volume 39 number i january/february/march 1997laboratory measurement and modelling the effects of...

cSAE

5CGR

The Journal of the Canadian Society of Agricultural EngineeringLa Revue de la Societe Canadienne du Genie Rural

CAE 39(1) 001-076 (1997)CN ISSN 0045-432X

Soil 01/(1 II'aler

LABORATORY MEASUREMENT AND MODELLING THE EFFECTS OF MULCI-IlNG ANDF RROWING ON POST·HARVEST SOIL WATER EROSION ON POTATO LAND

J.e. Leyte. Linnell Ed\lJ.lrds rind l.R. Bunlcy 01

Ellergy alld Processing

IMPROVING SMALL·SCALE COMPOSTING OF APPLE WASTES.F. Barringlon. K. EIl\1oueddcb and B. Porter 09

PERFORMANCE OF DIFFERENT BINDERS DURING ALFALFA PELLETINGL.G. T'lbil, Jr.. S. Sokhansanj and R.T. Tyler . .

Structures Gild Environment

.......................................... 17

POTENTIAL FOR THE PSYCHROPHILIC ANAEROBIC TREATMENT OF SWINE MANUREUSING A SEQUENCING BATCH REACTOR

0.1. Masse. R.L. Droste. KJ. Kennedy. N.K. Palni and l.A. Munroe 25

MICROBIAL INTERACTION D RII G THE ANAEROBIC TREATMENT OF SWINE MANURESLURRY IN A SEQUENCING BATCH REACTOR

0.1. Masse and R.L. Droste 35

NUTRIENT CHARACTERIZATION OF STORED LIQUID HOG MANUREAJ. C.ullpbell. J.A. Macleod and C. Stc\van 43

A GRAIN STORAGE INFORMATION SYSTEM FOR CANADIAN FARMERS ANDGRAIN STORAGE MANAGERS

D.O. Mann. 0.5. Jayas. N.D.G. White. W.E. Muir and M.S. Evans ..

Food Engineering

. 49

................................................................................................... 73

MODELLING OF MICROWA VE DRYING OF GRAPEST.N. Tulasidas. C. Ratti and G.S.Y. Ragh:.tvan 57

Technical Noles

H·OI'CO$T: A SOFTWARE PACKAGE FOR ANALYZING THE COSTS OFOPERATING A MECHANICAL WILD BLUEBERRY HARVESTER

K.J. Sibley and D.L. Arsenault 69

POWER REQUIREMENTS AND BALE CHARACTERISTICS FOR A FIXED AND AVARIABLE CHAMBER BALER

D. Trcmblay. P. Savoie and Q. LcPhal

Volume 39 Number I January/February/March 1997

CANADIAN AGRICULTURAL ENGINEERING1997

January/February/MarchVolume 39, Number I

EDITOR

J.J.R. FeddesDepartment of Agricultural, Food

and Nutritional ScienceUniversity of Alberta

Edmonton, Alberta T6G 2P5

ASSOCIATE EDITORS

Presidem

R.L. KUSHWAHA(Power & Machinery)Department of Agricultural

and Bioresource EngineeringUniversity of SaskatchewanSaskatoon. Saskatchewan S7N 5A9

JJ.LEONARD(Structures & Environment)Department of Agricultural. Food

and Nutritional ScienceUniversity of AlbertaEdmonton. Alberta T6G 2P5

JJ.LEONARDDepartment of Agricultural. Food


K.C. WATTS Past PresidentDepartment of Agricultural EngineeringTechnical University of Nova ScotiaHalifax. Nova Scotia B3J 2X4

S.F. BARRINGTON President-ElectDepartment of Agricultural and

Biosystems EngineeringMacdonald College of McGill UniversitySte. Anne de Bellevue. Quebec H9X 3V9

R.L. KUSHWAHA Vice-Presidem (Technical)Department of Agricultural

and Bioresource EngineeringUniversity of Saskatchewan57 Campus DriveSaskatoon. Saskatchewan S7N 5A9

J.C. JOFRIET Vice-Preside", (Regional)School of EngineeringUniversity of GuelphGuelph. Ontario NIG 2Wl

S.F. BARRINGTON(Waste Management)Department of Agricultural and

Biosystems EngineeringMacdonald College of McGill UniversitySte. Anne de Bellevue. Quebec H9X 3V9

H.STEPPUHN(Soil & Water)Semiarid Prairie Agricultural

Research CentreAgriculture and Agri-Food CanadaBox 1030Swift Current. Saskatchewan S9H 3X2

CSAE COUNCIL 1996-97

DJ. NORUM SecretaryDepartment of Agricultural and

Bioresource EngineeringUniversity of SaskatchewanSaskatoon. Saskatchewan S7N 5A9

R. D. MacDONALD TreasurerAGVIRO Inc.14 Univeristy Ave. W.Guelph. Ontario NIGIN I

JJ.R. FEDDES EditorDepartment of Agricultural. Food


REGIONAL DIRECTORS

V. LALONDE British ColumbiaB.C. Ministry of Agriculture.

Fisheries and Food457 McCallum RoadAbbotsford. British Columbia V2S 8A I

M.V. ELIASON AlbertaAlberta Agriculture. Food and

Rural Development7000 - I I3th StreetEdmonton. Alberta T6H 5T6

D.S.JAYAS(Food Engineering/Energy & Procesing)Department of Biosystems EngineeringUniversity of ManitobaWinnipeg. Manitoba R3T 5V6

L.GAUTHIER(Information & Computer Technologies)Dcpartement des sols et de genie agroalimentaireUmversite LavalSainte-Foy. Quebec GIK 7P4

M.E. JORGENSON SaskatchewanDGH Engineering Ltd.• Sask. Office303 Railway StreetP.O. Box 310Muenster. Saskatchewan SOK 2YO

S. CENKOWSKI ManitobaDepartment of Biosystems EngineeringUmversity of ManitobaWinnipeg. Manitoba R3T 5V6

H.K. HOUSE OmarioOntario Ministry of Agriculture

and FoodBox 159Clinton. Ontario NOM ILO

O. MENARD QuebecMAPAQ3230 Rue Sicotte CP40St-Hyacinthe. Quebec 12S 7B2

KJ. SIBLEY AtlanticSibley Engineering8 Wharf RoadGreat Village. Nova Scotia BOM lLO

Canadian Agriculwral Engineering publishes papers covering the general fields of Agricultural. Food. and Biosystems Engineering that fit into one of the followingclassifications: (I) a scientific paper based on original research: (2) a technical paper based on design. development. testing or analysis of machines. equipment.structures. processes. or practice: (3) a general paper on education relative to curricula and philosophy or trends in science. on a surveyor investigation of somephase of research or research methods. or on extension or extension methods.

Manuscripts for publication should be submitted to the Editor. The papers must be original and must have not been published elsewhere in a refereed publicationor copyrighted. The author. not the CSAE/SCGR. is responsible for opinions expressed. Information published in Canadian Agricultural Engineering may bequoted in whole or in part provided that credit is given to the author and to the journal. Publication charges are $75/page plus cost of translation. if required. Reprintcharges are $16/page for 100 copies.

All claims for missing issues must be made to the address below.

Central Office Address: Box 381. RPO University. Saskatoon. SK S7N 4J8Published Quarterly

Canadian Publications Mail Product Sales Agreement No. 0466247Return Postage Guaranteed

Subscription rate: Canada $50.00 per annumOutside Canada US$50.00 per annum

----_. -- ---- - --- -- --

Laboratory measurement and modellingthe effects of mulching and furrowing on

post-harvest soil water erosion onpotato land

le. LEYTE1, LINNELL EDWARDS2 and l.R. BURNEy3

INova Agri Associates Ltd., RR#I, Port Williams. NS, Canada BOP ITO; 2Research Centre, Agriculture and Agri-FoodCanada, Charlottetown, PEl, Canada CIA 7MB; and 3Department ofAgricultural Engineering, Technical University ofNovaScotia, Halifax, NS, Canada B3J 2X4. Received 5 June 1996; accepted 15 February 1997.

Leyte, J.e., Edwards, L. and Burney, J.R. 1997. Laboratory measurement and modelling the effects of mulching and furrowing onpost-harvest soil water erosion on potato land. Can. Agric. Eng.39:001-007. A laboratory experiment was run on a short length ofreconstructed post-harvest potato furrow to facilitate evaluation ofrunoff and soil erosion parameters for modelling 'cool season' soilloss from potato lands. The experiment comprised the use of acold-room facility to evaluate the separate and integrated effects offurrow side (interrill) and furrow bottom (rill) erosion on a 3.15 mlong section of furrow subjected to prior freeze-thaw cycling. Afactorial arrangement was run with four replications of three treatments comprising bare and straw-mulched surfaces at slopes of 3, 5,and 7%, under sequential 20 min simulated rainfall applications of47 and 94 mmlh. Sediment flux was separately measured on twosections of furrow side and on the furrow bottom during each rainfallapplication. Flow velocity was measured in the furrow bottom usingthe dye method. The straw-mulch cover factor for furrow side erosion was 0.21 for the lower intensity rainfall and 0.44 for the higherintensity. However. the greatest effect of the straw mulch was indiminishing rill erosion in the furrow bottom. An example is presented illustrating the use of parameters calibrated from this andprevious experiments to facilitate modelling of soil loss from furrows of practical field length using the COSSEM simulation model.Keywords: soil water erosion, simulated rainfall, hill-and-furrow,freeze-thaw, slope, straw-mulch, COSSEM model.

On a construit, en laboratoire, un sillon semblable aceux laissesau champ apres la recolte des pommes de terre, afin d'evaluer lesparametres de ruissellement et d'erosion des sols necessaires a lamodelisation des pertes de sol d'un champ de pommes de terre durantla saison froide. Les experiences se sont deroulees dans une chambrefroide oil un sillon de 3.15 m soumis prealablement ades cycles degel-degel a servi a mesurer les impacts distincts et combines deI'erosion sur les cotes et Ie fond du sillon. On a mene une experiencea facteurs multiples comprenant quatre repetitions de trois traitements (surfaces nues et recouvertes d'un paillis avec des pentes de 3.5 et 7%) qui ont tous ete soumis ades sequences de precipitationssimulees de 20 minutes a des -intensites de 47 et 94 mmlhr. On amesure separement les charges de sediments provenant des deuxcotes et du fond du sillon au cours de chacune des sequences deprecipitations. On a utilise la methode de la teinture pour mesurer lavitesse du courant au fond du sillon. Le facteur de couvert du paillispour I'erosion provenant des cotes du sillon etait de 0.21 pour laprecipitation de plus faible intensite et de 0.44 pour la precipitationplus forte. Cependant. Ie paillis a surtout pennis de reduirc I'erosionen rigoles au fond du sillon. On presente ici les resultats d'unesimulation. avec Ie modele COSSEM. des pertes de sol provenant de

sillons de longueur reelle en utilisant les parametres calibres lars decette experience et d'experiences precedentes. Mots-cles: erosionhydrique, pluie simulee, gel et degel, pente, paillis.

INTRODUCTION

Farmland planted to potatoes is very susceptible to soil erosion under the humid climate of Prince Edward Island (PEl)where these fields have traditionally been planted in longnarrow rows, up-and-down the natural slope which commonly reaches about 7% (Nowland 1975). Most of the soilloss from these fields occurs during the cool period of late fallto early spring (Burney and Edwards 1995).

Potato land is especially vulnerable to soil erosion: (a)prior to canopy development where furrowed seedbeds aresubject to spring rains, and (b) following harvesting whichleaves remnant hills and furrows subject to cool-period rainsand snowmelt. Furrows concentrate flow and therefore serveas pre-formed rills.

The high-traffic practices of intensive tillage and mechanical harvesting of potatoes using heavy machinery cause soilstructural breakdown (Edwards 1988). The practice of topkilling the potatoes just before harvest and the resulting lackof crop residue offers no residual resource for ground protection. Surface freeze-thaw during the cool period reduces theaggregate stability of the soil (Edwards 1991) and, in conjunction with the low hydraulic conductivity of a frost layer,leads to an increased potential for soil loss (Edwards andBurney 1989).

Preceding erosion studies of PEl soils have used a varietyof laboratory (Frame et al. 1992) and field (Parsons et a1.1994) procedures with simulated or natural (Edwards et al.1995) precipitation to measure interrill or rill erosion fromsoils cultivated to potatoes; but none of these proceduressimultaneously measured the separate components in an integrated system.

Increased soil erosion in a hill-and-furrow system, withconventional steep furrow side slopes, has long led to speculation of the merits of flat cultivation or minimum tillagesystems for potatoes. Chow and Rees (1994), in the neighbouring province of New Brunswick, measured soil lossesfrom potatoes planted up-and-down slope in a hill-and-furrow

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No. I Janurary/Fcbruary/March 1997

of four times that from flat planted potatoes. The hill-and-furrow system is, nevertheless, deeply rooted in internationalpotato-growing tradition with sound agronomic rationalization. In PEl this system is accepted as integral in 'potatofarming culture' and is likely to remain so for the foreseeablefuture.

At present, potato production in PEl is caught up in thedynamics of greatly increased markets for potatoes with resultant pressure to expand onto steeper slopes and to increaseintensification of currently used land. Concurrently, there ispublic pressure for 'sustainable' production systems andfarmer sensitivity to society's desire for implementation ofenvironmental farm planning. As indicated in Edwards et al.(1995), the practice of post-harvest mulching of potato landsas a relatively inexpensive cool season soil conservationmeasure has expanded rapidly in PEL However, the quantitative effectiveness of this practice is unknown andprohibitively expensive to measure directly. The objective of

which the characteristics are described by MacDougall et al.(1981). Blanket pretreatment comprised saturation andfreeze-thaw cycling of soil formed into a central furrowflanked by adjacent furrow sides. Treatment variables wereslope, simulated rainfall intensity, and surface mulching.

Soil bin

The soil bin was 3.65 m long by 0.65 m wide and wasconstructed from sheets of 0.85 mm thick galvanized steel. A0.15 m wide trough (Fig. 2) was attached along each lengthwise side to balance soil splash into and out of the testsections. The bin was divided across its width to give: (i) a0.5 m upslope portion comprising two furrow sides (each 250mm long, projected horizontally) facing inwards to separatecollection troughs for furrow-side only measurements, and(ii) a 3.15 m long downslope portion for full furrow measurements. The downslope furrow bottom outflow was sampledfrom a vertically-adjustable end-gate which enabled the operator to preclude interference with rill development.

tWatersupply

" -Pressuregauge

t--- ------

Full Furrowfurro~l_sides

~Sump

o J:e>

\...- Fu II fu rrowsample

Fig. 1. Soil bin and rainfall simulator unit.

·2O?500l----.~250--------2-50-----:'500

Horizontal dIstance from furrow centre (mm)

Fig. 2. Cross-section of soil bin.

Fllrow side moritoredeach side at I4lslopeend

m ~j.<..~_I :.:....:.;.:...I~_~O_W-__2_50_mm--Jt;,.,~ "

E§. 400

EXPERIMENTAL EQUIPMENT

A laboratory system was devised to measure furrow-siderunoff and sediment flux together with integrated full furrowmeasurements in a single soil bin. The soil bin and ancillaryequipment (Fig. I) were constructed and housed in a temperature-controlled room described previously by Edwards andBurney (1989). The test soil was a Charlottetown fine sandyloam, the province's dominant arable soil under potatoes, for

this study was to obtain data under controlled laboratoryconditions for calibration and utilization of the COSSEMmodel (Burney and Edwards 1996) to enable prediction offield effects under a wide range of site and event specificconditions.

2 LEYTE. EDWARDS and BURNEY

Four equally-spaced perforated copper pipes which randown the full length of the bottom of the soil bin facilitatedsub-surface saturation (and removal of excess surfaceponding) of each soil bed prior to the freeze-thaw pretreatment. The entire bin was lined with 50 mm of styrofoaminsulation to ensure a natural pattern of freezing and thawing(from the surface downwards). The bin was supported by awooden platform mounted downslope on a pivot and upslopeon hydraulic jacks that permitted slope variation from 0 to 7%.

Rainfall simulation

Rainfall simulation at 47 and 94 mm/h (with measured uniformity coefficients of 85%) was supplied by threeindependent nozzle units each mounted centrally over theupper, central, and lower third of the soil bin. Each unitcomprised a pressure gauge, solenoid valve and a 6.4 mmlOW (Spraying Systems Co., Chicago, IL) brass full-jet nozzle (in the same series as used by Tossell et al. 1990). Any ofthe units could be adjusted vertically to enable the applicablenozzle to be set at I m above the mean of the bed surface.Spray overlap from the nozzles was curtailed with the use ofplastic curtains suspended mid-way between adjacent nozzles. A metal trough along the bottom edge of each side ofeach curtain diverted intercepted spray over the side of thesoil bin.

The higher intensity of 94 mm/h was obtained by usingtwo nozzles (at 180 mm apart) on each unit.

Furrow forming

Based on post-harvest microtopographical surveys using aprofile meter in the field, a wooden roller was constructed toproduce a full scale furrow profile comprising two half-hillsand a central furrow (Fig. 2). Cone penetrometer readingstaken in the field in late November, after potato harvest andbefore freeze-up, were used to determine the amount of addedweight and number of roller passes required for realisticcompaction of the furrow surfaces.

Freeze-thaw cycling

A timer was installed to automate daily freeze-thaw cycling.The minimum temperature was set to -5°C and the maximumvaried between 18 and 20°C as affected by the outside laboratory temperature. Rainfall application was initiated at theend of the last freeze period. The temperature of the watersupply to the nozzles and the air temperature in the enclosurewere maintained as close to freezing (OOC) as feasible duringtesting.

EXPERIMENTAL PROCEDURE

Experimental design

A factorial arrangement of a randomised complete blockdesign was used in four replications. The treatments were: (i)slope (3, 5 and 7%), (ii) rainfall intensity (47 and 94 mm/h)and (iii) surface mulching (bare surface and straw mulch at 4t/ha). The rainfall intensities were run sequentially on thesame bed surface.

Soil preparation

With the soil bin in a horizontal position, the soil bed for eachtest was shaped and compacted in layers using a fixed

number of passes of the wooden roller. Where appropriate,straw was spread on the surface at a rate of 4 t/ha as used inPEl farm practice (Edwards et al. 1995).

The soil was then saturated by subirrigation using thesubsurface pipes attached to a constant-head (0.3 m) watersupply. The pipes were capped for 30 min to ensure evensaturation, and excess water (localised ponding) was thenremoved by drainage back through the pipes.

With the soil bin still in a horizontal position, the refrigeration unit was activated to provide 4.5 daily freeze-thawcycles. Rainfall application began at the end of the last freezeperiod.

Test run procedures

For each bed surface, the soil bin was set to the appropriateslope and the 47 mm/h rainfall intensity was applied for aperiod of 20 min (5 year recurrence interval for PEl) followed by an approximate 10 min break during which each ofthe single nozzles was replaced by a double nozzle set. The94 mm/h rainfall intensity was then applied for a period of 20min (100 year recurrence interval). The same data collectionprocedures were used for each run.

Runoff was collected from three locations - the two furrowside plots and the full furrow plot. For the furrow side plotsthe volumes of runoff were relatively small and therefore theentire amount was collected for the periods of 0-2, 2-4, 4-8,8-12, 12-16 and 16-20 minutes. For the full furrow plot, a10-second sample was taken centrally within each of theabove time periods.

Runoff flow velocity was measured down the furrow bottom during the last half of each run using a fluorescent dye.The dye was injected into the flow 2.5 m from the outlet endof the furrow. Timing began when the most concentrated partof the plume passed a marker 2 m from the end, and the timeat which it passed a I m mark and the end were noted. Thisprocedure was repeated three times and the average velocityover each metre length was determined.

Data analyses

All data were subjected to analysis of variance (ANOVA)and mean separation. A 5% level of significance was observed unless otherwise stated.

RESULTS AND DISCUSSION

Changes in the slope of the furrow bottom had negligibleeffect on the furrow side slopes and therefore ANOVA wascarried out only for cover and rainfall intensity effects for thefurrow side plots.

Total sedimentAs shown in Table I for the furrow side plots, doubling of therainfall intensity resulted in a significant, approximatelyfourfold, increase in soil loss. When split according to cover(cover x intensity interaction) the straw mulch was moreeffective in protecting against soil loss at the lower intensity.

For the furrow plot (Table II), there was no significantdifference between the mean soil loss mass from the 3 and the5% slope treatments. However, each of these means wassignificantly different from that at the 7% slope. Meyer andHarmon ( 1985) similarly reported little effect of slope on soil

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I Janurary/FcbruaryIMarch 1997 3

Means of any treatment not followed by the same letter are

signifcantly (P ::; 0.05) different.

50

....

30 40

Approxinate '... ntenSity. 94 mmlh10 nmuteperiodbetweenruns

fo"~~~~te ntenSity • 94 mmJh

period ~between ' ,

k'r_un_s__~ ; L''O·--·O'',,19..t? --'''E)I ""'. suf,,,,

! 9J-.iil"I/ /'0,..80--0fP L Straw rmkh!;:'

ntensity • 47 mm/h

htensity • 47 mm/h

Sare surface

--0,r ....0--. -0-"'-0

! ~traw mulch .a? ~.e---a-"o _-8'"

01020

2

8

~ ~,r Bare. 7%

: '.: rBare. 5~.~\j r Bare. 3%' ••••,

: \~ '.; --• ..:.-..:.~=~~.::..

~Straw.7% ~

..... s~raw.5%-&;.--6. _

o1.IiI::L.:.L..:.....ii.e::::::!!=:3L ....o-.........I.-:L::.;S~tr~aw~...=;3L-~::-:::!.....o 10 ~ ~ ~ 50

BOO

400

1200

'5u.

25Intensity Cover Section

~--(mm/h) (from top. m)G-.___

Bare 215 - 315.-.~~ ~94

20

I~ 94 Bare 115 - 215

Vi" 15 • __ ._- -.- .---•• --- •••~ 47 Bare 215 - 315g

e __ -....------.e.-.--..--.-~ 47 Bare 115 - 215~U0

10Q3c;-94 Straw 215 - 315

>CD--.---------_..-.-.······~ 94 Straw 115 - 2.15.....

05 ~:.=:-=:-=~:::-::.:- tC:..-- 47 Straw 215 - 315

?------~----~47 Straw 115 - 215

02 6 8 10 12

Slope (%)

Fig. 5. Measured flow velocities in furrow bottom.

Time (mIn)

Fig. 4. Soil loss rate from furrow end.

However, velocity was not significantly affected by bedslope.

Furrow side erodibility

Soil interrill erodibility is commonly defined by an equationgiven by Elliot et al. ( 1989) and used in the WEPP model. As

Time (min)

Fig. 3. Soil loss rate from furrow side (interrill) plots.

94 mm/h

752.5b

47 mm/h

Rainfall intensity

176.0a

(Cover x rainfall intensity)289.2a 1045.3b

62.9c 459.8a

Slope

3% 5% 7%

787a 923u 166l b

(Slope x cover)1302a 14073 2874b

272c 439c 447c

(Slope x rainfall intensity)207c 353c 542c

13683 1493a 2779b

Treatment

BareStraw

Overall

BareStraw

Means of any treatment not followed by the same letter are

significantly (P ::; 0.05) different.

Overall

47 mm/h94 mm/h

Treatment

Table II: Effects of treatments on soil loss (g) for furrowplot

loss from furrows for slopes between 2 and 5%, but substantial effects as slope increased above 6.5%. When splitaccording to cover (cover x slope interaction), bare soil produced significantly greater soil loss than the mulched surfaceat each of the slopes. The higher rainfall intensity similarlyresulted in significantly higher soil losses.

Sediment rateThe average soil loss rate from the furrow side plots ispresented as a function of time in Fig. 3. The average soil lossrate from the full furrow is similarly presented in Fig. 4. Ineach of these two figures, simulated rainfall application was47 mm/h for the period of 0 to 20 min and 94 mm/h for theplotted period of 30 to 50 min. The fixed break of 10 min,used for plotting purposes, approximated reality.

Runoff flow velocityFlow velocity was measured over each of the lower 2 m ofthe furrow bottom during the latter part of each run. Velocityincreased only slightly over the last meter of flow and ispresented according to treatment in Fig. 5. As shown, velocity was highly dependent on cover and rainfall intensity.

Table I: Effects of treatments on soil loss (g/m2) forfurrow side (interrill) plots


(2) 86

Sod lossin furrow

Depositionn 116row

r

InterlSlty • 94 mmlh~StrawmUCh-1----------.--::---..:s: •

~-----------~~----'-Zt~~~:;-:47 nmJh

Intensity. 47 mm/h ", Bare surfaceStraw mulch ,

"

"

'~

-4000 L..- ~ ......I

2

2000

~

c:o

.;;:;;(ijoat -2000"0

Q)Z

used in the COSSEM simulation model (Burney and Edwards 1996) to incorporate a cropping-management factor,this equation may be written as:

Dj = Cj Ki A P SF (I)

where:Di =measured soil loss (kg/s),Cj =cropping-management factor,Kj = soil interrill erodibility (kg-s-m-4),A =area of surface (m2

),I =rainfall intensity (m/s), andSF = slope factor.

The slope factor (Elliot et al. 1989) is:

SF = 1.05 - 0.85 exp (-4 sin 8)

------------------ -----

Table III: Effects of treatments on K;C; (Gg_s·1_m-4) forfurrow side (interrill) plots

where: 8 =slope angle in degrees.

Based on an average uniform slope and soil loss during thefinal 8 min of each run, values of the product KiCi arepresented in Table III. The mulch factor is the ratio of KiCivalues of the straw mulch to bare soil and represents theeffect of mulching.

Furrow bottom erodibility

Soil detachment by hydraulic shear along the furrow bottomdepends on side inflow characteristics, soil resistance, cropping-management, and the transport capacity of the flow.Parameter values could not be directly evaluated for theconditions of these experiments.

Sediment source

The outflow of sediment as measured at the end of the furrowrepresented the sum of soil loss on the furrow sides and in thefurrow bottom, less deposition in the furrow bottom. Thiseffect is illustrated in Fig. 6, in which a value of zero deposition in the furrow means that the furrow bottom was neitheraccumulating nor losing soil. In the mass balance, therefore,soil loss comprised through-flow from furrow side erosion.

Slight deposition or soil loss occurred in the furrow bottom for all treatments other than the high intensity rainfall onbare soil. For this treatment combination, there was no effectof slope from 3 to 5% but a dramatic increase in soil loss from5 to 7% where rill erosion started to take effect. The protection afforded by the straw mulch, therefore, dramaticallyincreased with degree of slope beyond 5% under the higherintensity rainfall.

MODELLED FIELD APPLICATION

Furrow slope (%)

Fig. 6. Deposition and soil loss from furrow bottom.

Uniform planeFor comparison purposes, COSSEM also was used to simulate runoff and soil loss for a specific example of a uniformplane of 100 m length and 7% slope for the same rainfallevent and calibrated parameters. Predicted soil loss valueswere 16 t/ha (bare) and 0.65 t/ha (straw mulched), which islow compared to 61 t/ha (bare) and 2.6 t/ha (straw mulched)for up-and down slope furrows (Fig. 7).

Comparable USLE single storm soil loss estimates for baresoil (for a USLE soil erodibility, K, value of 0.04 t-h-Mrl-mm- I )

are 12 t/ha based on rainfall (EI) erosivity (Foster et al. 1981),

Up-and-down slope furrowsThe measured data set generated in the laboratory was necessarily restricted to a maximum furrow length of 3.15 m. Togenerate data for practical field furrow lengths, the values ofthe runoff and sediment parameters optimised for COSSEM(as presented for bare soil in Burney and Edwards (1996»were used to extrapolate the furrow length for the lowerintensity event. Cover-management factors were separatelyoptimised for this study.

The furrow side (interrill) soil erodibility, Ki, and covermanagement, Cj, (Eq. I) values used for modelling were 1.70Gg-s-m-4 , and 0.21 (Table III), respectively. The furrowbottom soil erodibility parameters used were as given inBurney and Edwards (1996). The cover-management factor(Cj) for furrow bottom hydraulic shear, determined by the useof COSSEM, was optimally 0.13. This compared favourablywith a calculated value of 0.12 for the rill erosion test data ofFrame et al. (1992) on the same soil type and similar cover.

Modelled soil loss, on a conventional USLE mass per unitarea basis, is presented in Fig. 7 for (a) a bare and (b) a strawmulched furrow of up to 100 m in length for the conditionsof this study (a uniform rainfall of 47 mm/h and 20 minduration). Under field conditions these curves illustrate (a)the severity of soil loss due to a single 'cool season' stormevent on furrowed bare soil and (b) the effectiveness of strawmulch in preventing or reducing rill erosion. However, infield practice displacement of the straw cover by wind orrunoff can reduce the overall benefit.

1.540.68

0.44

94 mm/h

Rainfall intensity

0.21

1.700.37

47 mm/h

Treatment

Mulch factor

BareStraw

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I Janurary/Fcbnmry/March 1997 5

3. When straw cover is used under the circumstances ofthis laboratory experiment, net deposition in the furrowbottom is positive or only very slightly negative for allslopes and rainfall intensities, indicating that only interrill erosion occurs. Extrapolation of furrow length bymodelling indicates the limitations of this effect asslope inclination and furrow length increase.

4. On steeper slopes up-and-down hill furrowing increasessoil loss approximately four-fold. Immediately following harvesting tillage should be carried out to leave theland surface level across the slope, roughened and compacted prior to mulching.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the assistance of JackVissers, Jim Godwin and Albert Murphy in the constructionof the rainfall simulation equipment and the soil bin, andAllan MacRae and Irene Power in data and laboratoryanalyses.

REFERENCES

Burney, J.R. and L.M. Edwards. 1995. Sediment monitoringin a Bedeque Bay watershed. In Ecological Monitoringand Research in Atlantic Canada: A Focus onAgricultural Impacts in Prince Edward Island, eds. R.D.Elliot and P. Chan, 31-40. Occasional Report No.5,Atlantic Region, Environment Canada, Dartmouth, NS.

Burney, J.R. and L.M. Edwards. 1996. Modelling coolseason soil water erosion on a fine sandy loam soil inPrince Edward Island. Canadian AgriculturalEngineering 38(3): 149-156.

Chow, T.L. and H.W. Rees. 1994. Effects of potato hilling onwater runoff and soil erosion under simulated rainfall.Canadian Journal ofSoil Science 74:453-460.

Edwards, L.M. 1988. The effects of slope position andcropping sequence on soil physical properties in PrinceEdward Island. Canadian Journal of Soil Science68:763-774.

Edwards, L.M. 1991. The effect of freezing and thawing onaggregate stability and aggregate size distribution ofsome Prince Edward Island soils. Journal ofSoil Science42: 193-204.

Edwards, L.M. and J.R. Burney. 1989. The effect ofantecedent freeze-thaw frequency on runoff and soil lossfrom frozen soil with and without subsoil compaction.Canadian Journal ofSoil Science 69:799-811.

Edwards, L.M., J .R. Burney and R. DeHaan. 1995.Researching the effects of mulching on cool-period soilerosion control in Prince Edward Island, Canada. JournalofSoil and Water Conservation 59: 184-187.

Elliot, W.J., A.M. Liebenow, J.M. Laflen and K.D. Kohl.1989. A compendium of soil erodibility data from WEPPcropland soil field erodibility experiments 1987 & 88.National Soil Erosion Laboratory Report No.3. The OhioState University & US Department of Agriculture Agricultural Research Service, Columbus, OH.

1)0

100

80

80

60

60

40

40

Furrow length (m)

20

20

(a) Bare surface

(b) Straw much at 4 t/ha

,,/,,/

~%sIope ",/

/"'""'""'""'""'""'"

"'""'""","'" ~5%slope/"'" /---

// _/-

...... .--__________-- __--- ,3% slope..~.c:::.~-==:.~':::'::::::=::::::::::::::=:::::__ ...... .. ~ •. _

oL...- -' -'__-"

o

2

75

60

CU 452,If)If)

.Q

<5 30(/)

15

00

CONCLUSIONS

Based on this study it is concluded that:

1. Single storm events on bare furrowed soil during the'cool season' in Prince Edward Island can lead to severesoil losses.

2. Straw cover at the conventionally used rate of 4 t/ha ishighly effective in reducing soil loss into and from thetroughs of furrowed land surfaces.

Furrow length (m)

Fig. 7. Modelled soil loss for uniform rainfall of 47 mm/hfor 20 min of uniform application for (a) baresurface and (b) straw mulch at 4 t/ha.

and 17 t/ha based on runoff (MUSLE equation) erosivity(Williams 1975). For a USLE cropping-management (C-factor) value of 0.04 (Wischmeier and Smith 1978) predictedsoil losses under straw mulch are then 0.48 t/ha and 0.68 t/ha,respectively.

For this specific example, up-and-down slope furrowsresult in an approximate fourfold increase in soil loss overthat of a uniform plane (as also found under growing seasonconditions by Chow and Rees (1994» for both cover conditions. Transverse levelling of up-and-down slope furrowedland following harvesting therefore appears to be highlybeneficial in reducing cool season soil loss. However, themagnitude of benefit can only be evaluated through sitespecific modelling.


Foster, G.R., D.K. McCool, K.G. Renard and W.C.Moldenhauer. 1981. Conversion of the Universal SoilLoss Equation to SI metric units. Journal of Soil andWater Conservation 45:355-359.

Frame, P.A., J .R. Burney and L.M. Edwards. 1992.Laboratory measurement of freeze/thaw, compaction,residue and slope effects on rill erosion. CanadianAgricultural Engineering 34(2): 143-149.

MacDougall, J.L., C. Veer and F. Wilson. 1981. Soils ofPrince Edward Island. LRRI Contribution No. 141.Agriculture and Agri-Food Canada, Research Centre,Box 1210, Charlottetown, PEL

Meyer, L.D. and W.C. Harmon. 1985. Sediment losses fromcropland furrows of different gradients. Transactions ofthe ASAE 28(2):448-453.

Nowland, J.L. 1975. The agricultural production of soils ofthe Atlantic Provinces. Monograph No. 12. Agricultureand Agri-Food Canada, Research Branch, Ottawa, ON.

Parsons, T.S., J.R. Burney and L.M. Edwards. 1994. Fieldmeasurement of soil erodibility and cover managementfactors in Prince Edward Island using simulated rainfall.Canadian Agricultural Engineering 36(3): 127-133.

Tossell, R.W., GJ. Wall, W.T. Dickinson, R.P. Rudra andP.H. Groenevelt. 1990. The Guelph rainfall simulator II:Part 1 - Simulated rainfall characteristics. CanadianAgricultural Engineering 32(2):205-213.

Williams, J.R. 1975. Sediment-yield prediction withUniversal Equation using runoff energy factor. In Presentand Prospective Technology for Predicting SedimentYields and Sources, 244-252. ARS-S-40. USDA-ARS,Washington, DC.

Wischmeier, W.H. and D.D. Smith. 1978. PredictingRainfall Erosion Losses: A Guide to ConservationPlanning. Agriculture Handbook No. 537. USDA,Washington, DC.

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No.1 Janurary!February/March 1997 7

Improving small-scale composting of applewaste

S.F. BARRINGTON1• K. EL MOUEDDEB1 and B. PORTER2

I Macdonald Campus ofMcGill University. 21111 Lakeshore. Ste Anne de Bellevue. aC, Canada H9X 3V9; and2Basco-Tech

Inc., Hemmingford. aCt Canada JOL IHO. Received 21 December 1994,· accepted 18 November 1996.

Barrington, S.F., El Moueddeb,K. and Porter. B. 1997. Improvingsmall-scale compoStfng of apple waste. Can. Agric. Eng. 39:009016. Small-scale composting operations of interest to fruit andvegetable processors use tractor-operated compost turners. The mixing and aeration performance of such compost turners wereinvestigated by monitoring the composting of apple residues mixedwith sawdUSL The results suggested: 1) incorporating liquids intosawdust by passing it through the tmner two or three times and usingan overdose of the liquid to be absorbed; 2) mixing compounds instages and in equal proportions to obtain unifonn pH, TKN, and drymatter; 3) using a porosity of 35 to 40% and a C:N of 20 to 25 toreach temperatures of 60 °C; 4) bulking the residues with a mixtureofsawdust and straw to maintain compost structure while increasingthe available C thus reducing N losses.

Les usines de transformation de fruits et legumes s'interessentaox petits postes de compostage qui utilisent un retourneur d'andaintrame par un traeteur de ferme. Des essais furent effectues poursuivre la perfonnance de ces retoumeurs d'andains. Des restes depommes furent compostes avec de la sciure de bois et les resultatssuggerent de: 1) incorporer les elements par etapes en utilisant desquantit&S egales; 2) incorporer les liquides au compost en deux atroisetapes et en utilisant une quantit&S excessive de liquide; 3) utiliser uneporosit&S de 35 a 40% et un C:N de 20 a 25 pour atteindre destemperatures de 60 °C; 4) utiliser un melange de sciure de bois et depaille pour un compost de bonne structure, un meilleur taux de Cdisponible et moins de pertes de N.

INTRODUCTION

To process organic wastes as organic fertilizers rather thandisposing into landfills, fruit and vegetable processing plantshave demonstrated some interest in small-scale compostingoperations using the turned windrow system. The wet andacidic residues generated by the industry are not readilyaccepted by solid waste disposal sites because of the acidleachate they produce. Small-scale compost operations usingthe turned windrow system are economical to set up andoperate while the low volume processed limits the intensivelabour and large space required. The residues are compostedin small windrows exposed to exterior conditions and aremixed and aerated by means of a tractor-driven compostturner. The efficiency of such equipment has never beentested.

A project was, therefore, undertaken to evaluate the mixing and aeration performance of compost turners used bysmall-scale composting operations and to formulate recommendations for their most effective use. Thus, the projectfollowed the evolution of the properties of apple waste compost for 69 days to establish how well, with each successivepass of the compost turner:

1) sawdust was wetted with apple wastewater;

2) amendments and bulk materials were mixed together;

3) oxygen was introduced into the compost at three different levels of dry matter (d.m.).

The composting process was evaluated for 69 days by:

1) monitoring the compost temperature with a metal stemthermometer;

2) monitoring the windrow 02 with such a probe. to establish the need for turning;

3) establishing the best porosity and dry matter level required to compost apple wastes with sawdust by testingin triplicate three levels of dry matter;

4) evaluating visually the compost quality at 69-day;

5) measuring nitrogen (N) losses during composting bymonitoring the evolution of the carbon to nitrogen ratio(C:N).

LITERATURE REVIEW

Except for space, small-scale composting operations usingthe turned window system require little investment A compost turner operated by a 60 kW farm tractor is the onlyequipment required for both mixing and aerating. This ttactor-operated turner costs Can $15000 to $20 000 as opposedto at least Can$200 000 for a commercial self-propelled unit(Diaz et al. 1993). As opposed to the commercial unit whichturns 6.0 m wide windrows, the tractor operated turner straddles 3.0 m wide windrows which triples the land baserequired for the same composting capacity. Nevertheless, therequired land base is limited because of the small volumeprocessed.

The turned windrow system also requires more energy andlabour than the static system which aerates the compost byventilating the mass with a fan and a duct system. The turnedwindrow system incorporates air into the compost by regularly turning and mixing the entire mass, thus using muchenergy and time and often leading to over-mixing and coolingof the compost. Again, this additional use of labour andenergy is limited by the small volume processed by thetractor-driven compost turner.

Sawdust or peat as bulking agent can minimize the frequency of windrow overturning by giving the compost amore porous structure. The pore space of the compost helps

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No. I, JanwuylFebroary/March 1997 9

Table I: Chemical properties of the experimental material

1 All properties are expressed on a d.m. basis.2 Bxpressed on a eLm. basis and calculated by dividing the organic

matter content by 1.83. The organic matter content is equalto one minus the ash content.

3 The values in parentheses are the standard deviations.

Because the apple residues were acid and the sawdust hada high C:N of 720, dehydrated lime (Ca(OH)2) and urea(46%) were used to correctthe pH to 7.0 and the C:N to 30.A C:N of 30 instead of20 was selected to reduce N losses andminimize amendment costs while still respecting the recommended ratio. The required quantity of lime was establishedin the laboratory by measuring the buffer capacity of theapple residues.

Apple pulp and apple processing wastewater as primarycomposting materials offered excellent testing conditionsfor.

1) (mixing as their acidity requires correcting with limitedquantities of limestone;

2) aeration as their high humidity reduces the porosity ofthe mass and increases the need for aeration.

The experimental equipment

The test compost turner (Model 1012, Sittler Inc., Linwood,ON) consisted of a 3.66 m-Iong rotating drum with knives

Material

Apple pulp Apple Sawdustwastewater

15.3 1.44 64.6(1.77)3 (0.058) (0.040)

3.3 4.0 5.9(0.14) (0.01) (0.01)

3.7 20.0 0.6(0.50) (2.36) (0.04)

53.5 44.4 55.2

27.3 0.97 7.65(1.502) (0.025) (0.103)

5.0 17.3 0.9(0.01) (1.83) (0.03)

1571 3700 256(767.6) (236.2) (113.7)

340 870 171(52.6) (109.3) (56.4)

20 460 71J)C:N

K total (mglkg)

P total ( mglkg)

NH4-N (mglkg)

TKN(glkg)

Properties

pH

d.m. (%)

METHODOLOGY

The experimental material

Apple pulp and apple wastewater were composted with sawdust used as a bulking agent (Tables I and m. The appleresidues were produced by an apple sauce and pie fillingmanufacturer in Franklin, QC. The sawdust was purchasedfrom a local saw mill cutting mainly maple, hemlock, and oak.

store and diffuse oxygen to the microbes (Diaz et ale 1993)and is influenced by the structural strength of the bulkingagent Wood shavings, sawdus~ and peat moss offer goodstructural strength even at humidity levels of 75% whilesttaw and paper residues tend to collapse when wet (Zhao etale 1992; Mathur et ale 1990). Thus, compost containingsawdust should require less overturning to maintain an oxygen level above 5% within the compost (Midwest PlanService 1985).

Sawdust as bulking agent leads to high N losses becauseof its high lignin content and low C bio-degradability. Whilea compost C:N of20 to 35 is preferred (Rodrigues et ale 1995;Martin et ale 1993), soil microbes possess a C:N of approximately 8 (Alexander 1977) and lo~e 60% of the C consumedas CO2 (Henis 1986). Thus, no N will be lost with a compostpossessing a C bia-degradability of 100% and a C:N of 20because after loosing 60% of the C as C02, the final C:N of8 corresponds to the biomass requirements of the microbes.A compost with a C:N of 30 will conserve all of its N only ifat least 66% of its C is bio-degradable (Barrington 1994).When composed mostly of lignin which has a C bio-degradability of 7% (Russell 1973), sawdust compost with an initialC:N of 30, may offer a C bia-degradability of 15% to 25%,and thus, lose 80 to 60% of its N, respectively. A C:N of 20to 30 is still required early in the composting process otherwise slower composting activity may be expected (Diaz et ale1993). Thus, sawdust as a bulking agent inevitably leads tohigh N losses despite its better structure and porosity reducing the need for turning.

Because composting is a conttolled aerobic decompositionprocess converting biodegradable solid organic matter intostable humus (Parent 1983), the perfonnance and timely useof the compost turner is critical. The following criteria canestablish the composting performance of the system:

1) temperatures of 60 to 65°C should be reached within 3to 7 days to efficiently stabilize the residues (Diaz et ale1993; Lau et ale 1991);

2) the final product should be dark in colour, odourless andwell decomposed, except for the wood residues(Mustin 1987; Lo et al. 1993);

3) the compost should have conserved the highest level ofnuments (Lo et ale 1993);

4) the compost 02 level should be above 5% at all times(Midwest Plan Service 1985).

When such criteria are not achieved, the composting processand the physical and chemical composition of the compostmixture requires correction. These criteria were used toevaluate the mixing and aeration efficiency of compost turners and to recommend guidelines for their most effective use.

10 BARRINGlON. MOUEDDEB and PORTER

• the values in parenthesis are the standard deviations.

Table D: Physical properties or the experimental materials

held at 300 to 600 mm above ground level and powered bythe power-take-off of a 60 kW fann tractor. The drum isrotated while being moved through the compost and heavyrubber mats on both sides of the housing direct the thrownmaterial to shape a windrow while protecting the tractor anddriver.

The temperature of the compost was measured by insertinga long metal-stem compost thermometer to a depth of 800mm within the windrows. The oxygen level of the windrowswas measured by pumping the compost's air against an oxygen sensor (Sensitron Inc., Readings, PA) by means of a rigidPVC (polyvinyl chloride) tube, perforated at its tip and inserted into the windrows to a depth of 800 mm. This PVCtubing could be unscrewed from the sensor for cleaning whennecessary.

The methodologyEighteen exterior windrows were built to test in triplicates forthree sawdust to apple pulp ratios and three sawdust to applewastewater ratios (Table ill). Three levels of dry matter(d.m.), and thus porosity, were tested for each residue. Thewindrows measured 3.0 m in width and 1.5 m in height andwere built outside in unprotected straight rows spaced 5.0 mapart. The windrows were built by dumping 3 wet tonnes ofsawdust in a row and topping it with the apple residues. ThepH and C:N of the compost were corrected by manuallyspreading over the crest of the windrows the required amountof lime and urea. All the ingredients then were mixed byoverturning the piles twice with the compost turner.

The wastewater was applied to the sawdust using a technique adopted by most small-scale composting operations.The sawdust windrows were manually trenched at their crestto receive the wastewater from a hose attached to the compostturner in operation. Then, the lime and urea were manuallyspread over the windrows and mixed using the compostturner. All sawdust and wastewater windrows were sampled

Sawdust

1.05(0.112)

0.68(0.021)

7.7 (0.26)15.0 (1.61)60.3 (1.38)15.4 (1.12)

1.4 (0.13)0.2 (0.09)

immediately after incorporation to establishthe level of liquid absorption.

The windrows were composted fromSeptember 2S to December 3, 1992. Thetemperature and oxygen level of all windrows were measured early in the morning,every four days from day 0 to 32 and oncea week thereafter. The windrows wereturned once every 7 to 10days. From day 20to 30, the frequency was changed to once aday to measure the effect ofadditional overturning on windrow temperature andoxygen level.

The mixing performance of the compostturner was measured by sampling all windrows on days 3, II, 41, and 69. Thevariation in pH, TKN, and dry matteramong four samples taken within the samewindrow served as an indication of mixingefficiency for the lime, urea, and apple resi-dues. The level of wetting of the sawdustwith the wastewater was observed from thevariation in dry matter of samples takenimmediately after mixing the windrows.

The best dry matter level was selected as that from thecompost demonstrating temperatures closest to 60 - 6SoC,developing a dark colour, having no odour, and maintainingthe highest level of nutrients (Lo et ale 1993). The effectiveness of the 02 meter in predicting the aeration needs of thecompost was tested by comparing 02 level in the windrowswith its fmal visual qualities. The level of N conservationduring composting was evaluated by comparing the initialand final C:N. The level of C bio-degradability of the sawdust and apple pulp was estimated by assuming that 60% ofthe bio-degradable C was lost as CO2 and an average biomassC:N biomass requirement of 8.

Sampling and analytical procedures

The apple residues and the sawdust were randomly sampledfour times from piles stored outside. The apple wastewatersamples were collected after agitating the contents of theplant storage tank. These samples were analyzed for particledensity and size distribution, dry matter, TKN, NH4-N, N03N, pH, and ash. The ash content was used to calculate organicmatter as (1- ash, expressed as a fraction) and C as (organicmatter 11.83) according to Lo et al. (1993) and ]im6nez andGarcia (1992).

Standard methods were used to analyze the samples(APHA 1990). Dry matter was determined by drying at 80°Cfor 23 hours and one hour at 103°C. Particle density wasdetermined on oven dried samples by soaking in kerosene(parent and Caron 1993). Particle size was determined bysieving dried samples using standard sieves with openingsizes ranging from 7S m to 4.76 mm. TKN was determinedafter digestion with sulphuric acid and hydrogen peroxideusing an ammonia-selective electrode connected to a voltagemeter. Ammonium and pH were measured using an ammonia-selective and pH electrodes, respectively, connected to avoltage meter.

Material

1.00(0.002)

Applewastewater

1.00(0.002)

Applepulp

0.83(0.086)*

22.6 (7.00)29.5 (12.34)

9.6 (0.42)3.9 (0.28)0.9 (0.54)

33.5 (6.54)

Particle size (%)>4.76rnm4.76 - 236 rnm2.36 - 0.42 mm0.42 - 0.25 mm0.25 - 0.075 rnm<0.75rnm

Particle density (kg/m)

Properties

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No. I, JanwuyIFebroarylMarch 1997 11

Compost properties

Performance 01 the compost turner

Mixing The Cv (Eq. 2) for the compost pH, TKN, and drymatter (Figs. 1,2, and 3) served as an indication of the mixingperfonnance of the compost turner with respect to the lime,urea, and apple residues. Initially, the Cv was quite high butdecreased with time to reach its lowest value at 41 days, after15 passes of the compost turner.

After two passes with the compost turner (day 3), the limeand urea were not effectively mixed as the Cv of the pH andTKN was 10 times that on day 41. The compost turner wasbetter able to mix the residues and the sawdust since the drymatter Cv was only 2 to 3 times that on day 3 than on day 41.

The major factors influencing mixing uniformity are (Larson 1978; Weeden and Norrish 1981):

1) Compound density and moisture, where compoundsdenser than 400 kg/m3 mix easier because of higher

454030

301812

O.M.

343436

303030

C:N

values obtained were 58.9,46.5, and 40.1%. Since thewindrows absorbed 1.20,3.20, and 9.08 m3 of wastewater when 4.85, 9.7, and16.5 m3 was applied, thelevel of wastewater absorption for treatments W-1,W-2, and W-3 was 2S, 33,and 55%, respectively. Toimprove sawdust wetting,the wastewater should beapplied in quantities exceeding that required andrepeating theprocedure withthe collected runoff mtil thedesired dry matter isreached. Adding large volumes of liquid improves thelevel of absorption. Treat-

ments W-2 and W-3 were designed to exceed the sawdustabsorption capacity of 3 times its weight as most bulkingagents reach their limit at a dry matter of 25% (Midwest PlanService 1983).

The porosity of the compost was measured by excavatingthe windrow to a depth of 600 to 800 mm and taking a corevolume. This procedure was repeated three times for eachwindrow. The mass of each volume was weighed wet andafter oven drying. The particle density of the compost wasmeasured and the porosity was calculated by:

P=(Vp+Vw)/V, (1)

where:Vp =Mp/Dp =volume of oven dried compost particle in

the core sample (m3), .Vw =volume of water in the core sample (m~,V, =total volume of the core sample (m3),Mp =oven dried mass ofcompost particles in the .

core sample (kg), andDp = density of the compost particles (kg/m3).

The compost pile samples were analyzed for Particle density and size distribution, dry matter, TKN, NH4-N, N03-N,pH, and ash.

The mixing perfonnance of the compost turner was calculated from the variation in compost pH, TKN, and dry matter(Larson 1978):

Table m: The experimental treatments

Treabnent Compost composition

Residue Lime Urea~ kg/t kg/t

Apple pulpA-I 0.75 3.00 36A-2 1.50 6.10 38A-3 3.75 15.4 43

Apple wastewaterW-l 2.50 0.50 35W-2 5.00 0.60 35W-3 8.50 1.0 35

• element per ton of sawdust on a d.m. basis


Sawdust absorption 01 wastewater

The initial dry matter of the wastewater windrows was higherthan that calculated (Table Dn because the slow absorptionrate of the sawdust produced ground runoff during mixing.Treaunents W-1, W-2, and W-3 should have possessed initialdry mauer of 30, 18, and 12%, respectively, whereas the

4111

100Treatment

-A-180

+A-2

"'A-3;;i' 60

"'W-1~

> *W-2(.)

~ 40 +W-3

20

Time (d)

Fig. 1. Coefficient 01 variation 01 the compost pH withtime.

(2)

=coefficient of variation (%),=standard deviation of the element within the

samples, based on the normal curve,=mean value of the concentration of the element

among the samples.x

Cv =l00s Ix

where:Cvs

12 BARRINGTON, MOUEDDEB and PORTER

00 8 16 24 32 40 48 56

Time (d)6911 41

Time (d)

100Treatment

20-- A-l

80 +A-2 Treatment

*A-3 -15 -- A-l- ~

C60 ~"'W-1 Q) +A-2

> *W-2 >0 .!! 10 "A-3Z +W-3 c~40 Q) "'W-1... C)

~ *W-20 5 +W-3

Fig. 2. Coefficient or variation or the compost TKN withtime.

Fig. 4. Oxygen concentration level in the compost withtime.

30

1) a lack of bio-degradable C, in the case of the wastewatertreatments;

2) excess cooling because the windrow size reduced aftersubsidence to a height of 1.0 m and a width of2.0 m andbecause night temperatures below OoC left frost on thewindrow surfaces by morning;

Aeration At the onset of composting, turning once a weekwas insufficient in maintaining 02 levels above 5% for alltreatments with less than 34% d.m. or 50% porosity (Fig. 4,Tables IV and V). For treatments A-2 and A-3 with 27.8 and22.7% d.m. or 30.8 and 20.2% porosity, respectively, 02levels fell below 5% and temperatures did not exceed 4SOC(Fig. 5). Turning once a day, from day 20 to day 30 in A-2and A-3, increased02 levels and temperatures by I to 2% and10 to 20°C, respectively.

For treatment A-I with 34.0% d.m. and 50.6% porosity,turning once a week maintained 02 levels above 5% at alltimes while turning once a day had the negative effect ofdecreasing its temperature either because of:

1) the cooling effect of frequent turning, since the exteriortemperatures were low and the windrows had subsidedto a small size, or;

2) less active microbe activity after an earlier, more activecomposting period (higher 02 levels and temperaturesthan the other treatments for day 0 to 20) leading to thedepletion of bio-degradable C or available N.

For treatments W-I, W-2, and W-3, with a dry matterexceeding 35% and a porosity of60%, turning frequency hadno marked effect on 02 levels which stayed above 15%.

The oxygen probe was effective in indicating a lack ofcompost 02. Also with treatments A-2 and A-3, turning oncea day was insufficient in maintaining the necessary 02 levelsand forced aeration should have been used.

The composting process

Temperature evolution and rmal visual quaDty No treatment reached the desired temperatures of 60 to 6SOC (Fig. 5)because of:

69

Treatment

-- A-1

+A-2

*A-3

"'W-1

*W-2

+W-3

4111

5

o'----------'-----~----'3

25

Time (d)

Fig. 3. Coefficient orvariation or the compost dry matterwith time.

frictional and shear forces leading to more particle interaction;

2) The mass proportions, where the probability of findingan ingredient in a sample is related to its presence withinthe mix. Proper mixing is more likely when incorporating two compounds of equal weight;

3) Particle dimension, where particles of smaller sizessegregate by moving and falling through the larger onesin the mix;

4) The mixing time, where a long mixing time segregatesthe particles, while a short time leads to non-uniformmixing.

The blending effectiveness of the compost turner can,therefore, be improved by mixing the compounds in stages,depending upon their mass. Mixing of the bulk materialsshould be carried out at near equal proportions. For example,3.75 t of apple pulp can be composted with 1.0 t of sawdustby initially mixing 1.5 t of apple pulp to 1.0 t of sawdust andthen, 2.0 t ofapple pulp to the 2.5 t mixture. The amendmentsshould be added to 100 kg of residue and then incorporatedto the rest of the residues before mixing with the bulkingagenL

CANADIAN AGRICULTURAL ENGINEERING Vol 39. No.1. JanuarylFebrumylMarch 1997 13

Table IV: Final chemical properties of the compost

ProPerties1Treatment

A-I A-2 A-3 W-1 W-2 W-3

d.m. (%) 34.0 27.8 12.7 38.4 36.8 3S.7(0.3)3 (0.3) (1.2) -(0.6) (1.4) (0.8)

pH 7.9 8.2 7.4 8.2 8.5 8.7(0.03) (0.06) (0.08) (0.11) (0.1) (0.03)

TKN(g/kg) 4.7 7.2 10.2 4.2 4.4 4.1(0.3S) (0.40) (0.42) (0.39) (0.23) (0.22)

P (mg/kg) 471 1294 lS64 284 282 280(IS) (142) (274) (22) (S2) (37)

K(mg/kg) 121 236 344 91 77 84(S) (S) (37) (6) (29) (33)

Ash(%) 4.9 4.6 20 1.2 1.2 I.S(O.OS) (0.3) (O.S) (0.3) (0.3) (0.1)

C:N2 S3 74 117 130 126 133.

1 All properties are expressed on a d.m. basis.2 The C is ~xpressed on a d.m. basis and is calculated by dividing the organic matter content by 1.83. The organic matter content is equal

to one mmus the ash content.3 The values in parenthesis are the standard deviations.

besides the sawdust TreabDent A-3 was too wet to decompose properly. Among other factors, 02 levels below 5% fortteabDent A-2 had the drawback of preventing temperaturesfro~ reaching to 60°C. Considering the performance of A-2,the Ideal sawdust and apple pulp compost mixture shouldgive a porosity between 30% and 50%, probably in the rangeof 35 to 40%, and a C:N lower than 30, probably between 20to 25.

Nitrogen losses All treabnents demonstrated a high N lossas the final C:N ranged between 52 and 133 (Fig. 6). Theinitial C:N varied from that corrected with urea to 30 becauseof poor mixing.

From day 11, N losses were obvious as all windrowssmelled of ammonia indicating that the mea was being degraded into ammonia, which in tum, was being volatilized.Only treabnents A-3 and W-l demonstrated little ammoniasmell during the experiment because, for A-3, a high moisture content helped dissolved the ammonia and createdanaerobic conditions which slowed the degradation of ureaand, for W-1, a low moisture content slowed the microbialactivity degrading urea (Alexander 1977).

The fmal C:N of52 to 133 indicated N losses ranging from40 to 80%, respectively. The final C:N would have beenlower if the bulking agent had offered more bio-degradableC. The fmal C:N (126 to 133) in the apple wastewater treatments indicated that the sawdust had a C biodegradability of10 to 15% while that of the apple pulp treatments (52 to 117)indicated that the apple pulp had a C biodegradability of 30

60

50Treatment

-40 -A-1e! 30

-+- A·2

:::J "'A-31U.. --W·1~ 20E *"W·2

~ 10 -+- W·3

0"'Min.

+ Max.

-10 ~--'-_....l....--------''-------'-_--'-----'_-J

o 8 16 24 32 40 48 56Time (d)

Fig. S. Temperature of the compost with time.

3) depletion ofbio-degradable C and N where a C:N of 20would probably have improved microbial activity, despite N losses.

Despite low 02 levels from day 8 to 40, treabDent A-2demonstrated the best final compost properties after 69 daysbased on its lack of odour, darlc brown colour, and no longerdistinguishable apple residues. Treatments W-l, W-2, andW-3 produCed a mass of sawdust with little colour changebecause of the low dry matter of the wastewater. TreabDentA-1 developed a light brown colour but lacked structure

14 BARRINGTON. MOUEDDEB and PORTER

Table V: Final physical properties of the compost

Property

A-I A-2

Wet density- bulk (kg/m3) 0.58 0.74

(0.04)1 (0.02)

- particle (kg/m3) 1.05 0.98(0.05) (0.04)

Compost porosity (%) 50.6 30.8

Particle size distribution (%)>4.76mm 34.7 19.2

(3.2) (2.1)

2.36-4.76 mm 29.9 41.1(8.2) (5.9)

0.84-236mm 17.8 17.0(6.8) (5.9)

0.25-0.84 mm 10.1 9.0(2.1) (1.8)

0.075-0.25 mm 0.4 0.4(0.2) (0.3)

<0.075mm 7.1 13.3(13) (3.0)

Treatment

A-3 W-1 W-2 W-3

0.84 0.46 0.48 0.49(0.03) (0.01) (0.01) (0.01)

1.02 1.11 1.10 1.09(0.03) (0.04) (0.06) (0.04)

20.2 66.4 64.9 65.7

17.0 8.6 10.6 13.5(2.0) (1.5) (1.2) (0.7)

425 42.6 50.7 47.5(1.4) (7.0) (7.3) (6.7)

28.2 40.7 31.1 30.4(3.5) (2.8) (8.5) (7.6)

3.7 1.6 4.2 4.2(0.6) (1.2) (0.6) (0.9)

0.5 6.5 3.4 4.4(0.3) (1.3) (2.4) (21)

8.1 0 0 0(2.6)

1 'The values in parenthesis are the standard deviations.

to 35%. The use of straw instead ofsawdust would have fixedmore N and produced a lower final C:N at the expense of lessair exchange.

Time (d)

Fig. 6. C:N of the compost with time.

Treatment

-A-1 +A-2 *A-3 "'W-1 *W-2 +W-3 CONCLUSIONS

Small-scale composting operations use a tractor pulled compost turner to mix and aerate exterior windrows. The projectdemonstrated that incorporating wastewater to sawdust required the application of liquid two to three times whileturning the windrows. The results of this study also indicatethat compost turners will be most effective if:

1) the mixing operations are carried out in stages, especially when adding small quantities of amendments;

2) the 02 levels of the windrows can be monitored with aprobe and the frequency of turning can be increased tomaintain an 02 level above 5%.

The compost could have reached temperatures of 60 to65°C with better aeration by:

1) using an initial compost porosity and dry matlel' of3S to40% and 28 to 34%, respectively;

2) an initial C:N of 20 to 25 to maximize microbial activity;3) the compost had been turned once a day for the fIrSt two

weeks and then once a week thereafter.

694111

40

160

120

~ 80(,)

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No.1. JanuarylFeblWUYlMardl 1997 15

Dryma~and porosity were found to be just as imponant~ overtummg frequency in maintaining proper 02 levels.Finally, sawdust as bulking agent was found to help maintaina .good compost structure through the process but to lead tohigh N losses. A combination of straw and sawdust should beinvestigated.

ACKNOWLEDGEMENTThe authors acknowledge the financial contribution ofLeahyOrchards Inc., Franklin, QC, the National Research Councilof Canada, and the Natural Science and Engineering Research Council of Canada.

REFERENCESAlexander, M. 1977. Soil Microbiology. New York, NY:

John Wiley and Sons.

APHA. 1990. Standard Analysis for Water and Wastewater.Washington, DC: American Public Health Association,American Waster Works Association, and WaterPollution Control Federation.

Barrington, S.F. 1994. La gestion des engrais de ferme. Courssur les programmes fertilisation int6groo. LaPocati~re,

QC: Institut Agricole de LaPocati~re.

Diaz, L.F., G.M. Savage, LL. Eggerth and C.G. Golueke.1993. Recycling Municipal Solid Waste. Boca Raton, FL:Lewis Publishers.

Henis, E. 1986. Soil microorganisms, soil organic matter andsoil fertility. In The Role of Organic Matter in ModernAgriculture, eds. Y. Chen and Y. Avnimelch. Boston,MA: Martinus Nijhoff Publishers.

Jimtnez, E.I. and V.P. Garcia. 1992. Relationship betweenorganic carbon and total organic matter in municipal solidwastes and city refuse compost. Bioresource Technology41:265-272.

Larson, F.D. 1978. Mixing fundamentals. In InternationalFeed Milling Production and Grain Processing Seminar.Winnipeg, Manitoba: Underwood McLellan Ltd.

Lau, A.K., K.V. Lo, P.H. Liau and J.C. Yu. 1992. Aemtionexperiments for swine manure composting. Bioresource

16

Technology 41:145-152.

Lo, K.V., A.K. Lau and P.H. Liao. 1993. Composting ofseparated solid swine manures. Journal of AgriculturalEngineering Research 54:307-317.

Mathur,S'p., N.K. PaUli and M.P. Uvesque. 1990. Staticpile, passive aeration composting of manure slurriesusing peat as a bulking agent. Biological Wastes34:323-333.

Martin A.M., J. Evans, D. Porter and TR. Patel. 1993.Comparative effects of peat and sawdust employed asbulking agents in composting. Bioresource Technology44:65-69.

Midwest Plan Service. 1985. Composting Organic Waste.Ames, IA: Iowa University Press.

Mustin, M. 1987. Le compostage:gestion de la· mati~reorganique. Paris, France: Les Editions Fran~ois Dubusc.

Parent, B.C. 1983. Composting technical and economicalaspects. In Disinfection of Sewage Sludge: Technical,Economical and Microbial Aspects,139-149. Brussels,Belgium: Elsevier Scientific Publishers.

Parent, L.E. and J. Caron. 1993. Physical properties oforganic soils. In Soil Sampling and Methods ofAnalysis,ed. M.R. Carter, 441-459. Boca Raton, FL: LewisPublishers Inc.

Rodrigues, A.M., LJ. Ferreira, A.L. Fernando, P. Urbano andJ.S. Oliveira. 1995. Co-composting of sweet sorghumbiomass with different nitrogen sources. BioresourceTechnology 54:21-27.

Russell, W.E. 1973. Soil Conditions and Plant growth.London, England: Longmans, Green and Co. Ltd.

Weeden, J.K. and J.G. Norrish. 1981. Stationaryblending/grinding - A field evaluation. CSAE Paper No.81-101. Saskatoon, SK: CSAE.

Zhan, W., L. Fernandes and N. Pabli.1992. Composting ofpoultry manure slurries. CSAE Paper No. 92-515.Saskatoon, SK: CSAE.

BARRINGTON. MOUEDDEB and PORTER

Performance of different binders duringalfalfa pelleting

L. G. TABIL, Jr. l, S. SOKHANSANJ 1 and R.T. TYLER2

I Departmellf of Agricultural and Bioresource Engineering, University of Saskatchewan, 57 Campus Drive. Saskatoon, SK,Canada S7N 5A9; and 2Departmellf of Applied Microbiology and Food Science, University of Saskatchewan, 51 CampusDrive, Saskatoon, SK, Canada S7N 5AB. Received 21 March 1996; accepted 17 February 1997.

Tabil, L.G., Jr., Sokhansanj, S. and Tyler R.T. 1997. Performanceof different binders during alfalfa pelleting. Can. Agric. Eng.39:017-023. Production of good quality dehydrated alfalfa pellets isessential in the reduction of dust and fines generated during tnmsponand handling. Durable pellets can be produced by controlling theproduction process and by using binders. The objective of this experimental work was to determine the durability and hardness ofdehydrated alfalfa pellets with the use of binders. The study wasconducted in two crop years. In the first crop year (1993) the factorsused in the experiment were chop quality and binder used. Alfalfachops used were of three qualities (low, medium. and high) based ondurability of pellets without binders. Five binders were mixed withground alfalfa: lignosulfonate, bentonite, pea starch, collagen protein, and hydrated lime. Results indicated that all binders improvedthe durability of pellets made from low quality chop but did notimprove the durability of pellets made from medium or high qualityalfalfa. Hardness of pellets increased with the use of binders. Hydrated lime and pea starch were funher tested in the second crop year(1994) using inclusion levels of 1.0 and 0.5%. An inclusion of 0.5%of either hydrated lime or pea starch was sufficient to increase pelletdurability.

II est essentiel de produire des granules de luzerne deshydratcesde bonne qualite si on veut reduire' les quantites de poussieres et departicules generees par la manutention et Ie transport. On peul augmenter la durabilite des granules en controlant Ie procedc deproduction et en utilisant des liants. L'objectif de ce travail experimental etait de determiner I' impact des liants sur la durabilite et ladurete des granules de luzerne deshydratees. Cettc etude s'est etcndue sur deux saisons de production. Lors de la premiere annee, lesfacteurs considcrcs ont ctc la qualite de la mouture de luzerne et Ietype de liant utilise. Trois qualites de mouture. etablies a panir de ladurabilite des granules sans liant, ont ete utilisees (faible. moyenneet haute qualitc). Cinq liants ont ctc mclanges avec la luzerne hachee:lignosulfonate, bentonite, amidon de pois, protcine de collagene etchaux cteinte. Les resultats ont montre que tous les liants amelioraient la durabilite des granules fabriquees avec la mouture deluzerne de faible qualite, mais n'affectaient pas la durabilitc desgranules faites a partir des moutures de qualites moyenne et elevee.Les liants ont permis d'augmenter la durete des granules. Au coursde la deuxieme annee, on a poursuivi les tests avec la chaux eteinteet I'amidon de pois, en les mclangeant ala luzerne a des taux de 1.0et 0.5%. Un taux de melange de 0.5% de chaux eteinte ou d'amidonde pois est suffisant pour augmenter la durabilitc des granules.

INTRODUCTION

Production of fines and dust from pellets during transport andhandling is a major problem in the export of alfalfa products.Proper design of the handling and transport system and minimizing the number of times pellets are handled help reduce

the problem. Producing more durable pellets is another approach to minimizing the dust problem. Durability can beenhanced by controlling the production process and by adding binders. Feed millers and compounders use binders toproduce feed pellets that do not crumble upon handling.Fluctuations in the quality of ingredients can have adverseeffects on the physical quality of pellets; this can be correctedwith the use of binders.

Review of the binding mechanism

The mechanism of binding during pelleting is made possibleby natural adhesion between particles and the mechanicalload which forces inter-particle contact. Rumpf (1962) andSastry and Fuerstenau (1973) described several binding phenomena such as solid bridges, capillary pressure, viscousadhesion, inter-particle attraction forces, and mechanical interlocking.

Solid bridges develop at elevated temperatures and pressures from chemical reaction, crystallization of dissolvedsubstances, hardening of binder, and solidification of meltedsubstances (York and Pilpel 1972, 1973). Capillary pressureand interfacial forces develop in pellets in the presence of aliquid, e.g. water, thereby promoting binding of particles.Viscous binders and thin adsorption layers result in immobileliquid bridges (Rumpf 1962; Ghebre-Sellassie 1989). Thinadsorption layers are immobile and can contribute to thebinding of fine particles under certain circumstances (Pietsch1984; Rumpf 1962). High binding forces are produced whenareas of contact increase during compression of solid particles. The attraction force between two solid particles iscaused by van der Waal's, electrostatic, or magnetic forces.However, their effectiveness diminishes when particle size orinter-particle distance increases. Form-closed bonds or interlocking occur in fibers and flattened or bulky particles whereparticles interlock or fold about each other causing the bonding (Pietsch 1984).

Binders are widely used in pelleting animal feeds. Severalcommercial alfalfa dehydration plants have tested and sometimes used binders. However, there is little scientificdocumentation as to their effectiveness. Alfalfa processorshave been improving the control of process variables duringpelleting of alfalfa, yet production of high quality pellets hasnot been consistent. Thus, hay quality is an important factorto be considered in pelleting. Different lots of alfalfa hay maydiffer from one another when judged by the durability of

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I. January/February/March 1997 17

pellets they produce. The use of binders may minimize theinconsistencies in the durability of the product.

Objective

The objective of this study was to determine how alfalfapellet durability and hardness could be improved with the useof binders. The energy consumption during pelleting wasmonitored to determine pelleting efficiency.

Binders

There is a variety of binders available on the market. Table Ilists five groups of binders used in this study.

Lignosulfonate binders are the most commonly used binderin pelleting animal feeds. These are sulfonate salts made fromlignin of sulfite pulp-mill liquors. Lignosulfonates come indifferent brand names with different claims, inclusion level,and cost. Lignosulfonates have been regarded to be consistently the most effective and popular binder for all types ofcompounded animal feeds (Anonymous 1983a; MacMahon1984). They are generally effective as a binding agent whenused at inclusion levels of 1 to 3% (Anonymous 1983b). Thelignosulfonate binder used in the experiments was Maxibond™ (AGResearch, Inc., Joliet, IL). It is a free flowingwhite powder which forms a colloidal suspension in water.The two main ingredients of this binder were urea formaldehyde resin and calcium sulfate.

Bentonite is the traditional binder used in iron ore and feedpelleting. It is a colloidal clay (aluminum silicate) composedchiefly of montmorillonite. This binder forms a gel withwater. Pfost and Young (1973) reported that bentonite, at aninclusion rate of 100 kg/t of feed mash, improved pelletdurability of three poultry formulas consisting mainly ofground yellow corn, ground sorghum grain, and soy meal.The bentonite (leN Biomedicals, Inc., Aurora, OH) used inthis study was sodium bentonite which has a very high swelling capacity (according to the product's technical literature).

Starches are extensively used in the food industry as thickeners or binders. Pre-cooked starch reportedly did well as abinder for animal feeds (Wood 1987) and pelleted iron ore(Haas et al. 1989). In this study, Starlite® (Parrheim Foods,Inc., Saskatoon, SK) pea starch, a product of dry processingof yellow peas (Pisum sativum, L.) was used. It contained82.1 % starch and 5.3% protein (from analysis at Grain Chemistry Laboratory, Crop Science and Plant EcologyDepartment, University of Saskatchewan).

Proteins also are potential binders. Alfalfa has high proteinwhich may help in the binding of particles. Natural protein willplasticize under heat, even from the frictional heat produced asthe material passes through the die, making good quality pellets(MacBain 1966). The protein used in this study was Agricolloid® (Swift Adhesives, Winnipeg, MB). It is a refinedanimal protein which is 94.8% digestible collagen.

Table I: Binders used in pelleting of alfalfa

Propertiesand other infonnation

Binder

Collagen protein Hydrated lime Lignosulfonate Bentonite Pea starch

Trade name AGRI-COLLOID® Maxi-bond™ Starlite®

Chemicalfonnula

rH0 R 1N,~/e'N~'e

RHOn

R • amino lIcld

ClI(OHh-HOH

-c:--e-

Ci~~OHo

R • H. CIIrbohydrata orlInothllr lignin unhe.H...0u(OCH~0.11 (SO,H)o.A

7.7±O.5

688±19

1237±11

29.1

89.51±O.15

-0.64±O.35

1O.0I±O.17

82.1 % starch5.3% protein1% fat1% ashParrheim Foods Inc.Saskatoon, SK

5.2±0.2

487±ll

I792±60

6.0

ICN BiomedicalsAurora,OH

75.44±O.35

-0.0I±O.10

6. 18±0.05

high swellingcapacity in water(l2x)

6.7±0.5

695±11

1291±44

37.6

78.43±O.12

1.77±0.07

8.18±6.15

1.7±0.5

361±20

1626±61

6.9

15% nitrogen7% calcium0.18% free

fonnaldehydeContinental Lime Ltd. AGResearch Inc.Exshaw, AB Joliet, IL

94.84±O.27

-0.42±O.17

1.74±O.42

94.5% Ca(OHh

7.6±0.5

446±7

1066±29

43.2

80.33±O.27

0.69±O.02

17.45±O.70

94.8% collagenprotein

Swift AdhesivesWinnipeg, MB

Moisture (% wb; n=3)

Bulk density (kg/m3) (n=6)

Particle density (kg/m3) (n=5)

Particle size (~m)

Color (n=3)Hunter L

Hunter a

Hunter b

Other properties

Source

18 TABIL. SOKHANSANJ and TYLER

---------------- -------------~---------------

------------------------ --- - -------

Table II: Binder inclusion levels in experiments for the 1993 crop and 1994 crop

Table III: Settings used during pelleting in theCPM CL-S pellet mill

Setting

235-250

134-136

250

6.144.8

7.31

>9215

Variable

1.00.5

1.0

0.50.50.5

Inclusion level(%)

Steam pressure (kPa)Steam temperature (oC)

Die speed (rpm)Die diameter (d) (mm)

Die thickness (length) (I) (mm)

Die lId ratioConditioner temperature (oC)

Pellet knife position, from face of die (mm)

Pelleting

Pelleting of the sample was done in a CPM CL-5 laboratorypellet mill (California Pellet Mill Co., Crawfordsville, IN)using the pellet mill settings listed in Table III. The steamfrom the supply line with a pressure of 1030 kPa was reducedto 235 - 250 kPa using a pressure reduction valve. Heat wasalso supplied by the heater strips attached to the walls of theconditioning chambers and by a blower-heater. Pelleting ofthe 1994 crop was conducted in the modified pellet mill(CPM CL-5) with double conditioning chambers. The retention time of alfalfa grinds at the conditioner was from 17 to20 s. Hot pellets, which were thinly spread on a polyethylenesheet, were cooled by a blast of room air (21°C) from a deskfan. Duration of cooling was 1 h.

To determine the energy consumption during pelleting, awattmeter was installed in the feed line to the motor of thepellet mill. The specific energy consumption was expressedas kW·h of energy consumed per tonne of pellets. Output ratewas measured by weighing the container with pellets collected for 90 s.

Physical quality tests

Procedures for testing the physical quality of the alfalfa grindand pellets were as follows. Moisture of the grind and the hotand cooled pellets were determined using the oven method,where 25 g of sample was placed in an air oven set at 103°Cfor a duration of 24 h (ASAE 1993). The durability of pelletswas determined by the DURAL tester developed at the Uni

versity of Saskatchewan(Sokhansanj et al. 1991;Larsen et al. 1996) using a 100g sample. The impeller speedof the DURAL tester was set at1600 rpm and time was set at30 s. Tested pellets were sievedon a 5.95 mm round hole sieve.The mass retained on the sieveexpressed in percentage of thetotal mass was the durability ofthe pellets. The hardness ofpellets was taken as the crushing strength (yield point) ofpellets during a compressiontest. It was measured by compressing pellets placed in their

Binder*

1994 cropHydrated limeHydrated lime

Pea starch

Pea starch

Hydrated lime +pea starch

Inclusion level(%)

Unit cost(Can $/kg)


Bindert

I. 1993 crop IICollagen protein 2.80 0.20Lignosulfonate 0.38 1.25Bentonite 0.08 5.00Hydrated lime 0.26 1.90Pea starch 0.68 0.74

t Grinds from low, medium, and high quality chops were used.

* Only the grind from low quality chop was used.

Calcium hydroxide is a chemical binder. It hydrates whenadded with water. When mixed with molasses, it forms acalcium salt of one or more sugars resulting in an insolublesalt which does not interfere with evaporation of water andwhich may aid in forming cross-linkages of carbohydrates(Pietsch 1976). Calcium hydroxide in the presence of carbondioxide is also a good chemical binder. "Type N" hydratedlime (Continental Lime, Ltd., Exshaw, AB) was used in thisstudy. It had 94.5% available calcium hydroxide.

Material

The material was dried chopped alfalfa. In this process, freshalfalfa or alfalfa from round baled hay is chopped and driedin rotary drum dryers. The dehydrated alfalfa is ground,conditioned with steam, and pelleted.

Three qualities of alfalfa chops were used. The quality gradingwas based on the durability ofpellets these chops had produced inthe plant without binders: less than 70% durability with lowquality chop; 70 to 80% durability with medium quality chop; andmore than 80% durability with high quality chop.

Two series of experiments were conducted, one in 1994using chops from the 1993 crop and the other in 1995 usinglow quality chop from the 1994 crop. The moisture contentsof the 1993 crop as received were: 6.1 % (w.b.) for low qualitychop, 6.4% for medium quality chop, and 8.1 % for highquality chop. The moisture contents of the grinds beforepelleting were 4.7, 5.3, and 6.9% respectively, for low, medium, and high quality chops. Moisture loss was 1.1 to 1.4%during 4 months storage in the laboratory. The moisturecontent of the 1994 crop as received was 8.2% (low qualitychop). Before pelleting, the moisture content of the grindsaveraged 6.9%. Moisture loss was 1.3%.

Five binders were used in the experiments performed onthe 1993 crop (Table II). Inclusion rate of these binders wasbased on a cost of $ 3 to $ 5 per tonne of pellets and is shownin Table II. Pea starch and hydrated lime were further studiedin the se~ond set of experiments performed on the 1994 crop.Hydrated lime and pea starch were each added to 5 kg ofgrinds at levels of 0.5% and 1.0% (Table II). A mixture ofthese was added at a level of 0.5% each. The alfalfa grindsand binders were mixed in a Hobart horizontal batch mixer.

CANADIAN AGRICULTURAL ENGINEERING Vol. :W. No. I. Janual)'/Fcbruary/March 1997 19

Table IV: Moisture content of hot pellets, cooled pellets, and alfalfa grinds; durability, hardness (yield point) andspecific energy consumption of alfalfa pellets, 1993 crop

Pellet type Moisture content (% wb) Durability Hardness Specific energy

Grind tconsumption

Hot pellets* Cooled pellets* (%)** (N)*** (kWeh/t)****

Low qualitywith binder 9.1tO.l 8.0tO.1 86.7tO.9 671tl8 30.3 (1.9)

Collagen protein 9.1tO.5 7.8tO.1 83.6t2.1 646t31 30.3 (1.1)

Hydrated lime 8.8tO.3 7.9tO.3 88.9t1.9 783t46 32.2 (1.1)Lignosulfonate 9.3tOA 8.0±0.1 85.8tO.7 654t36 30.7 (1.6)Bentonite 9.0tO.3 8.2±O.1 88.8t1.3 70lt29 29.2 (3.0)

Pea starch 9.3tOA 8AtO. I 86AtO.7 573t33 29.1 (2.1)

without binder 4.7tO.1 8.6tO.7 7.3tO.2 65.1t16.7 507t33 29.2 (104)

Medium qualitywith binder 6.6tO.3 5.9±O.2 77At1.7 737t20 35.0 (304)

Collagen protein 7.0tO.8 6.3±O.5 78.2tO.9 715t37 33.6 (0.2)

Hydrated lime 6.ltO.9 5.7tO.7 78.9t8.0 833t48 38.5 (1.1)

Lignosulfonate 6.3±O.2 5.5±O.1 74.lt5.3 676t34 3504 (4.6)Bentonite 7.1tO.9 6.3tO.6 78.2t2.7 716t46 31.3 (3.7)

Pea starch 6.5tO.2 5.8tO.1 77.3t1.6 747t45 36.5 (3.2)

without binder 5.3tO.1 6.9tO.7 6.2tOA 77.8t5A 669t40 35.9 (9.2)

High qualitywith binder 7.8tO.3 7.0tO.2 83.6t1.5 742±20 37.7 (6.5)

Collagen protein 7.5tOA 6.7tO.2 83.l±1.8 751±39 35.6 (3.6)

Hydrated lime 8.2±1.4 7.3tO.9 86.7±3.9 792t39 40.9 (5.7)

Lignosulfonate 8.0tO.8 7.0tO.6 84.9t1.9 737t50 36.5 (2.8)

Bentonite 7.5tO.9 6.9±0.6 79.9t5.9 756±48 42.1 (12.1)

Pea starch 8.0±O.6 6.9tO.7 83.8±3.6 674t33 3304 (8.5)without binder 6.9tO.7 9.2tO.6 8.0tOA 84.9±2A 632t39 40.3 (104)

t n=3.* n=6 (3 replicates for each of the 2 pelleting runs).** for pellets with binders n=30; for pellets without binder n=6.*** for pellets with binders n=300; for pellets without binder n=60.**** for pellets with binders n= 10; for pellets without binder n=2; number in parenthesis is standard deviation.

natural position (position at which a pellet rests when placedon a flat surface) using a 57.2 mm diameter flat plunger in theInstron® Modell 011 testing machine (Instron Corp., Canton,MA) with a crosshead speed of 10 mm/min. Hardness testwas repeated on 30 pellets selected randomly from a pelletingrun.


Tests on the 1993 crop

Moisture content Variation in moisture uptake was observed between different grinds from the three qualities ofalfalfa chops. Table IV lists the moisture content of thegrinds, hqt pellets, and cooled pellets. The grind from highquality chop had the highest moisture content of 6.9% followed by medium quality with 5.3% and low quality with4.7%. The input steam pressure was maintained between 235

20

to 250 kPa (Table III). About 1% difference in moisturecontent between the hot pellets and cooled pellets was observed. Grind from low quality chop adsorbed the highestamount of moisture during conditioning and pelleting. Theaverage moisture of hot pellets from low quality chop was9.0% from an initial grind moisture of 4.7%. The medium andhigh quality grinds adsorbed 1.2 to 1.4 percentage points ofmoisture, respectively. The moisture content of pelletsamong various binder and grind mixes was not significantlydifferent (a. =0.05). At the levels applied, the binders did notalter the moisture adsorption properties of the grind. However, it appeared that the variation in moisture content ofpellets from high quality grind was the highest.

Pellet durability and hardness and specific energy consumption Table IV also lists the durability and hardness ofpellets and the specific energy consumption grouped by chopquality. Among pellets without binder (control), the most

TABIL. SOKHANSANJ and TYLER

Table V: Moisture content of hot pellets, cooled pellets, and alfalfa grinds; durability, hardness (yield point) andspecific energy consumption of alfalfa pellets, 1994 crop

Binder and inclusion level

Hydrated lime (I %)Hydrated lime (0.5%)Pea starch (I %)Pea starch (I %)Hydrated lime (0.5%)+

Pea starch (0.5%)ControlMean

Moisture content (% wb) Durability** Hardness*** Specific energyconsumption****

Grind t Hot pellets* Cooled pellets** (%) (N) (kW-h/t)

6.7±1.2 8.6±1.0 7.5±0.6 70.2±5.3 502±24 34.5 (9.5)

6.9±1.5 9.4±1.0 7.7±0.6 69.7±5.4 471±23 31.5 (4.2)

6.9±1.7 8.8±0.7 7.2±0.4 62.7±2.3 467±22 29.4 (4.4)

7.0±1.7 9.0±0.6 7.4±0.5 62.9±3.9 439±24 30.6 (0.9)

6.9±1.7 8.2±0.6 6.9±O.5 54.0±7.0 472±26 34.0 (9.1)7.0±1.6 8.9±0.9 7.5±0.2 54.6±7.5 425±27 34.3 (6.4)6.9±OA 8.8±0.3 7.4±0.2

t n=4

*n=1O**n=12*** n = 120****n =2; numbers in parenthesis are standard deviations

durable were those from grind of high quality chop, followedby those from grind from medium quality chop. The leastdurable were pellets from grind of low quality chop withaverage durability of 65.1 ± 16.7%. There was a high variability in the durability values of low-quality-chop pelletsdue to variability between pelleting runs. The first pelletingrun produced pellets with an average durability of 50.6%(S.D. =1.5%) while the second pelleting run produced pelletswith an average durability of 79.6% (S.D. = 0.5%). In bothpelleting runs the durability of low-quality-chop pellets without binder was the lowest. In the first run, the moisturecontent of hot pellets from the low quality chop withoutbinder was 8.1 %, while for the second run it was 9.2%. Thelower moisture adsorbed in the first run may have adverselyaffected the durability of first run pellets (low quality chopwithout binder). The varying flow rate of the grind (whosecontrol by the vibratory feeder was not precise) may havealso contributed to the durability variation. Grind flow rateaffects the uniformity of compression of grind particles in thedie-roller assembly of the pellet mill.

With binder use, the durability of pellets from grind of lowquality chop improved considerably with durability comparable to pellets from high quality chop. Pellets from mediumand high quality chops did not respond favorably to theaddition of binders. It was also observed that the durabilityof pellets with hydrated lime was highest in all three chopqualities tested. The other binders (pea starch, collagen protein, lignosulfonate, and bentonite) also made more durablepellets out of low quality alfalfa chop than the control (nobinder).

The difference in initial moisture content between thethree qualities of grind may not have caused differences intheir durability values. Pellets from low quality grind withoutbinder, having an initial moisture content of 4.7%, adsorbed3.9% moisture but the mean durability was only 65.1 %. Onthe other hand, pellets from high quality grind without

binder, having an initial moisture content of 6.9%, adsorbed2.3% moisture and the mean durability was 84.9%.

The durability of pellets may be affected by amount ofmoisture adsorbed during steam conditioning. The durablepellets produced by the addition of hydrated lime may havebeen caused by the hydration of lime upon steam conditioning and pelleting. The calcium hydroxide (Ca(OH)2) inhydrated lime may also have reacted with carbon dioxide(C02) of the atmosphere in the presence of moisture (Lea1970). This formed calcium carbonate (CaC03), which mayexplain the hardening effect. The other binders (pea starch,collagen protein, lignosulfonate, and bentonite) rely on anincrease in moisture to release their binding power. Table IVshows that binders seemed not effective in improving thedurability of medium and high quality dehydrated pellets.However, binders worked well with low quality dehydratedpellets. During pelleting, steam added between 1.5 and 2.5%of moisture to the processed pellets. Table IV also shows thatlow-quality-chop pellets adsorbed the most moisture duringpelleting. With the presence of binders and free water fromsteam, binding of low quality grind particles may have improved.

Hardness is an important quality parameter for resistanceof pellets to compressive forces during storage. It affects thepalatability of the pellets as well. However, no studies wereconducted relating the palatability of pellets and their hardness. Table IV shows the hardness of pellets groupedaccording to chop quality. Hardness values had wide variability which may be due to: a) the variation in the length ofpellets, with longer pellets usually requiring higher breakageforce than shorter ones; b) presence of cracks in some pellets;and, c) in many cases due to non-uniform compression ofgrinds during pellet manufacturing resulting in unevenstrength within the pellet.

Among pellets without binder (control), those made frommedium and high quality chops were harder compared to


those from low quality chop. Pellets containing hydrated limewere the hardest regardless of the chop quality. For pelletsusing other binders (collagen protein, lignosulfonate, bentonite, and pea starch), higher hardness values were observedfrom pellets made from high and medium quality chopscompared to pellets from low quality chop.

The power measurement in each of the pelleting runs wasfor a duration of 90 s with sampling interval of 15 s. Thepower drawn by the motor fluctuated but averaged about 1.2kW. Power drawn by the pellet mill motor was affected bythe moisture and flow rate of conditioned grind. When thegrind was too dry or the flow rate decreased, the power drawnby the motor went down. High power draw was observedwhen the pellet dies were about to plug.

Table IV shows the average specific energy consumption(kW·h/t) in the two pelleting runs conducted for each of thegrind-binder mixes. Results showed that pellets made fromhigh and medium quality chops consumed more energy pertonne of processed pellets than from low quality chop. Thehigh specific energy consumption was associated with increased durability and hardness of pellets. The use of bindersin pellets did not affect the specific energy consumption forany of the chop qualities.

Tests on the 1994 crop

Test results from the 1993 crop showed that only low qualitychops had an improved pellet durability with the addition ofbinders. It was also found that hydrated lime consistentlyproduced durable and hard pellets. Pea starch was also foundto increase durability but the hardness was similar to thecontrol. The other binders used were either expensive or noteffective at the inclusion levels used. The inclusion level ofhydrated lime in the previous tests was 1.9% which is quitehigh considering that alfalfa hay has a natural calcium content of 1.4%. In the test based on the 1994 crop, the inclusionlevel of hydrated lime was reduced to I% or 0.5%.

Moisture content of pellets Table V shows the moisturecontent of hot and cooled pellets and the grinds. The grindfrom low quality chops had an average initial moisture of6.9%. The moisture content of hot pellets was higher and hadgreater variability than that of cooled pellets. Moisture content of hot pellets was on the average 1.4 percentage pointsabove that of the cooled pellets. The amount of moistureadded to the grind by steam during pelleting ranged between1.3 and 2.5 percentage points. The amount of steam adsorbedduring conditioning was not affected by the inclusion of thebinder to the grind. The moisture of hot pellets was 8.8 ±0.3% compared to the 7.4 ±0.2% of the cooled pellets. Uponcooling, the moistures reduced by 1.1 to 1.7 percentagepoints, which also was not affected by the binder added to thepellets.

Pellet durability and hardness and specific' energy consumption Table V also lists the durability and hardness ofpellets and the specific energy consumption grouped by thebinder used. Large variation in durability was observed between samples from the two pelleting runs in each of thetreatment combinations. This variation may have beencaused by the variability of grind flow rate. The results showthat pellets with highest durability were those with hydrated

22

lime either at I% or 0.5% level. Pellets containing pea starchwere second in durability. Pellets without binder, as well aspellets with both the lime and pea starch, had the lowestdurability. An inclusion level of 0.5% of either hydrated limeor pea starch was found adequate enough to improve pelletdurability.

Table V also shows that pellets containing I% and 0.5%hydrated lime were the hardest followed by pellets with 0.5%starch, and I% pea starch and combination of hydrated limeand pea starch. Pellets without binders had the lowest hardness.

The power of the pellet mill motor was not stable butfluctuated from one sampling time to the next which similarly was observed in pelleting runs for the 1993 crop. Theenergy consumption of the mill was taken as the area underthe graph of time versus the power. Specific energy consumption (kW·h/t) differed between pelleting runs andbetween binders. Pelleting run 2 had higher specific energyconsumption than run 1. This may have been due to flow rateand moisture variation of conditioned grinds. Specific energyconsumption was not affected by the binder.

The result in the 1994 test for low quality chop (1993 crop)showed that pellets without binder had an average durabilityof 65%, hardness of 507 N, and specific energy consumptionof 29.2 kW·h/t (Table IV). The low quality chop from the1994 crop had an average pellet durability of 54.6%, hardnessof 425 N, and specific energy consumption of 34.3 kW·h/t(Table V). The lower pellet durability and hardness of pelletsfrom low quality chop of the 1994 crop may have been dueto its lower crude protein content which was 17.6% (drymatter basis) compared to the 22.0% of the low quality chopfrom the 1993 crop (Tabil 1996). The bulk density of thegrind from low quality chop of 1994 crop was 202 kg/m3. Itwas lower than the bulk density of the grind from low qualitychop of 1993 crop which was 237 kg/m3

. Since materialswith low bulk densities require more energy to pellet(MacBain 1966; Leaver 1985), the specific energy consumption in pelleting the grind from low quality chop (1994 crop)was higher than the grind from low quality chop of 1993crop.

CONCLUSION

From the experimental data in this study, the following conclusions can be drawn:

I. The addition of binders to grinds from low quality chopsignificantly improved pellet durability and hardness tolevels comparable to that of pellets from high qualitychop. However, binders did not improve the durabilityof pellets made from medium and high quality chops.

2. Among the binders used in this study, hydrated lime andpea starch were found to be promising. Hydrated limeproduced the most durable and hardest pellets. Peastarch on the other hand, produced durable pellets without necessarily increasing pellet hardness.

3. An inclusion level of 0.5% of either the hydrated lime. or pea starch can be used. Significant increases in dura

bility and hardness of pellets were achieved at thisinclusion level for pellets from low quality chop.

TABIL. SOKHANSANJ and TYLER

4. Binder usage did not affect the specific energy consumption during pelleting.

ACKNOWLEDGMENT

This research was funded by the Strategic Research Programof Saskatchewan Agriculture and Food under the AgricultureDevelopment Fund program. The Saskatchewan Pulse CropDevelopment Board also contributed funds for this project.We acknowledge the contribution of Parrheim Foods, Ltd. ofSaskatoon, SK and especially that of Mr. Ken Fulcher.

REFERENCES

Anonymous. 1983a. Special report: binders. Milling 166(2):31-33.

Anonymous. 1983b. Conditioning mixed feeds for higherquality. Milling 166(10): 22-23.

ASAE. 1993. ASAE S358.2 - Moisture measurement forages. In ASAE Standards 1993, 451. St. Joseph, MI:ASAE.

Ghebre-Sellassie, I. 1989. Mechanism of pellet formationand growth. In Pharmaceutical Pelletization Technology,ed. I. Ghebre-Sellassie, 123-143. New York, NY: MarcelDekker, Inc.

Haas, L.A., J.A. Aldinger and R.K. Zahl. 1989. Effectivenessof organic binders for iron ore pelletization. Report ofInvestigations No. 9230. United States Bureau of Mines,Pittsburgh, PA.

Larsen, T.B., S. Sokhansanj, R.T. Patil and WJ. Crerar.1996. Breakage susceptibility studies on alfalfa andanimal feed pellets. Canadian Agricultural Engineering38( 1):21-24.

Lea, F.M. 1970. The Chemistry ofCement and Concrete, 3rdedition. London, U.K.: Edward Arnold (Publishers) Ltd.

Leaver, R.H. 1985. Pelleting dies: characteristics andselection. Sprout-Waldron Feed Pointers 26: 1-6.

MacBain, R. 1966. Pelleting Animal Feed, Arlington, VA:American Feed Manufacturers Association.

MacMahon, MJ. 1984. Additives for physical quality ofanimal feed. In Manufacture of Animal Feed, ed. D.A.Beaven, 69-70. Herts, England: Turret-Wheatland Ltd.

Pfost, H.B. and L.R. Young. 1973. Effect of colloidal binderand other factors on pelleting. Feedstuffs 45(49):21-22.

Pietsch, W. 1976. Roll Pressing. New York, NY: Heyden andSon, Inc.

Pietsch, W. 1984. Size enlargement methods and equipment- Part 2. Agglomerate bonding and strength. In HandbookofPowder Science and Technology, eds. M.E. Fayed andL. Otten, 231-252. New York, NY: Van NostrandReinhold Co.

Rumpf, H. 1962. The strength of granules and agglomerates.In Agglomeration, ed. W.A. Knepper, 379-418. NewYork, NY: John Wiley and Sons.

Sastry, K.V.S. and D.W. Fuerstenau. 1973. Mechanisms ofagglomerate growth in green pelletization. PowderTechnology 7:97-105.

Sokhansanj, S., R.T. Patil, 0.0. Fasina, J. Irudayaraj and G.Ahmadnia. 1991. Procedures for evaluating durabilityand density of forage cubes and pellets. CSAE Paper No.91-402. Saskatoon, SK: CSAE.

Tabil, L. 1996. Pelleting and binding characteristics ofalfalfa. Unpublished Ph.D. thesis. Department ofAgricultural and Bioresource Engineering, University ofSaskatchewan, Saskatoon, SK.

Wood, J.F. 1987. The functional properties of feed rawmaterials and their effect on the production and quality offeed pellets. Animal Feed Science and Technology18:1-17.

York, P. and N. Pilpel. 1972. The effect of temperature on themechanical properties of some pharmaceutical powdersin relation to tableting. Journal of Pharmacy andPharmacology 24:47P-56P.

York, P. and N. Pilpel. 1973. The tensile strength andcompression behaviour of lactose, four fatty acids, andtheir mixtures in relation to tableting. Journal ofPharmacy and Pharmacology 25:IP-IIP.


Potential for the psychrophilic anaerobictreatment of swine manure using a

sequencing batch reactor, I 3 3 2 ..,

D.I. MASSE, R.L. DROSTE, KJ. KENNEDY-, N.K. PATNI and J.A. MUNROE-

J Dai'l. and Swine Research and Development Centre. Agriculture and Agri Food Canada, Lennoxville. QC, Canada, JIM123; Centre for Food and Animal Research, Research Branch, Agriculture and Agri-Food Canada, Ottawa, ON, CanadaKIA OC6; and 3Department of Civil Engineering, Ottawa University, Ottawa, Ontario, Canada, KIN 6N5. Received 22September 1995; accepted I November 1996.

Masse, 0.1., R.L. Droste, Kennedy, K.J., Patni, N.K. and Munroe,J.A. 1997. Potential for the psychrophilic anaerobic treatment ofswine manure using a sequencing batch reactor. Can. Agric. Eng.39:025-034. The feasibility of psychrophilic anaerobic digestion(PAD) in intermittently fed sequencing batch reactors (SBR) wasinvestigated during the start-up run of an ongoing laboratory study.The start-up run results indicated that PAD in SBRs was efficient instabilizing and deodorizing swine manure slurry. The digester effluents had little odour when compared to the raw manure. Totalchemical oxygen demand (COD) was reduced by 58 to 73% andsoluble COD (SCOD) by 85 to 96%. Methane production variedfrom 0.30 to 0.66 L CH4/g volatile solids added and methane concentration in the biogas ranged from 50 to 80%. The biog<lsproduction rate continued to increase even when concentrations ofacetic acid and ammonia nitrogen were as high as 5500 mg/L and3700 mg/L, respectively. Keywords: anaerobic digestion, swine manure, biogas, manure treatment, psychrophilic process, anaerobictreatment.

Cet article presente les resultats preliminaires du projet d'etudesur la digestion anaerobie en condition psychrophile dans un bioreacteur it operation sequentielle. Les resultats experimentaux ontdemontre que cette nouvelle technologie desodorise et stabilise Ielisier de porco Le lisier traite est presque inodore comparativement aulisier de porc brut: La demande chimique en oxygene totale a elereduite de 58 it 73%. La demande en oxygene chimique soluble asubit une forte diminution variant de 85 it 96%. La production demethane etait de 0.30 it 0.66 litre de CH4 par gramme de solidesvolatiles alimentes aux bio-reacteurs. La concentration du methanedans Ie biogaz variait entre 50 et 80%. Ce procede est tres stable, iln'est pas affecte par des concentrations elevees d'acide acetique(5500 mg/l) et ammoniac (3700 mg/l).

INTRODUCTION

Animal manures have produced a growing public concern inCanada and the U.S.A. because of their potential to producestrong odours, encourage fly breeding, introduce weed problems, and pollute air, soil and water. The CanadianAgricultural Services Coordination Committee (CASCC1991) and Canadian Agricultural Research Council (CARC1991) recommended further research that would allow farmers to adopt sustainable and environmentally soundagricultural practices where animal manure is integrated intothe overall production system. A National Workshop onLand Application of Animal Manure in Canada, recommended that processes be developed to stabilize, deodorize,

and add value to animal manure (Leger et al. 1991).At the present time there is no economical, stable, and

easy-to-operate process to stabilize, deodorize, and add valueto, or recover energy from, animal liquid manure. Psychrophilic anaerobic digestion (PAD) at temperatures rangingbetween 5 and 25°C holds promise for success under Canada's cool climatic conditions compared to mesophilic andthermophilic anaerobic processes previously studied. Thefeasibility of using PAD at 20°C in intermittently fed sequencing batch reactors (SBRs) was examined as a possibletreatment to: a) reduce the pollution potential; b) recoverenergy; and c) reduce odours of swine manure slurry on bothsmall and large fann operations.

LITERATURE REVIEW

Feasibility of psychrophilic anaerobic digestion

Masse (1995) carried out an extensive literature search onPAD. A limited number of studies using municipal wastewater and animal manures (Balsari and Bozza 1988; Chandleret al. 1983; Cullimore et al. 1985; Kroeker et al. 1979; Lo andLiao 1986; Maly and Fadrus 1971; O'Rourke 1968; Sutterand Wellinger 1987; Wellinger and Kaufmann 1982; Zeemanet al. 1988) have demonstrated that PAD has the potential tobe used successfully as a low cost process to produce methane from animal manure; but there was a large variation inPAD process performance for unexplained reasons. The energy or fibre content of the diet of the animals and thepresence of antibiotics or food additives were not indicated.Also, several reports did not provide information on the ageof the manure or its characteristics. Most of the studiesconcentrated on biogas production while little considerationwas given to odour reduction, waste stabilization, or increasein availability of plant nutrients. Additional research is therefore necessary to evaluate precisely the feasibility of PAD inSBRs.

Description of SBR system

An SBR is a simple operating system (Fig. I). It consists ofa tank where the following five consecutive steps take place:I) fill; 2) react; 3) settle; 4) draw; and 5) idle. During the fillperiod the organic waste is loaded into the SBR. When theSBR is full, the react period starts and its length should be

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No. I. January/February/March 1997 25

Settle

Draw

React

offers (he Oexibilily of coordinating simultaneous operationof two or morc SBRs.

Potenlial ben eli Is or an SBR to carr)' out PAD ul' animalmanures

The microbial activities and ecosystem of anaerobic digestion are affected by digester design as well as byenvironmental and operational conditions (Harper and Suidan 1991). The SBR is highly suitable for trea{Jllcnt of animalmanure at ambient temperatures because it offers optimumconditions to retain a high concentration of slow growingmicroorganisms in the tank. Dague et al. (1992) indicatedthat with an anaerobic SBR the food to microorganism (F/M)ratio is high after the filling period and low just prior to theselliing period. Thcse operating conditions resulted in efficient bioflocculation and solids separation. Dague et al.(1992) also indicated Ihal wilh an SBR the partial pressure orC02 is maintained in the reactor during the settling period.As a result, no significant quantity of C02 is transferred tothe head space. This rcduces suspcnsion or resuspension ofpaniculates in the supernatant that can occur when C02 istransferred from the liquid to thc gas phase. The long biomassretention time in the SBR may allow PAD to adapt 10 environmcntal changes such as temperature variations. changes inorganic loading rate. and presence of inhibitory elements.

Another very important feature of an SBR is that it maynot rcquire continuous feeding. As a result. PAD in :.H1 SBRshould not illlerfere with regular farm operations as previoussystems did. It can be loaded during l1on11almanure removaloperations and the farmcr wilJ not have to deal with dailydigester efnuent. Thc SBR efnuent will need to be handledonce cvery one or two mOl1lhs. depending on operating conditions. Because intermittent fceding will make use ofcxisting manure handling equipmcnt at the farm and shouldnot disrupt regular farm operations. the SBR has thc potentialto successfully treat animal manure on small and large operations.

The main disadvantage of an SB R is that its biogas-usestrategy is more difficult to plan because the biogas production is not uniform during thc fill and react periods. Othcrdisadvantages arc that no control strategies and expcrimentaldata are available for PAD of animal manure in an SBR.

addanimalmanure

reactionperiod

removetreatedmanure

clarify

A A A A A A·

tF~· ·0°· ·0°· .0°. ·0·. ·0·· ·0°."0°·"••0°. ·00. ·0°••0°••0°. 0.0°. ·0°. ·0°·

. . '!0

0 o 0-a

00

~O O~ 0 0

,<>. 0 • 0 . O· OOo°(jb"00 0 000 0 0 0 0 0o 00" 0 .. 0 .. 0 0

000 0 0 0 000 0 0.. 0 0 0 .. 0 0 0 .. 0

Fill

A A A A A A A !o O\..T0 O\..T0 OVo OVo 0 0 0 0 0 0 0

0 0 0 0 0 0 0 00 o 0 0 0 o 0 0

wastewater

Idle EXPERIMENTAL PROCEDURE

Experiments were carricd out in laboratory scale digesterslocated in a temperaturc-controlled room maintained at 20oC.

wastesludge

Fig. l. Typical operation for a SBR during a completecycle (Metcalf and Eddy 1991).

sufficient to meet the IrcalmCIlI objectives. During the settleperiod 110 mixing is provided and quiescent settling conditions prevail lO allow treated liquid to be separated from thesolids and 10 retain bacteria in the system. During the drawperiod the trcalcd liquid is removed and finally the idle period

EXI>crimcntal design

To apply results to the farm, laboratory tests should closelysimulatc the actual farm opcrations. On a typical farm. manurc is generally rcmoved from the barn once to three timesa wcek. Therefore. the SBRs wcre intermittently fcd once tothree times pCI' week. The fill pcriod was limited to a monthto limit the volume of the SBR. The react period was selectcdto produce almost odourless effluent with reduced pollutionpotential and incrcased fertilizer value. For PAD in ,111 SBRto be cost effective, it is important that the operational cost iskept low. The operation should occur at ambicnt tcmperaIIIres and mechanical mixing should be minimized.

For the start-up run. effccts of inoculum type and loading

26 MASSE. DROSTE. KENNEDY. PATNI ami MUNROE

rate on the process were investigated. Fill and react periodlengths were kept constant and no mixing was provided to theSBRs. Operating conditions for the start-up run are given inTable I. Four pairs of bioreactors were used in this study.Each pair of bioreactors investigated a different operatingcondition. Therefore the experimental data represent the average response of two bioreactors.

Loading rates in Table I are calculated according to:

*Equivalent loading rate if the swine manure would have been fed continuously.

** A - Agropur Sludge

B - Mixture (79% Agropur and 21 % Municipal Sludge)

SBR Loading rate Fill React

No. period period(g COD-L-I-d- I)* (week) (week)

1 - 2 0.72 4 43-4 0.72 4 45-6 1.20 4 47-8 1.20 4 4

5 Influent line6 gas outlet7 gas meter8 feeder tube

Analytical methodsSamples were analysed for pH, alkalinity,solids, volatile acids (VA), total Kjeldahl nitrogen (TKN), ammonia nitrogen, total COD,and soluble COD (SCOD). Some samples

1 25 L nalgene digester2 sludge bed zone (7.5 L)3 variable volume zone (12.0 L)4 head space zone (5.5 L)

A

BA

B

type

Fig. 2. Schematic of laboratory scale SBRs used for thestart up run (test run no 4).

1.6 L of anaerobic non-granulated sludge obtained from theRobert O. Pickard Environmental Centre, Ottawa, ON). TheAgropur sludge substrate consisted mainly of fats and proteins. The anaerobic municipal sludge substrate came fromboth primary and secondary clarifiers. Municipal sludge thatis already acclimatized to compounds such as cellulose,hemicellulose, and lignin should increase treatment efficiency.

The feeding procedure consisted of adding fresh swinemanure slurry to the feeder tubes. Thereafter nitrogen gaswas used to pressurize the individual feeder to transfer themanure slurry to the SBR. This feeding method worked very

well and it took less than one minute to deliver the feed to an SBR. A mixed liquorsample of 100 mL was withdrawn from eachSBR at the beginning of the experiment andat 7 day intervals after the start of the experiment. Additional 100 mL samples werewithdrawn from the supernatant and settledsludge bed zones at the end ofthe experiment.Swine manure slurry was sampled immediately before it was fed to the SBRs.

Inoculum**

(I)L= VfCfVi If

where:L =loading rate (g COD.L-1.d- I ),

Vf =volume of feed (L),Cf = concentration of chemical oxygen demand (COD)

fn the feed (mg/L),Vi = volume of sludge in the reactor at the beginning of

the cycle (L), andIf =duration of the fill period (d).

Experimental equipment

The bench scale SBRs and feeding system used in this studyare illustrated in Fig. 2. Eight 25-L nalgene bottles were usedin the startup runs. Wet tip gas meters were used to measurebiogas production.

Swine manure slurry collection and storageManure slurry that was up to 4 days old was obtained fromgutters under a partially slatted floor in a growing-finishingbam at a commercial swine operation. It was screened toremove particles larger than 3.5 mm as these large particlestend to create operational problems with small scale laboratory digesters (eg. plugging of influent line). The raw manurewas mixed to reduce experimental variation and feed sampleswere prepared and stored in a freezer at -15°C to preventbiological activity. Manure feed samples were warmed to thedigester operating temperature (20°C) prior to feeding.

Start-up of the SBR

All eight digesters were initially started using 7.5 L of anaerobic granular sludge obtained from the AgropurCo-Operative anaerobic wastewater treatment plant at NotreDame du Bon Conseil, QC. Digesters 3, 4, 7, and 8 eachreceived a mixture of sludge (5.9 L of Agropur Sludge and

Table I: SBR operating conditions for the start-up run


EXPERIMENTAL RESULTS

*% Oxygen = 100% - (% Carbon + % Nitrogen + % Hydrogen)

DB-FFAP high resolution column. Biogas composition wasdetermined by using a Carle 400 AGC gas chromatograph. C,H, and N were determined using LECO CHN 600 analyzer.

were also analysed to determine concentration of C, H, andN. Biogas production was monitored daily and its composition analysed weekly. Gas samples were withdrawn with 10mL syringes through septums located in digester gas lines.SCOD was determined by analysing the supernatant of centrifuged slurry according to the method developed byKnechtel (1978). Alkalinity, pH, TS, TSS, VS, VSS, andTKN were determined using standard methods (APHA1992). TKN and ammonia nitrogen were determined using aKjeltec auto-analyzer model TECATOR 1030 (Tecator AB,Hoganas, Sweden). VA concentrations were determined by aPerkin Elmer gas chromatograph model 8310, that had a

Concentration

Mean±S.D.

4.8 ± 0.12

3.6 ±0.20

3.0 ± 0.16

2.6 ± 0.30

39 ± 9.00

84 ± 10.00

7.5 ± 0.35

5.8 ± 0.40

7.4 ± 0.30

19.0 ± 2.70

6.3 ± 0.40

1.9 ± 0.15

2.5

2.43

4.15

l.31

38.18

4.69

6.10

51.00*

Constituent

Total solids (%)

Total suspended solids (%)

Volatile solids (%)

Volatile suspended solids (%)

Soluble COD (gIL)

Total COD (giL)

TKN (gIL)

NH4-N (gIL)

pHAlkalinity (g CaC03!L)Acetic acid (gIL)

Propionic acid (gIL)

Butyric acid (gIL)

Cellulose (% TS)

Hemicellulose (% TS)Lignin (% TS)Carbon (% VS)Nitrogen (% VS)Hydrogen (% VS)Oxygen (% VS)

Table III: Composition of Swine Manure

*% Oxygen = 100% - (% Carbon + % Nitrogen + % Hydrogen)

Agropur and municipal sludges yield the followingstoichiometric formulations for the volatile solids (VS) composition:

Municipal sludge: Cs Hll.S NO.66 01.8Agropur sludge: Cs H9.2S NO.84 02.7

Biogas production

Gas production at 20°C occurred without any breakdown orsign of process instability for a 2.5 months period from June14 to September I (Figs. 3 and 4). Shapes of cumulativebiogas production curves are similar for the four treatments.The rate of gas production was low during the fill period andit increased during the react period. The 30 day lag phase inbiogas production probably resulted from acclimatization ofmicroorganisms to a lower temperature and a new substrate(swine manure). During the react period the biogas production rate increased exponentially until the end of the periodwhen it started to decrease as the availability of substratebecame the limiting factor. Substantial amounts of biogaswere produced beyond the react period. This indicates thattreatment was not complete at the end of the react period.Therefore, during startup the organic loading rate (OLR)should be reduced or the react period should be extendedbeyond the 77 days used here.

The digesters with combined sludge produced the highestamount of biogas, perhaps because of an increased hydrolysisrate. Cumulative biogas production was 30 and 70% higherin these digesters compared to the digesters seeded withAgropur sludge at OLRs of 0.72 and 1.20 g COD.L- l.d-1

,

respectively. This combined sludge was already acclimatized

MunicipalSludge

2.62.3

1.31.2

55.98.4

10.625.1*

3.08.2

1.01.8

0.843.98

2.97.36.0

35.02.0

AgropurSludge

1l.010.7

5.65.4

48.419.647.54

34.41 *10.073.0

1.3

7.90.700.731.567.6

16.035.0

26.0

Constituent

Total solids (%)

Total suspended solids (%)

Volatile solids (%)

Volatile suspended solids (%)

Carbon (% VS)Nitrogen (% VS)Hydrogen (% VS)Oxygen (% VS)Soluble COD (gIL)

Total COD (gIL)

NH4-N (gIL)

TKN (gIL)

Cellulose (% TS)Hemicellulose (% TS)Lignin (% TS)pHAlkalinity (g CaC03!L)Operating temperature (oC )Sludge residence time (week)

Table II: Inocula Characteristics

Composition of swine manure slurry and inoculum

The manure had a neutral pH and high concentrations ofTCOD, SCOD, TKN, NH3-N, VA, and alkalinity. Based onthe concentrations of C, N, and H given in Table III, thecomposition of the insoluble organic fraction of the freshswine manure slurry was Cl.O HI.9 01.0 NO.1. This composition is similar to the formula for carbohydrates [CH20]n.

The main characteristics of the Agropur granulated sludgewere that it had very high solids, TCOD, SCOD, and TKN(Table II). The municipal sludge was less concentrated thanthe granulated Agropur sludge, but it had a higher fibrecontent on a dry weight basis and also had a lower alkalinity.Both sludges came from digesters operated at 3SoC. Theconcentrations of C, H, and N of the organic fraction of

28 MASSE. DROSTE. KENNEDY. PATNI and MUNROE

.........'

.'" .~..,,- ---, '".~ \, '".

': \, '"., . , "

,'::' ~\" .....,: i " .

,:.: Agropur sludge" ",~ ~

i ~,

,.;' 'l~....c' '.,

, -;.' - - - - • average of SBRs 5-6 ..~.,..~.., .,,: ,. average of SBRs 7-8 '~''':.'':.,,:

o l.f:.,~-'~f--__+-__+_-_+--+--_+__-_t_-+--+--_t---+-'

Fill period -~~- React period

o 7 14 21 28 35 42 49 56 63 70 77

Time (d)

5

6,...-------y---------r--==----:-=-::---,

Fig. 6. Average acetic acid concentration as a function oftime for SBRs with a loading rate of 1.2 gCOO-L-1-d-1•

i4--FIII period React period - _ Extended-:+. ..react period

- - - -. SBR #1.- ..._- SBR#2........ SBR#3- - - SBR#4

,..-:'.,-:~.. "

~.. ,~.. ,

combined sludge «. <:'.... ,.,pott

.;....-.~.~. .:>tt.J>tt

./. ptfJil'"

././ fl.J>tt

.~ ..1"."'" ,..,.""~""

~.~'--=:-"" Agropur sludge'~"'-"'-

100

~c.2 60U:J

"D

e 40a-U)ca

CJ20

80

oo 7 14 21 28 35 42 49 56 63 70 n

Time (d)

Fig. 3. Cumulative biogas production as a function oftime for SBRs with organic loading rate of0.72 g COO-L-1-d-1•

n70

.. .." ..

\\\

\\

\\\

63

Extendedreact period

5649

React period

35 42

Time (d)

'" Jj' .combined sludge '

.. ---1---- ..,,,, Agropur slUdge

282114

,,,,,,,,,,,,I

7

Fill period

o

• - - - • average of SBRs 1-2.......; average of SBRs 3-4

o 7 14 21 28 35 42 49 56 63 70 77

Time (d)

0.0 L....o':.....--+--+--+--~f__--+---+--_+__-_+_-+----lf--__+_J

1.2

0.0

1.2

1.0

~ 0.8'g

'uca 0.6u'20'Q. 0.40a-D.

0.2

Fill period React perIod

1.0 Agropur sludge _" ---.. --~ , ..

0.8 ,,"C ,U , ...... ........... ..ca ......2 0.6c ."0 combined sludgea. 0.40a-D.

0.2 Prfed - - - - • average of SBRs 5-6.......... average of SBRs 7-8

Fig. 7. Average propionic acid concentration as afunction of time for SBRs with a loading rate of0.7 g COO_L-1_d-1•

n

- - - - average of SBRs 1-2..... , .. average of SBRs 3-4

7 14 21 28 35 42 49 56 63 70 77Time (d)

1-4- Fill period React periodExtended~.. - react period

. - -- SBR#5 ..;;-''-"-'SBR#6 .-Y

.;;;-········SBR#7 .-7- - -SBRI8 .j-

.-7combined sludge ..;

~.:I "..."'""....,

.,;1 ./

.'/ K./ fI~

,/ ,.~ Agropur sludge1,.-'''''' ,.'

.," -'--'.-' .... ".'"." ......, ...-..

.-"":"tt-'"

o

oo 7 14 21 28 35 42 49 56 63 70

Time (d)

20

~§''; 60o:J"Doa 40U)ca

CJ

80

" - ~::.",'.... ""." ""'"I v,

I ~

I '~"I • '"

, I .... combined sludge ""

, ~ ~.:.,.... ~,~~ropur sl\Udge,.!-.. '".'

.('" .. :::: ... - - .o L.F:

4_·...c::::jf--__+-__+_---1---+---+---+--4--+--.'+-'';..-'-'.--.......

100

5

Fill period - .....~- React period6 ........-------.......------.....,.------,

Fig. 4. Cumulative biogas production as a function oftime for SBRs with organic loading rate of1.2 g COO-L-1-d-1

Fig. S. Average acetic acid concentration as a function oftime for SBRs with a loading rate of 0.7 gCOO_L-1_d-1•

Fig. 8. Average propionic acid concentration as afunction of time for SBRs with a loading rate of1.2 g COO-L-1_d-1.


to compounds such as cellulose, hemicellulose, and ligninwhile the Agropur sludge was only acclimatized to proteinsand fats which are the major constituents of cheese plant wastewater. Also, the activity of the municipal sludge may have beenhigher than that of the Agropur sludge. Actual sludge activitiesof inoculum were not measured in this study.

During the first 60 days, increased OLR had no significanteffect on biogas production for the digesters with theAgropur sludge. However, for the digesters with combinedsludge there was an increase in biogas production of 40%when the OLR increased from 0.72 to 1.20 g COD.L-1.d- l .

For the four treatments tested, the methane fraction in thebiogas was not constant. It continuously increased with time.At the start of the fill period the methane concentrationranged from 47 to 63% while at the end of the react period itranged from 77 to 80% for all treatments.

Volatile acids accumulationAcetic acid accumulated rapidly from 0 to 5500 mg/L duringthe fill period in the SBRs fed 1.2 g COD.L-1.d-1(Figs. 5 and6). This accumulation is about five times larger than theamount of acetic acid fed to the digesters. Propionic acid wasaccumulating faster in digesters with the Agropur sludge thanin digesters with combined sludge during the fill period(Figs. 7 and 8). For digesters with combined sludge, thepropionic acid accumulations were equal to the cumulativeconcentration fed. Butyric acid was not accumulating duringthe fill period, but rather was consumed because its concentrations were substantially lower than the cumulativeconcentration fed (Figs. 9 and 10).

The rapid increase in acetic acid concentration during thefill period indicates that hydrolysis and acidification wereoccurring and that the utilization of acetic acid by the methane formers was the rate limiting step. The rapid increase inacetic acid is usually due to the faster growth rate of acidformers or inhibition of methane formers by an increase inconcentration of VA or other compounds. Comparing Figs. 3and 4 with Figs. 5 to 10 demonstrates that methane formerswere not inhibited by the increase in VA concentrationsbecause during the period of increased VA concentration themethane production rate also increased. Therefore the increase in VA is more probably due to the faster growth rateof acid formers. The large increase in VA did not affect theprocess stability because: 1) alkalinity in the SBRs was veryhigh (16000 mg CaC03/L) (the large increase in VAs causedonly a small drop in pH); and 2) pH was maintained between7.5 and 7.8 (unionized VA concentration was always low at6 mg/L). Several existing models assume that the growth rateof methane formers is affected by the VA concentrationwhereas preliminary results from this work show that thistheory does not apply for acetic acid concentrations up to6000 mg/L in SBR anaerobic digestion of swine manure at20°C.

During the react period there was rapid utilization of aceticand butyric acids (Figs. 5, 6, 9, and 10) indicating thathydrolysis and acidification were the rate limiting processesduring the react period. When the OLR increased from 0.72to 1.20 g COD.L-1.d- l

, the maximum acetic, propionic, andbutyric acid concentrations in the SBR increased by 25, 13,and 33%, respectively (Figs. 5 to 10).

30

Inoculum type did not have much effect on acetic acidconcentrations although the SBRs with the combined sludgeinoculum had higher CH4 production and lower propionicand butyric acid concentrations at all times. Thus SBRs weremore stable with combined sludge than with Agropur sludgeand for this reason all subsequent experimental runs werecarried out with the combined sludge inoculum.

Propionic acid is the only VA that substantially increasedduring the react period (Figs. 7 and 8). A mass balance onpropionic acid shows that it was being utilized during the fillperiod, but at a rate lower than the feed and production rate.The increase in propionic acid might be due to an increase indissolved hydrogen concentration (Mosey 1983). Fukazaki etal. (1990) stated that fermentation of propionic acid to CH4and C02 is inhibited by dissolved hydrogen and acetic acid.Results for SBRs 3-4 (Fig. 7) indicate that propionic acid wasutilized even when the concentration of acetic acid was high.Therefore the propionic acid accumulation in this study maybe attributed to the effect of dissolved hydrogen in the SBRs.Inhibition of hydrogenotrophic methanogens may be anotherfactor for the increase of propionic acid.

Process stability

The pH level, alkalinity, and ammonia concentrations as afunction of time for the SBRs with an OLR of 1.2g COD.L-1.d- 1

were similar to curves obtained at the lower OLR of 0.7 gCOD.L-1.d-1 (Fig. 11). The pH ranged from 7.4 to 7.8. Thehigher concentration of VA during the react period did notaffect the microorganisms because of the high initial alkalinity. The increase in VA slightly reduced the pH and alkalinityduring the fill periods. During the react period both thealkalinity and pH started to increase mainly due to VA utilization. The contribution of ammonia-N to the pH andalkalinity during the react period was negligible becausethere was no increase of ammonia-N during this period (Fig.11). The high concentration of ammonia-N did not inhibitmethane formers because both the methane production andthe ammonia-N concentration increased simultaneously.Kroecker et al. (1979) found that ammonia is inhibitory to themethanogenic bacteria when its concentration exceeds 2000mg/L. Melbinger and Donnellon (1971) found that ammoniais toxic only when its concentration exceeds the thresholdlimit of 1700 to 1800 mg/L and is increasing faster than theacclimatization of the methanogenic bacteria. McCarty(1964) indicated that an ammonia-N concentration exceeding3000 mg/L is toxic to the anaerobic bacteria regardless of pH.Henze and Harremoes (1983) indicated that dissolved ammonia is substantially more toxic than ammonium ions toanaerobic bacteria. They indicated that a dissolved ammoniagas concentration ranging between 100 and 200 mg/L shouldhave an inhibitory effect on the anaerobic process. In this test,the total ammonia-N concentration (3700 m~/L) representsthe sum of ammonium ions (3550 mg NH4 NIL) and dissolved ammonia (150 mg NH3-N/L). Inhibition byammonia-N was not observed in this study. It is likely thatthe long hydraulic and solids residence times provided in thisstudy allowed the microorganisms to increase their toleranceto high concentrations of ammonia-N. PAD in SBRs appearsto be suitable to treat wastewater with a high nitrogen content.

MASSE. DROSTE. KENNEDY. PATNI and MUNROE

--t~_ React period

- - - - - - average of SBRs 5-6. . .. . . . . . .. average of SBRs 7-8

F/II period ~~- React period0.7

0.6

~0.5

'tJ 0.413

CISu 0.3'i:~::J

0.2m

0.1

", ,t·····'.,·, .. :.:~:'\ ..., ,. , "

, !.' .~ Agropur slUdge, ,. j,',.... . :::;;.:.~. ..

combined sludge .".~0.0 a::..--+---+---+---I------.--+-.....,;::.+---l-.....;.:.,If---l~---l

o 7 14 21 28 35 42 49 56 63 70 nTime (d)

Extendedreact period

- - - - - - average of SBRs 1-2........... average of SBRs 3-4

... _-- "

:i _-- ..combined sludge . ' '. '. .. ....

Agropur sludge--- ....~

..

28 35 42 49 56 63 70 77

Time (d)21147

0.7

Fill period0.6

~0.5

" 0.4U

CISu 0.3'i:

5'0.2m

0.1

2.0 '--- --4- --o-_~ >____~

5.0 ..-----------.-------...,..---------.

8.0 ,...-------,-------.----------,

7 14 21 28 35 42 49 56 63 70 nTime (d)

Fig. 10. Average butyric acid concentration as a functionof time for SBRs with a loading rate of 1.2 gCOD-L-I_d-I.

Energy recovery

The CH4 production ranged from 0.30 to 0.66 L/g VS formost of the experiment runs (Table IV). Methane productionobtained in this study was substantially higher than methaneproduction from swine manure obtained by digestion at 35°Cin continuous flow digesters by Kroecker et al. (1979), whoreported methane production of 0.45 L CH4/g VS added fora loading rate of 2.5 kg VS_m-3.d-1 and by Hashimoto(1983), who reported 0.42 L CH4/g VS added for a loadingrate of 2.5 kg VS-m-3-d- I

• The higher methane productionper gram of VS fed to the SBRs obtained in this study couldbe due to: I) the lower OLR (0.45 kg VS.m-3-d- I ) and longerhydraulic retention time; 2) the fact that the measured VS inthe influent is lower than the actual VS concentration because some VA and other soluble organics are volatilizedduring the VS determination; and 3) the fact that the measured methane flow rate includes the methane produced frommicroorganism decay. Another possible reason could be thatthe lower operating temperature and absence of mixing maintain higher concentrations of hydrogen and carbon dioxide inthe liquid phase. As a result more carbon dioxide can beconverted to methane by the hydrogen utilizing methanogens. Also, with the continuous flow anaerobic processespreviously tried, some C02, H2, and CH4 were lost in thedigester effluent. A high rate of methane production was notthe main objective of this work but these data are useful inassessing system performance and stability. Steady production of methane per unit mass of VS fed indicates that PADof swine manure at 20°C in the laboratory scale SBR digesters was a stable process.

Treatment efficiency

Total COD removal ranged from 58 to 73% and the VSremoval ranged from 27 to 74% (Table IV). Results for VSand total COD were highly variable due to rapid settling ofheavy particulates which affected VS and total COD determinations as well as the calculated methane production pergram of VS. The SCaD test results were consistent. HighSCaD removal ranging between 85-96% was achieved.

Reduction in swine manure slurry odours was one of the

Extended--'react period

average of SBRs 5-6average of SBRs 7-8

I

. - - average of SBRs 5-6average of SBRs 7-8

I

React

React

7 14 21 28 35 42 49 56 63 70 nTime (d)

, __Agrr sludge .

"" -.~.~ .. ·~·~·~······~·~·~·~·~·I· ~.~ ._'" '" - - - - combined sludge

... - ••.• . . . . • - - average of SBRa 5-6.. ' average of SBRa 7-8

Extended--react~n~."

.- -;;,.,.Agropur sludge ,,~.~.~ ....... ,

:.:.-~.>~~.:~ ~~~~ ~ :.c.•/· - -~-':: -- --combined sludge

f4- Fill

.- Fill

I---- Fill

10

20

7.8

7.0

7.2

4.5

7.6XQ,

7.4

7 14 21 28 35 42 49 56 63 70 nTime (d)

~4.0

~ 3.5

£Z 3.0

2.5

React - Extended-'react period18 ...... "", __ -

- Agropursludge,,"" --' \ r.-::;.~.~.~... '···~16 :''':.,:.~.",,_!_/ "\'~'>'': ''eI-f" ······7············

12 combined sludge

Fig. 11. PH, alkalinity (g CaC03/L) and NH3-Nconcentration as a function of time for SBRswith a loading rate of 1.2 g COD-L-I-d-I.

Fig. 9. Average butyric acid concentration as a functionof time for SBRs with a loading rate of 0.7 gCOD_L-I_d-I.

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I, January/February/March 1997 31

Table IV: Average Methane Production per Unit of VS Fed to the SBR and Reduction in Total COD, soluble CODand VS.

SBR Loading Rate Siudge* CH4 Removal(%)number type production after 56 days

(g COD/feed) g COD_L-1_d-1 LCH4/g VSafter 56 days TCOD SCOD VS

1-2 12.6 0.72 A 0.50 60.0 90.0 29.0

3-4 12.6 0.72 B 0.66 70.0 96.0 74.05-6 21.0 1.20 A 0.30 58.0 85.0 27.07-8 21.0 1.20 B 0.52 73.0 91.0 56.0

* Inoculum TypeA-loo% Agropur SludgeB - Combined Sludge (79% Agropur and 21 % Municipal)

objectives of this study. The major volatile compounds thatproduce odours in animal manure slurries are VA, amines,carbonyls, esters, hydrogen sulphide, and ammonia. Laboratory staff observed that test runs that achieved completeremoval of VA and more than 85% removal of soluble CODproduced treated manure that was relatively odourless compared to raw manure. A large reduction in soluble COD mayresult in complete utilization of amines, carbonyls, and esters. The actual degree of reduction in odour intensity was notdetermined because the techniques recommended to measureodour intensity are complex, subjective, time consuming, andcould not feasibly be used within the time frame of this study.Quantification of odours will be addressed in future studies.

CONCLUSION

Anaerobic digestion of swine manure slurry at psychrophilictemperature (20°C), in non-mixed sequencinybatch reactors,at loading rates of 0.7 and 1.2 g COD.L-1.d- ,stabilized anddeodorized the swine manure slurry. The digester effluentwas almost odourless when compared to the raw manure. TheSBRs were efficient in retaining the biomass. Up to 73%removal of total COD was attained by the process operated ata cycle time and conditions that are suitable for typical farmoperations. Methane production up to 0.66 L CH4/g VS wasobtained, with a methane content varying from 50 to 80%.This high biogas production and quality were not affected byhigh concentrations of volatile acids (6000 mg/L or higher)and ammonia-nitrogen (3700 mg!L) in the digester mixedliquor.

PAD at 20°C in intermittently fed SBRs is technicallyfeasible, stable and easy to operate.

ACKNOWLEDGEMENTS

The authors acknowledge the contributions to this work bytechnologists C. Defelice, A. Olson, and M. Lemieux andgraduate student Bryan Graham.

REFERENCES

APHA. 1992. Standard Method/or the Examination o/Waterand Wastewater, 18th ed. Washington, DC: AmericanPublic Health Association.

32

Balsari, P. and E. Bozza. 1988. Fertilizers and biogasrecovery installation in a slurry lagoon. In AgriculturalWaste Management and Environmental Protection,Proceedings of 4th International Symposium of theInternational Scientific Centre of Fertilizers, eds. E.White and I. Szabolcs, 71-80.

CARC. 1991. Proceedings of the National Workshop onLand Application of Animal Manure, eds. D.A. Leger,N.K. Patni and S.K. Ho. Ottawa, ON: CanadianAgricultural Research Council, Agriculture Canada.

CASCC. 1991. Annual Report. Ottawa, ON: CanadianAgricultural Service Coordinating Committee.

Chandler, I.A., S.K. Hermes and K.D. Smith. 1983. A lowcost 75 kW covered lagoon biogas system. InProceedings of the Symposium on Energy from Biomassand Waste VII, 627-646. Lake Buena Vista, Fl. January24-28.

Cullimore, R.R., A. Maule and N. Mansui. 1985. Ambienttemperature methanogenesis from pig manure wastelagoons: Thermal gradient incubator studies. AgriculturalWaste 12: 147-157.

Dague, R.R., C.E. Habben and S.R. Pidaparti. 1992. Initialstudies on the anaerobic sequencing batch reactor. WaterScience Technology 26:2429-2432.

Harper, S.R. and M.T. Suidan. 1991. Anaerobic treatmentkinetics. Water Science Technology 24(8):61-78.

Hashimoto, A.G. 1983. Thermophilic and mesophilicanaerobic fermentation of swine manure. AgriculturalWastes 6:175-191.

Henze, M. and P. Harremoes. 1983. Anaerobic treatment ofwastewater in fixed film reactors - A literature review.Water Science Technology 15:1-101.

Fukazaki, S., N. Nishio, M. Shobayashi and S. Nagai. 1990.Inhibition of the fermentation of propionate to methaneby hydrogen, acetate, and propionate. AppliedEnvironmental Microbiology 56(3):719-723.

Knechtel, J.R. 1978. A more economical method for thedetermination of chemical oxygen demand. Water andWaste Engineering 14(4):25-28.

MASSE. DROSTE, KENNEDY, PATNI and MUNROE

Kroeker, EJ., D.D. Schulte, A.B. Sparling and H.M. Lapp.1979. Anaerobic treatment process stability. Journal ofthe Water Pollution Control Federation 51 :718-727.

Leger, D.A., N.K. Patni and S.K. Ho. 1991. In Proceedingsofthe National Workshop 011 Land Application ofAnimalManure, 1-176. Ottawa, ON: Canadian AgriculturalResearch Council, Agriculture Canada.

Lo, K.V. and P.H. Liao. 1986. Psychrophilic anaerobicdigestion of screened dairy manure. Energy illAgriculture 5:339-345.

Loehr, R.C. 1977. Pollution Control for Agriculture.London, England: Academic Press.

Maly, J. and H. Fadrus. 1971. Influence of temperature onanaerobic digestion. Journal of the Water PollutionControl Federation 43(4):641-650.

Masse, D.1. 1995. Psychrophilic anaerobic digestion of swinemanure slurry in intermittently fed sequencing batchreactor. Ph.D. Thesis. University of Ottawa, Ottawa, ON.

McCarty, P.L. 1964. Anaerobic waste treatmentfundamentals; Part Three: Toxic material and theircontrol, process desi gn. Public Works J ourllalOctober:91-94.

Melbinger, N.R. and J. Donnellon. 1971. Toxic effect ofammonia nitrogen in high rate digestion. Journal of the

Water Pollution Control Federation 43(8): 1658-1670.

Metcalf & Eddy. 1991. Wastewater Engineering: Treatment,Disposal and Reuse, 3rd ed. Toronto, ON: McGraw-HilI.

Mosey, F.E. 1983. Mathematical modelling of the anaerobicdigestion process: Regulatory mechanism for theformation of short-chain volatile acids from glucose.Water Science Technology 15:209-232.

O'Rourke, J.T. 1968. Kinetics of anaerobic waste treatmentat reduced temperature. Ph.D. Thesis, StanfordUniversity, Palo Alto, CA.

Sutter, K. and A. Wellinger. 1987. ACF-System: A new lowtemperature biogas digester. In Proceedings of the 4thIllternational Symposium of the International ScientificCentre of Fertilizers, Braunschweig-Volkenrode,Germany, March 11-14.

Well inger, A., and R. Kaufmann. 1982. Psychrophilicmethane production from pig manure. ProcessBiochemistry 17:26-30.

Zeeman, G., K. Sutter, T. Yens, M. Koster and A. Wellinger.1988. Psychrophilic digestion of dairy cattle and pigmanure: Start-up procedure of batch, fed-batch andCSTR-type digester. Biological Wastes 26: 15-31.


Microbial interaction during the anaerobictreatment of swine manure slurry in a

sequencing batch reactor, 1 ?

DJ. MASSE and R.L. DROSTE-

IDail] and Swine Research and Development Centre, Agriculture and Agri-Food Canada, Lenlloxvil/e, QC, Canada JIM123; Department of Civil Engineering, University of Ottawa, Ottawa, ON, Canada, KIN 6N5. Received 18 March 1996;accepted 1 November 1996.

Masse, DJ. and Droste, R.L. 1997. Microbial interaction duringthe anaerobic treatment of swine manure slurry in a sequencingbatch reactor. Can. Agric. Eng. 39:035-041. A simple model thatsimulates the Psychrophilic Anaerobic Digestion (PAD) of swinemanure slurry in a Sequencing Batch Reactor (SBR) is developedand verified. The model predictions have been compared with corresponding laboratory results. The trends in volatile acids and solublechemical oxygen demand (COD) accumulation as well as the methane production rate were reasonably well predicted. The percenterror of estimate ranged between 12 and 37. The model is a usefultool to study the influence of SBR operating strategies on the dynamic interaction between the acid and methane formers. Keywords:anaerobic digestion, methane production, modelling, process kinetics.

Un modele mathematique est propose pour simuler la digestionanaerobie du lisier de porc en milieu psychrophile dans un bioreacteur aoperations sequentielles (BOS). Les predictions du modele ontete comparees avec des resultats experimentaux. Le % d'erreur surl'estimation de I'accumulation des acides volatiles, la demandechimique en oxygen soluble (DCO) ainsi que la production demethane variait entre 12 et 37. Le model est un outil pratique pouretudier I' influence des differentes strategies d'operation du BOS surl'interaction dynamique entre les differents groupes de bacteriesanaerobies qui transforment Ie lisier de porco

INTRODUCTION

A comprehensive study on Psychrophilic Anaerobic Digestion (PAD) of swine manure slurry in Sequencing BatchReactors (SBR) was conducted by Masse (1995). The temperature range for growth of psychrophilic bacteria includesthe range of temperatures normally found in manure storagegutters in animal shelters in Canada (5 to 20°C). An SBRanaerobic process occurs in a tank or a reservoir in thesequences given in Fig. I: fill; react; settle; draw; and idle.During the fill and react phases, the soluble organics andsome of the suspended organic particulates are removed bythe anaerobic microorganisms. During the settling phasethere is no mixing; this provides quiescent settling conditions(Dague et aI. 1992) for the separation of treated manure andsuspended solids. SBR operation retains a very high concentration of microorganisms in the digester. During the drawphase the treated manure is removed. The idle phase allowssome flexibility for the operation and' maintenance of theSBRs.

Masse (1995) indicated that PAD of swine manure slurryin SBR was very stable. The anaerobic bioreactors were not

affected by high concentrations of volatile acids (6500 mg!L)and ammonia nitrogen (NH3 + NH4+ = 3700 mg!L). Theproposed process stabilized and deodorized swine manureslurry and also produced a significant amount of high qualitybiogas (0.33 to 0.66 L CH4 /g of volatile solids added). Theseresults indicated that PAD in SBR has the potential to be astable, easy-to-use, and cost effective process to treat swinemanure slurry on Canadian farms. But before this process canbe recommended, additional laboratory tests are required toinvestigate the effect of other factors such as loading rates,temperature variation, solids content, animal diets, and manure handling practices. One problem is that laboratory testsrequire substantial amounts of time, especially when they arecarried out at low temperatures.

The objective of this work was to develop a simple comprehensive model to predict the bioreactor performanceunder different operating and environmental conditions.Such a model would be useful to:

I. gain a better knowledge of PAD in SBR;

2. predict the rate limiting steps during fill and reactphases;

3. reduce the number of experimental tests; and

4. optimize the bioreactor design and control strategy.

MODEL DEVELOPMENTThe authors opted for a comprehensive model that involvedsome simplifying assumptions. As a result, the model doesnot require a large number of kinetic constants. A dynamicmodel similar to the models developed by Hill and Barth(1977), Droste and Kennedy (1988), Jones (1989), and Jonesand Hall (1989) was developed for PAD in an SBR process.These previous models considered the two phases (acid andmethane formation) in anaerobic digestion, but apply to onlycontinuous flow or steady state bioreactors.

PAD in SBR not only has different flow regimes but alsohas transient conditions that always prevail during the fill andreact phases. The bioreactor operates as a semi-batch systemduring the feed phase and as a batch system during thereaction phase. The rate limiting steps during the transientfeed and reaction phases also differ (Masse 1995). The modelpresented below makes use of different assumptions and

CANADIAN AGRICULTURAL ENGINEERJING Vol. 39. No. I. January/Fcbruary/March 1997 35

non-linear differential equations.The simplified scheme of PAD in SBR used in this study

to develop the model (Fig. 2) includes the two major microbial groups (the acid formers and methane formers). Theproposed model will be used to simulate the biological phaseonly. The parameters considered are soluble chemical oxygen demand (SCOD), volatile acids (VA), and methaneproduction. Removal of volatile solids (VS) and total COD(TCOD) are not considered because in a SBR their removalis due to both biological degradation and settling; but hydrolysis of particulate matter to SCOD is considered.

Model assumptions

The model proposed for the simplified scheme is based on thefollowing assumptions:

addwastewater

Fill phase

influent

o 0 0 0 0 o· 0 00 0 0 0 0 00 0

o 00 00 o· 0 o·. 0 .0 0 0 00.00

wastewater---......

ISBRI

React phase

00 00 00 0 00. 0 0 00 00 0

o () '0 0 0 0 0 ()0·0' 00 00 0 0

0·0' 00' 00 00"o. 0 '0' 00 00 0o 00 0 0 () 0 0

reactionperiod

a) Swine manure

SolubleCOD

Settle phase

clarify

Volatileacids

Draw phase

AAAAAAAAAAA!

Idle phase

••

removetreatedwastewater

b)

TeODo---~

seODo---~

VA o-----~

SBR

TeODseOD

VAVssXmX a

wastesludge

Fig. 1. Operation of the anaerobic SBR process.

Fig. 2.. a. Simplified scheme for anaerobic digestionb. Parameters considered in simple modeldevelopment.

36 MASSE and DROSTE

where: Kp = first order solubilization rate (d- I).

Gujer and Zehnder (1983), indicated that a first-order

Particulate organics Droste and Kennedy (1988) used asoluble substrate and, therefore, did not consider hydrolysisof particulates. In the manure slurry, hydrolysis of particulates is important. During the fill phase the material massbalance for particulates is:

(8)

(9)

(6)

XaK saKdaKdm

XmF

where:SSoVmaXa

dXm __ Q Xm (vmaxm Xm VAJ-d - V + Ym Kdm Xm

t L Ksm + VA

where: Ym = methane formers yield factor.

hydrolysis rate for particulate solubilization may be the mostappropriate expression for complex wastes and this wastherefore used in this study.

The mass balance for a SBR during the fill phase is identical to the mass balance for a continuous flow stirred tankreactor (CSTR). The term PQ in Eq. 4 does not represent theeffluent output, instead it represents the reduction in concentration due to dilution caused by the increase in SBR liquidphase volume.

Similar mass balances for the fill phase were also developed for SCOD (Eq. 6), volatile acids COD (Eq. 7), acid andmethane fonners (Eqs. 8 and 9, respectively), as well asmethane production (Eq. 10).

Soluble COD

dS Q (So - S) (vmaxa Xa SJ- +KpP-dt VL K.m+S

= SBR SCOD concentration (mg COD/L),= influent SCOD concentration (mg COD/L),= maximum specific SCOD uptake rate

(mg COD-mg- I Xa-d- I),= acid fonners concentration (mg/L),= saturation constant (mg SCOD/L),= decay rate constant for acid formers (d- I),= decay rate constant for methane formers (d- I),= methane formers concentration (mg/L), and= theoretical COD equivalent of VSS

(mg COD/mg VSS)

Volatile Acids COD

dVA Q(VAo - VA) (vmaxaXaSJ--= +YAdt VL Ksa+S

- (vmaxm Xm VA J (7)Ksm + VA

where:VA =SBR VA COD concentration (mg COD/L),VAO = influent VA COD concentration (mg COD/L),YA =true yield of VA COD from substrate,Vmaxm = maximum specific VA uptake rate

(mg VA COD-mg- I Xm-d- I), andKsm = saturation constant (mg VA COD/L)

Acids Formers

dXa QXa (vmaxaXaSJ-=---+Y -KlXcit VL (l K.sa+S ta a

where: Ya =acid fonners yield factor.

Methane Formers(5)

(4)

(1)

(2)

(3)

KpPdP Q (Po -P)dt - VL

d(PVL) dVL dP--=P-+VL-

dt dt dt

where:PVLtQPo

1. Swine manure contains both particulate and solublesubstrates.

2. Particulates are converted to SCOD.

3. SCOD is converted to VA and acid formers.

4. VA are converted to methane and methane fonners.

5. pH, VA, NH3-N, and NH4+-N concentrations do nothave an effect on PAD in SBR process kinetics assupported by Masse (1995).

6. Soluble substrate utilization follows Monod kinetics.

7. Both acid former and methane former populationschange during the simulation.

8..Methane solubility in the liquid phase is negligible.The rate of methane leaving the SBR with the biogas isequal to that produced by the methane fonners.

9. Only one population of acetoclastic methanogens ispresent in the digesters.

10. Psychrophilic conditions (T=20oC) are maintained inSBRs.

= particulate COD concentration in the SBR (mgL),=SBR liquid phase volume (L)= time (d),= influent flow rate (L/d),=particulate COD concentration in influent (mg/L),

andrp =utilization rate of particulates (mg COD-L-I_d-I)

In an SBR during the fill phase, both the particulatesconcentration and liquid phase volume are functions of time.

It is known that:

dVL-=Q

dt

Substituting Eq. 3 into Eq. 2 gives:

d(PVL) dP--=PQ+VL-

dt dt

Substituting Eq. 4 into Eq. I and simplifying further yields:

CANADIAN AGRICULTURAL ENGINEERJING Vol. 39, No. l. January/February/March 1997 37

II

10 mixed liquor or supernatent sampling port11 gas outlet12 gas meter13 thermocouple14 feeder tube15 gas pump16 hydrogen gas monitor17 liquid pump18 dissolved hydrogen gas monitor

bed zones. Swine manure slurry was sampled immediatelybefore it was fed to the SBRs. The samples were analysed forpH, alkalinity, solids, VA, total Kjeldahl nitrogen (TKN),ammonia nitrogen (NH3 + NH4+), TCOD, and SCOD. Thebiogas production was monitored daily and its compositionwas analysed weekly.

Analytical techniquesSCOD was determined by analysing the supernatant of centrifuged slurry. The SCOD was determined according to themethod developed by Knechtel (1978). The pH, redox potential, alkalinity, total solids, total suspended solids, volatilesolids, volatile suspended solids, ammonia nitrogen (NH3 +NH4+), and TKN were determined using standard methods(APHA 1992). TKN and ammonia nitrogen were determinedusing a kjeltec auto-analyser model TECATOR 1030. The

1 300 mm diameter plexiglas digester2 slUdge bed zone (7.5 L)3 variable volume zone (28.0 L)4 head sp~ce zone (6.5 L)5 gas recirculation line6 blogas recirculation pump7 Influent line8 effluent line9 sludge sampling port,

also use for sluage wastage

Fig. 3. Schematic of laboratory scale SBRs used for test runs 5, 6 and 7.

(11 )


Experimental design and procedures have been previously reported(Masse 1995). Only a summary isgiven in this paper. Experimentswere carried out in 12 laboratoryscale bioreactors located in a controlled temperature room. All thetests were carried out at a temperature of 20°C.

Experimental equipment

Figure 3 is a schematic diagram ofthe bench scale SBRs and feedingsystem used in this study. Theplexiglass SBRs provided mixingby recirculating the biogas 10 minevery 30 min through an inlet located at the bottom of the digesters.Manure slurry was obtained fromstorage gutters under a partially slatted floor in a growing-finishing bamat a commercial swine operation.The manure was as old as four daysat the time of collection.

A mixed liquor sample of 100 mLwas withdrawn from each SBR atthe beginning of the experiment andonce a week during the experimentalrun. At the end of the test, after thesedimentation phase, additional 100mL samples were withdrawn fromthe supernatant and settled sludge

where: Vo = SBR initial volume (L).

where: QCH4

= methane production rate (g COD/d).

Mass balances for SCOD, VA, acid, and methane formers(Eqs. 6 to 9) were developed using Monod kinetics (Monod1949) for substrate removal. The mass balance for methaneproduction (Eq. 10) depends on the total conversion of VAless the conversion of VA used for the growth of the biomass.The mass balances for the react phase are similar to Eqs. 5 to10. The only difference is that the influent flow rate term isequal to zero. The Runge Kutta method· (Carnahan et al.1969) was used to evaluate theseequations.

The liquid zone volume changeswith time during the fill phase.Therefore, in the Eqs. 5 to 10 theliquid phase volume is determinedas:

Methane Production

38 MASSE and DROSTE

5.-----------------------,

60

25 30

Vmax. :::: 0.045Vmax", :::: 0.120

- - _. measured....... predicted

- - _. measured........ predicted

40 50

Vmax" ::::0.110Vmax", :::: 0.190

20

...... ".... ~'"

\'.'~"'"

...... ~"'" .

30TIme (d)

15Time (d)

20

10

..... / ........ ",

5

10

SBRs11 -12, Test 5

SBRs 5 - 6, Test 6, Cycle 1""

o 'o

OL-:-----_-- -.I

o

4.----------------------.~ 3o:Eg 2

~i 1

~

~ 4

-8 3gCD 2

i'0 1>

Constant Acidogens Methanogens

Vmaxi (mgemg-I.d- I) 0.4* 1.0*Ks (mg/L) 800* 200*Yi (mg/mg) 0.12* 0.05*kdi (d- I) 0.025* 0.025*Kp (d- I) 0.05** 0.05**

Table I: SBR operating conditions

Run Digester Mixing Fill React Numbernumber number phase phase cycle

(week) (week)

5 II -12 Yes 4 4 16 5 -6 No. 2 2 2

Table II: Initial values of biological kinetic constantsused in the simple model

2.5 r-------------------.......,

605550

SBRs 5 - 6, Test 6, Cycle 20.0 L-__--:..._---:~ ---I

~ 30 ~ 40 ~

TIme (d)

~ 2.0

~ 1.5UQJ

.2 1.0

i'0 0.5>

Fig. 4. Comparison between the experimentalmeasurement of volatile acids and the simplemodel prediction, Ksa =1500 mg/L,Ks~ = 2500 "!F/L, Kp = 0.04 d- l ,Kdl = 0.001 d .

RangeIncremental

Constant Acidogens Methanogens values

Vmaxi (mgemg-I.d- I) 0.04-0.80 0.04-1.4 0.01Ks (mg/L) 100-2500 50-3000 10.0Yi (mg/mg) 0.05-0.25 0.01-0.20 0.01kdi (d- I) 0.0005-0.04 0.0005-0.04 0.0001Kp (d- I) 0.01-0.08 0.01-0.08 0.005

* from Droste and Kennedy (1988)** from O'Rourke (1968)

Table III: Range of values considered for each kineticconstant in the grid analysis

VA concentrations were detennined by a Perkin Elmer gaschromatograph model 8310, that had a DB-FFAP high resolution column. The biogas composition was detennined byusing a Carle 400 AGC gas chromatograph.

Model validation

This section examines the adequacy of the model in predicting the dynamic behaviour of the PAD of swine manureslurry in SBR. Experimental data from digesters II and 12 intest run 5 (cycle length of 56 days) and digesters 5 and 6 intest run 6 cycle I and 2 (cycle length of 28 days) werecompared to the dynamic model prediction. These runs wereselected because they had the most comprehensive data set(Masse 1995). The organic loading for those ASBRs was1.63 g COD.L-I.d- I. The other operating conditions for theruns selected for simulation are given in Table I. The parameters used in this evaluation were VA, SCOD, and methaneflow rate.

Model kinetic constants in Table II were obtained from theliterature (Droste and Kennedy 1988; O'Rourke 1968). Thesekinetic constants were detennined for bioreactor operated

under different environments (pH, temperature, alkalinity,mixing level, etc.) and hydraulic flow regimes. The kineticconstants in this study were expected to be lower because theprocess took place at a lower temperature. A grid searcharound these values was used in the model prediction. Therange considered for ~ach kinetic constant is given in TableIII. The incremental value used for each biological kineticconstant during the simulation runs is also given in Table III.The error of estimate and percent error of estimate for eachParameter (VA, SCOD, and methane flow rate) were calculated according to Eqs. 12 and 13.

.... /"L (CalculatedValue; - ExperimentalEE= 'J N (12)

[I Experi~~ntalvaluei]

PEE = N * 100 (13)

where:EE =error of estimate,PEE =percent error of estimate (%),

CANADIAN AGRICULTURAL ENGINEERJING Vol. 39. No. I. January/February/March 1997 39

Vmax. =0.110VmaXm =0.190

16,.---------------------,

6050

vrnax. .. 0.110Vm8Xen .. 0.190

40

. - - - - measured.... · .. ·predlcted

30Time (d)

20

, ,, ,

::;/</.~>'-,<:

10

,, ,

SBRs 11 -12, Test 5_14

~ 12C 10

8 8

~ 6.g 4 .'U) 2

0L..------------------------1o

- 10 r---------------------.~ SBRs 11 -12, Test 5~ 8 ~o / '- ,'\ ',,'I: ....;/·::r/'·'v';;;J\·~:·:,,/· .~:.- -' '. -,,." \.,.:.....-..i 2 : I • - - - - measuredi ,: predicted

:::E OL....:...----------------------~o 10 20 30 40 50 60Time (d)

30

--,

25

Vmax. '" 0.045VmoXm =0.120

-'-

- - - - • measured···· .... predlcted

2015Time (d)

105

..... :/I,~ ....

,,

O"'-----------------------~o

SBRs 5 - 6, Test 6, Cycle 1 , _,~ 8 "

.... _ ~:-.~ "'...: ...:.~ .8 6o.!!! 4.a~

'0 2en

10,.----------------------,

30252015Time (d)

105

...._:0' 6,....---------------------.

.... SBRs 5 - 6, Test 6, Cycle 1 Vmax. '" 0.045c 5 I, . VmaXm ",0.120o :\

1$ 4 : \::I ' " I \ ..... ','", ,,'- .0\\e 3 ,/ \~/.. ',' \, \

~ 2 ..../-<>\ /\_j -----measured! 1 .'-' • -' ........ predicted1):::E 00

6055

Vmax. =0.115VmOXm =0.250

• - - - - measured.. ·· .... predlcted

5040 45Time (d)

3530

SBRs 5 - 6, Test 6, Cycle 2~70'6o(,) 5.!!!.c 4~

~ 3

2'------------+---------__---l25

8,----------------------,

6055

Vmax. .. 0.115VmaXm .. 0.250

5040 45Time (d)

3530

," '\ "...... ,'- --~ ,'\

" " '" ',.,.,.t:::./. .- .. .\/\::;"'~::~:~::::='. '. ~{'".,,

-8~-------------------.

~5 61$~e 4a.~ 2CI:li SBRs 5 - 6, Test 6, Cycle 2

== 025

Fig. 5. Comparison between the experimentalmeasurement of methane production and thesimple model prediction, Ksa =1500 mg/L,Ks~ = 2500 ~f/L, Kp = 0.04 d-1,Kdl = 0.001 d .

Fig. 6. Comparison between the experimentalmeasurement of soluble COD and the simplemodel prediction, Ksa = 1500 mg/L,Ks~ =2500 ~f/L, Kp =0.04 d-1,Kdl = 0.001 d .

predicted parameters were expected. In this study it was notpossible to determine the relative contributions to the errorsbetween the measured and predicted values due to the simplemodel limitations or sludge acclimation. Independent sets ofexperimental data would be required to clarify this.

Figures 4 to 6 compare the calculated and measured concentrations of VA, SCOD, and QCH

4as a function of time.

Table IV: Lowest PEE for the final values of kineticconstant used with the simple model

N =number of estimates, andi = day number

The best fit yield factors for the acidogens and methanogens were 0.1 and 0.05 mg/mg, respectively. The other bestfit kinetic constants are presented in Figs. 4, 5, and 6. Someof these kinetic constants are slightly different for each testrun because sludge acclimation was still taking place and theoperating strategies were different.

Quantitative sensitivity analysis has not been carried outto quantify the relative influence of each kinetic parameter onthe prediction accuracy. Observations during simulationclearly indicate that the maximum specific substrate utilization rates of the acids and methane formers had the largestinfluence on the model prediction. The second most influential group of parameters was the yield factors for theacidogens and methanogens.

Table IV gives the lowest PEE obtained for the kineticconstants that provided the best fit in each test run. The PEEvalues for VA, SCOD, and QCH

4are similar and within a

reasonable range. These differences between measured and

Parameter

VASCOD

QCH4

Test run 5digesters 11-12

373727

PEE

Test run 6digester 5-6

cycle 1

2012

30

Test run 6digester 5-6

cycle 2

233428

40 MASSE and DROSTE

These figures show that the simple model predicted reasonably well (PEE ranged between 12 and 37) the general trendin methane production as well as VA and SCOD accumulations during the fill and react phases. This model provides auseful tool to better understand the dynamics of PAD in SBRprocess operated under different operating conditions.

CONCLUSIONS

A dynamic model based on a simplified scheme for the PADof swine manure slurry in a SBR was developed and verified.Considering the overall complexity of this process, ongoingsludge acclimatization and different operating strategies, themodel was judged to be acceptable for predicting the accumulated VA and SCOD in the SBR as well as the methaneflow rates. The sets of kinetic constants that provided the bestfit for each experimental run were of the same order. Thismodel is a useful tool to gain better knowledge of: I) thedynamic interaction between the acid and methane formers;and 2) microorganism response to different operating strategies.

REFERENCES

APHA. 1992. Standard MetllOdfor the Examination ofWaterand Wastewater, 18th ed. Washington, DC: AmericanPublic Health Association.

Carnahan, B., H.A. Luther and J.O. Wilkes. 1969. AppliedNumerical Methods. Toronto, ON: J. Wiley and Sons.

Dague, R.R., C.E. Habben and S.R. Pidaparti. 1992. Initialstudies on the anaerobic sequencing batch reactor. WaterScience Technology 26(9-11 ):2429-2432.

Droste, R.L. and KJ. Kennedy. 1988. Dynamic anaerobicfixed film reactor model. Journal of EnvironmentalEngineering 114(3):606-620.

Gujer, W. and AJ.B. Zehnder. 1983. Conversion processesin anaerobic digestion. Water Science Technology15(8/9): I27-167.

Hill, D.T. and C.L. Barth. 1977. A dynamic model forsimulation of animal waste digestion. Journal of theWater Pollution Control Federation 49( I0):2119-2143.

Jones, R.M. 1989. Dynamic modelling of a high rateanaerobic wastewater treatment process: ProgressReport. Unpublished Report WTC-B/O-O 1-1989.Burlington, ON: Wastewater Technology Centre.

Jones, R.M. and E.R. Hall. 1989. Assessment of dynamicmodels for high rate anaerobic treatment process.Environmental Technology Letters 10:551-566.

Knechtel, J.R. 1978. A more economical method for thedetermination of chemical oxygen demand. Water andWaste Engineering 14(4):25-28.

Masse, D.I. 1995. Psychrophilic anaerobic digestion of swinemanure slurry in intermittently fed sequencing batchreactors. Ph. D. Thesis. Departement of CivilEngineering, University of Ottawa, Ottawa, ON.

Monod, J. 1949. Recherches sur la croissance des culturesbacteriennes. Hermann et de. Paris, 371-394.

O'Rouke, J.T. 1968. Kinetics of anaerobic waste treatment atreduced temperature. Ph.D. Thesis. Stanford University,Palo Alto, CA.

NOMENCLATURE

EE error of estimateF theoretical COD equivalent of VSS

(mg COD/mg VSS)i day numberKda decay rate constant for acid formers (d- I

)

Kdm decay rate constant for methane formers (d- I)

Kp first order solubilization rate (d- I)

K.w saturation constant (mg SCOD/L)K sm saturation constant (mg VA COD/L)N number of estimatesP particulate COD concentration in the SBR (mg/L)PEE percent error of estimate (%)Po particulate COD concentration in influent (mg/L)Q influent flow rate (LId)QCH4 methane production rate (g COD/d)rp utilization rate of particulates (mg CODeL-led- l)S SBR SCOD concentration (mg COD/L)So influent SCOD concentration (mg COD/L)t time (d)VA SBR VA COD concentration (mg COD/L)Vao influent VA COD concentration (mg COD/L)VL SBR liquid phase volume (L)Vmaxa maximum specific SCOD uptake rate

(mg CODemg-1 Xaed- I)

Vmaxm maximum specific VA u~take rate(mg VA CODemg- 1 Xmed- )

Vo SBR initial volume (L)Xa acid formers concentration(mg/L)Xm methane formers concentration (mg/L)Ya acid formers yield factorYA true yield of VA COD from substrateYm methane formers yield factor

CANADIAN AGRICULTURAL ENGINEERJING Vol. 39. No. I. January/February/March 1997 41

------------

Nutrient characterization of stored liquid hogmanure

AJ. CAMPBELL, J.A. MacLEOD and C. STEWART

Agriculture and Agri-Food Canada, Charlottetown Research Centre, P.O. Box 1210, Charlottetown, PEl, Canada CIA 7MB.Contribution No. 837. Received 24 October 1995; accepted 2/ January /997.

Campbell, A.J., MacLeod, J.A. and Stewart, C. 1997. Nutrientcharacterization of stored liquid hog manure. Can. Agric. Eng.39:043-048. Liquid swine manure from eight Prince Edward Islandconcrete storage facilities was sampled at three depths in the fall of1992. Samples were analysed for dry matter (OM), NH4+-N, P, K,Ca, Mg, S, Cu, Zn, Mn, Fe, and B. Ory matter content and mostnutrient concentrations declined as sampling depth increased exceptK and B which had the highest concentration at the mid depth.Samples taken from the mid depth generally represented the averagemanure nutrient concentrations in the storage. There were significantdifferences between farms on a volume basis for all parametersexcept P, Mg, Zn, Mn, and Fe and on a OM basis except for NH4+-N,Mg, and Mn. High manure Cu concentrations on some farms suggests that Cu toxicity problems could occur over time with highannual application rates on low pH soils. A regression analysis of thenutrient concentrations against OM indicated that P, Ca, Mg, S, Mn.and Fe could be predicted if DM content were known but OM contentwas not helpful in predicting NH4+-N, Zn, K, B, or Cu.

Ou Iisier de porc provenant de huit reservoirs de beton situes surdes fermes de I'lle du Prince Edouard, a ete echantillonne it troisprofondeurs durant l'automne 1992. Chaque echantillon a ete analyse afin de determiner son contenu en matiere seche, NH4+-N. P, K.Ca, Mg, S, Cu, Zn, Mn, Fe et B. Le contenu en matiere seche dememe que la plupart des concentrations des composantes ont diminue avec l'augmentation de la profondeur de I'echantillonnage.exception faite de K et B qui ont eu une concentration maximale itune profondeur moyenne. Les echantillons obtenus it profondeurmoyenne ont generalement ete representatifs de la compositionmoyenne des Iisiers entreposes. II y a eu, de fa~on significative, desdifferences entre les lisiers de chaque ferme pour tous les parametresctudies sur une base volumetrique, exception faite de P, Mg. Zn. Mnet Fe, et pour tous les parametres etudies sur une base de matiereseche, exception faite de NH4+-N, Mg et Mn. De hautes concentrations de Cu dans Ie lisier de certaines fermes soulignent la possibilitcdu developpement de problemes de toxicite au cuivre pouvantdecouler de I'epandage d'une grande quantite de lisier sur des solsayant un pH peu cleve. Une analyse de regression, sur les concentrations des composantes par rapport a la matiere seche, a indique queles concentrations de P, Ca, Mg, S, Mn et Fe peuvent etre estimeesen fonction de la concentration de matiere seche. La concentration dematiere seche ne permet cependant pas d'estimer les concentrationsde NH4+-N, Zn, K, B, et Cu.

INTRODUCTION

Swine manure chemical characteristics are generally expressed in tenns of maximum, minimum, average and, insome cases, standard deviation values (Tunney and Molloy1975; Beauchamp 1983; Fraser 1985; Collins et al. 1988; andKumar et al. 1990). These characteristics vary widely, oftenby a factor of 3 to 4 times. Animal diet, animal age, type of

storage, manure handling system and water content are possible sources of this variation (Beauchamp 1983; Tunney1980).

Less infonnation on the micro-nutrient content of swinemanures is available. Zublena et al. (1990) provided infonnation on concentrations of micro-nutrients for l.iquid hogmanure. Maschhoff and Muehling (1985) and the ASAE(1990) list amounts of micro-nutrients produced per animalunit but do not refer to the concentration of micro-nutrientsin the stored manure.

Overcash et al. (1983a) discussed the qualities of manureproduced by various livestock sectors, including both macroand micro-nutrients based on animal production units. Overcash et al. (1983b) discussed the land-limiting constituent(LLC) approach for maximum manure application rates. Thisapproach is based on crop nutrient removal rates, nutrientlosses, and the ability of the soil to assimilate nutrients overtime. Christie (1990) examined micro-nutrient build-up on a .clay loam soil, over a 19-year manure application experiment. He detennined that Cu levels were within 86% of thesoils allowable maximum after 19 years of manure applications of 200 m3.ha- l .yr- 1 on low pH soils.

With the introduction of storage structures of known geometry, systematic sampling can provide farmers andresearchers with a means of detennining manure nutrientconcentrations. This would provide an improved method formaking comparisons of manure handling systems and theirinfluence on manure nutrient concentrations. This system ofsystematic sampling could provide the farmer with a prespring estimate of the manure nutrient concentrations andamounts. The objectives of this study were to select suitableequipment, develop on fann procedures for the systematicsampling of stored liquid swine manure, and characterize thestored manure of a number of Prince Edward Island hog fanns.

MATERIALS AND METHODS

Manure storage facilities on eight hog farms were sampled inthe fall of 1992. A list of swine production facilities wasobtained from the PEl Hog Commodity Marketing Board andeight fanns were selected at random, at least one from eachcounty across the island. The nutrient concentrations of manure in various concrete storage structures werecharacterized by sampling manure at different depths downthe side of each storage structure. Each storage was sampledat three depths, shallow (0.5 m from the top of the storage),mid (the centre of the storage) and deep (0.5 m from the

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I. JanuarylFcbruary/March 1997 43

bottom of the storage). Samples were taken from near theedge of each storage and were replicated twice for a total ofsix samples per storage.

For the purposes of this study an ISCO 3700 AutomaticSampler Model Number 68-3700-001 (ISCO Corp., Lincoln,NE) was selected. The unit has built-in software which controls a peristaltic pump and allows for sample line purgingand measured volume sampling. It also produces an outputreport of the sample number, date, and time. The unit contains a 175 mm long strainer with 7 mm inlets. These unitsare designed for waste water sampling, are weather prQtected,and have an internal power supply. This sampler has 24, 500ml bottles, thus allowing four farms to be sampled into onetray.

A sectioned rod, marked in 0.5 m intervals, was constructed and used to measure the depth of the storage pit. Therod with the strainer and connecting tube was lowered intothe storage and samples taken at the predetermined depths.

Chemical analysis

For NH4+ analysis, liquid manure samples were agitated anda 50 mL subsample was collected and centrifuged. A I: 10dilution was prepared in duplicate for each centrifuged sample. The AOAC approved determination of NH4+ as outlinedin Industrial Method No. 825-87T of the Technicon Traacs800 Auto Analyzer manual (Bran+Luebbe Analyzing Technology Inc. 1987) was used to determine the NH4+ levelsdirectly from each prepared sample.

For dry matter (DM) determination, a 20 g subsample wascollected in a tared crucible from the agitated liquid manuresample. The crucibles were placed in a laboratory oven at 80°C. for 24 h. The oven-dried samples were placed in a desiccator until they had cooled to room temperature after whichthey were reweighed. The DM was then calculated by difference.

Other macro-nutrients and micro-nutrients were determined by Inductively Coupled Argon PlasmaSpectrophotometry (lCAP). Oven-dried samples were placedin a cold muffle furnace and ashed for 3.5 h at 500°C. Thecontents of the crucibles were cooled, treated with 5 mL of2N HCI, transferred to a 50 mL centrifuge tube, and dilutedto 50 mL with distilled water. Samples were agitated andcentrifuged and two ICAP analyses were performed on eachsample. ICAP results provided values for P, K, Ca, Mg, S,Cu, Zn, Mn, Fe, and B.

Analysis of variance was conducted to determine differences between storages (main plots) and sampling depth(sub-plots). Regression analysis was conducted to determinethe relationship between DM and the nutrient concentrationsof the manure, and testing whether the slope (b) or y-intercept(a) was significantly different from zero.

Sampling depths


Sampling depths

For the eight storage facilities, the DM, NH4+-N, S, and Bconcentrations on a volume basis did not vary significantlywith sampling depth (Table I). Phosphorus, Ca, Mg, Cu, Zn,Mn, and Fe concentrations declined as sampling depth increased, while K concentration increased. The sample takenfrom the mid depth corresponded well with the average concentration across the storage (Table I).

Table I: Summary of nutrient concentrations forsampling depths reported on a volume basis,averaged over the eight farms

SED Average

ILl

11.0

1.12

0.27

3.53

46.4

1.64

31.2

0.77

0.30

28.8

0.71

1.67 0.05**

1.08 ns

36.1 6.6*

25.4 ns

3.59 ns

23.9 4.9*

8.12 1.4**

0.28 ns

0.59 0.12*

8.46 1.9*

0.20 0.05*

0.56 0.11 *

Deep

10.9

1.16

1.72

12.2

29.7

46.1

0.30

3.69

0.26

0.71

29.2

0.76

Mid

1.52

14.0

1.11

12.6

31.9

3.31

0.94

0.32

57.0

39.9

0.36

0.86

Shallow

Barong/m3

Potassiumkg/m3

Dry matterkg/m3

Magnesiumkg/m3

Manganeseg/m3

Nutrient

Ammoniumkg/m3

Phos~horus

kg/m

Zincg/m3

Calciumkg/m3

Copperg/m

Sulphurkg/m3

(1)[c] =a + b *DM

where:[c] = nutrient concentration,a = Y-intercept,b =slope, andDM =dry matter.

The analysis was conducted using Genstat V ( 1987).

SED Standard error of the difference* Significant at the 5 % level** Significant at the I % levelShallow - Sampled at 0.5 m from the storage surfaceMid - Sampled at the centre depth of the storageDeep - Sampled at 0.5 m of the bottom of the storage

44 CAMPBELL. MacLEOD and STEWART

Table II: Summary of nutrients for the farms sampled in the study reported on a volume basis averaged over depth

Fanns in the surveyNutrient

Dry matterkg/m3

Ammoniumkg/m3

Phosphoruskg/m3

Potassiumkg/m3

Calciumkg/m3

Magnesiumkg/m3

Sulphurkg/m3

Zincg/m3

Manganeseg/m3

Barong/m3

18.3

2.68

0.42

1.08

0.48

0.17

0.18

3.21

22.3

6.52

32.8

0.49

2

19.8

1.45

0.63

0.69

0.59

0.23

0.18

3.80

36.3

9.97

34.2

0.48

3

22.0

2.77

0.66

1.47

0.50

0.26

0.26

18.0

19.3

9.55

42.1

1.12

4

28.7

3.50

0.93

1.77

0.67

0.31

0.27

18.2

35.1

12.5

49.9

1.08

5

31.8

4.28

0.81

2.00

0.87

0.25

0.25

3.75

34.0

12.0

54.4

2.03

6

34.2

4.55

0.93

2.04

0.80

0.33

0.30

18.9

38.7

13.2

63.5

1.11

7

40.3

5.33

0.96

2.97

0.78

0.26

0.58

13.7

26.2

14.2

49.8

2.36

8

53.5

5.67

0.80

1.90

1.14

0.36

0.74

8.6

19.3

14.1

45.7

0.67

SED

6.4**

0.30**

ns

0.10**

0.20*

ns

0.42**

2.6**

ns

ns

ns

0.10**

SED Standard error of the difference* Significant at the 5 % level** Significant at the 1 % level

Between storage variations

Across all farms in this study, concentrations of OM, NH4+N, K, Ca, S, Cu, and B on a volume basis varied significantly,while P, Mg, Zn, Mn, and Fe concentrations did not differ(Table II). Most nutrient concentrations on a OM basis variedsignificantly among farms except NH4+-N, Mg, and Mn (Table III). This suggests that differences in concentrations arenot simply dilution effects from varying amounts of wateruse, but are due to differences in feeding and other management practices.

A comparison of results from other studies indicates thatmacro-nutrients levels in Prince Edward Island stored swinemanure were similar to those of other regions in Canada, the

United States, and Europe. Average values from Prince Edward Island were very similar to those reported byBeauchamp (1983) in Ontario and Kumar et al. (1990) in theUnited States. The nutrient values from England reported byTunney and Molloy (1975) were in line with readings fromPrince Edward Island when one considered the more thantwo-fold difference in OM for the English manure.

Zublena et al. (1990) reported levels of secondary andmicro-nutrients in liquid swine manure effluent of NorthCarolina. When compared to Prince Edward Island levels,Ca, Mg, S, Fe, Mn, Zn, and Cu were all within an order ofmagnitude. Boron was markedly different at 8.2 g/m3 inNorth Carolina compared to 1.12 g/m3 in Prince Edward

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I, January/Fcbruary/March 1997 45

Table IV: Summary of regression analysis - nutrientsversus DM

Nutrients Constant Coefficient R Squared Variance

NH4+-Nkg/m3 1.597** 0.0672** 0.45 34.54**

Phosphoruskg/m3 0.0078 ns 0.0024** 0.68 89.72**

Potassiumkg/m3 0.0912** 0.0025** 0.29 18.02**

Calciumkg/m3 -0.0019 ns 0.0025** 0.85 233.13**

Magnesiumkg/m3 -0.0016 ns 0.001** 0.70 100.09**

Sulphurkg/m3 0.0001 ns 0.0011 ** 0.71 103.41 **

Copper3.95 ns 0.2448** 0.14 7.71**g/m-

Zincg/m3 2.98 ns 0.977** 0.52 46.4**

Maganeseg/m3 -0.14 ns 0.3906** 0.73 111.89**

Irong/m3 9.04 ns 1.295** 0.61 67.15**

Borong/m3 0.598** 0.018** 0.14 7.69**

ns Not significant* Significant at the 5% level** Significant at the I % level

nitrogen and Zn R2 values were in the mid-range whichwould suggest dry matter could not be used as a predictivetool. The R2 values (Table IV) for K, B, and Cu the R2 arelow. In the case of the soluble nutrients K and B this could bedue to the fraction of the nutrients in the liquid portion. In thecase of Cu, it is due to the wide variations between storagesas discussed earlier.

CONCLUSIONS

The waste water sampling equipment used in this study, anISCO 3700 Automatic Sampler Model Number 68-3700-001,proved to be a valuable tool for consultants or researcherssampling manure slurries in concrete storages. A number ofdepths and locations could be quickly sampled and processedto get an accurate measure of nutrient concentrations in themanure storage.

Average nutrient concentrations for the swine slurries inPrince Edward Island manure storage facilities were withinthe range of those measured in other parts of Canada, theUnited States, and Europe. There were significant differences in DM and nutrient concentrations between storages;therefore for best nutrient utilization from land application,individual storage sampling is required. Among storage facilities, differences were found on both a volumetric and adry matter basis, indicating that differences were due to morethan just a dilution factor.

Although concentration differences were detected amongdepths of liquid manure storage facilities, in most casessampling from the mid depth provided results that wererepresentative of the storage.

High Cu levels in manure on some farms suggests Cu soiltoxicity problems could occur if high annual application ratesare used for a long time.

A regression analysis showed a relationship between DMand P, Ca, Mg, S, Mn, and Fe suggesting that these nutrientscould perhaps be predicted from DM content. Ammoniumnitrogen concentrations could not reliably be predicted fromthe dry matter content.

ACKNOWLEDGEMENTS

The authors acknowledge the assistance of L. Kerry and D..Grimmett in the collecting and analysis of the manure samples.

REFERENCES

ASAE. 1990. ASAE D384.1: Manure production andcharacteristics. In ASAE Standards 1990, 37th ed,463-465. S1. Jospeh, MI: ASAE.

Beauchamp, E.G. 1983. Manure for crop production.Publication Number 83-039. Agdex Number 100/538.Toronto, ON: Ontario Ministry of Agriculture and Food.

Bran+Luebbe Analyzing Technology Inc. 1987. TechnicalPublication Number DSM-0005-00.4. Elnsford, NY:Bran+Luebbe Analyzing Technology Inc.

Christie, P. 1990. Accumulation of potentially toxic metalsin grassland from long-term slurry application. InFertilizer and the Environment. eds. R. Merckx, H.Vereecken and K. Vlassak, 24-130. Louvain, Belgium:Leuven University Press.

Collins, E.R. Jr., T.A. Dillaha, F.B. Givens and C.D.Eddleton. 1988. Farm manure management planning forVirginia. ASAE Paper No. 88-2048. St. Joseph, MI:ASAE.

Fraser, H. 1985. Manure characteristics. Publication Number85-109. Agdex Number 538. Toronto, ON: OntarioMinistry of Agriculture and Food.

Genstat, V. 1987. Reference Manual. Oxford, England:Clarendon Press.

Gupta, U.C. 1971. Boron and Molybdenum nutrition ofwheat, barley, and oats grown in Prince Edward Islandsoils. Canadian Journal ofSoil Science 51 :415-422.

Kumar, D., C.D. Eddleton, E.R. Collins and T.A. Dillaha.1990. Characterization of manure nutrient values in

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. J. January/February/March 1997 47

Virginia. ASAE Paper No. 90-2024. St. Joseph, MI:ASAE.

Maschhoff K.D. and A.J. Muehling. 1985. Using nursetankers to transport liquid manure to the field from a650-sow farrow-to-finish operation. In AgriculturalWaste Utilization and Management, 55-62. St. Joseph,MI: ASAE.

Overcash, M.R., F.J. Humenik and J.R. Miner. 1983a.Livestock Waste Management, Volume I. Boca Raton, FL:CRC Press Inc.

Overcash, M.R., Humenik F.J.,and Miner J.R., 1983b.Livestock Waste Management, Volume /I. Boca Raton,FL: CRC Press Inc.

48

Pond, W.O. and J.H. Maner. 1984. Swine Production andNutrition. Westport, CT: AVI Publishing Company, Inc.

Tunney, H. 1980. Fertilizer value of animal manures. Farmand Food Research 11(3):78-79.

Tunney. H. and S. Molloy. 1975. Variations between farmsin N. P, K, Mg and dry matter composition of cattle. pigand poultry manures. Irish Journal of AgriculturalResearch 14:71-79.

Zublena. J.P., J.C. Barker and J.W. Parker. 1990. Swinemanure as a fertilizer source. Soil facts. PublicationNumber AO-439-4. Raleigh, NC: The North CarolinaAgricultural Extension Service.

CAMPBEll. MaclEOD and STEWART

A grain storage information system forCanadian farmers and grain storage

managersI I 2 I S3D.O. MANN, D.S. JAVAS ,N.D.G. WHITE, W.E. MUIR and M.S. EVAN

JDepartment of Biosystems Engineering, 438 Engineering Building, University of Manitoba, Winnipeg, MR, Canada R3T5V6; 2Cereal Research Centre of Agriculture and Agri-Food Canada, 195 Dafoe Road, Winnipeg, MR, Canada R3T 2M9;and 3Department ofComputer Science, University ofManitoha, Winnipeg, MB, Canada R3T 2N2. Received 15 January 1996;accepted 15 Novemher 1996.

Mann, D.O., Jayas, D.S., White, N.D.G., Muir, W.E. and Evans,M.S. 1997. A grain storage information system for Canadianfarmers and grain storage managers. Can. Agric. Eng. 39:049056. Stored grain is susceptible to invasion by pests or infection byfungi, either of which has the potential to cause economic losses.Researchers have developed improved management techniques, butthese new strategies must be made available to farmers and grainstorage managers. This paper discusses the development, capabilities, and testing of an expert system, the Grain Storage InformationSystem (GSIS), developed to transfer information to western Canadian fanners and grain storage managers. Much of the informationused in the GSIS was gathered from the published literature. Thecurrent prototype is capable of teaching users why certain actions aremore likely to be successful than others. Teaching rather than simplydictating instructions should help the users gain confidence in theprogram, allowing it to be used to its full potential. The GSISfunctions well as a teaching tool, although it requires field testing.

Le grain entrepose est susceptible d'etre envahi par des animauxnuisibles ou des champignons, ce qui peut entrainer des perteseconomiques. Les chercheurs ont ameliore les techniques de gestiondu grain. Cependant, ces strategies nouvelles doivent etre mises a ladisposition des agricuiteurs et des gestionnaires d'entrepot de grains.Cet article parle du developpement, du potentiel et de I'essai d'unsysteme expert, Ie Systeme d'Information sur I'Entreposage desGrains, qui fut con~u pour diffuser des informations aupres desagriculteurs de I'Ouest canadien et des gestionnaires d'entrepot degrains. Une grande partie des informations contenues dans Iesysteme expert proviennent d'articles deja publies. Le prototypeactuel peut montrer aux utilisateurs pourquoi certaines decisions sontplus profitables que d'autres. La demonstration plutot que I'imposition d'instructions devrait aider les utilisateurs a avoir confiancedans Ie programme, ce qui permettra d'exploiter plus it fond sonpotentiel. Le systeme expert est un bon outil d'cnseignement, bienqu'i1 doive encore etre teste aupres des utilisateurs.

INTRODUCTION

Stored grain is susceptible to invasion by pests (e.g. storedproduct insects, mites, birds, or rodents) or infection byfungi, either of which has the potential to cause economiclosses. Researchers have spent much time and effort trying tounderstand the stored-grain ecosystem with the goal of identifying improved management techniques. Significant gainshave been made to address some of the problems faced byfarmers and grain storage managers (hereafter "farmer" wi II

refer to anyone who manages stored grain). This new information, however, must be made available to fanners.

Computers can be used to transfer information from researchers to farmers. One type of computer program whichhas been used successfully for transferring grain storageinformation is the expert system (Wilkin and Mumford 1994;Longstaff 1994; Flinn and Hagstrum 1990; Ndiaye and F1eurat-Lessard 1994). These expert systems have emphasizedthe control of stored-product insects because the presence ofstored-product insects is a serious problem for grain storagemanagers in countries such as Australia and the UnitedStates. In these countries, grain is often harvested hot anddry, so there is little chance for fungal deterioration. InCanada, however, grain is often harvested tough whichmeans that fungal deterioration is much more prevalent.Stored-product insects are not a major problem, but must notbe present in the grain at the time of sale. Consequently, themanagement of stored grain in Canada differs from othercountries, making the development of an expert system forCanadian grain storage managers a necessity.

An expert system is defined as "... an intelligent computerprogram that uses knowledge and inference procedures tosolve problems that are difficult enough to require significanthuman expertise for their solution" (Feigenbaum 1982).There are many problems that, despite an incomplete understanding, need to be solved. When a person works in the samearea, or on the same problem for several years, that personbecomes an expert. Due to valuable experience, the humanexpert is capable of repeatedly finding the "right" answer,even when there is no proven method for solving the problem.

Like a human expert, an expert system should find the"right" answer in the absence of a proven algorithmic solution. This can be accomplished by emulating thedecision-making ability of the human expert (Holt 1989). Inother words, the computer program should use the same "bestguesses" and "rules-of-thumb" as the human expert.

The objective of this research was to develop an expertsystem program, the Grain Storage Information System(GSIS), for western Canadian farmers. The GSIS encapsulates expertise from researchers working in the area of grain

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No I January/Fcbruary/March 1997 49

storage management at the Cereal Research Centre of Agriculture and Agri-Food Canada in Winnipeg and theDepartment of Biosystems Engineering, University of Manitoba. This paper discusses the development, capabilities, andtesting of the current prototype.

DEVELOPMENT OF THE GSIS

Sources of knowledge for the GSIS

There were several sources of knowledge employed in thedevelopment of the GSIS. Drs. D.S. Jayas and W.E. Muir,Professors, Biosystems Engineering, University of Manitobaand Dr. N.D.G. White, Research Scientist, Cereal ResearchCentre of Agriculture and Agri-Food Canada were the mainsources of expert knowledge or expertise for this program.The large base of scientific literature on grain storage andgrain storage management was also used as a resource. Inaddition, several computer simulation models have previously been developed which model processes within grainbulks. Because much expertise has gone into the development of these models, they were an excellent addition to theGSIS. A final source of knowledge was the farm backgroundof the first author. Although difficult to quantify, a practicalunderstanding of grain storage techniques did contribute tothe development of the GSIS.

Knowledge acquisition for the GSIS

Knowledge acquisition is defined as "the transfer and transformation of problem-solving expertise from someknowledge source to a program" (Buchanan and Shortliffe1984). For knowledge acquisition to work effectively, theperson gathering the knowledge (e.g. the knowledge engineer) must be familiar with the terminology used in theproblem domain. Often, the knowledge engineer (~) mu~t

spend some time learning about the problem domam. ThiSproject was no exception. Despite having a practical unde~

standing of grain storage management as the result of hisfarm background, the KE (D.D. Mann) did not have a scientific understanding equivalent to that of the domain experts(Drs. Jayas, Muir, and White). The first stage of knowledgeacquisition, therefore, consisted of the KE reading and reviewing the scientific literature available on grain storageand grain storage management.

While reviewing the literature, the KE found preliminaryanswers to many of his unanswered questions about grainstorage. These answers were distributed to the experts andwere discussed in a group session. This procedure workedquite well and was used a number of times. This procedureput less demand on the domain experts' time than the "traditional" interview process that is described in most textbookson the'development of expert systems and, therefore, may beconsidered by future KEs as an alternative to the "traditional"interview process.

Due to the success of the group validation procedure, theKE gathered much of the knowledge for the GSIS by rea?ingpreviously published scientific papers. When questionsarose, however, it was not always possible to meet as a groupso the KE usually met with one or more of the experts on anindividual basis. This eliminated the problems involved withscheduling group meetings and can be used by future KEs

50

when dealing with multiple domain experts.The KE must be aware that experts are not alike. Each

expert has different areas and levels of expertise which canresult in questions being answered differently by each of theexperts. The KE must decide how to combine the differentperspectives or whether one should be chosen over the others.

When gathering information from the literature, the KEmust also remember that biological variability can exist whenresearch is conducted in various regions around the world. Insome cases, more than one value was found for the samecharacteristic. In these cases, the choice of one number overanother was an arbitrary decision.

In addition to gathering knowledge from the literature, theKE also attended a university course on stored-product entomology taught by one of the domain experts (Dr. White).This worked well because the information was presented inan organized and structured format.

THE GRAIN STORAGE INFORMATION SYSTEM

Program philosophy

At the conceptual design stage, our philosophy was that theprogram would collect all of the necessary -information fromthe user and then tell the user what to do to protect the storedgrain. During the development of the program, however, ourphilosophy for the program changed. Rather than simplytelling the user what to do to protect the grain, the programshould teach the user why certain actions are more effectivethan others. We believe this revised philosophy has improvedthe program for two reasons. First, farmers have gainedvaluable experience storing grain in previous years. Theymay have made observations which they are unable to measure or explain. If the program is able to provide them with anexplanation, their understanding of the dynamics of a storedgrain ecosystem will increase, thus helping them beco~e

better grain storage managers. The second reason for thiSrevised philosophy is that some farmers do not have confidence in computers. For the program to earn theirconfidence, they need to use it extensively. This will nothappen by dictating instructions without any explanation.The program must be designed in such a way that the userfeels it is user friendly.

Purpose of the GSIS

The management of stored grain is a challenging task. Toremain in business, farmers cannot afford to make grainmanagement decisions that result in reduced grain quality.Farmers need assistance to deal with the rapid changes in theagricultural industry. Researchers have recognized this need.The main purpose of the GSIS, therefore, is to teach users thecurrent knowledge of grain storage to enable them to becomebetter grain storage managers.

How does the GSIS work?

The underlying chain of reasoning for the GSIS remains thesame from' consultation to consultation (Fig. 1) and is described below.

1. The initial grain conditions are input by the user.

2. The grain storage life is calculated by mathematical

MANN, JAYAS, WHITE, MUIR and EVANS

Infonnation Modules (infonnation modules can be accessed at any point)

t t~I Near-Ambient -..Reduce Drying

~Grain

MoistureContent

~[ Heated-Air InsectDrying I ~ Control

StomgeLifeMeasures

DetennineStomge ~ >

1Life Storage Length I

A- t ~ t Yes

I Tough

Tough Low RiskInput of Insects 1-

Initial ---. or Fungal~ Present? 1- Conclusions

Conditions Dry --. Infection

AA- I Dry " A

YesI

Iy I No

I

Detennine Storage Life IStorage ~ > I Calculate Risk of

Life Stomge Length~ Future Insect

tConductive Infestation

.~ Cooling

Reduce I

GrainTemperature '\

~I I Run I

~Forced

~I Aeration -"i AemtionSimulation I

Make Changes (changes can be made at various points throughout a consultation)

Fig. 1. The chain of reasoning used by the GSIS in a given consultation.

equations developed by several researchers (for wheat:Fraser 1979; for canola: Muir and Sinha 1986; for barley: Kreyger 1972). The safe storage life is described asthe period of time before germination drops by 5% orvisible mould appears. The storage life is based on thegrain temperature and moisture content.

3. The storage life is compared with the intended storageperiod.

4. If the storage life is less than the intended storageperiod, either the grain moisture content or grain temperature should be reduced. If the grain condition is notdry, appropriate drying actions are explored using anear-ambient drying simulation program. High temperature drying is also discussed as an option ifnear-ambient drying equipment is not available. If thegrain condition is dry, appropriate cooling actions areexplored using an aeration simulation program. Both ofthese simulation models are accessed by the GS IS in theGRAIN89 program (Huminicki et al. 1986).

5. If the storage life is greater than the intended storageperiod, the grain should be safe from fungal deterioration, however, insects must still be considered. Ifinsects are currently present, possible control measureswill be indicated. An insect identification module isincluded. Even if insects are not currently present, therisk of a future insect infestation is presented.

6. Changes can be made to the initial grain conditions sothat the user can consider possible effects of several

storage options.

7. Information screens are available throughout the consultation.

Capabilities of the GSIS

Initial grain conditions The first step in a consultation is thecollection of information about the grain to be stored and thestorage structure in which it will be stored. The GSIS asks forinformation on the following items: preferred measurementunits for temperature (e.g. °C or OF), distance, and volume;grain type; moisture content; grain temperature; grain damage (e.g. frost, mechanical, sprouting); foreign material (e.g.small particles, large particles); granary sanitation; presenceof insects; previous infestations; harvest date; intended selling date; bin dimensions; aeration equipment; heated-airdrying .equipment; and availability of a pneumatic grain conveyor.

A beneficial characteristic of the GSIS is that the user isable to answer "unknown" to most of the initial-grain-condition questions. If the user answers "unknown", the GSIS willassume a value for the parameter and offer a short explanation of why the particular assumption was made. When all thequestions have been answered, the GSIS displays a summaryscreen'to remind the user of the selections which were made.At this point, the user can change one or more answers. Theuser can also read a brief explanation of how each parameteraffects the storage of the grain.

Calculation of storage life The storage life of grain is the

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No 1 JanuaryjFebruary/March 1997 51

Drying considerations Grain can be dried using heated-airor near-ambient air. The choice of one type of drying over theother depends on several factors. The first consideration mustbe given to the equipment which the fanner owns. If there is

canola: (Muir and Sinha 1986)

W < 11%, T= S to 2SoCe=exp(6.224 - 0.302 W -0.069 n (3)

W ~ 11 %, T =S to 2SoCe=exp(S.278 - 0.206 W -0.063 n (4)

barley: (Kreyger 1972)

W =12 to 16%, T =S to 2SoC

1m = 2.79 + 0.0417 exp(S.124 + (39.6 -0.8107 n[l/(W -12) - 0.031S exp 0.OS79 n} (S)

Cooling considerations If the grain is dry and the storagelife is still less than the intended storage period, the grainshould be cooled. Some cooling of the grain near the wall andtop surface occurs when the outside air turns cold. Unfortunately, this conductive cooling may not be sufficient to coolan entire grain bulk because grain bulks have low thennaldiffusivities (Muir et al. 1989).

Yaciuk et al. (197S) ran a series of computer simulationsto determine the length of time required to cool a bin of wanngrain in Winnipeg below 20°C without aeration. Their resultsindicated that grain stored in large bins (>4 m diameter)should be cooled by some external means. The data fromYaciuk et al. (197S) were used in the GSIS to predict anapproximate cooling time if aeration was not used.

Farmers often cool their grain by aeration because conductive cooling is slow when grain is stored in large bins.Aeration refers to the process of forcing ambient air [1-2L.s-t.m-3] through the bulk of grain by a fan. As long as this

no limitation due to lack of equipment, consideration must begiven to the time constraints on drying. A well-designedheated-air drying system provides a rate of drying that canaccommodate the rate of harvesting grain, allowing the grainto be dried in a short period of time. Near-ambient drying,however, requires a longer period of time. It is important thatthe entire grain bulk be dried before any significant amountof grain spoils. If the grain ·will not be dried before spoilagestarts, then it may be advisable to use heated-air drying.

When neither equipment nor time causes any constraints,the selection should be based on cost and convenience. Nearambient drying is usually more convenient because itrequires little labour. Near-ambient drying can be less expensive if only the energy cost is considered. When theoverdrying penalty is also included, however, it is difficult topredict which type of drying will be less expensive.

If the GSIS detennines that the grain should be dried andif the bin is equipped with a near-ambient drying system, theuser will have an opportunity to run the near-ambient dryingsimulation model. The GSIS selects an appropriate airflowbased on the recommended minimum airflow requirementsfor Manitoba using a fully-perforated floor and a level grainsurface (Friesen and Huminicki 1987). The user can viewthese recommended airflow charts and change the value ifdesired. The drying simulation program which was previously validated by Sanderson et al. (1989) predicts dryingbased on 10 selected years of historic weather data for Winnipeg. For each simulation year, the GSIS determineswhether the grain has dried before November IS. If thesimulation model predicts the grain will be dried by November IS in an average year, the GSIS will recommendnear-ambient drying. Otherwise, the GSIS will recommendheated-air drying.

The drying simulation model as included in the GSIS isvalid only for grain stored in circular steel granaries equippedwith a fully-perforated floor and located near Winnipeg. Ifthe bin is either filled to the peak or if a partially-perforatedfloor is used, the results of the drying simulation will not beaccurate. The program can easily be used for other westernCanadian locations with the addition of their weather filesand minor modifications.

(1)

(2)

where:WTe

predicted length of time for which the grain will remain ingood condition. Researchers have used different criteria fordeciding when the end of the safe storage period occurs.Fraser (1979) assumed that the safe storage life would be thenumber of days before gennination drops by S% or visiblemould appears on the grain. Kreyger (1972) detennined thesafe storage life as the time to the appearance of visiblemould in barley.

The storage life equations used in the GSIS are:

wheat: (Fraser 1979)

W = 12 to 19%, T= S to 2SoCe=exp(6.234 - 0.212 W -0.OS3 n

W =19 to 24%, T= S to 2SoCe= exp(4.129 -0.0997 W -0.OS7 n

=moisture content (% wet mass basis),=grain temperature (oC),=storage life before gennination drops by S%

or visible mould occurs (d), and1m =time for mould to appear (d).

These equations were developed for a limited range oftemperatures. In the absence of other equations, however, theGSIS allows the user to input temperatures outside of thisrange. Extrapolation outside the given range will introducesome error to the approximation.

The storage life of grain depends mainly on the graintemperature and grain moisture content. This implies that thestorage life of a bulk of grain will be altered if either thetemperature or moisture content is changed. Theoretically, itshould be possible to lengthen the storage life of grain bydrying, cooling, or both.

The predicted storage life is a very important componentof the GSIS. The need to dry (or cool) the grain occurs whenthe predicted storage life is shorter than the intended storageperiod. The storage life equations assume constant temperatures and moisture contents throughout the storage period. Ifeither parameter changes, the prediction may no longer beaccurate. The storage life equations provide a good starting

.point. The storage life equations, however, do not considerdamage caused by insects.

52 MANN. JAYAS. WHITE. MUIR and EVANS

operation occurs when the ambicnt air tcmperature is lowerthan the average grain temperature, cooling will take place.

Whcn it is determined that the grain should be cooled andthe storage bin is equipped with aeration equipmcnt. theGSIS uscr can run an acration simulation model. The simulation runs with an airflow of 1.5 L.s- 1.m-3

. The initial graintcmpcraturc is assumed constant throughout Ihe bin. Thetcmperatures within the bin are simulated from thc specifiedharvest date until November 15 for three years of historicwca,her da'a (1965. 1968. 1969). When the simulations SlOp.

Ihe temperature I 111 below the surface at the centre of the binis shown to the user for each of the three separatc simulationyears. This location was selectcd because it is the 1110st likelylocation of the maximum temperature in fall and winter (Muirc, al. 1989).

Insect control considerations Even when grain is safe fromfungal deterioration. stored-produci insects can still invadethe grain. If insects arc present, the GSIS will suggest somepossible actions to kill the insects because it is illegal inCanada to knowingly sell grain infested with live grain-feeding insects. From the list of all possible actions. the mostprobable ones are identified. The GSIS provides informationall the various actions. explaining how they work. but thedecision of which action to select is left to thc uscr.

The most appropriate control mClhod depcnds on Ihe Iypeof insects that are present. The GSIS has an insect identifica-

tion 1110dule which helps the user idcntify the inseci species.The insect identification screens contain a picture and information about the species (Fig. 2). Thc GSIS also cmphasizesthe difference betwecn grain and fungus feeding species.

f{isk offuturc insect infestations Even when insects arc nOIcurrcntly prcscnt in thc grain samplc. there is no guaranteethai the grain will remain free of insccts. The likelihood of aninsect infestation depends on mall)' factors. These factorsinteract with one another. making it difficult to dctermine theovcrall risk of a future infestation. The risk of a future insectinfestation is calculatcd by:

1'/= I F"",F.,.FMOFsOFLPFsl'Fc;sFpFIIDFIJ I x 100(6)

where:PI = potential infestation factor.Fmc = moiSlure content faCIOr.F-,. = grain tempcrature faCIOI'.Fi\lD= mcchanical damage factor.FSJ) = sprouting damage factor.FLI~ = largc dockagc particles factor.Fsp = small dockage particles factor.Fcs = granary sanilation faCial'.Fp = previous insect infestation faclor.FI/D = harvcst date factor. andF/) = bin size factor.

== INSECT INFORMATION MODULEPick Insect Species

2mm

20. <IO·C

40·95% R.H.

33 ·C

70· SOli. RH.

reddish·brown

flat, reclsngular

1.5· 2.5 rom long

Yes

The germ ofcereals; will feed on some fungi..

RUSlY GRAIN BEETLECtypio/estesfenugineus

OPTIMUM CONDITIONS:

RANGE FOR DEVELOPMENT:

COLOR:

SHAPE:

SIZE:

FLYING:

FOODS:

~'...y:y*",'~~<"\JC

, .•. '; ' •••.. I ~".'

. ~Ti-~·,,-!..:t~ \

I 3 mm IItis the most conunon and serious pest of stored grain on farms and elevators in western Canada. ~en grain isharvested we.nn, large populations can build up quickly, causing grain heating and spoilage. Insect-infested grain islikely to cake, to become moldy and musty, to sprout and to undergo loss in gennination and in milling and bakingquality. The rusty grain beetle is very cold-hardy. It can survive short exposures of -25 ce, although prolongedexposure to.5 ce will kill them. Typical damage to a grain kernel can be recognized by the presence of e. distinctburrowing hole in the genn area made by the emerging adult. Large populations can generate enough heat andmoisture to create hot spots in bulk grain in cold weather under favorable conditions. This species tends to movedownward in bins. It will also move towards e.reas of high moisture content or high carbon dioxide concentration.

Fig. 2. An example of an insect identification StTccn from the GSIS.

CANADIAN AGRiCULTURAL ENGINEERiNG Vol. :W. No I January/FclJru:u)'/March II)lJ7 53

Each factor is given a value less than or equal to one(Table I). The value assigned to each factor depends on theinitial grain conditions specified by the user. In an ideal case,each factor would have a value of one. This would yield a PIof 100 corresponding to a negligible risk of a future insectinfestation. When some of the factors are less than one, thePI decreases below 100. The lower the number, the greaterthe risk of a future insect infestation. A PI less than 10indicates a high risk. A PI between 10 and 25 indicates anintermediate risk and values greater than 25 indicate a lowrisk of a future insect infestation. The form of Eq. 6 waschosen because when two factors interact with each other theeffect is multiplicative not additive.

Information screens The GSIS is designed to be an information system. Throughout a consultation, the user is presentedwith an abundance of grain storage information. Many of thedisplay screens contain pieces of information that do not

Table I: Insect infestation factors for various grainconditions

Parameter

Fmc

FT

FSD

Fsp

FGS

Fp

Parameter value Numerical factor

very dry 1.0dry 0.9

tough 0.7damp 0.5wet 0.6

cold 1.0cool 1.0wann 0.7

hot 0.5

yes 0.7no 1.0

yes 0.9no 1.0

significant numbers 1.0none 1.0

significant numbers 0.8none 1.0

yes 1.0no 0.6

yes 0.9no 1.0

early 0.9normal 0.9

late 1.0

small 0.9intermediate 0.7

large 0.5

contribute to the functioning of the GSIS, but do provide theuser with additional help (Fig. 3). This is an excellent way totransfer research findings to the farmer.

In addition to the information contained in the main program, the GSIS also contains four separate InformationModules. These modules give the user information on insectidentification, grain sampling, granary sanitation, and othergrain storage topics. Only the insect identification module isreferred to during a consultation. A user does not have to readthese modules, but they have been included for those whowant to learn more about grain storage management.

Conclusions of the GSIS Once the end of the consultationis reached, the GSIS briefly summarizes the most importantinformation that was given throughout the consultation. TheGSIS does not tell the user what to do. Rather, the situationis explained and some possible courses of action are discussed. The user must then decide what will be done.

Comparisons Throughout the consultation, the user is ableto make changes to the initial grain conditions and availablestorage equipment. This allows the user to explore differentoptions and compare the effects of different actions. Observing the effects of these changes is a good way to learn aboutgrain storage management.

TESTING OF THE GSIS

Types of testing

The testing of an expert system consists of two parts: validation and verification. Validation refers to the process ofdetermining whether a chain of correct inferences leads to thecorrect conclusion. Verification refers to the process of determining whether the information contained within the expertsystem is correct. In other words, validation is concernedwith building the right product, whereas verification is concerned with building the product right (Giarratano and Riley1994).

Validation may be viewed as taking an overall view of theprogram. Is the program being built according to the conceptual design? Does the program meet the needs of the intendedusers? Questions such as these must be answered to validatethe program.

Verification requires a more detailed inspection of theprogram. Is the information within the program correct? Arethe recommendations correct? Verification is necessary because it ensures that the information provided to the user bythe program is the same as would be given by a human expert.

Testing criteria

The GSIS was designed to be an educational tool for farmersin Western Canada. A prime objective of the testing procedure was to determine the adequacy of the program as aneducational tool. To be effective, the GSIS should facilitatelearning and be easy to use. Information should be presentedin a logical manner so that it is understandable and considered reasonable by farmers. Presenting the information as itwould be in a scientific paper may not be suitable. A secondimportant function of testing was to verify the accuracy of theinformation contained in the program.

54 MANN. JAYAS. WHITE. MUIR and EVANS

224283271Days until Dry

r=.aJConsultation Information Modules Make Changes

n

Dry Bofure "'in1or Stop Da10 yes yes yes yes yes

Bin Average Moisture Content 14. 14.3 14.3 13.g 14.3

Based on the results of the simulation model (assuming a level-filled bin), thedrying reconunendation is as follows: near ambient drying

How was the bin nUed?

Results of one field test indicate that bins filled to the peak can require a muchlonger drying time. If this bin was filled to the peak, you might expect the dryingtime to be as much as 1.8 times that which is shown above.

For further e"l'lanation, view thetopic 'Drying Considerations' fromthe 'Generallnforrnation' menuabove.

level·fille d peaked +

Fig. 3. A typical display screen from the GSIS depicling Ihe exira information that is helpl'ul to the user, though nolvital to the functioning of the program.

Validation

To validate the GSIS. the adequacy of the program as aneducational 1001 was based on the user-friendliness of theprogram and irs ability 10 convey infomullion. A class ofengineering students enrolled in a grain slOrage course wereasked 10 use the GSIS 10 answer a sample problem afterwhich they were asked to complele a questionnaire on the use01' the GSIS. Only lout of 18 survey respondents describedthe GSIS as being "very difricult·· 10 use.

On a further question. 15 out of 17 survey respondentsindicated Ihey had learned something new about grain S(Qf

i.lge during Iheir consultation. Twelve out of the 17 also statedIhat Ihe GSIS has potential as a leaching tool for universityslUdents and fanners.

For the sample problem. the students were required toassume a grain temperature to complete the consultation.Because the lemperatures selected by the students covered awide range. the recommended storage management procedures also covered a wide range. This sensitivity to graintemperature is the result of the srorage life equations and not'he GSIS. In this way. the GSIS can be used '0 illus"a,e 'heimportance of carefully detennining grain temperatures.

Verification

To verify the GSIS, it was necessary to ensure that theinformation contained in the program was accurate. The veri-

fication has been limited and unstructured. The three domainexperts involved in the project have used the current versionand are satisfied with the infomlation contained within theGSIS. These expens were also involved throughout the development. so they are familiar with the program.

Much of the infonnation used in the GSIS was gatheredfrom previously published scientific papers. To ensure thatthe information within the program describes actual grainstorage situations. structured field testing of the GSIS isrequired. but is cost prohibitive.

SUMMARY AND CONCLUSIONS

The GSIS is designed to be a source of grain storage information. Rather than simply informing the user how to protectgrain from spoilage. the program teaches the user why certainactions arc more effective than others. If the GSIS functionswell as a teaching tool. it will earn the confidence of lheu!)crs.

Based on responses to a survey comple,ed by 17 GSISusers who were engineering sludeIlls. the program does function well as a Icaching tool. The survey indicated Ihat theprogram is casy 10 use and provides useful information. Thedomain experts are satisfied with lhe accuracy of the informalion contained in the GSIS. although field testing IS

required.

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No I January!fcbru<LI)'IM:tr(;h 1997 55

ACKNOWLEDGEMENTS

The authors thank the Canadian Energy Mines and ResourcesResearch and Development Department for funding thisstudy.

REFERENCES

Buchanan, B.G. and E.H. Shortliffe. 1984. Rule-BasedExpert Systems. The MYCIN Experiments of the StanfordHeuristic Programming Project. Don Mills, ON:Addison-Wesley Publishing Company.

Feigenbaum, E.A. 1982. Knowledge Engineering in the1980's. Department of Computer Science, StanfordUniversity, Stanford, CA.

Flinn, P.W. and D.W. Hagstrum. 1990. Stored GrainAdvisor: A knowledge-based system for management ofinsect pests of stored grain. Artificial IntelligenceApplications 4:44-52.

Fraser, B.M. 1979. Solar grain drying in Canada: asimulation study. Unpublished M.Sc. thesis. Departmentof Agricultural Engineering, University of Manitoba,Winnipeg, MB.

Friesen, O.H. and D.N. Huminicki. 1987. Grain aeration andunheated air drying. Manitoba Agriculture PublicationNo. Agdex 732-1. Manitoba Agriculture, Winnipeg, MB.

Giarratano, J. and G. Riley. 1994. Expert Systems. Principlesand Programs. Boston, MA: PWS Publishing Company.

Holt, D.A. 1989. The growing potential of expert systems inagriculture. In Knowledge Engineering in Agriculture,eds. J.R. Barrett and D.D. Jones, I-II. St. Joseph, MI:ASAE.

Huminicki, D.N., C.I. Kitson, O.H. Friesen and W.E. Muir.1986. Computerized design of ventilation systems forstored grain. ASAE Paper NCR-86-604. St. Joseph, MI:ASAE.

Kreyger, J. 1972. Drying and storing grains, seeds and pulsesin temperate climates. Publication 205. Institute forStorage and Processing Agricultural Produce,Wageningen, The Netherlands.

56

Longstaff, B.C. 1994. Decision support systems for pestmanagement in grain stores. In Proceedings of the 6thInternational Working Conference of Stored-ProductProtection, Volume 2, eds. E. Highley, EJ. Wright, HJ.Banks and B.R. Champ, 940-945. Canberra, Australia.

Muir, W.E., D.S. Jayas, M.G. Britton, R.N. Sinha and N.D.G.White. 1989. Interdisciplinary Grain Storage Research atthe University of Manitoba and Agriculture Canada.Powder Handling Proceedings 3(1): 281-295.

Muir, W.E. and R.N. Sinha. 1986. Theoretical rates of flowof air at near-ambient conditions required to dry rapeseed.Canadian Agricultural Engineering 28:45-49.

Ndiaye, A. and F. Fleurat-Lessard. 1994. Research on anexpert system for appropriate management of the qualityof stored grain for food and feed processing. '94International Symposium & Exhibition on NewApproaches in the Production of Food Stuffs andIntermediate Productsfrom Cereal Grains and Oil Seeds.Beijing, China.

Sanderson, D.B., W.E. Muir, R.N. Sinha, D. Tuma and C.IKitson. 1989. Evaluation of a model of drying anddeterioration of stored wheat at near-ambient conditions.Journal of Agricultural Engineering Research42:219-233.

Wilkin, D.R. and J.D. Mumford. 1994. Decision supportsystems for integrated management of storedcommodities. In Proceedings of the 6th InternationalWorking Conference of Stored-Product Protection,Volume 2, eds. E. Highley, EJ. Wright, H.J. Banks, andB.R. Champ, 879-883. Canberra, Australia.

Yaciuk, G., W.E. Muir and R.N. Sinha. 1975. A simulationmodel of temperatures in stored grain. Journal ofAgricultural Engineering Research 20:245-258.

MANN. JAYAS. WHITE. MUIR and EVANS

Modelling of microwave drying of grapesT.N. TULASIDAS 1, C. RATTI2 and G.S.V. RAGHAVAN3*

IDepartment of Dairy Engineering. University of Agricultural Sciences. Hebbal Campus. Bangalore. India. 560024;2Departement des sols etde ghlie agroalimemaire. FSAA. Universite Laval. Quebec. QC. Canada GlK 7P4; and 3Departmentof Agricultural and Biosystems Engineering. Macdonald Campus of McGill University. Ste-Anne de Bellevue. QC. CanadaH9X 3V9. Received 13 February 1995; accepted 16 October 1996.

Tulasidas, T.N., Ratti, C. and Raghavan, G.S.V. 1996. Modelling ofmicrowave drying of grapes. Can. Agric. Eng. 39:057-067.Thompson seedless grapes were dried into raisins under combinedconvective and microwave drying in a single mode cavity type ofapplicator at 2450 MHz with varying operating parameters viz., airtemperature, microwave power density, and superficial air velocity.A number of experiments designed to detennine the relationshipsbetween grape physical and electrical properties and grape moisturecontent were executed. These relationships were then used in thedevelopment of a model of microwave drying. The model was basedon the continuum approach and included the tenns describing internal heat generation, convective heat transfer, and evaporation. Theinternal resistance to moisture transfer was described by a Fick's typediffusion equation. The effective diffusivity parameter included thecombined effect of microwave power and liquid diffusion on moisture transport. A moving co-ordinate system based on non-variatedry solids was adopted to handle the problem of shrinkage. Themodel was solved by the "Method of Lines". The simulated resultswere tested with the experimental data and good agreement wasfound proving the validity of the procedures. Keywords: dielectricdrying, electric field strength, method of lines, grapes, moving coordinates, raisins, shrinkage, single mode cavity.

Des raisins frais Thompson sans-pepin, ont ete seches sous microonde et air chaud combines, a I'aide d'un applicateur de mode depropagation unique operant a2450 MHz avec des variables d'operation telles la temperature de I'air, la puissance et la vitesse de l'air ala surface des raisins. Un certain nombre de tests ont ete cffectuesatin de detenniner les relations qui existent entre les proprietesphysiques et electriques et la teneur en eau des raisins. Une fois cesrelations etablies, eUes ont ete utilisees dans Ie developpement d'unmodele du sechage par micro-ondes. Vue Ie type d'infonnationobtenu, un modele continu a ete choisi tenant compte de la chaleurinterne, du transfert de chaleur par convection et de I'evaporation. Laresistance interne au transfert massique a ete decrite par la loi de Fick.Un systeme de coordonnees mobiles a ete adopte afin d'exprimer Ieretrecissement du raisin. Le modele a ete resolu grace ala methodedes lignes. Les resultats du modele ont ete valides, de falt0n satisfaisante, avec les resultats experimentaux. Mots cle: seehage,micro-onde, raisins, raisins sees, modele.

INTRODUCTION

Dielectric heating with microwaves (MW) has proven togreatly reduce the drying time of many agricultural commodities. With proper control of the drying parameters(dielectric field strength, airflow rate, and inlet air temperature) a dried product with quality attributes equivalent tothose of convection-dried material can be obtained (Ottenand 51. John 1988; Gunasekaran 1990; Caldas 1992;Shivhare et al. 1992; Raghavan et aI. 1993). However, mostof the work on MW heating and drying of agricultural com-

modities has been done on low-moisture material such asgrains. Grapes are a high-moisture commodity and the feasibility of MW drying of grapes to obtain raisins has beenreported (Tulasidas et al. 1993). The development of a commercial-scale microwave drying process to produce highquality raisins could make a significant contribution to theraisin industry. The objective of this study was to develop amathematical model to describe microwave drying of grapesfor the purposes of scale-up and process simulation.

The volumetric nature of MW heating leads to a rapidtransfer of energy throughout the body of the wet material.Unlike pure convective heating during which the temperaturewithin the grape is limited to that of the flowing air, MWheating can induce an internal temperature greater than thatof the ventilating air and thus accentuate internal heat andmoisture transport (Perkin 1990). However, in the microwave drying situation, forced convection is also necessarysince it is the process by which water vapour driven from theinterior of the grape to its surface is then carried away fromthe immediate drying environment. To prevent condensationof the driven moisture on the surface of the grape, it isnecessary that the convective airflow be heated so as toimprove its moisture-carrying capacity. Since both the microwave and convective energies are initiated simultaneously,there is a short period which one may refer to as combinedconvective and MW heating. Clearly, this period lasts untilthe surface temperature of the grape reaches that of theconvecting air. Thereafter, one may refer simply to microwave heating, since no convective heat transfer towards theinterior of the body is possible and the convective energy isunderstood to be used entirely for moisture transfer awayfrom the surface.

A model describing all major aspects of this "combined"drying process is formulated. It is aimed at describing themoisture content of the grape at any given time during heating and is more fundamentally aimed at taking into accountshrinkage (as it occurs drastically in grapes) as well as thechanges in physical and dielectric properties that occurthroughout the process due to changes in moisture contentand temperature. The model is based on the continuum approach and incl udes terms describing internal heatgeneration, convective heat transfer, and evaporation (Turnerand Jolly 1991; Perkin 1990; Ptasznik et aI. 1990).

The MW energy source term was developed from the dataobtained in the experiments on the dielectric properties ofgrapes, during which measurements were made at many com-

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No.1 January/February/March 1997 57

(I)

In this moving coordinate system, the equation for masstransfer becomes:

where: Deff= effective moisture diffusivity parameter (function of temperature, shrinkage, water content, and geometryof the product). Equation 4 is similar to the equation fordrying of a sphere without shrinkage (Eq. I), except it iswritten in terms of moving coordinates. The boundary conditions are:

(7)

(4)

(5)

(6)

(8)

aX*/aA = 0

11w =-(CslRo) Deffax*/aA =kG Ap

=moving coordinate (dimensionless),=radius at t =0 (m),=vivo (local shrinkage coefficient),= volume at time t (m3), and= initial volume at t =0 (m3

).

A= 0

A= I

Mr Loo

6L2 (~~ J--I ex --D t

Moo - - 0 ~;, [~;, + L (L - I)] P R2 eff

where:

where:ARos*vvo

Initial condition:t =0 X* =Xo

where:11w = mass flux at the surface (kgem-2es- I

),

kG = convective mass transfer coefficient at the surface(k -2 -I kP -I)gem eS e a ,

p = partial vapour pressure (kPa), andXo = initial moisture content at t = 0 (kg/kg dry mass).

Effective moisture diffusivity Crank (1975) derived ananalytical solution for the general diffusion equation in anon-shrinking sphere (Eq. I) for the boundary conditionssimilar to those discussed above. The solution for the totalamount of diffusing substance, moisture (Mr), entering orleaving the elemental sphere is:

However, since grapes shrink noticeably during drying,the mass transfer equation has been modified to (Tulasidas1994):

ax* _y ax* __I ...L1.(D' ,2 ac) (2)at a,. - C s ,.z a,. eff a,.where: y = velocity of shrinkage (m/h).

Equation 2 represents drying in a shrinking particle and isdifferent from the equations normally used in the literature

. *due to the shnnkage term -y ax lar. To handle the problemof shrinkage, a moving coordinate that follows the shrinkageduring drying was defined (Tulasidas 1994):

dA=~(~Jdr (3)s* A2 R~

where:X* = local moisture content (kg/kg dry mass),t = time (h),C = concentration of water in the solid (kgjm3

),

Cs = concentration of solid (kg dry solid/m ),D'eff = effective moisture diffusivity (m2/h), andr = radial distance in spherical coordinates (m).

THEORETICAL DEVELOPMENT

Mass transfer equation

The model formulation begins with the mass transfer equation for a fixed coordinate system as given by Crank (1975):

ax* _i....L1. [D' ,2 ac]at - cs ,2 ar eff a,.

Assumptions

The following assumptions were made in formulating themodel:

a) Each particle (grape berry) is assumed to be spherical,homogeneous, and isotropic with initially uniformtemperature (TmO) and moisture distribution (Xo).

b) An air stream of constant temperature (T~) and relative humidity (RH) passes over the particle at aconstant velocity (V~).

c) The particle is exposed to MW radiation at 2450 MHzand the absorption is assumed to be uniform throughout the body.

d) The dielectric properties depend only on the moisturecontent and temperature of the material.

e) Moisture migration is one-dimensional (radial), fromthe centre towards the surface where the evaporationis occurring.

f) The vapour pressure of water in the solid is describedby sorption equilibrium expressions.

g) Shrinkage of the particle during drying is uniform andproportional to the average moisture content.

h) The combined effect of diffusion and internal pressureon water migration within the particle is expressedthrough a Fick's type equation with an effective diffusivity parameter.

i) The temperature difference between the centre andthe surface is small when compared to the temperaturedifference between the surface and the bulk gasstream.

binations of moisture content and temperature, and measure-.ments of electric field strength. The influence of thegeometry of the product on internal moisture diffusion isaccounted for by considering a moving coordinate systembased on invariate dry solids following shrinkage using theconcept presented by Crapiste et al. (1988a, 1988b) andincorporating the relationships between volume and moisturecontent deduced during the shrinkage studies. The modelpredicts the drying history of the grape and is also capable ofsimulating temperature and moisture profiles.

58 TULASIDAS. RATTI and RAGHAVAN

The total amount of diffusing substance related to that atan infinite time can be expressed as:

where: Xe = equilibrium moisture content (kg/kg dry mass).

Equation 10 then becomes:

:a~;, I3t [l3t:~~L - I)] exp ( - ~! D,1]1J (12)

The resulting slope (5 = BI 2 DeJtlR 2 ) of the plot of In[(X-Xe)/(Xo-Xe)] against time t, yields the value of the effective diffusivity, Def{- The effect of X e in Eq. 12 is onlyimportant at very low water contents. For example, in atypical drying situation where the initial moisture content isXo =3.5 kg/kg, the final moisture content is X =0.18 kglkg,the drying air temperature and relative humidity of Tg = 50°Cand RH= 12%, respectively, the difference between X/Xo and(X-Xe)/(XO-Xe) is only 7.3%. This is the maximum differencethat could be expected since the difference (error) is smalIerat higher values of X. Hence the plot of In X/Xo would sufficeto estimate Deff.

The effective diffusivity is a function of water content,shrinkage, and temperature. In the case of microwave drying,the effect of internal pressure is included in the effectivediffusivity parameter since it is based on data from wholedrying runs. A rough estimate of Deff is needed to model theactual process; the procedure adopted is outlined below.

Estimation of effective diffusivity parameter The combined effect of MW power and air temperature on moisturediffusivity during MW drying was accounted for by an effective diffusivity parameter evaluated from experimental data.The MW drying data was used for estimating this parameter.The slope of the plot In (X/Xo) versus t (5 = effective diffusivity parameter) was modelled as a function of microwavepower density (P) and air temperature (Tg) using the empirical model:

InS=PCI(C2+~:) (13)

where: CI,C2,C3 = coefficients which were evaluated usingexperimental data. The basis for choosing this specific model(Eq. 13) in preference to a curve fitting through non-linearregression analysis is described below. The analysis of theMW drying data revealed that the functionality of 5 and Tgwas of the type:

(16)

(14)

- 0.898+ 1.609

- 542.04

Air velocity

Vt: = 2 mls

- 1.707+0.192- 63.94

Air velocity

VK = I mlsCoefficients

of Eq. 13

In 5 = (a + b/Tg )

where:Pal' = average microwave power (W/m3

),

cO =permittivity of free space (F/m),E" =dielectric loss factor,

Heat transfer equation

Sensible heat gain by the material is described by the governing energy equation. A simplified form of the general energyequation applicable to high frequency heating and drying isgiven by Ptasznik et al. (1990):

ciTd(=EI+E2+ E3 (15)

where:E 1 = internal energy generation due to MW heating,E2 =heat transfer at the surface, andE3 = energy loss in phase change due to evaporation at

the mass transfer surface.

The procedure adopted to evaluate these terms is given in thefolIowing sections.

Internal energy generation (El) Average MW power dissipated in a material can be derived from Maxwell'sequations and on the assumption of a constant electric fieldin the material, the equation for Pm' is (Metaxas and Meredith1983):

where: a,b =constants. Equation 14 is of the Arrehenius typeand has been widely used in the literature to explain thedependency of moisture diffusivity on temperature. Usingthis relationship, a and b values were obtained for each powerdensity used in the present drying studies. When b wasplotted against P, a power relationship was obtained (b =p'\The intercept a also was found to have a power relationshipwith P. The slope 5 was therefore assumed to have a powerlaw relationship with P. On this basis, the empirical equation(Eq. 13) was used to describe the dependency of 5 on Tg andP.

The Levenverg and Marquardt method was used to determine the values of coefficients CI, C2, and C3 in Eq. 13 thatminimized the sum of squares of differences (SigmaPlot1992). These coefficients were obtained from data of experiments with different MW drying conditions of airtemperature (Tg) varying from 30 to 60°C and MW powerdensities on dry matter basis (P) varying from 0.5 to 1.5 W/gand at two levels of air velocities of 1.0 and 2.0 m/s (Tulasidas 1994) and are presented in Table I.

Table I: Values of Cl, C2 and C3 of Eq. 13

(I I)

(10)

(9)

8 is of

Mt Xo-X

Moo =Xo-Xe

L = mass transfer Biot number (a function of airvelocity), and

~Il =nth root of Eq. 9.

~Il cot ~Il + L -1 =0

For large drying times, only the first term of Eq.significance (Crank 1975), thus Eq. 8 simplifies to:

Mt 6 L2

(~T JMoo = 1- ~T [~T+L (L-I)] exp - R2 Deff!

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No.1 January/Fcbruary/March 1997 59

(17)

£rms = electric field strength (V1m), and(J) =2 1t f, frequency (Hz).

As the MW energy is absorbed, the material's temperaturerises at a rate depending on a number of distinct parameters.Substituting £() = 8.8xlO-12 F/m for free space and /=2450MHz in Eq. 16, the power required to raise the temperatureof a mass of a material over a time period is given by(Metaxas and Meredith 1983):

0.13622 e" £~msPav=-----

PmCm

where:pm = density of grape (kg/m3

), andCm = specific heat of grape (Jekg-1eK-1).

Equation 17 was adopted to calculate the energy absorbed bythe material due to microwave radiation. A model to predictthe loss factor of grapes as a combined function of moisturecontent and temperature was used (Tulasidas et al. 1994). Useof Eq. 17 needed £rms for grapes and the procedure that wasfollowed in its establishment is explained under a separatesection.

ms = mass of dry solid (kg), andX = average moisture content (kg/kg dry mass).

The equations that are used to represent the equilibriumproperties e.g. water activity (aw), and heat of sorption (Hs)

of the grapes are (Ratti et al. 1989a):

In aw =- dl Xdz + ql exp (- q2 X) XQ3 1n Pwo (22)

where the constants dl, d2, q I, q2, and q3 were determined bynon-linear regression of experimental sorption data of grapes(Maroulis et al. 1988). The results are given in Table II.

The vapour pressure as well as the heat of vaporization ofpure water was calculated using the correlations of the subroutine PSYCHRP (Ratti et al. 1989b).

Application of mass transfer and energy equations (Eqs. 3and 15) needed determination of several properties of grapeslike shrinkage, electric field strength, density, and specificheat corresponding to a given drying condition. These weredetermined experimentally. The procedures followed in thequantification of these properties and the results obtained arediscussed in the following sections.

The convective heat transfer term (E2) The convectiveTable II: Values of coefficients of Eqs. 22 and 23heat transfer term £2 is given by:

£2- hG A (Tm - Tg)

(18)d) d2 q) q2 q3

Pmcm v- 0.936 11.98 - 0.9680.015 0.023

(21 )

where:hG = convective heat transfer coefficient (Wem-2 eK- 1

),

A = area (m2)Tm =surface temperature of sphere (fruit) (oC), andTg = temperature of air (oC).

When a single sphere is heated or cooled by forced convection, Eq. 19 can be used to predict the average heat transfercoefficient (hG) for Reynolds Number (Re) of I to 70,000 andPrandtl Number (Pr) of 0.6 to 400 (Geankoplis 1993):

Nu = 2.0 + 0.60 Reo.s Pr l!3 (19)

For a sphere subjected to heating or cooling in air, Eq. 19reduces to:

(20)

where:Kg = thermal conductivity of air (Wem-1eK- 1

), andd =diameter of grape (m):

The properties of air were obtained from the literature(Geankoplis 1993) and regression equations were developedto obtain the values at any desired temperature.

The evaporation term (E3) The evaporation term is givenby (Ptasznik et al. 1990; Geankoplis 1993 ):

-AHsms dX

Cm Pm V dt

where:AHs = heat of sorption (J/kg),

60

QUANTI~ICATIONOF PROPERTIES OF GRAPES

Shrinkage of grapes during drying

The approach taken was to determine the relationship between volume and moisture content of grapes. Therelationship was studied for two methods of drying: convective drying and combined convective and MW drying.Possible effects of initial size on shrinkage were also considered.

Materials and methods for shrinkage studies The volumeof grape berries/raisins was determined by the liquid displacement technique using toluene at room temperature(Mohsenin 1986; Saravacos and Raouzeos 1986). A sampleof ten grapes/raisins was used to determine the average volume. The volume of fresh grapes was taken as Vocorresponding to the grape berry at its initial moisture content Xo (kg/kg dry mass) before drying. The grape berries,pretreated with 2% ethyloleate in 0.5% NaOH (Riva and Peri1986; Tulasidas et al. 1993), were dried by two methods: a)convective drying at 60°C, air flow 2.0 m/s and b) MWdrying, MW power density 0.5 W/g on dry basis, air temperature 50°C and air flow 2.0 m/s. At different stages of drying,samples of 1oberries/raisins of uniform size and appearancewere taken out and their average volume was determinedalong with their moisture content. Shrinkage studies wereconducted using grape berries of different initial size, but ofuniform size within a batch. The initial sizes corresponded tothree initial volumes: 3.1 mL (3.0±0.1 g), 4.0 mL (4.0±0.1 g)

TULASIDAS. RATTI and RAGHAVAN

(25)

0.0 L--l.._..L-----l'--.--L--...L--L-....L--"-----L-.........--I

0.0 0.2 0.4 0.6 0.8 1.0

XIX 0

Fig. 1. Volume reduction as a function of moisturecontent in convective and microwave dryingof grapes (air temperature 50°C, powerdensity 0.5 WIg dry mass basis and airvelocity 2.0 m/s).

(26)

o

o

observedmodel

o

1.0

0.8

00.6

;>............;>

0.4 0

0.2

is given by (Metaxas and Meredith, 1983):

1/2_[pm em (Tm - Tmo) / t]Erms -

0.556 x 10-10le"eff

Equation 26 was used to calculate the value of Erms based onexperimental measurements of the temperature rise (Tm-TmO)in the particle (grape/raisin) for a known interval of time (t).

Methodology for determination of Erms E rms for grapes ofdifferent moisture content corresponding to various powerdensities were experimentally determined. The size of thesample consisted of 24 particles. The sample was spread in asingle layer on a tray and subjected to MW radiation. Theseconditions were identical to the conditions which existedduring the MW drying experiments. Since the determinationof MW heating is the point of interest, these experimentswere conducted with no air flow through the cavity (dryingchamber). A fluoroptic sensor inserted at the centre of agrape indicated the temperature rise in the particle due toMW radiation. In the absence of air flow the temperature risein the particle was rapid. The experiments were concluded assoon as mass loss was noticed as Eq. 26 is valid for a constantmass. Continuous monitoring of the experiment through thedata acquisition system permitted observations to be recorded every 5 s. E rms was calculated using Eq. 26 for eachstep interval and the average of all the steps represented theactual Erms for a particular run.

These experiments were conducted with grapes of different moisture content, viz., 0.80 (fresh grapes), 0.60, 0.40,0.25, and 0.15 kglkg, wet basis. The experiments were conducted at three power densities, viz., 0.5, 1.0, and 1.5 W/g ondry matter basis. These power densities correspond to theones used in drying studies. All experiments were replicatedthree times and the averaged values of Erms were used for

~ = 0.147 + 0.839.!....vo Xo

For a known initial size of grape (volume), Eq. 25 predictsthe actual size (volume) of grape at a given moisture content.Equation 25 was retained for the modelling work.

Electric field strength

The electric field strength is the prime parameter in MWheating; it is the intangible link between the electromagneticenergy and the material to be treated. The complex interaction of electro-physical properties of the material underelectromagnetic radiation makes prediction of the electricfield extremely difficult. One way to determine the absolutevalue of the electric field strength is through calorimetry and

and 4.88 mL (5.0±0.1 g). Moisture content was determinedby the vacuum oven method at 70°C (Boland 1984). All theexperiments were conducted on a single purchased lot ofThompson seedless grapes.

Results and discussion of shrinkage studies The relationship between vivo and X/Xo was represented by a linearequation (Lozano et al. 1983):

~=A +B (.!..) (24)vo Xo

where:vo =initial volume of the grape (single berry) (mL), andXo =initial moisture content of grape (corresponding to

the initial volume yo).

The constants A and B were obtained through linear regression analysis for different size groups and drying methods.

Shrinkage in convective drying Equation 24 was applied ondata of duplicated convective drying experiments. The linearrelationship between vivo and X/Xo was significant and described the data well as evidenced by the high R2 values forall the sizes considered. The coefficients (B) of the regression(Le. the shrinkage rates) were found to be independent of theinitial size of the grapes (Duncan's Test, 0.05 level). The datawere therefore pooled and a new regression equation generated for convective drying conditions. The procedure yieldedan equation with intercept A = 0.159 and slope B = 0.854(R2=0.975). This equation compares well with that of Masiand Riva (1988) which was derived from similar work on sixvarieties of grapes (A = 0.167 and B = 0.833).

Shrinkage in combined convective and MW drying The sameprocedure as described above was performed to study grapeshrinkage under MW drying conditions. There were threereplicates in the case of size 4.0 mL and two in each of theremaining two treatments. Again, the coefficients (B) werefound not significantly different among sizes (Duncan's Test,0.05 level).

Finally, the coefficients (B) from both sets (i.e. convectiveand microwave) were tested for differences due to dryingmethod; none were found (PROC GLM: SAS 1989). Therefore, the final regression equation based on all the results(Fig. 1) and applicable to shrinkage under convective andmicrowave drying was generated:

CANADIAN AGRICULTURAL ENGINEERING Vol. 39, No. I JanuarylFebruarylMarch 1997 61

300 L.-J-_L.-.-J-_L..-.-J-_L-.......1..._.L..--.L..---J0.0 0.2 0.4 0.6 0.8

M (kglkg wet mass)

Fig. 2. Electric field strength of grapes as a function ofmoisture content (power density 0.5 Wig drymass basis) at 2450 MHz.

further analysis. The dielectric loss factor (e") as a functionof moisture and temperature was obtained using the proposedmodel for grapes (Tulasidas et al. 1994).

Results of E rms The Erms (Eq. 26) for grapes for a givenpower density is shown in Fig. 2. A linear relationship between Erms and moisture content (M, kg/kg wet mass) wasobserved for each power densi~y:

Erms = Al + 81 M (27)

The linear models were found to be significant at the 0.05level (Steel and Torrie 1980). The resulting values of A I and81 obtained from linear regressions are given in Table III.The Erms values generally increased with decrease of moisture content in grapes. This behaviour is obvious as a highfield strength would be required for a low moisture materialso as to enable coupling of the electromagnetic field. Theinfluence of moisture dependent loss factor is also responsible for change in electric field strength. The values of Ermsobserved in this study appear to be small and this is due to avery low incident power applied on the material. As thepower density increased, the Erms values also increased (Table III). As reported earlier, the temperature measurementwas on only one particle. The power absorption was assumedto be uniform in all the particles. However, if the powerabsorption is not uniform in all the particles, then the errorassociated with the computation of Erms could be larger. Thisfactor was verified by conducting similar calorimetric studieswith crushed grapes in the cavity. Grapes were crushed intojuice (without addition of water) and the liquid material, ofequal mass corresponding to whole grapes in the previouscase, was used. Erms values determined for crushed grapescompared well with the corresponding Erms for whole grapes,thus validating the procedure used.

-~

-a)

Ew

600

500

400

o observedmodel

Table III: Values ofAl and BI of Eq. 27 for predictionE rms

Power density Intercept Slope R2

(WIg) (AI) (81)

0.5 570.57 - 292.41 0.991.0 757.99 - 434.39 0.931.5 784.74 - 303.88 0.87

Density (pm)

The density of grapes as a function of moisture content wasexperimentally determined. The volume of grapes at a givenmoisture content was determined with the pycnometermethod using toluene at ambient temperature (Mohsenin1986). Mass per unit volume was expressed as density at agiven moisture content. A linear relationship between densityand moisture content was observed for raisins and the linearregression yielded the equation:

Pm = 1408 - 428 M (28)

where: pm =density (kg/m3). An R2 value of 0.95 indicated

a high degree of correlation for the linear relationship.

Specific heat (em)

Specific heat of raisins was computed using the method ofdistribution. The major components of the material, viz.,water, total sugars, proteins, and total ash (minerals) wereconsidered to compute the specific heat value using Eq. 29(Buffler 1993):

Cm =4190 M + 1420 C + 950 A + 1780 P (29)

where:Cm = specific heat (Jekg-1eK-1),C =total sugars (fraction),P =proteins (nitrogen) (fraction), andA =ash content (fraction).

The composition of grapes/raisins was obtained from published work (Miller 1963). During drying, the water contentof raisins reduces and is associated with a proportionateincrease in the other constituents. Cm values were computedfor 10 different moisture contents by applying Eq. 29. Therelationship between em and M was found to be linear with ahigh degree of correlation (R2 =0.999). The resulting relationship is:

cm=1510 + 2671 M (30)

The resulting values of Cm with the use of Eq. 30 was in goodagreement with the reported values of Cm for fresh grapes of0.81 kg/kg (wet basis) and raisins of 0.15 kg/kg (Mohsenin1986); however, it was not possible to verify the results forthe other moisture content for the lack of reported values inthe literature. The variation of Cm with temperature is assumed to be negligible and Eq. 30 was used to obtain Cm

values at a given moisture content.

Mass transfer coefficients

The convective mass transfer coefficient (kG) at the evaporat-

62 TULASlDAS. RATTI and RAGHAVAN

Equation 31 was applied, as shown by Saravacos and Marousis (1988), to estimate kG using convective drying data ofgrapes (without microwave power) obtained in the new microwave drying apparatus. The term ~P was evaluated usingpsychrometric properties of air corresponding to drying conditions. The mass transfer coefficient values obtained at airvelocity of 2 mls were transformed to represent equivalentvalues at air velocity of 1 m/s. The transformation was:

kG. Imls =[kG, 2mls 1'-/2] (32)

The '-/2 in Eq. 32 comes from the power relationship betweenmass transfer coefficient and the Reynolds Number (heat andmass transfer analogy, Bird et al. 1960).

The calculated values of kG corresponding to air velocitiesof 2.0 mls and 1 mls at different air temperatures calculatedaccording to Eqs. 31 and 32 are tabulated in Table IV.

Table IV: Overall mass transfer coefficients ofpretreated grapes under convective dryingconditions

ing surface is given by (Geankoplis 1993):

11= kG f!.p (31 )

point, Le., the spatial derivatives were approximated interms of the neighbouring node points using finite difference approximations. This resulted in the partialderivative of the dependent variable at each node pointbecoming a total derivative.

c. The ODE for the nodes adjacent to the boundary nodeswere modified to satisfy the boundary conditions of theproblem (Eqs. 5 and 6).

d. The conditions of IVPDE at t =0 provided the initialconditions of the system of coupled first order ODE's.

e. The nodal equations were integrated forward in timeusing an ordinary differential equation solver.

A central second order finite difference scheme was applied to the spatial derivatives of the mass transfer equation.The surface boundary condition was represented by backward discretization while the centre point condition by aforward discretization, both having second order truncationerror. Table V lists the finite difference formula used in thisstudy.

Table V: Second order approximation for first andsecond derivatives

NUMERICAL PROCEDURE

.Air temperature (oC)(Tg )

30405060

Vg = 2.0 mls

0.18000.18000.16300.1285

Vg = 1.0 mls

0.12730.12730.11550.0940

Forward difference

Backward difference

Central difference

ax* 4 xi+ I - xi+ 2 - 3xi - 2-= +O(A)aA 2M

ax* xl- 2 - 4xl- I + 3xl - 2-= +O(A}aA 2M

ax* xi+ I - xi-I - .,aA 2M +O(A-}

A system of ordinary differential equations was obtainedas a result of the discretization:

(34)

(33)

(35)

i = 2 to NN:

dXj DI [* (1 +iJ * (1 -i)~a;-= RO~2 Xi+ I -i- -2XI -X;_I -i- ~

i = I:

i = NN + I:

* 1[* * 2~ kG ~ P JXNN+ 1='3 4XNN-XNN-1 - (ps)DNN+ I Ro

Thus the system of equations was transformed into a vector of 4*(N+ 1) temporal derivatives. The resulting vector wassolved simultaneously with the energy equation using a software based on the Gear method (Gear 1971; Hindmarsh1972, 1974). This method is appropriate for the integration ofstiff systems of ordinary differential equations. The simulation model coded in FORTRAN was run for different input

Simulation

The model discussed in the preceding sections (Eqs. 4 and15) was used to simulate the MW drying process with theinitial and boundary conditions given as Eqs. 5, 6, and 7. Themodel is highly nonlinear with a system of two simultaneousinitial value differential equations (moisture and temperature). The mass transfer equation (Eq. 4) is a second orderpartial differential equation with time and space as independent variables, whereas the heat transfer equation is anordinary partial differential equation with time as the independent variable (Eq. 15).

The "Method of Lines" (MOL) was used to solve thesystem of differential equations. This numerical method hasbeen applied successfully to simulate the drying process(Ratti 1991; Ratti and Mujumdar 1993). The initial valuepartial differential equation (lVPDE), i.e, the mass transferequation (Eq. 4), was converted into a system of ordinarydifferential equations (ODE) by using finite difference approximations for the spatial derivatives. As a result, an ODEin time was written for each interior spatial node point. Theprocedure for implementing the MOL is (Raghavan 1994):

a. The mass coordinate A of the sphere was divided intoN spherical slices of thickness ~A.

b. The IVPDE (Eq. 4) was discretized at each interior node

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I January/Fcbruary/Mnrch 1997 63

(36)

o

10842o

4

A Tg= 30 0e3 0 Tg =40

0e-en[J Tg= sooeen

asE v Tg= 60

0e~

2 simulation"Cen.:.::Q.:.::->< 1

6

t (h)Fig. 3. Predicted average moisture content of grapes by

the numerical model compared to theexperimental observations (power density0.5 WIg dry mass basis and air velocity 2.0 m/s).

Discussion of model performance

The simulation model (Eqs. 4 and 15) was run using thespecified initial and boundary conditions (Eqs. 5, 6, and 7).The simulated results for a given set of MW drying conditions were compared with the experimental data. The averagemoisture content as a function of time predicted by simulation is compared with the actual observed values in Fig. 3where the effect of air temperature on drying kinetics ofgrapes is illustrated. The effects of microwave power densityand air velocity on drying kinetics is shown in Figs. 4 and 5,respectively. The predictions of the model are also presentedin these figures. As can be seen, the model accurately simulates the microwave drying of grapes for many differentoperating conditions. The contrast with the behaviour ofconventional convective drying, an increase in air velocity inMW drying led to an increase in drying time (Le., a lowerdrying rate; Fig. 5). Higher air velocities resulted in fastercooling thus lowering the material temperature and hencelowering the drying rates.

x-xMR = e =exp (- kr')

Xo-Xe

where: k, n =parameters for a given drying condition whichwere obtained by fitting the experimental data. This modelwas found to fit the MW drying data of grapes adequatelywell (Tulasidas et al. 1993). The parameters k and n wereobtained for each of the drying conditions in this study usingnonlinear regression analysis (SAS 1989).

NT

b) Values of the average moisture ratios (MR) calculatedthrough Eq. 36 were used to estimate the accuracy ofpredictions by the numerical model (Eqs. 3 and 15).The relative errors of approximations (Eqs. 37 and 38)were used to perform the comparisons (Weres andJayas 1994):

IMR (tt) - MRexp (tt) I (37)et (tt)

MRexp (tt)

drying conditions and the simulated results were comparedwith the experimental data.

Model validation procedure

The solution obtained from the simulation was comparedwith the experimental data from a 4x3x2 factorial dryingexperiment on grapes aimed at evaluating the quality anddrying time withrespect to the process parameters (Tg, P, andVg). This experiment is fully described in Tulasidas (1994).The comparison was done according to the procedure presented by Weres and Jayas (1994) which they used tovalidate a numerical structural model for thin layer drying ofcom. They first represented their experimental data by thebest fitting of a number of empirical equations. The accuracyof the numerical structural model was then determined bycomparing data predicted by the structural model to theexperimental data as represented by the best fitted empiricalequation. The procedure of Weres and Jayas (1994) wasapplied to test the accuracy of simulation results of MWdrying of grapes as described below.

a) The experimental data were fitted by the modified logarithmic model (Page 1949):

where:NT[Ilt:: I

(38)

= set of instants bisecting two consecutive timeintervals (h),

=number of time intervals= local relative error of approximation at time, and= global relative error of approximation throughout

the drying period.

The accuracy of the numerical model was tested by adopting the procedure of Weres and Jayas (1994) as explainedearlier. The average moisture ratio predicted by simulationwas compared to the experimental data represented by themodified logarithmic equation (Eq. 36). The global relativeerror of the numerical model for the entire drying period werealso calculated. The details of these MW drying treatmentscan be found in Tulasidas (1994).

The modified logarithmic equation (Eq. 36) accuratelypredicted the MW drying of grapes (R2 ~alues > 0.99). Th.efitness of the modified logarithmic equatIOn and the numencal model with the observed data is presented in Fig. 6 forMW drying of grapes at an air velocity of 2.0 mIs, MW powerdensity of 0.5 WIg and air temperature of 30°C. Figure 7shows the plot of the relative errors of approximation (Eq.

64 TULASIDAS. RATTI and RAGHAVAN

0.0o 2 4 6 8 10

t (h)Fig. 6. Comparison of observed moisture ratio in

microwave drying of grapes with the predictionsby numerical model and Page's equation (airtemperature 30°C, power density 0.5 WIg drymass basis and air velocity 2 m/s).4

4 1.0

p= 0.5 Wig0 observed

0 Eq.36b. P= 1.0 Wig 0.83 o P=1.5W/g numerical model-(I) - simulation -(I) 41

as ><0.6E •c

~2 ><:s;:

~41

>< 0.4

~ ><- ->< 10.2

°0 2 3 4 5 6t (h)

Fig. 4. Effect of microwave power density on drying ofgrapes (air temperature 50°C and air velocity2.0 m/s).

SUMMARY

The numerical procedure predicted the MW drying behaviourof grapes adequately. The simulation which accounted for

0l.-J1...-1---.1.--L--L--L--L--L-...l--...L.-.L.-.L-'--'~~---'----'

023 4 5 6 789t (h)

Fig. 5. Effect of air velocity on microwave drying ofgrapes (air temperature 40°C and power density0.5 WIg dry mass basis).

6-~0-..0....CD 4CD>;:as'iia:

2

o 0 2 4 6 8 10 12

t (h)Fig. 7. The relative error of approximation for

experimental data represented by Page'sequation (Eq. 36) compared with the numericalmodel.

8

shrinkage based on the moving coordinate system was foundto describe the diffusion equation quite well as demonstratedby a good fitting with the experimental data. The model alsoaccounted for the changing physical and dielectric propertieswith the change in moisture content. The proposed numericalmodel is based on a semi-theoretical approach and thereforelends itself for adaptation to scale-up purposes. Given thenature of a material with its physical and electro-magneticproperties, it is possible to predict the MW drying behaviourby applying the numerical procedures elaborated. It is to benoted that this procedure is not restricted in its application to

o Vg = 1 mls

o Vg =2 mls3

>< 1

38) of experimental data with the numerical model for thiscase. For this case, the global relative error (e,F), calculatedthrough Eq. 38, was found to be 3.23% and can be considered as an excellent fit. However, the global error in four outof the 24 treatments exceeded 10%, indicating a poorer fit inthese cases (Tulasidas 1994). A maximum global error of13% occurred in one particular case. Since the actual dryingtime requirements under MW drying conditions are small,even an error of 13% would not make a substantial differencefrom a practical point of view.

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I January/FcbruarylMarch 1997 65

grapes but can be extended to other materials (fruits andvegetables). Although the modified logarithmic model fittedthe data very well, it lacked in versatility because of its purelyempirical nature.

ACKNOWLEDGEMENT

The authors gratefully acknowledge the Canadian International Development Agency funding through theirInstitutional Co-operation and Development services for thestudy; study leave granted to the first author by the University of Agricultural Sciences, Bangalore, India; and FCARand NSERC Grants for the study.

REFERENCES

Bird, R.B., W.E. Stewart and E.N. Lightfoot. 1960.Transport Phenomena. New York, NY: John Wiley andSons, Inc.

Boland, F.E. 1984. Fruits and fruit products. In AOACOfficial Methods of Analysis, ed. W. Horwitz, 413-418.Washington, DC: AOAC.

Buffler, C.R. 1993. Microwave Cooking and Processing.New York, NY: Van Nostrand Reinhold.

Caldas, F.P. 1992. Experimental high frequency dryer forgranular materials. Transactions of the ASAE35(4): 1275-1282.

Crank, J. 1975. The Mathematics ofDiffusion. OXford, U.K.:Oxford University Press.

Crapiste, G., S. Whitaker and E. Rotstein. 1988a. Drying ofcellular material I. A mass transfer theory. ChemicalEngineering Science 43(11 ):2919-2928.

Crapiste, G., S. Whitaker and E. Rotstein. 1988b. Drying ofcellular material II. Experimental and numerical results.Chemical Engineering Science 43(11 ):2929-2936.

Geankoplis, C.J. 1993. Transport Processes and UnitOperations. Englewood Cliffs, NJ: Prentice Hall.

Gear, C.W. 1971. Numerical Initial Value Problems inOrdinary Differential Equations. Englewood Cliffs, NJ:Prentice Hall.

Gunasekaran, S. 1990. Drying com using continuous andpulsed microwave energy. Drying Technology8(5): 1039-1047.

Hindmarsh, A.C. 1972. Linear multistep methods forordinary differential equations. Method formulations,stability, and the methods of Nordsieck and Gear.UCLR-51186. Rev. 1. Livermore, CA: LawrenceLivermore Laboratory.

Hindmarsh, A.C. 1974. Gear - Ordinary differentialequations system solver. UCID-3001. Rev. 3. Livermore,CA: Lawrence Livermore Laboratory.

Lozano, J .E., E. Rotstein and M.J. Urbicain. 1983.Shrinkage, porosity and bulk density of food stuffs atchanging moisture contents. Journal of Food Science48(5): 1497-1502.

Maroulis, Z.B., E. Tsami and D. Marinos-Kouris 1988.Application of the GAB model to the moisture ~orpt~on

isotherms for dried fruits. Journal of Food Engllleenng

66

7( 1):63-78.

Masi, P. and M. Riva. 1988. Modelling grape drying kinetics.In Preconcentration and Drying ofFood Materials, ed. S.Bruin, 203-214. Amsterdam, The Netherlands: ElsevierScience Publishers.

Metaxas, A.C. and R.J. Meredith. 1983. IndustrialMicrowave Heating. Hefts, England: Peter PeregrinusLtd.

Miller, M.W. 1963. Grapes to raisins. In 20 Years ofRaisinsResearch, 19-24. Fresno, CA: California Raisin AdvisoryBoard.

Mohsenin, N.N. 1986. Physical Properties of Plant andAnimal Materials. New York, NY: Gordan and BreachScience Publishers.

Otten, L. and C. St. John. 1988. Thin-layer microwave dryingof peanuts. CSAE Paper No. 88-502. Saskatoon, SK:CSAE.

Page, G.E. 1949. Factors influencing the maximum rates ofair drying shelled corn in thin layers. M.Sc. Thesis.Purdue University, West Lafeyette, IN.

Perkin, R.M. 1990. Simplified modelling for the drying of anon-hygroscopic capillary porous body usingcombination of dielectric and convective heating. DryingTechnology 8(5):931-951.

Ptasznik, W., S. Zygmunt and T. Kudra. 1990. Simulation ofRF-assisted convective drying for seed quality broadbean. Drying Technology 8(5)977-992.

Raghavan, G.S.V. 1994. Numerical analysis applications infood and biological engineering processes. ASAE paperNo. 94-3582. St. Joseph, MI: ASAE.

Raghavan, G.S.V., P. Alvo and U.S. Shivhare. 1993.Microwave drying of cereal grain: advant~ges andlimitations. Post Harvest News 4(3):79N-83N.

Ratti, C. 1991. Design of dryers for veg~table and fruitproducts. Ph.D. Thesis, Universi dad Nacional del Sur,Bahia Blanco, Argentina.

Ratti, C., G.H. Crapiste and E. Rotstein. 1989a. A new watersorption equilibrium expression for solid foods based onthermodynamic considerations. Journal of Food Science54(3):738-742.

Ratti, C., G.H. Crapiste and E. Rotstein. 1989b. PSYCHR: Acomputer program to calculate psychrometric properties.Drying Technology 7(3):575-580.

Ratti, C. and A.S. Mujumdar. 1993. Fixed-bed batch dryingof shrinking particles with time varying drying airconditions. Drying Technology 11(6):1311-1335.

Riva: M. and C. Peri. 1986. Kinetics of sun and air drying ofdifferent varieties of seed-less grapes. Journal of FoodTechnology 21(2):199-208.

Saravacos, G.D. and S.N. Marousis. 1988. Effect of ethyloleate on the rate of air drying of foods. Journal of FoodEngineering 7(4):263-270.

Saravacos, G.D. and G.S. Raouzeos. 1986. Diffusivity ofmoisture in air-drying of raisins. In Proceedings of theInternational Drying Symposium. IDS '86, ed. A.S.

TULASIDAS. RATII and RAGHAVAN

E"

KgkG

eoe'

x*~Id

pmpsW

vVVgX

thermal conductivity of air (Wem-1eK-1)average mass transfer coefficient based on

-2 -I P -I)pressure (kgem eS e aL mass transfer Biot number (dimensionless)M moisture content (kg water!kg wet mass)m mass (kg)n parameter in Eq. 36NT number of time intervals, Eq. 38Nu Nusselt number (dimensionless)P microwave power density ~W/g of dry solids)Pal' average MW power (W1m' )Pr Prandtl number =Cg J.lgIKg (dimensionless)p partial water vapour pressure (kPa)q I, q2, q3 coefficients in Eqs. 22, 23R radius (m)R2 coefficient of determinationRH relative humidity of air (%)

Re Reynolds number =d Vg pg/~g (dimensionless)r radial distance in spherical coordinates (m)S slope of the plot In [(X-Xo)/(X-Xe)l against

time, tlocal shrinkage coefficient, vivotemperature of solid (oC)air temperature (oC)solid temperature at surface (oC)time (h)

set of instants bisecting two consecutive timeintervals, (Eqs. 37, 38) (h)volume (m3)velocity (m/s)velocity of air in the drying chamber (m/s)average moisture content (kg water!kg drymass)average moisture content (kg!kg dry mass)equilibrium moisture content of grapes(kg/kg dry mass)local moisture content (kg water!kg dry mass)roots of Eq. 9incremental changepermittivity of free space (F/m)dielectric constanteffective dielectric loss factorvelocity of shrinkage (m/s)moving coordinatemoving coordinate (dimensionless)viscosity of air (kgem-1es- l )

fl k -" -I)mass ux ( gem -esdensity of material (kg/m3

)

dry matter concentration (kg/m3)

angular frequency = 21tf

A

mg

11

S*

TTgTmt

NTlth= 1

Subscripts

e equilibriumeff effectiveg gas, airm materialo initialS dry solidw water

NOMENCLATURE

A surface area of particle (m2)

A coefficient in Eq. 24A1 coefficient in Eq. 27B coefficient in Eq. 24BI coefficient in Eq. 27aw water activityC concentration of species (kg/m)CI. C2. C3 coefficients in Eq. 13em specific heat of material (Jekg-1eK- 1)

"D'e/f effective moisture diffusivity (m-/s) "De/f effective moisture diffusivity parameter (m-/s)d diameter (m)dl, d2 coefficients in Eq. 22Erms electric field strength (V1m)et local relative error of approximation at timeiF global relative error of approximation

throughout the drying periodf frequency (Hz)H s heat of sorption (Jlkg)H IV heat of vaporization (J/kg)h enthalpy of air (kJ/h)hg average heat transfer coefficient (Wem-2eK- 1)k exponent in Eq. 36

Mujumdar, 487-491, New York, NY: HemispherePublishing Corp.

SAS. 1989. SASISTAT User's Guide, Version 6, 4th ed.,Volume 2, Cary, NC: SAS Institute Inc.

Shivhare, U.S., P. Alvo, G.S.V. Raghavan, and F.R. van deVoort. 1992. Application of response surface methods ingrain drying research. In Proceedings of theInternational Drying Symposium. IDS '92, ed. A.S.Mujumdar,1549-1559. Amsterdam, The Netherlands:Elsevier Applied Science Publi~hers.

SigmaPlot, 1992. SigmaPlot Scientific Graph System:Transforms and Curve Fitting. San Rafael, CA: JandelScientific.

Steel, R.G.D. and J .H. Torrie. 1980. Principles andProcedures of Statistics. New York, NY: McGraw-HillBook Co., Inc.

Tulasidas, T.N. 1994. Combined convective and microwavedrying of grapes. Ph.D. Thesis. McGill University,Montreal, QC.

Tulasidas, T.N., G.S.V. Ragahavan and R. Girard. 1994.Dielectric properties of grapes and sugar solutions at2450 MHz. ASAE Paper No. 94-3005. St. Joseph, MI:ASAE.

Tulasidas, T.N., G.S.V. Raghavan and E.R. Norris. 1993.Microwave and convective drying of grapes.Transactions of the ASAE 36(6): 1861-1865.

Turner, I.W. and P.G. Jolly. 1991. Combined microwave andconvective drying of a porous material. DryingTechnology 9(5): 1209-1269.

Weres, J. and D.S. Jayas. 1994. Thin-layer drying of corn:Experimental validation of a new numerical structuralmodel. Canadian Agricultural Engineering 36(2):85-91.

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No.1 January/FebruaryIMarch 1997 67

TECHNICAL NOTE

H-OPCO$T: A software package foranalyzing the costs of operating a

mechanical wild blueberry harvesterK. J. SIBLEY and D.L. ARSE AULT

Sibley ElIgilll!eri1/g~Di\'i.\·iofioj Paige Holdillgs Ltd.. 8 \1111(11/ Road. Great Village. NS. Canada BOM ILO. Reeeh'ed /8Seplelllber/995: accepled 24 OClol,er /996.

Fig. I. Page I of lhe Operating Cost Statement window.

OPERATING STATEMENT

Su,,,h

If Print FOIm

.1 Next Paye

$ 13057.51

III

$ @181.75

SOFTWARE DESIGN

$ 13603.75

S 1128.00

$ @50.00

s r72U6

S 1977.11

S 1173.84

s r81.80

SfiOW Delia

own operations to get a detailed statement of annual. per hourof opcration. and per pound harvesled operating costs. Thesoftware can also be used to project fUlUre operating costs forvarious scenarios. Cost calculations are performcd using themost up-to-date generally accepted accounting principlesand fonnulae and are adjusted for inflation to give the mostaccuratc results possible.

H-OPCO$T was developed using Microsoft Visual Basic 3.0for Windows. Visual Basic's Setup Wizard was thcn used tocreate a stand alone software package lhat runs under thcMicrosoft Windows 3.1 environment. \Vindows based devel·opmcnt software was used to give the user a visually pleasingand easy-to-use working environment. The program is mouseand kcyboard driven. Results windows can be viewed onscrccn by clicking on the appropriate buttons. As wcll. each

Total

Tolal

Lubrication

Depreciation and interest

Fuel

Hydraulic oil

Housing

Insurance

Repairs and maintenance

Annual Variable Costs

Sibley. KJ. and Arsenault. D.L. 1997. I-I·OPCO$1': A SOnWOlrc

package for an:.tl)'zing the costs of operating a mechanical wildblucbcrr)' harvester. Can. Agric. Eng. 39:069-072. Operating costanalysis software was developed for a mechanical wild blueberryharvesting operation. It allows growers and harvcs[cr owners 10

cstim;l1c totnl anllual opcr:lIing costs and operation ('omponcill costs.Costs c:m also be calculated on a per hour and per pOllnd h;tr\'c~lcd

basis. User defined v:lriablc data allow for lhe c<llctllatiol1~ 10 bewilored to a specific operation or to prcdici future costs resultingfrom lhc altcration of any operating v;lriablcs. Resulls call be vicwedin financial SlalClllcnL tabular. or graphical form. The program wasdeveloped using Microsoft Visual Basic 3.0 1'01' Windows and nmson an IBM or compatible computcr undcr Ihc Windows 3.1 environ·mcnl.

Un logicicl pour analyscr Ics frais d'operation a CIC devclopp'::pour Ia cucillcllC mccaniqllc du blclici sauvage. II pcnnct au culti·valcur ct all proprictairc dc I'apparcil de rccollc (I'cslimcr Ics frais<i'operation annucls 10laux ct lcs frais <I'operation par scclcurs. Lesfrais pcuvcnl aussi elrc caleulcs sur line base hOl'ain: OlJ par livre deblcucls rccollcs. Les donnees dcfinies par

l'us;lgcr pcrmellcnt Ic caleul d'clrc ada piea une operalion spCcifiqllC all de prcdirc lesfrais fUHlfS resultant de l'allcralion desvariables d·operation. Les rcsllltats pCllvcnt elrc ulilises dans des rapponsfinancicrs. SOliS forme dc tablcaux OLl dcgraphiqllcS. Lc programlllc a etc elaborcavec un logicicl base sur .. Microsoft Vis-

Annual Fixed Costsual Basic 3.0 .. pour .. Windows .. ct sllr linordinatcur IBM ou compatiblc avec" Windows 3.1 .. au plus clcve.

INTRODUCTION

H-OPCOST (Harvester OperatingCost) is a software program that allows the operating costs of amechanical wild blueberry harvcsterto be estimated based on uscr-definedinput variable valucs. It was created toaid blucberry farmers and extensionworkers by eliminaling Icnglhy handand/or cumbersome spreadsheet calculations, Each usercun tailor the COSI

calculations specifically to his/her

CANADIAN AGRICULTURAL ENGINEERING Vul. :1'). No. I. J;ltIuary/Fehruary/Man:h 1997 69

Operating Slatcmen1..Pagc 2

Totol vorioble $ 16063.18

Annual Administration nnd Overhend $ 1508.14

Greph

Print Form

I Provious Page 1

correspond with the limber or Picking Heads on the harvesler indicatedby Ihe user in the Harvester dala entrywindow.

The Operating CosI Graph window(Fig. 5) is a graphical presentation ofIhe Operating CosI Matrix. showingIhe Operaling Costs per pound versusEffective Field Capacity for a selecledHarvesled Yield. Various harvestedyields can be selected using a dropdown box and an updaled graph isaUlomatically drawn afler each newselection.

Data can be enlered or changedfrom any of the program's resultsviewing windows by clicking on thenppropriale data entry butlon and enlering new dala in the pop-up dataentry window. Figure 6 shows a sample data entry window as wouldappear on the screen when running theprogram. Closing the window thenautomalically initiates costs recalculalion using the new data and displays

the new coslS in the current results viewing window. Thisallows easy comparisons of previous costs with the new COSISresulting from the alteration of any operating variable. Thedata in each dala entry window can also be reset to Ihepre-programmed defauil values.

H-OPCOST will run on any computer running \Vindows3.1 (or later) and supporling SVGA resolulion.

The user inpul variables built inlo I-I-OPCO$T for eachdata entry category. along wilh thcir dcfault values. areshown in Tablc I.

1

II

S 13715.20

S 13456.00

S 1155.52

s roJ.611Total

Wages

Benefits

Bonuses

labor

Tractor Harvester

Depreciation and interest S 11442.56 S 12439.05

Insurance S 140.00 S 100.00

Housing S 190.00 S 10.00

Total S 1'500.56 S 12519.05

I! - I IIOle: P,int

Interest on Operating Copitol $ ,18,,4'-.',,2'---_--"

Totol Annual Operating Cost $ 110670.93

S/Hr Operated $ 015:;:9:.:.2:;:8__-,

Fig. 2. Page 2 of the Operating Cosl Statement window.

Fig. 3. Annual Fixed CO,st detail window. COST CALCULATIONS SUMMARY

window can bc printed to give hard copy of a scenariocreated.

H-OPCOST is set up 10 display three restlll vicwing windows: (I) Ihe Operaling Cost Sialement window. (2) theOperating Cost Malrix window. and (3) the Opcrating CosIGraph window.

The Operating Cost Statement (Figs. I and 2) is set up asa standard financial statement. Ownership. Variablc, Labour..Operating Capital. and Administr<uion and Overhead costsarc displayed along wilh Total Annual Operating and Dollarsper Hour Operated costs. E,lch cost total (Variable. Fixed.and Dollars per I-Iour Operated) has a pop-up detail window(aclivated by the appropriate Show Detail bUllon) whichshows a breakdown of costs for each component of theoperation and cach component or the cost tOlal. Figure 3shows an example of Annual Fixed Cost details.

The Operating COSI Malrix window (Fig. 4) displays operating costs on a per pound harvcslCd basis in a tabular fonnat.Cost per pound values arc shown for various combinations ofHarvesled Yield and Effcctivc Field Capacity (work rate inthe field). The Effective Field Capaeily choices displayedacross thc lOp of thc Matrix will change aUlomatically to

H-OPCOST's opcrating statement breaks the lotal operatingCOsl down into fixcd. variable. and administration and overhead COSls.

Annual fixed costs

Annual fixcd costs are those associalCd with owning lhecquipmcnt (ie. traClor and harvester). They arc conslanl nomaHcr how much or how lillIe the cquipment is used. 1--1OPCOST includes Depreciation and Interesl on Investment.Insurance. and I-lousing in its Annual Fixed cosls calculations.

Depreciation and Interest on Inveslmcnt are determinedusing the Capital Recovery with Return Method. Thismethod accounts for Ihe "opporlunity COSI" of nOI being able10 invest the purchasing funds of Ihe equipment clscwhere (ifpurchaser uscs c.ash-on-hand) or for the cost of borrowing (hepurchasing funds. It uscs a rcal rate of interest bascd on Ihecurrenl markel interest rale and Ihe average expected inOalion rale over thc useful life of the equipmcnt. Depreciationis based 011 induslry averages of remaining valuc as rcportedin the ASAE D497 .1, Agricultural Machinery Managemell1Data (ASAE 1993). The Capilal Recovery with Return

70 SIBLEYand ARSENAULT

Fig. 4. Operating Cost Matrix window.

OPERATING COST MATRIX h!/LBIEffective Field (apllcil,y(lIc!tv)

Melhod has been reported by Bartholomew (1981) to give Ihe111051 accunlle estimate of true Depreciation and Interest.especially in times of high inOalion. Calculations based 011

the alternate Straight Line Depreciation, Interest on AverageInvestment Method can underestimate Depreciation and interest costs by 15% for a market interest rate of IOo/r and auseful life of 20 years. Default Annual I-lours of Usc for the

6,....

Prinl Watlill

repairs and maintenance costs. and a decreased useful life.Both are determined based on the purchase price of theequipmcnt as defined by the user. Default multiplicationfactors arc industry averages as reported in the ASAE0-1.97.1. Agricultural Machinery Managcment Data (ASAE1993) <lnd in the Atlantic Committee on Agricultural Engi·neenng publication Minimum Cast of Custom Rates for

Agricultural Machines (WailS 1986).

Unlikc other machinery costing software.Annual Fixed costs may also be increased ordecreased by a Severity of Use Faclor builtinlo H-OPCOST's financial model. In real life.if the cquipment is harshly used or not maintained properly. its uscful life and remainingvalue can be decrcascd significantly. Decreasing the useful life and remaining value of theequipment increases the actual ownershipcosts. The converse is also true for equipmcntthat is properly used and well cared for. InH-OPCOST. increasing Ihe Severily of UseFactor above 1.0 decreases useful life. Decreasing it below 1.0 increases useful life.Since this is a new concept bcing built into afinancial model. there arc no background datain thc Iileraturc and collecling such data wasbeyond the scope of this projecL Jt is availablefor users 10 use if they feel their particularcircumstances warrant. Otherwise it can be leftset to its default value of 1.0 which will haveno effect on the cost calculations.

.2 .2< 1.28 1.32 1.36'000 28.G 24.1 21.2 18.5 16.5

1500 19.8 16.5---14.1 '2.-,--11.0

2000 ' •. 8 -n.. 10.6 '.3 B.2

"'"'" lUI ••• B.S ~ G.G

JOBB ••• B.2 7.1 G.2 5.5,350B B.S I 7.1 ' G.O 5.3-- V

'080 7.4- 1G.2 , 5.3 '.G ,. I

'''B G.G 15 .5 '.7 1.4.1 3.7

500B 5.' ,.. '.2 3.7 3.3

55BB 5.' '.5 3.B 3~.'-- 3.0

6000 ,.. ,.I 3.5 3.1 2.7

650B '.G 3.B 3.3 2.. 2.5

7000 3.5 - 3.0 2.G 2.''.21500 '.0 3.3 2.B 2.5 2.2

BOOO 3.7 3.1 2.G 2.3 2.1

B500 3.5 2.. ---':5 2.2 I.'9000 3.3 2.7 2., 2.-,- ..B

9500 3.' 2.G 2.2 2.0 1.7

HOlvestedYield (Ibs/lle)

.0

Harvested Yield: 1000 lbs/acreAnnual variable costs

Annual Variablc costs arc those assochlled with operating theequipment. They vary according ta how much the equipmentis used. H-OPCOST includes Repairs and Maintenance. Fuel.Lubrication. Hydraulic Oil. Labour. and (I1lerest on Operat·ing Capital in dctermining Annual Variable Costs.

Repairs and Maintenance costs arc determined based onthe annual usc of the equipment and its purchase price.Default Illultiplication factors for Rcpairs and Maintenancccosts arc industry averages as reported in Saskatchewan. Agriculture's Farm Machinery Custom & Relllal R{{{e Guide(Saskatchewan Agriculture 1987).

Fuel COSIS are detennined based on the annual use of the0.1'0.1

to

Fig. 5. Operating Cost Graph window.

tractor are based on industry averagcs as reponed in AlbertaAgriculture's publication Farm Machil/l'ry Co.\'!S as a GHideto Custom Rates (Alberta Agriculturc 1990). Annual I-lour:.;of Use for the harvestcr are based on thc number of days used,Ind thc daily operating time as input by the user.

Insurance and Housing costs have been included in An·nual Fixed Costs. even though some owncrs may not insurcthcir equipmcnt or may leave it outside unprotected from theweather. Inclusion of thcse COSIS gives a more realistic estimate of the true costs of owning the equipmcnt. espcciallysince not propcrly housing the equipmcllt results in incrcased

I Res:ello default yalues , Done]

Fig. 6. Tractor data entry window.

Print _,

CANADIAN AGRICULTURAL ENGINEERING Vol. 39. No. I. J;t1HI;lIy/Fcbnlary/March 1997 71

Table I: User input variables and their default values

Tractor Purchase price $40,000Annual hours of use 600 hUseful life 10,000 hSeverity of use factor IInsurance rate factor 0.4%Housing cost factor 0.75%Repair and maintenance factor 0.0833 $/h/$IOOO listFuel use 13.24 L/hFuel price $0.41/LLubrication cost factor 15%

a user input percentage of wages paid.Interest on Operating Capital accounts for the "opportu

nity cost" of not being able to invest funds used duringharvesting (ie. to pay variable costs) elsewhere, if cash-onhand is being used to float the operation of the equipment, orfor the cost of borrowing funds during such time. It is determined based on the Operating Capital Repayment Period,current market Interest Rate, and Total Variable costs.

Administration and overhead costs

Administration and Overhead costs are determined based ona user input percentage of total Annual Fixed and AnnualVariable costs.

Purchase price $20,000Number of heads IUseful life 2000 hSeverity of use factor IHousing cost factor 0.75%Repair and maintenance factor 0.3125 $/h/$1000 listHydraulic oil use 90 L/yrHydraulic oil price $2.02/LLubrication cost factor 15%

equipment and the current price for fuel. Lubrication costsare determined as a percentage of Fuel costs. The defaultpercentage is based on industry averages as reported in theASAE D497.1, Agricultural Machinery Management Data(ASAE 1993). Hydraulic Oil costs are determined based onannual oil use and the current price for oil as input by theuser.

Labour costs are determined based on the number ofequipment operators, the annual hours the equipment is operated, average daily service time, and average daily travel timeto and from the work site as input by the user. Benefits andBonuses costs are also included and are determined based on

Harvester

Labour

Operating

Bank

Wage rate for driverWage rate for boxer(s)Number of boxersWage benefit factorWage bonus factorAdministration and overhead

factor

Number of days usedAverage daily operating timeAverage daily service timeAverage daily travel time

Operating capital repaymentfactor

Current interest rateAverage inflation rate

$8/h$6/hI4.5% of wages3% of wages

5% of total costs

18 dlOh

I hI h

60d8.5%1.5%

CONCLUSIONS

H-OPCO$T gives the user the following capabilities:

I. Growers can estimate mechanical harvesting costs andnegotiate the best price for having their fields custompicked.

2. Harvester owners and custom operators can use it topredict operating costs.

3. Ability to analyze labour costs versus operating costsfor different levels of operating performance.

4. Ability to analyze operating costs per pound versusharvested yield.

5. Ability to analyze the implications of changing operating variables on costs.

Currently, H-OPCO$T is being used by Farm BusinessManagement and Agricultural Engineering Extension personnel in New Brunswick, PEl, and Maine, USA. Commentsreceived from these users indicate that the software is extremely easy to use, is visually pleasing, and producesaccurate cost figures.

REFERENCES

Alberta Agriculture. 1990. Farm Machinery Costs as a Guideto Custom Rates. Agdex No. 825-4. Olds, AB: AlbertaAgriculture, Farm Business Management Section.

ASAE. 1993. ASAE D497.1 - Agricultural machinerymanagement data. In ASAE Standards 1993, 328-334 . St.Joseph, MI: ASAE.

Bartholomew, R.B. 1981. Farm machinery costing underinflation. Transactions of the ASAE 24(4):843-845.

Saskatchewan Agriculture. 1987. Farm Machinery Custom& Rental Rate Guide (Revised). Regina, SK:Saskatchewan Department of Agriculture, EconomicsBranch, Farm Management Section.

Watts, K.C. 1986. Minimum Cost of Custom Rates forAgricultural Machines. Agdex No. 740. Truro, NS:Atlantic Committee on Agricultural Engineering.

72 SIBLEYand ARSENAULT

TECHNICAL NOTE

Power requirements and bale characteristicsfor a fixed and a variable chamber baler

D. TREMBLAy2, P. SAVOIE1.2 and Q. LePHAT3

JSoils and Crops Research and Development Centre, Agriculture and Agri-Food Canada, 2Departement des sols et de genieagroalimentaire, Universite Laval, Sainte-Foy, QC, Canada G1K 7P4; and 3Ministere de I' Agriculture, des Pecheries et deI'Alimentation (MAPA), 200A, Chemin Ste-Foy, Quebec, QC, Canada GIR 4X6. Received 18 March 1996; accepted 6November 1996. Contribution No. 548, SCRDC.

Tremblay, D., Savoie, P. and LePhat, Q. 1997. Power requirementsand bale characteristics for a fixed and a variable chamber baler.Can. Agric. Eng. 39:073-076. A variable-chamber baler (VCB) anda fixed-chamber baler (FCB) were compared with respect to powerrequirements, field capacity, and bale mass. The fixed-chamber balerwas operated with and without a coarse cutting mechanism (FCB/Cand FCB, respectively). Average PTO power requirements weresimilar (10 to 15 kW) but peak power requirements were considerably greater with the FeB (38 kW) and the FCB/C (42 kW) than withthe VCB (15 kW). The VCB had a material flow of 4.1 tonne of drymatter (DM) per hour (t/h) compared to 4.8 and 4.6 t/h with the FCBand FCB/C respectively. Bale dry matter mass was not affected bybaler type nor by cutting and averaged 263 kg for a standard 1.22 mwide by 1.22 m diameter bale. Bale dry matter density varied,however, between 152 and 192 kg/m3 because of variations in material flow and moisture. Forage chopped with the FCB/C had anaverage length of cut of 150 mm. Keywords: forage, round bales,power, mass, density, chop length.

Une presse aballes rondes achambre variable (PCV) et une autreachambre fixe (PCF) ont ete comparees au niveau de la puissancerequise, de la capacite de recolte et de la masse des balles. La pressea chambre fixe pouvait etre operee avec ou sans mecanisme dehachage (PCF/H et PCF respectivement). La puissance moyenned'operation etait sembiable pour les deux presses etudiees (10 a 15kW). Par contre, la puissance maximale etait considerablement pluselevee avec la PCF (38 kW) et la PCF/H (42 kW) qu 'avec la pev (15kW). Le debit des fourrages a ete de 4, I tonnes de matiere seche al'heure (t/h) avec la PCV comparativement it 4,8 et 4,6 t/h avec laPCF et la PCF/H, respectivement. Pour une balle de 1,22 m delargeur et de 1,22 m de diametre, la masse de matiere seche moyenneatteint 263 kg sans etre affectee par Ie type de presse. La densite dematiere seche des balles a toutefois varie entre 152 et 192 kg/m3 acause de variations du debit et de la teneur en eau. Les fourrageshaches avec la PCF/H avaient une longueur moyenne de 150 mm.

INTRODUCTION

Round balers were introduced in the 1970's to facilitatehandling of large quantities of field-dried hay (Renoll et al.1978). The round baler has also proven to be a convenient andeconomical alternative to chopped forage for silage making(Kjelgaard et al. 1981; Harrison 1985). Several manufacturers modified their round balers to meet some of the newrequirements associated with handling wet windrows andheavy bales. The Prairie Agricultural Machinery Institutepublished 24 reports on as many different commercial round

balers between 1977 and 1992 (PAMI 1992), illustratingsome of the rapid changes in round baler equipment.

A survey of new commercial balers available in the province of Quebec (Canada) in 1991 revealed that there were 56round baler models distributed under 13 brand names(MAPA 1992). Of these, only five were among the 24 modelstested by PAMI (1992). The above data indicate how difficultit is to keep track of all the new technology.

Freeland and Bledsoe (1988) reported that balers withfixed-geometry chambers required larger tractors than balerswith variable-geometry chambers. They tested six round balers in either grass or alfalfa crops harvested at moistureranging between 21 and 38% on a wet basis (w.b.). This rangedoes not cover the typical round bale silage moisture range(40 to 70% w.b.).

The objective of this experiment was to compare a FCBwith a VCB according to power requirements, field capacity,and bale mass and density. A commercial model of eachmachine was selected to harvest windrows at moisture contents in the range of 20 to 75% (w.b.).

PROCEDURE

Baler description

The FCB used was a Welger RP200 model with bale diameterof 1.25 m and width of 1.22 m (Trade names are used solelyto provide specific information. Mention of trade names doesnot constitute an endorsement of the product to the exclusionof other products not mentioned). The windrow pickup widthwas 2.00 m. The forage chopping mechanism on the FCBwas a series of 14 static knives spaced 82 mm apart andinstalled vertically behind the windrow pickup. When thecutting mechanism was activated (FCB/C), forage was pulledthrough the 14 knives before reaching the bale formationchamber. The set of knives could be pulled down and therefore inactivated, representing the conventional method ofmoving full-length forage directly into the bale chamber(FCB). The baler was operated at 540 rpm. Chamber pressurewas set at the maximum level for haylage and silage and atthe level just below maximum pressur~ for hay.


Table I: PTO power requirements and material flow for round bale harvesting

§ based on data recording during bale fonnation.t based on bale mass and baling, tying, and ejection time.


PTO power§ (kW) Material Fieldflow§ . tcapacity

Mean Peak (t DM/h) (t DM/h)

8.5 12.1 5.55 4.1511.1 15.8 3.88 3.5011.2 16.0 4.39 3.90

14.3 40.7 8.51 5.5911.4 32.1 3.82 3.2813.7 38.6 5.66 4.12

16.5 44.6 7.78 5.3812.2 39.3 3.76 3.1414.8 42.8 5.63 4.17

Power requirementsAt running speed under zero forage flow, the PTO powerrequirement was 3.2 kW for VCB and 1.0 kW for FCB. Themoving belts in the VCB caused more friction at zero-flowthan the rotating rolls in the FCB.

Table I summarizes power measurements with the ninetreatments as forage was picked up and compressed in thebale chamber. Values are averages for 4 or 5 bales pertreatment. There were small but significant differences(p=0.05) in the mean PTO power requirements between theVCB (10.2 kW) and the FCB (13.3 kW) or FCB/C (14.5 kW).The additional 1.2 kW of average power required for cuttingwas not significant by Duncan's multiple range test (MRT)at p=0.05. The VCB had a peak power requirement of 14.5kW while the FCB and FCB/C had peak requirements of37.5and 42.2 kW, respectively. The differences between the threemachines were significant at p=0.05 by Duncan's MRT.

Typical time related curves of power during bale formation are illustrated in Figs. 1 and' 2. A relatively uniformrequirementof 10 to 15 kW was observed with the VCB (Fig.1). In contrast, the power requirement increased gradually up

the dimensions to estimate volume. Three random foragesamples totalJing 500 g were hand-collected in the windrowjust in front of the baler and oven-dried at 60°C for 72 h toestimate the moisture at baling (ASAE 1993a).

For FCB/C, samples were taken to measure the actuallength of cut. This was done by unrolJing each bale from thepower measurement trials (15 in total) and taking a I-kgsample that was immediately frozen. The unchopped forageheight was obtained by taking samples of the wilted crop inthe windrow prior to baling and freezing them. After allsamples were subsequently thawed, the mean length of cutwas estimated according to a standard method (ASAE1993b).

HatHaylageSilage

HayHaylageSilage

HayHaylageSilage

Conservationmode

Chambertype

Fixed

Fixed withknives

Bale physical characteristicsAn additional 114 bales (between10 and 15 per treatment combination) were harvested for balecharacteristics evaluation. Eachbale was weighed on a flat scale(Weigh-Tronix DSL) with a capacity of 1000 kg and a resolutionof ±0.2 kg. Two measures of circumference and width provided

Measurements during baling

Two main measurements were taken during baling: materialflow and power requirements. Material flow was obtained bymeasuring bale formation time in the chamber, tying time,and ejection time from the chamber to the field. Power requirement was measured continuously at a 10Hz frequencyby a torque meter attached to thetractor's PTO shaft (Tremblay etal. 1994). Mean power is the average during bale formation; peakpower is a I-s average of 10 measurements. Power was measured atrunning speed for 20 s (zero-flow)and during the complete cycle(bale formation, tying, ejection). Atotal of 43 bales (4 or 5 bales for Variableeach of 9 treatment combinations)was harvested for material flowand power measurements.

The VCB was a Gehl RB 1470 model with maximum balediameter of 1.52 m and width of 1.14 m. The actual diameterwas set at 1.22 m and the windrow pickup width was 1.40 m.Two pressures were exerted sequentially against the bale inthe variable-geometry chamber: a pneumatic pressure of 0.8MPa to form the core, and a hydraulic pressure through aseries of belts to maintain the bale under tension. The hydraulic pressure could range between 0.5 and 3.3 MPa and themanufacturer's recommended pressure of2.6 MPa was used.For high-moisture bales (moisture 50-75%, w.b.), the corediameter was set at 0.33 m. For dry bales (moisture around20%, w.b.), the core diameter was set at 0.69 m. The VCBwas operated at 450 rpm and it did not contain a choppingmechanism option.

Experimental design

Both balers were used on a commercial farm in Saint-Stanislas (QC) near the Normandin Experimental Farm (48°50' N;72°31' W) during first cutting of grass between June 29 andJuly 3, 1992. Grass, mainly timothy and bluegrass, was at theboot and early heading stage, with a yield of 1.35 t/ha of drymatter. A 3x3 factorial experimental design was used withthree machine treatments: VCB, FCB, and FCB/C and threetarget moisture levels: 75% moisture (wet silage), 50% moisture (haylage), and 20% moisture (hay). The grass wasmowed using a conventional mower-conditioner. Silage andhaylage treatments were harvested approximately after awilting period of 4 hand 24 h, respectively. The hay washarvested third when there were two successive sunny daysat the end of the week.

74 TREMBLAY. SAVOlE and LePHAT

Tn:mblay/14 Tn:mblay/IS

45 45

40 40

....... Zero-flow ....... Zero-flow35 35i -- Baling - --Baling~C 30

TyingC 30 Tying- ---- EJecting - - - - - Ejectingc c

II) II)

E 25 E 25! !~ 20 ~ 20II) II).. .... ..; 15 II) 15

-J.'.'.I.I...==0 0Do I Do10 I 10

III

5 I 5II

"0 0

o 50 100 150 200 250 300

Cumulative time (s)Fig. 1. PTO power requirements versus time to form

round bale of non-chopped haylage with a VCB.

o 50 100 150 200 250 300

Cumulative time (s)Fig. 2. PTO power requirements versus time to form

round bale of non-chopped haylage with a FCB.

Table II: Average moisture content, bale DM mass, and bale DM density of eachtreatment

to 35 kW for FCB (Fig. 2). The curves obtained for both balertypes corresponded well to the general curves presented byFreeland and Bledsoe (1988). The FCB would require aconsiderably larger tractor than the VCB because of the highpower exerted to complete bale formation just before tyingand ejection.

Material flow

Averaged across moisture, material flow during bale formation was 4.61 t OM/h for VCB, 6.00 t OM/h for FCB, and5.72 t OM/h for FCB/C (Table I). The actual field capacities,which include tying and ejection times, were 3.8,4.3, and 4.2t OM/h, respectively. Because the FCB had a wider pickup(2.0 m vs 1.4 m) and faster turning PTa (540 rpm vs 450

Bale mass and density

Table II shows the moisture content, mass, volume, and density for the nine treatment combinations. Values are averaged over 15 to 20 bales pertreatment, a combination of bales from the power and fieldcapacity trials. The three target moisture levels (20, 50, and75%) are identified by the conservation mode: hay, haylage,and silage, respectively. The actual average moisture contents were 17.8,54.7, and 63.6%, respectively. Although balemass ranged between 236 and 30 I kg OM, there was nosignificant difference (p=0.65) between baler type: the average mass per bale was 266, 262, and 259 kg OM for the VCB,FCB, and FCB/C, respectively. The cutting mechanism didnot have any effect on bale mass, probably due to the relatively coarse chopping.

Chamber

type

Variable

Fixed

Fixed with

knives

Conservation Moisture content Bale mass

mode (%, w.b.) (kg DM)

Hay 16.7 238.5

Haylage 52.9 295.8

Silage 62.2 272.0

Hat 20.1 251.6

Haylage 55.1 300.7

Silage 64.8 236.1

Hay 16.4 265.4

Haylage 55.9 260.9

Silage 63.8 251.8

Volume Density(m3) (kg DM/m3

)

1.57 151.9

1.58 187.2

1.67 162.9

1.59 158.2

1.57 191.5

1.60 147.6

1.61 164.8

1.56 167.2

1.63 154.5

rpm), it could move faster in thefield than the VCB which was especially useful in low yield.

The hay crop was systematically harvested at a higher fieldcapacity (5.0 t DM/h) than silage(4.1 t OM/h) or haylage (3.3 tOM/h). This was because haywindrows were generally tripledbefore baling while silage andhaylage windrows were baled singly. Haylage plots correspondedto the lowest yields; baler feedingcapacity was therefore limited bysoil topography and forward speedrather than machine type or moisture.


Moisture caused significant differences (p=O.OOOI) in baleDM mass with an interaction (p=O.OOOI) due to machinerytype. Overall, haylage bales contained more dry matter (286kg DM) on average than wetter silage bales (253 kg DM) ordrier hay bales (252 kg DM). Considering only the VCB andFCB without the cutting mechanism, haylage bales wereconsiderably heavier (297 kg DM) than average silage (254kg DM) or hay bales (245 kg DM). In the case of FCB/C, Le.with the cutting mechanism, haylage bales were not heavier(261 kg DM) than silage (252 kg DM) or hay bales (265 kgDM). It should be pointed out that haylage was generallyharvested at a slower material flow rate, which providedmore time to increase compactness during bale formation. As

3a result, haylage bales were denser (182 kg DM/m ) thansilage bales (155 kg DM/m3

) and hay bales (158 kg DM/m3).

However, the baler type did not significantly affect baledensities which were 167, 166, and 162 kg DM/m3 for VCB,FCB, and FCB/C, respectively.

Chopped bale particle length

The forage was chopped coarsely with the FCB/C. The freshcrop itself was relatively short (298 mm on average). Themean particle lengths (MPL) were 145, 135, and 162 mm forhay, haylage, and silage, respectively. The non-dimensionalgeometric standard deviations calculated according to ASAE(1993b) were 2.1, 2.1, and 1.9, respectively. These resultsshow that the cutting mechanism chopped stems in only twoor three pieces which was insufficient to affect bale drymatter density.

Power and throughput measurements were made underactual commercial harvest conditions to compare performance of three baler configurations at three moisture levels.Other factors such as yield, windrow handling prior to baling,and material throughput during baling could be significantfactors. Further research on the effect of these factors onpower and throughput is required for a more complete understanding of round baler performance.

CONCLUSIONS

I. The variable-chamber baler (VCB), the fixed-chamberbaler (FCB), and the FCB with a coarse cutting mechanism (FCB/C) used in this experiment required anaverage PTO power of 10.2, 13.3, and 14.5 kW, respectively. However, the peak PTO power required wasconsiderably lower for VCB (14.5 kW) compared toFCB (37.5 kW) and FCB/C (42.2 kW). A larger tractorwould normally be required to operate the FCB and theFCB/C.

2. Bales of standard dimensions (1.22 m wide by 1.22 mdiameter) weighed between 236 and 301 kg DM, withan average of 263 kg DM. Bale density varied between152 and 192 kg DM/m3• Bale DM mass and bale densitywere not influenced by baler type but they varied withforage moisture.

76

3. A series of 14 static knives on the FCB/C providedcoarse chopping with average cut pieces ranging between 135 and 162 mm in length. The cuttingmechanism required slightly more average power but itdid not increase bale mass or densi ty.

ACKNOWLEDGEMENTS

The authors thank the Engineering Services of the QuebecDepartment of Agriculture (Service du genie, MAPA) forfunding this project. They also acknowledge the cooperationof Roliso Farm in Saint-Stanislas which provided foragefields for the experiment and the Welger fixed-chamberbaler. The authors also thank the Quebec distributor of Gehlequipment for providing the variable chamber baler. Technical assistance was provided by Sylvain Fortin and MichelPouliot of Universite Laval. Mr. Jean-Marie Wauthy of theNormandin Experimental Farm of Agriculture Canada provided an instrumented tractor during the experiment.

REFERENCES

ASAE. 1993a. ASAE S358.2 Moisturemeasurement-forages. In ASAE Standards 1993,451. St.Joseph, MI: ASAE.

ASAE. 1993b. ASAE S424.1 - Method of determining andexpressing particle size of chopped forage materials byscreening. In ASAE Standards 1993, 459-461. St. Joseph,MI: ASAE.

Freeland, R.S. and B.L. Bledsoe. 1988. Energy required toform large round hay bales - Effect of operationalprocedure and baler chamber type. Transactions of theASAE 31 (I ):63-67.

Harrison, H.P. 1985. Preservation of large round bales at highmoisture. Transactions of the ASAE 28(3):675-679,686.

Kjelgaard, W.L., P.M. Anderson, L.L. Wilson, H.W.Harpster and PJ. LeVan. 1981. Round bale silage. ASAEPaper No. 81-1520. St. Joseph, MI: ASAE.

MAPA. 1992. L'inventaire descriptif des presses a ballesrondes. Service du genie. Agdex 745. Ministere deI' Agriculture, des Pecheries et de I' Alimentation,Gouvemement du Quebec, Quebec, QC.

PAMI. 1992. Index of Publications - July 1992. PrairieAgricultural Machinery Institute, Humboldt, SK.

Renoll, E., L.A. Smith, J.L. Stallings and D.L. Hess. 1978.Machine systems for handling and feeding round bales. InProceedings of the First Imernational Grain and ForageConference, ASAE Publication 01-78, 296-299. St.Joseph, MI: ASAE.

Tremblay, D., P. Savoie, M. Drouin, A. Amyot and R.Theriault. 1994. Power requirements to macerate freshforage and chop wilted mats. Transactions of the ASAE37(4): 1037-1042.

TREMBLAY. SAVOlE and LePHAT

NOTES TO CONTRIBUTORSThe Editorial Board will assess suitability and esselltial detail ofpapers submitted for public:.uion in COlladioll Agricultural E"gil/{'('I'ing. One or more reviewers will be used. Their comments andsuggestions will be compiled and submitted to the :'lUthar. The reviewwill ensure thai:

1. A research paper presents a piece of research carried to :1well-defined stage of advancement and the conclusions arc adequatelysupported by the experimental results.

2. A u'c!l1Iical paper presents .1 <:lear. concise. and f<lelUal outlineand interpretation of the development. design. test. or ;:malysis underconsider-Ilion and that il is a contribution in onc of the fields ofagricuhural. biosystems. or food engineering.

3. A gel/eral papa on education. rescarch. or cXlcnsion is pcrtincnl10 major changcs in curriculum. research. or cXlcnsion or to forwardlooking developmcnls in thcse areas.

4. A fecJmicalnote on equipmcnt dcvclopmcnt. techniquc of measurement, or mclhod of analysis will havc an applicalion for otherworkers in the fields of agricultural. biosystems. or food engineering.

MANUSCRIPTThc manuscript should be typed doublc-spaced on papcr 216 x 279mm (8.5" x II ") with margins not less than 30 Illm. The first pageshould comain only Ihc title. authors' names. addresses (includingposlal codcs). and conlribution number whcre applicable. The lelephone number. FAX number. and E-mail address (if available) of lhccorresponding aUlhor should also be included. Tables and capliolls forillustrations should be on separate pages. placed after lhe tex\. Manuscripl paper with numbered lines is preferred. Six copies arc required.After a p'lper has becn accepted for publication. Ihe author will beexpectcd to provide a copy of Ihc paper on Iloppy disk in .1 formatcompalible wilh MS-DOS or Macilllosh syslcms.

The litle of lhe paper should give .111 accurale descriplion of Ihearticlc. using key words thai can be uscd for computer-indexing.

ORGANIZATIONThe p'lpcr should be organized to conform with presenl Journal practice. Sec Norum and J'lyas (1995). All papers must include a shOl'labslract seclion of about 200 words. Authors are encour<l2cd 10 submilthe abslract in both English and French. There will be-a chargc forlranslalion services Ihal musl be provided by Ihe Journal.

Major hcadings - Cenler on lhe page wilh all words in capitallellers.

Subheadings - Slart at left-hand margin. capitalize fil'sl leller.

Sub-subheadings - Same as subhc'ldings bUI underline.

Technical and detailcd infonnation should be included only in thefoml of descriplioll. lable. graph. chan or photograph. In general.follow Ihe slyle given in Norum and Jayas (1995).

ReferencesList rcferences alphabelically by aUlhors allhc end. Follow the 1'01'111011sel by Norum and Jayas (1995). Maleri<ll in press. wilh the namc orthcjournal. may be used as a reference. Privale communicalions andunpublished rcports should be referred 10 in parenthcses in Ihc tex!.Privale communicalions should include the person's tille and address.Avoid thc use of foollloles. Use Ihe author-date systcm in Ihe manuscripl when referring 10 articles in lhe Reference scction.

TablesDesignale lables <II the top by table number (Roman numerals) andtille. in upper and lower case letters. All headings and other infonlla·tion in tables arc to be in lower case cxcept firsl letter of first word.Keep the t;lble compacl and place it <1cross the page whercvcr possible.Do not use vcnicallines.

MeasurementsOnly mctric systcm (51) units arc to be used.

Equations

Equations and formulas musl be set up c1carly. Usc capilals for symbols 'IS much 'IS possible and lower C<ISC for superscripts andsubscripls. Grcek and olher characters should be identified c1e'lrly.Equalions should be numbered on the right-hand margin and in linewilh Ihe cenlcr of the equation.

AbbreviationsAbbrcviale unils of mcasurc only when used with nUlllcrals. UsecorreCI 51 unit abbrevialions. Do not use abbrevialions in Ihe titlc.

ILLUSTRATIONSEilhcr original drawings or glossy pholographs are acceplablc forilluslralions. An illustration should be planned 10 fit. after reduction.inlo a spacc 90 mill wide (one column) or 183 wide (two columns). Theoriginal should be nOlmore than three limes the size of Ihe linalligure.For idCnlilicalion. Ihe figure number and author's name should bewrillcn on Ihe lower left comer wilh soft pencil.

Line drawings should be machine produced on while drawing paperor tracing paper. Authors are cncouraged 10 produce drawings usingonc of lhe commonly used computer packagcs. Lellers, numcrals.labels and 'lxis captions should havc only the first word capilalized.Axis captions should be followed by a comma. lhc symbol in il{l1ics.and the unils in parcnthcses [i.c. Acceleration of panicle. Al' (m/s-)l. Ifa symbol is not used. omit Ihe comma. Lcucrs and numemls must beal least 1.5 mill high and prefcwbly 2 mill high in final fonn. Curveson graphs must be 0.3 mill wide after reduclion. Axes and grid linesshould be c1carly visible bUI inconspicuous: a width of 0.2 nUll afterrcduction is suggestcd. Figure llumbcrs and c",ptions should be Iypedon a scparatc page. not on Ihe original illuslralions. When a paper issubmiHcd for publicalion. the original illuslr.llions necd not be pro·vided so long as lhc copies arc of such quality Ihal reviewers canundcrst<:lnd thcm. Original drawings mUSI be provided when the p'lperis .Iccepled for public.nion. If .1 dr<lwing has been produced by acompuler puckage. a copy oflhe file should be sllbmilled on disk allhelimc thai the manuscript is submillcd on disk.

DISCUSSIONSDiscussions may be submillcd on any paper or lechnical notc published in Ihe Journal for a period of nOI more Ihan four months followingpublicalion. Discussion of.a paper or technical nOle is open 10 anyonewho has significant commems or qucslions about lhe conlenl of thepaper/lechnical note. A discussion will nOI be accepted for publicationif it comains material readily found elsewhcre. is purely speculative.inlroduces personalities. or olherwise falls below the standards of aIcchnic<ll paper in a professional journal. Authors will be given anopportunity 10 reply to discussions.

The fonnal for discussions differs from those of papers in thailigures are (0 be idenlified by c'lpital IeHers 10 avoid confusion withthe original papcr. The discusser should refcr to him/herself as "thewriter" or "I" and to Ihe aUlhor or the original paper as "Ihc author."The firsl pagc shows lhe litle of lhe original paper wilh .1 foolnole 10identify the aUlhor. volumc. p'lgC. and date. N.une and address of Ihewriler of thc discussion follow Ihe litle.

Discussions will be reviewed by the Editorial Board and possiblyIhe reviewers of thc original paper. The lenglh of a discussion isrestriclcd to one journal page. Lengthy discussions will be returned forshortcning. or the writer may be encouragcd 10 submil a paper ortcchnical nOIC.

REFERENCEonllll. D.1. and D.S. J<:lyas. 1995. Instruclions for preparing a paper

for Canadian Agriculfllral Engineering. Canadian Agricultural fngi/u'erillg 37(3):239·243.

volume 39 number i january/february/march 1997laboratory measurement and modelling the effects of...

Documents