.59 - 0.61 cpa Pr ( a ma) Ga , when Re>50 By substituting the properties of air at 10°C: cpa = 1005 J /kg°C, ka = 0.0248 W/m K, pa = 0.06354 kg/m hr, Pr = 0.7167 and a = 784 mz/m3 (239 ftz/fts) the above expressions are simplified to: 049 {2.3278 Ga ° , when Ga< 2400 kg/hr/m2 h .— (6.29) 059 . 1.0556 ca ' , when 63 2400 kg/hr/mz where h has the units of W/m‘?‘ K and air properties are assured to remain constant and are approximated by the values at 10°C. Equation 6.29 indicates that the convective heat transfer coefficient is only a function of the flow rate. 146 5—C. Estimation of Mass-Transfer Coefficients at Low Mass-Transfer Rates 5—C.l. Ratio of Mass and Heat Transfer Coefficients The determination of mass transfer or diffusion coefficients are usually more difficult than the heat transfer coefficients since solute concentrations are not like terperatures which can be easily recorded continuously for a number of measuring points at any specified scanning periods. Nevertheless , the solutions of many mass-transfer problers at low mass-transfer rates may be obtained by analogy with corresponding problers in heat transfer. The sorption processes in most biological products are relatively slow compared to other chemical-based sorption processes. A process with slower rate of mass transfer is expected to be solved with less error using this analogy. (he of the Chilton-Colburn (1934) analogies states that Colburn factors for heat and mass transfer are identical and are a function of Reynolds number, or 3D = 3H = a flmction of Rs (5.30) and the Colburn factor for mass transfer, jD, is expressed by, iD =(StD) (Sc) 2/ 3 (6.31) where Sh St.D = W (6.32) _ - = M Sc — Schmidt number (3 OLD (6.33) D Sh = Sherwood number = OLD (6.34) (ID = mass diffusivity (mz/hr) hD = mass transfer coefficient (mu/hr) 147 and Sc corresponds to Pr in heat transfer; Sh is in analogy with Nu and represents a dimensionless concentration gradient at the wall. Equation 6.30 implies that, Sh Nu (6. 35) (Re) (Sc) ”3 (Re) (Pr) V37 or by substituting the original form for dimensionless numbers, Equation 6 . 35 becomes: hD 0‘13 a 1/3 “D 1 OD 2/3 _h = reg-or") “-3” The mass transfer coefficient, hm’ used in the model equation (Equation 6.11) is related to hD by: hm = o hD (6.38) where hm has the units of kg/(hr-mz) while it is m/hr for hD. Replacing hD by If" in Equation 6.37 gives: h 712:};— (—:2.)2/3 (6.39) P 5-C.2. Estimation of the Mass Diffusivity The only unknown parameter in computing the mass transfer coefficent from Equation 6.39 for ammonia diffusion problem is the mass diffusivity, “D The experimental values of “D for ammonia sorption processes in corn or other grain products have not been acknowledged in the litera- ture. Unlike the determination of moisture diffusion constants in which moisture contents are easily determined by standard air oven method 148 (103 °C for 72 hours), chemical procedures in quantitizing the ammonia concentrations in the corn (as described in Chapter 5) are rather. complicated and time consuming. Chu and Hustrulid (1968) developed the following empirical equation for moisture diffusion coefficient of the corn kernel: 4 2513. 00 Oh = 1.513 x 10 eXp[(0.045 6 + 6.806) M - 6 + 273. 1 (6.40) where 6 is the grain temperature in °C and M is the average corn moisture content in decimal d.b. The above equation was derived from the experi- mental results of the following test conditions -- grain temperature 49—71°C, relative humidity 12-70 percent, and corn moisture content 25-35 percent d.b. Equation 6.37 indicates that the values of water diffusion coeffi- cient range from 5.89 x 10.7 (for 25 percent m.c. at 49°C) to 3.38 x 10'6 mz/hr (for 35 percent m.c. at 71°C) under the above-specified conditions. A product with higher moisture contents at higher tempera- tures has a higher diffusion coefficient. On the other hand, the experimental value of self-diffusion coefficient for water is 8.78 x 10’6 [112/hr at 25°C (Robinson and Stokes, 1959). (he would assume moisture in the corn as "free water", OLD should be equal to the self—diffusion coefficient of water. Comparing these two quantities, it is fomid that they are in the same order of magnitude and the self-diffusion coefficient is higher. This is an indication of extra diffusion resistances exerted by the solid portion of the product which hinders water from diffusing freely. The experimental value of diffusion coefficient for ammonia gas at very low concentrations in liquid water was measured as 5.90 x 10-6 149 mz/hr at 12°C (Sherwood, 1975). This value may be considered as the maximum possible diffusion coefficient for ammonia diffusion in the corn at 12°C since in most cases the moisture content of shelled corn seldom exceeds 35 percent w.b. The value of OLD also varies with grain temperature besides the moisture content. As a matter of convenience, the diffusion coefficient (mass diffusivity) of ammonia in corn is assumed to be a constant of 5.90 x 10'6 mZ/hr throughout this investi- gation. The ratio of mass transfer to heat transfer coefficient can thus be approximated by substituting a = k/pcp = 0.0473 mZ/hr and cp = 2161 J/kg°C at 10°C (Table 3.1) in Equation 6.39 as follows: h l(_o_‘g_)2/3 6 0. Kb 959—;- C = 1.156 x 10' kg °C/J (6.41) 9 Thus, the mass transfer coefficient for ammonia is estimated as: 6 hm = Kh h = 1.156 x 10 h (6.42) 6. Dry Matter Decomposition The deterioration of grain was indexed by the measurelent of carbon dioxide production generated from both grain respiration and flmgi growth (Steele and Saul, 1962). The measured amount of carbon dioxide is then translated into dry matter loss of grain. Saul and Steele (1966) suggested an allowable dry matter losslof 0.5 percent for field-shelled corn without a reduction in grade. The percent dry matter decomposition was given by Thompson (1972) as follows: [M = 0.0884 I exp(0.006 te) - 1 ] + 0.00102 te (6.43) 150 where t e = equivalent storage time at the reference conditions of 15.6°C, 25 percent m.c. (w.b. ), and 30 percent mechanical damage. The relationship between the equivalent storage time and the actual storage time lmder specified conditions was described by Steele et a1. (1969) as: t =___1=__ e “MMTMD where MM' MT" and MD stand for moisture, temperature, and mechanical (6.44) damage multipliers respectively. These multipliers were determined experimentally by Steele (1967) and Saul (1970) as follows: (a) Moisture Multiplier (MM) _ 1.53 _ MM - 0.103 [ exp( 455mdb ) 0.00845 Mdb + 1.558 ] for 15_<_Mdb<54 (6.45) (b) Temperature Multiplier (MT) MI. = 128.76 exp( -2.592 - 0.146 T ) , T:l$.6°C (6.46a) MT = 32.3 exp( -l.856 - 0.1044 T ), T>15.6°C and Mwb<_19 (6.461)) MT = 32.3 exp( -1.856 - 0.1044 T) + [( Mwb - 19 )/100] exp( -0.2847 + 0.0183 T ) , T>15.6°C and 1915.6 and Mwb>28 . (6.46d) 151 (c) Mechanical Damage Multiplier (MD) Values for MD at various levels of mechanical damage were given by Steele et al. (1969) as shown in Table 6.1. Ammonia has been known as a very effective mold and bacteria inhibi- tor for stored grain. The dry matter loss is partly due to the micro- organism growth which might overwhelm the effects of grain respiration under storage conditions favorable for molds and bacteria growth. The primary function of ammonia in ammonia-treated corn is to wipe out the initial or field contamination of microorganisms and further slow down or inhibit the growth of storage fungi and other microorganism; “the dry matter loss is thus reduced. The evidence of this ”killing effect" on storage microorganism (Bothast et al. , 1973) due to proper ammonia treatment leads to the definition of an Ammonia Multiplier, or M . A Equation 6.44 can therefore be expanded to: t t = e “n MT “n “n The values for M A is not presently available and it is only known that MA > 1 with ammonia treatment and equals unity when no ammonia is (6. 47) applied. Experimental correlations must be established in order to determine the values of M A in Equation 6. 47. The experimental determina- tion of MA will not be performed in this investigation, however, it is suggested to take the following precautions: l) The primary inhibitory function is due to free ammonia concentra- tion both surrounding corn kernels and dissolved in corn' (Lancaster et al., 1975). 2) The value of MA is a function of the total amount of armronia applied as well as the rate of ammonia application. The diffusion of ammonia 152 TableGSLI. The mechanical damage multipliers. Percent Mechanical Damage MD 2 3.22 S 2.90 10 2.41 15 1.98 20 1.60 25 1.29 30 1.00 35 0.76 40 0.55 Fromm Steele et al. (1969). 153 into the pores and dissolved in corn is a slow process. The faster the application rate is, the more ammonia is lost in the air. 7. Equilibrium Moisture Content The experimental values and model equations of equilibrium moisture content (EMC) for various cereal grains were reviewed and summarized by a umber of researchers: Brooker et al. (1974), Pfost et a1. (1976), and Bakker-Arkema et al. (1978). Bakker-Arkema et a1. (1974) analyzed the EMC data of Rodriguez-Arias (1956) for shelled corn and developed the following empirical equations: S rh3 F M = 1 + —1—-— 0.028333 51) rh , 0.0 S rh< 0.17 (6.48a) e 1.02 0.17 s (0.34-rh)3 82(rh - 0.17)3 ( 52 + M = _L + - 0.028333 52) (rh-0.l7)+ e 1.02 1.02 0.17 J"‘1 (——-- 0.028333 51) (0.34-tn) . 0.17 s rh < 0.34 (6,4313) 0.17 s (0.51-rh)3 E3 52 M = —r + (rh-0.34)+( - 0.028333 52) (0.51-tn) , e 1.02 0.17 0.17 0.34 5, rh S 0.50 (6.48a) 53(rh-0.49)3 F5 F4 M =- -— + - 0.028333 53) (rh-0.49)+ (0.66-rh) , e 1.02 0.17 0.17 0.50 < rh < 0.66 (5.43.3) $3 (0.83-r11) 3 S‘1(l:'h-O.66)3 F M = + + —L- 0.028333 84) (rh-O.66)+ e 1.02 1.02 0.17 F5 ( - 0.028333 5 ) . 0.66_<_ rh< 0.83 - (6,439,) 0.17 3 84(1.oo-rh)3 F7 56 M = + (rh-0.83) +( e 1.02 0.17 0.17 0.835 rhé 1.00 (6.48f) - 0.028333 54) (1.00-rh) r 154 where: F1: -0.0007060 T + 0.0874 Sf 13.83(-9F1+6F2-F3) 52: -0.0007835 T + 0.1189 52: l3.83(4F3-9F2+6F1) F3= -0.0009646 T + 0.1475 $3= 13.83(4F4-9F5+6F6-F7) F4= -0.0009675 T + 0.1452 s4= 13.83(4F7-9F6+6F5-F4) r = —0.0012735 T + 0.1849 5 T = °c F6= -0.0013408 T + 0.2294 M deem d b r7= -o.0019278 T + 0.3588 e’ ' ' ° Pfost et a1. (1976) gathered EMC data for various grain from several sources and developed the well-known equilibrium relative humidity- moisture content equations . They concluded that Henderson-Thompson and Chung-Pfost equations give results nearly as good as the other more complex equations . The Henderson-Thompson equation of desorption me for yellow dent corn is: 1n(1-rh) l M = [ _ ]— -8.6541 x 10 3(T + 49.81) “3634 e (6.49) and the corresponding Chung-Pfost equation is: Me = 0.379212 - 0.058970 1n[ -R'(T + 30.205) 1n(rh) 1 (6.50) where R'is the gas constant, T is in °C, and Me in dry basis decimal. As it was indicated by all the EMC equations that EMC depends only on temperature and humidity for a particular kind of grain. The air humidity in equilibrium with a certain moisture content of grain is referred to as equilibrium relative humidity (ERR) which can be solved explicitly only by simplified EMC equations such as Equations 6.49 and 6.50. 155 8. The Finite Difference method 8—A. Finite Difference Formulation The finite difference approximation method is a.powerfu1 technique in solving partial differential equations. The formulation of finite difference equations is based on the Taylor's Series expansion which approximates the value of a function by its neighboring point and an infinite series of derivatives of such function. Basically, there are three ways of expressing the derivative functions: fOrward difference (implicit), bankward difference (explicit), and.central difference. For example, the first derivative of T(z,t) can be expressed by: '+l,j - '.j T2 = Tl Az Tl (implicit) (6.51a) . [j - 1.1,]. T2 = T1 A: (explicit) (6.51b) Ti+1,j _ Ti-1,j . T2 = 2Az (central difference) (6.51c) where superscripts i and j stand.for position and.time indices respectively. The indices (irl), (jtl) represent forward and badkward positions and times relative to (i,j). Similar equations for 8T/3t can be formulated. The finite difference equations for state variables x1(z,t), H(z,t), x2(z,t), M(z,t), and.T(z,t) (Equations 6.4, 6.6, 6.11, 6.12, and 6.19) are formulated by explicit differences as follows by assuming xé°= 0 in Equation 6.11: 1 1 i,j-l _ '- .j i 3'- i 3-1 i,j = Ga At x1 + 6 pa Az x1 (x2 ' x2 ' )ppAz(l+6) X1 Ga At + 8 pa A2 (6.52) 156 . . ._ . 0 AZ - - - -_ HI") = H1 1;] __éLA?( M11] _ Ml'j 1) (6.53) a . . h a At lelj-l +£1.7— i . X2 I] = 1‘“ a Atp (6.54) 1 + i’j“ * I Dp'xz M1,] = f»)! ( T1,]: Hl'JI Mel-,3! t) (6°55) i,j _ 1 1-1 3 1.3-1 _ T "At+PAz [AtT ’ +PAZT hf At ( Hilj _ Hi-l'j ) - Had At ( i,j - i-l'j )] (6.56) -—Jl-- x1 X1 C C pa Pa € 9 D C where P = G a.+7§i7§31. a a Pa 8-B. Solution of Finite Difference Equations The solution of the above finite difference equations are not straightforward since the values of these state variables at (i,j) are interrelated. Their relationships are shown as follows in functions which contains state variables only: i,j .. i,j x1 .. f1(x2 ) (6.57) 111'] = f2( 141'] ) (6.58) i,j _ i,j . x2 _ f3( x2* ) . (6.59) i,j ._ i,j x2* - f4( x1 ) (6.60) Ml'J = f5( Hl'J, Tl’j ) (6.61) 157 Ti'j = f6( Hi'j, xli'j ) (6.62) where the specific heats of both bulk stream (cpa) and product (opp) are assumed to be constant in Equation 6.61 within the finite difference grid. It is apparent that x1 and x2 must be solved simultaneously. Equations 6.57, 6.59, and 6.60 may be written as follows in accordance with Equations 6.52, 6.54, and 6.10: i,j _ id x1 - Al + A2 x2 (6.63) . . A x *i’j X2113 = 4 2 i’j (6.64) A3 + "2* . . xli'j x2*l’3 =A5 ( - . )A7 (6.65) A6 + x11":l where, assuming I-Ii’j can be approximated by 111-1"], the absolute humidity at the previous node, and the following coefficients are known: i'lrj i,j-l i,j-1 A1 2 Ga At x1 + spa Az x1 + pp Az x2 (1+0) Ga At + spa A2 0 A2 (1+0) A2 = - P G At + 6.0 A2 a a A3==100hmaAt/0p = i,j-l = i-l,j A4 A3 + x2 , A6 1+ H A5 = A(T) Mdb , A7 = B(T) 158 Equations 6.63, 6.64, and 6.65 can be solved by iteration or they may be combined into a single equation with one unknown variable; e.g. substituting Equation 6.65 in Equation 6.64 which is then substituted in Equation 6.63 yields: i,j A2 A4 A5 ("in )A7 i,j A3 + A5 “‘15 )A7 where lel’J=xli’J/(A6 + xll’J). The value of x11“) is solved by trial and error using the subroutine ZEROJN. Qice x11"j is known, )C2*i"j can = 0 (6.66) g(xli’j ) =A1+ be calculated from Equation 6.65 and x21"j from Equation 6.64. The model equations are left with three equations, 6.58, 6.61, 6.62, for H, M, and T which must be solved by a search algorithm. The general order of the computation in the search algorithm is outlined below: 1) Guess mi;j = 111'”, i.e. take the humidity of the previous position as a guess. 2) Calculate ’I‘i’j from Equation 6.56. 3) If RI-Ii"j < 100 percent, go to 5 . 4) Simulate condensation and find Ti’j, mi’j. Set flag. 5) Calculate Mi’j from Equation 6.55. 6) If condensation flag is set, exit. 7) Calculate (EMC)i"j from Equation 6.48. 8) Estimate new Hi’j from (£14ch - 111:3. 9) If AH < 0.0001, go to the next node. Subroutine ZEROIN is utilized to perform the search for an acceptable H value. Condensation occurs whenever the temperature is below the dew point. The condensed water is assumed to be added to the grain moisture according to Equation 6.6 and the estimation of EMC is skipped 159 because it is not applicable. 8-C. Temperature condensation Simulator During the solution of temperature and absolute humidity, the corresponding relative humidity is also Checked for condensation. If the relative humidity exceeds 100 percent in which the psychrcmetric chart does not apply, H must be readjusted. The new value for H is computed along the constant enthalpy line. The intersection point with the saturation line gives the desired temperature and absolute humidity. The grain moisture content must be reevaluated because of the assumption that any condensed.water be restored.to the grain moisture content. The search for a new T‘and H in this temperature condensation simulator is illustrated in Figure 6.2. The temperature at the inter- section can be solved from the simultaneous equations for the constant enthalpy line and for the saturation curve which is approximated by a straight line. The temperature equation (Equation 6.56) can be written as follows: i,j _ l '-l,j i,j-l _ _ T -————At+PAZ(AtT1 +PAzT 01 02) (6.67) where h At . . ._ . 01: f9 (H‘l’J-Hl 1’3) c Pa H At . . . . _ ad 1,] _ 1-1,j 02‘ c ”‘1 x1 ) Pa A saturation point (TS, HS) can be located by assuming Hl’J = Hl‘"1'J and TS lies between T and T': T = ___];_( At 'I'J'ml’:J + P Az Tl'j-l - Q s At + P Az (6'68) 2) 160 Temperature Figure 6 . 2 . Graphical presentation of the condensation simulator . 161 and HS is known once Ts is solved on saturation line. The slope of the line which approximates the local saturation line is then, %= HS - HS' (6.69) where Hs' is evaluated at terperature (TS-1) on saturation line. The equation for the constant enthalpy line with respect to (T,H) and (T',H') is: (At+PAz)(T'-T)+QB(H'-H)=0 (6.70) where 03 = hngt/cpa and the superscripts on T and H are emitted. The equation for the saturation line with respect to (TS, HS) and (T', H') canbe expressed as: AH ' ' - ' - — Substituting Equation 6.71 into Equation 6.70, the readjusted temperature T and humidity H are: (At+PAz)T+Q3(AH/AT)T -Q3(HS-H) T'= S (6.72)‘ At+PAz+Q3 (AH/AT) AH HS+(T-TS) — (6.73) I H AT 9. The Flowcharts The flowchart analysis is a very important and powerful step in solving the complicated differential equations and numerical logic by digital computers. The most important matter in using a digital computer to solve nurerical problem is the solution sequence which can be easily followed by a flowchart. The flowchart is essentially a detailed algorithm. 162 The computer program for ammonia adsorption and drying in a fixed- bed system consists of a main program (ADSFIX) which serves as a skeleton for the entire computation (Figure 6.3). The main program is composed of a depth loop enclosed by a time loop. The depth loop computes the results of state variables in each node at the same instant while the time loop performs the calculations in the depth loop at each specified carputing time increment throughout the entire process . The computing and print time scheme in the main program is flexible and has the capability of handling irregular input and output time intervals (Figure 6.4). The input conditions can be from any irregular time interval but computes at a constant and specified computing time interval. The printed outputs are controlled by either a fixed value of print time increment (TBTPR) or at each time the new input conditions are read. The major subroutines used in the main program are as follows: 1) mm: finds the grain moisture content by a thin-layer drying equation (6.12) and a rewetting equation (6.13). 2) SOLVE: computes terperature, moisture content, and relative humidity by given absolute humidity, also perform the condensation simulation described in Section 8—C. 3) ZERDIN: solves an equation by guessing a high and low limit for the unknown variable and narrowing the range for the solution within an acceptable error. 4) ADSPT (Figure 6.5): computes the ammonia concentrations in the air as well as in the corn by solving Equation 6.66 using ZEROIN. The solution for x 1’3 must lie between the guessed high and low values. 1 If multiple solutions occur, the solution range should be narrowed. 163 I READ input informationsand initial conditions I 1, [Compute initial equilibrium rel. humidity I J, I Initialize arraysgI 1 Compute heat and mass transfer coeff. I l IPRINT input conditions and properties I l (2) >1 t=t+A t l < Time to read Inew data ?W/1 Y READ inlet air temp., wet bulb temp.. and ammonia concentrations in the air 1 (End of file 7 N A (:Time to enter new data ?_)> )1, [Enter new input data I jfip Compute outputs for the inlet node I i‘ z=z+Az 1 I A y (C End of depth loop ? J>T J," I Guess Hi’JsHI'I’j I (56> Figure 6.3 . The flowchart presentation of the main program ADSFIX. 164 (B TL Set condensation flag ICON=l , :fiL IChlculate M1'5(Eq.6-55)I ' L (ICOndensation simultor: Find T ’ , Calculate temp.(Eq.6-56) I v /---41----> \ RH1'5<1.0 7 iJHi..i I - Calculate ammonia concentrations I _ (CALL ADSPTl and ADSPT) J) (icons )1 0 Moisture change with time ? ("1 ’J’Mi .J Y 1'} -1) l I J Drjing I [Guess low and high Hi’3 I Absorption I _§E_ J) ICuess low and high Hi’j I l _ C2) Solve H1’3.T1’J,M1’J simultaneously I (CALL ZEROI [ N) I Shift time arrays I l IEstimate dry matter loss I J 6 Figure 6.3 . continued Calculate average moisture contents, DM loss. ammonia concentrations and temperatures 165 99 Y (:15 total drying time over ? )>"""'“ LN N {Time to print 1:) Yr Calculate additional outputs I I PRINToutputsJ ® Figure 6.3. continued 166 LTDTE=I~IR=HR1=TBTPR=PRI=O . j J [READ TT,DE:LT,IPR] Irml 2 N or READ TBTPR] y _ T [Timur-DEE (TIME>HR 2 J. v EREAD Inlet Conditions] I) [HRl=HRl+DELT W N i I-IRN>I-IR1 ?fY L—‘LEnterN Inlet ConditionsJ NT=1 [_Depth 1009ij ( TummETIPTT? < IPR=1N? }— mam ‘ TIME’PRT? > [Make Final Calculations I I TD’IE=TIME+DELT I I PRT=PM‘+‘I‘BTPR I Lmuwmr ] L Figure 6.4. The computing and print time scheie in the main program ADSFIX. 167 [—Guess x1(1ow)=0.| [Guess x](high)=max.(xlinlet.x11'1’J,x1i’j-])I I! 1 . [~CALL ZEROIN to solve x1 ’j(Eq.6-66)l ’5: Eq.(6-65)[ 1.3 *2 = Eq.(6-64) I x2*1"j< .1540 f in ~ I Sorption rate: R= Eq.(6-8)I m _ F ‘ <:x11’3=0. ?:>’N , 'W Y A lwx 1’j=m‘in.(xzi"j.].xzi’j)J 2 Figure 6.5. The flowchart presentation of the subroutine ADSPT. 5) 6) 7) 168 The maximum possible value for xli'j will not exceed the whim values of: (1) inlet x1 concentration, (2) x1 concentration of the previous position at the same time, and (3) x1 concentration in the previous computing time interval. During the amnonia desorption processes, the total anmonia concentration in the corn, xzi’j , cannot be lower than the residual concentration (XZR) which is not removable by aeration. Another precaution in subroutine ADSPT is trade that the anmonia concentra- tion in corn does not increase with time if there is no ammonia in the air. ADSPTl (Figure 6.6): calculates the amnonia concentrations for the first inner node by further dividing the conputing time interval into NI‘ intervals. By doing so, the errors due to discontinuities of boundary and initial conditions are reduced and the numerical solution stability is improved. In other words, the effects of a step input on the evaluation of x at the first time step will be 1 damped. The rate of sorption is taken at the first subdivided computing time increment (At/NT) for the first computing time interval. EMC: calculates the equilibriun misture content of the corn according to Equation 6.48. Psychrometric package: includes several subroutines to estinate all of the air properties related to psychrometrics. It serves as a psychrometric chart for digital canputers. 169 At ' = At/NT P , Redefine time-related coefficients I ‘r, _ o mo copuoswumm on» :o ~o>cmaca xenon m:_u=asoo new no cocoapech .m.> mczmpa so me.omx ._o>couce space messaaeoo m N _ o q - °o° V A a . w. a. o new a? g o‘l‘ll-IIIIIO W v w. D.- l.. 00H 3 0 u m on; u x n 0. D5 \ l ”on" Al.- O I... \ 0 o u \ WM. 0 “message . 1 its sv “E G p w. 177 .choo >5th \mmzuox euooaeoumx an den EFQQmmzuox euoosméuex one Beams monument men use .. HKm p.mm a.m.” o.o a.mn m.cT a.mo mée v.5 vé a.m.. m4... oo.m .75 56m méa o.~ .77. ~.ml «.3 5.3 m.cm m6 m4: m6: oozm a.m.m a.mm mod Hé HA7. «6| m.cn p.mv exam mg. Won m.ml om.o ~.mm m.cw «.3 H4 a.m.. a.mu m.-. Nuom m.mm o.m o.o m.m| mmé m.mm 93 m.- H6 o.m a.m.. mos. a.mm mfim mt: m.~ m.~| 31o «Sm oém p.mm m.ca H6 o.~ p.mm v.3 mém a.ma «Hm m4 no.0 v.3 0.3 m.cm a.ma p.m. Tm v.3 otS m.cm N.ma mg. 41m so. ufl 32 nd 35 Amos 500 5 ucmucoo mg 00st are HE may 5 .980 mg uq manage mo muouum 09350qu 93ng mmnum>m mo muouum mmflcwoudm 9393mm .ouuQ can onud up Egan? «magnum H8395: 9.5 05 on 3% mace/House flame En 9:3 599.8 amount, 05m: going no mason m Hound 500 5 QB new on» 5 mcowunuusmoso mg monomers mo c0333 m5. 5 muouum mmcueaoumm puffing . a . s canoe 178 predict the average ammonia concentrations in a long-time simulation, a larger computing time increment has to be accompanied by a larger computing depth interval. Nevertheless, the average ammonia concentra- tions alone do not describe the entire nature of the accuracy in the prediction model. Other important results which may indicate the accu- racy of the prediction model are the ammonia concentration profiles in the grain bed which will be discussed in the next section. In general, larger Az and smaller At values create larger predic- tion errors on the right—hand side of the zero-error line in Table 7.1. However, on the left-hand side of the zero-error line, a smaller Az and a larger At impose a larger error. This phenomenon is evidenced by the sharp upward turn of the curves close to the zero computing time incre- ment in Figures 7.1 and 7.2. The error in the prediction of the ammonia concentrations in the air and in the corn decreases as the ammoniation time increases up to about five hours as shown in Figures 7.4 and 7.5. After five hours of ammoni- ation, the variation in errors are not as significant as at smaller times. The reason for choosing two hours of ammoniation time in the analysis of relative errors in Table 7.1 is that the error at two hours ammoniation seems to be at the position which imposes an average amount of error. The rate of ammonia adsorption also afffects the prediction error. Generally speaking, as the adsorption rate decreases a smaller error is found in the computation of the average ammonia concentrations. 179 .Hflm may 5 :oflnflfiocoo mace—en mango on» so Hoe/H35 58o 02.39960 «0 cosmosflcH .vK panama so mean" Jeanna fiche 93:98 m N e o _ _ o o Hoénfigfimfiugg O H C o m 2 [s l o C V. N -q-p Ibo-[x 'xte aqq, u‘; nopenteouoo etucmme efimnv Hnoa o.m 180 .500 5 uGREoo mango 00H“ mango 0c» :0 H§5 53o 5.550 no game—H .mK. mafia .5 means 4535 find 33298 m m H o . . O O Eoéuusgmnofinfisfiasoo I V. a a m a: m. s... W a: no; a 5 m. u a: m. 4 m; m .33 w. mu“? n. O N 181 3-B. The Ammonia Concentration Profiles in the Air as Influenced by the Computing Scheme The ammonia concentration profiles can give another view of the accuracy of the prediction model as influenced by a different computing time and/or depth interval. Comparison of the concentration profiles using various computing schemes are shown in Figures 7.6, 7.7, and 7.8 for A2 = 3.05, 15.24, and 30.48 cm, respectively. In general , the model predicts higher concentrations at the lower portion of the bed as a smaller Az is used. The variations in the prediction of the concentration profiles are less significant for the results using different computing time increments when a larger com— put ing depth interval is employed. Thus , even though a larger At accompanied by a larger Az predicts the average concentrations more accurately than using a smaller Az, the deviations in the prediction of ammonia concentration profiles are larger. 3—C. The Selection of a Proper Combination of the Querating Time and Depth Increments The influence of computing intervals on the estimation of the predict ion error is mixed according to the above discussion. In summary, a smaller Az predicts more accurately both the average concen- trations and the concentration profiles when the computing time increment is fixed. On the other hand, a smaller At predicts the concentration profiles more accurately but the results of the average ammonia concen- trations deviate further from the true numerical results for the same computing depth interval . The selection of the proper computing time and depth interval depends on the following factors: (1) the time of armoniation, (2) the 182 6 5 _ C l l. o t l. 0.05 hr 2. 0.10 hr 3. 0.50 hr 4. 1.00 m 5. 2.00 hr 4 3 \ 5 a) "i C m x s s 2 I. E \ «8 '5 0 1 I- 0 l l l I 0.0 1.0 2.0 ‘ 3.0 4 4.0 5.0 Ammonia concentration in the air,x10 ,d.b. Figure 7.6. Influence of computing time increment on the ammonia concentration profile in the air using Az=3.05 an after 2 hours of ammoniation. Grain depth, x30.48 cm Figure 7.7. 183 5 , 5 Outputing time increment l. 0.05 hr 2. 0.10 hr 3. 0.50 hr 4. 1.00 hr 4 _ 5. 2.00 hr 3 .. 2 .- 1 l— 0 l L l l 0.0 1.0 2.0 3.0 4 4.0 5.0 Mmmia concentration in the air,x10 ,d.b. Influenceofcarputingtimeincretentontheammonia coroentration profile in the air using Az=15.24 cm after 2 hours of ammoniation. 184 6 5 P Ccmputing time increment l. 0.05 hr 2. 0.10 hr 3. 0.50 hr 4. 1.00 hr 4 '- 5. 2.00 hr 5 CO “3 o m x i; 3 I '8 '5 33 2 n— 3 1 "" .4 5 1 O 1 l I 0.0 1.0 2.0 3.0 4 4.0 5.0 Ammonia concentration in the air,xlO ,d.b. Figure 7.8. Influence of computing time increment on the amrenia concentration profile in the air using Az=30.48 an after 2 hours of ammoniation. 185 total simulation time, (3) if accuracy is required in either the concentration profile or the average concentration, (4) the tolerable errors, (5) the inlet ammonia concentration, and (6) the initial grain moisture content. As the time of ammoniation increases, the error in the concentra- tion profile and the average concentration decreases. In the practical application of the amnonia sorption and drying model, a long-time simu- lation process is usually encountered. Tb avoid using an excess amount of canputing time it is advisable to employ the largest At and Az values possible within the acceptable range of error. A larger computing time increment can be applied if the average amnonia concentrations are .more important than the concentration profiles in the analysis. The computing depth interval is chosen depending on the physical dimensions of the system and the sensitivity of grain depth interval in the analysis. Values of A2 = 15.24 cm and At = 1 hr were chosen in the simulation of the ammonia grain drying process. The ammonia-supplemented.ambient air drying requires several months before the grain has dried to 15.5 percent w.b. The total time to be simulated for the experimental test was 1744 hours. With such a long-time test and the main concern being the average amnenia.ooncentrations, it seems appropriate to use one hour as the computing time increment. The total grain depth of the experimental test bin was 183 an (6 ft). Thirteen nodes within the grain bed were assigned by using 15.24 cm (0.5 ft) as the computing depth interval. The prediction errors due to using At=1 hr and A2 = 15.24 cm are 6.5 percent in the average amnonia concentration in the air and 2.0 percent in the average ammonia content in the corn (Table 7.1). There errors are considered tolerable 186 in this investigation. Even thougm the inlet ammonia concentration and the grain moisture content influence the selection of At and Az, their variations in practical application are usually very small and do not significantly alter the selection criteria for At and Az. 4. The Ammonia Breakthrough Characteristics 4-A. The Adsorption Concentration Profiles The analysis of the ammonia concentration profiles is one of the methods of investigating the breakthrough characteristics. The concen- tration profiles of ammonia in the air in the experimental bin are shown in Figure 7.9. The applied ammonia is completely adsorbed before one hour of atmroniation and no ammonia is found in the exhaust air. As time progresses, the adsorption of ammonia in the corn cannot keep up with the amount of ammonia applied, and 121118 more ammonia is emausted in the air. After about 15 hours of ammoniation under the standard conditions, the corn in the bin is almost completely saturated with ammonia and thus the amount of ammonia applied equals the amount of ammonia lost in the exhaust air. The ammonia adsorption profile of free ammonia at different times in the corn is shown in Figure 7.10. The shape of the curves is similar to the ammonia concentration profiles in the air except that the curves are shifted upwards beginning at a lower portion of the grain bed. The free ammonia content at positions very close to the air inlet decreases with time due to the drying of corn. As the grain loses moisture, the free ammonia content also decreases since the water moisture in corn is the only active component in the grain holding the free ammonia. 187 5 ‘ 0 9 8 4 i— 5 m 6 v .9; x 3 "' ‘3 4 8 .g 2 (’3 2 l p- l ADSORPTIQ‘I 0 L I L l 0.0 0.2 0.4 0.6 0.8 1.0 Antonia concentration in the air, dimensionless Figure 7.9. The concentration profile of ammonia in the ‘air at different times (in tours). 188 10 Grain depth, 1:30.48 cm h ABSORPTION 0 l l 0.0 0.5 1.0 4 1.5 Free armonia content in corn,x10 ,d.b. different times (in hours). 20 Figure 7.10.T‘he concentration profile of free ammonia in corn at 2.0 189 The free ammonia holding capacity of the corn becomes larger as time progresses. The free ammonia concentration profiles are shifted toward the right-hand side (higher concentration) in Figure 7.10 due to the combination of the following effects: 1) Changes in grain moisture content: Some drying is taking place 2) near the bottom of the grain bed (the relative humidity of the inlet air is 80 percent). The inlet air soon becanes saturated with moisture as it travels through the grain bed. The grain moisture content is not significantly changed at bed positions higher than 61 cm (2 ft) at the end of the 20-hour treatment. Thus the drying effect only influences the portion of the grain below 61 cm. Changes in grain temperature: The drying of grain tends to decrease the grain temperature due to evaporative cooling. ()1 the other hand ammonia adsorption tends to increase the grain temperature due to the release of adsorption heat. The combined effects was found to decrease the grain temperature under the standard conditions. Thus, the heat of adsorption is not large enough to compensate for the evaporative cooling of the grain. The air temperature, and accordingly the grain temperature, is lowered as the air passes through the drying zone. It is this cooling effect which increases the free ammonia-holding capacity of the grain since cooler grain can adsorb a larger amount of free ammonia (see the ammonia adsorp- tion isotherm in Figure 3.5). At grain positions closer to the bottom more drying takes place and the free ammonia adsorbing capacities are reduced due to lower grain moisture content . However, the reduction is not large enough to compensate for the increase in the ammonia-holding capacity due to the 190 temperature decrease . 4-B. The Desorption Concentration Profiles The desorption concentration profiles were also calculated for the standard conditions except for the initial and inlet ammonia. condi- tions. The initial ammonia concentration in the air in the bin was 0.05 percent. The free ammnia concentration in the corn was calculated and defined as the equilibrium ammonia content at 10°C. The inlet air was ammonia-free. The desorption profiles for the amronia in the air (Figure 7.11) are similar, but not identical, to the minor image of the adsorption profiles. This is due to the difference between the desorption profiles for the free ammonia in corn (Figure 7.12) and the desorption profiles (Figure 7 .10). The lower portion of the curves in Figure 7.12 is differ- ent from that in Figure 7 .10. Despite more significant drying in the portion of the grain closer to the bin bottom, the desorption process removes the ammonia from the air in the void spaces and the corn in those positions very rapidly. Therefore, the shifting of desorption curves is not apparait (Figure 7.12) due to changes in the grain moisture content , temperature, and accordingly, the holding capacities of free ammonia in corn close to the bin bottom. Nonetheless, the decrease in the grain temperature is more pronOmeed in the desorption than in the adsorption process. The combination of the evaporative cooling due to drying in parallel with desorption cooling of ammonia is expected to further decrease the grain temperatures. As a result, the prbcess of free ammonia desorption in corn is slowed down. 191 5 / 114/// 13 2 11 s 1 9 a 5 4 no [I 7 “I 5 $3 I'd v 4 '5 33 3 2 y 2 1 1 I masomxou 0 l l I l 0.0 0.2 0.4 0.6 0.8 1.0 Ammonia concentration in the air , dimensionless Figure 7.ll.The desorption profile of ammonia in the air at different times (in hours). 192 6 15 1o 5 — 7 6 4 8 m 4 "i 0 “£2 3 3 5 g 2 a 2 ' 1 ,1, 0330mm 0 1 J p 0-0 0.5 1.0 1.5 4 2.0 Ammonia concentration in com,xlo ,d.b. Figure 7. 12. The desorption profile of free ammonia in corn at different times (in hours). 193 4-C. The Ammonia Adsorption and Desorption Rates The rates of ammonia adsorption at different levels in the grain bin under standard amroniation conditions are shown in Figure 7.13. The adsorption rate is the highest when the ammonia-free corn first encounters amronia in the air. This is expected since there are more vacant "adsorption sites" available. The adsorption rate drops rapidly as fewer vacant adsorption sites become available and approaches zero asymptotically. At higher bed levels in the bin, less ammonia is left in the air stream since most of the ammonia applied has already been adsorbed by the corn at the lower bed levels. The adsorption rate reaches a maxi- mum value of hm a (from Equations 6.8 and 6.9), or 0.0937 kg-NH3/(hr m2) under standard conditions, as soon as the ammonia-free corn (x2" = x2 = 0 in Equation 6.9) first encounters amrenia. The ammonia desorption rates are shown in Figure 7.14. The shape of the curves resembles the rate of drying curves. The absolute values of the ammonia desorption rate increase as the aeration time increases. In the beginning of the desorption process, the desorption is not significant at the upper grain levels. The inlet ammonia-free air mixes with the armenia "saturated" air and corn (saturated at 0.05 percent ammonia concentration in the air) and soon becomes saturated in the air; no more desorption occurs once the ammonia concentration reaches 0.05 percent in the air. As more and more ammonia is removed, the air in the bin becomes more unsaturated. The absolute value of the desorption rate increases as shown in Figure 7 . l4 . The desorption rate at the lower bin level decreases rapidly and the corn at that level soon becomes practically ammonia-free; the desorption zone thus moves 194 100 Grain level 80 - 1. 30.5 cm (1 ft) 2. 61.0 cm (2 ft) 3. 91.4 cm (3 ft) 4. 121.9 cm (4 ft) 5. 152.4 cm (5 ft) 6. 182.9 cm (6 ft) ax o l Ammonia adsorption rate, kg-NH3/ (hr. m2)x103 N ob O O l l 0 2 4 6 8 10 12 Time of ammoniation, hr Figure 7. l3. Ammonia adsorption rate at different grain levels. 195 0 6 ma —30~ NA 4 6 E. \1 3 i -60_ 5’4.” .8 ii -90 r- 8 r8 _o 2 Grain level _ 1. 30.5 cm (1 ft) 12° "2. 61.0 cm (2 ft) 3. 91.4 cm (3 ft) 4. 121.9 cm (4 ft) 5. 152.4 cm (5 ft) 6. 182.9 cm (6 ft) -150 l ' 1 1 I l 0 2 4 6 8 10 Time of aeration, hr Figure 7.14.Ammonia desorption rate at different grain levels. 196 forward. 4-D. The Breakthrough Curves The breakthrough curves under standard conditions are shown in Figure 7.15 for the adsorption and Figure 7.16 for the desorption. The smaller the grain depth the steeper is the breakthrough curve. It requires 34 hours to reach a 90 percent breakthrough for 6.1 m of grain. Thus, the ammonia concentration in the air will reach 90 percent of that of the inlet air after 34 hours of ammoniation. (h the other hand, it takes about four hours for 0.31 m of grain to reach 90 percent break- through. The shape of the breakthrough curves is greatly influenced by the following factors: (1) the airflow rate (larger airflow will shorten the breakthrough times), (2) the mass transfer rate (larger mass transfer rates will generate steeper breakthrough curves), (3) the grain moisture content (larger grain moisture contents will adsorb more ammonia and thus prolong the breakthrough time), and (4) the grain temperature (grain at lower temperature tends to prolong the breakthrough time since the cooler grain has higher ammonia-holding capacities). The desorption breakthrough curves (Figure 7 .16) are also influenced by the above factors in parallel to the adsorption process. The 90 percent desorption breakthrough time is about 38 hours for 6. l m of grain which is longer than that in the adsorption process . The desorp— tion process lowers the grain temperatures and increases the breakthrough time. 5. The Sensitivity Tests of the Important Parameters The purpose of the sensitivity tests was to investigate the influence of certain parameters in the simulation model on the predicted ammonia 197 noon—380 gm Moos: 936 aggxowun 93 so flame 56.3 no gag .34. canoes .5 5033.958 no Q5 2. mm on ma em 3 S m o q . _ _ _ 7 1‘ o o 3. 285185 1 to .. we e 36 a mm: a 86 e «a.m. a 84 a 8.0 a and 1 we otn ssermrsuamrp'rre sum up uonerrueouoo 12eran 198 6:033:00 g Moon: 960 swag—noun 83388 on» so fiooo fiche mo 8533625 canoes 83mg mo SHE. ov mm om mm on ma ssetuorsuamtplrre sq: u'; uorqenueouoo etuounmz 199 concentrations. The investigated parameters included are: (l) the initial grain moisture content, (2) the initial grain terperature, (3) the airflow rate, (4) the ratio of the mass/heat transfer coefficients (KI-I), and (5) the division factor (NF), i.e. the nmber of time intervals in which the first inside node is subdivided at the first computing time interval. 5—A. The Initial Grain Moisture Content The influence of grain moisture content on the ammonia adsorption process is rather straightforward since water in the corn is assumed to be the only active component in holding the free ammonia. The more ammonia is adsorbed by the com, the less ammonia is left in the air under the same inlet conditions. The average amrenia concentration for different grain moisture content is shown in Figure 7.17. The range of grain moisture content tested is from 15 to 36 percent, w.b. It shows that the higher the grain moisture content, the longer time the ammonia takes to break through the bed. This is further evidenced by the difference in the ammonia concentration profiles in the air after three hours of armenia- tion at different grain moisture contents (Figure 7.18). The correspond- ing concentration profiles for free ammonia in the corn (shown in Figure 7.19) demonstrate an interesting variation at the lower portion of the gain bed. At grain” moisture contents higher than 20 percent (w.b. ), the S-shaped curves are due to the evaporative cooling effect . However, the concentration profiles at 15 percent moisture are shifted'to the left at the lower bed positions. This is the result of moisture read- sorption instead of drying since the grain moisture content is lower than the corresponding equilibrium moisture content. The combined 200 noose—mung 3g 0:» so young regimen: 5M5 amend...“ mo ousooflfi .54. 0563 H: 50.33% mo gm. ON 3.. 0.... 3” NH 3.. m w v N o o 4 _ _ _ n _ . . o o m \\\\\ s \ R \o‘ E ['I'l'l'lll""|WI-Illllll llallllll'llll'l"lll..ln|llilllunll“\ \\ Q \ \\\\s\\\\ \\\ \_.1 o.H choc sun 3:956 8km \\ \ \ m ..\.\\\\... . \ a. III- .I I. 9 II IIIIIIIIIIIIIIIII ‘IIIIIII‘ \\ \\ m \ x J o.~ m . x m s \ \ n s \ D. \ x n. m . \ o.m m 6.3 «mm .m \ x 4 . . U 6.3 mmm .~ .W x o .63 $4 A \. \ .o..v \ \ o €380 onstage fine .\ not an 5 \H 1 o e o \ mg \\ t\\\ \\\\ \\ \\ InWIIIIII. E11 O.m 201 5 - Grain moisture content 1. 15% w.b. 5 2. 20% w.b. 3. 25% w.b. 4. 30% w.b. 5. 35% w.b. 4 . 1 6 2 3:? 3 .. g 3 x E 2 .. s d l .. 0 1 1 1 1 0.0 0.2 0.4 0.6 0.8 Ammonia concentration in the air, dimensionless Figure 7.18. Influence of grain moisture content on the armenia concentration profile in the air after 3 hours of 1.0 202 6 5 Grain moisture content 1. 15% w.b. 2. 20% w.b. 3. 25% w.b. 4 __ 4. 30% w.b. 5. 35% w.b. E 2: 3 . O m >< «8 a 2 ‘ \. d \ 1 b 3 4 0 I l . 0.0 0.5 1.0 1.5 2.0 2.5 4 Free ammonia content in corn,x10 ,d.b. Figure 7.19. Influence of grain moisture content on the free ammonia concentration profile in corn after 3 hours of ammoniation. 203 effect of amonia adsorption and grain rewetting increases the grain temperatures which subsequently results in a decrease of the free annonia contents in the corn. As the noisture content is lowered, the free ammonia is more evenly distributed throughout the bed in the first few hours of the annoniation process. The lower the grain moisture content, the uore annonia is left in the grain void fraction and nore annonia is thus lost in the exhaust air (see Figure 7.18). Knowing that grain at a lower moisture content adsorbs anmonia at a slower rate, it is advisable to apply a smaller inlet annonia concentration or use an air recycling systen in the amronia grain drying process. An air recycling system will not only reduce the ammonia loss but also reduce the possibility of grain rewetting. 5-B. The Initial Grain Temperature The initial grain terperature was varied from -l.l to 32.2°C. This range covers the temperature for nost anbient air drying conditions in the U.S. The variation of annonia concentration in the air with the initial grain terperature is not large as shown in Figure 7.20. The average annonia concentration in the air is slightly higher for a higher initial grain terperature in the first few hours (about nine hours under standard conditions) of amnoniation. The buildup of free aunonia in the corn at a lower grain temperature is faster than at a higher grain temperature as a result of the changes of the equilibrium annonia content with temperature (Figure 7.21). The average annonia adsorption rate is proportional to the slope of the average free armonia concentration curve in Figure 7.21. The adsorption rate of the grain with a lower initial grain tenperature is faster in the beginning but slower after about nine hours of annoniation. 204 .Hflo my? 5 83935050 3:95 93 so 03989.3 fioum Hand: mo 09832.88 893m “5.853535 mo mans ma ca m o _ 41 c6 O H °q°p' orx'rge em tr; uopenueomo ermnmz 0%.? .q 95.3 .m coda .m oomém A 33m? 53m amid: o.N C O M V O O V o.m 205 .930 G." accuse «Ag oouu mango may no 838095”. fioum Hag 05 no 8533 .HmKounmflm .5 50.3ng no mafia. _ _ 0.0. unannoo £8.55 moon 1 °. H N I"! I *1 o .-| r m o m 3,, ' emaiedxm utters 96mm: H I 1 - q' ¢\ 0 ' n3 q p' orx'iueiuoo manna sex; 959mm: 17 coat“: v S .. 9.0.3 .m 9.23 .m oo~.~m .H mHfimnomeu 59.6 Agog O V mm J o.m 206 This result is consistent with that of the average ammonia concentra- tions in the air (Figure 7.20) as the curves cross and switch relative positions at about nine hours. The upper group of curves in Figure 7.21 indicates convergence of the average grain temperatures as time increases. The inlet air tempera- ture and humidity remain the same for different initial grain tempera- tures. The two groups of curves for the grain terperature and the free ammonia content in Figure 7 .21 are expected to converge into twoseparate fixed values after a long-time treatment . The telperature curve at 10°C initial grain temperature drops slightly with time because the drying-cooling effect is larger than the ammonia adsorption heating effect. 5-C. The Airflow Rate The airflow rate is one of the most important parameters in all types of grain drying problems. The unsaturated air humidity (RH < 1.0) is the driving force for the drying process in the case of ambient air grain drying. The drying rate of the total grain bed depends on the degree of unsaturation of the inlet air as well as the rate of the supply of the unsaturated air. Analogously, the ammonia adsorption rate is determined by a driving force and the airflow rate. The range of airflows investigated falls between 0.305 cum/m2 (1.0 cfm/ftz) to 6.096 elm/m2 (20 cfm/ftz). It should be noted that as the airflow rate increases, the rate of ammonia input (kg/hr) increases proportionally if the inlet ammonia concentration (in percentage) remains the same . Thus, more ammonia is applied at larger airflow rates during the same period of ammoniation. 207 More ammonia is lost in the exhaust air after three hours of amronia— tion when a higher airflow rate is used as shown in Figure 7.22. The buildup of ammonia in the air is faster at a higher airflow rate after the same period of ammoniation under standard conditions. Unlike the influence of the initial grain temperature and moisture content on the ammonia concentrations, the rate of accumulation of both ammonia in the air and in the corn is increased by a larger airflow rate as shown in Figures 7.23 and 7.24. This is an indication of larger amoimts of ammonia being supplied when larger airflow rates are used. The sensitivity of ammonia accumulation to the airflow is more significant at smaller airflow rates. At an airflow rate larger than 2.29 (mm/m2 (7.5 cfm/ftz), the increase in the speed of ammonia penetra- tion is less sensitive than that of smaller airflow rates. Further increase in the airflow rate (larger than 6.10 elm/m2) will not only increase the ammonia loss in the exhaust air but also do very little in improving the ammonia breakthrough time. Similar shapes for the curves were found in the penetration curves for the free ammonia concentration in the corn (Figure 7.24) and the ammonia concentration in the air (Figure 7 .23). This conclusion is based on the fact that a larger airflow rate will not only increase the ammonia concentration in the air but also will increase the mass transfer rate (Equations 6.29 and 6.42). 5-D. The Ratio of the Mass/Heat Transfer Coefficients The estimation of the value of the heat transfer coefficient has been presented in Chapter 6 . Assuming constant thermal properties for the air, the heat transfer coefficient is solely dependent on the airflow rate as shown in Equation 6.29. Thus, the mass transfer rate is determined 208 6.10 4.58 4 h- 3005 2.29 1.43 0.76 Grain depth, x30.48 cm 0.31 0 l -[ l I 0.0 1.0 2.0 3.0 4 4.0 5.0 Ammonia concentration in the air,x10 ,d.b. Figure 7.22.Influeme of airflow rate (cmm/mz) on the ammonia concentration profile in the air after 3 tours of ammoniation. 209 .uwn on» 5 Sflnbnmogo mg 3396 on» so Ame—>59 mum“ scam-HQ may no. 005335 .mmfimusmfim as .833..th no man. en es «H on m h e N o _ _ _ - _ _ _ 0.0 .86 I7 °q°p' o-[x'rge an tr; uormenueouoo erucmme ebexaAv 210 .500 5 unouooo £958 8mm 93 co Amgv gnu scammed 05 mo moon—sag .vmfioupmfim us 50338.5 no 9:9 ma ea NH OH m m w . _ A. _ _ _ Hm.o $5.0 mv.H mN.N o. mm.v 3H6 o.o Ln 0 o O H m...” C 0 N °q-p' porx’uotaenueouoo eruaime 831; afiezenv 211 by the airflow rate and the specified ratio for mass /heat coefficients (KH). This section focuses on the investigation of the influence of the ICE ratio on the ammonia adsorption process. At constant airflow rate the mass transfer coefficient is proportional to the KH ratio which is assumed constant. The influence of the ICE ratio on the free ammonia concentration profile is shown in Figure 7.25. The test range of the KH ratio was 3 —6 between 1.0 x 10‘ and 1.0 x 10 . As the KH ratio is higher than 1.0 x 10"3 , the mess transfer coefficient is almost large enou@ to have instant equilibrium between the ammonia in the air and in the corn. The crossover of the curves in Figure 7.25 is due to the higher armonia concentration in the air at the upper portion of the bed when a smaller mass transfer coefficient is used. In other words, more ammonia is left in the air and carried to the upper levels when the mass transfer rate is lower. An extremely slow adsorption process is found at the smallest KH ratio. The free ammonia content remains almost constant at bed levels higher than 15 an after three hours of ammoniation as indicated in Figure 7.25. This is because the adsorption rate is so small that most of the ammonia in the inlet air is carried through the bed with only a mall amount transferred to the corn. The entire grain bed encounters about the same ammonia concentration in the air and adsorbs ammonia at the same rate. The bottom level was assumed to reach equilibrium with the inlet ammonia instantaneously no matter how small the mass transfer rate is. The free ammonia concentration after the inlet node drops considerably and remains essentially constant throughout the bed for the smallest KH ratio. 212 Grain depth, x30.48 cm 0 JL 1 0.0 0.5 1.0 4 1.5 2.0 Free ammonia content, x10 ,d.b. Figure 7.25.Influence of the ratio of mass/heat transfer coefficients on the free ammonia profile in corn after 3 homrs of ammoniation . 213 The ammonia concentration profiles in the air are expected to be of similar shape to the mirror image of Figure 7 .25. The penetration curves of the average ammonia concentration are presented in Figure 7.26 for different KH ratios. The accumulation of ammonia is more sensitive to the mass transfer coefficient when a smaller ratio is used as indicated by the shape of curves using KH 8 1.0 x 10"6 in Figure 7 .26. The crossover phenomenon of the ammonia penetration curves in. the air in Figure 7.26 can be explained by considering the value of the KH ratio. After about eight hours of ammonia treatment under the standard conditions, the concentration curves for KH = 1.0 x 10-5 or higher are getting closer to the saturation. The adsorption is thus reduced considerably. On the other hand, the free ammonia contents in corn are still far from saturation for values of KH = 1.0 x 10.6, and also the reduction in the adsorption rate is less significant. The "delayed" adsorption at small m-I ratios explains a lower ammonia concen- tration in the air after the crossover region in Figure 7 .26. 5-E. The Inlet Almonia Concentrations The sensitivity test of the inlet ammonia concentration was designed to investigate its effect on the change in the ammonia concentration profiles. Che of the major problems in applying ammonia grain drying on a practical basis is the selection of a proper ammonia flow rate to utilize the armonia efficiently by reducing the ammonia loss in the exhaust air to a minimum. The ammonia concentration profile in the air is shown in Figure 7.27 for various levels of inlet ammonia concentration. At the end of a three- hour ammoniation, the ammonia concentration gradient across the bed is larger as a higher inlet ammonia concentration is applied. The amount 214 .cuoomfifiguwnflflficoflouucoosoo mag momenta as» so 3:33:80 young “gnome mo canon 05 no 8§HHGH .mmfi museum us schemes—9:5 no go. 2 3 m o n _ o o m H e a - O; N W m . as m 1o m m m m n. .m mucus Es m N o.m m. Toggle .m n or euocsénfi A n. n. . H 13. n2 o.m 215 6 5 Inlet ammonia concentration in the air 1. 0.05% 2. 0.10% 3. 0.30% 4 _ 4. 0.50% E 00 “I i 3 " r8 .5 2 _ 83 2 3 4 l .. 0 J . l 0 10 20 30 4 40 Atmonia concentration in the air, x10 ,d.b. Figure 7. 27 . Ammonia concentration profile in the air at different inlet ammonia concentrations in the air after 3 hours of ammoniation. 216 of ammonia lost in the air is not significant after three hours of ammonia application even at an ammonia concentration in the inlet air of as high as 0.5 percent. The comparison between the fraction of ammonia loss at different amionia inlet levels in the exhaust air is presented in Figure 7.28 for different inlet ammonia levels. It is surprising to find that the relative concentrations (based on their corresponding inlet concentrar tions) of ammonia.in the exhaust air are nearly independent of the applied ammonia concentrations. About half of the ammonia is lost in the air after eight hours of ammonia treatment at a constant inlet ammonia level. Despite the similarities of the relative amount of ammonia loss, the results in Figure 7.28 still imply that a larger (absolute) amount of ammonia is lost in the air at the higher inlet ammonia concentrations. The ammonia loss will increase as the ammonia- tion time increases. This is the reason why the inlet ammonia concentra- tion should.be maintained at a lower level in order to reduce the ammonia loss. Although a smaller inlet ammonia concentration has the advantages of reduced ammonia loss and an even distribution of ammonia during the early stages of ammoniation, the maximum level of free ammonia adsorp— tion in the corn is limited by the ammonia concentration levels in the air. There exists a minimum inlet ammonia level in order to maintain a minimum level of free ammonia content in the corn to be able to effectively inhibit microbiological activities. The determination of a minimum level of inlet ammonia depends on the temperature, humidity, grain moisture content, and.the physical condition of the grain. The corresponding profiles of the free ammonia content for differ- ent ammonia inlet concentrations are compared in Figure 7.29 after 217 .maosroa and: 3.5.53 nonhuman up Han anagram. may 5.” poo." mg no 838nm.mm.h 9:53 on .53....«558 no mere ma 0H m o . _ o.o I V 6 91:11me rem an ;o tIOTZIDEH .. as m some .m some .m shoe 4 he an 5 cowuouuowocoo mason-so #35 1 m .o Grain depth, x30.48 cm 0 Figure 7.29. 218 I. l Inlet ammonia concentration in the air 1. 0.05% 2. 0.10% .. 3. 0.30% 4. 0.50% t. l 2 3 4 j l 0 5 10 15 4 20 Free ammonia cmtent in the corn,x10 ,d.b. Free ammonia concentration profile in the corn at different inlet ammonia concentrations after 3 tours of ammoniation . 219 three hours of ammonia treatment. The shape of the curves resarbles that of the ammonia concentration profiles in the air (Figure 7 .27) except that the lower portion of the curves is slightly shifted in the direction of the higher concentration. The O. 5 percent concentration curves in Figure 7.29 indicate that even at 0.5 percent inlet ammonia concentration, the amount of adsorption heat released due to ammonia adsorption is still less than the latent heat loss due to the drying process and therefore the grain terperature decreases . 5-F. The Division Factor The division factor in the simulation model was designed to over— come instability often encomtered in numerical intemtion processes using finite difference methods. Unexpectedly, the instability did not occur throughout the entire tests. The significance of using the division factor (NT) is shown in Figures 7.30 and 7.31. The following three computing schemes are compared: (1) At = 0.1 hour, NT = 1, (2) At = 1.0 hour, NT = 10, and (3) At = 1.0 hour, NT = 1. ' The ammonia concentration profile in the air after two hours of ammoniation are compared in Figure 7.30 using the above NT values. Even though the division factor was only applied once in the computation at the first inside node and the first computing time increment, signifi- cantly different results are obtained. The computation using At = 0.1 hour gives a more accurate numerical result than a At = 1.0 hour. The computing scheme using At = 1.0 hour and NT = 10 predicts a more accurate result at the lower portion of the bed (Figure 7.30) but not as good as using At = 1.0 hour and NT = 1 at the upper levels. The improvement in the computational accuracy closer to the bed bottom 220 6 , ‘i l I ‘\ \ 5 \‘ \‘ Computing time scheme “ 1. At==0.l hr, NT:- 1 h 2. At=-1.0 hr, N'I‘=10 \ 3. At=1.0 hr, mr= 1 4 ‘ ' 5 a) “I C '33 3 r8 '5 5 2 o l L l l 0.0 1.0 2.0 3.0 44.0 5.0 Ammonia concentration in the air, x10 ,d.b. Figure 7.30. Amronia concentration profile in the air using different carputjng sclerms after: 2 hours of ammoniation . 221 3.0 1. At=0.l hr, me 1 2/’ A 2.5 .. 2. At=1.o hr, NT=10 , 3. AFLO hr, m 1 3/ ’ / I, / I ’ I . // , 2.0— ’1 .9 / rq I [I v , / o Ammonia I I: in the air / ~ I 1. i / é 5’ / , to I // 4’3 ,/ / 3 / ' 8 1.0— / ' / ° / Free ammonia ’4’,’ / incorn /’,’3 I I ’,” a) .. ,’ 8‘ h.— I, a ” a; 0.5 ’1’ /’}’//’ a: I/ ’/ /’,/, 0.0 / l l, ' 0 l 2 3 4 Time of ammoniation , hr Figure 7.31.Average ammonia concentrations in the air and in the corn using different computing sctanes. 222 is expected since that is where the division factor functions. The results of the average ammonia concentration in the air and in the corn (Figure 7.31) show the same trend in improving the accuracy close to the bed bottom for the computation using NI‘ = 10. After the first hour of computation, the division factor is no longer operative and the second and third computing schemes using At = 1.0 hr predict parallel results. Nevertheless, the residual effect of the division factor indicates a slightly better accuracy at a value of NT = 10 at prolonged treatment. Both results using the second and third cmputing schemes are farther apart from the results using At = 0.1 hour after four hours of ammoniation. 6. Experimental Verification of the Ammonia Sorption and Drying Model The experimental results presented in this section were obtained from the bin tests performed at Michigan State University. The tests were designed to verify the simulation model as well as to gain practical ammonia-assisted ambient air grain drying experience. The sampling frequency was limited due to the long-time nature of the test and the lack of sufficient manpower to perform the necessary chemical analyses regularly as would have been desired. 6-A. The Ammonia Concentration in the Air during Adsorption (hoe the ammonia treatment has been started, the residuals built up in the corn. The residuals are assured to be a permanent part of the ammonia-treated corn. Further aeration can only remove the free ammonia in the corn and the ammonia in the void spaces. The major concern in this part of the experiment was the change of the ammonia concentration in the air. 223 Before the adsorption test began, the corn was aerated with ammonia- free ambient air for about three days. The purpose of this process was to make sure that there was no free ammonia within the system. Each air sampling period lasted 10 minutes; a set of six samples fram six levels in the him was taken at the same time as described in Chapter 5. The samples were taken continuously for about two hours. The measured average ammonia concentration values in the air are shown in Figure 7.32 along with the simulation model results predicted by the simulation model. The predicted curves were generated by defining the initial condi- tions as the experimentally measured conditions at the start of the ammonia treatment. The measured initial conditions include the grain moisture content and the temperature. The predicted results in Figure 7 .32 are found to have satisfactory correlation with the experiment results. The correlation coefficient between the experimental and the predicted results is 0.918. The model predicts lower values in the early stages of the test but higher values after about one hour of ammoniation. The reason for this deviation can be explained by either of the following factors or their combined effect: (1) an incorrect value of the mass transfer coefficient, (2) the error in the measurement of the airflow rate and the deviation of the airflow from ideal plug-type in the actual bin system, and (3) the error in the prediction of the grain and air temperatures. The results in Figure 7.26 imply that a curve with a smaller slope can be obtained by using a smaller mass transfer coefficient which would have reduced the predic- tion errors. The airflow deviates somewhat from the ideal plug-type. In addition, the uneven distribution of ammonia at the_ same depth level will create significant errors in the experimentally determined ammonia Ammonia concentration in the air, x104,d.b. 224 5.0 :“ o l ABSORPTION w 0 o l N O o l 1.0 r- 0.0 l I L l 1 0.0 0.3 0.6 0.9 1.2 1.5 1.8 Time of ammoniation, hr Figure 7.32.Compariscm of the predicted and the experimental values of tl'eammoniaconcentrationintheairintheamronia adsorption test. 225 concentrations. The error in the prediction of the grain terperatures is less significant since the ammoniation tests lasted less than two hours during which the temperature fluctuations were small. The comparison between the experimental concentration profiles in the air and the predicted profiles is presented in Figure 7.33. Even though the experimental results for the concentration profiles at a specific ammoniation time do not form a smooth curve, the relative positions of the measured results agree reasonably well with the pre- dicted results throughout the entire bed. The measured ammonia concen- trations in the air at a specific bed level, in general, increase with time within the prediction range. 6-B. The Ammonia Concentration in the Air during Desorption The desorption tests are similar to the adsorption tests except for the initial and inlet conditions. The entire grain sample in the bin was treated with a constant ammonia flow for an extended period of time (about two days) to guarantee an ammonia-corn equilibrium condition before the desorption tests began. The profiles of grain temperature, moisture content, and free ammonia content at the start of the ammonia desorption tests are shown in Figure 7.34. The ammonia concentration in the air was 0.0475 percent throughout the bed before the tests started. The initial free ammonia concentration at the different bin levels is different even in equilibrium with a constant ammonia concen— tration in the air. The difference is attributed to the variations in the grain moisture content and temperature. The inlet air was the ambient air free of ammonia. Grain depth, x30.48 cm o l l l l 0.0 1.0 2.0 3.0 4 4.0 Ammonia concentration in the air, x10 ,d.b. Figure 7.33.Camparison of the predicted and the experimental ammonia concentration profile in the air in the ammonia adsorption test. 5.0 Grain depth, x30.48 cm 227 6 5 r- 4 .— 3 .. 2 _— l _. 0 l m 0.0 0.5 1.0 1.54 2.0 2.5 Free ammonia content, x10 ,d.b. L 1 L l 1 1 .14 .15 .16 .17 .18 .19 Grain moisture ccntent,w.b. 1 l J I 1 I 2.0 4.0 6.0 8.0 10.0 12.0 Temperature, °C Figure 7.34.The measured grain temperature, moisture content and free ammonia content at the start of the ammonia desorption test. ' 228 The experimental results of the desorption test are carpared with the predicted values in Figure 7.35. The model predicts the average ammonia concentrations with reasonable accuracy. The deviation between the predicted and experimental results can be explained by the same reasoning as for the adsorption process. The correlation coefficient is 0. 782. Most of the experimentally determined desorption concentration values fall within the neighborhood of the predicted values as shown in Figure 7.36. The experimental results after one hour of desorption are closer to the predicted results than those at earlier times. Despite the deviation of experimental results from the predicted values, it is encouraging to have such a close comparison between the experimental results with the "ideal" model with so many assumptions and necessary guesses of some 1mknown parameters such as the mass transfer rate, heat of adsorption, ammonia sorption isotherm, and the amount of residuals . 6-C. Comparison of the Experimental and Predicted Ammonia Concentration Profiles The experimental results which are compared with the predicted results were obtained from the second part of the experiment - the constant ammonia application rate (0.0475 percent ammonia) period. The grain and air sample were obtained at most of the test period. The selected experimental results were taken at the following times: 10, 22, 140, 161, 242, and 262 hours. The total test period in the second part of the experiment was 384 hours; the ammonia application was ended at the 264th hour. Ammonia concentration in the air,x104,d.b. 229 5.0 X 4.0 .— 300 — 2.0 - DESORPTICN 1.0 r- 0.0 L L J l l 0.0 0.3 0.6 0.9 1.2 1.5 1.8 Time of aeration, hr Figure 7 35 Comparison of the predicted and the erqaerimental values oftreammoniaconcentrationintleairintteamronia desorption test. 230 6 Time of aeration 5 _ l. X 0.33 hr 2. O 0.67 hr 3. A 1.00 hr 4. D 1.33 hr 4 .. Grain depth, x104,d.b. o acae'x 1 ‘ 000 1.0 200 300 4400 ’ 500 Ammonia concentration in the air, x10 ,d.b. Figure 7.36.Ccmparison of the predicted and the experimental values of the ammonia concentration profile in the air in the ammonia desorption test. 231 The’measured.ammonia concentration in the air, the total ammonia content and the free ammonia content in corn were compared with the predicted results. The ammonia was applied four hours after the fan had started to stabilize the grain terperature with the ambient air. At the 10th hour (Figure 7.37), the corn had been treated with ammonia continuously for six hours . The predicted ammonia concentration pro- files had not yet completely broken through as indicated by the unsaturation of ammonia in the air at grain levels higher than one meter. The experimental results of the armonia concentration in the air are lower than the predicted values at the tenth hour. Although the pre- diction model accepts variations in the inlet conditions at any input time interval, the prediction of ammonia contents in the corn is compli- cated'by the constant varying inlet ambient air'temperatures and humidities as indicated in Figure 5.4. The corn had been treated with a constant inlet amronia concentra— tion continuously for 18 hours in the samples taken at the 22nd hour (Figure 7.38). The air in the bin was predicted to be completely saturated with ammonia at the inlet concentration level. However, the experimental results are lower. Furthermore, the model predicts higher ammonia.concentrations in the corn at higher levels in the grain bed.due to higher moisture contents and lower grain temperatures. The experi- mental results showed a.simdlar trend in that region. The changes of grain temperatures lag behind.the inlet air temperature which was constantly changing. The density of the corn is much greater than that for the air. At higher levels of the grain bin, the time lag is even greater. 232 um. msfloufldwocoo mg no magma» 32.898 on» not @3033on 9.3 mo 53%.? . N. muomfim w .uggmogggfigfioag .6 $5? .596 Quad v m N H o. n. _ . _ oo 104 58528.5 r 4. 4 low m. P cuoofinflcosfiaoupe m < X 4 m. X l . e X on n. .m m o o o .0... 1 p. 0 ov mu. flamed? kw o.m 233 .uggmogggfigfigmg uo maggcoog mg no mum—nag 32.398 of can @03on 05 mo 83.3950 .mmfi magmas so aroma .fiamo cacao e m a m N H o . _. n} _ . _ o o lo...” < choofioflcghoofim m. V a a n a m. Q o.~ m . X W»! X 3. X K 9 50053555395 U. 1 i; .m R 0 .7 .m. 0 o. G O toe 0 . agfiflg k or W o.m 234 At the 140th hour, the grain had been treated with ammonia continu- ously for 27 hours (Figure 7.39) as shown in Figure 5.6. The ammonia concentrations in the corn were lower than those at the 22nd hour due to the higher ambient air temperatures and, consequently, the higher grain temperatures. The ammonia concentrations were formd to be consistently higher at higher levels in the grain than at the lower portions. This is due to the higher moisture contents at higher levels. Good agreement was observed between the experimental and.predicted results at the lGlst hour of treatment (Figure 7.40) which is at the end of the second ammonia injection period after 48 hours of ammonia treatment at a constant inlet ammonia concentration. Air samples were not taken at the l6lst hour. The results in Figure 7.41 show the same degree of agreement as the previous sample at the 242nd hour which is after 24 hours of ammoniation in the third ammonia injection period as indicated in Figure 5.6. The last ammonia samples were analyzed at the 262nd.hour (Figure 7.42), 44 hours after the ammoniation at constant inlet ammonia level in the third amronia injection period. Better agreerent between the experimental and predicted results were found for the amronia concentrations in the air at the 262nd hour than for the samples at the other times. However, the experimental results of ammonia concentra- tions in the corn failed to show higher concentrations at higher levels in the grain as did the predicted results. 235 #2393. no nude cocoon map 5 Has 5o: 0:» um. ncoflonucmocoo mg Mo was? Bow—named 93 not mongoose on» no anemone—comm .5 9:53 :0 maiomx .5306 5mg o m e m m a o a q _ _ _ 500538.538?“ L a 1 T < _ LV\<‘ 0 cuoofiofiDEBHonDu I a . x 11 v x x 1 0 L H.335; A.» d... 05 can o.m °. N °q'p'vo‘[x 'uorranueomoo epnmrv o4 o.m 236 eggs... mo ”rune 9808 m5 5 .50: #33 on» no gfloficmocoo 3.555 mo mmsdg Hogan—fig as» one oouoeooum m5 m0 83.398 .315. magma EU meémx .23 a5 0 m e m N H o _ a _ _ . o.o :Hoo 5 deco—=5 monk X can fl .4 A V x \ X X 5 an 5 mg L 23 C AW ‘ o.m 239 6—D. Comparison of the Experimental and Predicted Grain Moisture Contents and Temperatures Even though the prediction of the ammonia concentrations is the major concern in the ammonia sorption and drying model, the simulation model also predicts the grain moisture content and the temperature. The reason for the emphasis on the prediction of ammonia concentrations in this investigation is that the simulation of the chemical preservation process in a fixed-bed grain drying system is a new research area. Fixed—bed grain drying models have been developed, tested, and modified by various researchers (Roth and Bakker—Arkema, 197 7 , and Rugumayo , 1979). The addition of the ammonia sorption process to the fixed-bed grain drying model necessitated major modifications in the fixed-bed grain drying model developed by Roth and Bakker-Arkema (1977) . The ammonia drying model not only describes the ammonia sorption behavior but also predicts the grain temperature and moisture content as influenced by the ammonia sorption process. The verification of the ammonia sorpt ion characteristics has been presented in the previous sections. This section will cover the verification of the predicted grain moisture content and temperature. The ammonia grain drying experiment lasted for 1744 hours. The experimental values of the grain terperature and moisture content of the ammonia preservation and drying test were compared at different bin levels with the predicted results at the following times: 13, 392, 1447, and 1744 hours after the start of the experiment. . The initial grain moisture content was 25.61 percent w.b. At the 13th hour of the experiment, there was little drying in the grain and only the portion of the grain near the bottom of the bin was predicted 240 to have lost some moisture as shown in Figure 7.43. The thermocouple readings of the temperature actually read the air temperatures in the void.spaces. The grain temperature was assumed to be the same as the air temperature. With varying inlet air temperatures, this assumption .might be partially responsible for the error in the prediction of grain temperatures as compared to the experimental results. Only the latent heat and sensible heat changes are taken into account in the fonmulation of the temperature equations (Equations 6.14 and 6.16). In addition, the heat generation due to the dry matter degradation, and the respiration of grain and of the microorganisms:may attribute the terperature increase. Still the maximum deviation of the predicted results from the experimental results was less than 3°C as shown in Figure 7.43. After 392 hours of ambient air drying (Figure 7.44), the upper 2/3 of the grain still stayed at the initial moisture content. The grain was drier as it was closer to the bin bottomlin.the lower 1/3 portion of the grain. The predicted grain moisture contents are in good agreement with the experimental results. The simulation model predict higher temperatures in most of the grain bin. The shape of the temperature profile reflects the variations of the inlet air condi— tions. Theoretically, the inlet air reaches the top of the bin in less than 30 seconds at an airflow rate of 1.43 m/min in a 1.83 m—high bin of corn with 38 percent void spaces. The density of corn is approxi- mately 580 times greater than that of the air while the specific heat of the air (0.242 cal/g°C) is about the same as that for the corn (0.268 cal/g°C). In the case of using an infinite heat transfer coefficient, the temperature variation in the corn at different levels 241 Turperaturefic 5 A10 15 20 6 = 1‘1 I I F i A l \ 5‘- ‘ A x i . \ A 52m 5 4 - \ A Jr \ 94°: \ 3 \ A ~ \ *5; 3 " \ A " rc3 .5 \ A a \ 2 - \ A x \ A \ l _T‘enperature \\ A X \ 0 l A I 'K .10 .15 .20 .25 .30 Figure 7.43. Ccmparison of the predicted and the experimental values of the grain moisture content and temperature at the 13th hour of experiment. Terperature,? -6 -4 -2 O 2 6 A I 1 [Pr 1 ,- ’fr A l l s .. A : x l A l I Temperature boisttn'e 5 4 P A : x content 2: A l 9; I x 3 A I x 5 I 8* l 'U A I .5 I 5 2 - A { \ A \ \ 1 — A A 0 I h I .10 .15 .20 .25 .30 misture oontent,w.b. Figure 7.44.Canparison of the predicted and the experimental values of the grain moisture content and terperature at the 392nd hour of experiment. . 243 will follow the air temperature change at a speed of 1/600 of that in the air. In other words, it would take five hours for the grain to reach the same terperature as the inlet air of 1.43 m/min is used and the terperature change due to mass transfer is neglected. The results of the above two samples were taken during the first part of the experiment which lasted 1360 hours. The wet grain was kept in the bin throughout the winter without aeration due to the severe cold which lasted about three months. 'Ihe drying front had traveled to about 1/3 fran the bin top at the l447th hour of the experiment (Figure 7.45). 'lhe model predicts the grain moisture contents accurately except at the top level. The temperatures throughout the bin were nearly constant and good correlation was found between the experimental and predicted results in Figure 7 .45. The last set of samples for the grain moisture determination was taken at the end of the experiment after 1744 hours fan operation (Figure 7.46). The model predicts higher moisture contents and lower temperatures than the experimental results throughout the entire grain bed. The desirable final average grain moisture content was 15. 5 percent w.b. Both the experimental and the predicted results showed some over- drying near the bottom of the bin. The model predicts higher moisture contents at positions close to the bin top than the experimental values. A second drying front is predicted at an equilibrium moisture content of about 13.5 percent moving from the bottom of the bin as shown in Figure 7.46. The experimental results also indicate the existence of a second drying front. The formation of the new drying front is due to the warm and dry weather of the spring which is capable of drying the grain to a moisture content lower than the desired final grain moisture Grain depth, x30.48 cm Tarperature,°c -2 0 2 4 6 7‘ I I I: I l '9 l. 5 - X \ \\ A \ \ 4 — A \ } A I] 3 — Nbisture X A I content I A! l 2 I— X A! 'Tenperature AI \ \ 1 — x A\ A \ o ‘a 1 ’ fir—L .10 .15 .20 .25 .30 .35 Moisture content,w.b. Figure 7.45. Comparison of the predicted and the experimental values of the grain moisture content and temperature at the l447th hour of experiment. Grain depth, x30.48 an 245 'Ie'aperature,°c 8 10 12 JD! 16 18 6 I 1‘ I I “T I * I \ \ A \ 5 - \ A X \ A l | A x‘ 4 "' l l ‘ A mist-me 'IVemperature “ content 3 - A \ \ \ A \\ 2 P- \ \ \ \ \ l h- 0 J I 1* .00 .05 .10 i .15 .20 .25 Moisture content,w.b. Figure 7.46.Cauparison of the predicted and the experimental values of the grain moisture content and terperature at the l744th hour of experiment. 246 content of 15. 5 percent . As more moisture is evaporated near the bottom of the bin, not much drying takes place at the top of the bin since the air is saturated before it reached the top. The final average moisture content was determined as 15.61 percent w.b. while the model predicted a final moisture content of 17.40 percent w.b. The grain drying rate depends on the ambient air temperatures and humidities as indicated in Equation 6.12. A higher grain temperature (6) and a lower relative humidity (RH) in the inlet air will predict a smaller moisture ratio (MR) and, consequently, a lower moisture content. If the model predicts lower terperatures in the grain for an extended period of time, the predicted grain drying rate is decreased and thus higher grain moisture contents are predicted as shown in Figure 7.46. 6-E. The Experimental Results of Mold Contamination The mold contamination was determined at various stages of the corn drying and preservation test . The grain sample taken immediately after harvest was free of mold contamination. The average mold number during the test is shown in Figure 7.47 together with the change in grain moisture content and the amount of ammonia added during the various test periods. During the first 360 hours of the drying test, the average mold number increased rapidly despite the addition of ammonia. The average mold count was less than 10,000/g which is still below the dangerous level. Still, this is an indication that the ammonia treat- ment was not sufficient for the wet grain to completely inhibit the mold growth. Had a higher dosage of ammonia been applied, the mold count would have been lower. After 360 hours, the mold growth leveled off because of the cold terperatures. Significant mold growth was found at the top after the 600th hour in spite of the cold ambient 247 .383 8d... 2.” 82%.... 3:85.. mo ”Enoch Ba. .3893 053 58 mg dado m5 9% chafing 5.on no» can 9:380 gnu acne mango .9860 39: manage 05 5933 cEmooflnHon 05..va 353m u: .059 coma coma OONH com com com o a. a _ I _ _ _ _ . a o 939v \ I” m // 9.31m .386 . ,m S. I xx 334 Boom I o N m. // m a. x m a x: o m // £500 SQ: mango I. I . I l 06 w cm I I T. m . I, w u I m . I m d I/I/ .M . // 9:380 8330: o . o m mu. mm. I I!!! 9.33 amouo>¢ J III m ”woman? . 95538 I/ a mo puma 380m _ no Hana umHE // em. I! 0.x 248 temperatures. The mold growth at this stage was mostly attributed to the significant growth at the upper portion of the grain him as shown in Figure 7 .48. The mold growth at the lower portion of the grain bed was not obvious during this period even without the ammonia. The application of ammonia at around the l400th hour reduced the mold comit significantly at all levels (Figure 7.48). The mold growth then increases again after the armoniation has been stopped at the 1623rd hour. Visible molds were detected at the grain level just below the top surface at the 1360th hour. The infected portion of the grain was less than 20 on deep. The visible mold growth was concentrated on the broken corn kernels and was contained significantly after the ammonia was applied. The inhibitory effects of ammonia on the grain molds could readily be observed by the naked eye. A representative sample of ammonia-treated corn taken at the end of the experiment was sent to the U. S. Grain Marketing Research Laboratory (Miller, 1979) for determination of the mold count. One seed in 25 and one seed in 100 were found containing F‘usarium in two separate tests. The ammonia-treated sample was fomd to be cleaner than the regular commercial samples. No Aspergillus and Penicilliun were found. Thus no aflatoxin test was performed. 6-F. Energy Consumption in Ammonia Grain Drying The ammonia grain drying is one of the alternative energy-saving grain drying techniques. In this particular bin experiment, 3048 kg (120 bu) of corn was dried from 25.6 percent (w.b.) moisture content to 15.6 percent. The test weight (density) for the wet corn is 682 kg/ms. Thus, a total amount of 342 kg water was removed throughout the entire 249 .mmmoouc 055 860 3:95am 0.4.35 can ung can 93 5 mafia» can magma ucmummmao no 358 can! no £03an 9?. we . N. Roman Hz . 959 coca coma coma com coo c.~ O Q' etdmes 111216 rad sprout go :umoo 601 O \O c.c 250 drying process. The l/20-hp fan Operated for 1744 hours and the total amount of ammonia applied was 18.4 kg. The energy requirement in the ammonia grain drying process consists of two major parts (excluding loading and unloading): the energy in producing ammonia and the fan operation energy. The energy required in the ammonia porduotion is 5.598 x 104 KJ/kg and the total amount of energy in producing 18.4 kg of ammonia is 1.030 x 106 KJ. The l/20-hp_ fan consumes energy at a rate of 37.3 watts (1.343 x 102 KJ/hr) and the total fan energy amounts to 2.342 x 105 KJ. The total amount of energy required in the ammonia grain drying was 1.264 x 106 KJ, among which 81.47 percent for the production of ammonia and only 18.53 percent for the fan operation . The dryer efficiency is 3696 KJ/kg—water (1589 BTU/lb-water) removed which is a saving of over 50 percent compared to 7500 KJ/kg-water removed for high-temperature dryer and 33 percent saving compared to 5500 KJ/kg-water for him drying systems. CHAPTER 8 (INCLUSIONS A theoretical analysis of the ammonia-assisted grain drying process has been made, a five-equation model was developed and solved numerically. An experimental bin test of ammonia grain drying has been performed employing 4.35 m3 (120 bu) of com. The 1.83 m deep corn was dried from 25.6 to 15.6 percent (w.b.) moisture within 72 days and 16 hours of fan operation at 1.43 (mm/m2 (1 cfm/bu) without unacceptable grain spoilage. The total amount of anhydrous ammonia applied was 18. 3 kg. The five—equation model solves the dynamic changes of the ammonia concentration in the air and in the corn with respect to time, position, temperature, humidity and grain moisture content. A sensitivity model established the important parameters in the predic- tion model. A newly-developed ammonia sorption isotherm, which is approximated from the ammonia-water solution isotherm, serves as an essential stepstone in solving the equilibrated condition between the grain and the ammonia. The computer simulation program of the five—equation model was tested and compared with the experimental results. Good agreement 251 10. 252 between the experimental and predicted results was obtained. The prediction model is dynamic and is capable of handling varying ambient conditions , inlet ammonia concentrations , and other grain properties at any specified time interval which can be variable. The ammonia sorption and drying model was tested for numerical accuracy. A set of values for the computing time and depth intervals was established at 1.0 hour and 15.3 cm (0.5 ft) which were shown to have reasonable prediction accuracy and do not consme an excess amotmt of computing time. The sensitivity of the important parameters in the prediction model has been determined. It was fomd that the ammonia adsorption rate could be significantly increased with a higher grain moisture content and a lower grain temperature . The prediction model can be used to simulate a practical-scale experiment with a set of predetermined initial and operating condi- tions without actually performing the time—consuming experiment . Ch the other hand, the predict ion model enables the design of future experiments in a more efficient manner and is able to predict the results of a future experiment . The general grain drying simulation model using fungicidal chemicals was established. The general model can easily be made specific to simulate the treatment of a certain type of grain with a specific fungicidal chemical by specifying relevant property parameters. CHAPTER9 SUGGESTICNSFURFUIURERE‘SEARCH Conduct experiments to determine the following important but unknown parameters: (1) the mass transfer coefficient between the ammonia and the grain, (2) .the heat of adsorption and desorption of ammonia in the grain, (3) the amount of fixed-ammonia (residual) in the grain as a function of ammonia concentration in the air, free ammonia content in the corn, airflow rate, grain moisture content , temperature and time, and (4) the ammonia multiplier (MA) in the determination of dry matter loss and/or mold growth in the grain. Conduct an experiment to determine the ammonia adsorption and desorp— tion isotherms and compare with the equations developed in Chapter 3. Conduct an experiment to establish the dry matter loss or the mold growth as a function of the temperature and the free ammonia content in the grain. Perform experiments with grain samples other than corn, such as wheat , sorghum, rice, soybeans, barley, peanuts, pea beans, etc. , and compare the experimental results with the predicted ones of the amrenia sorption and drying model. Perform experiments with fimgicidal chemicals other than anhydrous ammonia, such as propionic acid, acetic acid, methylene bis-propionate 253 254 (MBP), formalin, sulfur dioxide, etc. , and compare with the pre- diction model. Compare the fungicidal effect of the same amount of ammonia in two identical bins at the same location by using a trickle (intermittent injection) process and a continuous (at a constant rate) process. APPENDI® 255 APPENDIX A UNIT CUWERSIONS Quantity Unit Equivalent Airflow m3/unn (dun) 3.532 x 101 gm dun/m2 3.281 cfm/ft calm/kg 8.971 x 102 cfm/bu cum/ton 8.830 x 10-1 cfm/bu Area m2 1.076 x 101 ft2 Convective heat W/m2 K 1 . 762 X 1031 BTU ft °C transfer coefficient 3.600 x 10 J /hr K Convective mass m/hr 3.281 ft /hr transfer coefficient Density kg/m3 6.242 x 10‘: 1b/ft3 7.768 x 10" lb/bu (corn) Diffusivity mZ/hr 1.076 x 101 ft2/hr Energy J 9.479 x 10:: BTU 2.390 x 10 Kcal Force N 2 .260 x 10"1 1bf Heat content J /kg 4.299 x 10‘4 BTU/lb Heat transfer rate W/m3 9.662 x 10.2 BTU/hr ft3 length m 3.281 ft Mass kg 2.205 lb _3 1.000 x 10 _4 tonne 9.843 x 10 ton Mass transfer rate* kg/hr m2 2.046 x 10':1 lb/hr ft2 Power W 1.341 x 10.3 hp 3.414 BTU/hr Pressure N /m2 9. 872 x 10'? atmosphere 4.014 x 10:4 in.water 1.450 x 10___3 psi 7.500 x 10 mm Hg * The mass transfer rate can also represent the mass transfer coefficient as defined in Equation 6.11. 256 APPENDIX A Quantity Unit Equivalent Specific heat J/kg‘C 2.388 x 10'4 B'IU/lb°F Specific surface area m2/m3 3.048 x 10"1 ftZ/ft3 Thermal conductivity W/m K 5.778 BTU/hr ft°F 3.600 x 103 J/hr m K Velocity m/hr 3.281 ft /hr Viscosity kg/m hr 6.719 lb/ft hr Volume m3 3.531 x 101 ft3 2.838 x 101 bu 2.642 x 102 U.S. gal APPENDHB FORIRANLISTDIGOFTHEAMMNIASGRPTIG‘IAMDDRYDBDDDEL 257 37 615: .296507.018ré96307515§49g0751857.9 63074185296307é193329 03071. 1357.9 01 5.? 518.); 96307415529530. 09.12....3465077399C11235456572. 3.701..4.433455407589001 3:. .. 15.267730. 901. :. I .92....691 599.....- 1... ..... .. .15/.867 0 at... 33. . 2., 1. 111111.... 110 1122222362?.r......2....:.u.{1.333.515 .7..<.?:«.?..1.2th1~1~ hL‘LLL/n......h .....5<..555$r (Jr _. .Sc....é.€6 1C5 6A. 01.7055 14.77777 uw E R F. S M. «I 7 D T n... R 9 c. .q TL A E c T I R Y 8 P .3 THE C F I .U . F U c \I .6 D G \l TAD P S ’1 I. an Mn N T P D .O 0 o I F. U IPr .c ( 31 C R Y. c. D. V. E R :4 I 8 E Lon T uh T SEM C F I6 . T AU U T U R I D )R \I 01 50.» V F F. .C Rn... E T ED 5 I Z or; 7H RF. T 3 .. CRF XE 6) 7 .n o .9 EU U C 1 C RACR IE (.3 9 7. I 51‘. T 05 s C H 5 EEC F 8.0 1 I o P T. NE M. L E T E VGELMS D \I \lI-l\ 90 VI I Iv... GO TA H F F 40 one: 5 5 IR I C. I 5645 D I N NU T S E A INTTRA x .9 6)? N 30 TNIE EHT OT 0 9:0 I C T CG F... B It 1 (1C 0 9/ INT TT Ll. A II N w.» I hTRECA CF I. u. 2 02E 3 c7. TPCh #9.... C F CS 1.1 P. C T JRT I R C 5(2 P 91 Asuuuun LEA RN R 1...: o I R .N U LRDCFN NA. I I I ZFX v. 92 Eva. UCT EC an can" ET F A F P. U n... 06 \l 9 XE I 0 9X AJAN v. 0 CC. F NN .ow . F. Pour TN MI...- 5 C IF.) u" )9 I I T. IGT E D 2.5 .9195 R S G I Cum TI X” a I )‘IDS T 29N NYF Sun 3 TL F. CAT-REF I a. he 2 [EN In.» A I I I\ 9.. 5.) IO .( t of. 050.8 IT Na. 8 C UFPI A a I C A . LN v.0. M. C 66" Q/Z v... I ETT TI APII I C T l. TNIPU I UC FC X I ((SR N ”BX KGNN EU I T C. RI,“ C G N N :. ECH NT. C... I 7. RRCF nu X3 I ABC... F TOT. C tunnel—PTO“ 3.. T. F AIAKXE IT IR \1 C IR(R I tKN IInr. Own INEN F0. IL I \I I6 F R A RII SF T I 2 I )202 T 1 f. RRTM TOREHI T Snuv. IEH 5 C. G P T TGRAFS R LU I 6 ZXEX C 0N1 0E9..." CUPT r. NHRTDK T. RE I . H EC D EC 5 C I IT I N 00X FER R..TECSB AAEN I I at 07.. R R T SNGR I HS F. D I 5 I 5) I) U SC I NC:- EXAR:UU .Q R VETH I F9 E \I F. U EH5 TD 5 IV ( 5 65).) F +KD TISU YLCC C IPS TEEHTF 0 7 E 1 F F C KTN A U TC H. V. C (6.36 T .1. A60 R5193” (Scum/Mn E E {L o RU P“ 0 T K01 R H I D IN I I Z(,O( H (KC ENAN CDDITNFENFTRTRRT 0 h. c U 0 S U N DARH C.» S E as? C‘ICE cu XR(R R IJV HE G N UE TFCEFACF I 1 F... K C H C .b C F! v... 5 IC RZCZ C (can? 7. II FT. F. IEFGN C IF. No.3... :- a L 0 T I 8...... a. nu A .8 T I P. TX I 1X02 H STD CLUT U I MTCIM... UHETIP.FS .4 T) . I o D. T .56 urn... P. K Han IS I I 1. 7. I IX U 31A I... U HHIUVLAT. U/HU TCS I Hi» CC + E U GAB" N AP. D G C TC ICNINA C (I) I I .7." FTTL STTTC..CRR£..L¢.TFE I I X E5 CZ VI nu F 9 .0 R” E . 0 A I IflEl‘I-ITC I .355) R g E I TRAP” NP... HGPTFS TVXNO S 2.9 52.1 .L up. IR Fara GA 5‘ R P TowTMIT 4C IRZ 6.0.05 8 (IL CE.CS AUEZLGTM. GnhHEFUIl—IT I) C o SC C. .C T I C UT 9. NSAA JXA IC (((6 T. P C EYIH IDCNRCL RI“...U...TEH .15 F1 4 o C U! (x HUD RE A H 8 T X/FN/ IN”. I 1XR( L XE I P..." F T IONA... A 9T0 II To 73:. / PL X NFH SD TN 1 U R Y/IS/NXHII OXAFR ET. ENG SUEU INTTH TAU FNTT I I6 I I Ut ole... H TE 1.. CTE C T CKQTITT SEA lTS/YSI/ IC 1 PEP VUO. . IC RH o DIN USRTIQLUrUNNHE R 0 010». a. RT T U0 F IN NCEUCPP VHC NRSEAG1Tnu/TC 231 L06 oT I ETTTO own-«IF OITLA ITITHI)... . .OEF TFUS P P...” S ILX TID IQTAPSS NLCN TPCFRAXHXSN/ XXX CC .1 IR HV ACOURRuUQIFNIFTUTRRFOTT 00 oFC TU.... E D a... Poem.“ TLhEECD CUYU naRRARL/S/N/C S 980H/TNSET:CGPrh= TNRCICPQLCN. .5 11C ( Iran-4: UTE—P. A ”UT. Iran UBLRZAA 1559. M FPNAF1DNUT1NNNN 5.3) I I . EIHI T : =CT1=TITT3 I: DXQT FFTYQ FPFTPTo T... R. C T I’ll/lXfiXCNCCUCC LNLVuCEh S CRTHR..N:3:A: ARTE" I It ($7.6.FGH HFHLFYCUS n. CF. C R C [Ill/ITIIIHATO XHA... IunPflAPUTPTITnaN : NR... : F01 .2U((.(RT : H .. FTTNFI A S .b N NNNNNNNNNNNNSSSSKNFT (.513 GCCH:PPDLRI.UIVF: =9T99R0.NTTTEPQR:RMEFIV .. R I F. U U UOUPUCUUUUGOCNNNN RF... T ILRRF... PNENEFNEPFCHCPE P I.» fifiucfl.OU=.InJTI:r P G C 5 I. HMMSMMHNMMMHEECELEO)(AA5CFHFPPIDTDIIXTCTKRITDDICTw“ u. FED CPRC: =1CD U R YMMVWMRMMKMPNNHVAT THTT3 DT AA(ANKRR =3hUU TDD T R 4'. «b.p.leruflIfilCCCCCCIIIIE x (RAIN o IRE EEC. EC.D can. prF K T I... n. N 6L F In CCCCCCCCCCCCDDUDRE. FEED I ”T 9.2.1:.FFTF HPH...F DIV. ...J It it..otttoifitttt..titi.t * +.... I............. a t .. ation..ttoi.ttcti*..Qoia... a .-.. o............. t o o In t9..pat9.alIum*..thti.o * .... t............. t a . into....tett.riaiii..ittt.t I .... t............. c. I - 1900'..+th.OOI§t¢-.uptat.s .I g... b............a 5C1)... Q v CCCCCCCCCCCCCCCCCCCCCCCCCC C CCCC CCCCCCCCCCCCCC :6 09C C C 258 3071:3522» 650741851~962a071~135~¢96307§15<¢29 L.!..U.I 418.529 0.507 7 5 .. 8.35.9 6.. 0761352967 .075 a. 85296307418§29 OO11137~5567G 996.012.); 45567.73. 90 n. 12236556....7xr $93112... 172.706 .7... 31577390. 017.. :7- 61336778990, . 7.35 656.. 78.69nvCOCOCCCOC000t.1111111111111.-12. L. s... .32.... Li}: 2.4./.22.. .1..5..J<.7..r 7:7 .5 «I77777w v77.....5.8388888€s388889999990.990.999 90 11d- 1 al.?! 11 a. 4. 1a . 140c£111a011111-. 1.1.1.: I 11.0 117. I 1..I.1I a.I.TQ.a..I.1 R E F c. .4. \I an N T I U H T T R T A N A F. n E C E Z 1. H I H N . T 0 E E C 0 D. uw 2 U. E T .U S I V C T A... TT N L L . U 1 CN 3 A B o 1 EE C N N I 1 .. VT C I s c T N“ L I H s N 1 S I 50. A T T N 0 7 N v" (C P R 1.. 0 I 9 C T. L U r» T. T c T. CE T F T I o T F TR. A a... S T D 1 T. I UU nu R v'. .0 H I\ 0 PT F J” N U 9 N T «MS I n. . C t G L 31 9 N T C S T C N CC 8 I A .\u L T C o o H- a U L C A N L 7.7. C F 0 A I I I A T I/ \I . NM...) 5 F E I T T 1 I )) F 1 O )‘I O ) ‘LUN I a: T I I X T) 22) +2 ))2 .I II .3 v. T. D N .1 I2) I 5.- 5 K] 11’ )( RRH C C 5 N N \I T ) N211I . :)oII) )H HEI H I IL N .3 I1 I001 3 4.311.) TR. IT.) R D C I s .3 D1N~D X P1/FP1 ( I ELM R F. E S H U 1 Star-TIL... L K {IKKF v.) LIT n... F SF U L )R 0 . UIN I 1. E 11((K )X) UH F S zUC F r. 9 l 0 I E1. 1 0 I111 I D ?I.12( \I (1 COT N H P. CO TL).M\I N6 - E1 I8 81 ‘KPXXR I) I TE‘ E 1. SR D N1 TEXN CZ ( M. 31.11: D CK+ +2 . (1.1 F U9 R T/ v» E I In. (I. G I 1 CI: I f N T()\Ix o um... II\ “no.1 L8 T ‘n o .1 GS NONH 0C * G :1?!) I an 1+vu11+ .. XIT )) CVAQ. .- o F #0 I nLC I». IT ”.6 Tun. T C) 11.) . )X I I) P. a (()IT. L3H I V: SF )TC 1 «.1 YH‘ 1 02 1 II)1 .1 N1) I G... N K1+111 U1 oTF1(( 1 1 F 319 O. C SE 051 A... r. s . TC 0 Ht, 0 NHODTTN o fIIC)I 3 TIM-MM. SO)1.I\I\ I?” 10 U RTD l.) I. V." ‘C n. In ' u. H) U .(Ix N \I N IT oNIII HII\I\(ROB 111KKK HNT(HHIXP 10rd. .N. INS? . oU) NC XIT I 1 C1HH I T 22 I 1oCINNZ ”(M4 9:4 . A.V.MV((I\ RI(/RF(+ + 0...}... I .nCV t 9 v.9 +CL I F0 C(RR IN T =0 I XQTIIIXT CTXXVXOTIKKK12R :IR)ESHUN ”II I 0 TI)! .1? GR 0 E 1 no TO ”/58 11 o "5 TO 1 o o 53.11291 “(((TC1F1 ((XX.‘ v.1ZIICO LU T/R. 10C TC]... .5 3.... o “up: TECT 8 8 R C Eh. 0 0T 8 H0011 ICC: 0° 0 1.I.092..XXCN 1 I I I I It: .— TM..I\I.\VI. U: XTAfAuvh 935 [V U = 0: RIFfl.G GI 0 IF (T a L I 19.555 oEth9inIT-8 ”Hun-Era oUI 3 8 0:261:31 3 8 RICZEROCUOCT‘LISI‘X : 2.13-5.51: I.“ 0H. U a. PC? USEIOJEEF C1 0. o n» TEP SIC/LIP... 1KTUCTUCU oHN 3)X:RUITN )\J:U12:T )):2U1UES555<.( :(\I)aU..n )XO:$XRUUSHU sIUVFISLtfir. LUURHIUPSC 0+ JUNEINuu V.0:IC)I:\I8 N 11111)“... T3011)XN NTZSEZTR‘ )31)NX71::)H .. ..NSMRN1\I(NNC .- 7 9»: 3:33. a “.0 Q. CKTGIHNICX .. TRO1(\I1)IFNTT I TTUDIu...1 I CI... 10.1”" A0021) I IKI:O(\J)1: NPIHX: IOIXICCORLCRHCISEHY .Lu .u 1JP+ TCITH :r..t.=1 RI fTTIN1IT‘TTANRTTTl‘ZTTTEDDDDDHFIIt IKKKIxTMu1RIT. IN UUT : 2 NT1(§.TCC¢G .1 .THTATK :CHG :ECN (N 01““ IRCICN (0 ((RN CID ((PTN N AARRAR IK(((RNU R((IUSSNNQIN OWN =(: 3 (V :8 :: r..KF...C FD uxHITQTqéfll‘HO FCC127.CCR2FU1222000 EECEEO P...(M1ZCUSCEM.H(Sv.HCIn.-P.a...vC.cn. «AF. CCC 3.7. N. .C 17.. .UJthLr. IC IRHHDTY.Y.HRC I OXXXCG x DXXXXCGC «Rnhpp-FEF DKTXXIAXCHDVIXRHUIxgnCHXQCDTan GIHC H IC P U CC 9 a o o o t i a - c t o. . g o a II. t 1 2. ’3. — i Av. . c . o 0‘ t a o a t a . O - 8 o 1 1 O 2 o S. 0 . L 7 1 I. i Q o o g o . nv . C c (u 4| ‘. 6 1 PJ . (J. a”. o .5 I 0 c a o o c TC TC 2 1 2 7.. 1 1c 2c Q 1 1 CC C CC C 259 #18529 6307.51.85295507£18529b3075135296307418529 630 74135296307‘ 1852963 07618529 630316135. 996307 6. 72309 n .01233‘5567730 CO1..<217~ 456... 780.90 11... .34 5.5.0 6739...9 0... 17.338556 7 389 C .2122? .6 r 567.. 4...». 9. . 122344.356? 89. v 0 93.5.2.5) 44.1» 44.544441. I» I. 4LSSSI..S:I..)I.1DSSSIJ.DI1010.06 6.6.06.0 8.6!.” .0 I419. 7. . 777.7 7.7 777.71 2... .9. 3.2.1.. 9. 50.5...83 cl... s... .v C 019C529? 90.0.9 C93. 0 Ollollp. I 1T Ioo. 14.1.1 TQ|4I1.1108,1114 14. 0014.920.10. at. 19.059 14. a. IIQIQIT RITQSTQIOOTJIT Tolllunvs. a. 11500. 140.... 11 I 19.3.1; .1111. loo-IQ... .a.14. I .0 «In A T EI L \I 3M...“ F. M. II...... C nu I NT” E T nu r.“ T U .9 “NA 5 I 7.. NC A. V . A T F. I E NE L C .4 TUL .3 EIN T E 6 . T S LSI .A 85 E 1. F. S c .NI M». D o L 5 an IVN. 7 TE r... 1. «n IT n1. II 3 LL N . I TU D. oID E8 1. RD. 5 Y5... 5 DC. 0T 7 .n F.T c. TU .. N R 05 c. I S PU R ISN I To. >R C A 5 CC R R D .n S S qu . .2 R u. n... IFP A. RV. IF I: T .U PG .u C "CC E I... P o 1 ( A G h K TL RR FI C F X... I PC A T UHF... GIBB L C .P H a I .h L NI. I U 9. EC NALLTI L .L FT K 1. C ... A R F HEB I UUAB S T I I ( I I .e P I T S IVMT £38 A N T R 1. G T T 55 nu. A U V. IUL EH... CT I . U R C X F. A I .JI H. U C A ATNE BTYTDS P I E... I. 2 L H. .1.-. C T 0* TA 5 c. R—CCN R N I r. \I X C I IR I E R L ALE NTuuCHNIz E «(I n.” T I .L T... n. I E A DEHT CST I I T T B NU 1‘ \I \I R ) IH I T T I RTS IU EETTN N 090 H I .I... \I \I G .L 2 UT 0 T A T U R Tr... .RRSN I U TET TR X 1. 1. N F J N0 V L I L L EUDI I ICUUREI. .U N D \l 5.: I I . I I u. n. I c R ) EONNN‘. DRNTTIVN c o O N \I RI ) 4.. H. T H I C I AA I F Hva 01. AA NEUA...FF.I cu GRG A N IL .3 I\ .H R \I \I T 566 T CJ ID T . n...TCC.R RT 1R E I FI S \l 2 I C \I F \I Fug I If. a I I To); oRnAR Cum 6.3.23va) U \IT) 9.. T U T \I v.» \I 1. P. 2 29.9 UP I 0 H1. SIEPI I UAFPN (500 \IN) 7. D H \I R0 . Z . \I an I I T DPHR I u... R)Ru...n.c.q TGIFFCCZITH 9..an S I. N «LE 1. I \l 2 E 1. E .s. EEO 0.03 (C r. I I3 E D UCNEEITR. I . . 2 M F. I F . 1.. a... I Y I U I. GOT P IN (PL 9... I t..—UTNCN P .LTTS ICU... E... I n... I V 1. T. I.\ I\ I 1 A. E F L I\ 3. PUKVDE I...) P ET INI ARM IDT oGV TTT 1. R \I P 5 Sun I. 7. O I. L I C 3. T PTO Iv. ICAn. PI 0 NT R IUu..RRVE I0 I o o 5 F n...) I I T.» I. X :. T. \I I. V I RG IXC fin I I\ C C .EINDN G.nIIINZT\IT OCT 2 I)“ C UT ..u I 7. f. In T L F R0) .2 INF [H SD. L OVBCNG CH AACGNFOA TN 0 1 H TNI .3 F5 7312 X FII H T S I\ EH1. MN ICM.TTHE SRI IOI I T II). IL LIT 8) R (IS ‘1 TN TZTZI (2H T HE U A .05 o RIM. TPLTE E THTTTCT MSETTFST IEU ETG T . FTP . “I a .21.. H. T ( TN n” H VA 0 NHFaAEEPD .5 UI EAEA ASLEEOEXOHKTD . NC .L ((( 0 n. .\ .TXZE R1 r. FI I F LET... ITCCPDDEI I P oFLRSR NENLL .U I . .ULI\F\I ) a T 8.7V - GI8((7H 91‘...— ( E I VID EH: 0 I I I I I I I0 I T NTINT T YCINN o RSICF. IT 7.7. OT A “(P E o o . ENTCZTI. I NuuD.I.T\I AC C... F3...) 0 TTonFuTZJIucr I! T .16 Inns...” DO:IIUSHFOCDE D. N." 0L L 855 Lnucc TIXII.AXC o IIqrvnIAT)HET .. 9. IV 0 Tn..N4.Iq31.7.1.31 TTEN L: 0‘ IZHEC. 9.7.: :NI £...OIH+L‘HZII 9 o UEDDR at o: = = .61.»... : zanPOETHH: U IIIc.Yu..N.J TFPI FrhII-I‘CZZEZZNZUI EUDECDCTCU DRNZN: 012‘ :4. cc. NTT EN78CUVSH8 UN)\I\I EFU))\IMI:OU=R:)LT(DAXI . J . .H.IRR I N6 IONAMEAN NNNA:IM..IZZ.CTF1.RTERIHXZ .I o LNFPR “7.22410 vrSZCTAITN)a)TA(MuHL= 6219151.? HRPDTTTTTTTTRTIEO:IEIHEOFUIIERTITTT AUPHRHHT: .. TTT 30:21.. 2 :0 I1 I I‘3QUS I I(..T..I2\I.). IVTXR )50 . JSH .. RFIINNNNRNNNFNTB:ETRTCRCDCTQ.RHXTUNNDMEH :HCI\ :NNNNH NTCTNNNTTTXTIIG)TC£1IIUZ(\IT 3. TE: :..L1.P.1.J.. U: A .. (IIIIIIII IN TEN I\ N X InRT II( (NII .. (Ix I ”III a N(((ET ((ETRINTTTI‘ TTTLI\ .. v“ 20 to FREE. RRRPR Rn. RIG .5. C 8 EC... R.) FITTTF‘ Hnb nLVSHUTnH..FT7.2(P.. TZfiLZZICTu (a... ”ruin—“5.... nu. ...r JP C IFPDPFPDP PC HTC I CY. RFI HI ITnuXu. II RG CFPRGNCIKXXD G. XXXYXRCTO an! TXRCX DJJ u... I. t a .t..—. ....... . . . .. I I a t. c i .0.... ....... . . . .. i t t n. 9 fi .9.... ....... . . . .. t a I I. I t .t.... ....... . . 1.2 .. .s I I o. ‘9‘ . §.... .......C .0. S...) a». ..T 2 I p . CC C 9. C I» c (CC-LS; CCCCCCC 8 CI. .2 7 7 C). 0.30 c) I. C.\ t. C C C 260 1585296307‘ 185-(.9 6307‘183961u07‘18529 SO7IQTRJSq‘9630q/‘18529 6.35 .7 “1852‘? 6.55.7.5 1?:3529 550741893259. 633.74 10 ...lo.....£.295) .1. )TD (J I: I C. F M. I T E TIRCCSB )P)(J NC CU ( X EIR H C ITC I I E12 XI...- I ...E Ih J 2 t T T93). O. J T M: US §E+H A \IAZTC IT CC H I.\ "IRS—LTD; 2 R2... T 2.) ...] N CTUG ...... IST J». C .6 S 1‘ I.\ EJ) NET P)RET 2 ((T C: O I\ T IRFC E Y... IC.» ETC . 6)EI LICI Is J..C I X/ I ..4 5 1H H H( . AAIINM.Z [Hts IEJQC RT 0) i F TG "HUSH RIM U H..- CCU... 1 LH I... (PI 5/ v.10 9 E I. . TTO) TTFJE ISTH JH‘TE 0H LJ 6 S U GENAI NXHTC ant. . II +RLX 171... V TUD YSITCT99 U 0) \I ... .)H(\IITMB R (TTDC CT UTHN D ONuaRAE I/ IC/CRtT w” 7.... Art ..DXX l TE LGNP/Cfic G 12 2 ETTZRACT JAE? H. {HA C _ C390 A RITE I T TTuu/FNAI . XIII- avian}: {+ ... SEN R‘IPS/ o I . I I TLL I HHAETTTDéT. ETDH A! HXIC E VFNDN XHXSFUPTT *CTT XTLTu. .a S ESP FLARNCC/ L2 )J 2 AE oJEP.RSTT(ARCDSTT§c.. Q7.e.. T. E UDIFIEE [SILEF i7 )+ X(T.A. .vIIn. T ATRFOTS RJ Z( I LC)(T0.NAOHLH+HHICSD RIEC TA) N RNRIBTN TENUU 5.9 RT¢9$9EH CL ..n Ufa... III’CCle IHT .H6HEUIETUH)EMTRU. TCTN.(HH JY/EF. UJ I .ICDRUG YRXCCchECITEE....E...TGEC4.. .... FV l/OH T .. . C(¢.(UC.. ITPRJDILBCSLI II: ..R D(A2VXO( T S GFF III/[AIS .ETZCCH: .TVEX I » LE NNNNRNTROIIVPEQ ONL)JFH ..(NTEULHE??T)\I:TNHLCL:Eu...“ U IREE "no.“ NNNHRNFTTTAXI. . GH=HLZ( T CC H UCOCUC: : Mug/.1. .707IRZ(: C)Hn..5auloh=DCCszq/.):IX :n..E XR . . HIHHNCO 30405.net: .. (.. .. .. ..ISVJ. n. .. C CST FFRMI.N\IAHI I..JCTCTC TSCJRCE(HCT: : ..E I IJNE..L6$VF .. U R TRTTICC ”MPH”. ED))9.ACDEHIPJXSL2 N Fm whiny. TTO JHIx: 3 N JIxB (I. z a 0123 JJTU IL" L TTD .6 PM“ MFT TT .. NHRN HLC La... U a... F...H.CILI\AC (MF9C90 (FA NF 55 H111 (THC "p.60 C. PEN U FCCP.C..I ((Ionucnfiu 1...... ..1 F CCCCCCEPI HJICraCC TIT RI TH DCCC THRI XCC 5 KM... S CCCCCS. 55.4 CCCC XXI CV, IIIII . . t t. . . t . g ......... . . . wowot . . t 9.7. . *1 . 9 ......... _ . . ‘66:: o . t .6- . . 61. .t ......o.. . . . III... . . ... ... . . a... ... ......... . . . «Ito. . 1C6. I h. . . I ... ......... . . . CCCCC C 4. I. TC C CC C C C C C CCCCCCCCC .L C C. 265 5 C57&I0050505050090 00000000050000000050000004£5500 0.430057000 .. ...: 4.577....699Gn 2.6:}... 783.012-.. 455679023455680 91222:...IIZ 5......)6 7775.16. 777.....777777796u Vac...» 996.063 Gian-0.06.001; 11111.... . 27...? ...... .c . .227... .(7. . .355...55r $35.3: ... 5.3.3.. .... 5...: 56€66.06?!6.65666666666666bbfi. 6 tr. blufbeZAEI. 7. N. .... C I N Il- S A I X 5. nu. "a. . C G T OH \I R N RE I... C D A F7. 5 \J In N T i T I A 5 EN a 7. N EI H ET T R I N 2 5X C IE I1. 7. A I h“ (AD 9| 5 flux T E). C S L N X R). N MN R A nu:- .. C1. In I I U C“. In B.X C \I E II. D To C( X ID I TN 0 1 m. A SI T S XC1 C T X I . VI. "5 A E I2 .9 0A A .... CN R FTC T NV. R X NI E CU I C A G I C T I. L9 ( T: 2 CT A 0 N.LC I 52 In 3. S S R ...S I ) U2 I) 7. AR T N I 83 a v.21 2 X. II N E P. TECE c. X 2... I NF. .4. \IH I CH I I t 2) C? I: h. T 2T T S LT7E I. 2) ”X 2 HE S 2 A T U CN H 1. I C I 7.. HH N 7.5 L N FR .0 7. X0 C1. X AT C XA U A Céc .I. I I 7. I C I C) .l AFC I 1. SF. (... 2 S I TE LX S In x .EN 0X 2 EE D 2.! AA N 555”) i C. I E R.‘ T T I F. «CA CH C F'ACHI. I an AR ... ET X UI T XR 7. C NO CD N CI X ZN I F I E A I EX TIA C IC RII ) I I II Hr. 2 L 2N H I .U FTCCTOC CX 9.1») SH. «.... 0.0 .x ...... 7.7.0 TT E EA IN]: # NC 7. T IA II CO I R III 2 T DU1.C2HE T I IX37. .. X N N OXT R7. A \I QCCZZI. 0 0+ 2 R2 ( E I C E T In. 5X L 2E... I IIIZC quasIITX 1...... 1 N0 22R vb TRNF I F. 7.... «(AXTC CIIEECC. II? T TN XC I I CZCX 2 R TIECNI X... N IIZCTE VI. F To I I" T GTSaxz XT’. oh 1).... flJTt X/ . 2 .N) C. UC N1..." X «CR7. .5. (UXTCZ IN CYTEEX I o o nuE TIC I G \l CRYX Mu Cn. CICOO 2... IGITC C A Rs TINT“ TYH)./.I\ I ..NRD/F( IC TETOX It I I B XSXNI NCTOXT T U3NSF/T( UGX Ii Ium CG I... ”up... [NI/N IT. I X FUAKC7.L+ P IOTCTH NF. N SH TONTI EPEGEAZZT N 5 0a.... .0 C2». AL» L I (In. I I T XCXNFTT. GSNENNZZ+ E PTT N ZCCIaHN 2 IUIIN Ian I 33. UC). T S [III—CNN .011 II ATZTTU Uni-In CSX/IXZC 0N 1OECE0 N 2...... U IR NHNND/IN TAFTF..XXN um TI. «N. II INN..2CN .U .. IF XC... .../.55 21:2. R C H0 0000.....12ION ENEX .. ..IOIZETZR THTCCHII .. R TZTII ..XZXIITOTIXU R TF HMMMRCCT ..ILDID AIIII: .ITRRCCU CT X...M...I..XCU 2N (El‘ZI\= .. H (To a "Marv; .. .. XTIL ( ".22.... N: .. ..TD “1. MM2(NTD 01D FZFZFNULC TEN U 0060 12.. MFA Ir ZTZNCCN al./IEN ..L C. .TFUEH GXC IXIXIXGXC III... S CCCC CCYIIC I XXXIGCX . CCRE F CCXIFRE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I . . . . . . . . . . I. ...C I. 1 5C CCCC C II C C «I. .( CC BIBLIGEAPHY BIBIICERAPHY Agrawal, K. K., B. L. Clary and G. L. Nelson. 1969. Investigation into the theories of desorption isotherms for rough rice and peanuts - I. ASAE Paper No. 69-890. American Association of Cereal Chemists (AACC) . 1975. 1975 Revisions to AACC Approved Methods 45-05:1-9, St. Paul, Minnesota. Associaticn of Official Analytical Chenfists (AOAC) . 1970. Official methods of analysis, 11th ed. 26.051—26.055, Washington,DC. Association of Official Analytical Chanists (AOAC) . 1975. Official methods of analysis. 12th ed. 26.044-26.047. Washingtm,DC. Anderson, H. W., E. W. Nehring and W. R. Wichser. 1975. Aflatoxin oontaminatim of corn in the field. J. of Agric. and Food Chem. 23:775-782. Aris, R. and N. R. Atmmdson. 11973. Mathematical Methods in Chenical Engineering. Vol. 2. Prentice Hall, Englewood Cliffs, N. J. BaJd