CANADIAN

AGRICULTURAL
ENGINEERING

The Journal of the Canadian Society of Agricultural Engineering

La Revue de la Societe Canadienne du Genie Rural

CAE 29(2) 99-230 (1987)

CN ISSN 0045-432X

EDITORIAL
Edward McKyes Inside front cover

Soil and Water

THE SEALING OF SOILS BY MANURE. I. PRELIMINARY INVESTIGATIONS

S.E Barrington, P.J. Jutras, and R.S. Broughton 99
THE SEALING OF SOILS BY MANURE. II. SEALING MECHANISMS
S.F. Barrington, P.J. Jutras, and R.S. Broughton 105
SOIL EROSION LOSSES UNDER FREEZE/THAW AND WINTER GROUND COVER USING A LABORATORY
RAINFALL SIMULATOR
L.M. Edwards and J.R. Burney '09
TILE DRAIN DISCHARGE UNDER DIFFERENT CROPS
G.J. Bryant, R.W. Irwin, and J.A. Stone "7
MEASUREMENTS AND ANALYSIS OF RUNOFF FROM SUBSURFACE DRAINED FARMLANDS
S. Natho-Jina, S.O. Prasher, C.A. Madramootoo, and R.S. Broughton 123
SOIL CHARACTERISTICS AROUND MOLE DRAINS INSTALLED AT DIFFERENT DEPTHS
T.G. Sommerfeldt, C. Chang, T. Entz, and B.J. Lamond 131
SUBSURFACE IRRIGATION AND DRAINAGE ON SANDY SOIL IN SOUTHERN QUEBEC
N.A. Nemon, B. von Hoyningcn Huene, J. Gallichand, and R.S. Broughton 137
PERFORMANCE OF LOW-PRESSURE CENTER-PIVOT SPRINKLER IRRIGATION SYSTEMS IN
SASKATCHEWAN
K.P. Thooyamani and D.I. Norum 143
APPLICATION RATES AND UNIFORMITY UNDER CENTER-PIVOT SPRINKLER IRRIGATION SYSTEMS
USING SPRAY NOZZLES
K.P. Thooyamani, D.I. Norum, and S. Dubetz '49
A PORTABLE RAINFALL SIMULATOR
R.W. Tossell, W.T. Dickinson, R.P. Rudra, and G.J. Wall 155
EVALUATION OF BEST MANAGEMENT PRACTICES TO CONTROL PHOSPHORUS NONPOINT SOURCE
POLLUTION
A. Rousseau, W.T. Dickinson, and R.P. Rudra '63

Power and Machinery

EVALUATION OF POTATO-STONE-CLOD SEPARATION MECHANISMS

G.C. Misener and CD. McLeod 169
Continued on back cover

Volume 29 Number 2 July 1987

 CANADIAN INFORMATION TRANSFER II

AGRICULTURAL Edward McKyes

ENGINEERING Editor, Canadian Agricultural Engineering
JULY 1987 Two years ago John Ogilvie, then President of theCanadian Society of Agricultural Engineer
VOLUME 29, NO 2 ing,asked members what our Society should doto help inthecataloguing and transfer of technical
informationto all of our members. There is a great deal of information in this country and in others
EDITOR which could be of assistance in our endeavours, and discussions are underway in various commit
E. McKYES
Department of Agricultural Engineering. tees now asto how much collectinganddistributing can be accomplishedconsideringour available
McGill University,
Box 950, Macdonald College,
resources. Canadian Agricultural Engineering can help in some ways in this respect without any
Stc. Anne dc Bcllcvuc. Quebec H9X ICO changes to its mandate or regulations. The Notes to Contributors on the inside back cover state
ASSOCIATE EDITORS clearly that papers for our Journal are welcomed in both research and technical fields, among
F. DESIR (Power & Machinery)
Department of Agricultural Engineering, others. While most of the papers published have been of a research nature, there is no reason why
Nova Scotia Agricultural College, the number of submissions treating the development, design and evaluation of new devices or
Truro, Nova Scotia B2N 5E3
J.C. JOFRIET (Structures & Environment)
systemscannot be increased. Suchinformation wouldbeof great interest anduseto researchers and
School of Engineering, designor supervising engineers alike, andwouldnotdetract atall from the quality orquantity of the
University of Guclph,
Guelph, Ontario NIG 2W1 pure research works.
K.V. LO (Electric Power & Processing) Not only would information on Canadian agricultural engineering design technology or evalua
Department of Bio-Resource Engineering, tion be interesting for members to read, whether in machinery, soil and water, structures and
University of British Columbia,
2357 Main Mall, environment, energy, processing or foods, but the international experience of Canadianor foreign
Vancouver, B.C. V6T IW5
engineers would also be of value for those who also give technical assistance overseas, or those
K. POHJAKAS (Soil & Water)
Resource Planning Division, who could adapt foreign experiences to localconditions. Inorderto increasethe input of design and
Land Classification Branch, evaluation engineers to our Journal, all such readers are encouraged to contemplate submitting
Alberta Agriculture,
Agriculture Centre, technical manuscripts. And allCSAE members could discuss the question with acquaintanceswho
Bag 3014, Lcthbridgc, Alberta TIJ 4C7 are involved in the design andtesting fieldsof activity. The message should be that technical papers
R. KOK (Food Engineering)
Department of Agricultural Engineering,
have always been welcomed in our Journal, as have research papers, and that the Editorial Office
McGill University, (i.e. the Editor) is more than willing to give advice on the suitability or layout of such articles even
Box 950, Macdonald College,
Stc. Anne dc Bcllcvuc, Quebec H9X ICO before they are in fact written. It would be encouraging to see increased cooperation among
CSAE COUNCIL 1987-88 researchersand designers in agricultural engineering, and the mere fact of seeing published articles
M.G. BRITTON
Department of Agricultural Engineering,
in both of these fields side by side in our Journal could well be one step in promoting such
University of Manitoba, collaboration.
Winnipeg, Manitoba R3T 2N2
D.A. ALLEN Past President
Department of Agricultural Engineering
Nova Scotia Agricultural College,
Truro, Nova Scotia B2N 5E3
G.C. MISENER President Elect
Agriculture Canada Research Station,
P.O. Box 20280,
Frcdericton, New Brunswick E3B 4Z7
P. SAVOIE Vice-President
Dcpartcmcnt dc Genie Rural, (Technical)
Univcrsite Laval,
Stc-Foy, Quebec GIK7P4
KG. BOYD Vice President
Guclph Agriculture Centre, (Regional)
Box 1030,
Guclph, Ontario N1H6NI
G.M. OMICHINSKI Secretary-Treasurer
P.A.M.I.,
P.O. Box 1060,
Portage la Prairie, Manitoba R1N3C5
REGIONAL DIRECTORS
S.T. CHIENG British Columbia
Department of Bio-Resource Engineering,
University of British Columbia,
2357 Main Mall,
Vancouver, British Columbia V6T 1W5
J.J.R. FEDDES
Department of Agricultural Engineering,
University of Alberta,
Edmonton, Alberta T6G2HI
R.B. ROGERS Saskatchewan
Rogers Engineering Inc.,
1510 Hilliard Street East.
Saskatoon, Saskatchewan S7J 0G4
A. ROBERTS
223 Glcnbush Street,
Winnipeg, Manitoba R3R 0N9 Canadian Agricultural Engineering publishes papers covering the general field of Agricultural Engineering that fit into
M.POTTER one of the following classifications: (1) a scientific paper based on original research; (2) a technical paper based on design,
Masscy Ferguson Research Ltd., development, testing, or analysis of machines, equipment, structures, processes, or practice; (3) a general paper on education
C.E.Q. 5735 McAdam Road, relative to curricula and philosophy or trends in science, on a survey or investigation of some phase of research or research
Mississauga, Ontario L4Z 1N9 methods, or on extension or extension methods. The Editorial Board may also publish abstracts published elsewhere and
D. MILLETTE Quebec interesting news items from members of Agricultural Engineering.
U. Del isle and Associates,
P.O. Box 60,
Manuscripts for publication should be submitted to the Editor. The papers must be original and must have not been
St. Charles dc Richelieu, published elsewhere or copyrighted. The author, not the CSAE, is responsible for opinions expressed. Information published in
Quebec J0H 2G0 Canadian Agricultural Engineering may be quoted in whole or in part provided that credit is given to the author and to the
K.C. WATTS journal. Publication chargesare $40/pagepluscostof illustrations etc. and reprintchargesare [$10.00 + (numberof setsof 100
Department of Agricultural Engineering, reprints ordered x $3.00/100 reprints)] x (number of pages per reprint).
Technical University of Nova Scotia,
Box 1000,
Central Office Address: Suite 907, 151 Slater Street, Ottawa, Ontario K1P5H4
Halifax, Nova Scotia B3J 2X4
Published Semi-Annually (January and July)
Subscription rate: $30.00 per annum, single issue $17.00
 THE SEALING OF SOILS BY MANURE.
I. PRELIMINARY INVESTIGATIONS

S. F. Barrington1, P. J. Jutras2, and R. S. Broughton1

'Department of Agricultural Engineering, McGill University, P. O. Box 950, Macdonald College, Ste.
Anne de Bellevue, Quebec H9X ICO; and 2USAID, do Embassade des Etats-Unis, Dakar, Senegal.
Received 2 May 1986, accepted 15 October 1986

Barrington, S. F., P. J. Jutras, and R. S. Broughton. 1987. The sealing of soils by manure. I. Preliminary inves-
tigations. Can. Agric. Eng. 29: 99-103.

Infiltration rates, groundwater contamination and changes in soil profile nutrient composition were observed for four
experimental field reservoirs filled with 10% total solids (TS) dairy manure. The results indicated poor correlation between
manure soil infiltration and soil saturated hydraulic conductivity with water. These findings were confirmed in the
laboratory using nine columns of four different soil textures exposed to 6% dairy manure.

INTRODUCTION TABLE I. PREVIOUS WORKS ON THE SEALING OF SOIL BY MANURE

The intensification of Quebec's agri Infiltration rate Manure
culture during the 1960s contributed to the Initial Final Head Type Soil Experimental
deterioration of the quality of surface Author (10-5 m/s) (10-" m/s) (m) texture conditions

waters of the southern part of the Province. California, U.S.A.

Faced with this problem, the Quebec Gov Hart et al. 0.9 620-30 2.2 Poultry, Sand Laboratory
ernment introduced, from 1972 to 1981, (1965) dairy, hog cores kept
outside
laws and bylaws aimed at curtailing this
Meyer et al.
deterioration. Among other measures,
(1972)
livestock enterprises in the process of and
establishment or expansion, along with Oliver et al. 0.7 - 2.48 12 3.0 Screened Sand, Lagoon
those located within heavily polluted river (1974) dairy clay loam
basins, were required to store their Davis et al. 15 60 3.0 Screened Clay loam, Lagoon
(1973) dairy sand
manures in concrete structures. Other
Robinson
materials were accepted as long as their
(1973) 1.3 35 1.0 Beef Clay loam, Lagoon
permeability was of the order of that of silty clay
concrete structures. Soils for example, had Chang et al. 120 - 0.3 486-11 Dairy Sand, loam, Laboratory
to demonstrate a saturated hydraulic con (1974) (k)t (lOt Silty clay
ductivity not exceeding 8.64 x 10~5m/d New Zealand
(10~7 cm/s). Concrete manure storage Hills

facilities being expensive, government
schemes to improve manure storage pro
gressed at a slow pace. Priority was there
fore given to the development of
construction guidelines for environmen
tally acceptable low-cost manure storage
facilities, such as earthen reservoirs. Such
development would provide a more favor
able economic climate in which to acceler
ate environmental schemes improving
surface water quality, by reducing pollu
tion from manure.
The authors summarized in Table 1 have
suggested three possible sealing mecha
nisms: (1) physical mechanisms by which
soil pores become clogged, (2) biological
mechanismsthrough bacterial activity, (3)
chemical mechanisms where soil clay par
ticles deflocculate and soil structure is
destroyed through reductive processes.
The last two mechanisms were identi
fied as temperature dependent. Biological
mechanisms rely on bacterial enzyme
activity which follow the general tem
perature rate relationship expressed by the
(1976) 0.4 - 0.8 10
Nova Scotia
Lo
(1977) 1.16 - 0.06 3.5-7.75
tk refers to hydraulic conductivity value rather than an infiltration rate.
growth factor, Q'. Chemical mechanisms
require temperatures above 15C during
the soil reduction process (Mirtskhulara
et al. 1972).
Nordstedt et al. (1971), Collins et al.
(1975), Sewell (1978), Ciravolo et al.
(1979) and Patni et al. (1981) investigated
groundwater contamination from earthen
reservoirs via seepage losses. They identi
fied bacteria (total and fecal coliforms,
fecal streptococci), ammonia, nitrates and
chlorides as major contaminants. But they
also observed that the concentration of
groundwater contaminants varied widely
with time and location for any given reser
voir.
The various levels of manure infiltration
2.5 Dairy Sand to Laboratory
clay
Canada
2.4 Diluted Sand to Laboratory
dairy clay
rates, as well as groundwater contam
ination, provided no bases for definite
guidelines for the construction of earthen
storage facilities as of 1980. At that time,
authorities based their bylaws on soil
hydraulic conductivity values. In Ontario,
a value of 8.64 x 10~2 m/d (10"4 cm/s)
was required for the soil on site. Quebec
increased its 1975 value of 8.64 x 10-5
m/d (10 "7 cm/s) to 8.64 x 10~4 m/d
(10-6 cm/s) in 1981. The U.S. state of
Pennsylvania required a permeability
value of 8.64 x 10'4 m/d (lO"6 cm/s)
after sealing had taken place.
A research project was initiated in 1981
to determine the extent of the sealing of
soils by manure under Quebec conditions.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2,SUMMER 1987 99

 Ground Level
Embankment

7 ) t /
7
/ *
111/11
-N '///>}

Ora/n
^200 H 450 cm ->lO0 H /Water 7ab/c?

Perforated Well

Figure 1. Profile of experimental reservoirs.

The results of this project wereto verify the
Quebec norm requiring a soil hydraulic
conductivity under 8.64 x 10~4 m/d
(10"6 cm/s).
PROCEDURE
To investigate sealing efficiency under
natural conditions, four small manure res
ervoirs were built on sites of various soil
textures (Site no. 1 of clay, Site no. 2 of
loam, Site no. 3 of coarse sand and Site no.
4 of 1.2 m of coarse sand over a grey
structureless clay). These reservoirs were
built of minimum dimension to reduce
costs, thus a floor dimension of 1.0 m x
1.0 m, but of typical hydrauliccharge, thus
of 3.0 m depth (Fig. 1). Their side had a
slope of 1.5:1.0 for stability, considering
especiallythe sandy sites. At the firstthree
sites, the groundwater table wascontrolled
below the bottom of the reservoir, while at
the fourth site, no such control was under
taken. This allowed the observation of
groundwaterlevel effects. Twoof the con
trolled reservoirs required a peripheral
drain 0.30 m below the floor levels (sites
no. 1 and no. 2), emptying into a nearby
ditch. Site no. 3 demonstrated a ground
water table maintained naturally some
0.60 m below the reservoir floor.
These four reservoirs were filled in
October 1981 to a depth of 2.4 m with 10%
total solids (TS) liquid dairy manure. They
were observed for infiltration rates and
groundwater contamination until No
vember 1982. Reservoir no. 4 had to be re
plenished in May 1982.
Infiltration rates were monitored reg
ularly by measuring the levelof the surface
crust of each reservoir, using an engineer's
level and a bench mark established within
5 m of each site. Evaporation rates and
rainfall data were obtained from weather
stations located 2 km away (sites nos. 2, 3
and 4) and 10 km away (site no. 1). This
experimental method was based on pro
cedures used by Meyer et al. (1972), Davis
et al. (1973) and Robinson (1973). Infiltra
tion data were compared statistically using
the method of analysis of variance.
Groundwater quality was monitored
through the sampling of water from a well
2.8 m deep within 3 m of each reservoir
and from a control unaffected by the reser
voir seepages if any, thus located more
than 200 m upgrade from each site. Water
samples were tested for bacteria(total and
fecal coliforms and fecal streptococci),
ammonium, nitrate, phosphorous and
potassium. In November 1982, the reser
voirs were emptied and their soil profiles
were analyzed for pH, phosphorous,
potassium and total Kjeldahl nitrogen.
Core samples were taken for this purpose
at two places per site on the reservoir side
walls, 100 cm above their floor. These
cores were taken perpendicular to the res
ervoir surface, at 10-cm intervals to a depth
of 90 cm.
In September 1982, nine dairy manure
columns (three of clay, three of loam, two
of sand and one of gravelly silty clay) were
set up in an unheated laboratory (Fig. 2)
This second project was conducted to
obtain more accurate readings on infiltra
tion rates and groundwater pollutioi haz
ards. Each column held a 10-cm bigh by
10-cm diameter undisturbed B Horizon
soil sample (sample dimension recom
mended for soil hydraulic conductivity
measurements), exposed to a head of 1.80
m of 6% TS dairy slurry. The seepages
(exfiltrates) recuperated under each soil
column were used to measure infiltrat/on

100
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
rates. These exfiltrates were collected dur
ing the second month for nutrient and min
eral analyses. The columns were observed
for 3 mo, October through December
1982.

RESULTS AND DISCUSSION

The four experimental reservoirs were
monitored for infiltration rates throughout
a 12-mo period (Table II). The rates of
infiltration were not significantly different
among sites (confidence level exceeding
95%) despite the wide variation in hy
draulic conductivity values between the
different soil textures. However, infiltra
tion rates were significantly lower (95%
confidence level) for all sites during the
period from 52 to 54 wk as compared to
the initial 0- to 2-wk period. Variations in
reported rates of 21.6 x 10"4 m/d
( 2.5 x 10~7 cm/s) are due to experi
mental error in measuring such low infil Pfox/g/<3s Column
trations. 45mm0 x 1800 mm
Only the reservoir with no groundwater
control, site no. 4, had to be partially
refilled in May 1982because of its substan
tial infiltration rates during the Winter of
1981-1982. The reservoir being half full of
water when filled in October 1981, it was
presumed that subsequent manure and
water separation during the winter resulted 3 Soil Corp lOOmmOf
in increased seepage from the separated x 100mm
low solids layer.
The substantial difference in infiltration Seeled Cas/ng
rates obtained with manure as compared to
water demonstrated that the sealing mech
anisms were effective, even under cool F/ask
November temperatures (less than 10C).
Because biological and chemical mecha
nisms are known to be temperature-depen
dent, physical mechanisms were pre
sumed to dominate the sealing processes
which had occured at all four reservoirs.
Groundwater sampling could not
provide any definite conclusions pertain
ing to the level of contamination produced
by each reservoir,as only the 1981 October
and November as well as the 1982 April,
May and June samples were taken. Frozen
conditions from November 1981 to April Figure 2. Laboratoryinfiltration columns using 6% total solids dairy slurry and varioussoil types.
1982 inclusive, and very low groundwater
levels from July 1982 to October 1982 did
not allow for further sampling. For the five
samples analyzed, no significant dif
TABLE II. SATURATED HYDRAULIC CONDUCTIVITY AND INFILTRATION RATES FOR
ference (95% confidence level) was found EXPERIMENTAL RESERVOIRS
between the observation wells near the res
ervoirs and the control wells. Accumula Hydraulic Manure infiltration rate
Soil textural conductivity (k) 0-2 wk 52-54 wk
tion of some odorless black matter within
Site class (10"6 m/s) (10~9 m/s) (10~9 m/s)
thereservoirwell wasobservedfor the clay
(no. 1) and the coarse sand (no. 3) sites. 1 Clay 9.25 0-14
2 Loam 3.00 10-24 (Too low to
The analysis of the soil profiles for all be observed)
four reservoirs demonstrated some ac 3 Coarse sand 30-60 24-29 0-18
cumulation of phosphorous, potassium 4 Sand over clay 15 (sand) 24-29 7-23t
and nitrogen (N-Kjeldahl, ammonium and tMeasurements subjectto error broughtabout by rainfall interference.

CANADIANAGRICULTURALENGINEERING, VOL. 29, NO. 2, SUMMER 1987 101

 TABLE III. EXPERIMENTAL RESERVOIR SOIL PROFILE ANALYSIS have acquired a greycolor, indicating some
Concentrations (ppm)
gleysation.
The laboratory trial provided data
Site 4
,

Site 3 (sand over

which could be related to those of field
Depth Site 1 Site 2
(loam) (sand) clay) work. The measured infiltration rates are
Analysis (cm) 0clay)
272 600 600 315 115
summarized in Table IV, while Table V
P,Os 0-5 400 168 390
10-20 270 162 148 126 150 115
102 illustrates the quality of the exfiltrates col
30-40 280

148 135 132 110 135 82 lected. Again, the sealing of soils by
50-60 275
105 97 145 212 67 manure was successful in reducing infiltra
90 116 130 201 114 93 71

tion rates to less than 8.64 x 10~4 m/d

K70 0-5 600 600 600 600 600 600 600 600 (10~ 6cm/s). This was true even for soilsof
600 600 590 575 600 600
10-20 590 515
500 600 310
saturated hydraulic conductivity (k) as
30-40 309 28 206 460
50-60 188
22
504 485 144 41 high as 8.64 m/d (10"2 cm/s). A low cor
90
24 27 328 35 41 22 relation coefficient of 0.32 was established
49 42 21
between the soil sample saturated
NH4 + 0-5

10-20 32 32 10 10 hydraulic conductivity values and their

30-40 8 21 5 7 manure infiltration rates after 5 wk. The
50-60 5 10 8
presence of cool ambient temperatures
90 8 8 10 Trace
(less than 10C) during this trial indicated
58
N-K 0-5 176 81 95 78 75 93
that physical sealingmechanisms played a
61 42 47 64 16 10 25
10-20
8 33 5 8 22

11
major role; biologicaland chemical mech
30-40 22

50-60 22
5 11 8 5 Trace anisms were secondary, as their activity is
90
8 8 14 0 8 2 only significant at temperatures above
15C.
Laboratory column exfiltrate analysis
TABLE IV. LABORATORY COLUMNS INFILTRATION RATES
showed poor correlation between soilclay
content and sample contaminant con
Manure infiltration
centration. The color of the exfiltrate
rate
Hydraulic clearly indicated that the amount of
Column conductivity 48 h 840 d Exfiltrate
(10-6m/s) (10-8m/s) (10-9 m/s) turbidity t organic matter being carried through the
no. Soil
soil was a function of soil clay content.
1 Sand 2.35 6.7 5.0 Heavy
2.27 1.3 6.0 Light Furthermore, comparison of concen
2 Loam
3 Clay 0.06 0.01 2.2 Clear trations among all three clay soil assays
4 Sand 2.11 2.56 6.6 Heavy leads to the probable link of soil structures
5 Loam 3.58 1.50 7.0 Light or permeability to the quantities of exfil
6 Gravelly 1.74 6.6 Light

trated elements. Columns 9, 3 and 8, in

silty clay
7 Loam 23.0 2.55 6.5 Light order of permeability, show a progres
8 Clay 0.00015 0.01 2.5 Clear sively higher cation exfiltration. By con
9 Clay 2.12 0.0036 6.5 Light trast, total organic carbon showed no
tVisual evaluation. relation to hydraulic conductivity.

TABLE V. LABORATORY COLUMNS ANALYSES OF EXFILTRATES COLLECTED DURING

THE SECOND MONTH OF EXPERIMENTATION CONCLUSIONS
The four experimental reservoirs along
Element concentration (ppm)
Column Soil with the nine laboratory columns demon
no. texture TOCt Ca2 + Mg2+ K + N-K P CI
strated the lack of correlation between ini
1 Sand 400 59 567 1360 430 23 167 tial soil hydraulic conductivity (k) and
412 6 25 49 233
2 Loam 62 970 final manure soil infiltration rates. Further
Clay 87 150 156 5 10 1 100
3
524 1260 300 46 180
more, excellent surface sealing leading to
4 Sand 350 126
Loam 74 472 418 3 17 2 225 infiltration rates lower than 8.64 x 10~4
5
6 Gravelly silty clay 54 391 272 2 12 2 165 m/d (10~6 cm/s) were obtained with soils
7 Loam 56 701 503 3 24 3 295 exhibiting initial k values in excess of
6 6 3 1 10
8 Clay 26 15
4 185
0.864 m/d (10~ 3 cm/s), values far exceed
Clay 66 384 495 12 24
9
ing those specified by the Ontario and
tTOC = total organic carbon. Quebec environmental authorities as being
safe for earthen manure reservoirs.
nitrates), resulting from manure infiltra and site clay contents. For site no. 4, with Physical mechanisms seemed to playa
tion. The quantities accumulated varied out groundwater control, elements had major role in the sealing process. Biolog
among sampling locations as well as accumulated within a band some 30- ical and chemical mechanisms tended to
amongsites (Table III). For sites no. 1, no. 60 cm deep. This band formation was be secondary in effect.
2 and no. 3, all elements had accumulated attributed to the higher infiltration rates at Soil clay content and structure were
from the soil surface to a depth varying this site, thus creating a displacement of also found to influence, to some extent, the
between 0 and 10 to 60 cm. This depth the cation band deeper into the soil profile. quality of the seepage exfiltrated from the
varied directly with the elementsolubility All soil reservoir profileswere observed to experimental columns.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

102
 ACKNOWLEDGMENTS
This research project was made possible
with funding from the Quebec Ministry of
Agriculture and the collaboration of the follow
ing farmers: Harold Merson, Gilles Quenne-
ville, Alain Bergeron and Claude Quesnel.
REFERENCES
BARRINGTON, S. F. 1985. The sealing of
soilsby manure. Ph.D. Thesis. Agricultural
Engineering Department, McGill Univer
sity, Montreal, Que.
CHANG, A. C, W. P. OLMSTEAD, J. B.
JOHANSON, andG. YAMASHITS. 1974.
The sealing mechanism of waste water
ponds. J. Water Pollut. Control Fed. 46(7):
1715-1721.
CIRAVOLO, T. G., D. C. MARTENS, D. L.
HALLOCK, E. R. COLLINS, Jr., E. J.
HOMEGAY, and H. R. THOMAS. 1979.
Pollutant movement to shallow groundwater
tables from anaerobic swine lagoons. J.
Environ. Qual. 8: 126-130.
COLLINS, E. R., Jr., T. G. CIRAVOLO, D. L.
HALLOCK, D. C. MARTENS, H. R.
THOMAS, and E. J. KORNEGAY. 1975.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Effect of anaerobic swine lagoons on
ground water quality in high water table
soils. Rages 303-305, 313, in Managing
livestock wastes (Proc. Int. Symposium).
Publ. PROC-275. Am. Soc. Agric. Eng.,
St. Joseph, Mich.
DAVIS, S., W. FAIRBANK, and H.
WEISHEIT. 1973. Dairy waste ponds effec
tively self-sealing. Trans. ASAE (Am. Soc.
Agric. Eng.) 16: 69-71.
HART, S. A. and M. E. TURNER. 1965.
Lagoons for livestock manure. J. Water Pol
lut. Control Fed. 37(11): 1578-1596.
HILLS, D. J. 1976. Infiltration characteristics
from anaerobic lagoons. J. Water Pollut.
Control Fed. 48(4): 695-709.
LO, K. V. 1977. A study of the infiltration
characteristics of dairy waste storage la
goons. Can. Soc. Agric. Eng. Paper 77-
208. CSAE, Ottawa, Ont.
MEYER, J. L., E. OLSAN, and D. BAIER.
1972. Manure holding ponds found self-sea
ling. California Agric. May 1972. p. 14.
MIRTSKHYLARA, Ts. E., G. B. ABE-
LISHVILI, and M. N. TERLETSKAYN.
1972. Effects of microbiological processes
in percolation of water through soil. Proc.
IAHES-ISSS Symposium, University of
Guelph, Guelph, Ontario, pp. 713-721.
NORDSTEDT, R. E., L. B. BALDWIN, and
C. C. HORSTENSTINE. 1975. Multistage
lagoon systems for treatment of dairy farm
waste. Rages 77-80, in Livestock waste
management and Pollution Abatement
(Proc. Int. Symposium). Publ. PROC-275.
Am. Soc. Agric. Eng., St. Joseph, Mich.
OLIVER, J. C, W. C. FAIRBANK, J. L.
MEYER, and J. M. RIBBE. 1974. Subfloor
monitoring of shady grove dairy liquid
waste holding pond. California Agric. April
1974. pp. 6-7.
PATNI, N. K., P. A. PHILLIPS, F. R. HORE,
and J. L. B. CULLEY. 1981. Groundwater
quality near concrete manure tanks and
under heavily manured cropland. Can.
Agric. Eng. 23(1): 37-43.
ROBINSON F E. 1973. Changes in seepage
rate from an unlined cattle waste digestion
pond. Trans. ASAE (Am. Soc. Agric. Eng.
16: 95-97.
SEWELL, J. I. 1978. Dairy lagoon effects on
groundwater quality. Trans. ASAE (Am.
Soc. Agric. Eng.) 21: 948-952.
103
 THE SEALING OF SOILS BY MANURE.
II. SEALING MECHANISMS

S. F. Barrington1, P. J. Jutras2, and R. S. Broughton1

Agricultural Engineering Department, McGill University, Box 950, Macdonald College, Ste. Anne de Bellevue,
Quebec H9X ICO; and 2USAID, do Embassade des Etats Unis, Dakar, Senegal
Received 2 May 1986, accepted 15 October 1986

Barrington, S. F., P. J. Jutras, and R. S. Broughton. 1987. The sealing of soils by manure. II. Sealing mecha-
nisms. Can. Agric. Eng. 29: 105-108.

The sealing mechanisms for soils being infiltrated by dairy and swine slurries were investigated using laboratory
infiltration columns. Physical mechanisms were found to be predominant especially for dairy slurries. Biological
mechanismsinterveneto strengthen physical seals only where ambienttemperatures exceed 15C. Biological mechanisms
are especially important in the sealing of soils by swine slurries as their physical mechanisms are weaker than those of dairy
slurries. Chemical mechanisms were found to be insignificant.

INTRODUCTION
A preliminary project (Barrington et al.
1987) established that physical mecha
nisms play a primary role in the sealing of
nutrients accumulated within the reservoir TABLE I. PARTICLE SIZE ANALYSIS OF
COARSE SAND UTILIZED IN
profiles. Miller and Robinson (1981) also
INFILTRATION COLUMNS
monitored groundwater quality below a
Fraction
beef manure earthen reservoir to observe Particle size

the presence of nitrate and chloride micrometer (%)

soils by manure. Furthermore, dairy
manures of 6% and 10% total solids (TS) seepages. Barrington et al. (1987) report <2 Trace

were successful in sealing soils to infil several other authors having monitored 53-2 Trace

tration rates below 8.64 x 10-4 m/d groundwater quality around earthen 53-75 1
75-150 9
(10~6 cm/s), even on soils with initial per manure storages. Some of the authors 150-250 82
meability values exceeding the limits set observed seepage effects but not under a >250
by the Ontario and Quebec environmental systematic basis.
authorities. This project, following from that of
These findings were in agreement with Barrington et al. (1987), aims at investi by returning to the individual column, the
those of several previous research works. gating the specific mode of action as well liquids collected as seepages (exfiltrates).
DeTar (1977, 1979) demonstrated, using as the relative importance of the physi The second trial pertained to the sealing
manure seeping throughsiltysoils, thatTS cal, biological and chemical sealing effects of chemical mechanisms. Eight
content above 1.8% was far more signifi mechanisms. clay soil samples, 0.048 m in diameter by
cant in controlling final infiltration rates 0.010 m high, were compacted at the bot
than either soil saturated hydraulic con PROCEDURE tom of aluminum tubes (Fig. 2). After
ductivity to water(k) or manure hydraulic The investigation of the three sealing measuring the saturated hydraulic con
head. Lo (1977)found, using dairy lagoon mechanisms was carried out through four ductivity of each clay core, four columns
liquids over columns of various soil tex laboratoryinfiltration trials. Exceptfor the were soaked in tap water and four more
tures, that final infiltration rates were not chemical mechanism investigation, all columns were soaked in swine slurry. All
significantly different despite wide varia assays were performed using plexiglas eight columns were kept under tempera
tions in hydraulic heads. Rowsell (1980) tubing of 0.14 m insidediameter. A coarse tures of 0-5C for minimum bacterial
compared sealing effects of screened beef sand column, 0.15 m high, was placed at activity. Carewastaken as to soaktheclay
manure, formaldehyde-sterilized screened the lower extremity of the tubing and was samples from the bottom of the tubes to
beef manure and two salt solutions on vari subjected to various constant manure prevent any manure solids from infiltrating
ous soil textures. He concluded (1) that the hydraulic heads (Fig. 1). Both dairy and from the surface and thus creating a phys
sealing seemed to reside within theorganic ical seal. The saturated hydraulic con
swine slurries were used for these trials,
mat which accumulated at the soil surface, the former containing 9% total solids (TS) ductivity of each column was measured
(2) that biological sealing mechanisms and latter 6% TS. The particle size analy initially as well as after 1 wk and 2 wk of
were insignificant and (3) that chemical sis for the coarse sand is presented in soaking. These permeability measure
mechanismsresulted perhaps from the soil Table I. ments require the infiltration of a volume
destructuring effect of the manure's neu The first trial compared sealing effi of water, representing four times that of
tral pH. ciency between physical and biological the soil cores. Thus, the cores' per
Miller etal. (1976) sampled four earthen mechanisms. Using the setup described in meabilityto water was expectedto change
manure reservoirs, two in use for 2 yr and Fig. 1, formaldehyde-sterilized slurries between trials and soaking periods. Nev
two others in use for 8 yr. Their sampling (Rowsell 1980) and natural slurries were ertheless, chemical sealing effects should
indicated significant contaminant accumu tested for infiltration rate over coarse sand produce an additional change in per
lation as deep as 150 cm below the reser columns 0.14 m in diameter by 0.15 m in meability among core groups, if signifi
voir soil-manure interface. Sampling only height. The coarse sand was assumed to cant. Analysis of variance was used as
four reservoirs, Miller could not establish demonstrate minimal chemical activity. statistical method to compare the effects of
a correlation between soil clay content or Slurry hydraulic head as well as total the two treatments, water and swine slurry
reservoir age and the quantities of solids content were maintained constant soaking.

105
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 TABLE III. HYDRAULIC CONDUCTIVITY
CHANGES RESULTING FROM SOAKING
WITH SWINE SLURRY

Hog slurry Water

soaked cores soaked cores
Time
(d) Mean SD Mean SD

Hydraulic conductivity values (10 5 m/s)

0 6.70 3.49 5.64 3.16
7 3.63 1.43 4.13 1.83
14 1.75 0.38 1.50 0.50

Confidence level of 95% for no significant differ

ences among treatment.

TABLE IV. INFILTRATION RATE FOR

P/pvig/as Column VARIOUS MANURE HYDRAULIC HEADS

Average infiltration ratet

(10"9m/s)
750
450 Manure
nyurauuc neau
750 type 125 mm 440 mm 700 mm
mm
Dairy 16 22 24
Rfezometers Hog 65 27 42

t After400 h of infiltration and from two replicates

per manure type hydraulic head combination.

contents of 3, 6 and 9% were investigated

50 Sand Core
as to sealing efficiency, over fine sand
columns.
Heat-treated and natural dairy slurries
75
(both 6% TS) were compared for infiltra
Reducer tion rates over coarse sand columns. The
heat-treated dairy slurry was subjected to
121C for 15 min under 105 kPa of steam
Collation Rrdkrr pressure. This heat-treated slurry was
demonstrating a finer particle size distribu
tion from laboratory analysis using soil
screens and a water washing method. This
last trial was performed using aluminum
tubing 0.048 m in diameter by 0.580 m in
height, and a slurry hydraulic head of
0.45 m.

Figure 1. Experimental infiltration columns with piezometers to measure pressure drops. RESULTS AND DISCUSSIONS
The comparison of sealingperformance
TABLE II. RELATIVE PERFORMANCE OF PHYSICAL AND BIOLOGICAL SEALING between physical and biological mecha
MECHANISMS nismsis presented inTableII. Hogslurries
Infiltration ratef
were observed to ferment heavilyfrom the
(10~9 m/s"1) 12th to 48th h of eachtrial. Thisbiological
Temperature activity was, by far, greater than that of
Assay Dairy slurry Hog slurry (C)
dairy slurry. This superiority probably
Initial (water) 72 000 72 000 0-5 resulted from the better biodegradability
Sterilized slurry or lower fiber content of the hog manure
(physical mechanism) 8.0 400
Nature slurry
0-5
solids. The chemical sealing effects were
(physical and found to be insignificant, as no significant
biological mechanisms) 6.5 70 0-5 difference was found between the
7 20-30 hydraulic conductivities of the clay col
tlnfiltration rates are reported for an hydraulic head of 0.700m above thecoarse sand column. umns soaked in hog liquids and those
soaked in water (Table III).
The use of piezometers duringthe dairy
A fourth procedure was used to investi sand columns, with two replicates each for and hog slurry infiltration trials demon
gate various physical sealing mechanisms. dairy and hog manure. Piezometers were strated pressure gradient losses directly
Manure hydraulic heads of 0.125, 0.440 used inside the columns to locate the above and at the soil-manure interface but
and 0.700 m were established over coarse sealed layers (Fig. 1). Also, total solid not below. It was therefore concluded that

106 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 TABLE VI. INFILTRATION RATES FOR
HEAT-TREATED AND NATURAL DAIRY
SLURRIES (6.0% TS)

Infiltration ratef
Slurry type (10-9 m/s)
Heat-treated 1040 260
Natural 120 44

t Range indicates a 95% confidence interval obtained

with eight replicates for each slurrytype, for infiltra
Aluminum Cylinder tion rates after 10d and, hydraulic head of 0.40 m
48 mm 6 x 580 mm above the coarse sand column.

sealing mechanisms such as gleization

(Barrington 1985).
Recognizing the importance of the
physical sealing mechanisms, their mode
of action was further investigated. The
use of several different hydraulic heads
produced no significant changes on infil
tration rates (Table IV, level of confidence
exceeding 95%). This was probably the re
Chij Core sult of a thicker mat formation with
increased manure head.
Toffon Mexh
Variations in TS of hog slurries over
Soaking Reaker columns of fine sands failed to produce
any conclusive results, as no correlation
/ / / / / ///>/// could be established between infiltration
rates after 400 h and manure TS content.
Nevertheless, this assay demonstrated that
the hog manure mat accumulating at the
soil surface had a tendency to gain per
meability under cool conditions or under
weak biological mechanisms. This
occurred through the leaching out of the
finer organic particles, manure slurries
solids not being uniformly distributed in
particle sizes (Table V). Such leaching
was not noticed for dairy slurries as the
sealprobably forms as a randomly deposi
ted fibrous mat consisting of a more even
particle size distribution (Table IV). This
leaching out process of the hog mat was
observed while the slurry was fermentat-
Figure 2. Clay columns utilized in the evaluation ofchemical sealing mechanisms. ing under warm ambient conditions (over
15C) duringthe second dayof infiltration.
TABLE V. PARTICLE SIZE DISTRIBUTION FOR MANURE SOLIDS
The gas bubbles produced through fer
(Overcash et al. 1983) mentation agitated the mat, settling the
finer particles at the soil surface. Under
Particle size distribution for manure solids (%) warm conditions, these fine particles
Manure
<50|xm 50--250 |xm 250 -500 |xm > 7500 |xm would cake in a solid layer at the soil
type

Hog 45 2.5 2.5 50 interface, produce a second mat and

Dairy 36
Poultry 45
Beef 30
the sealing was taking place within the
organic mat accumulating over the soil
column at the soil-manure interface. This
illustrates the basic screening role of the
soil leading to the establishment of the
impermeable organic mat. The efficiency
of the sealing depends on the ability of
the soil to retain the manure solids at its
surface and to trap them within its sur
10 8 46
10 12.5 32.5
13 17 40
face pores. Soil voidgeometry anddimen
sion become more important than soil
hydraulic conductivity (k) with regard to
sealing. It also appears that soil particle
size distribution, rather than soil k, could
better describe a soil's ability to become
sealed by manure. Soil aggregate size
intervenes to a lesser degree as it is
believed to be destroyed by biochemical
strengthen the seal. No such activity was
noticed with hog slurry which had been
sterilized or for columns under cool tem
peratures (less than 5C), thus suggesting
the work of biological mechanisms.
Infiltration rates for a dairy slurry of
finer particle size solids (heattreated) and
of natural, coarser particle size solids,
over coarse sand columns, were signifi
cantlydifferent (Table VI). Slurry particle
size was found to influence physical seal
ing mechanisms and exfiltrate total
organic carbon content (Table VII).

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 107

 TABLE VII. EXFILTRATE CATION CONCENTRATION Chemical sealing mechanisms through
NATURAL ANDHEAT-TREATED DAIRY MANURE OVER COARSE SAND their deflocculating effects on clay parti
Exfiltrate quality (ppm)f cles were insignificant. This does notelim
S04 CI
inate the possibility of biochemical soil
Exfiltrate NH4 N03 Ca Fe Mg K Na P
reduction resulting in soil structure
11 43 1230 49 8 90 153
Natural 55 0.1 56 destruction or gleysation.
242 200 890 116 170 120 506
Heat-treated 660 1.6 21
Although the sealing extent of soils in
tExfiltrates collected after 10 d of infiltration. contact with manure is of the order of 8.64
x 10"5 m/d (10_7 cm/s), groundwater
TABLE VIII. EXFILTRATE ANALYSIS FOR COLUMN OF NATURAL DAIRY AND HOG contamination is still probable for soils of
SLURRIES OVER COARSE SAND low CEC especially under conditions of
Ca Mg Mn K Na NH4
long exposure to hog manures.
Total
Exfiltrate (meq/L)t ppm

Dairy$ 60 191 104 3.2 104 55 60 ACKNOWLEDGMENTS

The author wishes to thank the Quebec Min
Hog
istry of Agriculture for its financial help, and
After 1 wk 224 302 139 4.2 35 168 650
After 2 wk 254 280 252 12.4 846 284 850
Alvin Barrington and Jean Aime Parent for the
After 6 wk 284 163 428 8.0 1560 500 1200 provision of the required manure slurries.

t Values corrected for 700 ppm and 3000ppm of NH4 for dairy and hogexfiltrates, respectively.
tExfiltrates collected after 50 d of infiltration.
TABLE IX. SOIL SATURATION RATEt
FOR EARTHEN RESERVOIR PROFILE
Cation saturation rate
Soil
(m/yr)
texture CEC* Hog manure Dairy manure
Sand
5 0.18 0.04
Clay 25 0.04 0.01
tThese rates assume 100% soil saturation.
$CEC, cation exchange capacity.
Throughout this project, exfiltrate sam
ples were analyzed for nutrient and min
eralcontent. Because most of the samples
were collected from coarse sand columns,
soils of low filtration capacity, the quality
of theseliquids wasassumed equivalent to
that seepingtowardsthe groundwaterfrom
earthen reservoirs. Because of ammonia
volatilization during collection, sample
cation concentrations were corrected in
Table VIII to more probable ammonia lev
elsof 700 and 3000 ppm fordairy and hog
exfiltrates, respectively. These new levels
of ammonia (based upon data from the
Quebec Ministry of Environment) gave
final liquid cation concentrations of 60 and
280 meq/L. Table IX illustrates the rate of
cation saturation for the soil of earthen
reservoir profiles subjected to a regular
infiltration rate of 8.64 x 10~5 m/d
REFERENCES
BARRINGTON, S. F. 1985. The sealing of
(10-7 cm/s). Assuming a groundwater soils by manure. Ph.D. Thesis. Agricultural
table some 60 cm below a hog manure Engineering Department of McGill Univer
earthen reservoir, contamination would sity, Montreal, Quebec.
become obvious within 3.3 and 15 yr for BARRINGTON, S. F., P. J. JUTRAS, and
sandy and clay soils, respectively. This R. S. BROUGHTON. 1987. The sealing of
soils by manure. Part I. Preliminary inves
analysis indicates a need for further
tigations. Can. Agric. Eng. 29: 99-103.
groundwater protection for soils of low DeTAR, W. R. 1977. The concentration of
cation exchange capacity (CEC), even liquid manure effects its infiltration into
though infiltrations were reduced to 8.64 soils. Annual Meeting, Am. Soc. Agric.
x 10-5 m/d (IO-7 cm/s). Eng., St. Joseph, Michigan. Paper no.
Column assays throughout the project, 77-2060.
indicated a final infiltration rate tending DeTAR, W. R. 1979. Infiltration of liquid dairy
towards 8.64 x 10~5 m/d (10~7 cm/s). manure into soil. Trans. Am. Soc. of Agric.
Perhaps this represents the maximum seal Eng. 22:520-528,531.
ing extent of soils in contact with manure. LO, K. V. 1977. A study of the infiltration
characteristics of dairy waste storage
lagoons. Can. Soc. Agric. Eng. Paper
CONCLUSIONS
77-208. CSAE, Ottawa, Ont.
The investigation of individual sealing MILLER, M. H., J. B. ROBINSON, and
mechanisms demonstrated the superiority D. W. GALLAGHER. 1976. Accumulation
of thephysical processes whereby organic of nutrients in soil beneath hog manure
solids are trapped within soil pores and at lagoons. J. Environ. Quality5(3): 279-282.
the soil surface. The efficiency of this MILLER, M. H. and I. B. ROBINSON. 1981.
physical process is a direct function of soil Natural sealing of earthen manure storage
void geometry as well as dimension, and ponds. Workshopon swinemanure manage
manure solids particle size distribution. ment technologies, Agriculture Canada,
Biological sealing mechanisms inter Ottawa, Ontario, Canada, pp. 72-90.
OVERCASH, M. R., F. J. HUMENIK, and
vene as a function of manure solids bio-
J.R. MINER. 1983. Livestock waste man
degradability and ambient temperature. agement. CRC Press, Boca Raton, FL.
This mechanism binds manure solids to ROWSELL, J. G. 1980. Infiltration of liquid
soil particles thus strengthening the phys manure into soil materials. M.Sc. Thesis,
ical sealing process. Guelph University, Guelph, Ont.

108
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 SOIL EROSION LOSSES UNDER FREEZE/THAW AND WINTER
GROUND COVER USING ALABORATORY RAINFALL SIMULATOR
Linnell M. Edwards1 and J. R. Burney2
'Research Station, Agriculture Canada, Charlottetown, Prince Edward Island CIA 7M8 and
Department ofAgricultural Engineering, Technical University ofNova Scotia, Halifax, Nova Scotia
B3J 2X4

Received 21 May 1986, accepted 15 October 1986

Edwards, Linnell M. and J. R. Burney. 1987. Soil erosion losses under freeze/thaw and winter ground cover using a
laboratory rainfall simulator. Can. Agric. Eng. 29: 109-115.

Alaboratory rainfall simulator was used to test three Prince Edward Island agricultural soils (varying in soil texture) for
runoff and splash volume and sediment loss under varying conditions of freeze/thaw, ground cover and erosivity. Wooden
soil boxes and ancillary collection frames (termed cassettes) were designed tofit four atatime under the rainfall simulator.
With bare soil, freeze/thaw significantly increased sediment loss by about 90%: additionally, sediment in runoff varied
significantly with soil type and, for a loam soil, was 15 and 31%, respectively, ofthe amounts for a fine sandy loam and a
sandy loam. Where the soil was seeded to a winter ryecover, sediment loss was reduced by 70-80% with no significant
effect of soil type or freeze/thaw. Where, however, theerosive force was increased by adding overland flow to simulated
rainfall, there was a significant increase insediment loss even with ground cover. Sediment splash was sampled foralltests,
and only ground cover indicated a significant effect.

INTRODUCTION
In the Atlantic Provinces of Canada,
as in many other parts of the world, the
Universal Soil Loss Equation (USLE) is
routinely used to predict soil loss on agri
cultural land. However, as with any statis
tically fitted regression model, the
application accuracy of the USLE is con
fined to the range and nature of its data
base. In particular, the data base used for
the USLE is predominantly continental
U.S.A. The erosivity factor (R) is a mea
sure of the summed product of the kinetic
energy (E) and the maximum 30-min
intensity (/) within each storm period over
an average year. The data base therefore is
heavily biased towards high intensity sum
mer storm conditions as a causative factor
and takes no account of erosion caused by
non-precipitation events such as snow-
melt runoff or the stress of alternate freez
ing and thawing.
The applicability of the USLE under
the light rainfall conditions of the Pacific
Northwest of the U.S.A. was examined by
McCool et al. (1982) and Onstad and
Young (1982). In general it was found that
the erosivity value (R) and the cropping
value (C) required substantial modifica
tion. In a related sense, Morgan (1979)
presented European data which showed
substantial soil loss for low intensity, long
duration storms on saturated soil.
On Prince Edward Island (P.E.I.) soil
erosion is a major concern particularly on
potato lands where the crop is often culti
vated up and down the slope, and the soil
usually remains bare during the cool-sea
son after potato harvesting. The cool sea
son is moist and the soils could easily be at
or near saturation during periods of snow-
melt. The cool season of PE.I. is charac
terized by frequent showers and occur
rences of soil freeze/thaw cycles; and most
of the soil erosion in P.E.I, seems to occur
during this period under these predispos
ing conditions.
The effects of frost action on soil sta
bility have been studied to some extent
(Logsdail and Webber 1959; Sillanpan and
Webber 1961; Hinman and Bisal 1968;
Benoit 1973; Richardson 1976), and results
have varied from no effect to an effect in
either direction, viz. improved aggrega
tion or degradation depending on soil tex
ture, initial soil moisture, freezing rate and
temperature. Undercircumstances of slow
freeze/thaw cycling with high initial
moisture in the soil, as typify PE.I. natu
rally and as prevailed in this experiment,
Bryan (1971) found significant soil phys
ical degradation and thus susceptibility to
erosion.
The first documented study on soil ero
sion in PE.I. (Harza of Canada Ltd. 1968)
advocated, among other things, that the
erosivity factor (R) and erodibility factor
(K) in the USLE be independently evalu
ated in the Atlantic Provinces. Wall et al.
(1976) computed rainfall erosivity (R)
values for Canada east of the Rocky Moun
tains and Wall and Dickinson (1979) and
Wall et al. (1983) developed seasonal rain
fall distribution curves. All of these are,
however, limited in applicability under the
stochastic erosivity and erodibility condi
tions which prevail on PE.I.
The only available soil loss data for
P.E.I, were reported by Himelman and
Stewart (1979) based on observations of
erosion plots on a fine sandy loam soil
between 1973 and 1977 inclusive. Yearly
soil loss on a 12 and 7% slope averaged 411
ha-1 and 19 t ha-1, respectively, under
fallow conditions; 20 t ha-1 and 10 t
ha-1, respectively, under potatoes culti
vated along the slope; and 7 t ha~ *and 41
ha-1, respectively, across the slope. Fur
thermore, erosion during the snow-melt
period of March-April contributed sub
stantially to total annual erosion load, thus
strengthening general observations in
P.E.I, that erosion losses are highest dur
ing the cool season. Stewart (1979) con
tends, moreover, that the low erosivity
value (R) of 50-60 projected from rainfall
data alone seems inadequate to account for
annual soil losses in P.E.I, which com
monly exceed 10 t ha_i. It would there
fore appear that both increased erosivity
forces and factors which markedly reduce
soil erodibility resistance substantially
increase erosion during the cool seasons.
None of the aforementioned studies on
soil erosion in P.E.I, provided definitive
information on the quantitative impact of
cool-season meteorologic processes on
soil loss or runoff. The present study was,
therefore, instituted using soil boxes and
ancillary collection frames (termed cas
settes) in a laboratory rainfall simulator to
investigate the effect of freezing/thawing
on P.E.I, soils under the influence of a cool
season erosive force and ground cover.
MATERIALS AND METHODS
General
Samples of three agricultural soils of
P.E.I. (Table I), taken following a barley

109
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 TABLE I. SOME PHYSICAL CHARACTERISTICS OF THREE PRINCE EDWARD ISLAND
Dunblane
Characteristics
Organic matter (%)
Textural analysis
Clay (%)
Silt (%)
Fine and very fine sand (%)t
Sand (%)
Bulk density (Mg rrr3)
Hydraulic conductivity
(cmrrl)
Aggregate stability {%)$
tSand fractions between 250 Ltm and 50 luti.
tKemper (1965).
crop, were tested in a laboratory rainfall
simulator for runoff volume and soil loss.
Two replicates of a factorial arrangement
were used to test the following factors:
three soils (Charlottetown fine sandy
loam, Dunstaffnage sandy loam and Dun
blane loam); two temperature states
(frozen and unfrozen); three ground cover
states (bare soil, winter rye early cover
seeded 30 Sept. and late cover seeded 31
Oct.); two erosion agents (simulated rain
and simulated rain plus overland flow).
There were 72 experimental units.
Equipment
Seventy-two wooden soil boxes (Fig. 1)
were constructed from 18-mm Douglas fir
select plywood which was pressure-treated
with a preservative. A frame of 50 x 50 x
3-mm rust-proofed steel angle provided
110
were designed to fit easily over any ofthe
soil boxes during rainfall simulator runs. A
Charlottetown fine Dunstaffnage soil box complete with splash frame and
loam sandy loam sandy loam runoff collector, termed a soil cassette, is
2.7 3.1 3.7 shown on the left in Fig. 4. Four splash
frames and four runoff collectors were
18.0 11.7 5.6 constructed so that four soil cassettes could
37.1 34.0 23.7
44.4 53.3
be run at once in the rainfall simulator.
37.4
44.9 54.3 70.7 Each splash frame was constructed
1.09 1.18 1.25 from 22-gauge (0.84-mm) galvanized iron
26.2 26.2 28.2 and consisted of the following main parts:
(1)A continuous splash collection
97.0 86.1 80.1 trough which ran across the top and down
each side; the edge above this trough lying
directly above the inside edge of the box,
and the section below the trough fitting
support along the bottom and side snuggly into the inner edge of the box.
plywood butt joints. The frame was firmly (2) A cover at the lower end to prevent
fixed to the plywood by stainless steel raindrops from falling onto the flow con
screws, and evenly spaced 20-mm-diame- centrating section.
ter drain holes were drilled in the bottom of (3) A small perforated pipe across the
each box. A sheet of plastic flyscreen over upper end to spray water against the inside
the inside bottom of each box prevented back edge and from there onto the soil
soil loss through the drain holes. Two sta surface. Each pipe was designed to deliver
ples were welded to each end of the boxes water at an equivalent rate of 150 mm h ~'
and a short length of 20-mm rope was evenly across the top end of a soil box,
fitted to facilitate handling. effectively extending the plot length.
The soil boxes were designed to have a The runoff collector was fitted sepa
45 bevel on the overflow, lowerend (Fig. rately over the sill of the soil box. For each
1). In order to provide for a sharp raindrop run, the edges between the splash frame
separation edge for erosion tests, a con and the runoff collector were sealed with a
centration of runoff for collection at the silicon sealer.
lower end, collection of splash, and the As shown in Fig. 5, the basic rainfall
provision of overland flow simulation at simulator unit consisted of a 3.05-m-long
the upper end, a specialsplash frame (Fig. x 0.94-m-wide and 0.25-m-high soil
2) and a separate runoff collector (Fig. 3) trough suspended beneath a matrix of
SCALE:
NOTE'. ALL VALUES IN mm.
18 mm PRESSURE
TREATED PLYWOOD
Figure 1. Soil box.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 SCALE:

500
230
905
NOTE:

1. ALL VALUES IN mm
2. MATERIAL: 22 gge g.i.

PLAN

GZ mr
r. 130
l END
ELEVATION ELEVATION

Figure 2. Splash frame which fits over a soil box shown in Fig. 1.

box. Drops were formed at the ends of 25-

mm lengths of 0.58-mm i.d. polyethylene
scale: tubing spaced on a 32-mm matrix in the
lower side of each box. Each of the inlet
nozzles was connected to the equivalent
200
nozzle in each of the other boxes in the 2 x
7 matrix of boxes. Water flow to each of the
four sets of nozzles was controlled by a
N0Tr 1. Al L VALUES IN mm
solenoid valve. By opening combinations
2. MATrKIAL: ?? gge g.i.
of the four controlling solenoid valves,
rainfall could be instantly applied over the
soil trough area at rates from 25 to 187mm
h-1 in 12.5-mmh-1 increments. Rainfall
could also be stopped instantaneously
when the master switch closed all the sup
ply solenoids. The soil trough beneath the
PLAN
rainfall boxes could be tilted up to a 5%
slope.

ELEVATION
Figure 3. Runoff collector which fits overthe sill of a soil box shown in Fig. 1.
drop-former perspex boxes. The entire
apparatus was enclosed in an insulated
cooling chamber.
The rainfall drop-formers consisted of a
set of 430-mm-square x 30-mm-deep
SILL FLEVATION
hollow perspex boxes based on the design
of Chow and Harbaugh (1965). Four
nozzles in the upper surface of each box
injected water at rates equivalent to 12.5,
25, 50 and 100 mm h ~' over the area of the
Procedure
Soil boxes were filled in the field at
three sites of different soil types. Care was
taken to maintain the original profile orien
tation of the soil. The boxes were taken to a
common outdoor site, cover crop treat
ments seeded to Kodiak winter rye at 4 g
box-' (140 kg ha- *) and left under natu
ral meteorologic conditions until mid-
November when they were moved to the
laboratory. All boxes were kept at near
field capacity through regular watering.
Rainfall simulator runs were done in three
time blocks: first, with bare soil treat
ments; second, with early seeded treat
ments; and third, with late seeded
treatments. For each time block, 12 boxes

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 HI

 Figure 4. Assembled soil cassette in rainfall simulator trough (left) and minus splash frame (right).

INLET NOZZLE MANIFOLD -SOLENOID

-R 15 INSULATION x 4/BOX x4 VALVE x 4

-WATER
SUPPLY

VACUUM
LINE
REFRIGERATION
UNIT
2x7 MATRIX OF HOLLOW
0.43 m SQ. PERSPEX BOXES
-RIBBED
ALUMINUM
SHEET
(WALLS AND
CEILING) -TROUGH SUPPORT ROD x 4

SOIL TROUGH
HINGE-
SCISSOR
JACK

Figure 5. Rainfall simulator in insulated and refrigerated enclosure.

112 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 were placed at field capacity in cold stor TABLE II. MEANS OF MAIN EFFECTS
age at - 15C for 10 d, and the remaining Mean value
12 were subjected only toa laboratory tem Runoff
perature of about 20C. Splash
Volume Sediment Volume
For practical reasons, the frozen treat Sediment
Cell (mL) (g) (mL) (g)
ments and the non-frozen treatments were Crop
run at opposite ends of each time block at Bare 4191a 60.2a 434a 1.50a
ambient temperatures of 1C and 20C, Early seeded 3917a 10.1/7 390a 0 54/7
Late seeded 4943a 17.0/7 437a O.86/7
respectively. For each subset of four boxes
Soil
of any particular soil type within the tem Loam 2226/? 9.4c 407a 0 82a
perature set, two boxes were selected at Fine sandy loam 5888a 47.6a 419a 1 03a
Sandy loam
random to receive rainfall plus overland 4937a 30.3/7 434a 1.05a
(upslope) flow, and the remaining two to Temperature
Normal 3287a 21.2/7
receive rainfall application only. Twelve Frozen 4874a 37.0a
397a
444a
0.84a
1.09a
cassettes were then run randomly, four at a Erosion agent
time. All tests were run under a rainfall Rainfall 1402/7 12.2/7 411a 1 1 la
setting of 50 mm h~l for a period of 30 Rainfall + overland flow 7299a 46.0a 430a 0.83a
min. a-c Means followed by the same letter in any column for any treatment are not significantly different at
In total, six rainfall simulator runs were
conducted (four cassettes per run) for the
bare soil block. Prior to each run, the rain TABLE III. LEVELS OF SIGNIFICANCE OF MAIN EFFECTS AND INTERACTIONS
fall rate was calibrated byusing a tarpaulin
Runoff Splash
under the rainfall boxes for a period of Source of
variation Volume Sediment
5 min. Fourboxes were then placed in the Volume Sediment

soil bin pre-set to a 5% slope. A splash Main effects

Ground cover status (GrC) NS **
NS **
frame was set over each box and a runoff Soil type (S) ** **
NS NS
concentrator inserted over each sill. An Temperature status (T) NS *
NS NS
Erosion agent (EA) ** *M=
overland flow applicatorpipe was inserted NS NS
Two-way interactions
in each of the appropriate boxes. Unused GrC x S **
NS NS **

pipes were left on the bottom of the rainfall GrC x T NS NS NS NS

simulator trough where their discharge GrC x EA NS **
NS NS
S x T #*
NS **
preserved the calibration of the four pipes S x EA ** **
NS
NS NS
as a unit. Siliconsealer was used along the T x EA NS NS NS NS
joint between the splash frame and the Three-way interactions
runoffcollectorto preventleakage. Splash GrC x S x T NS NS NS NS
GrC x S x EA NS **
NS
was collected from the side troughs of the GrC x T x EA NS NS NS
NS
NS
splash frame and runoff was collected S x T x EA **
NS NS NS
from the end of the runoff collector for *,**/> = 0.05and/> = 0.01, respectively; NS == not significant at P = 0.05.
each cassette.
Following each run, the volumes of
runoff and splash from each of the four
TABLE IV. MEANS OF MAIN EFFECTS BARE SOIL
cassettes were measured and separately
retained for sediment determination Mean value

together with the wash from the runoff Runoff Splash

collector, splash trough and the collection Volume Sediment Volume Sediment
and measuring vessels. Cell (mL) (g) (mL) (g)
Sediment dry mass for both runoff and Soil
Loam 1673/7 17/7 413a 0.82/7
splash from each soil box was obtained by Fine sandy loam 6179a 110a 424a \.12ab
filtering through a 0.47-u.m filter, then Sandy loam 4721a 54/7 465a 1.97a
oven-drying the retained soil at 105C. Temperature
Data on runoff volume and sediment loss Normal 2767/7 42/7 466a 1.38a
Frozen 5615a 79a 401a
from bothrunoffand splashweresubjected 1.63a
Erosion agent
to analysis of variance and mean separa Rainfall 1797/7 26/7 407a 1.59a
tion. Rainfall 4- overland flow 6585a 94a 461a 1.41a
a,b Means followed by the same letter in any column for any treatment are not significantly different at
P = 0.05. & J
RESULTS
As reflected in Table II, ground cover
significantly decreased runoff sediment sediment loss by 75%. Sediment loss and compared to the sandy loam. When over
loss to 17% of that for bare soil for early runoff volume varied significantly with land flow was added to rainfall (to form a
seeded rye and to 28% for late seeded rye, soil type. The loam soil showed only 20% single erosion agent), runoff sediment loss
and also significantly decreased sediment of the runoff sediment loss and 38% of the increased 277% and runoff volume 420%.
splash to 36% for early-seeded rye and runoff volume as compared to the fine As Table III shows, therefore, all of the
57% for late seeded rye. The freezing pre- sandy loam, and 31% of the runoff sedi main effects, viz. ground cover status, soil
treatment significantly increased runoff ment loss and 45% of the runoff volume as type, temperature status, and erosion agent

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 113

 TABLE V. MEANS OF MAIN EFFECTS EARLY SEEDED
Mean value
Splash
Runoff
Sediment Volume Sediment
Volume
(mL) (g)
Cell (mL) (g)
Soil
1143/7 3a 382a 0.74a
Loam
Fine sandy loam 6163a 18a 404a 0.46a/?
383a 0.40/7
Sandy loam 4147a 10a
Temperature
368a 0.44a
Normal 3504a 10a
Frozen 4330a 10a 411a 0.63a
Erosion agent
Rainfall 1082/7 4/7 409a 0.69/7 the influence of rainfall plus overland flow.
370a 0.39a
Rainfall + overland flow 6753a 17a
This is a significant saving of soil.
a,b Means followed by the same letter in any column for any treatment are not significantly different at
P = 0.05.
were significant on runoff sediment loss. of maintaining soil profile orientation, and
Soil type and erosion agent significantly in testing four experimental units at once,
affected runoff volume. Ground cover sta thus quadrupling the turn-over rate of the
tus significantly affected sediment splash. original single-trough design.
Significant two-way interactions on sedi As indicated previously, Bryan (1971)
ment loss occurred between ground cover found significant soil aggregate degrada
status x soil type, ground cover status x tion and thus susceptibility to erosion
erosion agent, and soil type x erosion under conditions of slow freeze/thaw cyc
agent; while significant three-way inter ling with high initial soil moisture content,
action on sediment loss occurred between such as typify P.E.I, and as prevailed in
groundcoverstatus x soil type x erosion this experiment. Such a finding supports
agent. Significanttwo-way interactionson the significant increase in runoff volume
runoff volume occurred between soil type and sediment loss on bare soil observed in
x temperature status, and soil type x this study due to freeze/thaw. The results
erosion agent; while significant three-way of the application of this technique to the
interaction on runoff volume occurred agricultural soils of P.E.I, point unmis-
between soil type x temperature status x takeably to the importance of ground cover
erosion agent. Significant two-way inter in soil and water conservation. For,
action on sediment splash occurred whereas with ground cover, temperature
between ground cover status x soil type, status had no significant effect on sediment
and soil type x temperature status. loss when the soil was bare, freeze/thaw
Separateanalysis of varianceon data for caused significant increases in sediment
ground cover status showedthat, with bare loss and runoff volume as previously
soil, the freezing pretreatment increased shown.
the sensitivity of the tests. Temperature Based on the results of this study, there
status had a significant effect on sediment were distinct differences between soil
loss and runoff volume from bare soil, but types in their basicabilityto resisterosion;
had no significant effect on either the but, again with a cover crop, the severity of
early- or late-seeded (ground cover) treat erosion was reduced to such an extent that
ments. As the treatment means for bare differences in sediment loss were insignifi
soil in Table IV show, the freezing pretreat cant.
ment significantly increased sediment loss The greater resistance to erosion shown
as well as runoff volume by 88% and by the loam soil in this study was based on
103%, respectively. However, as indicated a textural advantage. It showed no advan
in Table V freezing pretreatment had no tage in organic matter content, but had a
significanteffect on either sediment loss or higher clay content than either of the other
runoff under ground cover conditions. two soils, which is a highly desirable
characteristic for good aggregate stability
and, in turn, for successful soil and water
DISCUSSION conservation management. It is imprac
tical, however, to manage any soil for
This technique was operationally inno
vative, in that it took the basic Chow and increased clay content, although consid
Harbaugh (1965) rainfall simulator design erable improvement in stability can be
as previously used by Burney and Huggins attained through increased soil organic
(1973) and adapted it for more efficient matter and one of the surest ways to man
usage. There was an advantage in transpor age for this is to seed a winter cover crop Lafayette, Ind. 137 pp.
tation as well as in handling the samples in and incorporate it as green manuring after CHOW, V. T. and T. E. HARBAUGH. 1965.
the field and laboratory, in the relative ease the winter.
114
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Although the results of this study make
it clear that, regardlessof the groundcover
status, there will be significantly increased
sediment loss with an increase in erosive
force, there would still be considerable
wisdom in establishing a winter ground
cover to minimize such loss. The effect of
erosion agent on sediment loss has shown
significant interaction with ground cover
status; moreover, early- and late-seeded
cover, respectively, showed 18 and 29% of
the bare soil treatment sediment loss under
The principle findings of this study
leave no doubt of the quantitative, bene
ficial impact of winter rye cover on soil
erosion, and its ability to significantly
attenuate the effects of such naturally
uncontrollable factors as erosion agent,
freeze/thaw and soil type on sloping land.
Cover cropping as a management measure
to minimize soil erosion has long been
acknowledged globally, and the benefitsof
its application in the cool season are well
known in P.E.I., at least, qualitatively.
Nevertheless, popular adoption has been
constrained, mainly by the late harvest of
potatoes, although there is some adoption
after tobacco, corn and soybeans.
If, therefore, potato farmers on P.E.I,
are to adopt winter cover cropping and
make it an integral part of annual-crop
management, they must be prepared to
trade off the higher profits of late potatoes
against the benefits of cool-season soil ero
sion control by this means. This will,
undoubtedly, involve some sacrifice by
farmers in the short term, but can be justi
fied through stewardship in the pursuit of
long-term benefits to the farm as well as to
the wider resource as a whole.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the
assistance of Allan MacRae and Paul Frame in
conducting the field work and the rainfall sim
ulator runs, and Jack Vissers and Jim Godwin in
the design and construction of the soil cassettes.
REFERENCES
BENOIT, G. R. 1973. Effect of freeze-thaw
cycles on aggregate stability and hydraulic
conductivity of three soil aggregate sizes.
Soil Sci. Soc. Am. Proc. 37(1): 3-5.
BRYAN, R. B. 1971. The influence of frost
action on soil aggregate stability. Trans.
Inst. Br. Geogr. 50: 71-88.
BURNEY, J. R. and L. F. HUGGINS. 1973.
Hydraulics of shallow flows over stable
eroded sand surfaces defined by area spec
tra. Water Resources Research Center Tech.
Rep. No. 36, Purdue University, West
Raindrop production for laboratory water-
 shed experimentation. J. Geophys. Union
Res. 70: 6111-6119.
HARZA OF CANADA, LTD. 1968. Soil con-
servation program for the Atlantic
Provinces. Report for Dept. of Forestry and
Rural Development of Canada. 60 pp.
HIMELMAN, D. and N. E. STEWART. 1979.
Soil erosion studies in Prince Edward
Island. Paper, Soil Erosion Workshop,
NSAC, Truro, N.S. 17-19 April. 9 pp.
H1NMAN, W. C. and F. BISAL. 1968. Altera
tions of soil structure upon freezing and
thawing and subsequent drying. Can. J. Soil
Sci. 48: 193-197.
KEMPER, W. D. 1965. Aggregate stability.
Pages 511-519 in CA. Black ed. Methods
of soil analysis. Part 1. Agronomy 9. Am.
Soc. of Agron., Madison, Wis.
LOGSDAIL, D. E. and L. R. WEBBER. 1959.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Effect of frost action on structure of Haldi-
mand clay. Can. J. Soil Sci. 39: 103-106.
McCOOL, D. K., W. H. WISCHMEIER, and
L. C. JOHNSON. 1982. Adapting the uni
versal soil loss equation to the Pacific north
west. Trans. ASAE (Am. Soc. Agric. Eng.)
25: 928-934.
MORGAN, R. P. C. 1979. Soil erosion. Aca
demic Press, Don Mills, Ont. 113 pp.
ONSTAD, C. A. and R. A. YOUNG. 1982.
Erosion characteristics of three north
western soils. Trans. ASAE (Am. Soc.
Agric. Eng.) 25: 367-371.
RICHARDSON, S. J. 1976. Effect of artificial
weathering cycles on the structural stability
of a dispersed silt soil. J. Soil Sci. 27:
287-294.
SILLANPAN, M. and L. R. WEBBER. 1961.
The effect of freezing-thawing and wetting-
drying cycles on soil aggregation. Can. J.
Soil. Sci. 41: 182-187.
STEWART, N. E. 1979. Some problems asso
ciated with implementing soil conservation
programs. Paper, Soil Erosion Workshop,
NSAC, Truro, N.S. 17-19 April. 14 pp.
WALL, G. J. and DICKINSON, W. T. 1979.
Soil erosion prediction in Canada. Paper,
Soil Erosion Workshop, NSAC, Truro, N.S.
17-19 April. 25 pp.
WALL, G. J., L. J. P. VAN VLIET, and W. T.
DICKINSON. 1976. The universal soil loss
equation a method for predicting soil loss
in Canada. Abstract, Annual Meeting, Can.
Soc. Soil Sci., Halifax, N.S. 4-7 July.
WALL, G. J., W. T. DICKINSON, and J.
GREVEL. 1983. Rainfall erosion indices
for Canada east of the Rocky Mountains.
Can. J. Soil Sci. 63: 271-280.
115
 TILE DRAIN DISCHARGE UNDER DIFFERENT CROPS

G. J. Bryant1, R. W. Irwin1 and J. A. Stone2

'School of Engineering, University of Guelph, Guelph, Ontario NIG 2WI; and 2Agriculture Canada Research
Branch, Harrow, Ontario NOR JGO

Received 10 July 1986, accepted 8 December 1986

Bryant, G. J., R. W. Irwin andJ. A. Stone. 1987. Tiledrain discharge under different crops. Can. Agric. Eng. 29:
117-122.

Tiledrain discharge for a 14-yr period hasbeen examined to determine theeffect of cropcoveron the flow. Discharge
characteristics wereanalyzed from 12 drainage research plots having fertilized andnonfertilized cropsin a 4-yr rotationof
corn-oats-alfalfa-alfalfa, continuous bluegrass andcontinuous corn. Fertilized cropscontributeda greater volumeof drain
flow than unfertilized crops. The majorcontribution to draindischarge occurred in Marchand April. Continuous corn and
bluegrass contributed a largerdrainage volume thanrotation crops.Theeffectof cropcoveron drainagewas the leastfor a
2-yr return period. The analysis indicates that the crop grown and level of fertility have an importanteffect on tile drain
discharge in Brookston clay loam soils.

INTRODUCTION
Discharge from agricultural subsurface
drains is of concern to many people. Intu
itively, one feels that the installation of
subsurface drains must increase the vol
ume and rate of flow of water from the land
thus creating streamflow problems such as
flooding, increased erosion and scour,
damage to ecological systems, to list but a
few. The downstream effects of subsur
face drainage have been a concern for
many years.
Analysis of Ontario streams by McCub-
bin (1938) and Archibald (1939) concluded
there was little indication of increased
flooding due to agricultural drainage
activity. More recently Eddie (1982) stud
ied the annual mean, annual maximum,
and annual minimum streamflows of 10
rivers in western Ontario but failed to find
any trend that might be related to agri
cultural drainage. Irwin and Whiteley
(1982) reviewed 48 papers which dealt
with the effect of drainage on flow regimes.
The authors concluded that changes were
difficult to separate from other effects and
in many cases a reduction in flow occurred
rather than increased flow.
The watershed of the Middle Thames
river in southern Ontario has undergone a
substantial increase in drainage intensity.
Serrano et al. (1985) examined a 30-yr
record of this river and found no evidence
of appreciable change in volume of runoff
for individual storms, annual streamflow
volume, or changes in seasonal distribu
tion of streamflow.
Two major investigations which ana
cantly influenced by cropping sequence.
Daily flow rates for corn preceded by oats
was higher than for corn preceded by sec
ond year meadow. He attributed this to the
greater evapotranspiration of the meadow
which left the soil drier the next season.
Overall, there is a lack of knowledge
regarding the influence of crop cover and
fertility level on subsurface drain dis
charge characteristics. The objective of
this study was to determine the effect of
crop cover and fertility level on tile drain
discharge from a Brookston clay loam soil
in southwestern Ontario by analyzing
long-term tile drainage records.
EXPERIMENTAL DATA
Soil Characteristics
Data were obtained from an experiment
at the Whelan Experimental Farm (ap
proximately 4213'N, 8244'W) of the
Harrow Research Station, Agriculture
Canada, (Fig. 1) on Brookston clay loam.
The Brookston series is a naturally poorly
drained lacustrine soil (orthic humic
gleysol). It is an important agricultural soil
in Ontario representing 1.1 million ha. The
typical surface slope on these soils is
0.05-0.10%. Physical properties of the
soil are presented in Table I. Crop and
fertility treatment effects on soil structure
for the experiment described in this study
are presented by McKeague et al. (1987).
This soil often cracks extensively in late
summer. Hydraulic conductivity is there
fore difficult to measure with consistent
TABLE I. PHYSICAL PROPERTIES OF BROOKSTON SERIES
results. Field tests range from 0.05 m/day
to 5.00 m/day; laboratory tests range from
0.12 m/day to 1.60 m/day, with a median
value of 0.85 m/day.
Plot Description
Twelve plots, each 12.2 m by 76.2 m,
were established in 1956. A 100-mm tile
drain, spaced 12.2 m apart, was installed
in the center of each plot. The plot layout,
drain gradient (%), corner surface eleva
tions (m), and tile flow collection system
is shown in Fig. 2. The average depth of
drains is 70 cm which is existing practice
in the area.
Each drain discharged into an access
well which contained a flow meter equipped
with a linear head-discharge weir plate.
The access wells were heated for year-
around operation.
Crop Cover
Crops grown on the plots were ran
domized duplicate plots of continuous
corn (CC), continuous bluegrass sod (B)
and a 4-yr rotation of corn (C), oats (O),
alfalfa (Al) and alfalfa (A2). Superim
posed on the cropping treatments were two
levels of fertility, unfertilized (U) and fer
tilized (F). Plots 1to 6 received 16.8, 29.6,
and 27.9 kg/ha of N, P and K, respec
tively. Corn plots received an additional
side-dressing of 112kg/ha/yr of N. Plots 7
to 12 were unfertilized. Each crop was
grown each year. Figure 3 shows the crop
ping system for the plots during each
month of the year.

lyzed the effect of crop cover on drain Girain size {%)

Depth Dry density Organic matter
discharge were conducted in Ohio. The
Horizon (cm) Sand Silt Clay (g cm"3) (%)
Tiffin drainage plots were described by
Goins (1956) and Hoover and Schwab Ap 0-16 25 38 32 1.40 2.5

(1969) and the Sandusky drainage plots by Bgl 16-45 21 34 45 1.58 1.0
Bg2 45-116 21 34 45 1.58 0.4
Schwab et al. (1961). Hoover and Schwab
Ckg 116 + 21 39 40 1.52 0.6
(1969) found tile discharge was signifi

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 117

 Ottawa
Figure 1. Location of Woodslee drainage research plots.
0.12%
I I
29.40, ,1^ , ^ /^ ^<X .o. ;x: ^ \ 29.05
N
29.08 29.02
12 2m +>
29.26
29.11
O Measurement Well BG Bluegrass
Plot Boundary CC Contimuous Corn
Subsurface Tile Drains
Figure 2. Layout of drainage plots.
118 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 The quality of the crop cover varied
SEASON PLANTING GROWING DORMANT with annual variation in climate and with
fertility. For example, Table II gives the
MONTH A M J JASONDJFM
average yield and average plant height for
corn. The yield of fertilized rotation corn
was 64% higher than the unfertilized. The
PLOT DISC CORN PLOWED unfertilized continuous corn was nearly a
crop failure in every year. The yield of
other crops were not measured; however,
PLOT BLUEGRASS
the corn yield and plant height data should
be indicative of yearly variation in oat,
PLOT OATS + ALFALFA ALFALFA
alfalfa and bluegrass yields.

Hydrologic Data
PLOT ALFALFA 1 A continuous record of tile flow from
the 12 plots was analyzed for the years
1969 through 1982. Stage-time data from
PLOT ALFALFA PLOWED
the strip charts (Fig. 4) were digitized and
stored on a microcomputer hard disk in the
Figure 3. Time distribution of cropping and tillage operations. form of discharge. The discharge was in
millimeters of depth.
Friday The digitized data were used in a com
10 XII 2 4 puter program to develop time-to-peak
flow, total event flow duration, total event
volume, peak flow and maximum flow
volume expressed as an equivalent depth
for 12, 24, 36 and 48 h. These data were
then analyzed on a seasonal and annual
basis (Bryant 1986).
Climatological data were collected at
the site as part of the Environment Canada
meteorological observation network but
were not used in this analysis.
Surface runoff was measured from four
plots for the years 1980-1982. Runoff was
negligible except for one major storm in
this 3-yr period. Water table elevations
were not measured.

ANALYSIS OF DATA
Summary of the Data
There were 1927 tile flow events during
the 14-yr period, 1969-1982, of which
1641 were over 12 h in duration. The rec
Figure 4. Tile discharge hydrograph from plot showing time periods analyzed. ord was divided into three seasons. The
Planting season was the months of April,
TABLE II. YIELD OF GRAIN CORN FROM PLOTS May and June. The Growing season was
July, August, September and October. The
Rotation corn Continuous corn
balance of the year was termed a Dormant
Fertilized Unfertilized Fertilized Unfertilized season. Table III shows the distribution of
Plant Plant Plant Plant events by duration and season.
Yield height Yield height Yield height Yield height During the analysis it became apparent
Year (t/ha) (cm) (t/ha) (cm) (t/ha) (cm) (t/ha) (cm)
that maintenance and operation of the
1969 7.76 254 3.61 190 6.66 242 0.42 137 flow-measuring instruments was not con
1970 8.04 244 5.11 222 5.86 221 0.96 165 sistent over the period of record. The
1971 7.70 235 4.76 226 5.75 230 0.87 190 major problem was in defining the actual
1972 7.80 239 6.66 233 5.81 226 1.87 196
end of an event. Due to this uncertainty in
1973 7.03 258 4.93 224 5.50 244 0.42 158
1974 4.22 193 2.64 184 3.16 173 1.13 166
the data it was concluded that the max
1975 6.65 226 4.20 217 6.22 213 2.48 167 imum time period analyzed would be 48 h.
1976 5.82 235 4.01 212 3.71 189 0.81 131 Event volumes over 48 h and seasonal and
1977 8.75 NR 3.71 NRt 6.84 NR 1.41 NR
annual volumes are not reported here due
1978 6.20 NR 2.92 NR 3.86 NR 0.45 NR
to this uncertainty.
Mean 7.00 236 4.26 214 4.72 217 1.08 164
The 24-h discharge volumes and peak
tNR, no record. flows are often used in the design of field

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 119

 TABLE III. NUMBER OF TILE FLOW Four nonparametric tests were made on ing more continuous biopores due to root
EVENTS, 1969-1982 each of the 24 data sets (six crops x two channels and more cracking due to evapo
36 48
fertility levels x two parameters). The transpiration. In addition, the greater bio-
Duration (h) 12 24
results indicated that the data analyzed were mass produced by the fertilized corn and
Planting season 530 435 343 295 independent, without trend, homogeneous alfalfa crops may have resulted in more
Growing season 233 171 138 116
and random. residue being incorporated, thereby pro
Dormant season 878 765 678 612
1023
moting infiltration.
Total 1641 1371 1159
RESULTS AND DISCUSSION The high volumes for the continuous
Table IV shows the seasonal distribu corn plots in Figs. 5 and 6 are not readily
tion of annual discharge maxima. The explained as continuous corn has been
TABLE IV. SEASONAL DISTRIBUTION OF
ANNUAL 24-H AND EVENT VOLUMES majority of annual maximaoccurred in the shown to result in the most compact sur
Dormant season due to snowmelt. Ap face soil structure (McKeague et al. 1987).
Season 24-h volumes Event volumes
proximately 50% of the drain discharge However, the continuous corn plots, notably
Planting 29.0ft 24.47c volume occurred in March and April. the unfertilized plot, have been observed to
Growing 18.7ft 8.3ft Figures 5 and 6 show the 1.05-, 2-, 5- be the wettest treatments and to exhibit the
Dormant 5237c 67.3ft and 10-yrreturn period values of 24-h vol most ponding. Longer wet periods and a
March + April 44.57c 53.8ft
umes and event volumes for the different ponding head coupled with a smaller leaf
crops and two fertility levels. Figure 5 area (Table II) therefore less evapotranspi
demonstrates that, except for bluegrass, ration, may result in more water even
drainage systems. A frequency distribu the fertilized crops produced 24-h volumes tually moving to the tile drain.
tion analysis was conducted on the data which were 1.5 times greater than those of It is interesting to note that the
using the annual maximum 24-h tile vol the unfertilized crops. This is supported by 24-h volumes under bluegrass were the
umes and event volumes. An event was Fig. 6, which shows that the fertilized least of all the crops, whereas fertilized
defined as a single peaked, hydrograph crops, except continuous corn and alfalfa-2 bluegrass produced the largest event vol
shaped period of flow (Fig. 4). The three- (second-year), produced tile event vol umes. This suggests that the tile flow
parameter lognormal distribution (Kite umes which were approximately 1.4 times under bluegrass is long, low flow, and that
1977) was chosen to describe the annual greater than those of the unfertilized crops. a 24-h drainage coefficient does not ade
maxima. The Consolidated Frequency It is hypothesized that fertilizer pro quately represent the contribution of tile
Analysis computer program developed by motes the proliferation of roots and there flow to streamflow under bluegrass.
Pilon et al. (1985) was used to fit the three fore evapotranspiration. The greater number Figures 5 and 6 also show that the least
parameters of the distribution to the data. of roots may increase drainage by produc amount of variation among crops occurred

Woodslee, Ontario

a
9

E
3
O
>

I
CM

E
3

E
X
o
2

o
3
C
C
<

ZZI 1-05 yr CSS 2yr E2Z 5yr r^ 10 yr

Figure 5. Annual maximum 24-h volumes for each crop (F, fertilized; U, unfertilized; B, bluegrass; CC, continuous corn; C, rotation corn; O,oats;
Al, first-year alfalfa; A2, second-year alfalfa).

120 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 Woodslee, Ontario

E
E
a
o
E
3
0
>

c
0

E
E
X
0
2

o
3
C
c
<

FB FCC FC FO FA1 FA2 UB UCC UC UO UA1 UA2

ZZI 105 yr ESI 2yr Um 5yr r^ 10 yr

Figure 6. Annual maximum event volume for each crop (F, fertilized; U, unfertilized; B, bluegrass; CC, continuous corn; O, oats; Al, first-year
alfalfa; A2, second-year alfalfa).

TABLE V. COEFFICIENTS OF VARIATION FOR RETURN PERIODS The least variation among crops occurred
24-h volumes Event volumes
at the 2-yr return period. This was
Return
Period
expected since return periods greater than
X CT CV X CT CV
2-yr are associated with more extreme
1.05 12.5 5.8 0.467 33.3 10.7 0.323 storms. Extremity of storms may also
2 25.5 4.1 0.159 75.5 15.7 0.207 explain the greater differences in the tile
5 34.8 7.3 0.209 121.5 33.0 0.272
10
discharge among crops.
43.4 12.0 0.277 161.1 50.1 0.311

at the 2-yr return period for both the 24-h
volumes and event volumes. The coeffi
cients of variation in Table V support this
observation since they are lowest at the
2-yr return period.
It should be stressed that this study is
site specific. Overland runoff at the site
was negligible, and thus, was neglected.
In areas where overland runoff is signifi
cant, the combined effects of overland
runoff and subsurface drainage must be
considered.
CONCLUSIONS
The results indicate fertilized crops
potentially contribute a greater volume of
tile discharge to a stream than unfertilized
crops. Fertilizer appears to promote root
proliferation and therefore more continu
ous biopores and cracking through the sur
face profile which results in a more direct
channel to the drain thus increasing the
volume of water discharged.
Continuous corn and bluegrass gener
ally contributed a greater tile discharge
volume than rotation crops.
For the bluegrass plots this may be
attributed to a more favorable soil struc
ture, therefore more water storage in the
surface profile, which results in a long,
low flow. For the continuous corn plots it
may be associated with a compact soil
structure which results in longer wet peri
ods and ponding. Legumes in a rotation
increase the number of biopores in the
surface soil and evapotranspiration from
the crop. Therefore, less water may be
available for drainage.
The largest contribution of tile dis
charge to a stream occurred in March and
April, due primarily to snowmelt and a
greater depth of rainfall in these months.
REFERENCES
ARCHIBALD, S. W. 1939. Ontario stream
control. Can. Eng. Apr. 9: 4.
BRYANT, G. J. 1986. The effect of crop cover
on subsurface tile drain flow. M.Sc. Thesis,
University of Guelph, Guelph, Ont.
EDDIE, J. D. 1982. Temporal variations of
southwestern Ontario streamflows. Ontario
Ministry of Environment, Toronto, Ont.
Water Resources Paper 15.
GOINS, T. A. 1956. Tile flow characteristics.
Agr. Eng. 37: 30-32, 35.
HOOVER, J. R. and G. O. SCHWAB. 1969.
Effect of tile depth, spacing and cropping
practice in drain discharge. Trans. ASAE
(Am. Soc. Agric. Eng.) 9(2): 150-152.
IRWIN, R. W. and H. R. WHITELEY. 1982.
Effects of land drainage on streamflow.
Conf. on Water-Based Recreation Issues,
Conflicts, Strategies, Minett, Ont.
KITE, G. W. 1977. Frequency and risk analy
sis in hydrology. Water Resources Publica
tions. Littleton Co.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 121

McCUBBIN, G. A. 1938. Agricultural drai-
nage in southwestern Ontario. Eng. J. 21(2):
66-70.
McKEAGUE, J. A., C. A. FOX, J. A.
STONE, and R. PROTZ. 1987. Effects of
cropping system on structure of Brookston
clay loam in long-term experimental plots at
Woodslee, Ontario. Can. J. Soil Sci. 67: (In
Press).
PILON, P. J., R. CONDIE, and K. D.
HARVEY. 1985. Consolidated frequency
analysis package. User Manual, Water
Resources Branch, Inland Waters Dir.,
Environment Canada, Ottawa, Ont.
SCHWAB, G. O., J. L. FOUSS, and A. P.
LEECH. 1961. Rate, duration and frequency
of tile flow. Agr. Eng. 42: 546-549, 555.
SERRANO, S. E., H. R. WHITELEY, and
R. W. IRWIN. 1985. Effects of agricultural
drainage on streamflow in the Middle
Thames River, Ontario 1949-1980. Can.
J. Civil Eng. 12: 875-885.

122 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 MEASUREMENTS AND ANALYSIS OF RUNOFF FROM SUBSURFACE
DRAINED FARMLANDS

S. Natho-Jina, S. O. Prasher, C. A. Madramootoo, and R. S. Broughton

Department of Agricultural Engineering, Box 950, Macdonald College of McGill University, Ste. Anne de Bellevue,
Quebec, H9X ICO

Received 26 March 1986, accepted 8 December 1986

Natho-Jina, S., S. O. Prasher, C. A. Madramootoo, and R. S. Broughton. 1987. Measurements and analysis of
runoff from subsurface drained farmlands. Can. Agric. Eng. 29: 123-130.

The correlation between agricultural subsurface drainage systems and peak flow rates was investigated by measuring
bothsurfaceand subsurface componentsof runoffon an 11-ha field in Ormstown,Quebec. Depth of runoff, peak flow rate
and time parameters were estimated from measured hydrographs. Longer times to peak were observed on the subsurface
flow hydrographs compared to the surface runoff hydrographs.

INTRODUCTION
Drainage improvements are essential in
the humid regions of Canada. Approx
imately, two-thirds of the land in southern
Ontario, Quebec and the Maritime
Provinces that has soil and climate suitable
for food production needs drainage
improvement to achieve economic crop
production. It is estimated that as of March
1981, about 332 583 ha of a total of
2 534 540 ha in Quebec have been drained
by subsurface drainage systems. In
1980-1981, 50 404 ha of land in Quebec
were drained by subsurface drainage sys
tems, and large amounts of monies are
spent on this drainage work every year.
Some of the rivers which receive water
from drainage ditches have flood and pol
lution problems. There is a need for mea
surements and theoretical analysis to
determine the quantitative and water
quality effects of various land drainage
practices on the flood flows and the low
flows of the rivers downstream of the drai
nage works (Hill 1976; Skaggs 1982;
Madramootoo 1984).
Irwin and Whiteley (1983) pointed to the
fact that the magnitude of flood volumes
from drained and undrained land depended
on several factors including antecedent
soil moisture, rainfall intensity and loca
tion of drainage works. More field work is
needed to quantify the effect of subsurface
drainage on surface and subsurface runoff,
on soil properties, drainage intensity and
drainage conditions.
In this paper, field measurements of
some actual runoff and peak flow rates are
presented, and hydrographs due to surface
and subsurface runoff are also compared.
EXPERIMENTAL FIELD
Site
The research was done on a farm located
approximately 3 km west of Ormstown,
Quebec, near the Chateauguay River. The
research site is cropped with continuous
grain corn. The field is 11 ha with an aver
age slope of 0.2% towards the Cha
teauguay River, and has two open ditches
for removing surface runoff, and subsur
face drains for subsurface water.
Soil Description
Particle size analyses were carried out
for a number of soil samples. The distribu
tion was found to be 12% clay, 68% silt and
20% sand and the soil was classified as a
silt loam.
Drainage System
Thirty-two laterals, complete with
envelope materials, were installed in
October 1983. Observation chambers were
installed at the lateral/mainline connection
of 30 laterals. The observation chambers
allowed measurements of flow rates from
individual laterals which were spaced 15 m
apart. The average depth below the soil
surface of all laterals ranged from 1.14 to
1.62 m.
FIELD MEASUREMENTS
Hydrometric Measurements
Three 90-degree V-notch weirs were
placed in the field in order to measure the
subsurface and surface flows. One weir
was installed to collect subsurface runoff
at the outlet of the subsurface drainage
system. An adjacent weir measured sur
face runoff from the northern half of the
field. The third weir was installed at the
southeastern corner of the field to measure
the surface runoff. The locations of these
weirs are shown in Fig. 1. The discharges
from the weirs were measured with a vol
umetric cylinder and a stop watch follow
ing the cessation of a precipitation event.
The corresponding heads of water above
the V-notch weir were also measured. A
calibration curve of head versus discharge
was then plotted.
Belfort F.W. 1 portable continuous water
level recorders with a weekly clock and
chart provided measurements of water
level above the weirs. The recorders were
fixed to the top of a stand protected by a
wooden shelter. From the records of head
on the weir and the calibration curve,
hydrographs of outflow versus time were
drawn for selected runoff events.
Climatological Measurements
The hourly rainfall data were measured
by a tipping bucket rain gauge. Due to
gauge malfunction during some storm
events, some rainfall data were used from
the closest weather station at Ste.
Clothilde. In some cases, the total daily
precipitation of the Ormstown Station was
used.
Soil Physical Properties
Hydraulic conductivity was measured
by the single auger hole method (van Beers
1976). The saturated hydraulic con
ductivity values, Ksat, ranged from 0.17 to
0.29 m/day, with a mean value of 0.23 m/
day.
Soil samples were taken at different
sites at depths of 10, 35 and 65 cm from the
soil surface. The soil moisture charac
teristic curve was obtained by both the
sand suction table (Rashid-Noah 1981) and
the pressure plate apparatus. The soil
moisture content at saturation (zero suc
tion) and field capacity (1/3 bar) are shown
in Table I.
Bulk density was measured by drying
the soil core samples after the desorption
TABLE I . SOIL MOISTURE CONTENTS
Depth Saturation Field
(cm) (ft) capacity (ft)
10 48 43
35 46 40
65 52 31

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 123

C/3
C Figure 1. Plan of the subsurface drainage system at the research site.

m
73
 1 1 measurements were completed. The mean
bulk density was found to be 1.70 g/cm3.

-J RESULTS AND DISCUSSION

Storms producing a significant amount
J P of runoff during the spring and summer of
<
CC 1984 and 1985 were identified. The hydro-
/ i \ ^ CREST SEGMENT graphs were plotted with the correspond
ing rainfall hyetographs.
The hydrograph parameters under study
TL / ' \ are illustrated in Fig. 2.
Table II summarizes the rainfall-runoff
ratio and the peak flow rate deduced from
different hydrographs.
| T5 Tp / 1 \ Table III summarizes the time param
eters estimated from all of the hydrographs
Li
/ 1 \ ^- RECESSION shown in Figs. 3-10.
rr
<
x Subsurface Runoff
o
if) The greatest rainfall did not produce the
RISI NG
greatest surface runoff. The snowmelt may
LI MB
have contributed to the total runoff giving
greater depth of runoff with less rainfall in
the 15-20 April 1984 event compared to
TIME the 7-11 May 1984 event. In summer,
when the evaporation and transpiration
Figure 2. Hydrograph parameters. from the plants is greater, there is less

APRIL 15 APRIL 16 APRIL 17 APRIL 18 APRIL 19 APRIL 20 DATE

12 24 12 24 12 24 12 24 12 24 12 24^ HOUR
"T" I 0.0

PERIOD: APRIL 15-20, 1984

1.0
(STE-CL0THILDE RAINFALL DATA)

8.0 2.0
SSF=SUBSURFACE FLOW

rr

4.0

0.0 12 24 12 24 12 24 12 24 12 24 12 24 HOUR
APRIL 15 APRIL 16 APRIL 17 APRIL 18 APRIL 19 APRIL 20 DATE

Figure 3. Hydrograph for the 15-20 Apr. 1984 event.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 125

 TABLE II. RUNOFF AND PEAK FLOW RATES runoff. Peak flows are significantly influ
Runoff Runoff/
enced by the antecedent soil moisture con
Qv
(mm ) rainfall (mm/day) dition. When antecedent soil moisture was
Number Rainfall
low, as in summer months, subsurface
Date of days (mm) SSRt SRt SSR SR SSR SR
drainage produced relatively small in
15-20 Apr. 1984 5 36.2 27.4 NA 0.76 NA 6.1 NA
creases in depths of runoff.
7-11 May 1984 4 40.6 9.6 NA 0.24 NA 3.4 NA
In 1985, the maximum peak flow rate
29-31 May 1984 3 31.2 16.4 NA 0.53 NA 4.4 NA
27 Mar.-1 Apr. 1985 6 27.7 17.3 3.7 0.62 0.13 5.5 6.6* observed was 7.79 mm/day during 2-7
2-7 Apr. 1985 6 18.5 26.3 7.5 1.42 0.40 7.8 5.2 April arising from 18.5 mm rainfall. It is
18-25 Apr. 1985 8 16.4 21.8 1.0 1.33 0.06 5.1 2.3 less than the value of 10 mm/day com
17-21 June 1985 5 33.2 3.0 0.1 0.09 0.04 2.2 1.6*
monly used in subsurface drainage design
15-19 July 1985 5 23.4 2.54 0.4 0.11 0.02 2.3 3.2*
as the drainage coefficient. The pipes in
tSSR, subsurface runoff; SR, total surface runoff weir no. 2 and weir no. 3. the experimental field have sufficient
^Surface runoff from weir no. 3.
capacity to allow drainage rates of up to 18
TABLE III. TIME PARAMETERS
mm/day. The maximum drainage rate
observed to date has been limited by soil
Tlag T
1p hydraulic conductivity and height of the
(h) (h) (h)
water table in the soil rather than by the
Date SSRt SRt SSR SR SSR SR pipe capacity. There may also be resis
15-20 Apr. 1984 7 NA 18 NA 3 NA tance to flow due to the blockage of the
7-11 May 1984 13 NA 12 NA 6 NA filter material around the pipes.
29-31 May 1984 6 NA 6 NA 4 NA
27 Man-1 Apr. 1985 NA NA 12 7 0 0
Surface Runoff and Subsurface Runoff
2-7 Apr. 1985 NA NA 30 15 NA NA
18-25 Apr. 1985 16 8 12 4 6 6 The peak flow rate from surface runoff
17-21 June 1985 12 13 7 1 8 14 was greater than from subsurface flow for
15-19 July 1985 9 3 7 2 4 3 some events. In the 27 March - 1 April
tSSR, subsurface runoff; SR, total surface runoff weir no. 2 and weir no. 3. 1985 event, the runoff may have beer

MAY MAY MAY MAY 10 MAY DATE

12 24 12 24 12 24 12 24 12 24a() HOUR
1 "T" r
IT

1.0

PERIOD: MAY 7-11, 1984

(STE-CL0THILDE RAINFALL DATA)
2.0

3.0

SSF : SUBSURFACE FLOW

J2
5.0 4.0
rr

4.0 5.0

.3.0 6.0
Id
O
rr
<

o 20 7.0
CO

i.o-

1 i i i i i 1
0.0
12 24 12 24 12 24 12 24 12 24 HOUR
MAY 7 MAY 8 MAY 9 MAY 10 MAY II DATE

Figure 4. Hydrograph for the 7-11 May 1984 event.

126 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 MAY 28 MAY 29 MAY 30 MAY 31 JUNE I DATE
24 12 24 12 24 12 24 12 24 12 ^4aoHOUR
nrmrn i

10

2.0
PERIOD^ MAY 29-31, 1984
(STE-CLOTHILDE RAINFALL DATA)

3.0 ?
<
rr

6.0 SSF =SUBSURFACE FLOW 40

5.0 5.0

uj 40
CD
rr
<
x

co 30
ou

2.0

00 L- 12 24 12 24 12 24 12 24 12 24 HOUR
MAY 28 MAY 29 MAY 30 MAY 31 JUNE I DATE

Figure 5. Hydrograph for the 28-31 May 1984 event.

MARCH 27 MARCH 28 MARCH 29 MARCH 30 MARCH 31 APRIL I DATE

24 12 24 12 24 12 24 12 24 12 24 12 24 HOUR
0.0

SSF SUBSURFACE FLOW

80 PERIOD^ MARCH 27 - APRIL Ist, 1985 1.0
SR SURFACE RUNOFF (ORMSTOWN RAINFALL DATA)

TOTAL PRECIPITATION
70 2.0
4.5mm
18.2 mm
5.0mm
6.0 3.0

5.0

UJ

2 4.0
<
X
O
CO

Q 3.0

2.0

10

0.0,
4 12 24 12 24 12 24 12 24 12 24 12 24 HOUR
MARCH 27 MARCH 28 MARCH 29 MARCH 30 MARCH 31 APRIL DATE

Figure 6. Hydrograph for the 27 Mar.-l Apr. 1985 event.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 127

influenced by the frozen soil in the field for total flow. The subsurface drainage time to peak and recession time of the
which affected the infiltration process. maintained discharges at greater levels for surface runoff hydrograph are relatively
During all storm events, the subsurface longer periods than surface runoff (Figs. short. Surface runoff from the field ditch
flow was the major component responsible 8-10). Due to the small drainage area, the alone caused the peak flows to occur dur
ing and shortly after the rainfall when the
APR. 2 APR. 3 APR. 6 APR. 7 DATE water is removed rapidly from the field.
24 12 12 24 24 12 24 12 24 HOUR
r Subsurface drainage removes excess soil
water more slowly over a longer period of
time. It lowers the water table and
increases the storage volume available in
the soil profile prior to a storm. The hydro-
graphs in Figs. 9 and 10 show the result of
storms starting with a rather dry initial
condition. The storm intensities exceeded
the soil infiltration rate and produced
runoff without raising the water table to the
surface.

Time Parameters
Time to peak, 7p, represents the time
interval from the beginning of the runoff
until the peak discharge rate, Qp, occurs.
Time to start, Ts, is the time interval from
the beginning of rainfall to the beginning
of the runoff. Lag time, 7,, is defined as the
time interval from the maximum rainfall
until the peak discharge rate, Qp, occurs.
Time to peak, Tp, occurs later during the
early spring storm runoff than during other
periods. Thegreater Tp values observed in
spring may be due to the water flow altered
by the high soil moisture condition or to
either the gradual release of water from the
snowpack or the remaining snow and
12 12 24 12 12 24 12 24 HOUR
rough plowed ground retarding runoff.
APR. 3 APR. 4 APR. 5 APR. 6 APR. 7 DATE Time to peak, Tp, was also longer for
Figure 7. Hydrograph for the 2-7 Apr. 1985 event

APR. 17 APR. 19 APR. 21 APR. 23 APR. 25 DATE

12 24 12 24 12 24 12 24 12 24 12 24 12 24 12 24 12 24^ HOUR
n TT 0.0

1.0
PERIOD- APRIL 18-25, 1985
(ORMSTOWN RAINFALL DATA) 2.0 |
SSF* SUBSURFACE FLOW

SR * SURFACE RUNOFF 2
z

4.0 <

UI2 24 12 24 12 24 12 24 12 24 12 24 12 24 12 24 12 24 HOUR
APR. 17 APR. 19 APR. 21 APR. 23 APR. 25 DATE

Figure 8. Hydrograph for the 18-25 Apr. 1985 event.

128 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 JUNE 17 JUNE 18 JUNE 19 JUNE 20 JUNE 21 DATE
24 12 24 12 24 12 24 12 24 12 24_ HOUR
0.0

1.0

PERIOD= JUNE 17-21 , 1985 - 2.0

(ORMSTOWN RAINFALL DATA)

3.0

SSF' SUBSURFACE FLOW

SRI' SURFACE RUNOFF WEIR *3 H4.0 <

<
a:
5.0

3.0" 6.0

,2.0 h 7.0
e>
108mm
rr
<

5 1.0 8.0
CO

0.0 l/V
24 12 24 12 24 12 24 12 24 12 24 HOUR
JUNE 17 JUNE 18 JUNE 19 JUNE 20 JUNE 21 DATE

Figure 9. Hydrograph for the 17-21 June 1985 event.

JULY 15 JULY 16 JULY 17 JULY 18 JULY 19 DATE

24 12 24 12 24 12 24 12 24 12 24 HOUR
Tl r 1ll r T"

1.0

E
E
2.0
_T
<
PERIOD' JULY 15-19, 1985
(ORMSTOWN RAINFALL DATA) 30 2

SSF' SUBSURFACE FLOW 4.0

SRh SURFACE RUNOFF WEIR *3

5.0

3.0

r 2.0
9.8mm

1.0

0.0
24 12 24 12 24 12 24 12 24 12 24 HOUR
JULY 15 JULY 16 JULY 17 JULY 18 JULY 19 DATE

Figure 10. Hydrograph for the 15-19 July 1985 event.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 129

subsurface runoff when compared with the
surface runoff due probably to the longer
path of water to flow into the drain and
from the drain to the outlet. The longer
delays associated with subsurface flow are
also partly due to the increased resistance
of flow through the soil matrix. More time
is required for the water to go into the
drainable pore space raising the water table
and producing the head which causes flow
toward the drains.
The surface runoff response to rainfall is
not always the same as subsurface flow
response. The greater 7S values for surface
runoff compared to subsurface flow are
due to the fact that more rainfall is needed
to fill the depressional storage. During the
first hours of rainfall, most water infiltrates
into the soil resulting in a longer time for
surface water to flow to the drains.
Large variations of 7, were due to the
variations in rainfall duration and to the
nonuniform distribution of the rainfall over
the field. Most of the large values of 7, for
subsurface flow were from the short dura
tion rainstorms of less than 6 h.
The time parameters were found to be
lowest for the month of July. This may be
130
due to the crop interception. The corn
plants are tall during this period.
CONCLUSION
The subsurface flow hydrographs
exhibited flatter and longer recessions
compared to the surface runoff hydro-
graphs. The subsurface drainage contrib
uted the most to the water removed from
the field. The subsurface drainage peaked
at 7.79 mm/day and decreased as the water
table receded. Most flow occurred during
and immediately after the snowmelt
period. Subsurface and surface flows of
26.3 and 7.45 mm, respectively, were
measured in a spring event and 2.53 and
0.38 mm during a summer storm.
Hydrograph time parameters were
dependent on the storm season, storm
duration and rainfall intensity. The time to
peak, Tp, was the greatest in April (30
hours), and it was always later for subsur
face flow than for surface runoff. The time
to start, Ts, ranged from 3 to 8 h for subsur
face flow and from 3 to 14 h for surface
flow. The lagtime, 7,, varied from 6 to 16h
for subsurface runoff and from 3 to 8 h
from surface runoff.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
ACKNOWLEDGMENTS
The research reported in this paper was sup
ported by funds provided by the Counseil des
Recherches et Services Agricoles du Quebec
and Environment Canada. The authors also
acknowledge the support of Mr. Peter Finlayson
for allowing us to use his farm and for his
numerous helpful suggestions.
REFERENCES
HILL, A. R. 1976. The environmental impacts
of agricultural land drainage. J. Environ.
Manage. 4: 251-274.
IRWIN, R. W. and H. R. WHITELEY. 1983.
Effects of land drainage on stream flow. Can.
Water Res. J. 8(2): 88-103.
MADRAMOOTOO, C. 1984. An agricultural
land drainage simulation model. Ph.D. The
sis. McGill University, Montreal, Que.
RASHID-NOAH, A. B. 1981. Drainage subsur
face drainage systems to avoid excessive
drainage of sands, Ph.D. Thesis. McGill
University, Montreal, Que.
SKAGGS, R. W. 1982. Predicting effects of
drainage runoff. Drainage Contractor 8(2):
28, 30-31, 34.
VAN BEERS, W. F. 1976. The auger hole
method. International Institute for Land
Reclamation and Improvement. Wage-
ningen, the Netherlands.
 SOIL CHARACTERISTICS AROUND MOLE DRAINS INSTALLED AT
DIFFERENT DEPTHS

T. G. Sommerfeldt, C. Chang, T. Entz, and B. J. Lamond

Research Station, Agriculture Canada, Lethbridge, Alberta T1J 4B1

Received 22 October 1986, accepted 7 January 1987

Sommerfeldt, T. G., C. Chang, T. Entz, and B. J. Lamond. 1987. Soil characteristics around mole drains installed
at different depths. Can. Agric. Eng. 29: 131-135.

A study was carried out to determine the bulk density distribution of soil around mole drains and also the smear and
conditions at the surface of the mole. Drains were installed at 50, 70 and 90 cm depths in a clay loam soil. Three water
treatments were usedon the soil: not flooded (D), flooded 1wk prior(Fl) to installation, and flooded 1day prior (F2) to
installation of the drains. The water treatments affected the conditions of the moles, the fracturing of the soil above the
moles, and the bulk density of the soilaround the moles. Thegreatest amount of smearing around the perimeter of moles
was observed in the F2 treatment and the least was in the D treatment. Visiblefracturing of the soil above the drains was
affected by depth of installation as well as by watertreatment. In 50-cm-deep drains, fracturing started at the side of the
mole and fanned out upward to the surface. In 90-cm drains there was minimal fracturing, starting at about the 50-cm
depth. Most fracturing occurred under the D water treatment, whereasfracturing was least in the F2 treatment. The mean
bulk density of the soil above the moles was smaller than that below the moles, where it was similar to that of the
undisturbed soil. Compaction of the soil around the mole, though significant, shouldnot impededrainage. Drag between
the bullet and the soil should remove segments of the smeared skin, leaving sufficient areas of unsmeared soil through
whichwatercan easilyenter the drains. Thus, waterentry intothe moles shouldnot be restricted by compactionor smear,
regardless of depth of installation or water content of the soil during installation.

INTRODUCTION
Soil bulk density and pore size distribu
tion are determining factors for the perco
lation of water through the soil (Wingate-
Hill 1978). When mole drains are installed,
soil material is displaced by the bullet,
which exerts compressive forces in all
directions. These compressive forces
should increase the soil bulk density and
reduce pore size distribution around the
drain, which should restrict water entry
into the drain. However, because of the
shape and dip of the bullet, which is
tapered from top to bottom on the leading
end and set on sufficient pitch to cause
suction, the major thrust is upward (Childs
1942). With a tillage tool, such as a plow
share or cultivator blade, there can be a
drain, which affect water entry into the In previous studies (Sommerfeldt
drain. 1984), the moles were installed for recla
If the mole is installed shallow enough, mation of a waterlogged saline soil with
the upward thrust should cause soil frac out regard for the critical depth and
turing to the surface. The depth at which conditions of the mole. The major concern
the upward thrust does not fracture the soil was whether the drains functioned or not.
to the surface is known as the critical Mole drainage has good potential for
depth. To produce well-formed, stable draining fine-textured soils in southern
mole channels, the moles must be installed Alberta (Sommerfeldt 1984; Sommerfeldt
deeper than the critical depth (Godwin et and Chang 1986), either by itself or in
al. 1981), but installation below the critical combination with subsurface tube drai
depth may cause soil compaction around nage. If soil compaction around the mole
the moles, which limits water entry into is critical to water entry (Naarding 1977),
the drains (Naarding 1977). This compac then drains installed below the critical
tion can persist for years. depth may not function well. As indicated
TABLE I. WATER CONTENT OF THE SOIL AT EACH DRAIN, TO DRAIN DEPTH, AT TIME
OF INSTALLING THE MOLE DRAINS

small build-up of soil at the tool point

which exerts forces downward, to cause Water content (%) at mole depth (cm)
Water Sample depth
compaction beneath the tool (Gill and treatment t (cm) 50 70 90
Vanden Berg 1968). Presumably, this
14.1
build-up and compaction could also occur 0-20 13.0 16.5
20-40 20.8 20.2 22.4
at the point of the bullet of a mole plow. 40-60 23.1 23.0 25.3
Also, in wet soils smearing can occur 60-80 23.0 22.7
when the mole is pulled through, which 80-100 23.3
can prevent water entry into the moles, but Fl 0-20 20.7 23.1 21.7
friction forces between the bullet and the 20-40 21.6 23.2 23.6
soil can cause cracking of the smeared 40-60 25.3 24.0 23.5
60-80 22.8 24.0
surface of the mole. These cracks or fis
80-100 24.4
sures, created also by the shank to which
F2 0-20 25.3 24.7 25.0
the bullet is attached, are an important
20-40 25.0 24.7 23.9
feature of mole drainage, especially near 40-60 25.6 25.5 26.5
the mole (Leeds-Harrison et al. 1982; 60-80 25.6 26.2
Youngs 1985). Consequently, it appears 80-100 24.0
that there can be undesirable and desirable t D, not flooded priorto installing drains; Fl, flooded 1wkpriorto installing drains; F2, flooded 1day priorto
effects on the soil at the surface of a mole installing drains.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 131

by Gill and Vanden Berg (1968), the
amount of compaction on the bottom of
the mole should probably be small and
insignificant. Also, because of drag, the
soil could be pulled apart, leaving cracks
around the perimeter of the mole for water
entry. Flow of water to a drain is mainly
radial and entry into the drain is mostly
through the bottom, if the drains are
spaced several meters apart and depth to
the impermeable layer is much deeper than
the drain (Kirkham et al. 1974). Thus the
performance of the mole drains should not
be seriously affected by depth of installa
tion, if indeed the bulk density of the soil
beneath the mole is minimally affected. A
study reported here was carried out to
determine the effects of installing mole
drains at different depths and different soil
water contents on the bulk density dis
tribution of the soil around the mole.

MATERIALS AND METHODS

Mole drains were installed at three
depths in a clay loam soil under three
water treatments in a split-plot designed
experiment with three replications. The
established grass field had a water table
generally at 1.25 m depth. Characteristics
of the mole plow have been reported
(Sommerfeldt 1984). Depth settings for
the mole plow, in a floating position to
reduce upward thrust (Childs 1942), were
established previously.
Three lines of drain were installed paral
lel to each other 5 m apart. Within each
line there were three depths of drain (50,
70 and 90 cm), each 55 m long, in a
random position. Buffers, 5 m long, were
left between each section to provide space
to establish the new plow depth. Prior to
installing the drains, three water treat
ments were established in the soil within
each depth section, namely: D, no water
applied; Fl, 20 cm water applied 1 wk
before installing the drains; and F2, 20 cm
water applied 1 day before installing the
drains. These were established directly
over the drains. The plots were 15 m long
and 2 m wide with a 5-m buffer between
each. Immediately after the drains were

Figure 1. Surface conditions around the

perimeter of the mole drains installed at 70-cm
depth, as affected by water treatment prior to
installing the drains. Arrow A indicates a
smeared area and arrow B indicates area where
smeared layer has been pulled away. (D, no
flooding (top), Fl, flooded (with 20 cm water)
1wk before installing the drains (center), and
F2, flooded I day before installing the drains
(bottom).)

132 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

Figure 2. Fracturing of the soil that had not been flooded (Dtreatment) caused while installing themole drains at 50-, 70-and90-cm depth. Arrows
indicate fracture lines on the edge of the fan-shaped area affected.

Depth (cm) depths, with clay contents of 35.0, 35.6

and 35.5% at 50, 70 and 90 cm, respec
1.70- tively.
Soil samples were taken along radial
1.65- lines at 45 intervals around the mole,
except at the vertical (shank) position. The
radial positions were numbered consecu
1.60- tively from 1 to 7 in a clockwise direction
> around the mole, starting at 45 from the
% 1.55- vertical (shank) position. Samples, 14.6
c mm in diameter and 15.0 mm long, were
CD taken simultaneously at 10, 35, 60, 85,
Q 1.50-
110, 135, 160, 185, 210 and 235 mm from
the edge of the mole to the center of the
"5 1.45- sample. A specially made sampler was
m used to take all the samples on one line
1.40- simultaneously, which thereby prevented
disturbance of the soil at the adjacent sam
ple site. A sample at 335 mm was taken to
1.35-
represent the bulk density of the undis
turbed soil on the assumption that compac
1.30- tion by the mole did not extend to that
i i r
distance.
3 4 5
The bulk density data were statistically
Radial Arm analyzed by analysis of variance.
Figure 3. Mean bulk density of the soil as affected by the mole depth and radial position
around the drain. RESULTS
The water treatments affected the condi
installed, soil samples were taken from the around the drains exposed in the pits, on tions of the mole and the fracturing of the
midsection of each plot, at 20-cm depth the side opposite to the direction of travel. soil above the mole. Because of the shal
intervals to 10 cm below drain depth, for Soil disturbance and the amount of smear low water table, water treatment effects on
soil moisture (gravimetric) and particle caused by the plow were examined and the water content of the soil below 60-cm
size distribution (hydrometer, Day 1965) studied on the same side of the pit. Study depth were small (Table I). Above this
determinations. After the drains were of the smear around the perimeter of the depth these treatment effects were greater,
installed, pits were dug across the drains at mole and surface conditions of the mole especially in the 0- to 40-cm depth.
the midsection of each plot to expose the was done primarily on the other side of The greatest amount of smear on the
soil and drains. Soil samples for bulk den the pit. perimeter of the mole was in the moles of
sity determinations were taken from The soil texture was similar at all mole the F2 treatment and the least was in the D

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 133

 1.70 Radial 1 tion. Because this characteristic was gen

1.65
1 Radial 2,6
Radial 3,4,5
eral, the cause is attributed to the mole
plow, which could have become slightly
misaligned over the years. At 70-cm
Radial 7
depth, fracturing began at the top of the
1.60 moles and extended upward to the suface,
fanning out with distance. The angle of the
1.55 fan was smaller than that of the 50-cm-
deep moles. Where lateral fracturing
occurred above the moles at 90-cm depth,
1.50 it started at a depth similar to that of the 70-
cm moles, but the angle of the fan was
1.45 - small or nonexistent, depending on the
water treatment.
The width of fracturing was also indi
1.40-
cated by the mound of soil deposited at
\ ground level while the drains were being
1.35 installed. The greatest amount of mound
50 cm depth ing was observed when the moles were
installed at 50-cm depth. The mounds,
1.30
TI 1 I I I 1 I I I I I I I 10-15 cm high, extended laterally 45-60
CO
cm from the shank. After the plow had
c passed through, the soil settled to leave
0
only a small mound (see surface Fig. 2,
O
50-cm depth). Mounds for the 70-cm deep
moles were similar to but smaller than
DO those for 50-cm-deep moles. For 90-cm
deep moles, the mound was limited to an
area near the shank. With the F2 treatment
there was essentially no mound.
At the 50- and 70-cm depths the soil
bulk densities in radial positions 1,2,6 and
7, at and above the center line of the
moles, were smaller than those in radial
I I I I I I I I positions 3, 4 and 5 (Fig. 3). The soil in
radial positions 1 and 7 would be in that
1.65 -. location of disturbed soil, above the crit
ical depth, while drains were installed.
Radials 3,4 and 5 were below the center of
1.60 -
the moles and below the soil disturbed
during drain installation, and the densities
1.55 - were not significantly different from the
undistrubed soil. For the 90-cm-deep
1.50 - moles the greatest bulk densities were lat
eral at the depths of the mole, along radials
1.45 - 2 and 6, while the smallest was along
90 cm depth radial position 4, at the bottom of the
1.40 mole. The mean bulk densities of the soil
"1 1 I I I I I I I I I I I I in radial positions 1 and 7 were signifi
10 60 110 160 210 260 310
cantly different ( P < 0.05) from each
Distance (mm) other, while those in positions 2 and
6 were not significantly different
Figure 4. Mean bulk density ofthe soil along radial arms, asaffected by mole depth and radial (P < 0.05) and neither were those in posi
position around the drain. tions 3, 4 and 5. The soil density in radial
positions 1 and 7, zone of visible fractur
treatment (Fig. 1). Within each water Soil disturbance caused by the moles at ing (Fig. 2), was significantly less
regime, sections of the smeared perimeter, 50-cm depth started at the sides of the (P < 0.01) than that in positions 2 and 6,
5 to 10 mm deep, had been pulled away by mole and extended upward to the ground and that in positions 2 and 6 was signifi
the bullet, leaving areas of exposed surface, fanning out with distance (Fig. 2). cantly less (P < 0.01) than that in posi
unsmeared soil. The width of the fan was greatest for the D tions 3, 4 and 5.
Visible fracturing of the soil above the water treatment and least for the F2 treat The bulk density varied along the radi
drains, more evident in the field than in the ment. The angle of the fan, from the ver als, depending on the radial position and
photographs, was affected by depth of tical, was larger on the left side, number 7 the depth of the drain. According to the
installation as well as by water treatment. position, than on the right, number 1posi bulk densities at 335 mm, the bulk density

134 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 1.65-1
1.60-
>
CO
c mole was installed.
<D
G
1.55-
m during installation, as long as the soil is
1.50-
1.45-
10 60 110
Distance (mm)
Figure 5. Mean bulk density of the soil as affected by water treatment prior to installing the
mole drains. (D, noflooding, Fl, flooded (with 20cm water) l wkbefore installing
the drains, and F2, flooded 1 day before drains were installed.)
of the undisturbed soil at 50-cm depth was
about 1.48, whereas at 70 and 90 cm it was
about 1.55. At all depths, most radials had
a zone of compaction at 35-mm distance
from the edge of the mole (Fig. 4). For the
50- and 70-cm-deep moles, the mean bulk
density along radial positions 1 and 7
decreased with distance from the 35-mm
point and there was a trend for the soil bulk
density in the 2 and 6 positions to decrease
with distance also. However, in positions
2 and 6, for the 50-cm-deep moles, the
bulk density reached a minimum in the
110- to 210-mm range and with the 70-cm-
deep moles it was minimum in the 110- to
160-mm range, whereas in positions 3, 4
and 5 the soil bulk density increased with
distance for the 50-cm-deep moles but
tended to be relatively constant for the 70-
cm-deep moles, with a small increase in
the 135-to 210-mm region. For the 90-cm-
deep moles, there was a trend for the bulk
density to reach a minimum at 185- to 210-
mm distance. Also, the bulk densities in
radial positions 3, 4 and 5 tended to be
greater than those in the other positions.
The smallest bulk densities, as affected
by water treatment, were generally in the
soil that had been flooded 1 wk before the
drains were installed (Fl) (Fig. 5), while
the largest was in the soil that had not been
flooded (D). The rate of reduction of the
bulk density with distance was greatest
with the D treatment and least with the F2.
DISCUSSION
The critical depth for this soil and
implement combination was between 50
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Flooded
None
1wk before
1d before
i r
160 210 260 310
and 70 cm and the thrust from the bullet
was primarily upward. At the 50- and 70-
cm depths, the overburden was suffi
ciently thin that the drag between the soil
immediately above the bullet and that
adjacent to this vertical plane were suffi
cient to allow lateral fracturing of the soil,
which fanned upward to the surface. With
the mole at 70-cm depth, some of this
upward thrust appears to have dissipated at
the mole, as fracturing began above the
mole and the width of the fan was not as
great as it was where the mole was
installed at 50-cm depth. At 90-cm depth
there was sufficient overburden to further
limit lateral fracturing; no fracturing was
apparent in some locations.
The data indicate that the soil around the
moles installed at 50-cm depth had been
disturbed during installation of the drains
in such a way that the bulk density to the
side of the mole was lessened, although
this effect was not observed during exam
ination of the soil. There is also evidence
of lateral disturbance at the 70- and 90-cm
depths. The minimums along radials 2 and
6 at 10-cm depth in the range of 110- to
225-mm distance suggest a reaction to the
mole. At 90-cm depth there is a general
minimum around the 185-mm distance.
These values are all less than those at 335
mm. Neither these differences nor signs of
effects were apparent while observing the
cutaway. Evidently there were compac
tion actions and rebound counteractions
while installing the moles.
Compaction of the soil around the mole
by the bullet, although statistically signifi
cant, appeared to be unimportant. The re
duction in porosity of the soil relative to
that of the undisturbed soil was minimal,
especially around the bottom portion of the
moles, where most of the water sup
posedly enters the drains. The lower bulk
densities at the perimeters of the drains are
attributed to rebound of the soil after the
Water entry into the moles should not be
a problem in this soil, regardless of depth
of installation or water content of the soil
sufficiently wet to be cohesive. The drag
between the bullet and the soil is sufficient
to pull away segments of the smeared skin,
leaving areas of unsmeared soil through
which water can easily enter the drains.
REFERENCES
CHILDS, E. C. 1942. The mechanics of mole-
draining. Emp. J. Exp. Agric. 10: 169-181.
DAY, P. R. 1965. Particle fractionation and
particle size analysis. Pages 545-567 in
CA. Black, ed. Methods of soil analysis.
Part 1. Agronomy No. 9. Am. Soc. Agron.,
Madison, Wis.
GILL, W. R. and G. E. VANDEN BERG.
1968. Soil dynamics in tillage and traction.
ARS, U.S. Dept. Agric. Washington, D.C.
Handb. 316.
GODWIN, R. J., G. SPOOR, and P. LEEDS-
HARRISON. 1981. An experimental inves
tigation into the force mechanics and result
ing soil disturbance of mole ploughs.
J. Agric. Eng. Res. 26: 477-497.
KIRKHAM, D, S. TOKSOZ, and R. R. VAN
DER PLOEG. 1974. Steady flow to drains
and wells. Pages 203-244 in J. van
Schilfgaarde, ed. Drainage of agricultural
soils. Agronomy No. 17. Am. Soc. Agron.,
Madison, Wis.
LEEDS-HARRISON, P., G. SPOOR, and
R. J. GODWIN. 1982. Water flow to mole
drains. J. Agric. Eng. 27: 81-91.
NAARDING, W. H. 1977. A review on inter
national experience with trenchless versus
trenching drainage-machines. Govt, of the
Netherlands, Ministry of Agriculture and
Fisheries. Aasen, The Netherlands.
SOMMERFELDT, T. G. 1984. Performance
of unlined and lined mole drains in a saline
clay loam field. Can. Agric. Eng. 26: 1-5.
SOMMERFELDT, T. G. and C. CHANG.
1986. Desalinization of an irrigated, mole-
drained, saline clay loam soil. Can. J. Soil
Sci. 67:263-269.
WINGATE-HILL, R. 1978. Tillage require
ments for cereal crop production on their
relationship to their development of new
tillage machinery. Pages 363-370 in W.W.
Emerson, R. D. Bond and A.R. Dexter, eds.
Modification of soil structure. John Wiley
and Sons, New York.
YOUNGS, E.G. 1985. An analysis of the
effect of the vertical Assuring in mole-
drained soils on drain performances. Agric.
Water Manage. 9: 301-311.
135
 SUBSURFACE IRRIGATION AND DRAINAGE ON SANDY SOIL
IN SOUTHERN QUEBEC

N. A. Nemon1, B. von Hoyningen Huene2, J. Gailichand2, and R. S. Broughton2

'Sind Agricultural University, Sind, Pakistan; 2Macdonald College of McGill University, Box 950, SteAnne de
Bellevue, Quebec H9X ICO

Received 26 July 1985, accepted 23 January 1987

Nemon, N. A., B. Von Hoyningen Huene, J. Gailichand, and R. S. Broughton. 1987. Subsurface irrigation and
drainage on sandy soil in southern Quebec. Can. Agric. Eng. 29: 137-142.

An experiment was conducted in 1982 and 1983 on flat sandy soil in southern Quebec to study the feasibility of
subsurface irrigation inthis region. The field experiment was designed on 10 hawith two treatments and eight replicates.
These treatments consisted of irrigated andnonirrigated maize plots. Itwas found that thesubsurface irrigation raised the
water table satisfactorily toa predefined water table position. This resulted inadequate water supply to thecroproot zone.
Subsurface irrigation treatment produced grain maize yields twice as high as the nonirrigated treatment.

INTRODUCTION
Drainage involves removal of excess
surface and subsurface water by means of
ditches and subsurface drain pipes to
provide sufficient aeration in the root zone
in a humid climate and control of salinity
in soils in irrigated semi-arid areas.
In some humid areas where water sur
pluses and deficits occur, drainage alone
may not be able to provide suitable moisture
conditions for optimum crop growth. Sup
plementary irrigation is needed to provide
water during periods of drought to achieve
MATERIALS AND METHODS ment was designed to use an existing sub
A field experiment was conducted in the surface drainage system, which had laterals
1982 and 1983 growing seasons on the 29 m apart and 1.05 m deep. This system
farm of Mr. Leandre Charbonneau, St. was modified to give a randomized com
Louis Parish, Richelieu county, Quebec. plete block design with two treatments and
The soil is a St. Samuel sand. The experi eight replicates. The treatments consisted
\ 1 \
> 29m
TYP.
A-1
<D A-2
<S> B-2

adequate plant growth.

Climatically, the southern Quebec farm o
t - *- -

ing region is suitable for subsurface irriga O

tion because there is excess precipitation

in winter and early spring, and often short
age of rainfall in summer. In the flat sandy
A-3
<?> B-3 B-4
<$> A-4

soil in this region, water can be removed

from the root zone in early spring by sub t 3CV *
surface drainage and in summer, irrigation
can be economically provided by adding
water to the drain pipes.
Subsurface irrigation systems must
B-5
<J> A-5 A-6 <t> B-6

meet two main requirements. That is, the

system must be capable of removing
excess water after periods of heavy rain t
fall, and supply the evapotranspirational
demand during dry periods.
B-7 A-7 A-8 B-8
Subsurface irrigation is practiced in
some parts of the Netherlands (Kalisvaart 0 4

1958)and the United States (Renfro 1955).
The design procedure for drainage is more
established than for subsurface irrigation.
t > I -

ROAD
In recent years subsurface irrigation has
O MUNICIPAL WATER COURSE
begun to be used on field scale in North
and South Carolina and Florida in the LEGEND^ fy NUMBER OF THE REPLICATION
U.S.A. (Kriz and Skaggs 1973; Skaggs et DIVISION BETWEEN REPLICATIONS ^ _ _ ^ _ ^
al. 1972; Doty et al. 1975;Doty and Chris- -e- CONTROL CHAMBER EEEE3ZZZEE 3
tenbury 1979; Doering et al. 1982). D DRAIN OUTLETS 0 25 50 100m
A PLOTS SUBSURFACE IRRIGATION
This paper describes the first series of B PLOTS SUBSURFACE DRAINAGE ONLY
experiments to determine the feasibility of
subsurface irrigation in southern Quebec. Figure 1. Drainage system after modification and location of the plots.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 137

 I 7 10 13 16 22 25 26 28 of irrigated and nonirrigated grain corn.
H ' <> <><>
Each plot contained two subsurface drains.
There were buffer zones between irrigated
and nonirrigated plots. Four water table
< ' !><> <) control chambers were installed in such a
t ^G- t way that water could be added at those
chambers and flow up the subsurface
10 13 14 16 17 20 26
drains to provide subsurface irrigation to
F<< i <> the eight irrigated plots. Water was pumped
into the control chambers from a reservoir
established in a nearby water course.
><>
Unfortunately the reservoir was not large
t t enough and did not receive sufficient sup
plemental inflow to supply the complete
3 7 13 15 21 24 25 27 water needs of the irrigated plots for the
4

dry summers. A well was drilled in late

I I August 1983 about 1.5 km to the west of
the experimental field and water was
<<>
pumped to the control chambers at the end
x
t t of growing season.
Sf B HI <><) (> (I o
During the subsurface irrigation in
1982, the water table responded satisfac
torily in the plot which had a new drain
A ' <<> << i <>
lateral added and in one other plot, (A-2
and A-4, Fig. 1). The water table did not
t t rise properly in the other six plots. Those
six plots did not absorb water as quickly as

=3 ROAD
MUNICIPAL WATER COURSE
expected when water was added to the
water level control chambers. It was found
LEGEND--0-CONTROL CHAMBER that some of the drain pipes were blocked
WATER TABLE PIPE
I 1 I- with sand. They had been installed in 1972
TENSIOMETERS 0 25 50 100m without being wrapped with an envelope
SUBSURFACE DRAIN material. New subsurface drain laterals,
enrobed with a knitted polyester envelope,
Figure 2. Location of water table pipes and tensiometers.
were installed in October 1982 on those six
field plots.
During irrigation in 1983, all plots
J L-
-PT-
TT accepted water adequately from the sub
120 . surface irrigation system approximately as
10 might be expected.
100 Water table observation pipes were
y\
s installed in each of the irrigated and non-
20
E
<
irrigated plots. Four tensiometers were
E 80 installed at 15-, 30-, 45-, 60-cm depths in
SOIL AVAILABLE
DEPTH WATER
30 each of the 16 plots (Fig. 2).
60 cm While irrigation was underway, mea
I IRRIGATED
surements were made of the depth to the
UJ
30 water table. The soil moisture content was
NON-IRRIGATED
-j 40 measured gravimetrically for samples
CD
<
IRRIGATED
taken at 7-, 22-, 37-, 52-, 67-cm depths to
| 20 45
NON-IRRIGATED 10 E represent the depth ranges of 0-15, 15-
E 30, 30-45, 45-60, and 60-75 cm,
respectively. The bulk density was mea

1 cr
sured at 5-, 20-, 35-, 50- and 65-cm depths
to convert moisture content by weight
basis to a volume basis. The soil moisture
o tension was measured with tensiometers.
r"-1"i 1 1 1 1 1 The amount of water added to the cham
10 20 30 10 20 30 10 20 30 bers was measured with water meters. The
JULY AUGUST SEPTEMBER
yield of maize was measured in Octoberof
Figure 3. Average remaining available water in 30- and 40-cm soil depths in irrigated and each year following the methodof Raghavan
nonirrigated plots, 1982. and McKyes (1977).

138 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 time during the field experiments of 1982.
TH*-
120 While the 30- to 45-cm soil layer stayed
above field capacity in the subirrigated
10 plots. The soil moisture content in the non-
100 irrigated plots was found to be below field
2 capacity in all four layers during the field
20
E experiments of 1982. Similar results were
<
E 80
SOIL AVAILABLE
DEPTH
obtained for the 1983 experiment (see
WATER
of cm 30 Table II).
UJ
60 IRRIGATED Referring to Fig. 4, it is seen that the
i 30
NON-IRRIGATED
remaining available water decreased when
the subsurface irrigation stopped after 19
-j 40
GO
IRRIGATED July and 10Aug. The water table dropped
45
<
NON-IRRIGATED- due to evapotranspiration. This shows that
< 20 10 the upward flow from the water table was
not sufficient to supply ET demands. This
resulted in a loss of soil moisture and a
reduction of the available water in the soil
5 5 o
layers.
cr
Crop Yield
r t 1 r
o s During the 1982 growing season, yields
10 20 30 10 20 30 10 20 30 of maize were obtained from two irrigated
JULY AUGUST SEPTEMBER and two nonirrigated plots. The results are
Figure 4. Average remaining available water in 30- and 40-cm soil depths in irrigated and presented in Table III.
nonirrigated plots, 1983. The analysis of variance of the 1982
grain yield results was not carried out
because there were no observations col
RESULTS AND DISCUSSION remaining available water in the 0- to 45- lected from the six other replicates of irri
Soil Water Response to Subsurface cm soil profile. The remaining available gated treatment due to blockage of the
Irrigation water in the 0- to 30-cm soil profile drain lines. However, the magnitudes of
In order to observe the response of sub remained nearly constant during the sup yields are given in Table III. On the aver
surface irrigation during the 1982 and 1983 ply of subsurface irrigation water. There age the irrigated treatment produced 40%
field experiments, the remaining available was approximately twice as much avail more grain yield than the nonirrigated
water within 30- and 45-cm soil columns able water for the irrigated as for the non- treatment. Table III shows a considerable
was plotted in Figs. 3 and 4 along with the irrigated treatment. For the subsurface variation in yield of maize within the irri
recorded subsurface irrigation volumes irrigation treatment, as shown in Table I, gated plots. A similar situation was observed
and the rainfall data. Remaining available the soil moisture in the 15- to 30-cm soil in nonirrigated plots. The lower yields in
water is defined as the difference between layer stayed near field capacity most of the the A-2 plot might be due to the water table
the observed soil moisture content and the
TABLE I. VARIATION OF WATER TABLE AND SOIL MOISTURE CONTENT IN IRRIGATED
permanent wilting point. AND NONIRRIGATED PLOTS OF ST. SAMUEL SANDY SOIL IN 1982 FIELD EXPERIMENTS
Figures 3 and 4 show the results of the
Average soi 1 moisture content in percent by volume for 15-
soil moisture content at various depths Water table
cm soil layers at the following depths (cm)
during the growing seasons of 1982 and depth
Date (cm) 0-15 15-30 30-45 45-60
1983, respectively, for irrigated and non-
irrigated treatments. Soil moisture content Irrigated
is presented in the form of remaining 20 July 58.0 15.0 16.0 34.7 36.0
available water for the 0- to 30-cm and 0- 21 July 70.0 14.0 24.4 29.0 37.1
22 July 78.0 13.6 21.0 29.0 36.0
to 45-cm soil profiles. For the nonirrigated
23 July 70.2 7.4 29.0 29.0 35.0
treatment, there was little remaining avail 04 Aug. 77.0 7.5 27.4 35.0 38.0
able water during both the 1982 and 1983 05 Aug. 72.0 7.4 29.8 35.0 38.0
growing seasons. During the field experi 13 Aug. 90.6 18.0 27.4 34.7 37.1
ments of 1982, the remaining available 20 Aug. 102.5 14.0 19.7 27.4 34.7
24 Aug. 110.0 15.2 17.9 26.0 32.0
water for the irrigated treatment in the 0- to Nonirrigated
30-cm and 0- to 45-cm soil profiles was 20 July 127.2 9.5 9.9 18.5 20.5
never less than 38 mm and 75 mm, respec 21 July 128.6 9.0 9.7 16.7 18.5
tively. On the other hand, remaining avail 22 July 129.0 8.6 10.2 15.7 16.8
23 July 130.1 8.8 9.2 17.7 18.5
able water for the nonirrigated treatment
04 Aug. 134.0 9.0 9.0 16.7 18.5
was below 20 mm and 40 mm for the 0- to 05 Aug. 135.4 9.0 9.8 16.7 17.7
30-cm and 0- to 45-cm soil profiles, 13 Aug. 131.0 16.5 11.9 20.5 21.8
respectively. It can be seen in Fig. 3 that 20 Aug. 132.0 14.1 11.0 17.8 20.0
the upward flow from the water table 24 Aug. 135.4 15.1 9.2 17.8 23.0

increased the soil moisture content in the Field capacity is equal to 25.7% by volumebased on a fieldcondition with approximately 600-mm suction in
30- to 45-cm soil layer, thus increasing the the pore water. Permanent wilting point is 3.1% by volume at 15 bars suction.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 139

TABLE II. VARIATION OF WATER TABLE AND SOIL MOISTURE CONTENT IN IRRIGATED beyond the tensiometric range.
AND NONIRRIGATED PLOTS OF ST. SAMUEL SANDY SOIL IN 1983 FIELD EXPERIMENTS In 1983, all irrigated plots received
Average so il moisture content in percent by volume for 15-cm soil water from subsurface drain pipes satisfac
Water table
depth
layers at the following depths (cm) torily due to improvements made to the
Date (cm) 0-15 15-30 30-45 45-60 60-75 subsurface irrigation system in 1982 and
1983. Grain and ear yields were measured
Irrigated in the 1983 growing season to examine the
20 July 86.2 23.5 22.1 35.1 33.0 42.2
22 July 96.8 18.8 19.4 32.2 33.9 41.2
effects of subsurface irrigation. Table IV
26 July 111.9 15.7 20.7 28.7 30.0 39.8 gives the analysis of variance on the
27 July 109.7 17.2 19.1 28.3 30.6 38.4 results of maize yields. This table indi
28 July 111.7 16.7 18.8 28.1 30.3 37.5 cates that the increases in grain yield and
29 July 117.6 18.6 20.9 28.0 27.3 37.5
02 Aug. 118.0 17.2 20.5 22.0 25.6 34.0
ear yield of maize are significant at the
05 Aug. 129.7 13.6 16.7 19.6 17.0 25.2 0.01 probability level. Duncan's new mul
31 Aug. 99.6 23.0 21.9 25.9 30.0 37.2 tiple range test was used to compare the
02 Sept. 95.4 19.3 22.6 26.3 29.0 33.9 mean yields of irrigated and nonirrigated
05 Sept. 88.4 18.8 22.0 29.5 34.2 42.2
treatments, the results of which are pre
08 Sept. 72.8 16.1 20.5 31.4 37.2 42.2
Nonirrigated sented in Table V. The results indicate that
20 July 131.2 14.0 14.7 19.7 24.5 33.6 the mean yields of the irrigated treatment
22 July 132.6 12.3 13.4 20.4 25.7 35.0 are significantly different than the mean
26 July 131.0 10.1 12.5 19.0 29.5 34.9
yields of the nonirrigated treatment at 0.05
29 July 132.0 10.3 11.5 17.6 24.8 34.5
02 Aug. 133.8 10.9 12.5 19.0 25.4 35.1
probability level. The table also shows that
05 Aug. 134.8 8.5 12.9 19.3 25.1 36.8 the mean yields of the irrigated treatment
10 Aug. f 10.7 12.5 17.3 23.6 31.1 are double those of the nonirrigated treat
23 Aug. t 7.8 7.8 11.6 17.0 t ment.
08 Sept. t 7.5 7.8 14.6 26.4 29.4
Table VI gives the results of maize
t Indicates that the water table was below the observation pipe. yields obtained from individual plots of
^Indicates that the moisture content was not realized in the field. irrigated and nonirrigated treatments in
Field capacity is equal to 25.7% by volume. Permanent wilting point is 3.1% by volume at 15 bars suction.
1983. It was observed from the yield mea
surements that the cobs were healthier and
better filled in irrigated plots than in nonir
TABLE III. AVERAGE YIELD OF MAIZE being closer to the soil surface over the rigated plots. For example, the A-6 and
FROM SUBSURFACE IRRIGATED AND NON- drain. During field observations, the water B-2 plots produced an equal number of
IRRIGATED PLOTS ON AN OVEN-DRY
table depth at the G-22 pipe was found to cobs but the ear yield and grain yield were
BASIS FROM 1982 MAIZE CROP
be 18-22 cm below the soil surface for 7 higher in plot A-6 than in plot B-2. Similar
Grain yield Average consecutive days at the time of tasseling results were obtained from A-7 and B-5.
on dry basis grain yield and silking. The yield reduction in the B-2
Plots (kg/ha) (kg/ha) Although one irrigated plot (A-4) has the
plot may be due to the deep water tables. same number of cobs as that of A-6, it
Irrigated The water table in this plot was found to be produced a lower grain yield as compared
A-2 4413
5314
5-6 cm deeper than the B-4 plot during to plot A-6. This may be due to delayed
A-4 6215
the field observations. Also in the B-2 tasseling and silking in this plot and also
Nonirrigated
B-2 3259
plot, the soil at 30-cm depth below the soil many cobs from plot A-6 were affected by
3788 surface was quite dry from 20 July through
B-4 4316 white smut.
4 Aug. and the soil suction readings were One factor which might have caused the
lower yield of nonirrigated plots was stunt
ing of the plants due to soil moisture defi
TABLE IV. ANALYSIS OF VARIANCE FOR YIELDS OF MAIZE IN 1983
cit. In addition to the observed stunting,
Source DF Sum of sq. Mean sq. F Pr>F R CV the moisture deficit also delayed tasseling
A. Grain yield
and silking on nonirrigated plots by about
Model 15 160 081 205 10 672 080 9.46 0.0001 0.689 28.65
4 to 5 days. Table VI indicates that the soil
Error 64 72 165 114 1 127 579 Root MSE Grain yield Mean moisture deficit stunted the size of the cobs,
Corr.Tot 79 232 246 319 1061 3705 thus reducing the ear yield. There is also
Source DF ANOVA SS
evidence that some of the nonirrigated plots
F Pr>F
did not yield as many cobs due to a severe
Block 7 28420836 0.90 0.5539 soil moisture deficit. The last column in
Treatment 1 100054301 22.16 0.0022**
Table VI gives the grain yield of maize
B. Ear yield obtained by the combine harvester. Ap
Model 15 214 451 160 14 296 744 9.22 0.0001 0.684 28.35 proximately the same yield was obtained
Error 64 99 226 863 1 550 419 Root MSE Ear yield Mean by hand sampling.
Corr.Tot 79 313 678 023 1245 4392 Crop yield response to subsurface irri
Source DF ANOVA SS F Pr>F
gation draws attention to the results of soil
moisture data discussed above. These results
Block 7 38049104 0.89 0.5598
reveal that the available water in the irri
Treatment 1 133595389 22.85 0.0023**
gated treatment was never limiting in the
**Yields from irrigated plots were greater than from nonirrigated plots at the 0.002 level of significance. uppermost layers during the time when

140 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

TABLE V. AVERAGE YIELD OF MAIZE IN SUBSURFACE IRRIGATED AND NONIRRIGATED ing season of 1982 and 1983, the supply
PLOTS ON AN OVEN-DRY BASIS FROM 1983 MAIZE CROP was adequate before and after the tasseling
Machine
period. For that reason, tasseling began 4
Hand sampli ng harvest or 5 days earlier in irrigated plots than in
No. of Ear yield Grain yield Grain yield nonirrigated plots. In addition to the
Plots ears (kg/ha) (kg/ha) (kg/ha) observed tasseling, the plants in the irri
gated plots were found to be taller than in
Irrigated 425644 56854 48244 4445
Nonirrigated 347975 31005 25875
the nonirrigated plots. The difference was
2431
approximately 40-50 cm at maturity.
All the figures are average of eight plots. The weather data in Table VII show that
4,5Duncan's new multiple range test was used for comparing means. Means with the same letter are not the 1983 growing season was relatively
significantly different at the 5% probability level.
drier than 1982 (from 15 May to 10 Sept.).
The total rainfall in 1982 was lower than
TABLE VI. YIELDS OF MAIZE FROM SUBSURFACE IRRIGATED AND NONIRRIGATED
the 40-yr average at the St. Hyacinthe sta
PLOTS ON AN OVEN-DRY BASIS FROM 1983 CROP
tion. The monthly rainfall for 1982 and
Machine 1983 was also lower than the 40-yr average
Hand sampling harvest
monthly rainfall at the above station. It
No. of Ear yield Grain yield Grain yield appears that both experimental years were
cobs on dry on dry on dry
per
relatively dry years indicating the need for
basis basis basis
Plot no. hectare (kg/ha) (kg/ha) (kg/ha) irrigation water for better crop production.
The relatively low yields of maize in non-
Irrigated irrigated plots may be attributed to low
A-l 45981 5700 4815 4206
A-2 42253
rainfall in those years. Total amounts of
5059 5260 3882
A-3 45981 6922 5861 4673
subsurface irrigation supplied in 1982 and
A-4 39768 3503 2936 2955 1983 were 60.3 and 167.6 mm, respec
A-5 43496 5980 5131 4625 tively. Total rainfall and the subsurface
A-6 39768 5334 4521 3943 water applied raised the 1982 water supply
A-7 42253 6560 5610 5814
A-8 41010 6423 5457 5465
in irrigated plots to 440.3 mm and the 1983
Mean 42564 5685 4824 4445 total to 548.1 mm. May 1983 rainfall was
Nonirrigated higher than that of May 1982; therefore,
B-l 34379 3138 2611 2301 the planting in 1983 was delayed for 15
B-2 39768 3967 3332 2118
B-3 37382 3860
days. Comparatively lower yields obtained
3222 3128
B-4 34797 2785 2301 2278
in 1983 than 1982 might be due to delayed
B-5 42253 4722 3085 3490 planting. Also, the maize matured very
B-6 34797 2829 2371 2828 rapidly and dried due to hot dry windy
B-7 27340 2058 1715 1722 weather at the end of August and the first
B-8 27340 1443 1159 1583
Mean 34797 3100 2587 2431
days of September 1983.
From the above results and discussion,
subsurface irrigation appears to be bene
TABLE VII. MONTHLY RAINFALL, PE AND SUBSURFACE IRRIGATION VOLUME FOR 1982 ficial in flat sandy soil in southern Quebec.
AND 1983
Controlling the subsurface drain outflow
Monthly Monthly conserves soil moisture and provides more
average subsurface water for dry periods resulting in better
rain irrigation
Rainfall PE 40-yr
plant growth and increased yields.
volume
Month (mm) (mm) (mm) (mm)
REFERENCES
1982
15-31 May DOERING, E. G., L. C. BENZ, and G. A.
25.1 37.9 84.3 0.0
June 97.5 118.8
REICHMAN. 1982. Shallow-watertable
102.1 0.0
July 25.0 157.8 109.7 44.2
concept for drainage design in semi-arid and
August 120.9 117.4 109.0 16.1 subhumid regions. In Advances in drainage.
1-10 Sept. 64.8 26.4 110.6 0.0 Proceedings of the Am. Soc. Agric. Eng.,
Total 333.3 458.3
Fourth National Drainage Symposium, St.
515.7 60.3
Joseph, Mich.
1983 DOTY, C. W. and G. D. CHRISTENBURY.
15-31 May 89.6 66.2 84.3 0.0 1979. Controlled and reversible drainage,
June 24.9 119.2 102.1 0.0 past, present, and future. Paper No.
July 38.6 125.0 109.7 68.8 79-2545, Am. Soc. Agric. Eng., St. Joseph,
August 69.2 111.2 109.0 55.8 Mich.
1-10 Sept. 7.4 44.3 110.6 43.0 DOTY, C. W., S. T. CURRIN, and R. E.
Total 229.7 465.9 515.7 167.6 McLIN. 1975. Controlled subsurface drain
age for southern coastal plains soil. J. Soil
Water Conserv. 30: 84-85.
subsurface irrigation was applied. When most layers supplied the ET needs as KALISVAART, I. C. 1958. Subirrigation in
the water table receded below 100 cm from evidenced by soil moisture depletion in the Zuiderzee polders. Int. Inst, for Land
the soil surface, the upward flux was lim Figs. 3 and 4. Although subsurface irriga Reclam. and Improv. Wageningen, The
ited and the available water in the upper tion was not uniform throughout the grow Netherlands. Publ. 2.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 141

KRIZ,G. J. and R.W. SKAGGS. 1973. Water
management using subsurface drains. J. Soil
Water Conserv. 28: 216-218.
RAGHAVAN, G. S. V. and E. McKYES.
1977. Study of traction and compaction
problems on Eastern Canadian agricultural
soils. Report presented to Agriculture Can-
ada Engineering Research Services. Mac-
donald College of McGill University, Ste
Anne de Bellevue, Quebec, pp. 63-64.
RENFRO, G. 1955. Applying water under the
surface of the ground. USDA Year Book
of Agriculture, Washington, D.C. pp.
273-278.
SKAGGS, R. W., G. J. KRIZ, and R. BER-
NAL. 1972. Irrigation through subsurface
drains. J. Irrig. Drain. Div,. (Am. Soc. Civil
Eng.), 90(IR3): 363-373.

,42 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 PERFORMANCE OF LOW-PRESSURE CENTER-PIVOT SPRINKLER
IRRIGATION SYSTEMS IN SASKATCHEWAN

K. P. Thooyamani and D. I. Norum

Department of Agricultural Engineering, University of Saskatchewan, Saskatoon, S7N 0W0

Received 19 November 1986, accepted 23 January 1987

Thooyamani, K. P. and D. I. Norum. 1987. Performance of low-pressure center-pivot sprinkler irrigation systems in
Saskatchewan. Can. Agric. Eng. 29: 143-148.

Uniformity of water distributions under six low-pressure center-pivot sprinkler irrigation systems was measured to
evaluate the systems. For a uniform overlapped spray pattern, uniformity of water distribution as per design was estimated
from sprinkler charts for each system. This design uniformity was compared to uniformities measured under field
conditions to find the suitability of design and/or management of the systems for field conditions. It was concluded that the
low-pressure center-pivot systems were properly designed for sprinkler discharge rates, but the resulting uniformity of
overlapped spray pattern was not properly considered. Thus, there is room for improvement in the design of low-pressure
center-pivot systems.

INTRODUCTION
Center-pivot irrigation systems are clas
sified as low-pressure or high-pressure
systems based on the operating pressure at
the pivot. Although there is no definite
boundary between low and high pressure,
in general, a system is classified as low
pressure if the pivot pressure is less than
350 kPa.
In spite of high capital investment,
sprinkler irrigation systems are replacing
surface irrigation systems because of auto
mation and flexibility of management. Out
of 76 890 ha of total irrigated area in Sas
katchewan, 37% was irrigated by sprinkler
systems in 1978. It was 3% in 1950 and 8%
in 1960 and 1970 as shown in Fig. 1
(Prairie Farm Rehabilitation Administra
tion (PFRA), 1982). In 1985, 70% of total
irrigated area was serviced by sprinkler
irrigation systems (Hill 1985).
To avoid runoff and erosion a center-
pivot system is usually designed and oper
ated to apply water at an application rate
equal to or less than the infiltration rate of
ate towers. Thooyamani (1982) conducted
16 field trials to measure the circular uni
formity, and reported that the circular uni
formities were above the acceptable level
of 0.8. Thus, radial uniformity was
selected to evaluate the systems.
The objectives of this research were:
(a) To evaluate the performance of
center-pivot sprinkler irrigation systems
under field conditions, in terms of water
application uniformity.
(b) To evaluate the design of the center-
pivot systems in terms of the uniformity.
(c) To compare the design uniformity
to the observed uniformity to find the suit
ability of design and/or management for
field conditions.
REVIEW OF LITERATURE
Murray et al. (1982) studied the perfor
mance of low-pressure and high-pressure
center pivots in terms of uniformity of
110
application and energy consumption. The
performance was compared to that of an
automated surface irrigation system. They
reported that energy requirements of high-
pressure and low-pressure center pivots
were approximately 100 and 6.50 per
ha-mm of water, respectively, based on
1980electrical power rates (approximately
4.20 kWh). It was concluded that an effi
cient automated surface irrigation system
can be designed to distribute water as
uniformly as low-pressure center-pivot
systems.
Norum et al. (1983) studied the perfor
mance of low-pressure center-pivot sprin
kler irrigation systems in the Outlook
Irrigation District, Saskatchewan. Unifor
mity of application in both the circular and
radial directions was evaluated under eight
different systems. It was concluded that all
the systems were distributing water above
the generally accepted level of uniformity
(0.8).

the soil. The speed of rotation around the 100

field, and to some extent the operating /98
o
pressure, can be regulated to avoid runoff 90

and erosion. Average depth of application

80
can also be regulated by speed of rotation. b
The main problem under center-pivot irri o 70 Z11
#
-

o
gation systems is the nonuniform applica
tion of water. Thus, uniformity of applica 60 .
/ 70%
tion was selected to evaluate the systems. 50
&/
Uniformity of application in circular as
well as radial directions should be consid 40

V &/
ered for a complete evaluation of a sprin V
30 -

kler system. Circular uniformity depends /37%

on radius of throw of the sprinklers and the 20 .

maximum allowable offset for intermedi- 20

10 J5
3% 8% /8% 1978 1985

This paper was presentedat the annual meetingof the 1950 I960 1970 1980 1990
Canadian Society of Agricultural Engineers, 6-10 YEAR
July 1986, Saskatoon, Saskatchewan. Paper no
86-303.
Figure 1. Growth of sprinkler irrigation in Saskatchewan.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 143

 TABLE I. SYSTEM IDENTIFICATION AND DESCRIPTION 2tt
Ucc = 1
Ad
% [di-d]S-lRi (1)
Sprinkler
Pressure (kPa) Data
System Flow rate Length'!" spacing where Ucc = the Christiansen uniformity
ID (L/sec) (m) (m) Pivot End* Year Trials
coefficient, dx = the depth of application
A 46.00 377 2.53 260 138 1980 6 at the /th station (mm), d = the average
1981 4
depth of application (mm), S{ = the radial
B 49.07 409 2.49 240 133 1980 5
1981 3
increment represented by the /th station
CH 55.20 476 Variable 355 205 1982 2 (m), Rx = the radial location of the /th
DH 55.20 415 Variable 265 140 1982 3 station (m), and A = total area of irriga
El 79.73 554 Mixed 380 170 1982 2 tion (m2).
Fl 46.00 418 3.05 240 150 1982 2
Depths of water applied in radial direc
t Includes effective end gun radius. tions were measured under the six systems
$ Before booster pump, if any. during the summers of 1980-1982.
Data from Murray et al. (1982).
11 Data from Norum et al. (1983).
System identification and descriptions are
given in Table I.

Estimation of Design Uniformities

It is usually assumed that there is equal
tradeoff of water between adjacent sprin
kler heads. In this paper, thus, it was as
sumed that
(a) The radius of throw touches the
bases of adjacent sprinklers.
(b) An individual sprinkler sprays
water in a triangular (conical) spray pat
tern shape.
From the sprinkler specifications for a
system, the location of each sprinkler and
its discharge rate can be obtained. But the
o. o
discharge volume (or depth of application)
from each sprinkler is required to estimate
o. o 100.0 200.0 300.0 400. O
Ucc, not the discharge rate. Since unifor
RADIAL DISTANCE FROM PIVOT, m mity of application is independent of rota
tion speed of a system (Thooyamani 1982),
PERCENTAGE OF IRRIGATION AREA
any reasonable duration of application per
0.0 25. 0 50. 0 75. 0 100.0 revolution can be assumed. In this paper, a
1 , 1 1ii j 1 i r - | -r t ii i^
UJ
0.0 1 1 1 r
duration of 50 h per revolution was chosen.
LO
<
Based on the triangular pattern, the depth
/ " LOW of application at each sprinkler location
J^^^
a:
uj
>
-
QUARTER was estimated as
<
AVERAGE
1.0- Qt
a
--^""""'average di- (2)
UJ
2-rrtfjSi

a. f where dx = depth of application at the

a.
< 2 base of the /th sprinkler (mm), Q = the
discharge rate from /th sprinkler (L/sec),
u

-/ and t = the duration of application per

2.
revolution (sec). Ucc was then estimated
from Eq. 1.
3.0

Figure 2. (a) An example of a poor distribution of water in radial direction. (System A, average RESULTS AND DISCUSSION
application = 10.2mm, Ucc = 0.54); (b) Infiltration profile under System A.
Radial Uniformity for Field Conditions
The average depths of application in the
PROCEDURE operating under normal field conditions, 27 trials ranged from 7.5 to 24.2 mm, with
except that the test periods were chosen so an overall average of 15.0 mm. Figures 2
Estimation of Uniformities under Field that the wind was relatively light. One- and 3 show examples of two radial dis
Conditions liter cans of 100 mm diameter were posi tributions.
All field tests were carried out on far tioned on stands approximately 0.90 m From Fig. 2(a), (b), it can be seen that
mer-operated center-pivot machines in the above ground level, and were spaced System A performed poorly. Only 43% of
vicinity of the Outlook Irrigation District mostly at 5.0 m, but in some instances at the area received an irrigation of average
in Saskatchewan (Murray et al. 1982; 10.0 or 15.0 m. depth or greater. Another parameter to
Norum et al. 1983). Water distributions For each trial, a weighted Christiansen evaluate the water distribution is the aver
were measured while each machine was uniformity coefficient was estimated from age depth of application in the least irri-

144
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 Figure 3(a), (b) shows the performance
20. 0 of a well-designed and operated center-
Q pivot system. Even though only 45% of
UJ
the land received a greater-than-average
5 15.
Q.
depth of irrigation, the low quarter average
< was about 90% of the average depth. Max
cc imum depth of application was only one
10.0 and one-half times the average depth.
<
Thus, for System F, most of the land was
u. adequately and efficiently irrigated.
o
5. 0 Radial uniformity of application, as
X
h-
observed under field conditions for the 27
Q_
UJ
trials under the six systems is given in
_L _L _J_
Q
0
_i i i_
Fig. 4. Taking 0.8 as the acceptable uni
0. 0 100.0 200.0 300.0 400.0 formity level, the uniformities for 17trials
were below the acceptable level. Thus, out
RADIAL DISTANCE FROM PIVOT, m
of 27 random samples, only 37% of the
PERCENTAGE OF IRRIGATION AREA time did the low-pressure systems perform
0.0 25. 0 50. 0 75. 0 100.0
adequately under field conditions.
0. 0| 1 r -j 1 1 1 n 1 1 1 1 1 1 r- -r-ii r-

UJ
o
Radial Design Uniformities
<
Radial design uniformity of a system is
UJ
> the uniformity that can be obtained under
< 0.5
ideal field conditions such as no wind,
level ground surface, etc., provided the
Q
UJ
LOW
overlapped spray pattern is uniform. Thus,
~ 1.0-
_l
Q_
the design uniformity is the theoretical
AVERAGE QUARTER
Q_
<
maximum possible uniformity of applica
CC
AVERAGE tion that can be expected under a system.
1.5
UJ
Examples of radial distributions that could
be expected according to the designs for
two systems, namely System A and Sys
2. 0L tem F, are given in Figs. 5 and 6.
Figure 3. (a) An example of good distribution of water in radial direction. (System F, average Figures 5(a), (b) and 6(a), (b) show that
application = 12.0mm, Ucc = 0.87); (b) Infiltration profile under System F. efficient and adequately uniform distribu
tion of water can be achieved under the
system if a uniform overlapped spray pat
1. 0 tern can be produced. Under System A
more than 50% of the area can be expected
z 0. 9 to receive an irrigation greater than the
UJ
-
s average depth, while the maximum depth
(J 4

0. 8
_ _ _ ^_ _ of irrigation is only 1.25 times the average,
u.
u_ m and the low quarter average is 90% of the
UJ
o
0. 7
average. Under System F, more than 55%
u
of the area can be irrigated with at least the
average depth, while the maximum and low
0. 6 "*

2:
quarter average depths are only 1.25 and
CC
o
0.9 times the average depth, respectively.
0. 5
Ll Design uniformities for the six systems,
z
1 1 1
as operated on the Outlook Irrigation Dis
n a 1 1 1
trict area, are given in Fig. 7. The design
A B C D uniformity for the systems ranged from
SYSTEM IDENTIFICATION
0.90 to 0.95. Apparently all six systems
had the capability to distribute water well
Figure 4. Radial uniformities under low-pressure sprinklers.
above the acceptable level of uniformity
(0.8). Thus, a low-pressure center-pivot
gated 25% of area, known as the low the average. Thus, some areas had a sprinkler irrigation system can be designed
quarter average (Hart and Heermann severe moisture deficiency while other to distribute adequate water uniformity, if
1976). Essentially, the low quarter average areas had heavy deep percolation losses. a uniform overlapped spray pattern can be
indicates the severity of moisture defi Crop yields may be reduced in both areas produced under the system.
ciency in the least irrigated area. The low resulting in a double loss to the farmer. A
quarter average for System A was less than well-designed and operated surface irriga Comparison of Uniformities
one-half of the average. Maximum depth tion system can distribute water better than Design uniformity and uniformities
of application was two and one-half times the infiltration profile for System A. under field conditions were compared by

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 145

 TABLE II. WATER JET VELOCITY AND
30. 0
EQUIVALENT WIND SPEED FOR SYSTEM F
Q
Distance Water jet Equivalent
from pivot velocity wind speed
OL
(m) (m/s) (km/h)
2: 20.
5.0 19.5 70.2
cr
UJ 63.0 18.5 66.6
<
124.0 17.6 63.4
185.1 16.7 60.1
u. io. 243.1 16.1 58.0
o 304.0 15.7 56.5
365.0 15.5 55.8
395.7 15.4 55.4
Q_
UJ
Q _L _L J_
0. 0
water jet from System A was 27 km/h.
0. 0 100. 0 200. O 300. 0 400. 0
Under System A, a wind of 30 km/h can
RADIAL DISTANCE FROM PIVOT, m effectively reduce the windward radius of
throw to zero and can elongate the leeward
PERCENTAGE OF IRRIGATION AREA
radius of throw to more than double. A
0.0 25. 0 50. 0 75. 0 100.0 light wind with a speed of 10-15 km/h can
0. Qiii.>p -i 1 1 r-
T TT 1 1
"T considerably distort the spray pattern
UJ

< under System A and can cause low unifor

en
UJ mity of application. Since the initial vel
>
< 0.5 ocity of the water droplets from System F
was much greater than the speed of the
a light wind, the light wind had little effect
UJ
~ 1.0- on spray patterns under System F. Thus, a
_j LOW
Q_
Q_ AVERAGE
wind of 10 km/h is negligible wind for
< QUARTER System F, but detrimental for System A.
cr
UJ 1.5 AVERAGE System A was made vulnerable to wind by
<
installing flow regulators, even though the
flow regulators solved the problem of
2. 01- unevenness of the field. Flow regulators,
of course, should not reduce nozzle pres
Figure 5. (a) Expected radial distribution under System A. (average application = 18.9 mm, sures more than necessary. This could
Ucc = 0.93); (b) Expected infiltration profile under System A. explain the poor performance of System A
compared with System F, even though
both systems were operated at 240-260
calculating the ratio tion. Wind continuously distorts the spray kPa with the same flow rate, 46 L/sec.
Ucc field conditions
pattern shape of the sprinklers before Irrigation during unfavorable wind condi
Ur
(3) water droplets reach the ground surface or tions can be avoided but the wind speed
Ucc design capability
crop canopy. When a center-pivot system and direction cannot be controlled. Thus,
The ratio UR indicates the effects of field is operated during a windy period, the wind and excessive reduction of water jet
topography, history of wind speed and leeward radius of throw is elongated and velocity could be one of the reasons for the
direction, management of the systems, the windward radius of throw is reduced. center-pivot systems performing below
and the design procedures. If field topog Because of the distortion, water is applied their expected level of uniformity.
raphy was uneven and the factor was con in a pattern different from that of the de
sidered during design to produce a uniform sign, resulting in a nonuniform overlapped Nonuniform Overlapped Spray Pattern
overlapped spray pattern, UR will be close spray pattern and low uniformity of appli When estimating design uniformity of a
to 1 under no wind or negligible wind cation. system, a uniform overlapped spray pat
conditions. Figure 8 compares the perfor The magnitude of the distortion depends tern was assumed. An ideal design should
mance of the low-pressure systems. on the height of the sprinklers, the wind produce a uniform overlapped spray pat
Hart (1961) reported that depths of speed and the velocity of water droplets as tern, regardless of individual spray pattern
application under sprinklers can be charac they leave the sprinklers. For all six sys shapes. Due to a lack of knowledge about
terized by a normal distribution. Based on tems, the height of the sprinkler could not the effects of spray pattern shapes on uni
the normal distribution, possible varia be adjusted. For some field tests, the wind formity of application, this is often
tions in the uniformity coefficient at 95% was relatively calm but, in some tests, the ignored during design. The ill-effects due
confidence are shown as vertical lines in wind speed was as great as 30 km/h. The to high application rate can be seen during
Fig. 8. Note that all six systems consis velocity of the water jet and the equivalent irrigation as surface runoff, thus it gets
tently distributed water below their ex wind speed at various points along the higher priority. The ill-effects of non
pected (design) level. lateral of System F are given in Table II. uniform application of water can only be
In the case of System A, the flow rates seen after the damage has been done, that
Wind Speed and Direction were controlled by regulators and the pres is as an uneven crop stand.
Wind speed and direction during irriga sure was maintained at 138 kPa at all sprin Under low-pressure sprinklers, spray
tion influence the uniformity of applica klers. The equivalent wind speed for the pattern shapes are usually "donut" shaped

146 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 30. 0
spacings, based on average application
rate only, but not on uniformity of over
a
UJ lapped spray pattern, could be the other
t

_i cause for the poor performances of the

a.
a. systems.
< 20. 0
The topography of the land can affect
cr
LU line pressure causing uneven discharge
i-
< from sprinklers. However, the effects can
be taken into consideration during design,
u_ 10. 0
a and pressure regulators can be used if
necessary.
CL
Management practices such as main
UJ
Q -j i i L J i L_ J i 1 _j L_J
taining pivot pressure, keeping nozzles
0. 0
free from clogging, etc., help the system
0-0 100.0 200.0 300.0 400.0 500. 0 to perform according to design. Failure to
RADIAL DISTANCE FROM PIVOT, m maintain the system may result in low uni
formity of application of water.
PERCENTAGE OF IRRIGATION AREA The uniformities of application under
0.0 25.0 50.0 75.0 100.0
the center-pivot systems may be improved
0. 0|iiii| r- ~i 1 1 n1~
considerably if the redistribution within
T" inir-
UJ
CD the soil profile is taken into consideration.
<
cr Hart (1972) showed examples of a unifor
UJ
> mity coefficient of 0.60 for the water dis
< 0.5
tribution at the ground surface, becoming
0.76 and 0.86 after redistribution in the
soil for 1 and 2 days, respectively. The
- 1.0
CL LOW
effect of redistribution within the soil is,
CL
<
AVERAGE of course, much dependent upon the spa
QUARTER
cr
tial distance between above-average and
1.5
UJ AVERAGE below-average applications.
<

Seasonal Uniformity
2. O1- Crop yield is not determined by a one
time distribution of water under a pivot
Figure 6. (a) Expected radial distribution under System F. (average application = 14.8 mm,
Ucc = 0.96); (b) Expected infiltration profile under System F.
system. Seasonal uniformity of applica
tion and overall spatial distribution of
water should be considered. Such exten
1. 0 |- sive data were not collected due to time
and cost constraints. Nevertheless, one
z 0. 9 time distribution of water does influence
UJ
yield potential. For example, for System
CJ
0. 8
A, an area may be repetitively under-irri
gated or another area may be under-irri
UJ
o 0. 7
gated at different times. In any case, crops
u
are subjected to moisture stress, and yield
0. 6
may be reduced. Thus, repetitive good
z water distributions help to realize full
cr
a
potential yield.
0. 5
u_

z
0. 4 J_
A B O D E
SYSTEM
Figure 7. Radial design uniformities for six low-pressure systems.
(Thooyamani 1982). If the sprinkler spac-
ings were designed based on only average
application rate, and not spray pattern
shape, the resulting overlapped spray pat
tern may be quite nonuniform. The ulti
mate result is nonuniform application of
water with very high peak application
rates. To avoid surface runoff associated
_L _L _L
IDENTIFICATION
with the high peak application rate, farm
ers operated the systems at high rotational
speeds, resulting in shallow depths of
application. This could explain the reason
for the average depth of application to be
below 10 mm in six field tests and the
overall average depth of application to be
at 15 mm. Thus, the design of sprinkler
SUMMARY AND CONCLUSIONS
Distributions of water in radial direc
tions under six low-pressure center-pivot
sprinkler irrigation systems were mea
sured to evaluate the systems. Twenty
seven trials were conducted under the sys
tems as operated by the farmers. The
expected distributions of water under the
systems were estimated from sprinkler
specification charts. To evaluate the sys
tems for their suitability, the observed and
the design distributions of water were
compared.
It was found that the low-pressure sys
tems distributed water at or above the gen-

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 147

 u 1. 2 Colorado State University Expt. Station,
u
Fort Collins, Co.
SYSTEM-A SYSTEM-B C D E F
HILL, H. M. 1985. Irrigation the option
-

z
o for economic development. Proceedings
1. 0
CO
X Fourth Annual West Prov. Conf., Ration
UJ
a Q * * A mm alization of Water and Soil Research and
\
Management. Saskatchewan Water Corp.,
0. 8 Moose Jaw, Sask. pp. 233-245.
u
L>
D
+vtfft MURRAY, J. M., D. I. NORUM, and G. E.

f 'i
DYCK. 1982. Development and evaluation
a
UJ
of automated surface irrigation system.
0. 6
> Final Report on DSS Contract No. OSZ79-
cr
UJ 00061 to Canada Department of Agriculture.
CO
CD 61pp.
o 1 1 1 1 1 1 1 1 1 i i i i i i i i i i i i i i i i i i
0. 4 NORUM, D. I., J. M. MURRAY, and
1 4 7 10 13 16 19 22 25 28 S. DUBETZ. 1983. Evaluation of, and rec
ommendations for, operation of low pres
TRIAL NUMBERS sure center pivot sprinklers. Prairie Regional
HQ, Agriculture Canada, Ottawa, Ont.
Figure 8. Performance of six low-pressure systems. Publ. no. 1-4791. 136 pp.
PRAIRIE FARM REHABILITATION AD
erally acceptable level of uniformity co ciated with high peak application rates MINISTRATION. 1982. History of irriga
efficient (0.8) only 37% of the time. It was which, in turn, were due to the noncon- tion in Western Canada. PFRA, Regina,
also found that the six systems distributed sideration of spray pattern shapes during Sask.
water consistently below their design design. This is an additional factor causing THOOYAMANI, K. P. 1982. Efficiency of
(expected) uniformity level. poor uniformity. low-pressure sprinkler irrigation systems.
During the field tests, wind speeds were The potential detrimental factors caus M.Sc. Thesis, Department of Agricultural
ing the poor uniformity of the systems can Engineering, University of Saskatchewan,
in the range of 0-30 km/h. Installation of
Saskatoon, Sask.
flow regulators and excessive reduction of be controlled by design. Thus, there is
nozzle pressure (from line pressure) can room for improvement in the design of
make the systems more vulnerable to center-pivot sprinkler irrigation machines. LIST OF SYMBOLS
wind, resulting in nonuniform overlapped A = total area of irrigation (m2).
spray patterns and low uniformity of dx = depth of applicationat/th station, or at
application. Thus, wind and excessive REFERENCES base of /th sprinkler (mm).
reduction of nozzle pressure were identi HART, W. E. 1961. Overhead irrigation d = average depth of application (mm).
fied as potential factors causing the poor pattern parameters. Agric. Eng. 42(7): Q = discharge rate from /th sprinkler
354-355. (L/sec).
performance.
HART, W. E. 1972. Subsurface distribution /?j = radial location of /th station (m).
For the 27 field tests, overall average of nonuniformly applied surface waters. 5j = radial incrementrepresented by the/th
depth of application was 15 mm. The low Trans. ASAE (Am Soc. Agric. Eng.) 15(4): station (m).
average depths of application were the 656-661,666. / = duration of application per revolution
result of high rotational speed of the sys HART, W. E. and D. F. HEERMANN. 1976. (sec).
tems. The high rotational speed was due to Evaluating water distribution of sprinkler Ucc = Christiansen uniformity coefficient.
the attempt to avoid surface runoff asso irrigation systems. Technical Bulletin 128. UR = uniformity ratio.

148 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 APPLICATION RATES AND UNIFORMITY UNDER CENTER-PIVOT
SPRINKLER IRRIGATION SYSTEMS USING SPRAY NOZZLES

K. P. Thooyamani1, D. I. Norum1, and S. Dubetz2

'Departmentof Agricultural Engineering, University of Saskatchewan, Saskatoon, Sask. S7N 0W0, and
2Soil Science Section, Agriculture Canada, Research Station, Lethbridge, Alberta TIJ4BI
Received 19 November 1986, accepted 18 February 1987

Thooyamani, K. P., D. I. Norum, and S. Dubetz. 1987. Application rates and uniformity under center-pivot
sprinkler irrigation systems using spray nozzles. Can. Agric. Eng. 29: 149-154.

Uniformity of water application in circular and radial directions under various center-pivot irrigation systems using
spray nozzles was recorded inthefield onfarmer-operated systems. Forthe eight systems studied, theaverage application
was 15.2 mm and the Christiansen uniformity coefficient values were all above 80%. Acomputer model was developed to
simulate field water application distributions using single sprinkler distributions that had been determined in the
laboratory. These distributions were used to predict potential runoff rates under various operating conditions. Simulation
showed that it is possible to obtain a high uniformity coefficient with any spray pattern provided the nozzles are spaced
properly; however, the result may be an application rate that is great enough to produce runoff.

INTRODUCTION FIELD TESTS cans of 100 mm diameter were positioned

As the cost of energy increased, many
sprinkler irrigation farmers purchased
low-pressure (under 300 kPa) center-pivot
systems that use spray nozzles. The radius
of throw for these nozzles is only 25-35%
of that for conventional high-pressure
sprinklers, and the pumping rate for either
a low-pressure or high-pressure system,
covering the same area, must be approx
imately the same. Therefore, the instan
taneous application rate is much greater
for the low-pressure system than for the
high-pressure system. This high applica
tion rate can often lead to runoff and ineffi
cient irrigation. In addition, for electri
cally driven machines, the distance trav
eled between stops by any tower of the
center pivot system can be a significant
fraction of the radius of throw of the spray
nozzles. This may also contribute to non
uniform application of water.
The objectives of the work described
here were:
(a) To determine, in the field, water
application distributions and uniformities
along radial and circular lines from low-
pressure center-pivot systems using spray
nozzles.
(b) To simulate, by means of computer
modeling, field application distributions
by numerically overlapping single nozzle
distribution patterns obtained from labora
All field tests were carried out on on stands approximately 0.90 m above
farmer-operated center-pivot machines on ground level. When determining the dis
or in the vicinity of the Outlook Irrigation tributions along circular arcs, the cans
District in Saskatchewan. Eight elec were spaced 0.30 m apart, while the spac
trically driven machines were tested. No ing for most radial line tests was 5.0 m;
hydraulically driven machines were avail however, in some instances spacings of
able for testing. 10.0 and 15.0 m were used. Depending
Water application distributions were upon the length of arc that was required
measured while each machine was operat to obtain a representative distribution,
ing under normal field conditions, except 20-50 cans were used to obtain each cir
that the test periods were chosen so that the cular arc distribution.
wind was relatively light. One-litre catch Thirty-two field tests were made, eight
40r
e
E
o 30
UJ
-j
CL
Q.
<
20 -
UJ
<
O
X
r-

tory tests. Q_
UJ
(c) To use simulated distributions to cal Q
culateuniformityof applicationand poten
tial runoff. 0
IOO 200 300 400

Thispaper waspresented at the1984 Annual Meeting DISTANCE ALONG A RADIAL LINE, m

oftheCanadian Society of Agricultural Engineering,
19-23 August, Winnipeg, Manitoba. Paper no. Figure 1. Example of water application distribution taken along a radial line for a center-pivot
84-307. irrigation system.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 149

 30r SIMULATION OF FIELD
DISTRIBUTIONS
E Single-nozzle distribution tests were
E carried out in the laboratory to determine
the water distributions from four types of
Q
UJ spray nozzles operating at various pres
sures and heights above the ground sur
0^20 face. Type 1 nozzles were Nelson nozzles
with a flat spray plate; Type 2 were Rain-
<
bird with a flat spray plate; Type 3 were
rr Nelson with a convex spray plate; and
UJ
Type 4 were Nelson with a concave spray
plate. Details of these tests are given by
Thooyamani (1982). Field application dis
tributions were then simulated numeri
cally by overlapping patterns from the
laboratory tests.
x Chu and Moe (1972) have shown that
CL
for a well-designed center-pivot system
UJ with a large number of nozzles, the flow
Q rate, Q, and the pressure, P, at any point in
the lateral can be defined by
0 4 8 12 16
Q = <2oO -x2) (1)
DISTANCE ALONG A CIRCULAR LINE.m
Figure 2. Example of water application distribution taken along a circular line for a center-pivot I5r
l - M -X3 + -X5 (2)
irrigation system. Po - Pe 8 L 3 5

where Q0 = flow rate at the pivot, P0 =

25 pressure at the pivot, Pe = pressure at the
E end of the lateral, L = length of the lat
E eral, s = distance along the lateral from
the pivot, andX = s/L. Equations 1 and 2
20 are valid only for a horizontal lateral of
Q
UJ
uniform pipe diameter.
-J
To simulate the movement of the center
Ol pivot lateral, it was assumed that the posi
< 15 COMPUTED tion of the lateral line could be modeled by
moving the end tower for a certain per
rr centage of a minute and that the maximum
UJ "out-of-alignment" for any intermediate
< tower was 1.0 m.
10
A computer program was developed to
UL OBSERVED
calculate the time rate of application of
O water at any point in the field using the
nozzle sizes, lateral line pressure, and lat
X 5- eral line position. In addition to using the
h-
Ql single sprinkler distributions obtained in
UJ the laboratory, simulations were carried
Q
out using single sprinkler triangular and
JL elliptical distributions with the same areal
0 4 8 12 16 coverage and average application rate.
DISTANCE ALONG A CIRCULAR LINE.m
Distribution and Uniformity
Figure 3. Observed and computed application distributions along a circular line with a nozzle
Simulations were carried out for a
height of 3.7 m, a speed of rotation 40% of full speed and a distance from the pivot of
487 m.
number of systems using pivot rotational
speeds of 10-100% of maximum speed in
steps of 10%. Distributions were obtained
along radial lines and 24 along circular (UCC) for radial lines ranged from 82 to for radial lines and for circular arcs at 30,
lines. The average depths of application 88%, and for circular lines ranged from 81 60 and 90% of the lateral length from the
for the 32 tests ranged from 9.6 to 27.9 to 94%. Figures 1 and 2 show examples of pivot. Figure 3 is an example of the sim
mm, with an overall average of 15.2 mm. the distributions obtained from radial lines ulation for a circular arc. Figure 4 shows
The Christiansen uniformity coefficients and circular lines, respectively. the simulation for a radial line.

150 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 125 erally were about the same magnitude as
produced by the flatter elliptical pattern.
The shape of the application rate pattern at
a point is, for practical purposes, indepen
dent of the speed of rotation of the lateral.
It is only the time scale that changes with
speed. Figures 6 and 7 show average
application rates along the lateral when the
nozzles were assumed to be at two dif
ferent heights, 1.2 and 3.7 m. In these
cases, increasing the nozzle height from
1.2 to 3.7 m decreased the average appli
cation rates by approximately 40%.

Potential Runoff
Potential runoff is defined as the per
centage of total depth of water applied,
that is, applied when the application rate is
greater than the soil intake rate. For sim
ulation purposes, it was assumed that the
ponded water infiltration rate for the soil
could be described by the Kostiakov equa
0 100 200 300 400 500 600 tion

DISTANCE ALONG A RADIAL LINE , m I = Kt" (3)

Figure 4. Predicted application distribution with nozzles operating at 3.7 m height (28.8 h per where / = infiltration rate under ponded
revolution at 100% speed). conditions, (mm/h), t = time since water
was first applied (h), and K and n are
coefficients dependent on the soil type. As
for much of the time (preferably all the
TRIANGULAR time) the application rate from the sprin
V PATTERN kler system is less than the infiltration rate,
Eq. 3 must be modified to take that into
account. The method used was that of
n ELLIPTICAL
Cook (1946) and Kincaid et al. (1969) in
PATTERN which for nonponded conditions

ACTUAL (4)
/, - /
PATTERN D,

where /, = modified infiltration rate,

p = depth of water infiltrated under
ponded conditions, and Da = depth of
water actually infiltrated. Once ponding
reoccurs, the infiltration rate reverts to the
regular Kostiakov equation except that the
time is offset by a value Ar so that
/ = Kit - A0n (5)

satisfies the current infiltration rate, /m.

The rate equation that was assumed for
simulation purposes was
5 10 15 25 / = 14.3 f--46 (6)

TIME DURATION , min which corresponds to a medium textured

Figure 5. Predicted application rates using actual, triangular and elliptical spray patterns with a soil (sandy clay loam).
nozzle height of 3.7 m, a speed of rotation 30% of full speed and a distance from the Potential runoffs were calculated for a
pivot of 523 m. number of systems and operating speeds,
for various distances from the pivot. Table
I is an example for one system showing the
Application Rates In most cases the peak application rate for potential runoff as a percentage of the
Figure 5 shows an example of the actual patterns was less than that for the applied depth at various points away from
application rate at a fixed point as simu triangular pattern because the actual pat the pivot, and as a function of the speed of
lated using actual sprinkler patterns and tern generally had a doughnut shape. The rotation of the system. The bottom line of
assumed triangular and elliptical patterns. double peaks from the actual pattern gen Table I gives the percentage runoff for the

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 151

 I25r off. The main factors that affect this objec
tive are:
(a) system pipe size, nozzle sizes, spac
ing and height, and operating pressure;
(b) distribution pattern from individual
nozzles;
(c) speed of rotation of machine around
the field;
(d) wind;
(e) crop canopy;
(0 infiltration rate characteristics of the
soil;
(g) topography; and
(h) ground surface condition (ability to
pond water).
Factors (a), (b), and (c) are essentially
' 'hardware" factors which can be charac
terized in sufficient detail by testing, so
that the water application from the system
can be described accurately. In some
cases, topography may have an effect on
this distribution if there are sufficient dif
0 100 200 300 400 500 600 ferences in field elevations that the pres
sure distribution in the lateral line is
DISTANCE ALONG A RADIAL LINE , m affected. However, if the topography is
known, it is still possible to take that into
Figure 6. Predicted average application rates along radial lines for four sprinkler types with consideration when simulating the distri
nozzles at 1.2 m height (28.8 h per revolution at 100% speed). bution from the system.
Wind is the one factor that distorts the
distribution pattern as it comes from the
system, and before it reaches the crop can
opy or the ground surface. Although wind
has a major effect on the ground distribu
tion it is extremely difficult to characterize
the effect as the distortion depends not
only on the wind speed and direction, but
also on the droplet size, nozzle height and
trajectory of the droplets as they leave the
nozzle.
Distortion of the distribution pattern by
wind can have a very marked effect on the
uniformity coefficient for the distribution.
This is especially so under prairie condi
tions. Although field tests and simulations
may indicate high uniformity, it is possible
that on an entire field basis, say a 60-ha
circle, the uniformity will be considerably
less, because of the distribution pattern
being distorted and shifted as the wind
changes speed and direction during the
time the system is making one revolution
0 100 200 300 400 500 600 of the field. Under these conditions there

DISTANCE ALONG A RADIAL LINE ,m may be some advantage in using small

depths of application as the irrigations
Figure 7. Predicted average application rates along radial lines for four sprinkler types with must be frequent and there is a reasonable
nozzles at 3.7 m height (28.8 h per revolution at 100% speed). probability that each time the machine is
situated at a particular point in the field,
the wind speed and direction will be dif
entire system. Figures 8 and 9 show the DISCUSSION ferent. Therefore, the accumulated dis
potential runoff at points away from the As with most irrigation systems, the tribution for a number of irrigations will be
pivot for nozzles at 1.2 and 3.7 m height, prime objective of operating a low-pres more uniform than for a distribution result
respectively. It is clear that increasing the sure center pivot system is to apply an ing from one equivalent depth irrigation.
nozzle height reduces the potential runoff adequate depth of water uniformly over The irrigation efficiency may be relatively
considerably. the field with a minimum amount of run poor, but the overall uniformity may be

152 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 TABLE I. EXAMPLE OF SIMULATED POTENTIAL RUNOFF (%) ALONG A RADIAL LINE some point either:
FOR A SYSTEM OPERATING AT VARIOUS PERCENTAGES OF CONTINUOUS OPERATING (a) infiltrates directly into the soil at that
SPEED
point;
Distance
Speed ofrevolution (b) becomes ponded at that point and
(m) 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
infiltrates at some later time;
(c) flows to some other point in the field
37 22.8 5.9 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0
64 18.5 1.4 0.0 0.0 0.0
and infiltrates at some later time;
0.0 0.0 0.0 0.0 0.0
91 30.6 12.5 2.9 0.0 0.0 0.0 0.0 0.0 0.0 0.0
(d) flows off the field entirely and is lost
118 25.3 8.0 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 to drainage; or
145 18.5 2.9 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 (e) evaporates after becoming ponded
172 21.4 5.9 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 somewhere in the field.
199 34.7 17.1 6.5 1.0 0.0 0.0 0.0 0.0 0.0 0.0
226 43.2 26.6 16.2 7.9 3.5 1.3 0.0 0.0 0.0 0.0
In most situations (e) can be neglected as it
253 44.2 28.3 16.8 9.2 4.7 1.5 0.0 0.0 0.0 0.0 has little effect on the distribution. Con
280 41.6 23.9 12.6 5.6 1.1 0.0 0.0 0.0 0.0 0.0 dition (d) is generally not significant for
307 42.6 25.4 14.9 8.3 4.7 1.4 0.0 0.0 0.0 0.0 most sprinkler systems. If it is significant,
334 42.1 23.3 13.1 6.2 2.9 0.7 0.0 0.0 0.0 0.0
361 44.0 25.6 13.5 8.1 3.9 0.9 0.0 0.0
there is something radically wrong with
0.0 0.0
388 49.4 34.9 23.4 14.4 9.5 5.9 3.3 1.1 0.0 0.0
the system and major changes must be
415 56.1 43.8 35.7 28.2 20.5 14.2 10.0 7.5 4.8 3.0 made in its design and/or operation. Con
442 54.4 40.8 32.8 24.9 18.4 12.4 8.3 4.4 1.5 0.1 ditions (b) and (c) are first of all dependent
469 52.8 38.4 29.4 21.0 13.3 8.2 5.3 2.8
496 52.0
0.9 0.0 upon the application rate and the infiltra
35.8 24.7 15.4 11.0 7.5 4.5 2.2 0.9 0.0
523 53.0 38.3 26.6 17.4 11.8 8.3 4.8 2.1 0.4
tion rate characteristics, but then become
0.0
550 56.0 42.8 31.4 23.7 16.7 11.5 7.6 4.0 1.7 0.0 very much dependent upon the topography
and the surface condition.
System 46.0 30.5 20.6 13.7 9.0 5.6 3.4 1.9 0.8 0.2 Hart (1972) has shown that the unifor
mity of water infiltrating at the ground
60 surface may be improved considerably if
the redistribution within the soil profile is
TYPE-2 taken into consideration. He shows exam
A
/ ples of a uniformity coefficient of 60% for
50 the water distribution at the ground surface
A
becoming 76 and 86% after redistribution
in the soil for 1 and 2 days, respectively.
TYPE-4
The effect of redistribution within the soil
is, of course, much dependent upon the
. 40
spatial distance between above average
and below average applications.
O The results of the simulated tests show
z
O that good uniformity can be obtained by
30- proper spacing of the nozzles on the lat
eral, but the resulting application rate may
produce runoff. The application rate can
<
be reduced by increasing the height of the
z 20 nozzles but this may lead to greater wind
111 drift losses and reduced uniformity. From
I-
o the simulations it would appear that it is
Q. feasible to calculate potential runoff for a
10 system if the infiltration characteristics of
the soil are known. However, potential
runoff does not take into consideration the
topography and surface conditions, which
are entirely site specific. Potential runoff is
only a measure of the amount of water that
0 100 200 300 400 500 600
is applied that may either pond or redis
DISTANCE ALONG A RADIAL LINE , m tribute on the surface due to the applica
tion rate being greater than the intake rate.
Figure 8. Potential runoff along radial lines for four sprinkler types with nozzles at 1.2 m height
and 30% speed of rotation (28.8 h per revolution at 100% speed).
SUMMARY AND
RECOMMENDATIONS
improved by small applications. that it never reaches the ground surface. The water application distributions
Once the water reaches the crop canopy Although this interception is only of the along both radial and circular lines were
there may be some additional distortion of order of a few millimeters of water, it may determined for eight low-pressure center-
the distribution pattern; however, from a be quite significant when only small pivot sprinkler systems that were operat
practical standpoint, the main effect of the depths of water are being applied. ing under normal field conditions. The
canopy is to intercept some of the water so Water that reaches the ground surface at average application was 15.2 mm and all

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 153

 50 r
0 100 200
DISTANCE ALONG
Figure 9. Potential runoff along radial lines for four sprinklertypes with nozzlesat 3.7 m height
and 30% speed of rotation (28.8 h per revolution at 100% speed).
UCC values were above 80%.
A computer program was developed to
simulate the distribution from a center-
pivot system using single-nozzle distribu
tions that had been determined in the labo
ratory. By assuming a particular infiltra
tion function for the soil, it was possible to
use the program to determine potential
runoff for various operating conditions.
Simulation showed that it is possible to
obtain a high uniformity coefficient with
any spray pattern provided the nozzles are
spaced properly; however, the result may
be an application rate that is great enough
154
300 400 500 600
A RADIAL LINE f m
to produce runoff. The application rate can
be reduced by increasing the height of the
nozzles but this may result in increased
losses due to wind drift. Results from the
simulations showed that the application
rate was essentially independent of the
rotational speed of the system.
ACKNOWLEDGMENT
Financial aid by the Government of India to
the first author is hereby acknowledged. This
work was supported by funds provided by the
Department of Supply and Services, Govern
ment of Canada under contract OSU81-00005
to the University of Saskatchewan.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
REFERENCES
CHU, S. T. and D. L. MOE. 1972. Hydraulics
of a center pivot system. Trans. ASAE (Am.
Soc. Agric. Eng.) 15(5): 894-896.
COOK, H. L. 1946. The infiltration approach
to calculation of surface runoff. Trans. Am.
Geophys. Union 27: 726-747.
HART, W. E. 1972. Subsurface distribution of
nonuniformly applied surface waters. Trans.
ASAE (Am. Soc. Agric. Eng.) 15(4): 656-
661, 666.
KINCAID, D. C, D. F. HEERMAN, and
E. G. KRUSE. 1969. Application rates and
runoff in center pivot sprinkler irrigation.
ASAE Trans. (Am. Soc. Agric. Eng) 12(6):
790-794, 797.
THOOYAMANI, K. P. 1982. Efficiency of
low pressure sprinkler irrigation systems.
M.Sc. Thesis, Department of Agricultural
Engineering, University of Saskatchewan,
Saskatoon, Sask.
LIST OF SYMBOLS
= depth of water actually infiltrated
(mm).
Dn = depth of water infiltrated under
ponded conditions (mm).
= infiltration rate under ponded condi
tions (mm/h).
= modified infiltration rate (mm/h).
K = infiltration coefficient dependent on
the soil type (mm/hn).
L = length of the lateral (m).
n = infiltration coefficient dependent on
the soil type.
= pressure at point s (kPa).
= pressure at the pivot (kPa).
= pressure at the end of the lateral
(kPa).
Q = flow rate at point s (L/sec).
= flow rate at the pivot (L/sec).
S = distance along the lateral from the
pivot (m).
t = time since water was first applied (h).
X = s/L.
At = time offset to make / equal to /m (h).
 A PORTABLE RAINFALL SIMULATOR

R. W. Tossell1, W. T. Dickinson2, R. P. Rudra2, and G. J. Wall3

'Land Resource Science Department, and2School of Engineering, University of Guelph, Guelph, Ontario NIG 2WI;
and 3Agriculture Canada, Ontario Institute of Pedology, Guelph, OntarioNIG 2WI
Received 2 September, 1986, accepted 27 January 1987

Tossell, R. W., W. T. Dickinson, R. P. Rudra, andG. J. Wall. 1987. A portable rainfall simulator. Can. Agric. Eng.
29: 155-162.

Anew portable rainfall simulator has been developed for both field and laboratory soil erosion research attheUniversity
of Guelph. Employing several low- to medium-flow-rate, full-jet nozzles in a continuous-spray design, the simulator is
capable of reproducing storm intensities ranging from 17.5 mm/h to inexcess of 200mm/h. Simulated rainfall intensity
varies with nozzle size, water pressure atthenozzle, and height ofthe nozzle above theplot surface. Uniformity of rainfall
intensity overa 1-m by 1-m plot averages 88.33% overallnozzles andsettings, with many nozzles producing averages over
90%. The simulator and support equipment are portable to the extent that they can be loaded onto a half-ton truck for
transport to the field.

INTRODUCTION
Rainfall simulation has become a very
effective technique for assessing soil ero
sion, particle detachment, overland flow,
and chemical runoff (Meyer and Harmon
1979; Bubenzer and Jones 1971; Turner
1965; Beasley et al. 1985). The benefits of
using a rainfall simulation approach to soil
erosion research are well documented by
Meyer (1965), Neff (1979), and Molden-
hauer (1979), with versatility being the
foremost advantage. Rainfall simulation
provides control of precipitation charac
teristics such as intensity, and spatial and
temporal frequency and duration in both
field and laboratory studies (Pall et al.
1983).
Several types and designs of rainfall
simulators have been proposed to meet the
objectives of researchers. These objectives
in turn have often dictated the type or
design of simulator appropriate to the
research studies. For example drop-former
simulators (Ellison and Pomerene 1944;
Adams et al. 1957; Chow and Harbaugh
1965; Langford 1970; Black 1972) produce
a relatively uniform, narrow, drop-size
distribution. These simulators are there
fore suitable for studies of rainsplash or
particle detachment, where known homo
geneous drop-size characteristics are
desired. Recent design improvements in
sprinkler nozzles have greatly improved
the drop characteristics produced by
nozzle-type simulators. A wide range of
randomly-distributed drops are produced
by such simulators, rendering the simu
lated rainfall more comparable to natural
rainfall. Devices employing nozzles
include: F-type simulators producing con
tinuous spray upward (Hart 1984; Meeu-
wig 1969), rotating-boom and oscillating
simulators (Luk and Morgan 1981; Meyer
1979), rotadisk simulators (Morin et al.
1967; Pall et al. 1983), and continuous
downward flow spray simulators (Lyles
etal. 1969).
A number of rainfall and operational
characteristics, identified to be important
for rainfall simulation for soil detachment
and transport studies, have been docu
mented by Bubenzer (1979), Meyer (1965,
1979), and Mutchler and McGregor
(1979). These include:
1. Drop-size distribution similar to natu
ral rainfall given comparable rainfall
intensities.
2. Drop impact velocity approximating
terminal velocity of natural raindrops.
3. Rainfall intensity representing the
geographical region where studies are to
be conducted.
4. Uniform rainfall over the study area.
5. Energy characteristics correspond
ing to natural rainfall for comparable
intensities.
6. Rainfall intensity continuous over the
storm event.
7. Storm pattern reproduction.
8. Sufficient area of coverage.
9. Drop impact angle near vertical.
10. Site to site portability.
ftill et al. (1983) have added that a sim
ulator should be efficient and simple to
operate to facilitate reliability.
The Guelph Rainfall Simulator I (GRS I)
was developed on the basis ofthe above list
of criteria for both laboratory and field
erosion research (Pall et al. 1983). It em
ploys a large-capacity wide-angle spray,
rotadisk system. The GRS I has proven to
be very useful for laboratory studies; how
ever, the system has not been sufficiently
portable to be readily used at various field
locations.
A second rainfall simulator, desig
nated as the Guelph Rainfall Simulator II
(GRS II), has been developed. This sim
ulator incorporates many of the good
qualities of the original simulator, with a
number of new features to provide an
improvement over the rotadisk system.
The GRS 11 is completely portable and can
be assembled at field locations with a mini
mum of site disturbance. The continuous
spray generated by the small nozzle design
is also more physically realistic than the
intermittent spray produced by the rotadisk
and some other simulator systems.
As noted briefly above, each rainfall
simulator type meets only a subset of the
desired criteria. Drop formers have a nar
row range of drop sizes, and a nonrandom,
biased, drop-size distribution. F-type
designs are usually nonportable and are
susceptible to wind-sorting effects. Porta
bility is also a problem with many rotating
and oscillating simulators. A noncon-
tinuous spray, combined with a narrow
range of drop sizes at all intensities are the
major drawbacks of the rotating disk
design. On the other hand, the continuous-
spray, single-nozzle simulators have not
been very useful for reproducing low
intensity storms, and do not readily allow
the changing of intensity over the duration
of a simulated storm. However, a wide
range of nozzles are now available to re
produce a variety of storm intensities.
The purpose of this paper is to provide a
summary of the design and development of
the GRS II rainfall simulator, including
preliminary calibration results regarding
simulated intensities and spatial unifor
mity. A thorough analysis of drop-size,
drop velocity, and energy and momentum
characteristics will be presented in a sub
sequent paper.
DEVELOPMENT AND
DESCRIPTION OF THE GUELPH
RAINFALL SIMULATOR II
In light of the merits and shortcomings
of the various simulator types, a new con
tinuous-spray simulator design has been

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 155

 1 Water barrel
2 Pump system
3 Plot boundary
4 Nozzle
5 On-off valve
6 Pressure gauge
7 Pipe
8 Ingoing hose
9 Return line hose
10 Support poles
11 Clamps
12 Support cables
13 Clips

Figure 1. Rainfall simulator assembly.

selected and developed for both field and TABLE I. SPECIFICATIONS FOR NOZZLES EMPLOYED IN GRS II DESIGN
laboratory soil erosion studies at the Uni
Flowrate capacity at
versity of Guelph. Special attention has
specific pressure (kPa) Spray angle
<

been given to the portability ofthe system, (L/min) (degrees)

Nozzle Orifice diam.
simple operational features, and cost. A
number (mm) (inches) 34.5 48.3 69.0 103.4 137.9 275.8 34.5 69.0 275.8
schematic ofthe GRS II system is shown in
1/8GG 2.8W 1.588 1/16 NAt NA 1.06 1.25 1.44
Fig. 1, and a photograph of the field set-up 2.32 NA 120 120
1/8GG 4.3W 1.984 5/64 NA NA 1.63 1.93 2.20 3.59 NA 120 120
is given in Fig. 2. 1/4GG 10W 2.778 7/64 2.80 3.26 3.79 4.54 5.30 8.18 112 120 120
The GRS II has a downward-oriented, 1/4GG 14W 3.572 9/64 3.79 4.54 5.30 6.44 7.19 11.82 114 120 120
single-nozzle, continuous-spray design, 3/8GG 20W 4.366 11/64 5.68 6.44 7.57 9.09 10.22 16.82 114 120 121
employing a selection of low- to medium- 1/2HH 30W 5.556 7/32 8.33 9.84 11.36 13.63 15.52 25.00 114 120 121

flow-rate, fulljet nozzles (Spraying Sys

tems Co., Wheaton, 111.). Six nozzles with
a full-cone spray angle of 120 (at 70 kPa),
including the 1/8GG 2.8W, 1/8GG 4.3W,
1/4GG 10W, 1/4GG 14W, 3/8GG 20W and
the 1/2HH 30W nozzles (Spraying Sys
tems Industrial Catalog 27), have been
selected to provide a range of rainfall
intensities. Table I illustrates the nozzle
specifications. Water is supplied through a
1.25-cm (1/2-inch) galvanized pipe, by a
1475 L/h rotary gear pump.
A high-pressure gauge (1400 kPa) mea
sures water pressure originating at the
pump, while a second gauge (200 kPa)
monitors the pressure at the nozzle. The
pump is driven by a 0.75 kW, 220-V, 1725-
rpm motor, which, in turn, is driven by a
3000-W gasoline-powered generator.
tNot available.
Note: Nozzle specifications are supplied by the manufacturer.
Pressure in the system can be adjusted
by a pressure regulator, and by applying
back pressure on the outflow end of the
simulator system by means of a Janesbury
clincher "shut-off" valve (Type: 1.9 cm
21-1136-TT-O supplied by Janesbury Cor
poration, Worcester, Mass. Another Janes
bury clincher valve has been used adjacent
to the nozzle to facilitate accurate control
of spray timing and to eliminate nozzle
drip problems. For more convenient con
trol of spray timing, a solenoid valve can
be employed in place ofthe manual control
valve. Rubberized garden hose distributes
water from the pump system to the sim
ulator and back to a 210 L water storage
barrel.
The unique, light-weight support sys
tem consists of two 2.54-cm- (1-inch)
diameter galvanized pipes, 2.25 m in
length (divided into two sections for porta
bility), supported by vinyl-coated, gal-
vanized-steel guide lines, producing a
tent-like structure. Steel stakes are used to
anchor the guide lines. Dual, adjustable
clamps are employed to facilitate adjust
ment of the simulator height (Fig. 1).
Under field conditions, simulation is
conducted over a 1-m by 1-m plot. Three,
1-m by 0.1-m, 16-gauge, galvanized,

156 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987


Figure 2. Photograph of the Guelph Rainfall Simulator II in the field.
sheet-metal strips are used to delineate the
plot boundary and to provide adequate pro
tection against any rainsplash entering the
study area from outside the plot. At the
base of the plot, a galvanized steel trough
(with cover) channels runoff water and
sediment through a sieve system into a
collection apparatus (Fig. 3).
The simulator, water supply system,
generator, and all related facilities are suf
ficiently portable to be loaded onto a half-
ton truck and transported from the labora
tory to field sites.
AREAL UNIFORMITY AND
INTENSITY CALIBRATION
Performance tests regarding rainfall
intensity and areal rainfall uniformity over
the plot area were conducted as described
by Pall et al. (1983). The procedure
involved placing collection apparatus or
gauges throughout the target area of
application, and taking a representative
sample of simulated rainfall. In the case of
the GRS II, the 1-m by 1-m plot was the
target area. The collection gauges were
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
1 jM '%k
placed throughoutthe plotarea as shownin
Fig. 4. The volume of simulated rainfall
collected in each gauge was recorded
according to its position within the plot
area (recorded by gauge number). Average
rainfall intensity over the plot was deter
mined by the following equation:
/ =10 [(2 VjAf)/n) x60/rj (I)
where /p is the plot-average intensity
(mm/h); Vt is the volume in the ith gauge
(cm3); Ag is the gauge collection area
(cm2); / is the time of each run in (min);
and n is the number of gauges; the coeffi
cient 10 converts measurements from cm/h
to mm/h.
Collection containers of the same size
were used (i.e., 108-mm-diameterby 178-
mm-high cylindrical cans); however, the
number of cans was reduced to 17 from 49,
and they were placed in strategic positions
within the calibration area, as shown in
Fig. 4. The reduced number of containers
helped to decrease the time needed for
calibration by a factor of 6 for each run,
thus greatly speeding up the calibration
process. The only drawback to decreasing
the number of containers is the accuracy of
the estimate for overall plot average rain
fall intensity. Covering the entire plot area
with gauges would yield the most accurate
estimate of rainfall intensity; however,
time efficiency necessitated reducing the
number to a manageable figure. The
reduction in the number of gauges from 49
to 17 represented a difference in measure
ments of 1and 3% for intensity and unifor
mity, respectively.
Uniformity of rainfall application was
determined using a procedure developed
by Christiansen (1942). The Christiansen
method is somewhat of a unwritten stan
dard procedure for determining the overall
uniformity of rainfall application over the
study area (plot area). As is the case with
the intensity measurements, the number
and also the size of the gauges will ulti
mately affect the results of the uniformity
trials. An in-depth look at uniformity of
application would involve several small
gauges covering the entire plot area. This
can be extremely time consuming, and the
amount of information obtained is offset
by the time involved. Conversely, a few
large gauges covering the entire plot area
may lead to misleading results. Therefore,
it is more informative to use small gauges
equally spaced throughout the plot area,
than it is to use the same number of large
gauges covering the whole plot. Unifor
mity of rainfall application, often referred
to as uniformity coefficient, was deter
mined by the following Christiansen
(1942) equation:
UC = 100 (1.0 - X \x,\/mn) (2l
where UC is the uniformity coefficient as a
percentage; jc, is the deviation from the
mean value m; and n is the number of
observations or gauges employed. Two
replications were conducted for each inten
sity/uniformity trial, and a third replicate
was added when observations from the first
two replicates were dissimilar. Simulated
rainfall intensity and uniformity measure
ments were determined for simulation runs
of 20-min duration.
RESULTS AND DISCUSSION
Rainfall intensity and uniformity were
observed to be dependent upon nozzle
size, water pressure at the nozzle, and the
height of the nozzle above the plot surface.
Each nozzle was tested using operating
pressures of 48.3, 69.0 and 96.5 kPa;
while the distance from the nozzle to the
plot was varied from 0.8 m to 1.7 m in
increments of 0.1 m or 0.2 m. The 1/8
157
 RAINFALL SIMULATOR PLOT SYSTEM
TOP VIEW
ft""
2
U 1\
SIDE VIEW
Figure 3. Field plot system for rainfall simulator.
2.8-W nozzle was not tested at the 48.3
kPa pressure because the specified mini
mum operating pressure was 69.0 kPa. The
1/2 30-W nozzle was not tested at 96.5 kPa
because rainfall intensities were consid
ered to be unrealistically high for Ontario
conditions. Not all heights were tested for
every nozzle type because of several over
lapping rainfall intensities. In such cases a
coarser increment of nozzle height (i.e.
0.2 m) was used. Results ofthe intensity/
uniformity trials are shown in Table II.
With the exception of the 1/8 2.8-W
nozzle, the rainfall intensity increased
with increasing nozzle size. The smaller
flow-rate nozzles (i.e. the 1/8-inch
nozzles) showed the smallest range in
intensity as a function of nozzle height.
158
3-
The larger nozzles (i.e. the 3/8-, 1/2-inch
nozzles) provided the largest range of
intensity with changes in height. This
effect was more evident at the lower
heights with the rate of change in intensity
decreasing with increasing height. Inten
sity tended to level off with nozzle heights
above 1.7 m for the 1/8 2.8-W and 1/8 4.3-
W nozzles, while the intensities generated
by the larger nozzles leveled out at greater
heights. Results were not presented for
nozzle settings above 1.7 m in height
because of the leveling out of intensities
with some nozzles. Furthermore, opera
tion of the simulator above the 2.0-m
height becomes functionally more diffi
cult, and the spray distribution becomes
more susceptible to wind-sorting effects.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Plot boundary
Plot surface
Surface-trough
interface or 'lip'
4 Collection trough
5 Collection trough cover
6 Collection pit
7 Collection system
8 Sieves
9 Sieve support
ring
Collection bucket
With the exception of a few anomalies
(i.e. the 1/8 2.8-W nozzle and 1/8 4.3-W
nozzle at 48.3 kPa), an increase in water
pressure at the nozzle increased rainfall
intensity. The smaller 1/8-inch nozzles,
however, showed a small change in inten
sity with pressure. In fact, the 1/8 2.8-W
nozzle at 96.5 kft* and the 1/8 4.3-W
nozzle at 69.0 kPa produced lower inten
sities than the corresponding lower pres
sures. Originally it was hoped that the 1/8
2.8-W nozzle would produce intensities
down to 10.0 mm/h, but the lowest inten
sity attained was 20.7 mm/h. The 1/8
4.3-W nozzle produced the lowest inten
sity of 17.5 mm/h.
The various combinations of nozzle
size, nozzle height, and water pressure
 _ _inint--NONOOoo 0 CN O
QQyoqoNO-oq
ZZoNONodod\dr-^r-^ ZZZdZriZriZZ
00ON000000000000
o .* m
"~ o
<N ON
ON
5^
sO qqqJq^q'
si z z zszsz; : z z
eg
ON
Q m \o N oo r^ a vO '
Z d ri < ZZhddcioiri-ri
On On On < *. r-oooooooooooooo
O M
<n/")~-cn m m
or-ONTtONoomONO
QTfiOON'OvOMOOiO^ Q Q t ^ "! u-nnOooonco
51 5 6 Z>7TtON ^tONTtodd
*h ONOO^ininrorf
Cs,^-
C/3
X
o mor-cnoN ininr*-)
QtJ-tJ-- -vqoo<N^q
w ZinoddfNcioNO-- - onoo ZZZZZ
X 0000ONONON00ONONON On 00 ON On On
H O *L -*
"" CO
5od On CN tj- (N vO fN m - XJ m 00 On
Q r-^CNOlOO ^00- (N(NQQQQQ
s 5 6 Zrisd-r^ioinfodd
-I ^r^o^ON'rrZZZZZ
^ on oo r^ r^
E
mOOO^moO
S QQ-NmO^ONnQ m\Or- O Q Q Q Q Q NO
H ZZon'ddoNvdodZ on'ooooo Z Z Z Z Z ON
oo ON ON ON oo oo oo 00 OO 00 ON ON
><
O -*
H
"^t m ^
00 so
QQr-00\0-"t-fNQ
ON NO (N
ON
rr-oiiri-ZZZZZ
O
fa 2I ZZ^N^-dddZ
N (N M N M (N (N on oo r^ \0 no
z
Ot^CNCNOmNO^
QQromoo^rn^tooQ oovOOvOCQQQQQ
< ZZ^dri- (NOn'oZ Ttr-ododoN'ZZZZZ
00 On On On On OO On oo oo oo oo oo
I ^ o
-*
OOONSOONON^^^__
QQOio-(NTrin^Q 8 \OOr~-r-TtQQQQQ
Q si ono^ ZZZZZ
Z 2I Z Z o <n ^ on oc oo Z
M(NfN(Nl-- - oo vo no in v>
H m m o r- r- ommm>nn
QQQQQONO^toc) QQ(^XONffl'tvO,t(N|
SS ZZd-doNOodoNd
z ZZZZZoddojd-
OO ON 00 ON ON ONONON00ON0000ON
w
H 12s
Z
r^rtsooN "^moN
oo od r no m o no
QQoOOMnt^Ov^vO-
QQQQQOn^ ^ ~
ZZdoNVr^dodd,t
2I ZZZZZofjo^hh
CN CN " 5I r^NONONONOnnn
z
CN m o O O c n in A _
QQinONONTtmOQQ
u ^ <~2 *: ZZri-dolddZZ
D b ZZZ-Zc^Z-Z^ On On On 00 ON ON
on oo oo oo

~"
0O sO
QQQ^-QOnQcnQno 3 QQootnr-oocNnQQ
5 g ZZZ<nZo<Zo^Zoo 51 ZZfiooNfodt^ZZ
vo n >n in >n Tt
in incar-ocNO^ _
QQQMQOQinQ^ QQinmO^'<NQQ
ZZZraZ-ZoZo NO ZZdrJr^-ON'-ZZ
ON ON 00 00 ON ON ON ON OO ON
*.
oo 2 M
T t *
(N o
&c= in
inONON00J__
00 ON QQm^ooNooqQQ
nO
5^? ZZZcjZroZoZoj 2I ZZm'oj^rn-jdZZ
in Tf ^t Tt Tl" r t
cs cn cn <
c its ^
"SCO
OQOOOOOQP
oo 2 ooONO-^rgmTj-inNor^
'5 3 'S 3 ai 3 Z
4 - -H- &00
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 159
 provide a broad range of simulated rainfall
intensities. Figure 5 summarizes the
range: the 1/8 4.3-W, 1/4 10-W, 1/4 14-W
and 3/8 20-W nozzles provide a range of
intensities from 17.0 mm/h to 110.0 mm/h;
and the 1/2 30-W nozzle produces higher
rainfall intensities, ranging from approx
imately 140.0 mm/h to 215.0 mm/h. A gap
exists between 110.0 mm/h and 140.0
mm/h, where the production of rainfall
intensities for the 3/8 20-W and 1/2 30-W
nozzles did not overlap. However, this gap
is not a major concern with this simulator
design, because new nozzles with inter
mediate flow rates are available (e.g.
3/8GG 24-W and 3/8GG 27-W nozzles)
1 m and can be calibrated quickly. The wide
range of rainfall intensities possible with
the GRS II provides a great deal of flex
ibility in the development stages of erosion
research employing rainfall simulation.
Uniformity of simulated rainfall appli
cation revealed no distinct relationship
with either nozzle water pressure or nozzle
size, as shown in Fig. 6. This result is in
contrast to the observations found by Pall
et al. (1983) which revealed that pressure
was the primary factor in determining the
uniformity coefficient for the GRS I. On
an individual nozzle basis, only the 3/8
Diameter of rainfall gauge 0.108 m 20-W nozzle indicated any increased uni
Figure 4. Distribution of rainfall gauges for intensity and uniformity calibration. formity with an increase in pressure. The
240
Nozzle
Nozzle Water
220 H
Size Pressure

200 X " 1/2 30W 48.3 kPa

A - 3/8 20W 96.5 kPa
0 - 1/4 14W 96.5 kPa
180
+ - 1/4 10W 48.3 kPa
n - 1/8 43W 96.5 kPa
\ 160
E
E
140 X

120

100
o
**-
c 80
o
a:
60

40

20 4

0 T T~ I I I 1 1 1
0.80 0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70
Height (m)
Figure 5. Range of rainfall intensities produced by the GRS II.

160 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 91 -

90 ~

89 -

c 88 -
0

87 -
0
o

86 -

85 - Nozzle \\
c Water \\
Pressure \\
84 -

D - 48.3 kPa \\
83 - -F - 69.0 kPa Ml
0 - 96.5 kPa ^
Ave V
82 -

81 - I 1 I I I i '

1/8 2,8W 1/8 4,3W 1/4 10W 1/4 14W 3/8 20W 1/2 30W
Nozzle Size

Figure 6. Average uniformity coefficients for all nozzles used with the GRS II.

larger nozzles (i.e. the 1/2 30-W and 3/8 research needs. Toprovidea wide range of distribution which is too narrow to be con
20-W)exhibited a slight increase in unifor rainfall intensities, three nozzles will sidered as a representation of natural rain
mity with nozzle height, whereas the re cover many research requirements. The fall for the same given intensity based on
maining nozzles showed no distinct rela 1/8 4.3-W nozzle will provide relatively preliminary investigations into drop size
tionship between rainfall uniformity and low rainfall intensities, while giving rea measurements. Even with an increase in
height. sonably consistent uniformity coeffi dropvelocities associated with an increase
The 1/4 14-W nozzle had the highest cients. Intermediate intensities can be in nozzle water pressure, the increased
average uniformity coefficient of 90.78%, covered with nozzles such as the 1/4 14-W velocity effect is offset by a reduction in
and the 1/2 30-W nozzle produced the or 3/8 20-W. Higher rainfall rates, such as theproportion of larger drops which make
lowest at 81.49% (averaged over all thunderstorms, can be produced with the up a large percentage of the rainfalls ero
heights andpressures). Otheraverage uni 1/2 30-W nozzle. If there is an option, sive characteristics.
formity coefficients were 89.93, 89.47, choosing between one of two nozzles, the
CONCLUSIONS
89.34 and 85.47% for the 1/8 4.3-W, 1/4 larger nozzle is recommended because of
10-W, 3/8 20-W, and the 1/8 2.8-W the widerdrop sizedistributions produced. The Guelph Rainfall Simulator II em
nozzles, respectively. The most inconsis The simulators nozzle height above the ploys a continuous, downward-flow wide-
tent nozzle proved to be the 1/8 2.8-W at study area, should be kept within func angle full-jet nozzle spray system which is
69.0 kPa, producing uniformities which tional limits, such as 1.0-2.0 m. It is suitable for both field and laboratory
ranged from 80.45 to 92.80% (Table II). suggested, however, that if drop velocities research. A selection of low- to medium-
All other nozzles had uniformity coeffi are of importance, the greater heights are flow-rate nozzles provides a goodrange of
cientranges of less than 7%. Average uni recommended. The increase in fall height, rainfall intensities and uniformity charac
formity for all nozzles and settings was will increase individual drop velocities. teristics. The simulator is completely por
88.33%. Operational nozzle water pressure table for field experimentation, and can be
In terms of uniformity of rainfall should also be kept within certain limits. transported in a half-ton truck. Thecontin
application, the GRS II outperformed its Pressures ranging from 45 to 100 kP& are uous-spray, random-drop distribution pat
predecessor, the GRS I, by approximately recommended. The smaller nozzles (1/8 tern is superior to the intermittent spray
4% on average. The uniformities of ap 2.8-W and 1/8 4.3-W) will not provide a pattern providing a closer representation of
proximately 90%are among the highest of full 120 spray angle and may yield low the natural rainfall phenomenon.
any presented in the literature. uniformity coefficients if nozzle pressures The simulator design provides a flex
The GRS II rainfall simulator has the are below 45 kPa. Conversely, pressures in ibility that renders it suitable for a wide
potential to accommodate a wide range of excess of 100 kPa will produce a drop size variety of research applications. To sup-

161
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
plement the great range of nozzle sizes,
combinations of nozzle heights above the
study surface and nozzle water pressures
will provide other desirable rainfall
characteristics. However, the combination
of such variables will ultimately depend on
the objectives of individual research
projects.
ACKNOWLEDGMENTS
The authors would like to thank D. Beals for
his constructive suggestions with the initial
simulator design. Special appreciation is
extended to B. K. Hohner, I. J. Shelton,
E. E. Vaughan and B. G. Johnson for their
helpful suggestions regarding the development
of the simulator and the calibration procedure.
Development funds were provided by the
Ontario Ministry of Agriculture and Food, the
Natural Sciences and Engineering Research
Council and Agriculture Canada (Great Lakes
Working Group).
REFERENCES
ADAMS, J. E., D. KIRKHAM, and P. P.
NEILSEN. 1957. A portable infiltrometer
and physical assessment of soil in place. Soil
Sci. Soc. Am. Proc. 21: 473-477.
BEASLEY, D. B., E. J. MONKE, E. R. MIL
LER, and L. F. HUGGINS. 1985. Using
simulation to assess the impacts of con
servation tillage on movement of sediment
and phosphorus into Lake Erie. J. Soil Water
Conserv. 40: 233-237.
BLACK, P. E. 1972. Hydrograph responses to
geomorphic model watershed characteristics
and precipitation variables. J. Hydrol. 17:
162
309-329.
BUBENZER, G. D. 1979. Inventory of rainfall
simulators. Rainfall Simulator Workshop
Tucson, Arizona., Agric. Review Manual
ARM-W-10. pp. 120-130.
BUBENZER, G. D. and B. A. JONES, Jr.
1971. Drop size and impact velocity effects
on the detachment of soils under rainfall sim
ulation. Trans. ASAE (Am. Soc. Agric.
Eng.) 14: 625-628.
CHOW, V. R. and T. E. HARBAUGH. 1965.
Raindrop production for laboratory water
shed experimentation. J. Geophys. Res. 70:
6111-6119.
CHRISTIANSEN, J. E. 1942. Irrigation by
sprinkling. University of California Agric.
Exp. Sta. Bull. 670.
ELLISON, W. D. and W. H. POMERENE
1944. A rainfall applicator. Agric. Eng. 25:
220.
HART, G. E. 1984. Erosion from simulated
rainfall on mountain rangeland in Utah. J.
Soil Water Conserv. 39: 330-334.
LANGFORD, K. J. 1970. A review of designs
for simulators. University of Melbourne.
Agric. Eng. Rep. 17.
LUK, S. andC. MORGAN. 1981. Spatial vari
ations of rainwash and runoff within appar
ently homogeneous areas. Catena
383-402.
LYLES, L., A. DISRUD, and W. P. WOOD
RUFF. 1969. Effectsof soil physical proper
ties, rainfall characteristics and
velocity on clod disintegration by rainfall
simulation. Soil Sci. Am. Proc.
195-205.
MEEUWIG, R. O. 1969. Infiltration and soil
erosion as influenced by vegetation and soil
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
in northern Utah. J. Range Manage. 23:
185-188.
MEYER, L. D. 1965. Simulation of rainfall for
soil erosion research. Trans. ASAE (Am.
Soc. Agric. Eng.) 8: 63-65.
MEYER, L. D. and W. C. HARMON. 1979.
Multiple intensity rainfall simulation for ero
sion research on row sideslopes. Trans.
ASAE (Am. Soc. Agric. Eng.) 22:100-103.
MEYER, L. D. 1979. Methods of attaining
desired characteristics in rainfall simulation.
Rainfall Simulator Workshop Tucson, Ari
zona. Agric. Review Manual ARM-W-10.
pp. 35-44.
MOLDENHAUER, W. C. 1979. Rainfall simu
lation as a research tool. Rainfall Simulator
Workshop Tucson, Arizona. Agric. Review
Manual ARM-W-10. pp. 90-95.
MORIN, J., D. GOl DBERG, and I. SEGINER.
1967. A rainfall simulator with a rotating
disk. Trans. ASAE (Am. Soc. Agric. Eng.)
10: 74-79.
MUTCHLER, C. K. and K. C. McGREGOR.
1979. Geographical differences in rainfall.
Rainfall Simulator Workshop Tucson, Ari
zona Agric. ReviewManual ARM-W-10. pp
8-16.
NEFF, E. L. 1979. Why rainfall simulation?
8: Rainfall Simulator Workshop Tucson, Ari
zona.,- Agric. Review Manual ARM-W-10.
pp. 3-7.
PALL, R., W. T. DICKINSON, D. BEALS,
wind and R. McGIRR. 1983. Development and
calibration of a rainfall simulator. Can.
33: Agric. Eng. 25: 181-187.
TURNER, A. K. 1965. The simulation of rain
fallsfor studiesin overland flow. J. Inst. Eng.
Austr. 37:9-15.
 EVALUATION OF BEST MANAGEMENT PRACTICES TO CONTROL
PHOSPHORUS NONPOINT SOURCE POLLUTION

A. Rousseau1, W. T. Dickinson2, and R. P. Rudra2

'Department ofAgricultural Engineering,Cornell University, Ithaca, New York 14853-5701; and 2School of
Engineering, University of Guelph, Guelph, Ontario NIG 2W1
Received 12 August 1986, accepted 4 February 1987

Rousseau, A., W. T. Dickinson, and R. P. Rudra. 1987. Evaluation of best management practices to control
phosphorus nonpoint sourcepollution. Can. Agric. Eng. 29: 163-168.

The Guelph model for evaluating the effects of Agricultural Management Systems on Erosion, Sedimentation and
Phosphorus yields (GAMESP) has been used toevaluate potential reductions in downstream sediment and phosphorus
loadings from agricultural land when best management practices are applied on problem areas of a small Ontario
agricultural watershed. This paper introduces a brief overview of different means of controlling phosphorus nonpoint
source pollution, and reveals that best management practices using a crop residue cover such as no-till and spring
moldboard plowing can potentially reduce downstream phosphorus loadings, attributed tosheet and rill erosion during the
hydrologically active spring period byas much as 83 and 74%, respectively, when compared to fall moldboard plowing
without a crop residue cover.

INTRODUCTION
In 1970, the International Joint Com
mission reported to the Governments of
Canada and the United States that 72% of
the phosphorus inputs to Lakes Erie and
Ontario was from municipal sources, 7%
from industrial sources, and 21% from
agricultural sources. In the past 15 yr,
since phosphorus inputs from municipal
and industrial sources have been reduced
by advanced waste water treatments,
phosphorus from agricultural sources
(nonpoint sources) have received a great
deal of attention. Phosphorus is recog
nized to be the nutrient limiting the pro
duction of plant biomass in surface waters
(Lee 1973; Lee et al. 1978); hence, non-
point sources of phosphorus have been
identified to be the main cause of surface
water pollution (Myers 1986). Inputs of
phosphorus from agricultural sources have
been observed to be mainly associated
with sediments (Miller et al. 1982; Nelson
and Logan 1983) carried by cropland
runoff during the spring period (Coote et
al. 1982), consequently, it is expected that
land management practices prescribed to
reduce sediment loadings to surface waters
would reduce phosphorus nonpoint source
pollution (U.S. Army Corps of Engineers
1982). However, since sediment loadings
are spatially and temporarily variable, it
becomes a major task to identify which
erodible land in an agricultural watershed
requires alternative land management
practices from the perspective of down
stream water quality impacts (Ogg and
Pionke 1986).
One cost-effective approach to identify
seasonal source areas of sediment-associ
ated phosphorus and to investigate the
impact of alternative land management
practices on phosphorus nonpoint source
pollution is to select and use a deter
ministic watershed model. GAMESP, the
Guelph model for evaluating the effects of
agricultural management systems on ero
sion, sedimentation and phosphorus yields
was developed by Rousseau et al. (1985)
and has been used for the delineation of
seasonal source areas of sediment and
phosphorus attributed to sheet and rill ero
sion in small agricultural watersheds. The
model is easy to use and makes use of
limited amounts of readily available data
bases and of easily quantifiable physical
and hydrological watershed charac
teristics. The model is based on the dis
cretization of a watershed into land and
stream cells of various shapes. Each land
cell is a field-sized cell with homogeneous
characteristics of land use, soil type and
class of slope. This modelling approach
offers an excellent means to define and
evaluate the implementation of different
environmental management strategies on
specific sites of an agricultural watershed.
The GAMESP model consists of two
major components: the potential soil loss/
sedimentation component and the phos
phorus component. The potential soil loss/
sedimentation component estimates for
each land cell the seasonal potential soil
loss, using a seasonal application of the
universal soil loss equation, and the subse
quent seasonal amount of sediment deliv
ered to the adjacent downstream land cell
and to the downstream stream cell, using a
seasonal sediment delivery ratio function.
Based on the assumption that no instream
sedimentation processes occur, the sea
sonal sediment load at the outlet of the
watershed is estimated by summing up all
predicted cell-to-stream sediment loads.
The phosphorus component predicts for
each land cell the seasonal phosphorus
load by the amount of soil-derived particu
late phosphorus in overland runoff, using
the average total phosphorus con
centration ofthe surface soil (10 mm), the
potential soil loss, the sediment delivery
ratio, and the phosphorus enrichment
ratio. This concept is based on the hypoth
esis that most, if not all, ofthe phosphorus
carried in overland runoff is derived from
soil particles detached from the land sur
face by sheet and rill erosion. The result
ing seasonal phosphorus load at the outlet
ofthe watershed is calculated by summing
up all predicted cell-to-stream phosphorus
loads.
This paper introduces a brief overview
of different means of controlling phos
phorus nonpoint source pollution, and
presents an application of GAMESP to
evaluate the potential reductions in down
stream sediment and phosphorus loadings
from cropland areas when selected best
management practices (i.e. a combination
of land management practices) are applied
on problem areas of a small Ontario agri
cultural watershed.
BEST MANAGEMENT PRACTICES
Land management practices that can be
used to control phosphorus nonpoint
source pollution and that are socially and
economically acceptable are termed "best
management practices" (BMP). The types
of best management practices that are
technically feasible and socially and eco
nomically acceptable are limited. Best
management practices have been grouped
by Baker and Johnson (1983) into those
which reduce: (i) the volume ofthe carrier
(amount of sediments or runoff volume),
(ii) the phosphorus concentration in the
carrier, or (iii) the delivery from the field
to surface waters.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 163

 The volume of the carrier from the field strip tillage, (ii) winter cover crops, (iii) 1986) have observed that extractable-
can be reduced by use of conservation conservation cropping systems (rotations) phosphorus soil test values are generally
tillage practices, contouring practices, and (iv) critical area planting. much higher with conservation tillage (no-
close-grown crops and tile drainage. The The erosion control provided by con till) than with plowing. With plowing,
phosphorus concentration in the carrier servation tillage nearly always results in a concentrations of extractable phosphorus
can be decreased by the incorporation of reduction of total losses of phosphorus rel are approximately uniform to the max
chemical fertilizer in the soil surface, the ative to losses with conventional tillage imum depth of cultivation (20 cm). Drew
application of chemical fertilizer during (fall moldboard plowingfollowed by seed and Saker (1978) have also observed that
the spring period and the application of bed preparation). Decreases in total phos with no-till, the concentration of extracta
chemical fertilizer during a dry period. phorus loss are less than the decreases in ble phosphorus in the first 2.5 cm of soil is
The delivery from the field to surface soil erosion because phosphorus enrich more than 50% greater than with plowing,
waters can be diminished by the imple ment ratios associated with conservation but the degree of increase is highly vari
mentation of terraces, grassed waterways tillage systems are greater than phos able. Further, they pointed out that below
and vegetative filter strips. phorus enrichment ratios associated with 10 cm, the concentration falls to about
The U.S. Army Corps of Engineers conventional tillage systems (Logan and 25% of that in the plowed soil.
(1982), in its final report on The Lake Erie Adams 1981; Logan and Forster 1982). The aforementioned overview of best
Wastewater Management Study, focussed Logan and Adams (1981) have defined a management practices has focussed on the
on cultural and land management practices coefficient of phosphorus reduction effi use of conservation tillage systems as
as cost-effective means of reducing Lake ciency as the ratio of the actual reduction means of reducing phosphorus loadings
Erie phosphorus nonpoint source pollu in total particulate phosphorus to the actual from erosional or cropland sources.
tion. The practices in this category deal reduction in sediment delivered. In their Results of conservation tillage studies
principally with cultural and land manage study, they have indicated that on average (Romkens et al. 1973; McDowell and
ment decisions made by land owner, oper the coefficient of phosphorus reduction McGregor 1984; Mueller et al. 1983;
ator and user. The land management efficiency is in the order of 0.89. The Angle et al. 1984) have suggested that: (i)
practices are most suitable when phos coefficient of phosphorus reduction effi conservation tillage systems are expected
phorus losses from a watershed are domi ciency is less than 1, since reductions in to greatly decrease sediment and phos
nated by cropland or erosional sources. sediment and particulate phosphorus phorus losses to surface waters, (ii) the
These practices primarily affect the soil losses resulted in greater phosphorus concentrations of extractable phosphorus
surface and its protection from the impact enrichment ratios. of the surface soil are likely to increase
of falling raindrops. The practices have Several authors (Shear and Moschler with the implementation of conservation
been classified by the U.S. Army Corps of 1969; Triplett and Van Doran 1969; Drew tillage systems and (iii) the main effect of
Engineers as follows: (i) conservation and Saker 1978, 1980; Oloya and Logan conservation tillage systems on phos
tillage, i.e. no-till, reduced tillage, and 1980; Ketcheson 1980; O'Halloran et al. phorus loadings to surface waters is to

/<')

Water Survey (
of Canada Downie Township \
Stratford/Avon
Station 02GD01& '' . Demonstration
South Easthope TownshipL_-Watershecj

0 1 2 3 4 5 km.

Figure 1. Location of the Stratford/Avon Demonstration Watershed in the Avon River Basin.
(After Stratford Avon River Environmental Management Project 1983.)

164 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

significantly decrease the phosphorus outlet 163 t of sediment and 257 kg of applied to problem areas of the study
yield from agricultural land with a coeffi phosphorus (i.e., 1.6 g of phosphorus per watershed, a variety of sets of input data
cient of phosphorus reduction efficiency 1000g of sediment) have been estimated to have been developed in accordance with
close to one. bedelivered at the watershed outlet during selected land management scenarios. The
the February through May period (Rous selected land management scenarios
STUDY AREA AND FARMING seau 1985). This initial application of included: (i) fall moldboard plowing with
PRACTICES GAMESP has revealed that: (i)the spatial maximum or without crop residues left on
The Stratford/Avon Demonstration variability of cell-to-stream phosphorus the field, (ii) spring moldboard plowing
Watershed has been selected as the study yield reflected the spatial variability of with maximum or without crop residues
area. This study watershed of 5.37 km2 is cell-to-stream sediment yield(Figs. 2 and left on the field, (iii) conservation tillage,
located in North Easthope Township, 3), and (ii) mostofthe phosphorus load to i.e. planting the crop with a minimum of
Perth County, within the 165-km2 Avon stream emanated from steeply sloped corn seedbed disturbance in the remaining resi
River Basin (Fig. 1),a subwatershed ofthe and small grain fields located near the dues of the previous crop by a coulter,
North Thames Basin in Southwestern drainage channel. single chisel, or similar tool with seed
Ontario. The land is distinctlyrolling. The In this study, to illustrate the potential placement at the same time and (iv) cross-
predominant soil in the watershed is silt reductions in downstream sediment and slope farming with fall moldboard plow
loam (79% ofthe area); some clay loams, phosphorus loadings from agricultural ing without crop residues left on the field.
loams, and sandy loams are also present. land when best management practices are Each land management scenario has been
The agricultural activities ofthe area are
mostly livestock based, including swine,
beef and dairy operations. Cropping prac STRATFORD/AVON
tices reflect the emphasis on livestock DEMONSTRATION WATERSHED
operations, with rotations of hay/grain/ - SEDIMENT LOAD FROM CELL TO STREAM
corn or hay/grain being preferred. Contin 1983 spring conditions
uous row cropping with corn is the next
most frequent system. For the 1983 crop
year the percentages of the watershed in
corn fields, hay/pasture fields, small grain
fields, and woodlot were 48, 23, 15 and
14%, respectively.
An almost universal tillage practice in
the watershed is fall plowing with a mold- 1 km
board plow. Another popular practice in
the fall involves either discing or cultivat
ing prior to or following plowing. Second
ary tillage in the spring usually involves
two or three passes with a cultivator. The
use of fertilizers in the spring season is
another universal practice throughout the
watershed. The average drainage density
of the watershed is 6.5 m ha- {.

METHODOLOGY
To explore the potential impact of best
management practices during the hydro-
logically active spring period in South
western Ontario, GAMESP was first used
to identify the source areas of sediment INTERVALS C t ]
and phosphorus delivery to surface waters
000-0.49
of the Stratford/Avon Demonstration
Watershed for the 1983 spring season 0.50-1.24
(Rousseau et al. 1985). The application of 1.25-2.49
GAMESP involved: (i) the discretization 2.50 -4.99
of the study watershed into land and >5.00
stream cells, (ii) the determination of the
drainage network of each cell to the water
shed outlet, (iii) the quantification for each
cell ofthe input parameters required by the
model, and (iv) the estimation of sediment
and phosphorus loads at the watershed out
let (attributed to sheet and rill erosion from
cropland for the season under considera
tion). From water discharge, sediment and Figure 2. Spatial variability of cell-to-stream sedimentyield in the Stratford/Avon Demonstra
phosphorus data collected at the watershed tion Watershed for the 1983 spring conditions.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 165

 applied on corn and small grain fields, and
STRATFORD/AVON
has required adjustments in the following
seasonal parameters: crop and manage
DEMONSTRATION WATERSHED ment factor Cs, manningroughness coeffi
PHOSPHORUS LOAD FROM CELL TO STREAM- cient ns, and NaHC03-extractable phos
1983 spring conditions phorus concentration of the surface soil
APS (Table 1). The crop and management
factor Cs was computed according to pro
cedures proposed by Cook et al. (1985)
and Wischmeier and Smith (1978). For
corn fields (silage corn and grain corn) and
small grain fields a good productivity level
was assumed when the crop and manage
ment factor was estimated. Hence, for
1 Urn
corn fields the crop residue cover was
assumed to be the same for silage corn and
grain corn fields. The Manning roughness
coefficient, ns, was modified according to
Cook et al. (1985). When conservation
tillage practices were simulated, the
NaHC03-extractable phosphorus of the
surface soil APS was increased by 30% to
account for surface applied fertilizer
(Drew and Saker 1978, 1980). A support
practice factor (Ps) was used to represent
cross-slope farming and was taken into
account by multiplying the crop and man
agement factor Cs by a value of 0.6
(Wischmeier and Smith 1978).
INTERVALS Ckg] The selected land management sce
narios were applied in a first series of
0.00 - 0.49 simulation runs on all corn and small grain
0.50-1.24 fields found in the watershed. In a second
1.25 - Z49 series of simulation runs, the selected land
management scenarios were applied on
2.50 - 4.99
corn and small grain fields located in prob
^5.00 lem areas of the watershed. The problem
areas were defined according to van Vliet
et al. (1978)and Dickinson et al. (1983) by
cells having a gross spring erosion rate
greater than 2.5 t ha-' or a cell-to-stream
spring sediment delivery rate greater than
1 t ha ~'. The problem areas represented
45% of the watershed area. Since fall
Figure3. Spatial variability of cell-to-stream phosphorus yieldintheStratford/Avon Demonstra moldboard plowing with crop residues
tion Watershed for the 1983 spring conditions.
removed was the universal tillage practice
TABLE I. BEST MANAGEMENT PRACTICES-INPUT PARAMETERS FOR CORN AND SMALL in the study watershed for the 1983 crop
GRAIN FIELDS year, it was taken as the basic reference
level for the determination of potential
Land management Silage corn fields Small grain fields
reductions in downstream sediment and
scenarios cst nsX APJ cs "s APS phosphorus loadings.
Fall moldboard plowing
Without crop residues 0.503 0.07 21.0 0.335 0.04 21.0
With maximum crop RESULTS AND DISCUSSION
residues 0.341 0.15 21.0 0.284 0.15 21.0 Table II presents the potential reduc
Spring moldboard plowing tions in dowstream sediment and
Without crop residues 0.420 0.07 21.0 0.268 0.04 21.0 phosphorus yields during the hydro-
With maximum crop
residues 0.234 0.15 21.0
logically active spring period (February
0.196 0.15 21.0
Conservation tillage through May), when best management
No-till 0.111 0.15 27.3 0.153 0.15 27.3
practices are applied on the Stratford/
Cross-slope farming Avon Demonstration Watershed. These
Without crop residues 0.302 0.10 21.0 0.201 0.10 21.0 results have prompted the following obser
tCs = seasonal crop and management factor (dimensionless). vations, (i) The estimated potential reduc
$ns = seasonal Manning roughness coefficient (s m-1/3). tions in potential soil loss, downstream
APS = seasonal NaHC03-extractable phosphorus ofthe surface soil (u>g g_l). sediment load, and downstream phos-

166 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

TABLE II. IMPACT OF BEST MANAGEMENT PRACTICES ON POTENTIAL SOU I OSS and 56%, respectively, when compared to
DOWNSTREAM SEDIMENT AND PHOSPHORUS LOADS, AND THE A^vSaGE CELL TO fall moldboard plowing without a crop res
STREAM PHOSPHORUS ENRICHMENT RATIO IN THE STRATFORD/AVON^SeMONSTOA-' idue cover, (vii) Finally, it can be
TION WATERSHED FOR SPRING CONDITIONS
observed that best management practices
Reductions in applied only on areas of high sediment
Potential Sediment Phosphorus Average
Land management soil loss load
Potential Sediment Phosphorus yields achieve almost the same potential
load cell-to-stream soil loss
scenarios (0 (t) (kg) PERf
load load reductions in downstream phosphorus
(%) (%) (%)
loads as if they were applied on the entire
Fall moldboard plowing watershed.
Without crop residues
Entire watershed 2582$ 163$ 257$ 2.4 0 0 o
Withmaximum crop residues CONCLUSIONS
Entire watershed 1830 43 90 3.2 29 This paper has introduced a brief over
74 65
Problem areas 1890 51 107 3.1 27 69 58 view of best management practices and
Spring moldboard plowing
Without crop residues
has emphasized the use of conservation
Entire watershed 2154 123 221 2.5
tillage systemsas meansof reducing phos
17 18 14
Problem areas 2213 137 227 2.5 14 16 12
phorus nonpoint source pollution.
With maximum crop residues GAMESP, the Guelph model for evaluat
Entire watershed 1278 30 67 3.4 50 82 ing the effects of agricultural management
74
Problem areas 1890 37 83 3.4 48 78 68 systems on erosion, sedimentation, and
Conservation tillage no-till
Entire watershed 703 17
phosphorus yields has been used to evalu
45 3.9 73 90 83
Problem areas 790 23 60 3.7 70 86 77
ate the impacts of best management prac
Cross-slope farming tices on specific sites of a small Ontario
Without crop residues agricultural watershed. This specific
Entire watershed 1576 56 111 3.0 39 67 56 investigation has revealed that best man
Problem areas 1648 63 126 3.0 36 61 51 agement practices using a maximum crop
tPhosphorus enrichment ratio. residue cover such as conservation tillage
$ Basic reference level. (no-till) and spring moldboad plowing
could potentially reduce downstream
TABLE III. THE EFFECT OF BEST MANAGEMENT PRACTICES ON THE COEFFICIENT OF phosphorus loadings during the spring
PHOSPHORUS REDUCTION EFFICIENCY period by as much as 83 and 74%, respec
tively. Further, reductions in downstream
Land management scenarios Entire watershed Problem areas
sediment and phosphorus loadings are
Fall moldboard plowing likely to be accompanied with greater
With maximum crop residues 0.88 0.84
increases in the average cell-to-stream
Spring moldboard plowing
Without crop residues 0.78
phosphorus enrichment ratio than the one
0.75
With maximum crop residues 0.90 0.87
reported for fall moldboard plowing on
Conservation tillage fields left without a crop residue cover. It
No-till 0.92 0.90 has also been observed that best manage
Cross-slope farming ment practices applied only on areas of
Without crop residues 0.85 0.84
high sediment yields located near the
drainage channel would achieve almost
phorus loads reveal that the best manage to-stream phosphorus enrichment ratio. the same potential reductions in down
ment practices considered have moderate Consequently, when best management stream phosphorus loads as if they were
to significant impacts, (ii) For spring and practices are applied, the number of grams applied on the entire watershed.
fall moldboard plowing on fields left with of phosphorus transported by 1000 g of
ACKNOWLEDGEMENTS
out a residue cover, reductions in down sediment is greater than the one reported
Sincere appreciation is expressed to D. J.
stream sediment load are similar to for fall moldboard plowing on fields left
Cook former research associate in Water
reductions in potential soil loss. For other without a residue cover, (v) For all the best Resources Engineering, University of Guelph,
land management scenarios involving a management practices considered, the for providing his computer expertise. In addi
crop residue cover on fields, reductions in coefficient of phosphorus reduction effi tion to the funds of Environment Canada and
downstream sediment load are larger than ciency varies between 0.75 and 0.92 Agriculture Canada through the Department of
reductions in potential soil loss. This non- (Table III), a range similar to that reported Supply and Services, the financial support of
linearity results from the fact that potential by Logan and Adams (1981). The coeffi the Natural Sciences and Engineering Research
soil loss is altered by changes in land use, cient of phosphorus reduction efficiency is Council of Canada and the Ontario Ministry of
i.e. the seasonal crop and management greater for land managment scenarios that Agriculture and Food is gratefully acknowl
factor Cs, while the sediment load is use a crop residue cover than for those that edged.
affected by the seasonal crop and manage do not use one. (vi) Conservation tillage,
ment factor Cs and the seasonal manning spring moldboard plowing on fields with a REFERENCES
ANGLE, J. S., G. McCLUNG, M. S.
roughness coefficient ns. (iii) Reductions crop residue cover, fall moldboard plow
Mcintosh, p. m. thomas, and d. c.
in downstream phosphorus load are correl- ing on fields with a maximum crop residue WOLF. 1984. Nutrient losses in runoff from
lated with reductions in downstream sedi cover, and cross-slope farming on fields conventional and no-till corn watersheds.
ment load, (iv) Reductions in downstream without a crop residue cover can poten J. Environ. Qual. 13: 431-435.
sediment and phosphorus loads are accom tially reduce the spring load of phosphorus BAKER, J. L. and H. P. JOHNSON. 1983.
panied with increases in the average cell- attributed to field erosion by 83, 74, 65, Evaluating the effectiveness of BMPs from

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 167

 field studies. Chapter 15 inF.W. Shaller and
G.W. Bailey, eds. Agricultural manage
ment and water quality. Iowa State Univer
sity Press, Ames, Iowa.
COOK, D. J., W. T. DICKINSON, and R. P.
RUDRA. 1985. GAMES, Guelph model for
evaluating the effects of agricultural man
agement systemson erosion andsedimenta
tion, user's manual. School of Engineering,
University of Guelph, Guelph, Ont.
COOTE, D. R., E. M. MacDONALD, W. T.
DICKINSON, R. C. OSTRY, and R.
FRANK. 1982. Agriculture and water
quality in the Canadian Great Lakes Basin.
I. representative agricultural watersheds.
J. Environ. Qual. 11:473-481.
DICKINSON, W. T., R. PALL, and G. J.
WALL. 1983. Identification of soil erosion
and fluvial sediment problem. ASAE Paper
NAR 83-204.
DREW, M. C. and L. R. SAKER. 1978.
Effects of direct drilling and ploughing on
root distribution in spring barley, and on the
concentrations of extractable phosphorus
and potassium in the upper horizons of aclay
soil. J. Sci. Food Agric. 29: 201-206.
DREW, M. C. and L. R. SAKER. 1980. Direct
drilling and ploughing: their effects on the
distribution of extractable phosphorus and
potassium, and of roots, in the upper hori
zons of two clay soils under winter wheat
and spring barley. J. Agric. Sci. 94:
411-423.
KETCHESON, J. W. 1980. Effect of tillage on
fertilizer requirements for corn on a silt loam
soil. Agron. J. 72: 540-542.
LEE, G. F. 1973. Role of phosphorus in
eutrophication and diffuse source control.
Water Res. 7: 111-128.
LEE G. F., W. ROST, and R. A. JONES.
1978. Eutrophication of water bodies:
Insight for an age-old problem. Environ.
Sci. Techcl. 12: 900-908.
168
LOGAN, T. J. and J. R. ADAMS. 1981. The
effects of reducedtillage on phosphatetrans
port from agricultural land. LEWMS, Tech
nical Report, U.S. Army Corps of
Engineers, Buffalo, N.Y.
LOGAN, T. J. and D. L. FORSTER. 1982.
Alternative managementoptions forthe con
trol of diffuse phosphorus loads to Lake
Erie. LEWMS Technical Report, U.S.
Army Corps of Engineers, Buffalo, N.Y.
McDOWELL, L. L. and K. C. McGREGOR.
1984. Plant nutrient losses in runoff from
conservation tillage corn. Soil Tillage Res.
4:79-91.
MILLER, M. H., J. B. ROBINSON, D. R.
COOTE, A. C. SPIRES, and D. W.
DRAPER. 1982. Agriculture and water
quality in the Canadian Great Lakes Basin.
III. phosphorus. J. Environ. Qual. 11:
487-493.
MUELLER, D. H., B. J. ANDRASKI, T. C.
DANIEL, and B. LOWERY. 1983. Effect of
conservation tillage on runoff water quality:
total, dissolved and algae-available phos
phorus losses. ASAE Paper No. 83-2535.
MYERS, P. C. 1986. Nonpoint source pollu
tion control: the USDA position. J. Soil and
Water Cons. 41(3): 156-158.
NELSON. D. W. and T. J. LOGAN. 1983.
Chemical processes and phosphorus trans
port. Chapter 5 in F. W. Shaller and G. W.
Bailey, eds. Agricultural management and
water quality. Iowa State University Press,
Ames, Iowa.
OGG, C. W. and H. B. PIONKE. 1986. Water
quality and the new farm policy initiatives.
J. Soil Water Cons. 41(2): 85-88.
O'HALLORAN, I. P., M. H. MILLER, and
G. ARNOLD. 1986. Absorption of Pby corn
(Zea mays L.) as influenced by soil distur
bance. Can. J. Soil Sci. 66: 287-302.
OLOYA, T. O. andT. J. LOGAN. 1980. Phos
phate desorption from soils and sediments
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
withvarying levelsof extractable phosphate.
J. Environ. Qual. 9: 526-531.
ROMKENS, M. J. M., D. W. NELSON, and
J. V. MANNERING. 1973. Nitrogen and
phosphorus composition of surface runoff as
affectedby tillage method.J. Environ. Qual.
2: 292-295.
ROUSSEAU, A. 1985. Development of a
phosphorus transport model for small agri
cultural watersheds. M.Sc. Thesis. Univer
sity of Guelph, Guelph, Ont.
ROUSSEAU, A., W. T. DICKINSON, and
R. P. RUDRA. 1985. A phosphorus trans
port model for small agricultural watersheds.
ASAE Paper No. 85-2526.
STRATFORD AVON RIVER ENVIRON
MENTAL MANAGEMENT PROJECT.
1983. Physical characteristics and land uses
of the Avon River drainage basin. Technical
Report R-9. Upper Thames River Con
servation Authority, London, Ont.
SHEAR, G. M. and W. W. MOSCHLER.
1969. Continuous corn by the no-tillage and
conventional tillage methods: a six year
comparison. Agron. J. 61: 524-526.
TRIPLETT, G. B. Jr. and D. M. VAN
DORAN Jr. 1969. Nitrogen, phosphorus and
potassium fertilization of non-tilled maize.
Agron. J. 61: 637-639.
U.S. ARMY CORPS OF ENGINEERS, BUF
FALO DISTRICT. 1982. Lake Erie Man
agement Study, Final Report, Buffalo, N.Y.
VAN VLIET, L. J. P., G. J. WALL, and W. T.
DICKINSON. 1978. Erosion losses and sed
iment delivery ratios for agricultural water
sheds. I.J.C. Tech. Rep. re: Great Lakes
Pollution from Land Use Activities, Wind
sor, Ont.
WISCHMEIER, W. H. and D. D. SMITH.
1978. Predicting rainfall erosion lossesA
guide to conservation planning. Agricultural
Handbook No. 537, USDA, Washington,
D.C.
 EVALUATION OF POTATO-STONE-CLOD SEPARATION MECHANISMS
G. C. Misener and C. D. McLeod

Research Station, Agriculture Canada, P.O. Box 20280, Fredericton, New Brunswick E3B 4Z7
Received 28 August, 1986, accepted 16 March, 1987

M^on^QG;^'and C*D'McLeod- mi- Evaluation ofpotato-stone-clod separation mechanisms. Can. Agric. Eng.
^v: lo" l/I.

Methods of separating potatoes from stones and soil clods in storage were evaluated and factors affecting performance
were determined. The three methods ofseparation included a rotating brush separator, an electronic separator and an
impact roller separator. All separators effectively sorted the stones from the potatoes but the brush separator failed to
differentiate between the soil clods and potatoes. Parameters affecting the performance ofthe brush separator were the
feed rate and the brush gap while the performance of the electronic and impact roller separators was affected by the feed
rate and the stone/potato ratio, respectively.

INTRODUCTION
Thedevelopment of mechanisms to sep
arate potatoes from stones and clods has
been directed towards mechanical devices
located on the harvesters. These devices
have been designed to separate potatoes on
the basis of physical differences between
the potato and the associated material. The
differences include rolling resistance,
elasticity, specific gravity, terminal
velocity, resistance to X-ray penetration,
optical reflectance, and vibrational fre
quency response. Sides and Smith (1970)
describe the mechanisms available for sep
arating potatoes and stones. Methods
which have been used or investigated over
the past several years have also been
described by Story and Raghavan (1976).
More recently several mechanisms have
been developed to operate in storages dur
ing postharvest potato handling. The
advantage of doing a portion ofthe separa
tion at storage rather than on the harvester
is to reduce the number of people working
on the harvester under unfavorable condi
tions. Often in Eastern Canada from two to
four people are required to sort the ma
terial on the harvester. Two approaches
for in-storage use appear promising and
have been developed into commercial ma
chines. A commercial separator developed
by Feller et al. (1984) is based on the
difference between the coefficient of res
titution of the products being separated.
The device involves bouncing the objects
ejected from the flow by means of high
speed pneumatic hammers. The basic pro
cess is similar to the roller with the advan
tage that the impact differences are
amplified electronically as opposed to
creating larger differences by changing the
contact angle on the roller. Larger dif
ferences increase separating sensitivity.
One disadvantage is having to align the
product in order to optimize the perfor
mance of the electronic and impact ham
mer systems. A third method of separation
which is less expensive is presently at the
experimental stage. This method utilizes a
transaxial rotary brush separator in which
the material flow is across the brush axis.
Separation is achieved by a differential
deflection of the bristles caused by the
difference in specific gravities of the
potatoes, stones and soil clods as
described by Eaton and Hansen (1969).
Figure 1. Cross-sectional view of brush separator.
DISTRIBUTION
The objective of the study described in
this paper was to quantify the performance
of the three types of in-storage separators
in terms of their capability to separate
potatoes from stones and soil clods.
TEST PROCEDURES
The laboratory arrangement included a
feeding conveyor on which the products
were placed and fed at controlled rates
onto each of the three separators. The
three separators are shown schematically
in Figs. 1, 2 and 3. Russet Burbank
potatoes were distributed evenly at a
known weight on the conveyor and then
either stones or clods were uniformly
mixed with the potatoes. The conveyor
speed was constant during the tests while
the weight of material per unit length on
the conveyor was varied to deliver various
loading rates to the separators. Prelimi-

on a revolving roller and separating them BRUSH

according to their trajectories. A second -SENSOR PLATE

type of separator available differentiates DISTRIBUTION | -REJECTION CYLINDERS

ROLLERS "
between potatoes, stones and soil clods
electronically. This separator has a series
Srfr fifcfl ffftQQvof (TV/
of sensors each consisting of a membrane
and an electromagnetic transducer. The
\

magnetic field is influenced by the vibra 1

tion of the membrane due to impact and %
the magnetic-field-influencing property of J
^

the products being separated. Using these

differences, the items not desired are Figure 2. Cross-sectional view of electronic separator.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 169


Feed Conveyor
A Roller Separator
Potatoes &
Figure 3. Cross-sectional view of roller separator.
nary tests were conducted to determine the
range of machine settings for each
machine when operating with various
stone/potato and clod/potato ratios.
The stones used in the tests had an aver
age weight of 79.4 g with standard devia
tion of 55.7 g. Clods were gathered in
the fall and stored either in opened barrels
or in plastic bags in order to obtain two soil
moisture levels of 3.5 and 19.0% WB. The
average tuber weight was 154.6 g with a
standard deviation of 89.8 g.
The stone/potato ratios and clod/potato
ratios were selected on a weight basis and
represent ratios often found under New
Brunswick conditions. The field stone/
potato ratio depends on the soil type as
well as to the extent of the stone removal
operation. The clod/potato ratio has a
higher variability depending on soil type
and moisture as well as tillage practices.
The feed rate selected during the evalua
tion of the separator represents sufficient
capacity to handle the output from two
harvesters under New Brunswick condi
tions. A large percentage of growers oper
ate two harvesters in order to meet the
requirements of a short harvest period.
Brush Separator
The critical parameter affecting ma
chine performance is the gap between
adjacent pairs of brushes. Eaton and
Hansen (1969) found that the separating
efficiency was independent of brush
speed, loading rate or separator slope.
Tests consisting of three replications were
randomly conducted with ratios of stones/
potatoes of 0.01, 0.055 and 0.1 over a
range of feed rates of 9.0, 18.0, 27.0 and
36.0 t/h over 1.3 m of brush width. The
tests included brush gap settings of 7.0,
10.0, 13.0 and 16.0 mm. Due to the small
difference in specific gravities of soil clods
and potatoes initial tests indicated that a
170
two moisture levels were included in a
second test. The tests were randomly con
ducted at four feed rates of 9.0, 18.0, 27.0
and 36.0 t/h, two soil clod moisture levels
of 3.5% WB and 19.0% WB, and one
clod/potato ratio of 0.05.
Clods
Feller et al. (1984) have done an exten
sive evaluation of the roller separator sort
ing onion-clod mixtures. They found that
the separator was very effective when sep
arating the clods from the onions. Gan-
Mor et al. (1985) suggested that a machine
A similar to the roller separator will not
effectively separate the stones from the
potatoes due to the small difference in the
coefficient of restitution between potatoes
and stones. However, in order for the
machine to be useful in Eastern Canada it
low separation efficiency was achieved
must be capable of handling stones as well
with the brushes when separating these
as clods. The initial tests indicated that
products. Consequently, only one set of
separation of stones from potatoes was
tests was undertaken with soil clods mixed
possible; consequently, a complete series
with potatoes. The ratio of clods/potatoes
of tests was performed. The tests included
was 0.05 with feed rates of 9.0,18.0, 27.0
four feed rates of 9.0, 18.0, 27.0 and 36.0
and 36.0 t/h at a brush gap setting of
13.0 mm.
t/h and three stone/potato ratios of 0.01,
0.055 and 0.1.
Electronic Separator Separation effectiveness, E, was deter
Preliminary tests were conducted with mined as the product of recovery and
the electronic separator in order that the rejection (Brown et al. 1951). Feller et al.
sensitivity (amplification) adjustments for (1984) also used this method of measuring
clods and stones could be selected. When
separation effectiveness:
the optimum setting was reached, tests p s
were conducted at four feed rates of 9.0, Pp + Pr sr +
18.0, 27.0 and 36.0 t/h over a 1.3 m width pr
with three stone/potato ratios of 0.01, (l -z)d
Pp + Pr sr + SD (1)
0.055 and 0.1. In order to measure the
effect that moisture content of soil clods where Pp = potatoes in the product exit;
going into storage in early fall and leaving = potatoes rejected with the stones;
during winter has on the performance of = stones in the reject exit; and
the electronic separator, soil clods with s = stones remaining with the product.
Brush Gap
. . . . 10 . 0 mm 1 fi fl mm
.0
L
O
"' ^^^><^_
-H
O
. 9
"^^^ -
CD
li
.8
en
"*^-
CO
CD . 7
c
CD
**N..
"^-^ . "
.6
- "
H
O "--
CD
. 5
0-
LJ . 4
I I I
. 3 I
9 18
Feed Rate t/h
Figure 4. Effectiveness of the brush separator as effected by feed rate and brush gap.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 L
BRUSH SEPERATOR
1 .0
O -_ ELECTRONIC SEPERATOR for separating stones from potatoes,
-H
U ._ ROLLER SEPERATOR
although Gan-Mor et al. (1985) indicate
CD otherwise.
LL
The performance ofthe three separators
Cfl
0. 9
when separating a mixture of potatoes and
CO stones is shown in Fig. 5. Overall, the
CD
C performance of the brush separator ap
CD pears superior.
>
r-i 0.8
-H
U CONCLUSIONS
CD From the results of the study, it can be
C^
Q- concluded that separating stones from
LU 0. 7 potatoes with any of the three separators is
0 18 27 36 effective. The feed rate and the width of
Feed Rate t/h the gap between brushes are factors that
affect the performance of the brush sepa
Figure 5. Comparison of the performance of the three separators over a range of feed rates.
rator with stones and potatoes. However,
this separator has limited capability to sep
TABLE I. COMPARISON OF PARAMETERS AFFECTING PERFORMANCE OF THE THREE
arate soil clods from potatoes. The elec
SEPARATORS
tronic separator is sensitive to the feed rate
Brush separator Electronic separator Roller separator when separating mixtures of stones and
Feed rate Brush gap Feed rate Stone/potato potatoes. Moisture of soil clods multiplied
(t/h) E (mm) E (t/h) E ratio E by feed rate is a significant factor when
9.0 0.81a 7.0 0.46a 9.0 0.89a 0.01 0.77a separating clods from potatoes. The stone/
18.0 0.77a 10.0 0.72b 18.0 0.88a 0.055 0.88/7 potato ratio becomes a significant factor
27.0 0.70b 13.0 0.90c 27.0 0.87a 0.1 0.86/7
0.82/7
for the roller separator when separating
36.0 0.70b 16.0 0.91c 36.0
stones from potatoes.
a-c Means notfollowed by same letter are significantly different at the 5% level of probability as judged
by Duncan's multiple range test.
REFERENCES
Analysis of variance was carried out on the mean effectiveness was only 0.15 BROWN, G. G., O. KATZ, and A. S.
the data using the program Genstat when attempting to separate the clods and FOUST. 1951. Unit operations. John Wiley
ANOVA 1982. Significant differences of potatoes. The small difference in specific and Sons, New York. p. 15.
means were examined utilizing Duncan's gravities of the two products probably EATON, F. F. and R. W. HANSEN. 1969.
multiple range test. accounts for the low performance. Mechanical separation of stones from
With the electronic grader, feed rate potatoes by means of rotary brushes. Am.
was found to be the only significant factor Soc. Agric. Eng. Paper No. 69-108.
RESULTS AND DISCUSSION
ASAE, St. Joseph, Mich.
The width of the gap between the affecting performance with a mixture of FELLER, R., O. NAHIS, and C. G. COBLE.
brushes was found to be an important stones and potatoes. Feed rates above27.0 1984. Separation of soil clods from onions
parameter for determining the separation t/h began having a negative effect on the using impact. Trans. ASAE (Am. Soc.
effectiveness of the brush separator. The effectiveness of the separator. Again, as Agric. Eng.) 27(2): 353-357.
analysis ofvariance ofthe data which were with the brush separator, the stone/potato GAN-MOR, S., A. ZACHARIN, R. FEL
accumulated in the process of evaluating ratio did not significantly affect perfor LER, and E. MAROGOLIN. 1985. Reduc
the sensitivity of the brushes to gap and mance. Although the moisture level ofthe tion ofthe restitution coefficient for stones to
feed rate with mixtures of potato and clods was found not to be a significant enable separation from potatoes. Am. Soc.
Agric. Eng. Paper No. 85-1569. ASAE,
stones indicated that both factors, brush factor on the performanceofthe electronic
St. Joseph, Mich.
gapand feedrate, werehighlystatistically separator,the interaction of moisture level SIDES, S. E. and N. SMITH. 1970. Analysis
significant. Their effects on the effec and feed rate was highly significant. and design of potato-stone separation mech
tiveness are shown graphically in Fig. 4. The test results with the roller separator anisms. Am. Soc. Agric. Eng. Paper No.
In Table I, the analysis indicates that a indicate that the stone/potato ratio was a 70-673. ASAE, St. Joseph, Mich.
brushgap ofat least 13.0mm is required to significant factor affecting performance. STORY, A. G. and G. S. V. RAGHAVAN.
maximize the effectiveness of the sepa The overall mean of the effectiveness fac 1976. Sorting potatoes from stones and soil
rator. As indicated by Eaton and Hansen tor of the roller separator for tests con clods by infrared reflectance. Quality detec
(1969), soil clods could not be separated ducted was 0.84. This value suggests that tion in foods. Am. Soc. Agric. Eng., St.
from potatoes using brushes. In our tests, the separatorcan be an effective machine Joseph, Mich. pp. 177-182.

171
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 SIMULATION OF ADSORPTION DRYING OF CORN, WHEAT,
BARLEY AND OATS USING SODIUM BENTONITE

K. C. Watts1, W. K. Bilanski2, and D. R. Menzies3

'Department of Agricultural Engineering, Technical University of Nova Scotia, Box 1000, Halifax, N.S.
B3J 2X4; 2School of Engineering, University of Guelph, Guelph, Ontario NIG 2W1; and ^Agriculture
Canada, Research Branch, Research Station, Vine Iand Station, Ontario LOR 2E0

Received 25 April 1986, accepted 10 October 1986

Watts, K. C. and W. K. Bilanski. 1987. Simulationof adsorptiondrying of corn, wheat, barley and oats using sodium
bentonite. Can. Agric. Eng. 29: 173-178.

Bentonite is an abundant clay mineral that is suitable for the drying of various grains by adsorption. The absorbency
characteristics are a function of the composition of the clay and the adsorbed ions. This paper presents a mathematical
model for adsorption drying of various grains using existing equations for drying individual kernels at low temperatures.
The equilibrium moisture content ofthe grain was derived from accepted data in the literature. It was necessary to obtain
the adsorption characteristics of the specific sodium bentonite being utilized using clay /grain mixtures. Comparison of
simulation results with experimental data showed satisfactory agreement when a mass ratio of grain to clay, which was
derived by imposing a moisture balance on the experimental data, was used in the simulation.

INTRODUCTION
One major source of energy consump
tion on the farm is the drying of grains by
conventional systems which force natural
or heated air through the bed of material to
be dried. In the search for alternate meth
ods of drying, three requirements were
determined: (1) low conventional energy
usage; (2) the material should not be sub
jected to high temperatures; (3) the dryer
must be available at all times, independent
of weather or season.
The only method of drying that meets
all three requirements is that of desiccant
drying in which the moisture which is
removed from the grains is temporarily
adsorbed onto the surface ofthe desiccant.
Low-grade forms of energy, e.g. solar,
mixed. This mixture can be analyzed as a
three-stage process:
(1) The moisture movement out of the
grain into the air which is assumed to
surround the grain.
(2) The moisture adsorption by the clay
from the air.
(3) The resultant relative humidity ofthe
air due to the resultant moisture content of
the clay.
When the clay is dry, the resultant rela
tive humidity will be low causing a greater
potential for extracting moisture from the
grain. As the clay adsorbs more moisture,
the relative humidity increases and it is
less able to extract moisture from the
grain. Since the relative humidity sur
rounding the grain particles changes with
bentonite would rapidly adsorb any
moisture that was on the surface of the
grain. Cereal grains are normally consid
ered to dry during the falling-rate period.
This is due to the reasonably low moisture
content at which grains are harvested for
optimal operation of machinery and mini
mal seed damage. Hence, all moisture
relationships noted assume that only fall
ing-rate drying is occurring where
moisture diffusion in the grain is the limit
ing moisture transfer mechanism.
The diffusion process has been model
led by various workers (Ficks second law
of diffusion; Graham et al. 1983; Keey
1972) and may be written, on simplifica
tion for constant diffusion coefficient, D,
as

could be used to dry the desiccant when
they are available so that the dryer is avail
able at any time throughout the year.
The desiccant that was identified by
Sturton et al. (1981) and Graham et al.
(1983) as having the most potential is ben
tonite, a widely distributed clay mineral
that is relatively inexpensive and yet has a
relatively high adsorption rate and capac
ity. A comparison of drying corn using
sodium and calcium bentonite was re
time, it is not possible to directly apply
= DV2M (1)
conventional drying equations that assume dt
a drying medium which has a constant
where / is time and M is the decimal
relative humidity on entering. On the con
trary, it is necessary to analyze this process moisture content (db) ofthe individual par
with time as outlined below.
ticle. Equation 1can be solved for different
shapes (Crank 1975) or as an infinite series
which is claimed to be independent of the
Grain Moisture Relationships
size and shape of the drying object:
Conventional drying of grain is gener
ally accomplished using deep-bed drying MR = 2 Ate1^ (2)
which is somewhat different than if the n = I

ported by Watts et al. (1986).
The objectives of this study are to pre
sent the mathematical models used in cal
culating the simultaneous drying curve of
the grain and the moisture adsorption
curve ofthe clay and to compare the results
to experimental data (from the literature)
obtained when using sodium bentonite to
dry four different types of grain.
MATHEMATICAL MODEL
Throughout this paper, it is assumed
that the clay and grain are intimately
kernels were dried separately. In this paper
or for spherical coordinates
each kernel is envisaged as being isolated
from all others and surrounded by the dry b V 1
ing medium. MR 2, e
tt" ., = . n~
(3)
A general drying process is considered
to comprise a constant-rate drying period and
and a falling-rate drying period. The con MR = (M - Mcq)/(A/0 - Mcq) (4)
stant-rate drying period occurs when the
internal resistance to moisture transport is where D = diffusion coefficient (m2 s~'
much less than the external resistance to or m2 h ~ l),M = volume average decimal
water vapor removal from the grain sur moisture content (db), t = drying time
face. This does not occur in the drying (hours or seconds), M0 = original
process under consideration here since the moisture content (db), Meq = equilibrium

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 173

moisture content (db),Aj = dimensionless
characteristics of the drying object, Bx =
constants characteristic of the drying
object (m~2), k = a "drying constant"
(s-1 orh"1).
The conditions imposed on the above
equations are that M(r,o) = M0 and
M(ro,0= Meq; that is, that there is an
infinite rate of mass transfer from the sur
face of the seed to the drying medium.
The constants Ax and Bx are chosen to fit
experimental data. The number of terms in
the series is adjusted for best fit of the data
normally 2 or 3. Researchers often
argue (Brooker et al. 1974) that drying is
analogous to Newton's law to cooling and
use only one term of Eq. 2 with A set at 1,
or one term of Eq. 3 with A set at 6/tt.
These equations are written as:
MR = ^xp(-Kt) (5)
and
Little information was found in the liter
ature on the drying of oats. Van Rest and
Isaacs (1966) proposed several forms of
drying equations and fit them to experi
mental data. They did not obtain an excel
lent fit for all forms of equations. For oats,
the best form of equation was found to be
Plage's
MR = 10-u,b. (13)
Van Rest and Isaacs (1966) presented data
of a and b for the best fit of several experi
mental runs at several temperatures, the
lowest being 80F. We averaged these 80F
data and obtained the coefficients a and b.
For consistency, Eq. 13 can be written as
MR = exp (- 2.303 atb)
= txp(-kt) (14)
where
k = 2.303 a/b-'
a = 0.2083
b = 0.6002 (15)
potential are also changing. The equi
librium moisture contents are calculated at
each time using the following equations.
Meqcorn = 0.33872 - 0.05897 x
ln[-(T + 30.205) In (RH)] (16)
(ASAE Yearbook, Chung equation.)
where M = moisture (db), decimal, T =
temperature (C), RH = relative humidity,
decimal.
The Chung equation did not accurately
represent the data given in the ASAE Year
book for barley or soft wheat at 25C.
Hence, the following equations were
derived from data (ASAE Yearbook 1982)
at 25C.
Meqsol(wheat = -0.0106 +
0.5923 RH - 1.0438 RH2 + 0.7708 RH1 (17)
and
Meqbarlcy = 0.0429 + 0.2344 RH -
0.3481 RH2 + 0.3646 RH3 (18)

MR = exp(-fo) An equation for the equilibrium moisture

(6)
where K and k are both termed drying
constants (h ~x or s ~ ') The value of K for
Equilibrium Moisture Content
content for oats has been approximated
The driving potential for moisture
extraction from grain is a function of the from data (ASAE Yearbook) at T = 25C
ratio of difference between the grain
as

corn has been experimentally determined
(Brooker et al. 1974) as
Kam =0.54 exp (- 5023\
where 0 is in R and K has units of s~ K
Values of the drying constant k for wheat
and barley are (O'Callaghan et al. 1971)
*whea, =2000 exp (-^2)
. lloo
^barley = 139.2 CXp /\~ ~Z
7976\
) (9)
moisture and the equilibrium moisture to
the difference between the initial moisture
(7)
and the equilibrium moisture (Eq. 3).
Thus, the value of the equilibrium
moisture content must be evaluated to use
the above equation. However, the equi
librium moisture content is itself a func
tion of the relative humidity of the air
surrounding the grain. Since the relative
(8) humidity ofthe air changes throughout the
drying period as the clay adsorbs moisture,
the equilibrium moisture and drying
Meqoa(s = 0.0022 + 0.5548 RH -
1.0911 RH2 + 0.8481 RH3 (19)
It might be argued that, due to the tran
sient process, a varying equilibrium
moisture content is unrealistic. It should
be noted, however, that at the start of the
process the final equilibrium moisture con
tent of either the grain, clay or air is
unknown. Also, because the kernels are
separated from one another by the clay,
equilibrium moisture contents are estab
lished quickly.

.14
These equations, as well as ones noted
below, will be used in predicting data for rs
the grains studied. The assumption inher
ent in using these equations, as well as
8
w
12
VISUM. CURVE FIT
ones noted below, is that the predominant
mechanism is diffusion within the kernel, 2SP . 1
UJ
independent of what mechanism is being l-
z
used to dry the air surrounding the grain Q
(e.g. moving air or desiccant). .08 * *
Because the above equations have UJ
C
/
sometimes been found to be unsatisfactory 3
.06
for conventional drying, workers have
CO :/ 0ATA FROM STURTQN t BILANSKI (I860)
proposed empirical drying equations for -*
o
cereal grains. For example, Becker (1959) .04
proposed the following drying rate equa
tion for wheat:

MR = 1 - 24.82 (Dt)W2 + 142.29D;

.02
1 VETTING CURVE FOR CLAY ABOVE
A SATURATED SALT SOLUTION
for(D0,/2 < 0.034 (,U)

or MR = 0.509 exp (-629.68Dt) 0 SO 100 ISO 200 250 300 350 400
for (D/),/2 > 0.034 (ID
TINE (HOURS)
u D
where r = 76.8
n^o exp II H080A1. (12) Figure 1. Wetting curve for sodium bentonite derived by placing the clay above a standard salt
V 6abs ' solution.

174 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 300 time delay) occurs, the moisture pick-up
can be considered reasonably instan
taneous, at least much faster than the
moisture diffusion from the grain. When
the grain and clay are mixed together, there
is very little distance for the moisture to
travel within the clay. Hence, there are not
extensive gradients in moisture content
throughout the clay. Data on moisture
diffusivity in clay as a function of moisture
content, temperature and compaction has
not been found in the literature.
Sturton and Bilanski (1980) noted con
siderable time-lag for full moisture
adsorption (Fig. 1). If the moisture adsorp
tion curve is approximated by a curve of
the form of Eqs. 5 or 6, an adsorption
constant is obtained that is larger than the
drying constant ofthe grains. However, the
size of the sample used by Sturton and
0 .2 .4 .6 .8 1 Bilanski (1980) gave a path length of diffu
RELATIVE PRESSURE (P/Po) sion (0.5-2.0 cm) much greater than that
required for the low mass ratios of grain to
Figure 2. Moisture adsorption curves for sodium bentonite (sodium-montmorillonite). 1. Thix- clay used for the grains.
otron Montmorillonite (Grim 1968)); 2. Orchiston, homoionic sodium montmorillonite
Montmorillonite clays exhibit quite a
(Orchiston (1955)); 3. Mooney, sodium montmorillonite (Mooney et al. (1952));
4. Mooney, Thesis (Mooney (1951)); 5. Mooney, natural montmorillonite (Mooney
significant hysteresis in the adsorption-
etal. (1952)); 6. Keren, sodium montmorillonite (Keren and Shainberg (1979)); desorption curves. Since this paper is con
7. Johansen and Dunning (Grim(1968)); 8. Derived fromgraindrying dataof Sturton sidering only the drying of grain, only the
and Bilanski (1980) and Graham et al. (1983); 9. Illite, Johansen and Dunning (Grim adsorption characterisitics of the clays are
(1968)); 10. Kaolinite, Johansen and Dunning (Grim(1968)); X, Adsorption data from of interest; hence, only these data are used.
Sturton (Sturton and Bilanski (1980)). According to Keren and Shainberg
(1979) the clay adsorbs water on its surface
TABLE I. NOMINAL AND COMPUTED MASS RATIOS (GRAIN:CLAY) in multiple monolayers. Hence, one con
Nominal Nominal Computed tinuous fourth order curve of relative
mass ratio mass ratio mass ratio humidity (RH) vs. moisture (db) was fitted
Runt (wb) (db) (db)
to the experimental data of Keren and
SWHEAT1 1:1 1:1.146 1:1.02*
SWHEAT2 12
Shainberg (1979) for sodium bentonite.
1:2.3 1:1.40
SWHEAT3 1 0.5 1:0.57 1:0.48 RH = 11.12086* - 83.7160A:2 +
SWHEAT4 1 1 1:1.074 1:0.955
SWHEAT5 1 0.5 351.4836X3 - 564.1115A:4 (20)
1:0.578 1:0.615
G WHEAT 1 1 0.5
GWHEAT2 1 1
1:0.578 1:0.455 Moisture adsorption data from other
1:1.146 1:0.935
GWHEAT3 12 1:1.292 1:1.28
workers, including Sturton and Bilanski
SCORN1 1 1 1:1.191 1:1.52$ (1980) were compared to data of Keren and
SCORN2 12 1:1.238 1:2.47 Shainberg (1979) and are shown in Fig. 2.
SCORN3 13 1:1.357 1:5.67
SCORN4 1 1 Considerable variation can be seen to exist
1:1.30 1:1.95
SCORN5 12 1:2.6 1:4.21 among the results of the various workers,
SCORN6 13
GCORN1 1 1
1:3.9 1:5.60 and it is most likely that this variation is
1:1.382 1:1.61
SBARL1 1 1 1:1.076
due to the different composition of the
1:1.29
SBARL2 1 0.5 1:0.598 1:0.77*
clays, even though they are all termed
GBARL1 1 1 1:1.151 1:0.811 sodium bentonite or sodium mont
SOAT1 1 1
SOAT2 1:2
1:1.227 1:1.64* morillonite. Also noted in Fig. 2 are one
1:1.533 1:2.20
SOAT3 1:0.5 1:0.619 1:0.704
investigator's values for moisture adsorp
tRunsstarting with "S" refer towork bySturton etal. (1981). Runs starting with 'G' refer to work by Graham tion isotherms for illite and kaolinite clays,
et al. (1983). which sometimes are "impurities" in ben-
Runs for which predictionbecame unstable. tonites. It can be deduced that the greater
tRuns plotted in Figs. 4-7.
the component of illite and kaolinite, the
less moisture adsorption capability the
Moisture Adsorption by Bentonite exchangeable cation (Watts et al. 1986). clay has. The effect of adjusting the ions of
Bentonite is a naturally-occurring The time for saturation of the mont a given clay can also be seen from Fig. 3.
montmorillonite clay mineral that has morillonite clays is a function of its Curve 3 represents the moisture adsorption
excellent hygroscopic properties. The exchangeable cation but, in general, it isotherms for a natural montmorillonite
wateradsorption capabilities are a function occurs reasonably rapidly (Fig. 1, Watts et whereas curve 5 represents the same clay
ofthe purity ofthe montmorillonite (often al. 1986). Certainly over the small that has been saturated with sodium ions.
considerable kaolinite is present) and a moisture pick-up encountered here, where Curve 8 on Fig. 2 is discussed in the next
function ofthe type (calcium or sodium) of very little swelling(the processcausingthe section.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 175

 .2 tal data obtained are used in this section to
/ ^ evaluate the performance ofthe simulation
EXPERIMENTAL DATA / ? model being presented.
_ B = BARLEY /c On examining the data of Sturton and
.18 Bilanski (1980), or Graham et al. (1983) it
C = CORN Ci
. 0 - OATS was found that the moisture transferred
W - WHEAT between the grain and the clay was not
.16

/ I balanced. The reason for this discrepancy

/ c/ is unknown. However, because the com
puter simulation assumes a moisture bal
/*
/^
77 A /
/^ '
ance, a mass ratio of grain to clay was
computed such that the moisture balance
/^ c C/ 0 '

condition was valid (see Table I). It was

-

/4? 7 */ this computed mass ratio that was used as

imput to the computer program.
-

/
Q) V /
N /
-

/ / Clay Moisture Adsorption Isotherm

/ / / Deduced from the Experiments
All experiments were carried out for
/ /
-

/
#
, <*. / /
about 180 h, well beyond the time when
equilibria of the grains and clay were
A0/ /
/WB
established. It was thought that the
/ hygroscopicity characteristics of the
;
/ wV M
grains were better established than those of
the clay because of the variation of the
/ /
.06 - chemical make-up and adsorbed ions of
the clay. With this assumption, it is possi
PURITY = .78-. 11 RH +. 28 RH2 ble to determine the equilbrium relative
humidity of the air surrounding both the
.04 1 i 1 i 1 . 1 i, 1 i
clay and the grain by substituting the final
.4 5 .6 7 8 9 experimental moisture content ofthe grain
RELATIVE HUMIDITY into the appropriate equation (Eqs. 16-19).
Knowing the equilibrium moisture content
Figure 3. Moisture isothermdatafor sodium bentonite determinedfrom grain drying experiments
of Sturton and Bilanski (1980) and Graham et al. (1983).
of the clay, it is then possible to derive a
moisture adsorption isotherm for the clay
(Fig. 3). The individual data points are
noted for the different grains by their first
letter. A fairly consistent moisture iso
DRYING CORN WITH SODIUM BENTONITE
therm is obtained, although some scatter is
J.25 evident. Also plotted on the graph are lines
of constant "purity", where purity is
2.2 defined as the ratio ofthe relative humidity
predicted from the data of Keren and
O # * # * Shainberg (1979) and a given relative
".IS humidity at the same moisture content. It
UJ can be seen that the grains that equilibrated
x "
| at higher relative humidity (>0.7) follow a
o> . 1 X 0.85 purity line, and those that equili
brated at lower relative humidities (<0.7)
EXPERIMENTAL CORN DRYING DATA
follow a 0.78 purity line. However, since
.05 x EXPERIMENTAL CLAY WETTING DATA
only one clay was being used, one continu
- THEORETICAL PREDICTIONS
ous curve as noted on the graph was deter
i 1 i 1 i_ 1 , 1 . 1 . 1 . 1 . mined for use in computing the drying
0
10 20 30 40 SO 60 70 80 curves for all grains. The equation for that
curve was
TINE (HOURS)
Purity = 0.78 - 0.11 RH + 0.28 RH2 (21)
Figure 4. Comparison of experimental and predicted results of drying corn using sodium ben
tonite. Data from Sturton and Bilanski (1980). Run SCORNI. It is evident that this equation is valid only
for the sodium bentonite that was used in
SIMULATION AND PREDICTION reported in Sturton et al. (1981) and the work of Sturton and Graham. Further
Drying trials with freshly harvested Graham et al. (1983) in which the authors work at the University of Guelph using
wheat, corn, oats and barley were carried fitted their experimental data with expo nominally the same clay of a different
out at the University of Guelph. The nential equations but did not actually mesh size found a slightly lower purity.
experimental method and results are model the drying process. The experimen Thus, it is evident that to use Black Hills

176 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 ticles does occur as moisture is adsorbed
.2 (Grim 1968). Greater confidence is placed
DRYING BARLEY WITH SODIUM BENTONITE on moisture isotherms characterized by
mixtures of grain and clay than by using
clay in isolation above a standard salt
solution.

Prediction Results
Twenty-one experimental drying
u .1 curves were predicted with the initial
UJ
moisture contents of grain and clay and
OC their mass ratios as the only inputs. The
3
r- data represented a variation of initial
U)
S . OS
moisture contents of grain and clay and
EXPERIMENTAL BARLEY DRYING DATA mass ratios of grain to clay. As space did
x EXPERIMENTAL CLAY WETTING DATA not premit inclusion of all, one example of
- THEORETICAL PREDICTIONS the prediction curves of wheat, corn, oats
I I and barley are shown in Figs. 47, respec
10 20 30 40 SO 60 70 80 tively. The drying curves for the wheat
were virtually identical whether Eqs. 6 and
TINE (HOURS) 8 or Eqs. 10-12 were used. The predic
tions for barley used Eqs. 6 and 9. The
Figure 5. Comparison of experimental and predicted results of drying wheat using sodium corn drying curves were obtained using
bentonite. Data from Sturton and Bilanski (1980). SWHEAT1.
Eqs. 5 and 7. The equations used by Flood
et al. (1972) were empirically derived for
.2 air drying, hence, apparently do not apply
DRYING WHEAT WITH SODIUM BENTONITE to adsorption drying, as they predicted a
curve which did not reach equilibrium
moisture content even after 180 h. The
IS prediction of those runs which had a low
clay content relative to the grain became
UJ unstable, probably due to the inaccuracy of
the clay moisture isotherms at high
.1 moisture contents.
One interesting and fairly consistent
observation that can be seen in the experi
mental data (e.g. oats data) and that could
r: .05 not be predicted, is the interim plateau that
EXPERIMENTAL WHEAT DRYING OATA
is reached in the clay adsorption curve
x EXPERIMENTAL CLAY VETTING DATA
between 4 and 30 h. The wetting process of
- THEORETICAL PREDICTIONS
bentonites is quite complex. The initial
1,1,1,1,1,
i. moisture adsorption for single-layer wet
ting of the clay particles follows the BET
10 20 30 40 50 60 70 80 equation. As more water molecules are
TIME (HOURS) added, there is some reorientation of the
clay particles, a process which takes time
Figure 6. Comparison of experimental and predicted results of drying barley using sodium (Grim 1968). This reorganizing ofthe clay
bentonite. Data from Sturton and Bilanski (1980). Run SBARL2.
particles is thought to be responsible for
the hysteresis in the adsorption-desorption
Wyoming bentonite clay will not neces curves. It is postulated that this is what is
workers. This may indicate that there is
sarily yield consistent moisture adsorption more illite (curve 9) or kaolinite (curve 10) happening during this plateau. For the pre
isotherms for the clay. dictions of the grains it is assumed that the
present in the clay used by Sturton and
Instead of using only the moisture iso moisture adsorption is instantaneous;
Bilanski (1980) and Graham et al. (1983)
therms found experimentally, we thought hence, the plateau is not predicted.
than in those of other workers.
it wise to relate all moisture isotherms to a Also noted on Fig. 2 are two points
standard. Since Keren and Shainberg's which were experimentally derived CONCLUSIONS
data satisfies the BET theory at low relative moisture contents at known relative (1) Using the conventional single-ker
humidities, it was felt that these data would humidities for the clay used in the experi nel drying equations and the given
make a reasonable standard. ments. Both differ somewhat from the equilibrium moisture isotherms noted in
The experimentally derived moisture curve derived above. This discrepancy this paper, drying curves for wheat, corn,
adsorption isotherm is plotted as curve 8 may be due to the fact that clay mixed with barley and oats were predicted for one type
on Fig. 2. It is noted that this isotherm is grain cannot expand readily. Even though of sodium bentonite.
somewhat below most curves for sodium expansion is minimal at low moisture con (2) The interim plateau reached in the
montmorillonite determined by other tents, some reorganization ofthe clay par clay adsorption curve between 4 and 30 h

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 177

 .25 KEREN, R. and I. SHAINBERG. 1979. Water
vapour isotherms and heat of immersion of
DRYING OATS WITH SODIUM BENTONITE Na +/Ca+ montmorillonite systems. II.
Mixed systems. Clay Clay Miner. 27(2):
145-151.
MOONEY, R. W. 1951. The adsorption of
water vapour by the clay minerals, kaolinite
and montmorillonite. Ph.D. Thesis, Cornell
MOONEY, R. W., A. G. KEENAN, andL. A.
WOOD. 1952. Adsorption of water vapour
by montmorillonite. I. Heat of desorption
g.i
ui * -
and application of BET theory. J. Am.
Chem. Soc. 74(6): 1367-1370.
O'CALLAGHAN, J. R., D. J. MENZIES and
EXPERIMENTAL OATS DRYING DATA P. M. BAILEY. 1971. Digital simulation of
1.05 x EXPERIMENTAL CLAY VETTING DATA agricultural dryer performance. J. Agric.
- THEORETICAL PREDICTIONS Eng. Res. 16: 223-244.
ORCHISTON, H. D. 1955. Adsorption of
i 1 1 I i I i water vapour. III. Homoionic mont-
20 40 60 80 100 120 morillonites at 25F. Soil Sci. 79: 71-78.
TIME (HOURS) PAGE, G. E. 1949. Factors influencing the
maximum rates of air-drying shelled corn in
Figure 7. Comparison of experimental and predicted results of drying oats using sodium ben thin layers. M.Sc. Thesis, Purdue Univer
tonite. Data from Sturton and Bilanski (1980). Run SOAT1. sity, Lafayette, Ind.
STURTON, S. L. and W. K. BILANSKI.
1980. Indirect drying of cereal grains and
is due to the time needed for a reorientation ARKEMA, andC. W. HALL. 1974. Drying peanuts using a combination of osmotic
of the clay particles following their initial cereal grains, AVI Publishing Co., West- dehydration and a dehydrating medium
single-layer wetting. port, Conn. which is solar dried. Research Report
(3) For studies on grain drying the CRANK, J. 1975. The mathematics of diffu for Agriculture Canada, Contract
hygroscopicity ofthe clay should be deter sion. Clarendon Press, Oxford, U.K. 065Z.01843-9-1918, Ottawa, Ont.
FLOOD, A. S., M. A. SABBAH, D. STURTON, S. L., W K. BILANSKI, and
mined using grains rather than in the loose
MEEKER, and R. M. PEART. 1972. Sim D. R. MENZIES. 1981. Drying of cereal
state above standard chemical solutions.
ulation of a natural air corn-drying system. grains with the desiccant bentonite. Can.
Trans. ASAE (Am. Soc. Agric. Eng.) 15: Agric. Eng. 23: 109-112.
REFERENCES 156-159, 162. VAN REST, D. and G. W. ISAACS. 1966.
ASAE AMERICAN SOCIETY OF AGRI GRAHAM, V. A, W. K BILANSKI, and Exposed layer drying rates of small grains.
CULTURAL ENGINEERS. 1982. ASAE D. R. MENZIES. 1983. Adsorption grain ASAE (Am. Soc. Agric. Eng.) 47:
Yearbook. ASAE, St. Joseph, Mich. drying using bentonite. Trans. ASAE 688-689.
BECKER, H. A. 1959. A study of diffusion in (Am. Soc. Agric. Eng.) 26(5): 1512-1515. WATTS, K. C, W. K. BILANSKI, and D. R.
solids ofarbitrary shape, with application to GRIM, R. E. 1968. Clay Mineralogy. MENZIES. 1986. Comparison of drying
the drying of the wheat kernel. J. Appl. McGraw-Hill, New York. corn using sodium and calcium bentonite.
Polymer Sci. 1:212-226. KEEY, R. B. 1972. Drying principles and prac Can. Agric. Eng. 28: 35-41.
BROOKER, D. B., F. W. BAKKER- tice. Pergamon Press Ltd., Oxford, U.K.

178 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 AUTOMATIC TIMER CONTROL FOR A BATCH-IN-BIN DRYER

R. B. Brown1, L. Otten1, and J. E. Brubaker2

'School ofEngineering, University ofGuelph, Guelph, Ontario NIG 2W1; and 2Ontario Ministry ofAgriculture and
Food, Energy Branch, Toronto, Ontario.
Received 24 April 1986, accepted 11 February 1987

Brown, R. B., L. Otten, and J. E. Brubaker. 1987. Automatic timer control for a batch-in-bin dryer. Can. Agric
Eng. 29: 179-182.

Acomputer-based control system for abatch-in-bin grain dryer was developed and tested with afarm-scale grain dryer
in 1985. Nine drying tests were conducted with wet corn harvested at about 30% initial moisture content (wet basis). The
control system utilized an inexpensive personal computer (Radio Shack TRS-80 Model 100) and an A/D interface unit
designed tocommunicate with the computer through the RS232 port. The control algorithm was based upon simulation of
the drying process, with known initial conditions and parameters. Control to within 0.3 percentage points ofthe target final
moisture content of the corn dried was realized in three of the tests.

INTRODUCTION
Batch-in-bin dryers are commonly used
to dry corn on farmswherethe annualcrop
does not exceed about 500 t. This system
requires a relatively low initial investment,
is flexible and with careful management
can produce high-quality grain. In addi
tion, energyefficiency is potentially excel
lent especially with exhaust heat recovery
(Otten 1985).
A typical drying bin is about 7000 mm
in diameter and has a fully perforated dry
ing floor. A fan and burner unit delivers
heated drying air, usually at 60-100C, to
an underfloor plenum. This drying air
picks up moisture as it passes through a
1000- to 2000-mm depth of wet corn in the
bin, and is exhaustedthrough roof vents. A
level fill of wet grain is dried to an accept
able moisture content and then cooled in
the bin and unloaded to storage. The time
required for one cycle varies from about 12
to 20 h depending upon initial moisture
content, grain depth and drying tem
perature.
The deep bed of grain in the dryer cre
ates two problems with the system. First, a
steep gradient of moisture content exists
from the bottom to the top of the grain bed
at the end of drying. Although grain is
mixed as the dryer is unloading, the dif
ference in moisture content among indi
vidual kernels may be as great as 20 control system.
percentage points. Various schemes to stir
or mix the grain as it dries have been
introduced to eliminate the gradient, and Three approaches to the mathematical
dryer manufacturers supply either auger modelling of deep-bed corn drying were
stirring machines or grain recirculating investigated. The logarithmic model (Sab-
devices.
The second operational problem is one
of control. It is very difficult to monitor the
moisture content of grain in the dryer in
order to determine when to shut off the
burner. The axial moisture gradients men
tioned earlier and also radial fluctuations
resulting from uneven stirring or pockets
of fines make it necessaryto obtain a large
number of samples from the grain bulk.
These samples must be blended and cooled
before a moisture tester can be used to
measure average moisture content. The
procedure must be repeated until the batch
is finished. In practice, dryer operators do
not test the grain but rely upon experience
to time the batch. This approach is satisfac
tory when variations in grain moisture,
depth and ambient conditions are slight.
More typically, the operator acts in such a
way that the corn is always overdried by
2-3 percentage points to be on the safe
side.
Apart from a reduction in grain quality,
deliberate overdrying costs at least $5.00
per tonne of grain marketed at present
prices. The cost is due to additional fuel
costs for drying and to weight loss from
unnecessary moisture removal.
A control system to regulate final aver
age moisture content for a batch-in-bin
dryer is desirable. A closed-loopfeedback
control system for moisture content is not
possible because average moisture content
is not readily measured. An alternative
approach is to use a predictor, or a calcu
lated estimate of the moisture content, to
control the dryer. The objective of this
study was to develop and field-test such a
SIMULATION MODELS
bah et al. 1979; Barre et al. 1971) is the
simplest and fastest to use. It is suited to
near-equilibrium drying since it neglects
sensible heating of grain by the drying air.
Moisture and temperature gradients in the
grain bed are assumed to be logarithmic
functions of timeanddimensionless depth
units. Since batch-in-bin drying uses air
temperatures above 60C, equilibrium is
attained for only a small fraction of the
bed, and the method is inappropriate.
A set of four partial differential equa
tions can be derived to define the changes
in grain moisture and temperature and air
specific humidity and temperature for a
bed of grain (Meiering et al. 1977; Bakker-
Arkema et al. 1974). These may be solved
numerically, but very small increments of
time and depth are necessary for solution
stability. For a deep bed and a high initial
moisture content simulation time can
become excessive.
A similar approach is to perform heat
and mass balances around a thin section of
the bed, and then treat the bed as a stack of
adjacent layers (Thompson et al. 1968).
The thin-layer drying rate is calculated
from an appropriate empirical equation.
The resulting system of equations can be
solved quickly if several simplifying
assumptions are made. For example, ker
nel temperature is assumed to be in equi
librium with the drying air. It was also
assumed that although condensation
occurs in the upper grain layers, rewetting
does not occur due to the relatively short
drying time. This assumption greatly
reduced the computation time for the first
half of the drying period whengraincondi
tionsabove the drying front areessentially
constant.
The model used wasdevelopedfrom the
concepts presented by Thompson (1970).
A time increment of 300 s (5 min) and a
layer thickness of 50 mm (2 inches) was
used for simulation. Psychrometric rela
tionships given by Wilhelm (1976) and
published by the American Society of
Agricultural Engineers (1983, D271.2)
were used to calculatemoist air properties.
The Zeroin search technique reported by
Bakker-Arkema et al. (1974) was used

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 179

with the equation of the wet bulb line to for simulation and a new moisture ratio is exchange would be required. The model
determine the wet bulb temperature when calculated to determine the final moisture approach would also require a measure
condensation occurred in the grain bed. content after that time step. The water ment of ambient temperature at the initia
The specific heat of corn was deter removed is found from the change in tion of cooling since this temperature
mined by the relationship published by moisture content, and the absolute humid would almost certainly differ from the
Otten and Samaan (1980): ity of the drying air is increased by the measurement taken at the start of drying
same amount. some 12-20 h earlier. Due to these diffi
Cp = 1.178 + 0.0627 Mw -
The final drying air temperature was culties it was decided that rather than com
8.7 x 10-4MW2 (1) then calculated from a heat balance for the plicate the control system, a constant
air, grain and water as: cooling interval would be used. The expe-
Equilibrium moisture content of corn
was calculated using the Chung-Pfost (1.006 + 1.775 Ho + Cp/?k)7c - A//(2501 + L- 4.1868 Te)
equation for desorption and the coeffi 1.006 + 1.775//, + Cp/?k
(6)
cients developed by Gustafson and Hall
(1974):
The relative humidity of the air after rience of the previous year indicated that
ln[-/?7rln(RH)] = ln(A) - BMC (2) drying was checked for feasibility of the 2-4 h were required to cool the grain,
Dry matter bulk density of corn was final air state conditions. If the relative depending upon grain depth, final
assumed to be 620 kg/m3. This value, humidity was greater than 100% a con moisture content and drying temperature.
along with airflow rate and incremental densation subroutine was called. A new The operator would be free to change this
bed depth, was used to calculate a dry final temperature and absolute humidity of parameter, and often did, depending upon
grain-to-air mass ratio (Rk) which was nec the exhaust air at saturation was found the aeration capabilities ofthe final storage
essary for a heat balance. The ratio was using the Zeroin algorithm. The difference bin.
used to convert specific heat of grain to an in water removed between the two air Data from the previous tests were com
equivalent value expressed on the basis of states was assigned to the corn by adjusting pared with simulated drying runs for the
air mass rather than grain mass (i.e., kJ/kg the moisture content. same conditions. The mean difference
dry air). Several additions were necessary to between simulated and actual drying time
Latent heat of vaporization for water in complete the drying control program. The was 1.5 h, or about 7% ofthe actual drying
corn was determined from an expression airflow rate-pressure drop relationship of time. Since some moisture removal occurs
developed by Thompson et al. (1968) and Haque et al. (1982) was used to calculate with cooling, drying must stop and cool
converted to SI units the airflow rate from plenum static pres ingcommencebeforethe average moisture
sure and grain depth. A grain stirring sub content is at the target level. It was initially
L = (2502.22 - 2.3867) x
(1 + 4.35 exp (28.25M)) (3)
routine was developed assuming complete assumed that 1.5 percentage points of
mixing and instantaneous stirring at the moisture were removed in cooling, but
The drying temperature for each grain end of each stirring interval. The grain subsequent field tests in 1985 demon
layer was calculated by performing a sen temperature and moisture content of each strated that the actual reduction was about
sible heat balance of the air and grain level in the bed were replaced by the over 3%. In view of inaccuracies in the source
before and after thermal equilibrium. This all average value for the entire bed at the data and assumptions a mean error of 7%
was an intermediate estimate only since it end of each stirring cycle. was considered acceptable for the simula
does not account for moisture evaporation. The dryer used for field trials had been tion model.
The solution for the estimated equilibrium used in the previous year for an investiga
temperature is: tion of exhaust air heat recovery. Since the DRYER CONTROL SYSTEM
heat exchanger was also used in these The simulation model was the basis for a
(1.006 + 1.775 Ho) + Cp/?kG very simplecontrol programto operate the
T = (4) trials, the control program had to accom
(1.006 + 1.775//o) + Cp/?k modate any effects it had on the system. dryer. The control steps were as follows:
During the drying stage, the only signifi 1. Input grain and air parameters
The relative humidity ofthe drying air at
cant effects would be a slight reduction in moisture content, depth, target moisture
the new equilibrium temperature was then
airflow rate and lower fuel consumption content, grain temperature, plenum pres
calculated. If the air was saturated at that
for a given drying temperature. These sure, ambient temperature and humidity,
temperature then no drying waspossiblein drying temperature and duration of
effects would be accounted for in the
that layer. For an unsaturatedair condition,
parameters determined as inputs to the cooling.
the water removal for drying for the spe 2. Read start-up time and energize gas
control program.
cific time increment was calculated from solenoid and drying fan relay.
The effects during the cooling stage
the following thin-layer equation de 3. Simulate drying for input conditions
could not be compensated for easily.
veloped by Thompson et al. (1968) for a
Ideally, time for cooling and the amount of and store required drying time.
temperature range of 60-150C and an moisture removal during cooling would be 4. Monitor clock until elapsed real time
average drying air velocity of 0.10 m/s.
calculated by the simulation model. This equals required time and then close gas
t = a \n(MR) + b[\n(MR)]2 (5) would require knowledge or an assump solenoid to shut off burner.
A new drying curve is required for each tion of inlet air temperature to the dryer. 5. Monitor clock until cooling time has
layer as the drying air temperature Without the heat exchanger, this would elapsed, then open fan relay to stop air
increases. Therefore, an equivalent drying simply be ambient temperature plus a flow.
time for the new curve must be calculated small temperature increase due to the fan. The computer selected for the trials was
to account for the drying that occurred at However, with the exchanger in place, a Radio Shack TRS-80 Model 100. Advan
the lower temperature. This equivalent either real time measurements of inlet tem tagesof this machine were: RS-232-C port
time is then incremented by the time step perature or a very good model for heat for I/O, addressablesystemclock withcal-

180
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 TABLE 1. SUMMARY OF CONDITIONS FOR 1985 CONTROL SYSTEM TESTS
Ambient conditions
Initial Wet Bed
moisture weight depth Temp.
(% WB)
Test (tonnes) (m) (C)
1 29.9 28.1 1.0
2 31.4 29.0 0.9 16
3 27.3 37.5 1.3 11
4 28.0 37.2 1.3 16
5 30.4 35.8 1.2 10
6 31.5 33.2 1.2
7 28.9 32.8 1.6
8 27.5 43.5 1.5
9 27.0 40.0 1.3
endar, battery back-up power pack in case
of power fluctuations, small size, built in
display for data input prompts, and low
cost.
A controller and interface unit was built
to set the gas solenoid valve and dryer fan
relay. Two solid state relays were ad
dressed by setting the two bottom bits of a
control byte and writing the result to the
controller via the RS-232-C port.
FIELD TESTING
The dryer used for the field tests was
located near Ayr, Ontario. It was 7300 mm
in diameter and 6100 mm to the eave with a
full perforated drying floor. A 660-mm
diameter, 7.5-kW axial fan and a natural
gas burner provided heated drying air.
Grain was distributed by a centrifugal
spreader as it was loaded into the bin, and a
six-auger stirring machine traversed the
bin once every 4 h during drying.
An exhaust air heat recovery system was
installed on the dryer utilizing a crossflow
flat plate heat exchanger which was tested
the previous year (Otten 1985).
The first dryer test commenced 16Oct.,
and nine tests were conducted between that
date and 9 Nov. A poor field drydown rate
coupled with persistent rainy weather
made for sporadic testing. The conditions
for the tests are summarized in Table I.
The grain stirrer drive reduction ratio
was changed prior to test 4 since the oper
ator felt that better mixing and more uni
form drying would result from a 2.5-h
stirring cycle. The stirrer was not started
until 6 h of drying time had elapsed, in
order to establish a uniform drying zone
before stirring activity disrupted the air
flow pattern in the bed. Three tests (3, 6
and 9) were conducted without stirring.
The heat exchanger was disconnected for
two tests to evaluate its effect on dryer
performance.
RESULTS
Moisture removal during cooling was
greater than the 1.5 percentage points
anticipated, consequently the corn of the
Drying Grain Duration
RH temperature temperature of cooling
(%) (C) (C) (h) (% WB)
6 80 90 10 2.0
85 no 22 2.5
85 108 11 2.5
50 103 12 3.5
45 109 4 3.5
8 55 88 5 3.5
6 60 107 5 3.5
8 80 108 6 3.5
5 85 92 4 3.0
first three tests was severely overdried
(Table I). It was impossible to determine
the average moisture content at the time
the burner was shut off because there were
radially distributed zones of wet and dry
corn created by the stirring auger paths.
Therefore the third test was unstirred, and
the average moisture content was 17.3%
wet basis compared to a simulated value of
17.0%, a difference of less than 2%. After
cooling the average moisture content was
12.6% indicating that more than 4 percent
age points of moisture were removed with
cooling.
The control program was altered assum
ing a conservative value of 3 percentage
points moisture removal with cooling. The
subsequent two tests yielded a final
moisture content within 0.3 percentage
points of the target level of 15.5% wet
basis. Test 6 indicated that the moisture
removal during cooling was the same with
or without the heat exchanger. The seventh
and ninth tests both produced corn wetter
than desired. In both cases there was a
discrepancy of about 1 percentage point
between the moisture tester readings at the
farm and the oven-dry moisture deter
mination in the laboratory for initial grain
conditions. The tester results were used to
control the dryer, and these had been in
good agreement with the oven-dry method
in the earlier tests. Test 8 produced good
results, and the moisture tester determina
tion of initial moisture content was again
close to the oven-dry method.
The grain stirring augers did not do an
adequatejob of mixingthe grain, although
there was some reduction of moisture gra
dient in the dried grain. The delayed stir
ring scheme (Test 4 and subsequent tests)
improved the situation, but radial zones of
wet grain were encountered between adja
cent auger paths. The solution to this prob
lem would be more augers, or variable-
path stirring augers. The overall energy
efficiencyof drying and the accuracyofthe
simulation model did not appear to be
affected by stirring strategy.
Specific energy consumption averaged
2.45 MJ/kg of water removed for the dryer
with the heat exchanger. Without the heat
Final exchanger, specific energy consumption
moisture
was 2.98 MJ/kg water removed, an
increase of almost 22%.
11.9
12.2 DISCUSSION
12.6 Once the amount of drying that occurred
15.2
with cooling was known, the control sys
15.8
11.9
tem worked very well. A target moisture
16.4 set 3 percentage points above the desired
15.3 final value (i.e., 18.5% for 15.5% final
16.5 average) gave good results. For other dry
ers this parameter may vary, but it would
be determined as part of the tuning pro
cedure. Final moisture content of cool
grain was controlled to within 0.3 per
centage points for three of the test runs. If
the initial moisture content of tests 7 and 9
had been more accurately determined, it is
probable that the final moisture contents
would have been within the same range. It
is doubtful that greater accuracy could be
maintained consistently because commer
cial moisture testers have an error range of
the same magnitude.
Level filling and uniform mixing by the
stirring device was assumed in the model
and should be realized in the actual sys
tem. An implicit assumption in the control
scheme is that the ambient absolute
humidity is constant throughout drying.
Although ambient temperature and relative
humidity can vary significantly in any 10-
to 20-h period, it is unlikely that the
humidity ratio will deviate greatly over the
same period. This drying method is quite
insensitive to ambient conditions in any
event due to the large temperature increase
with the burner.
The drying temperatures used in the
tests were higher than normal for deep-bed
batch drying because the dryer was oper
ated for maximum throughput. For a tem
perature range of 60-70C moisture
gradients within the bed would be reduced
and a more uniform product moisture
could be maintained.
CONCLUSIONS
The main conclusions drawn from the
study are as follows:
1. The simulation model allowed an
accurate prediction of the time required to
dry corn to a specific average moisture
content.
2. With accurate determination of grain
and system operating conditions, the cal
culated drying time can be used to control
a batch-in-bin dryer and regulate the final
grain moisture.
3. Moisture removal during the cooling
stage averaged 3 percentage points for the
particular combinations of temperature
and airflow rates in this study.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 181

 4. The small personal computer and Wellington Engineering Ltd., Guelph, specific heat of grain (KJ/kg K).
interface controller comprised an inexpen Ontario, 31 January. G0 = initial grain temperature in grain
sive and effective control unit. OTTEN, L. and G. SAMAAN. 1980. Deter layer (C).
mination of the specific heat of agricultural Ho = initial absolute humidity of air for
materials: Part II experimental results. Can.
ACKNOWLEDGMENTS grain layer (kg water/kg dry air).
The research project was jointly funded by Agric. Eng. 22(1): 25-27.
Hf = final absolute humidity of air for
the Ontario Ministry of Agriculture and Food SABBAH, M. A., H. M. KEENER, and G. E.
MEYER. 1979. Simulation of solar drying of grain layer (kg water/kg dry air).
and the Ontario Ministry of Energy. We are
shelled corn using the logarithmic model. AH = H f //o.
grateful to Robert and Jeff Hall and to Bruce
Ford for their cooperation and assistance during Trans. ASAE (Am. Soc. Agric. Eng.) 22: L = latent heat of vaporization for corn
the drying tests. 637-643. (kJ/kg water).
THOMPSON, T. L. 1970. Simulation for M grain moisture content, decimal
=
optimal grain dryer design. Trans. ASAE dry basis.
REFERENCES
(Am. Soc. Agric. Eng.) 13: 844-848. Mc = equilibrium moisture content,
AMERICAN SOCIETY OF AGRI
THOMPSON, T. L., R. M. PEART, andG. H. decimal dry basis.
CULTURAL ENGINEERS. 1983. Agri
FOSTER. 1968. Mathematical simulation of
cultural Engineers Yearbook. ASAE, St. Mo = initial moisture content, decimal
corn drying a new model. Trans. ASAE
Joseph, Mich. dry basis.
(Am. Soc. Agric. Eng.) 11: 582-586.
BAKKER-ARKEMA, F. W., L. E. LEREW, Mw = grain moisture content, percent
WILHELM, L. R. 1976. Numerical calculation
5. F. DeBOER, and M. G. ROTH. 1974. wet basis.
of psychrometric properties in SI units.
Grain dryer simulation. Mich. State Univ.
Trans. ASAE (Am. Soc. Agric. Eng.) 19: M - Mc
Agric. Exp. Sta., Res. Rep. 224. MR =
318-325.
BARRE, J. H., C. R. BAUGHMAN, and Mo ~ Mc'
M. Y. HAMDY. 1971. Application of the R = gas constant 1.987 Btu/lb mole
logarithmic model to cross-flow deep-bed LIST OF SYMBOLS R.
grain drying. Trans. ASAE (Am. Soc.
a, b, coefficients in thin layer drying RH relative humidity of air, decimal,
Agric. Eng.) 14: 1061-1064.
GUSTAFSON, R. J. and G. E. HALL. 1974. equation *k dry grain to air mass ratio (kg dry
Equilibrium moisture content of shelled corn a = -1.706 + 0.008787 grain/kg air).
from 50 to 155F. Trans. ASAE (Am. Soc. b = 148.6 exp(-0.05947). T air or grain temperature (C).
Agric. Eng.) 17: 120-124. A,BfC,D}E,F, constants used in Tc intermediate equilibrium tempera
HAQUE, E., Y. N. AHMED, and C. W. Chung-Pfost equation for equi ture (C)
DEYOE. 1982. Static pressure drop in a fixed librium moisture content (devel Tf final equilibrium temperature for
bed of grain as affected by grain moisture oped in British Engineering grain layer (C)
content. Trans ASAE (Am. Soc. Agric. Units). To initial air temperature for grain
Eng.) 24: 1095-1098. layer (C)
MEIERING, A. G., T. B. DAYNARD, R.
A = exp[(C + DTT)(RTr)].
B = (E + FTr)(RTr). Tr grain temperature used in Chung-
BROWN, and L. OTTEN. 1977. Drier per
formance and energy use in corn drying. C = 1.544 x 10-2. Pfost equation (R)
Can. Agric. Eng. 19: 49-54. D = -1.383 x 105. t time (h)
OTTEN, L. 1985. Exhaust heat recovery in a E = 3.211 x 10"3.
batch-in-bin dryer. Report to OMAF by F = 2.069 x 10"5.

182 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 DISTRIBUTION OF FOREIGN MATERIAL IN CANOLA BINS FILLED USING A
SPREADER OR SPOUT

D. S. Jayas1, S. Sokhansanj2, E. B. Moysey2, and E. M. Barber2

'Department ofAgricultural Engineering, University of Manitoba, Winnipeg, Man. R3T2N2; and department of
Agricultural Engineering, University of Saskatchewan, Saskatoon, Sask. S7N 0W0
Received 28 August 1986, accepted 16March 1987

Jayas, D. S., S. Sokhansanj, E. B. Moysey,and E. M. Barber. 1987. Distribution of foreign material in canola bins
filled using a spreaderor spout. Can. Agric. Eng. 29: 183-188.

The distribution offoreign material (chaff and fines) was studied in a4.6-m-diameter bin filled with Tobin canola. The
bin was filled twice using a central spout and once using a conical spreader. The distribution ofchaff and fines can be
described using a quadratic equation with radial coordinates. The use ofa conical spreader to fill the bin with canola
increased the uniformity of chaff and fines distribution only slightly. Contrary to general belief, fines were not highly
concentrated in the center of the bin for either ofthe fill methods. The variation ofchaff and fines with height was
insignificant.

INTRODUCTION
The drying of farm crops is an integral
part of agricultural production systems.
The current trend is to dry farm crops,
including canola, using near-ambient tem
perature air. Aeration is the most com
monly used method for cooling stored
grains and oilseeds. In both the near-
ambient drying and the aeration systems,
air in varying quantities, 10-30 L/(s m3)
for near-ambient drying and 1-2 L/(s m3)
for aeration systems (Friesen et al. 1984),
is forced through a perforated area and is
then distributed throughout the grain
mass. The success ofthe systems is highly
dependent on uniformity of the airflow
within the stored mass. The uniformity of
airflow is affected by the configuration of
the perforated area used for introduction of
the air; bin geometry; resistance to airflow
of the product; and the amount, size and
distribution of foreign material in the
stored bulk. If a clean product at known
moisture content is used to fill the storage
bins, then theeffectofthe configuration of
the perforated area and the bin geometry
can be predicted reasonably well using the
mathematical models of several inves
tigators (Brooker 1969; Marchant 1976;
Segerlind 1982; Singh(Jayas)and Sokhan
sanj 1985; Miketinac et al. 1986).
The effect of the amount, size and dis
tribution offoreign material onuniformity
of airflow can also be incorporated in the
mathematical models if the distribution of
foreign materials in storage bins can be
estimated. Chang etal. (1981, 1983, 1984)
studied the distribution of foreign material
incorn and wheat bins. Toourknowledge,
studies on thedistribution of foreign mate
rials in other stored crops have not been
reported in the literature.
Canola is an economically important
oilseed crop for Canada. In 1984, about
3.2 million tonnes of canola worth 1.2
billion dollars were produced in Canada
(Anonymous 1984). Depending on the
weather conditions during harvest, drying
of canola may be a necessity. High- tem
perature air drying is undesirable for can
ola because of the fear of contamination
from burnt kernels, and nonuniform air
flow distribution in near-ambient drying
systems can result in localized regions of
wet seeds causing mouldy or heated ker
nels. Both the burnt and heated kernels
complicate refining ofthe canola oil. As a
first step in an attempt to predict the uni
formity of airflow within stored canola,
this study was undertaken to investigate
the distribution patterns of foreign mate
rial (chaff and fines) in canola bins as
affected by the method used to fill the bin.
Based on average particle diameter of
Tobin canola, chaff was defined as over
flow of No. 12 mesh US Standard sieve
and fines as underflow of No. 16 mesh US
Standard sieve.
MATERIALS AND METHODS
Sampling for Foreign Material
Distribution
The amount, size and distribution of
foreign material was studied in a farm bin,
4.6 m in diameter and 6.89 m in overall
height. The bin was erected on a farm near
Saskatoon. The perforated floor was 0.48
m above the concrete floor and the bin roof
had a 26.7 slope. The effective vertical
height for storage of canola was 5.26 m.
Canola, variety Tobin {Brassica camp-
estris), harvested during the fall of 1984
was used to fill the bin. Canola was first
swathed and then combined at moisture
contents of about 6-10%. The low
moisture contents were due to hot dry probe to sample the bin at grid points
weather at harvest. The amount of foreign spaced 300 mm radially and 280 mm ver
material in canola depends on many fac tically. This method of sampling was not
tors such as moisture content ofthe plants,
combine setting, presence of weeds and
atmospheric conditions at the time of com
bining. The amount of foreign material in
the harvested canola was difficult to con
trol. Therefore, instead of keeping con
stant foreign material at harvest, it was
decided to sample each truck load for for
eign material. Trucks were sampledat the
time of unloading. The dropping streamof
canola from the truck was collected every
120 sec. The canola stream was cut by
moving a container in one direction per
pendicularto the falling stream.The speed
of the container was controlled manually
and was assumed to be nearly constant.
The time to empty the truck ranged from
20 to 30 min depending on the setting of
the unloading gate. The samples collected
for an individualtruck were mixed to give
a representative sample for that truck.
The bin was filled on two occasions
using a central spout with a drop heightof
7.0 m (measured from the perforated floor
to the outlet of the unloading screw con
veyor). Additional canolawasaddedusing
a conical spreader to the partially filled
bin. The dropping height for the conical
spreader filled bin was 3.0 m (measured
from apex of the inverted bottom cone to
the leveled canola surface). Since the bin
was being monitored for another experi
ment, the canola was added on top of the
canola already present in the bin when
using the conical spreader. The arrange
ment of various components ofthe unload
ing system is shown in Fig. 1.
Stephens and Foster (1976, 1978) and
Chang et al. (1981,1983,1984) studied the
distribution of fines in corn, wheat and
sorghum. They used a multicompartment

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 183

 Figure 1. The schematic ofthe arrangement ofvarious components used to fill the bins.

used in the present study for several rea sampling and was reintroduced after
sons. First, it was decided that the opening unloading the next truck. Four truck loads
through which the sample enters the probe were used for the first bin fill and five truck
is small and could prevent entering of the loads were used for the second bin fill.
chaff due to bridging. Second, if a probe The profile after filling the bin usingthe
with a large opening and diameter is used conical spreaderwas not a conebuthadthe
it disturbs the sampling area during inser shape of a doughnut. The sampling grid
tion; also, the difficulty in inserting the for the spreader-filled bin is shown in Fig.
probe increases significantly with probe 3. In preliminary tests, filling with the
diameter. From studies of Stephens and conical spreader did not give an axisym-
Foster (1976,1978) and Chang etal. (1981, metric fill, rather canola sloped in a heap
1983, 1984) it can be assumed that the from one side ofthe bin wall. The problem
distribution of foreign material is sym was in the design of the spreader. The
metrical about the vertical axis of the bin opening between the two cones was too
when bins are filled using spreaders or a wide. Therefore, a slight angle from the
centralspout. Therefore, sampling of a bin vertical ofthe dropping canola guided it to
alongone radiuswas considered sufficient one side. The spreader was modified so
instead of along the bin diameter. The that the top cone was nearly full during
method of sampling a bin along its radius emptying. Also, thehanging portion ofthe
is described below. This method could be dropping-spout was tied so that it dis
extendedto samplealong the bin diameter. charged vertically downward. The modi
The first truck load (about 81) of canola fications resulted in axisymmetric dis
was transferred into the bin using the tribution of foreign material. Many
unloading system shown in Fig. 1. A farmers may not notice this problem and
plywood platform (250mmwide and2500 the conical spreader would result in bins
mm long) was hung from an inside ladder having all the chaff accumulated on one
in the bin. The platform was parallel and side.
close to the canola profile in the bin but
was not touching the canola surface. This Analysis of Samples for Foreign
arrangement allowed easy sampling of Material
canola at all sampling locations spaced Collected samples were subdivided
372 mm apart along the bin radius. The using a Boerner divider into samples of
Radius (r),m 250 g each. The 250-g sample was frac
samples were collected using a circular
sampler 100 mm in diameter and 145 mm Figure 2. Sampling points for spout-filled tionated into three components using two
long.The sampler wasinserted atthepoint bins. Tylerwoven wiremesh sieves (No. 12 and
of samplingand then wascoveredusingan 16). The opening of No. 12 sieve was 1.70
aluminum sheet at the bottom and lifted. 600 g. The samples near the bin wall mm and of No. 16 was 1.18 mm. The
The disturbance caused by sampling at one weighed less because of the presence of sieves were 51mm in depth and 203 mm in
point was not transferred to the adjacent higheramounts of lighter chaff. As shown diameter. The samples were shaken for 5
sampling point. The amount of sample in Fig. 2, the grid of sampling points in min using a Ro-Tap shaker. The shaking
collected at each point was about 800 g this method followed the profile of the time was determined based on preliminary
except nearthebinwall where itwas about canola. The platform was removed after tests in which material on sieves was

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

184

0 I
Radius (r), m
Figure 3. Sampling points for a conical
spreader-filled bin.
recorded at 2-min intervals. The dif
ference between amounts of material
retained on the various sieves after 4 and 6
min was negligible. Hence, 5 min was
arbitrarily chosen as the shaking time.
The material retained on the No. 12
sieve was defined as chaff, the underflow
of No. 12 and overflow of No. 16 was
defined as canola and the material that
passed through No. 16and was retained on
the pan was defined as fines. Canola at this
point contained a small amount of needle
likeparticles of plantresidue. Theseparti
cles wereseparated as overflow using rid
dle screen No. 000 with a feed rate set at 3
on a Carter Dockage Tester (Simon Day
Ltd., Winnipeg, Man.). This small frac
tion was added to the chaff because it
looked similar. The masses of fines, can
ola and chaff were recorded and were used
to give the distribution of foreign material
in thebin as affectedby the method used to
fill the bin.
Normalization of Foreign Materials
The amounts of fines and chaff deter
mined at seven locations along the radius
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
0-5 10 1-5 20 2-5
Radial Distance, (r), m
Figure 4. Distribution ofchaff along the radius ofabin which was filled axisymmetrically using a
central spout. The radial distances were measured from the bin center. The numbers 1to
4 are for first bin fill and 5 to 9 are for second bin fill.
2-5 -1 -5
-2 -6
3 -7
20 4
8
9
II 1-5
a>
10
o
e
!o-5
00
00 0-5 10 I-5 20 2-5
Radial Distance, (r), m
Figure 5. Distribution offines along the radius ofabin which was filled axisymmetrically using a
central spout. The radial distances were measured from the bin center. The numbers 1to
4 are for first bin fill and 5 to 9 are for second bin fill.
of the bin were used to estimate the fines in the mass of canola for spout-filled
weighted average of chaff and fines for bins are shown in Figs. 4 and 5, respec
each truck load. The volumes of canola in tively. The lines were drawn through the
the bin represented by each sample were mean values of chaff and fines. The means
taken into account when means of chaff were calculated from triplicate samples
and fines were estimated. collected 372 mm apart (radially), except
The individual amounts of chaff and for the sampling location, farthest from the
fines were divided by the mean values of center, where only duplicates were used.
chaff and fines to determine the nor The quantity of chaff and fines were nor
malized chaff and fines. malized by dividing the measured value at
each point by the mass averagequantity at
RESULTS AND DISCUSSION
the same level. The plottedlinesrepresent
Distribution in Spout-filled Bins two bin fills. The lines numbered 1 to 4 are
The radial distributions of chaff and for first bin fill and 5 to 9 are for second bin
185
 to 0.6-0.8 m away from the center. The
2.5r
distribution of chaff in spout-filled canola
a-Spreader fill
bins can be described by a quadratic equa

ormalized P-bai
| 1.5
2.0 -

-
a-Spout fill tion:

Cs/Cs = 0.722 - 0.740 r + 0.535 r2

(correlation coefficient = 0.98)

where Cs = mean chaff content for spout-

(I)

o
filled bin; Cs = chaff at radius r for spout-
filled bin; and r = location along bin
v radius measured from central axis (m).
k ^
Chang et al. (1981, 1983) have reported
very high concentrations of fines around
the center of the bins filled with corn and
wheat. We did not observe very high con
centration of fines near the center of the
o.qL X J_ bin. Rather, we observed that the con
centration of fines in canola bins near the
0.0 0.5 1.0 1.5 2.0 2.5
bin center was almost equal to that near the
Radial Distance,(r) , m sides ofthe bin. The minimum fines were
Figure 6. Distribution ofaverage chaff and best fit curves for spout- and spreader-filled bins. The found about midway between the center
radial distances were measured from the bin center. and the bin wall. The canola particles are
very small, and hence distinction between
canola and fines is not as markedly defined
2.5r- as in the studies of Chang et al. (1981,
a-Spreader fill 1983) with corn and wheat. The distribu
a-Spout fill tion of fines in spout-filled canola bins can
2.0 mathematically be given by:
FJFS = 1.093 - 0.474 r + 0.241 r2 (2)
(correlation coefficient = 0.88)
1.5
where Fs = mean fines for spout-filled
0) bins; and Fs = fines at radius r for spout-
=5 ,0 w^r-^-T--^irr: filled bins.

E
o Distribution in Spreader-filled Bins
z 0.5 Normalized distribution of chaff and
fines for spreader fill are given in Figs. 8
and 9, respectively. The lines were drawn
o.oL through the mean values of chaff and
0.0 0.5 1.0 1.5 2.0 2.5 fines. The means were calculated from
triplicate samples collected 372 mm apart
Radial Distance ,(r), m
(radially), except for the sampling loca
Figure 7. Distribution ofaverage fines and best fit curves for spout- and spreader-filled bins. The tion, farthest from the center, where only
radial distances were measured from the bin center. duplicateswere used. Linesnumbered 1,2
and 3 represent the samples taken after
fill. The variation among replicates for fill can be given. Most of the chaff was transferring first, second and third truck
each location was less than 5% for 80% of concentrated near the wall ofthe bin (Fig. loads of canola. The detailed data are
the samplesof chaff, and for all samplesof 6), and minimum chaff occurred between given elsewhere (Jayas 1987). The varia
fines. The detailed experimental data are 0.6 and 0.8 m from the center of the bin. tion among subsamples was less than 5%
givenin Jayas(1987). Asadditional canola This trend was also confirmed by visually for most samples of chaff and for all sam
was added in each filling, the drop height observing the canola profileafter the addi ples of fines. In the case of spreader fill, a
decreased for successive additions. The tion of each load. The cleanest spot was change in drop height had no significant
effect of drop height on radial distribution not the center of the bin but rather it was effect on radial distribution of chaff and
of either chaff or fines is negligible. the annular area between 1.2 and 1.6 m fines. The mean radial distributions of
Hence, the chaff and fines were averaged diameter of the bin. It was expected that chaff and fines for spreader fill are shown
for all samples at each sampling point minimum chaff should occur in the center in Figs. 6 and 7, respectively. The radial
along the bin radius. of the bin. During filling using the central distribution of chaff in spreader-filled bins
The mean values for chaff and fines are spout, particles larger than canola have a is described by:
shown in Figs. 6 and 7, respectively. In tendency to slide along the slope, but some CJCC = 0.960 - 0.866 r + 0.520 r2 (3)
calculating mean values, data points of these particlesdid not have enough time (correlation coefficient = 0.94)
marked "5" were not used. No explana and were trapped in the falling stream.
tion for such a distinct behavior in trans This trapping was probably the main rea where Cc = mean chaff content for
ferring the first loadduring the second bin son for higherchaff at the centercompared spreader-filled bins; and Cc = chaffcon-

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

186
 more uniform distribution of fines in other
2-5r
crops (Chang et al. 1981, 1983).
1

_ 2 Utility of the Results

20 3 To use Eqs. 1-4 for predicting radial
/
/ distribution of chaff and fines a represen
/ , tative sample from a bin must be frac

a

if tionated to find chaff and fines means for
that bin. By substituting values for radial
-

/? coordinates, mean chaff and mean fines in

^ 10 " ><
/'' an appropriate equation, the distribution of
-'/'' foreign material in the bins can be pre
dicted. In the event a representative sam
---'/'
"^* "~ "* ** ** -. ^v ^Z ../ ple cannot be obtained and analyzed, an
|o-5h
estimate of total foreign material in canola
can be used. The estimated foreign mate
rial can be divided by 2 to obtain chaff and
I 1 1 i 1 . 1 . 1 fines means.
00
Equations 1, 2, 3 and 4 are developed
00 0-5 10 1-5 20 2-5
from experimental data for 4.6-m-diame-
Radial Distance, (r), m ter bins. The use ofthe equations to predict
distribution of foreign material should be
Figure8. Distribution of chaffalongthe radiusof a bin whichwas filled axisymmetrically using a
limited to the bins having diameters of
grain spreader. The radial distances were measured from the bin center.
about 4.6 m. The main reason to limit this
study to a 4.6-m-diameter bin was its
2-5 availability and its commonness on most
farms in western Canada. An extensive
1
study with different diameter bins would
2 require substantial cooperation from the
20- 3 farmers and a significant amount of finan
cial resources. If funds become available a
general equation should be developed for
1-5 prediction of chaff and fines not only in
a>
canola bins but also in bins filled with
other cereals and oilseeds.

| 10
CONCLUSIONS
o
The distribution of chaff and fines in a
z
0-5 4.6-m-diameter bin can be modelled using
quadratic equations relating the fraction of
chaff or fines to the radial coordinate.
Contrary to general belief, fines were not
00 1
highly concentrated in the center of the bin
00 0-5 10 1-5 20 2-5 for either of the fill methods. The use of a

Radial Distance, (r), m grain spreader concentrated more fines

Figure 9. Distribution of fines along radius of a bin which was filled axisymmetrically using a
grain spreader. The radial distances were measured from the bin center.
tent at radius r for spreader-filled bins and
radial distribution of fines in spreader-
filled bins is given by
FJFK = 1.269 - 0.466 r + 0.166 r2
(correlation coefficient = 0.78)
where Fc = mean fines content for
spreader-filled bins; and Fc = fines con
tent at radius r for spreader-filled bins.
The effect ofthe conical spreader on the
distribution of chaff is to increase chaff
concentration at the center and reduce it
near the bin wall (Fig. 6), as compared to
spout-filled bins. The main reason for this
shift is the doughnut-shaped profile which
resulted during bin filling using the conical
spreader. The chaff did not have enough
energy to cross the ridge formed and was
(4)
trapped in the center portion of the bin.
The fines are more concentrated in the
center when a conical spreader is used to
fill the bins (Fig. 7). This also happens
because of trapping of fines in the central
portion of the doughnut-shaped profile.
These observations contradict the man
ufacturers claim that use of their spreader
will result in better distribution of fines
than spout fill. Some spreaders do result in
near the center and resulted in only slightly
more uniform distribution of fines than use
of spout fill. We experienced that unsuper
vised use ofthe spreader may cause filling
of the bin towards one side and result in
most of the chaff being concentrated
towards the opposite side.
REFERENCES
ANONYMOUS. 1984. Canadian grain indus
try statistical handbook - 1984. Can. Grains
Council, Winnipeg, Man.
BROOKER, D. B. 1969. Computing air pres
sures and velocity distribution when air
flows through a porous medium and non
linear velocity pressure relationships exist.
Trans. ASAE (Am. Soc. Agric. Eng.) 12(1):
118-120.
CHANG, C. S., L. E. SHACKELFORD, F. S.
LAI, C. R. MARTIN, and B. S. MILLER.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 187

 1981. Bulk properties of corn as affected by
multiple-point grain spreaders. Trans.
ASAE (Am. Soc. Agric. Eng.) 24(6):
1632-1636.
CHANG, C. S., H. H. CONVERSE, andC. R.
MARTIN. 1983. Bulk properties of grains as
affected by self-propelled rotational type
grain spreaders. Trans. ASAE (Am. Soc.
Agric. Eng.) 26(5): 1543-1550.
CHANG,C. S., H.H.CONVERSE, and F.S.
LAI. 1984. Distribution of fines and bulk
density of corn as affected by choke-flow
and dropping height. ASAE Paper No.
84-3531. Am. Soc. Agric. Eng., St. Joseph,
Mich.
FRIESEN, O. H., H. P. HARMS, and D. N.
HUMINICKI. 1984. Movement of natural
188
air through grain. Publ. No. Agdex 732-1.
Manitoba Agriculture, Winnipeg, Man.
JAYAS, D. S. 1987. Resistance of bulk canola
oilseed to airflow. Unpublished Ph.D. The-
sis. University of Saskatchewan, Saskatoon,
Sask.
MARCHANT, J. A. 1976. The prediction of
airflows in crop drying systems by the finite
element method. J. Agric. Eng. Res. 21:
417-429.
MIKETINAC, M. J., S. SOKHANSANJ, and
D. S. JAYAS. 1986. Graphical analysis of
airflow distribution in grain bins using finite
element method. Can. Agric. Eng. 28(1):
23-30.
SEGERLIND, L. J. 1982. Solving the non-
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
linear airflow equation. ASAE Paper No.
82-3017. Am. Soc. Agric. Eng., St. Joseph,
Mich.
SINGH (JAYAS), D. and S. SOKHANSANJ.
1985. Transient field problems: finite ele-
ment formulation. Proc. Can. Soc. of Civil
Eng., Saskatoon, Sask. pp. 425-441.
STEPHENS, L. E. and G. H. FOSTER. 1976.
Grain bulk properties as affected by mechan-
ical grain spreaders. Trans. ASAE (Am.
Soc. Agric. Eng.) 19(2): 354-358, 363.
STEPHENS, L. E. andG. H. FOSTER. 1978.
Bulk properties of wheat and grain sorghum
as affected by mechanical grain spreaders.
Trans. ASAE (Am. Soc. Agric. Eng.) 21(2):
1217-1218, 1221.
 THE EFFECT OF AIRFLOW DIRECTION ON THE RESISTANCE OF CANOLA
(RAPESEED) TO AIRFLOW

D. S. Jayas1, S. Sokhansanj2, E. B. Moysey2 and E. M. Barber2

'Department of Agricultural Engineering, University of Manitoba, Winnipeg, Man. R3T 2N2; and 2Department of
Agricultural Engineering, University of Saskatchewan, Saskatoon, Sask. S7N 0W0
Received 28 August 1986, accepted 6 March 1987

Jayas, D. S., S. Sokhansanj, E. B. Moysey, and E. M. Barber. 1987. The effect of airflow direction on the
resistance of canola (rapeseed) to airflow. Can. Agric. Eng. 29: 189-192.

The resistances to airflow of Tobin and Westar canola, of foreign material usually found in harvested canola, and of
several mixtures of Tobin and foreign material were measured experimentally for two airflow directions. The resistance to
airflow of individual samples for the horizontal airflow direction was 0.5-0.7 the resistance to airflow for the vertical
direction. From a practical point of view, the resistance for horizontal airflow direction can be estimated as 0.6 the
resistance for vertical airflow direction.

INTRODUCTION of mixtures of canola and foreign mate of 720 mm. A square (230 mm x 230
Resistance to airflow of various agri rials. mm) airflow straightener made from steel
cultural products has been studied for at wool was installed in the pyramidical sec
least 55 yr by many investigators including MATERIALS AND METHODS tion to diffuse the air uniformly. The air
Stirniman et al. (1931), Shedd (1951), Bar- Experimental Method was then forced through the canola col
rowman and Boyce (1966), Bakker- The effect of the direction of airflow on umn. Two possible directions of airflow
Arkema et al. (1969), Agrawal and Chand resistance to airflow was studied in a cubic (horizontal and vertical) were used. The
(1974), Bern and Charity (1975) and box (Fig. 1). Compressed air from a cen box was always filled from the top as
Abrams and Fish (1978). (An extensive list tral compressor was regulated by a ball would be the case in bins filled on the
of these studies is given in Jayas (1987).) valve and a pressure regulator. The air was farm. The direction of airflow was
In most of these studies on resistance to filtered before entering the pressure reg changed by rotating the air delivery sys
airflow of agricultural products it has been ulator. The air was then forced through a tem. The pressure drop, to an accuracy of
assumed that resistance to airflow is equal calibrated orifice plate. The measured vol 0.25 Pa (0.025 mm of H20), was mea
in all directions. Therefore, the reported ume of air was introduced into a pyra- sured across 250 mm of a canola column in
results were measured only for the vertical midical air plenum having an apex height both directions. The study was conducted
airflow direction. Many mathematical
models (Brooker 1969; Marchant 1976;
Haque et al. 1981) have been developed to
predict pressure patterns in grain bins.
.--<r
These models are also based on isotropic
behavior of agricultural products. Lamond
and Smith (1982) pointed out the pos 250
AIR
sibility of different resistances to airflow
based on airflow direction since most ker
nels (such as wheat and barley) lie with
their major axes horizontal. Recent inves
tigations by Kumar and Muir (1986) con
firmed the variability of resistance to
airflow with airflow direction for wheat
and barley. Canola seeds are spherical in
shape, hence the effect of direction of air
flow on resistance to airflow of clean can
AIR STRAIGHTENER 720
ola seeds might be expected to be
negligible. The resistance to airflow of
foreign material usually found in har
vested canola is expected to be different
for different airflow directions. To our
knowledge, no study has been reported on
the effect of direction of airflow on resis
ALL DIMENSIONS ARE IN MILLIMETERS
tance to airflow of canola, of foreign mate
rial usually found in canola and of
mixtures of canola and foreign materials.
The objectives of this study were to deter
mine the effect of direction of airflow on AIR
resistance to airflow of canola, of foreign Figure 1. Elevation view of the apparatus used to study the effect of direction of airflow on
material usually harvested with canola and resistance of various samples. The apparatus was supported on a light metal frame.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 189

for an airflow range of 0.0158-0.1709 m3/ TABLE I. THE PHYSICAL PROPERTIES AND THE COEFFICIENTS OF EQ. 1 FOR VARIOUS
SAMPLES AT A MOISTURE CONTENT OF 7.0 0.5%, FOR AN AIRFLOW RANGE OF
(sec-m2).
0.0158-0.1709 m3/(secm2)

Method of Filling Sample composition Physical properties Coefficients of Eq. 1

The cubic box in these tests was filled Bulk

using two methods (spout and sprinkle). In Method Direction density

Canola Chaff Fines of of (kg/ Porosity A
spout fill, which produces a very loose fill, variety fill airflow m-1) xlO6 B R2
(%) (%) (%)
canola was dropped into a funnel and then
Tobint 0 0 Spout Horizontal 694.4 38.9 494.9 0.8072 0.999
allowed to flow with nearly zero height of Tobin 0 0 Spout Vertical 694.4 38.9 283.6 0.8563 0.998
fall. Tobin 0 0 Sprinkle Horizontal 747.9 34.2 249.5 0.8326 0.997
In sprinkle fill, samples were dropped Tobin 0 0 Sprinkle Vertical 747.9 34.2 171.7 0.8616 0.999
about 200 mm into a U.S. standard sieve Westar 0 0 Spout Horizontal 677.4 38.7 604.8 0.8101 0.997
with9.5-mm x 9.5-mm square apertures. Westar 0 0 Spout Vertical 677.4 38.7 380.0 0.8491 0.998

The material then was dropped approx Westar 0 0 Sprinkle Horizontal 725.0 34.4 344.6 0.8151 0.998

imately another 800 mm in a rain-like Westar 0 0 Sprinkle Vertical 725.0 34.4 221.8 0.8608 0.997

manner resulting in a packed fill. The 0 100 0 Spout Horizontal 406.6 68.8 2071.3 0.8643 0.995
0 100 0 Spout Vertical 406.6 68.8 2089.8 0.7788 0.997
sprinkle-fill method was used to simulate
0 100 0 Sprinkle Horizontal 481.9 63.0 1037.9 0.8809 0.996
grain spreaders used in farm bins. The 0 100 0 Sprinkle Vertical 481.9 63.0 986.0 0.7811 0.998
design of a scale-model grain spreader was 15 0 Spout Horizontal 638.0 44.9 505.2 0.8431 0.994
Tobin
not considered a viable option. Tobin 15 0 Spout Vertical 638.0 44.9 271.8 0.8821 0.997

Tobin 25 0 Spout Horizontal 604.5 48.5 846.1 0.7788 0.999

Canola Samples Tobin 25 0 Spout Vertical 604.5 48.5 384.2 0.8521 0.997

0 0 100 Spout Horizontal 569.8 54.0 205.4 0.8661 0.997

Canola varieties can be classified in two
0 0 100 Spout Vertical 569.8 54.0 83.6 0.9090 0.998
groups: Brassica campestrisand/?, napus.
Tobin 0 15 Spout Horizontal 677.2 41.2 345.1 0.8284 0.999
Brassica campestris seeds are about 17% Spout Vertical 677.2 41.2 209.6 0.8659 0.997
Tobin 0 15
smaller in diameter than B. napus seeds. Spout Horizontal 666.8 42.5 296.9 0.8417 0.999
Tobin 0 25
Brassica campestris was represented by Tobin 0 25 Spout Vertical 666.8 42.5 174.3 0.8804 0.996
the variety Tobin, which was harvested at tThe equivalent particle diameter of Tobin seeds was 1.50 0.025 mm and of Westar seeds was 1.80
an initial moisture content of 6.5% in the 0.008 mm.
fall of 1984. The Tobin canola was cleaned
by a commercial cleaning plant in Saska and their physical properties are given in comparison pycnometer. The averagevol
toon, Saskatchewan. Brassica napus was Table I. ume of one kernel was then calculated.
represented by the variety Westar. The estimated volume of a single kernel
Cleaned Westar with an initial moisture Moisture Measurements was set equal to the volume of an equiv
content of 7.0% was purchased from a All tests were conducted at a moisture alent sphere having a diameter of d which
local seed company. content of 7.0 0.5% (wet basis). The was then determined. This method could
Foreign material usually found in can moisture contents of all samples were not be used for the estimation ofthe equiv
ola can be divided into two groupschaff determined according to American alent diameter of fines and mixtures of
and fines. Both these fractions were pur Society of Agricultural Engineers Stan Tobin, chaff and fines because of diffi
chased from a seed-cleaning plant. The dard S352.1 (ASAE 1984). Samples were culties in counting the fine particles. Chaff
chaff, which represents the fraction of for dried in an air convection oven at 130C particles were more close to needles and
eign material larger than average canola for 4 h. The decision to use this standard long cylinders in shape and were not repre
seeds, was overflow from the cylinder for chaff and fines was based on the fact sented as equivalent spheres.
machine in a commercial cleaning opera that these components were derived from
tion. The fines, which represents particles canola. AIRFLOW-PRESSURE DROP
smaller than average canola seeds, were RELATIONSHIPS
underflow of number 21 sieve in a com Bulk Density and Porosity Measure Several equations are available in the
mercial cleaning operation. Size analysis ment literature to represent airflow-pressure
on these samples was not done because of The mass of sample required to fill the drop data. A comprehensivediscussionon
lack of any standards available for this box divided by the volume ofthe box gave the use of these equations in mathematical
purpose but these fractions were defined in situ measurements of bulk density for modelling of airflow distribution in bins is
qualitatively. Chaff consisted mainly of all experiments. The particle volumes of given by Segerlind (1983). Based on his
wild oats and broken stems of canola small samples (about 15 g each) were recommendation, data were analyzed
plants. Fines consisted mainly of broken determined using an air comparison pyc- using Shedd's (1953) equation:
canola kernels, whole small canola ker nometer (Beckman Model 930, Beckman
v = A(AP)B 0)
nels, weed seeds and fine soil particles. Instruments Inc., Fullerton, Calif.).
Resistance to airflow for both airflow Porosities were then calculated according where V = airflow rate (m3/(sec-m2)), AP
directions was determined for clean to Mohsenin (1970). = pressure drop per unit bed depth (Pa),
Tobin, clean Westar, chaff, fines and vari and A and B = experimental constants for
ous combinations of chaff, fines and Determination of Equivalent Particle each test. The equation was fitted in linear
Tobin. Mixtures were manually prepared Diameter form to log-transformed data using GLM
using known masses of components. In To determine the equivalent particle procedure of the Statistical Analysis Sys
total, 22 experiments with triplicate runs diameter of clean canola, the volume of tem, Inc. (1982) and the constants A and B
were conducted. The sample compositions 1000 kernels was determined using the air were estimated for each test.

190
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
TABLE II. MEAN PRESSURE DROPS AND STANDARD DEVIATIONS AT VARIOUS AIRFLOWS extent, these factors. Considering all these
ACROSS 1 m OF 6.5% MOISTURE CONTENT TOBIN CANOLA FOR TWO AIRFLOW DIREC factors, a variation of 5-10% in experi
TIONS AND TWO METHODS OF FILL
mental data is justifiable, and data with
Pressure drop (Pa/m) such variability have an excellent practical
Horizontal Vertical value.
Spout Sprinkle Spout Sprink e
Sprinkle fill gave consistently higher
Airflow
(m3/(sec-m2)) Mean SDt Mean SD Mean SD Mean SD
bulk densities by approximately 7%, as
can be seen from Table I. Resistance to
0.0158 80.0 10.6 166.8 10.3 120.0 10.1 200.0 0.0
0.0217 105.2 6.1 210.0 11.0 158.8 7.9 280.0 0.0 airflow was also significantly affected by
0.0298 158.8 8.3 303.2 19.7 225.2 6.0 400.0 0.0 fill method. In Fig. 2, difference in airflow
0.0454 258.8 9.2 480.0 17.9 352.0 11.3 613.2 20.6 resistance due to fill method appears to be
0.0624 382.8 10.1 713.2 30.1 514.8 4.1 893.2 20.6
small, because it is drawn to a log scale; in
0.0857 573.2 16.6 1063.2 38.8 757.2 6.5 1300.0 17.9
733.2 24.4 1353.2 30.1 954.8 4.1 1653.2 20.6
fact it is a two to one ratio.
0.1032
0.1178 880.0 36.7 1613.2 30.1 1133.2 8.3 1926.8 16.3
0.1305 1000.0 28.8 1853.2 20.6 1288.0 7.2 2206.8 16.3
0.1419 1110.8 53.6 2070.0 32.9 1437.2 10.9 2456.8 15.0
Analysis of Data
0.1523 1240.0 55.2 2273.2 37.2 1576.0 13.4 2700.0 17.9 Equation 1 in its linear form was fitted to
0.1619 1336.0 41.6 2480.0 31.0 1716.0 9.8 2910.0 11.0 mean data for each of the 22 test condi
0.1709 1442.8 46.2 2666.8 41.3 1840.0 7.2 3140.0 17.9 tions shown in Table I. The estimated
tSD is standard deviation and is based on sample size of six. coefficients A and B along with correlation
coefficients for each sample are given in
00 Table I. The correlation coefficients for all
_

H Spout samples are greater than 0.99. Except for

* H Sprinkle 100% chaff, the B exponents are larger for
o V Spout the vertical than the horizontal direction of
E -

in airflow. The exponents for spout-filled

\ o V Sprinkle
ro "

chaff and fines are larger than other sam

E
ples. From a practical point of view, sam
D0 AOi0
Do ples of only chaff and fines can be
0-10
n r\
eliminated. Considering the remaining
z DO AC
samples it will be reasonable to assume
o DO A 0
_

that B exponents are nearly constant at

Do a O

0.82 0.02 for horizontal flow direction

- O A O
and at 0.86 0.01 for vertical flow direc

D O a O
tion. Statistically, the mean values of the B
-

exponents are significantly different

O A
I l 1
.Ll._l.u_
001
100
Pressure
Figure 2. The effect of method offill and of direction of airflow on resistance to airflow of Westar
canola at 7.0% moisture content. H indicates the horizontal airflow direction and V
indicates the vertical airflow direction.
RESULTS AND DISCUSSION
Resistance Data
Resistance data for the sample of Tobin
are shown in Table II and for the sample of
Westar are shown in Fig. 2. For other
samples, data have also been tabulated and
graphed (Jayas 1987). The mean values
given in Table II are based on a sample
size of six. For each replicate, data were
recorded for both increasing and decreas
ing airflows. For 80% of the measure
ments, the coefficient of variation was less
than 5%, and for only 4% of the measure
ments it was greater than 10%. The varia
tion in bulk density and porosity among
triplicates was less than 0.5% for most
samples. The moisture contents of sam
ples were within 0.5 percentage points
among triplicates.
The orientation of particles can also sig
O
i i i 1 i i i i 1 ,1 .i L.L.LJJL
1000 10000
Drop, (AP), Pa/m
nificantly affect the resistance of a product
to airflow. Consider a hypothetical prod
uct with cylindrical particles. If all the
particles are oriented in a cubic box with
their major axes being horizontal, then
resistance to airflow will be less in the
horizontal direction compared to the ver
tical direction. The quantitative assess
ment of orientation of particles was
impossible and might have contributed to
the variation among triplicates. Another
factor which might contribute to the resis
tance to airflow will be the weather condi
tions during harvest. These conditions
may change the particle sizes of canola
and the amount and size distribution of
foreign material within the canola. (As an
example, in a dry year stems of plants may
be ground to small pieces.) The artificially
prepared mixtures may simulate, to some
(P<0.01) but again from a practical point
of view exponent B can be assumed con
stant at 0.84. This assumption will intro
duce an error of 2% in the predicted
exponent B. As is also demonstrated in
Fig. 2, the plotted points form reasonably
parallel lines.
The intercepts A can be used to compare
the effect of direction of airflow on resis
tance to airflow of canola. The ratios ofthe
intercepts for vertical to horizontal direc
tion varied from 0.5 to 0.7 with a mean
and standard deviation of 0.570.09.
Resistance to airflow in the horizontal
direction is about 0.6 the resistance to air
flow in the vertical direction. This ratio of
the horizontal to vertical resistances to air
flow is very similar to the value of 0.63
reported for wheat at an airflow value of
0.077 m3/(sec*m2) but is higher than the
0.47 value reported for barley (Kumar and
Muir 1986).
In mathematical models used to predict
pressure patterns in bins filled with canola
the coefficient A for the horizontal direc
tion can be found by multiplying the
coefficient A for the vertical direction by
1.7. This should be incorporated in future
prediction models.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 191

Future Research
A substantial amount of energy is spent
in drying grains and oilseeds on farms. A
growing trend is to dry grains and oilseeds
by forcing near-ambient-temperature air
through the stored mass. This study leads
to the hypothesis that the bins which will
allow horizontal airflow through a stored
mass will reduce energy consumption in
drying by 40%. The major problem which
we foresee in the design of such bins will
be to eliminate short circuiting of air cur
rents. Studies must also be conducted to
determine the effective heat and mass
transfer properties due to direction of air
flow. If properties either encourage faster
removal of moisture or maintain the same
level of moisture removal for horizontal
airflow direction compared to vertical air
flow direction, then development of new
aeration systems will be worth pursuing.
CONCLUSIONS
Based on this study it can be concluded
that resistance to airflow of canola and of
mixtures of canola and foreign materials
for the horizontal airflow direction is 0.6
the resistance to airflow for the vertical
airflow direction.
REFERENCES
ABRAMS, C. F. and J. D. FISH. 1978. Resis
tance of sweet potatoes to airflow. ASAE
Paper no. 78-4523. Am. Soc. Agric. Eng.,
192
St. Joseph, Mich.
AGRAWAL, K. K. and P. CHAND. 1974.
Pressure drop across fixed beds of rough
rice. Trans. ASAE (Am. Soc. Agric. Eng.)
17(3): 560-563.
AMERICAN SOCIETY OF AGRI
CULTURAL ENGINEERS. 1984. Stan
dards 1984. Am. Soc. Agric. Eng., St.
Joseph, Mich.
BAKKER-ARKEMA, F. W., R. J. PATTER
SON and W. G. BICKERT. 1969. Static
pressure-airflow relationships in packed
beds of granular biological materials such as
cherry pits. Trans. ASAE (Am. Soc. Agric.
Eng.) 12(1): 134-136, 140.
BARROWMAN, R. and D. S. BOYCE. 1966.
Air distribution from lateral ducts in barley.
J. Agric. Eng. Res. 11(4): 243-247.
BERN, C. J. and L. F. CHARITY. 1975. Air
flow resistance characteristics of corn as
influenced by bulk density. ASAE Paper no.
75-3510. Am. Soc. Agric. Eng., St. Joseph,
Mich.
BROOKER, D. B. 1969. Computing air pres
sures and velocity distribution when air
flows through a porous medium and non
linear velocity pressure relationships exist.
Trans. ASAE (Am. Soc. Agric. Eng.) 12(1):
118-120.
HAQUE, E., D. S. CHUNG and G. H. FOS
TER. 1981. Pressure and velocity field in
airflow through packed bed of corn mixed
with fines under non-Darcy flow conditions.
Trans. ASAE (Am. Soc. Agric. Eng.) 24(6):
1595-1599, 1604.
JAYAS, D. S. 1987. Resistance of bulk canola
oilseed to airflow. Ph.D. Unpublished The
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
sis, University of Saskatchewan, Saskatoon,
Sask.
KUMAR, A. and W. E. MUIR. 1986. Airflow
resistance of wheat and barley affected by
airflow direction, filling method and dock
age. Trans. ASAE (Am. Soc. Agric. Eng.)
29(5): 1423-1426.
LAMOND, W. J. and E. A. SMITH. 1982.
Modeling low temperature drying of grain in
anisotropic beds. Third International Drying
Symposium, Birmingham, U.K.
MARCH ANT, J. A. 1976. The prediction of
airflows in crop drying systems by the finite
element method. J. Agric. Eng. Res. 21:
417-429.
MOHSENIN, N. N. 1970. Physical properties
of plant and animal materials. Vol. I. struc
ture, physical characteristics and mechan
ical properties. Gordon and Breach Science
Publishers, New York.
STATISTICAL ANALYSIS SYSTEM INC.
1982. SAS users guide: Statistics. SAS,
Inc., Raleigh, N.C.
SEGERLIND, L. J. 1983. Presenting velocity-
pressure gradient data for use in mathe
matical models. Trans. ASAE (Am. Soc.
Agric. Eng.) 26(4): 1245-1248.
SHEDD, C. K. 1951. Some new data on resis
tance of grains for airflow. Agric. Eng.
32(9): 493-495, 520.
SHEDD, C. K. 1953. Resistance of grains and
seeds to airflow. Agric. Eng. 34(9):
616-619.
STIRNIMAN, E. J., G. P. BODNAR, and E.
N. BATES. 1931. Tests on resistance to pas
sage of air through rough rice in a deep bed.
Agric. Eng. 12: 145-148.
 DESIGN, COST AND PERFORMANCE OF A FREE-ACCESS, TWO-LEVEL PEN
FOR GROWING-FINISHING PIGS

P. A. Phillips1 and D. Fraser2

'Engineering and Statistical Research Centre; and 2Animal Research Centre, Research Branch, Agriculture Canada,
Ottawa, Ontario K1A 0C6

Contribution no. I-8241 and no. 14012, received 30 May 1986, accepted 10 October 1986

Phillips, P. A. and D. Fraser. 1987. Design, cost and performance of a free-access, two-level pen for growing-
finishing pigs. Can. Agric. Eng. 29: 193-195.

Precast concrete panels and a ramp were installed on a steel support frame in a conventional 2.6 x 4.9-m slotted-floor
pen to produce a free-access two-level pen for growing-finishing pigs. The area ofthe two-level pen was about 1.7 times
that ofthe single level pen at an added material cost of approximately $70 per m2of upper level. Daily gain, feed conversion
efficiency and carcass grade index of pigs reared from 25 kg to market weight in the two-level pen compared favorably with
pigs reared in a one-level pen of similar total floor area.

INTRODUCTION
Two-level pig housing is sometimes
adopted as an economical method of using
barn space. Two-level "decks" are nor
mally used for pigs only at the weaner
stage, and they offer few advantages
beyond the economy of housing more ani
mals per unit of floor space.
The use of a ramp providing free access
to both levels creates some new pos
sibilities in pig housing and makes it feas
ible to house heavier growing-finishing
pigs in two-level pens. In such pens, the
two levels provide a choice of thermal
environment, and the exercise gained in
using the ramp could be beneficial to pig
health and condition. The upper level
might be used to increase the proportion of
solid floor area available to the pigs, and
the choice of levels could provide avenues
of escape from aggressive pen-mates.
Hence, such designs could provide a
means of addressing concerns over the
welfare of intensively housed pigs.
A previous study (Fraser et al. 1986)
used small prototype pens, 2.4 x 2.4 m,
to test the principle of free-access, two-
level housing for pigs. The animals were
studied for relatively short periods (8 wk)
over a weight range of about 20-60 kg at a
stockingdensity of 0.53-0.63 m2per pig.
The results showed that pig weight gain,
health and feed conversion efficiency were
at least equal to that of similar pigs housed
in conventional, single-level pens offering
the same total floor space.
The work reported here tested a modi
fied design intended to suit more normal
commercial requirements. The new pen
was designed to accommodate the entire
growing-finishing phase and to fit the com
mon 11-m-wide hog barn. This paper
describes the design, discusses costs, and
compares performance of pigs housed up
to market weight in two-level and con
ventional pens.
METHODS AND MATERIALS
Pen Description
A two-level pen (Fig. 1)was constructed
to fit into an existing pen measuring 2.6 x
4.9 m. The 0.84-m high upper platform
consisted of solid panels of precast rein
forced concrete 1.83 m long, with a 2.6-m
slotted panel (16% porous) along the back
ofthe pen. The panels were supported by a
frame of 76 x 76 x 8-mm steel angle.
Other important features of the pen,
detailed in Fig. 1, include the extended
partitions, the protective barrier on the
platform, and the ramp. The ramp frame of
25-mm square tubing was covered with a
double layer of expanded metal and cross-
laid with 18xl8x3-mm angle steps
spaced at 100 mm. The ramp was hinged at
the top end so it could be lifted out of the
way. If ceiling height is sufficient, a one
piece ramp is used; otherwise the ramp can
be made to fold by adding another hinge at
its center with adequate support provided
under the hinge.
The total area provided in the two-level
pen was about 21 m2, the bottom level
being totally slotted and 12 m2 (allowing
for space lost because ofthe ramp), and the
mainly solid top area 9 m2. Manure col
lected under the slotted floor where it was
aerated and circulated by means of an
oxidation ditch. Methods of moving pigs
into and out of the two-tiered pens are
described by Fraser et al. (1986).
Experimental Methods
The two-level pen was tested with four
consecutive groups of 24 pigs each in a
heated, insulated barn. This stocking rate
provided about 0.88 m2 of floor space per
pig, in accordance with recommended
Canadian standards for pigs of 75 - 100 kg
per pig (Anonymous 1984). Uniform
groups were selected from the Animal
Research Centre's specific-pathogen-free
Yorkshire herd. Each group was drawn
from approximately six litters maintain
ing, as closely as possible, a balanced sex
ratio. The pigs weighed about 25 kg aver
age body weight.
For the trials, a fortified pelleted corn-
barley-soybean meal diet was provided ad
libitum from two, two-hole feeders on the
upper level. Water was available from
three nipple drinkers on the lower level and
one on the upper level located over the
porous floor near the top of the ramp. On
the day after the pigs were put into the pen,
an attendant scattered feed on the ramp to
attract any pigs that had not yet climbed to
the upper level. Out of 96 pigs, four such
animals were identified; after these were
chased up the ramp once, they appeared to
adapt to the system.
Body weights of the pigs were recorded
at the start of the trial, on day 7, and on
shipping day when a market weight of
90-105 kg was reached. Total feed con
sumption was recorded for each group.
Mean carcass grade index for each group
(based on weight and thickness of backfat)
was computed from packing house
receipts. The work extended from June
1984 to December 1985, each group
requiring approximately 3 mo to attain
market weight.
Comparative data were available from
four similar groups of 24 pigs each, housed
in an adjoining room in a conventional
single-level pen measuring 1.83 x 11m.
This pen gave about 20 m2 (0.84 m2/pig)
of floor space, with 60% slotted, similar to
the two-level pen in total area and solid-to-
slotted ratio.
To determine how pigs used the two-

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 193

 A^
2.6 m

.^a%^^ 0fcf
IF ' FEEDERS
. _ ' *
^rrnnrrir
T$y\i TrrrrrrrrrrTPrripnprpnnr1"
mi 111II11111111111111! ij
ii.m.UAim
i
UIZ
SH
W\ W
/"'' iS^^^Pl'
By 1 LLi
S s s 13a a 3 oatfaMiiejsssoi'S
E
n O)
> t
R
ffiwS^^^^^I
SECTION A-A
z
> B 0.76 m |- B
; .;
A_ 1" "1-

>
z C c

> PROTECTIVE
c BARRIER
n
c fV
73
> ii
r

c c
z
g FEEDERS
z
m ! II I I

J'^

z
p SECTION B-B
*
1.83 m t-
< A
c
-

z
c Figure 1. The design and dimensions of the free-access, two-level pen.
us
a
2
2
en
90
TABLE I. PERFORMANCE VALUES (MEAN TABLE II. PERCENTAGE OF TIME SPENT The final installed cost of the second
SD) OF 96 GROWING-FINISHING PIGS IN ACTIVE AND RESTING IN THE TWO- level under farm economic conditions is
A FREE-ACCESS, TWO-LEVEL PEN, LEVEL PEN
difficult to estimate, but certainly must
COMPARED WITH SIMILAR GROUPS IN A
CONVENTIONAL PEN Time (%)t include labor, transportation, and extra
costs for small tools and fasteners and a
Level Active Resting Total
Pen design profit for the installer. As a first approxima
Upper 12.6(10.4-14.3) 31.4(12.2-46.3) 44
Variablet Two-level Conventional tion, these latter items can reasonably be
Lower 9.4 (5.5-14.0) 46.6(37.8-59.4) 56
Total 22 (15.9-28.3) 78 (71.0-84.4) 100
expected to result in an installed cost
Days to market 153.3 154.6
Initial weight (kg) 25.64.35 24.43.01
which is double the materials cost (i.e.
tMean of four groups, and range of group means.
Final weight (kg) 99.25.74 96.75.84 $140 per m2). This compares favorably to
Gain (g/day) 878 82.2 849 94.4 $230 per m2 for purchasing new swine
0- to 7-day gain extrapolated to a 12-h period of daytime building space (Bunn and Kains 1986).
(kg) 5.461.56 5.090.84 activity, suggest an average of five to six Where stocking density of a barn is
Feed conversion
ramp uses per pig per day. However, some increased by installation of two-level pens,
efficiency 2.70 0.095 2.72 0.054
Carcass index 100.7 3.82 100.63.11
pigs use the ramp more than others, while design ventilation rate would have to be
a few did not use the ramp at all until forced increased, and the air inlet configuration
tStandard deviations with n = 96 pigs (except feed
to do so. Ramp design features need to be might have to be altered to suit the two-
conversion efficiency, where n = 4 pens) are shown
to illustrate individual variability in performance in improved so that all pigs will use the ramp level system. Studies of ventilation
the two pen types. spontaneously. Work is underway to deter requirements are needed under full-scale
mine pig preference for ramp design fea conditions and with a variety of manure-
level pen, observations were made tures such as slope, width, step spacing handling systems.
between 0730 and 1000h for 1day approx and illumination.
imately every 2 wk during each trial. The Pigs dunged only occasionally on the
number of pigs active and resting and num upper level, and the dung did not accumu CONCLUSIONS
bers on each level were recorded at 10-min late because it was usually pushed off the Performance results indicate that two-
intervals. The number of trips up or down platform by normal pig traffic. level housing has no adverse effects on
the ramp was counted throughout the 2.5-h In this study, the pigs rested on the average daily gain, feed conversion effi
observation period. upper level much more than in our pre ciency and grade index of pigs grown from
vious study (Fraser et al. 1986) which 25 kg to market weight (100 kg) at a stock
RESULTS tested the smaller 2.4 x 2.4-m, two-level ing density of approximately 0.88 m2/pig.
Pigs performed very well in both the pen. The nearly twofold difference proba The cost of new floor space provided by
two-level and the conventional pens bly can be attributed to the much larger adding a second level is about 60% of the
(Table I). Days to market were similar for undisturbed area available for resting in cost of new building floor space. Further
both groups, averaging 153.3 for the two- the longer pen. This larger space seemed studies are needed on the ventilation of
level and 154.6 for the conventional pen. better suited to the pigs' resting habits as two-level housing systems, and to identify
Final market weights averaged slightly the animals appeared to prefer huddling in and quantify possible improvements in the
lowerfor the conventional group (96.7 kg) groups of six to ten rather than in groups of welfare of pigs housed in free-access, two-
as compared to the two-level group (99.2 two or three as required by the smaller pen. level pens.
kg). The two-level group had slightly bet The cost of adding the upper level was
ter average daily gain, feed conversion approximately $70/m2 for materials. The ACKNOWLEDGEMENTS
efficiency and 7-d weights, but the dif precast concrete panels are the main com We wish to acknowledge the helpful cooper
ferences were small and not statistically ponent and account for 70% ofthe material ation of the staff at the Animal Research Centre
significant. Seven-day weights indicate cost. Next, and about equal to each other in pig unit, and the Engineering and Statistical
how well pigs adapted initially to the two- cost, are the support structure for the plat Research Centre Shop staff for constructing the
level system. Carcass grade index was the form and the extra material required to two-level pen.
same for both pen systems. expand the partition to 1.55 m in height. In
Pigs distributed their time fairly evenly certain cases, the pen partitions could REFERENCES
between the upper and lower levels serve to support the platform. For exam ANONYMOUS. 1984. Recommended code of
(Table II). Because the feeder was on the ple, concrete pen partitions to a height of practice for the care and handling of pigs.
upper level, there was proportionally more 0.76 m could support the concrete panels, Agriculture Canada, Ottawa, Ont. Publ.
activity there than below. The lower level with spindle partitions on top. 1771/E.
was used mainly for resting, drinking and The barrier and ramp must each be BUNN, A. J. and F. A. KAINS. 1986. Sum
dunging. During the interval from 0730 to costed against the second level and, mary of pork production and costs, February
1986. Pork News and Views. Ontario Minis
1000 h the ramp was used an averageof 31 together, they account for 10% ofthe mate
times.
try of Agriculture and Food, London, Ont.
rial cost. A small area of the lower level is
Mar.-Apr. issue, p. 3.
required for the ramp and support posts FRASER, D., P. A. PHILLIPS, and B. K.
DISCUSSION and, strictly, the cost of this lost space THOMPSON. 1986. A test of a free-access
Pigs used the ramp willingly at all stages should also be assessed against the upper two-level pen for fattening pigs. Anim. Prod.
of growth. The behavioral observations, level. 42: 269-274.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 195

 USE OF PLANNED DRAFTS IN AN ATTEMPT TO REDUCE PREWEANING
MORTALITY IN BABY PIGS

P. A. Thacker1 and E. M. Barber2

'Department ofAnimal and Poultry Science and 2Department ofAgricultural Engineering,

University of Saskatchewan, Saskatoon, Saskatchewan S7N 0W0
Received 17 July 1986, accepted 8 December 1986

Thacker, P. A. and E. M. Barber. 1987 Use ofplanned drafts inanattempt toreduce preweaning mortality inbaby
pigs. Can. Agric. Eng. 29: 197-200.

A total of 69 litters from crossbred sows (Large White x Landrace) of mixed parity was used to evaluate the
effectiveness ofafarrowing crate attachment (Medata Blowaway Unit) designed to reduce baby pig mortality by directing
an air stream directly under the standing sow. Use ofthis attachment was hypothesized tocondition piglets toavoid the
areaunder the standing sowandthereby reduce theirchances of being crushed if the sow moved. All sows received the
same treatment during the experiment except that a Blowaway Unit was attached tohalf ofthe farrowing crates a minimum
of3days before parturition and remained there until 10 days postpartum. Sows farrowing in crates fitted with a Blowaway
Unit crushed significantly (P < 0.05) fewer of the total number of piglets born alive than did sows farrowing incrates
without a Blowaway Unit (5.3 vs. 2.3%). However, mortality occurring as a result of starvation andother miscellaneous
causes was slightly higher for sows farrowing in crates fitted with a Blowaway Unit than without (11.5 vs. 10.6%).
Therefore, under the conditions of this experiment, preweaning mortality was not significantly reduced by using the
Blowaway attachment.

INTRODUCTION
Preweaning mortality is a source of
majoreconomic loss to the swine industry.
It is estimated that an average of 15-25%
of all pigs farrowed die before weaning
(Fahmy et al. 1978). Most of these deaths
result from crushing by the sow, chilling,
starvation, weakness or disease (Bille etal.
1974; English and Smith 1975). In con
trast, some production units have a pre
weaning mortality below 10% (English et
al. 1982). Therefore, it may be possible to
save as many as one pig per litter through
improved facilities and management.
Preweaning mortality can be reduced by
providing a heated creep area for the baby
pig (Adams et al. 1980). This supplemen
tary source of heat encourages the piglets
to lie away from the sow and therefore
reduces their chances of being crushed
when the sow moves (Titterinton and
Fraser 1975). Unfortunately, the newborn
pig tends to lie against the sow's udder for
the first 2 or 3 days of life even when
heated creep areas are provided (Petherick
1983) and it is during these first 3 days that
the majority of mortalities occur (English
and Smith 1975).
A farrowing crate attachment (Medata
Blowaway Unit, Medata Sales, Elmira,
Ont.) has been developed recently which
is designed to reduce baby pig mortality by
creating a draft directly under the standing
sow. It was hypothesized that this draft
would condition the piglets to avoid the
area under the sow and to utilize the creep
as their sleeping/resting area. The purpose
of this experiment was to evaluate the
blowaway attachment in order to deter
mine its effectiveness in reducing pre
weaning mortality.
MATERIALS AND METHODS
This experiment was conducted from
April to October, on a 60-sow, farrow to
finish operation located near Vilna,
Alberta. The farrowing facilities consisted
of two 4.6 x 12.2-m farrowing rooms
located on either side of a 1.8-m hallway.
Each farrowing room contained seven
1.5 x 2.1-m wire-floored farrowing crates
(Faromor Inc., Waterloo, Ont.) located
over a sloped gutter (280-480 mm deep)
which was flushed with fresh water twice a
day to remove fecal material. The farrow
ing crates were equipped with heated front
creeps maintained at 27C with heat
provided from hot water pipes located in
the concrete floor of the creep.
Incoming air was drawn from the attic
of the swine barn and entered the farrow
ing room through an adjustable styrofoam
baffle located on the interior wall. Air was
exhausted through two 305-mm-diameter,
variable-speed fans located on the exterior
wall. Hot water heating was utilized to
maintain the temperature of the farrowing
room at222C.
A total of 69 litters, obtained from
crossbred sows (Large White x Land-
race) of mixed parity, were tested during
this experiment. Sows were moved into
the farrowing room on day 109 of gestation
at which time they were alternately
assigned to either the control or the experi
mental group. Sows in the experimental
group received the same treatment as those
in the control group except that a Blow
away Unit was attached to their farrowing
crates a minimum of 3 days before parturi
tion and remained there until 10 days
postpartum. Two Blowaway Units were
utilized during the experiment and were
moved from crate to crate as required.
The Blowaway Unit (Fig. 1) comprised
a small squirrel cage fan (76 mm diameter,
64-mm-wide blade) powered by a 115-V
motor rated at 0.7 amps. The fan directed
air into a 0.9-m downpipe constructed
from rigid plastic tubing with outside and
inside diameters of 60.4 and 51.7 mm, re
spectively. The downpipe was joined to
the center of a horizontal 1.4-m length of
plastic tubing which had 13 equispaced
20-mm-diameter holes drilled in it at 15
and 30 angles off a horizontal plane.
The horizontal tube of the Blowaway
Unit was mounted on the lowest bar of a
farrowing crate. The unit was secured to
the crate by means of an aluminum box-
section bracket and two clamping screws
fitted with thumb screws. This mounting
system enabled a Blowaway Unit to be
installed on a farrowing crate in less than
5 min.
An infrared sensing device was situated
on the top of the downpipe. The infrared
beam of the Blowaway Unit was adjusted
so that it passed just above the sow's back
when it was lying down. A 7.6-cm reflec
tor disc, mounted on the top bar of the
opposite side of the farrowing crate, re
turned the beam of light to the receiver.
Whenever the sow stood, the beam of light
was broken, which activated the blower
and directed a 12.45 m/s air stream (range
9.5-15.0 m/s) under the sow's body.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 197

 FAN
1NFA-RED BEAM (switch)
BLOW AWAY ATTACHMENT
FARROWING CRATE
HEAT LAMP
(in creep area)
When the sow lay down, the air stream
was automatically switched off. Use of a
solid partition between farrowing crates
prevented piglets in adjacent crates from
coming in contact with the airstream.
Air temperature at the Blowaway inlet
and at the discharge holes was measured
using a Fluke multimeter and temperature
sensing attachment and was the same at
both locations (22C). Total air flow, esti
mated using an aspect ratio of 0.6, 13
outlet holes 20 mm in diameter and an
average air velocity of 12.45 m/s, was
30.5 1/s.
During gestation, all sows were housed
individually and received between 1.8 and
2.7 kg of feed per day depending on con
dition. During lactation, the sows were fed
ad libitum the same ration as they received
during gestation. Chemical analysis ofthe
gestation-lactation diet (Association of
Official Analytical Chemists 1975) indi
cated that the ration met or exceeded
National Academy of Sciences-National
Research Council (1979) recommenda
tions for all nutrients.
Pigletswere individually weighed with
in 24 h of birth and at weaning (average
weaning age 26 days). Routine herd man
agement procedures were followed, in
cluding clipping teeth and tails and ear
notching at day 1,administering ironinjec
tions and castrating at day 3 and antibiotic
198
Figure 1. The Blowaway Unit.
TABLE I. PERFORMANCE OF SOWS AND LITTERS
Blowaway SEt
Control
Number of litters 35 34
Average sow parity 4.42 5.02 0.51
0.15
Pigs born dead per litter 0.91 1.05
Pigs born alive per litter 11.25 11.26 0.43
Birth weight (kg) 1.40 1.41 0.03
Pigs weaned per litter 9.45 9.71 0.30
Weaning weight (kg) 7.30 0.18
7.35
Mortality (%) 15.97 13.76 2.47
t Standard error of the mean.
treatment as necessary. Heat lamps were and immersing them into water.
provided in the creep area for the first The experimental data were tested for
week following parturition. Fresh creep normality by measuring skewness and kur-
feed (18% crude protein) and water were tosis following which it was subjected to
provided after the piglets reached 10 days analysis of variance according to the pro
of age. cedures outlined by Snedecor and Cochran
Records were kept of the number of pigs (1967). The incidence of specific causes of
born alive, born dead and weaned. Birth mortality were analyzed by Fisher's exact
weight and weaning weight were also method (Keeping 1962).
recorded. In addition, mortality from birth
to weaning was tabulated. Piglet mortali RESULTS
ties were subjectively grouped into pre- There were no significant differences
agreed categories by the barn manager. (P> 0.05) in the number of pigs born alive
Causes of piglet mortality included overly or weaned between sows farrowing in
ing (piglets which were obviously crushed crates fitted with or without a Blowaway
by the sow), starvation (showing signs of Unit (Table II). Sows on both treatments
emaciation), genetic defects (ruptures, farrowed an average of 11.3 piglets born
spraddle legs and other deformities), born alive per litter. Control sows weaned 9.4
weak (low piglet vigor at birth) and other piglets per litter while sows farrowing in
miscellaneous causes. Stillbirths were crates fitted with a Blowaway Unit weaned
confirmed by removing the piglets lungs 9.7 piglets per litter. Total preweaning
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 TABLE II. INCIDENCE OF PREWEANING
MORTALITY DUE TO SPECIFIC CAUSES t
Control Blowaway
Crushed
5.32a 2.34b
Starvation 2.53 3.39
Genetic defect 0.76 1.04
Born weak 4.82 3.65
Miscellaneous 2.54 3.39
t Percentage of totalpigsbornalive(394bornalivein
the control group and 383 in the blowaway group).
a,b Means followed by different letters differ
(/><0.05).
mortality averaged 15.97% for control sows
and 13.76% for sows farrowing in crates
fitted with a Blowaway Unit (P > 0.05;
Table I).
There were no significant differences
(P > 0.05) in the birth weight or weaning
weight of piglets farrowed in crates fitted
withor without a Blowaway Unit (Table I).
The average birthweight of piglets for both
treatments was 1.4 kg while the average
weaning weight of piglets was 7.3 kg.
Sows farrowing in crates fitted with
a Blowaway Unit crushed significantly
(P < 0.05) fewer piglets than sows far
rowingin crates without a Blowaway Unit
(Table II). Sows farrowing in standard
crates crushed 5.3% ofthe total number of
piglets born alive compared with 2.3% in
crates fitted with a Blowaway Unit.
However, mortality as a result of starva
tion, genetic defects and other miscel
laneous causes was slightly higher in
farrowing crates equipped with a Blow
away Unit compared with the control
(Table II).
There were no problems with the Blow
away Unit during the experiment. Care
was taken to ensure that dust did not
accumulate on the infrared reflector as this
would have caused the fan to blow contin
uously. Under normal use, it is unlikely
that a build-up of dust would occur, since
most commercial farrowing rooms are not
excessively dusty and the Blowaway Unit
would be cleaned prior to installing on a
clean farrowing crate.
DISCUSSION
The overall results of this experiment
indicatethat use ofthe Blowaway Unit can
result in a significant (P < 0.05) reduction
inpreweaning mortality due to crushingby
the sow. Crushing accounted for 33% of
piglet mortality for litters farrowed in
crateswithouta Blowaway Unit compared
with only 17% of piglet losses in crates
fitted with a Blowaway Unit. However,
more piglets were lost due to starvation
and other miscellaneous causes in crates
equipped with a Blowaway Unit than in
the control. As a consequence, total pre
weaning mortality was not significantly
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
different between the Blowaway and the
control group. Brabbs and Hogberg (1985)
also found a reduction in crushing losses
but no significant difference in total pre
weaning mortality in an experiment
designed to test the effectiveness of the
"Portable Pig Sitter", an apparatus based
on a principle similar to that of the Blow
away Unit.
The fact that there was an increase in
preweaning mortality due to starvation and
other miscellaneous causes for piglets far
rowed in crates fitted with a Blowaway
Unit suggests that at least some of the
piglets that were saved from crushing by
the Blowaway Unit later died due to other
causes. Therefore, if the full potential of
the Blowaway Unit is to be realized, addi
tional attention must be given to under
privileged pigs to ensure that they do in
fact survive until weaning and are not sim
ply saved from crushing so that they can be
lost at a later date from other causes. Man
agement techniques that would appear to
have potential for use in combination with
the Blowaway Unit would include cross-
fostering (transferring piglets to another
sow (Gadd 1985)), supplementing diets
with commercial milk replacer (Hogberg
1979) and creching (use of a supplemen
tary rearing box (Gadd 1984)).
Piglets farrowed in crates fitted with a
Blowaway Unit became conditioned to
avoid the air stream within the first 12 h of
life and, thereafter, they moved to the
creep area as soon as the Blowaway Unit
was activated. Even after the Blowaway
Unit was removed, the piglets continued to
move away from the sow's body whenever
it stood. Since the majority of piglet losses
occur during the first 3 days of life
(English et al. 1982), it is likely that the
Blowaway Unit could be removed from
the farrowing crate approximately 4 days
after parturition without reducing its effec
tiveness. Therefore, one Blowaway Unit
could be used for about six farrowings a
month which means that one unit would be
required for every 30 sows in the herd.
In assessing the potential of the Blow
away Unit as a means of reducing pre
weaning mortality, it is important to point
out that this experiment was conducted
under excellent management and environ
mental conditions with the result that even
the control sows had a preweaning mor
tality significantly below the level found in
most commercial operations. In a recent
survey of eight commercial swine units
located throughout the Canadian Prairies
(VIDO Swine Technical Group 1985),
crushing losses accounted for an average
of 42.2% of all piglet mortality whereas in
the control group of this experiment,
crushing losses accounted for only 33% of
piglet mortality. Therefore, with a poorer
environment and a lower level of manage
ment, it is possible that more pigs would
have been saved as a result of using the
Blowaway attachment. However, under
the conditions of this experiment, pre
weaning mortality was not significantly
reduced.
ACKNOWLEDGMENTS
This study was supported by the Farming for
the Future Program of Alberta. The authors
wish to thank Mr. Richard Scoffom for his
excellent work in the care and management of
the experimental animals.
REFERENCES
ADAMS, K. L., T. H. BAKER, and A. H.
JENSEN. 1980. Effect of supplemental
heat for nursing piglets. J. Anim. Sci. 50:
779-782.
ASSOCIATION OF OFFICIAL ANA
LYTICAL CHEMISTS. 1975. Official
methods of analysis. 12th ed. AOAC, Wash
ington, D.C.
BILLE, N., N. C. NIELSEN, J. L. LARSON,
and J.SVENDSEN. 1974. Preweaning
mortality in pigs. 2. The perinatal period.
Nord. Vet. Med. 26: 294-313.
BRABBS, E. K. and M. G. HOGBERG. 1985.
Effectiveness of the portable pig sitter on
baby pig loss. Michigan State University
Report of Swine Research. 469: 26-28.
ENGLISH, P. R. and W. J. SMITH. 1975.
Some causes of death in neonatal piglets.
Vet. Annu. 15: 95.
ENGLISH, P. R., W. J. SMITH, and A. Mac-
LEAN. 1982. The sow Improving her
efficiency. Farming Press Ltd., Wharfedale
Road, Ipswich, Suffolk.
FAHMY, M. H., W. B. HOLTMAN, T.M.
MacINTYRE, and J. E. MOXLEY. 1978.
Evaluation of piglet mortality in 28 two-
breed crosses among eight breeds of pigs.
Anim. Prod. 26: 277-285.
GADD, J. 1984. Creching is coming. Pig
Farming (Dec.) pp. 38-39.
GADD, J. 1985. Fostering saves you piglets.
Pigs 1(3): 10-13.
HOGBERG, MG. 1979. Reducing baby pig
losses through supplementation of sow's
milk with a milk replacer. Michigan State
University Report of Swine Research. 386:
5-7.
KEEPING, E. S. 1962. Introduction to statis
tical inference. Van Nostrod Co., Toronto,
Ont.
NATIONAL ACADEMY OF SCIENCES
NATIONAL RESEARCH COUNCIL.
1979. Nutrient requirements of domestic
animals. No. 2. Nutrient requirements of
swine. 8th rev. ed. NAS-NRC, Washington,
D.C.
PETHERICK, J. C. 1983. A note on nursing
termination and resting behaviour of suck
ling piglets. Appl. Anim. Ethol. 9:
359-365.
SNEDECOR, G. W. and W. G. COCHRAN.
199
 1967.Statistical methods. 6th ed. Iowa State piglets during early lactation in relation to Farrowing barn design and management.
University Press, Ames, Iowa. the positionofthe creepheater.Appl. Anim. VeterinaryInfectiousDiseaseOrganization,
TITTERINGTON, R. W. and D. FRASER. Ethol. 2:47-53. Saskatoon, Sask.
1975. The lying behaviour of sows and VIDO SWINE TECHNICAL GROUP. 1985.

200 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 AUTOMATED LINE-HEAT SOURCE SYSTEM FOR THE MEASUREMENT OF
THERMAL CONDUCTIVITY AND DIFFUSIVITY

Douglas S. McGinnis

Engineering andStatistical Research Centre, Research Branch, Agriculture Canada, Ottawa, Ontario K1A 0C6
Contribution no. 1-875, received 16 September 1986, accepted 9 March 1987

McGinnis, Douglas S. 1987. Automated line-heat source system for the measurement ofthermal conductivity and
diffusivity. Can. Agric. Eng. 29: 201-207.

A system was developed for the rapid, automatic, and simultaneous measurement of thermal conductivity and
diffusivity of food materials and, with knowledge of the food density, for determination of its specific heat. The system
utilizes the line-heat source method for conductivity and diffusivity determination, and operates under the control ofa
microcomputer which is interfaced with a data logger for temperature and voltage measurements, and with external
electronics linked to a line-heat source apparatus. Experimental measurements are made automatically after the user has
entered the necessary experimental conditions and desired experimental precision to the microcomputer. The system was
found to be extremely simple to use, and capable of providing accurate measurements, together with statistical precision
statements, in 6 min orless for agar-water gel. The system software was programmed in IBM-BASIC (version 3.0). The
software was written to provide maximum user flexibility, and contains an optional algorithm (subroutine) for automatic
calibration of the diffusivity measurement apparatus.

INTRODUCTION nent must be identified on the basis of tem with these attributes and to demon
Thermal diffusivity values of foods are largest size and/or lowest thermal strate its usefulness in making precise
used in a variety of food engineering diffusivity. measurements of thermal conductivity and
applications involving heat and mass A number of methods for thermal diffusivity automatically and rapidly.
transfercalculations for freezing, cooling, diffusivity determination have been
or other thermal processing operations. reported in the literature. These methods THEORY
The thermaldiffusivity (a) of a food mate generally require solving equations The history and development of a math
rial depends upon its composition and describing a transient temperature field ematical basis for the line-heat source
physical properties and is related to con resulting from an applied constant or con method of conductivity and diffusivity
ductivity (k), specific heat (c), and density trolled heat source or sink. Most of the determination is presented extensively in
(p) by the relationship methods are indirect in that the thermal the literature (Chang et al. 1980;Mohsenin
a = k/(p-c)
conductivity, specificheat, and density are 1980; Nix et al. 1967). The essential ele
measured separately, sometimes under dif ments of this theory are given here for
Thermal property values are reported in ferent experimental conditions, and a is discussion purposes.
the literature for a number of selected food calculated. A popular method for food Considering only radial conduction heat
materials, but for an ever increasing inven materials involves using a thermal con flux in an infinite cylinder, the tem
tory of processed and/or manufactured ductivity probe (line heat source) in con perature-rise at any point within that cylin
food products, thermal property data are junction with one or more temperature der following a suddenly initiated,
not available or are simply not reported. sensing probes at fixed radial distance(s) constant heating rate, line-heat source is
For foods undergoing thermal steriliza from it, as described by Mohsenin (1980) given by the following equation (Lowan
tion or pasteurization processes, the for and others. This technique has the advan 1940):
mulation of the product and the thermal tages of reasonably short test duration,
properties of its components comprise a small temperature change of the product, 7X6) = -ln0
(-Q"P2"1
set of critical factors which determine the and the ability to obtain the conductivity
2ttR 2 (2/i)(!)J (1)
thermal response characteristics of the and specific heat under the same test con For points close to the line-heat source
product under the applied heating condi ditions. The technique is well suited to such that p ^ 0.16, it has been found that
tions. Singh (1982) and others have dis biological materials, especially those Eq. 1 can be expressed to better than 1%
cussed the importance of using the which are significantly altered by changes error by using the first two terms (Hooper
component thermal diffusivity values in in temperature. and Lepper 1950), hence:
numerical or analytical solutions to heat Despite the short test times for this
transfer in composite foods, as opposed to method, extensive computations for the
the potentially erroneous use of "effec determination of the thermal diffusivity
7X6) = JL\zK
2-jtA'l 2
-Inp] (2)

tive" or "apparent" thermal diffusivity must be carried out using a trial and error The change in temperature for this point
values. It is therefore important that the method on a digital computer, and transfer between times 0, and 62 is, therefore:
thermal processing specialist know the ring data to the computer can be time con
thermal properties of each component suming. It was recognized that a 7(0.) - 7Xe2) = (fi/(4ir*)) ln(0,/e2) (3)
when applying that knowledge in the significant time saving could be realized if The conductivity may then be deter
design of processes to ensure food safety control of the experimental apparatus, mined from Eq. 3 knowing the heat source
and optimum quality. This is particularly gathering of the data, and subsequent data strength and the temperature versus time
true in the practical development of ther analysis could all be performed by a single relationship. Use of a time correction fac
mal processes for heat-sterilizable foods desktop computer. The objective of the tor has been widely accepted as a means of
wherein the slowest heating food compo work presented here was to develop a sys compensating for the finite diameter and

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 201


l^O'vl
LEGEND:
1. HEATING PROBE
2. THERMOCOUPLE LOCATIONS
3. THERMAL INSULATION
Figure 1. Temperature-controlled line-heat source apparatus for the measurement of thermal
conductivity and diffusivity.
unique thermal properties of the line-heat
source. The time correction factor (60) is
subtracted from each time value and the
slope(5) ofthe time-correctedtemperature
rise versus ln(0 - 60) plot can be used in
the calculation of k, giving:
k = Q/{4ttS)
where
T(9,) - T(92)
5 =
/e, - e\
To determine the thermal diffusivity,
and hence specific heat, the temperature-
rise versus time values for a point a known
distance from the line heat source are used
202
OIL OUT
4. SAMPLE CYLINDER
5. LEAD WIRES
6. OVERPRESSURE PORT
( OPTIONAL )
in combination with the known con
ductivity and density inEq. 1. To dothis,P
is found by nonlinear regression analysis
using a finite and sufficient number of
terms in that expression (e.g. 50 terms).
MATERIALS AND METHODS
(4)
A thermal diffusivity measurement
apparatus, described byMcGinnis (1986),
was modified and used for the purposes of
this work. This apparatus (Fig. 1) consists
of two sealed concentric cylinders. The
outer cylinder was designed to allow a
constant temperature fluid (e.g. oil) to cir
culate in a spiral fashion around the inner
sample cylinder. Thesample cylinder con
tains two parallel (1.7 mm o.d.) probes
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
1 cm apart: a heating probe (probe no. 1)
and a passive probe (probe no. 2). Both
probes contain a grounded thermocouple,
with its sensing junction at the mid-length
location. A constantan heating element is
packed inside the hollow probe together
with a thermal joint compound (Type 120,
Wakefield) to permit rapid heat transfer
from the heating element to the surround
ing food sample. Because pressure is an
important variable in many thermal pro
cesses for foods, such as retort processes
for foods packaged in flexible retortable
containers, the sample cylinder was fitted
witha port for controlled application of an
overpressure during diffusivity measure
ments. A third thermocouple is installedin
the outer cylinder to monitor the tem
perature of the circulating fluid.
A miniature thermal diffusivity probe
(i.e. pair of probes), similar in con
struction to the miniature thermal con
ductivity probe described by Timbers and
Robertson (1977), would be well suited for
the measurement of small and/or solid par
ticulate food materials.
The system configuration (Fig. 2)
provides for the measurement of probe
temperatures and DC voltage across the
probe heating element. A microcomputer
(PC/XT, IBM) connected to an intelligent
analog/digital signal converter (Isaac-911,
Cyborg) is used for control. A digital data
logger (Digistrip-4C, Kaye Instruments) is
used for logging, conversion, and scaling
of the thermocouple and voltage signals.
In this configuration, digital informa
tion is taken directly from the logger
through an asynchronous communications
port (RS-232) on the microcomputer, at
intervals specified by the user. Communi
cation software is combined with a timer
routineto requestreportsfromthe monitor
and to download each report to the micro
computer. These reports are stored as raw
data in array form, and broken down into
time, temperature, and voltage arrays fol
lowing each test for subsequent analysis.
The data aquisition interval of the system
is limited to a minimum of 2 sec by the
intrinsic sampling rate of the data logger.
During preliminary work, it was dem
onstrated that a faster logging rate is possi
ble in a configuration in which the analog
temperature signals are supplied by a ther
mocouple signal conditioner connected
directly to an A/D converter which itself
is connected to the microcomputer. Scal
ingand interpretation ofthe analog signals
can then be accomplished after the test
period, resulting in an increased potential
for high rate data acquisition (i.e. up to a
possible 10 000 readings per second per
channel for four channels). Such a system
 . HEATER PROBE IN SAMPLE CYLINDER

TEMPERATURE CONTROLLER
AND PUMP

OIL RESERVOIR fr

WATER OUT

DUPLEX THERMOCOUPLE LEAD WIRE

OIL CIRCULATING LINE

: COOLING WATER CIRCULATION LINE

Figure 2. Schematic layout of automated system for conductivity and diffusivity measurement.

can give accurate thermal conductivity put to a relay (SPDT, R10S-E1Y1-J10K, temperature pairs used in the regression
measurements in the shortest possible time Potter and Brumfield) which closes the may be specified by the user to better esti
(e.g. =^10 sec), with the advantage of mini heating circuit. The circuit and software mate the straight-line portion ofthe curve.
mizing temperature changes in the product were designed to permit setting of the In the final phase of the program, the user
near the heating probe, yielding a closer probe voltage prior to probe energization. is requested to provide values of param
estimate of the k value associated with the A simplified logic flow chart ofthe com eters by which the diffusivity can be deter
mean sample temperature during the test. puter program is presented in Fig. 3. Ini mined by nonlinear least squares re
A shorter test time may also be advan tially the user is prompted to input the gression, fitting Eq. 1 to the observed pas
tageous for some products which undergo desired experimental conditions. The pro sive probe temperature rise values. The
irreversible changes in thermal properties gram next monitors temperature condi sample density, number of regression data
at the prescribed test temperature. tions until a state of thermal equilibrium, points to use, the acceptable resolution
However, to obtain a sufficient temperature as defined by the user, is achieved between (maximum estimation bracket size) of the
rise at the passive probe location in such a the two probes and between the probes and best fit solution (a1), and a starting esti
short time for accurate thermal diffusivity the constant temperature fluid. Once equi mate of the specific heat for the iterative
determinations, considering temperature librium is attained, the user is prompted to solution, are supplied by the user. The
measurement precision (resolution initiate probe heating through key entry. "Golden Section method" as described by
0.1C) and error, would require that the The heating probe element is energized, Himmelblau (1972) is used for the uni-
probes be located very close together (e.g. and data are logged until the user-specified dimensional search algorithm for a1.
r ^ 1cm), and/ or that the heating power be test duration is completed, at which time Confidence limits for the experimental
increased substantially (e.g. Q 2^ 30 the probe is de-energized. The raw data are curve fit of Eq. 1are developed by a second
W/m). The finite diameter ofthe two pro then transferred to temperature, time, and Golden Section search algorithm. This
bes (e.g. 1.7 mm), and their combined voltage arrays. The resistance ofthe heat algorithm finds a value of a such that:
influenceon heat flow in the vicinity ofthe ing probe is calculated for each probe tem
probes imposes a limitation on the probe perature and used with the known voltage (T(a) =a(a')-(V 1+'(F(l,
m \
m- 1, 1 7))
spacing. In addition to this, the probe to calculate the average probe heat strength (5)
spacing is limited by the particle size and (Q) for the test (the constantan heating
uniformity of the product. For these rea element resistance varies slightly with where
sons, the potentially high rate of tem temperature, in a linear fashion). The slope
o-(a) = X (T -/(&, a))2,
perature data acquisition possible using of the probe temperature rise versus
this configuration was considered to yield ln(6 60) curve is determined by linear and
no significant advantage for this applica least squares regression, and the thermal
tion. conductivity is calculated (Eq. 4). Since a1 = value of a such that a is minimized.
Heating of the conductivity probe is ini heating probe temperature rise values dur The probability level (7) in Eq. 5 is only
tiated by the user from the computer key ing the first few seconds following the start approximate where/is nonlinear (Draper
board. This operates by way of a of probe energization can be dramatically and Smith 1981). However, Eq. 1 was
LABSOFT command to the A/D, D/A affected by the initial temperature found to be highly linear (linear/?2 = 0.95
converter producing a latched binary out response of the probe, the range of time- 0.99) for extended segments of the

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 203

 INITIALIZE VARIABLES AND INPUT DESIRED EXPERIMENTAL CONDITIONS
AND KNOWN CONSTANTS

MONITOR PRODUCT AND CONSTANT-TEMPERATURE BATH TEMPERATURES

UNTIL THERMAL EQUILIBRIUM IS ESTABLISHED

BEGIN PROBE ENERGIZATION AT THE PRESCRIBED VOLTAGE AND RECORD

RAW DATA AT SPECIFIED TIME INCREMENT UNTIL TOTAL TIME HAS ELAPSED

NO YES

INPUT PROBE N.1 TIME INPUT THE TRUE THERMAL

CORRECTION FACTOR CONDUCTIVITY

CALCULATE PROBE N.1 CALCULATE THE TIME

HEATING CURVE SLOPE, CORRECTION FACTOR
PROBE HEAT STRENGTH, BYTRIAL-AND-ERROR
AND THERMAL AND LEAST SQUARES
CONDUCTIVITY REGRESSION

INPUT STARTING ESTIMATE OF THE SAMPLE SPECIFIC HEAT (Cp) ANDTHE

CORRECTED PROBE SEPARATION DISTANCE

CALCULATE THE PROBE N.2 TEMPERATURE RISE FOR THE LAST TIME VALUE
USING THE SPECIFIC HEAT ESTIMATE; COMPARE THIS VALUE TO THE RECORDED
VALUE, AND REPORT THE ESTIMATION ERROR

INPUT NEW STARTING

ESTIMATE FOR
SPECIFIC HEAT

CALCULATE THERMAL DIFFUSIVITY BY NONLINEAR LEAST SQUARES

REGRESSION; DETERMINE 95% CONFIDENCE LIMITS, AND DISPLAY
FINAL RESULTS

INCREASE NO.
DATA POINTS FOR
REGRESSION

WRITE RESULTS TO OUTPUT ( FINAL REPORT )

Figure 3. Logic flow diagram ofprogram used to control the thermal property measurement system and to analyze the experimental results.

204
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 TABLE I. NOMENCLATURE
curve following the first 60-120 sec for the
Variable Definition Units selected range of experimentaldesign con
E Euler's constant = 0.5772157 ditions (i.e. probe spacing, heat source
F Frequency statistic of the F-distribution strength, etc.).
f Predicted temperature-rise using Eq. 1 C The confidence limits so developed
k Thermal conductivity W/(mC)
m Sample size provide the user with a measure of the
n Positive integers uncertainty in the results owing mainly to
Q Heat source strength (W/m) the imprecision of the temperature mea
r Distance from line heat source m surements. In order for the user to resolve
R2 Coefficient of determination (0-1)
S Slopeof temperature-rise vs. In (0 - 60) and correct for systematic errors, however,
for heating probe C
the user must first calibrate the apparatus
T Temperature rise C using a substance of known thermal con
Observed (/th) temperature-rise C ductivity and diffusivity. In this regard an
f Sample arithmetic mean temperature-rise C optional part of the computer program
Thermal diffusivity m2/s allows the user to automatically determine
Bestfit solution of thermal diffusivity using
Eq. 1 the time correction factor for the probe
m2/s
P r/(2(a0),/2) apparatus. In this operation the user
7 Probability level for null hypothesis provides the true (known) thermal con
(0.05 for 95% confidence limits) ductivity of a test material such as 0.5%
e Time from initiationof line heating
e0 Time correction factor
s
agar-water gel and the program finds, by
s
a Sum of squares oc2 trial and error, the time correction factor
for the probes that permits a match, to a
user-specified precision (0.01%, say) of
the measured and true thermal con
ductivity.
TABLE II. PROGRAM OUTPUT FILE PRODUCED BY THERMAL CONDUCTIVITY/
DIFFUSIVITY MEASUREMENT SYSTEM
Equation 1 is highly sensitive to probe
separation distance, and it will be neces
THERMAL PROPERTY REPORT DATE: 86-08-20 sary to either correct or check for minor
Test sample: 0.5% AGAR Calibration run
Datafile: A:TEST1.DAT
disturbances from true parallel alignment
Applied pressure (kPa) = 0.00
of the probes, or for off-center position of
RAW DATA: the thermocouples within each probe. The
Heating Passive corrected probe separation distance can be
Data set Time Voltage Heat Bath probe 1 probe 2 7-Rise obtained by trial-and-error calibration
No. (sec) (V) (W/m) (C) (C) (C) (C) using a homogeneous substance of known
0 0. 0.000 0.000 28.1 28.2 28.2 0.0 thermal diffusivity. Systematic or random
6 30. 9.990 28.920 28.1 40.7 28.2 0.0
12 60. 9.980 28.868 28.1 42.9
errors which cannot be corrected directly
28.3 0.1
18 90. 9.970 28.813 28.1 44.3 28.4 0.2
through calibration must be estimated by
24 120. 9.960 28.758 28.1 45.2 28.6 0.4 the usual experimental replication and sub
30 150. 9.950 28.702 28.1 45.9 28.9 0.7 sequent calculation of a mean and standard
36 180. 9.950 28.703 28.1 46.4 29.1 0.9 error for a and k.
42 210. 9.940 28.646 28.1 46.9 29.4 1.2
48 240. 9.930 28.590 28.1 47.3 29.6 1.4
As a means of assessing the relative
54 270. 9.930 28.591 28.1 47.7 29.8 1.6 degree to which the experimental passive
60 300. 9.920 28.534 28.1 48.0 30.0 1.8 probe data fits Eq. 1(for a), and the degree
66 330. 9.920 28.535 28.1 48.3 30.2 2.0 to which the heating probe data fits Eq. 4
(for k\ a coefficient of determination (R2)
Averages 9.949 28.695 28.1
value is calculated for the nonlinear and
SUMMARY:
Conductivity (W/mC) = 0.6115
linear regressions. Of the eight alternative
Density (kg/m3) = 996.49 and uniquely different expressions for R2
Specific heat (J/(kg C) = 4161.84 identified by Kvalseth (1985), the follow
(4137.80-4186.17) 95% CL* ing recommended form was selected for
Diffusivity (m2/sec) = 0.1474E-06
both regressions:
(0.1466E-06 - 0.1483E-06) 95% CL*
* CL = Confidence limits
Conductivity parameters
i - X if, - Tf-/Z (t - T)2 (6)

Slope of T vs. log (t) curve (C) = 3.73

Temperature intercept (C) =27.1
Time correction factor (s) = - 9.0417 RESULTS AND DISCUSSION
First and last points used 5 & 26 Typical program output is presented in
Coefficient of determination = 0.99845 Table II. Using 0.5% agar-water gel as a
Diffusivity parameters test substance, it was found that a suffi
Total test duration (s) = 360.0
cient passive probe temperature rise (e.g.
Distance between probes (m) = .009998
No. data points used = 25 2C) could be obtained to enable accurate
No. terms used in series solution = 45 thermal diffusivity measurements in from
Temperature rise error sum of squares = 0.207E-01 5 to 6 min using a heating strength of
Coefficient of determination = 0.98795 26 W/m and 1 cm probe separation.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 205

 2.2-j

2.1-

2.0-

1.9-

5 1.7H
<
QC

i.bH
5

1.5-

1.4-

1.3-

1.2-

m ri 111TtiniM i ii m 11 m i rnii 1111 ii i rji 11 m 11111 fi m int 1111111 riij 1111111 ii j11111111111 ii 11 m ii 11 n mi 11 j11111 m 11 j11 ni 111 i j11111 n 11 j111111 n 11 n 11 n 1111

200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360

TIME (s)

Figure 4. Experimental and predicted temperature rise of the passive probe as a function of heating time.

However, the optimum heating time could predicted temperature rise curve. Figure 4 damental electronic components into a sin
vary significantly depending upon the pro shows the two temperature rise profiles gle package, with the attendent advantages
bes used, their separation distance, the (dashed lines) for the upper and lower 95% of lower system cost and portability.
heating power applied, and the substance confidence limits of a. These curves are However, the system can be easily
being measured. Table II reports all rele not to be confused with 95% confidence assembled from components which may
vant information pertaining to the calcula contours for the temperature-rise versus already exist in many well-equipped ther
tion of the thermal conductivity and time data. The approximately uniform pat mal process development laboratories.
diffusivity respectively. The R2 value for tern of the temperature-rise error residual
the heating probe temperature versus loge (measured-predicted) indicates that Eq. 1 ACKNOWLEDGMENT
(time) curve was found to be no less than provides a good fit to the data, and The author is grateful for technical support
0.998 for the tests on 0.5% agar water gel provides evidence of good equipment per provided by Dave Dow of the Engineering and
Statistical Research Centre (E.S.R.C), and for
using 20 data points. Because the tem formance.
assistance provided by associates and staff of
perature rise for the passive probe is low It was found that approximately 12itera E.S.R.C. on this project.
compared to that of the heating probe, and tions are required to obtain a to within
because ofthe limited precision ofthe tem 0.1% of the "best-fit" solution using 50
perature measurements (0.1C), the terms of Eq. 1 and 20 temperature-rise vs. REFERENCES
number of data points required for accurate time data pairs. CHANG, C. S., F. S. LAI, and B. S. MILLER.
diffusivity determination was found to 1980. Thermal conductivity and specific heat
increase with decreasing test time or with SUMMARY
of grain dust. Trans ASAE (Am. Soc. Agric.
Eng.) 23: 1303-1306, 1312.
decreasing overall temperature rise of the An automatic system for the measure
DRAPER, N. R. and H. SMITH. 1981.
passive probe. Table II provides the ment of thermal conductivity and dif Applied regression analysis. 2nd ed. John
residual sum of squares value and 95% fusivity has been successfully developed. Wiley and Sons, Inc., Toronto, Ont.
confidence limits for this regression. Fig The development system reported here HIMMELBLAU, D. M. 1972. Applied non
ure 4 shows the last 25 measured tem could be further refined and redesigned linear programming. McGraw-Hill Book
perature rise values from the data towards production of a more compact Co., Toronto, Ont.
presented in Table II, together with the commercial system by combining the fun HOOPER, F. S. and A. R. LEPPER. 1950.

206 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 Transient heat flow apparatus for the deter-
mination of thermal conductivity. Am. Soc.
Heat. Vent. Eng. Trans. 56: 309-324.
KVALSETH, T. O. 1985. Cautionary note
about R2. The American Statistician 39(4):
279-285.
LOWAN, A. N. 1940. Tables of sine, cosine
and exponential integrals. Vol. 2. New York
Math Project, National Bureau ofStandards,
Washington, D.C.
MOHSENIN, N. N. 1980. Thermal properties
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
of foods and agricultural materials. Gordon
and Breach Science Publishers, New York.
McGINNIS, D. S. 1986. Prediction of transient
conduction heat transfer in foods packaged in
flexible retort pouches. Can. Inst. Food Sci.
Technol. J. 19(4): 148-157.
NIX, G. H., G. W. LOWERY, R. I. VACHON,
and G. E. TANGER. 1967. Direct deter-
mination of thermal diffusivity and con-
ductivity with a refined line-source tech-
nique. Progress in aeronautics and astro-
nautics. Thermophysics of Spacecraft and
Planetary Bodies 20: 865-878, Academic
Press, New York.
SINGH, R.P. 1982. Thermal diffusivity in food
processing. Food Technol. 36(2): 87-91.
TIMBERS, G. E. and G. D. ROBERTSON.
1977. An inexpensive thermal conductivity
probe for in-situ measurements with foods.
Report No. 7404-614 ofthe Engineering and
Statistical Research Centre, Research
Branch, Agriculture Canada, Ottawa, Ont.
207
 TECHNICAL NOTE

A CURVE-FITTING PROGRAM TO STRESS RELAXATION DATA

R. P. Rudra

School of Engineering, University of Guelph, Guelph, Ontario NIG 2W1

Received 18 December 1985, accepted 10 October 1986

Rudra, R. P. 1987. A curve-fitting program to stress relaxation data. Can. Agric. Eng. 29: 209-211.

A computer program has been developed for the analysis of stress relaxation data using the method of successive
residuals. This technique eliminates the use of graph paper and eyejudgement for developing a multi-term exponential
stress relaxation equation. This program hasthepotential to select best fit multi-term exponential equations to exponential
decay data.

INTRODUCTION
Viscoelastic materials show a stress
relaxation phenomenon which is one ofthe
most important parameters in characteriz
ing materials (Ferry 1961). In stress relaxa
tion the test specimen is suddenly brought
to a given strain, and the stress required to
hold the deformation constant is measured
as a function of time. The results are ex
pressed in terms of a time-dependent mod
ulus^/) in tension or compression, G(t) in
shear, or K{t) in bulk compression. These
parameters can be used either in direct
form or in their reciprocal form called
compliance in viscoelastic analyses.
One rheological model representing
stress relaxation is the generalized Max
well model as illustrated in Fig. 1. If this
model is subjected to constant strain e0 at
time t = 0, the total stress a in the model is
given by
<T = CT| + 0"2 + O3 + + CTn + (Tc
where ae is equilibrium stress.
The decrease in stress with time can be
represented by
v(t) = e0 [Edl e_,/T| + Ed2 e",/T2 + ...
+ dn e",/Tn + Ec] (2)
where t,, t2, ... Tn are relaxation times,
TreI, corresponding to various elements in
the model; E0 is the initial modulus; Ed is
thedecaymodulus (E0 - e); and Eeis the
equilibrium modulus.
To predict stress at any time, /, the stress
relaxation curve must be represented by a
mathematical expression depicting the
behavior of the material under constant
strain. The plot of the logarithm of stress
versus time is generally nonlinear. It has
been found that a three-term exponential
equation is usually a reasonably good fit
for practical purposes (Mohsenin 1978).
Various graphical techniques outlined by
Mohsenin (1978) and numerical analysis
methods given by Moore (1974) can be
used to express the relaxation curve in
terms of exponential functions. The most
commonly used method is that of suc
cessive residuals. This method, however,
is rather time consuming, and the accuracy
using graphical technique depends upon
personal judgement. It is the purpose of
this study to develop a suitable computer
program to be used to analyze the stress
relaxation data by the method of suc
cessive residuals.
DESCRIPTION OF THE
COMPUTER PROGRAM
If the stress relaxation is expressed in
the general form of
a(r) 0\. + 2/ 0~n
(3)
then the computer program begins by
looking at either slope or change in stress
values corresponding to equal time inter
(1)
vals in order to obtain the straight line
portion of the log-stress versus time curve
to compute an and Tn of Eq. 3 from the
slope and ordinate intercept, respectively.
The program takes the stress relaxation
data and converts the stress or stress relax
ation modulus into a logarithmic form.
The semi-log curve is then divided into a
large number of equal time segments. The
cubic spline function (Erh 1972) is used to
look at the incremental slopes or change in
mi1?, igg^ igi^ nig
Figure 1. Generalized Maxwell model representation.
stress values for equal time intervals. The
spline function is also available in a sub
routine form in the International, Mathe
matical and Statistical Library (IMSL)
package. The incremental slope or change
in stress values and time arrays are scan
ned backward to arrange them in a
decreasing order of time. Equal incremen
tal slope or change in stress values for
constant time intervals will detect the
straight line portion ofthe stress relaxation
curve. The computer, however, performs
very accurate analysis resulting in very
large numbers of a and t. Since, for most
practical purposes, three exponential func
tions are adequate for representing stress
relaxation data, the problem is solved by
taking the difference in the adjacent incre
mental slopes or change in stress values
and comparing them with a very small
value of tolerance. After obtaining the
points for the straight line portion of the
log-stress versus time curve, the data
points are arranged in an increasing order
of time, and the equation of the straight
line portion is obtained by simple linear
regression. The slope of the straight line
portion gives time constant t, and the
intercept on the ordinate gives the coeffi
cient of the exponential term a,. The first
residual is obtained by taking the dif
ference between the original stress versus
-!*

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 209

 TABLE I. STRESS RELAXATION DATA
FOR WHEAT DOUGH 1000-
(CUNNINGHAM ET AL. 1953)
900-
Time Tension
800- WHEAT DOUGH
(S) (g)
700 4 >
0.0 710.0
1.0 560.0
600 -
2.0 487.0 _____ Experimental
4.0 420.0
6.0 383.0 500 - \# Fitted Values

8.0 355.0
10.0 334.0
321.0 00
12.0 400 -
14.0 309.0
2
16.0 298.0 o

18.0 288.0 CO

W
300 _
20.0 280.0
H

time curve and the straight line portion

extrapolated to the ordinate. This pro
cedure was repeated two more times to
a(t) - 385.82e-t/65-469 + 262.29e"t/3-4464 + 82. 2ISe"^1'4186
200
obtain a2 and cr3, and t2 and t3. The
following steps describe the procedure in
detail.
(1) Read the data of stress relaxation.
(2) Compute the natural logarithm of
<>
stress or stress relaxation modulus values.
(3) Set the index value equal to the
number of exponential terms required. i r i

(4) Set the values for tolerance. 12 16 20

(5) Divide the time data into about 50
equal segments. TIME, s
(6) Use the spline procedure to com Figure 2. Comparison of experimental andcomputed values for stress relaxation v(t) for wheat
pute the incremental slopes (DS) or change dough. (Data from Cunningham et al. 1953).)
in stress values (DY) for equal time inter
vals.
RESULTS AND DISCUSSION TABLE II. STRESS AND TIME DATA FOR
(7) Compute the difference between THE FIRST AND SECOND RESIDUAL AS
The application of the method
incremental slope (DIF) or change in stress PREDICTED BY THE COMPUTER
developed has been described by an illus
values (DIFF) for equal time interval I and PROGRAM
trated example for wheat dough. The data
I + 1 as: First residual Second residual
for this example, given in Table I, were
DIF = DS (I + 1) - DS(I) taken from Cunningham et al. (1953). The Time Tension Time Tension
DIFF = DY (I + 1) - DY(I) (S) (S) (g)
analysis was performed by assuming a (g)

where DS = incremental slope, and zero value of equilibrium stress. The data 0.0 344.20 0.0 81.89

DY = stress value. representing the stress versus time rela 0.41 275.60 0.36 46.99
0.82 219.50 0.72 18.85
(8) Compare the difference (DIFF) tionship were divided into 50 equal time
1.63 149.80 2.15 16.32
with tolerance. segments. With a tolerance value of 0.005, 2.86 99.26 2.87 15.19
(9) Take the points of log-stress and the program generated values of 385.82 3.67 79.45 3.22 13.49

time for which DIFF or DIF is less than and -65.469 for a, and t,, respectively. 4.90 65.02 3.94 9.30
7.27
tolerance to obtain the points for the The corresponding stress values for the 5.71 48.62 4.30
6.53 39.70 4.66 5.45
straight line portion of the log-stress ver first residual are given in Table II. The 7.35 31.85 5.02 3.83
sus time curve. repetition of the procedure resulted in 8.57 21.94 5.37 2.44

(10) Compute the values of a and t by values of 262.29, 82.218, -3.4464 and 9.39 16.81 5.73 1.30

simple linear regression. -1.4166 for a2, a3, t2 and t3, respec 10.20 13.22 6.09 0.43
11.43 10.36 6.45 0.17
(11) Compute the differences (STMD) tively. The stresses associated with the sec
12.65 8.19
between the original stress and the stress ond residual are also shown in Table II. 13.47 6.60
Na obtained from the straight line to The curve and equations developed from 14.69 4.48

obtain the first residual.
STMD(I) = a (I) - Na (I)
(12) If STMD is less than or equal to
zero, then omit that value.
(13) Compute the natural logarithm of
the STDM values.
(14) Repeat steps 4-11 two more times
to obtain terms for the second and third
exponentials.
this analysis are shown in Fig. 2. Com 15.51 3.29
parison with the experimental data yielded 16.73 1.68
17.55 0.76
very good agreement with the fitted
values, with a determination coefficient
(r2) or 0.991. Statistical analysis indicated was used by manipulating the equilibrium
no significant difference between the fitted stress or modulus (ae or Ee) in a(0 or E(t).
values and experimental curve at the 0.05 Analysis performed on wheatdough witha
level. value of 40 g for ae and a value of 0.005 for
Some food materials do exhibit equi tolerance predicted the following equa
librium stresses. In such cases this method tion.

210
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

1000
EXPERIMENTAL VALUES, kPa
Figure 3. Comparison between the experimental and computer fitted values for stress relaxation
for food materials.
a(t) = 327.74e* + 258.86e~
+ 8l.56eT407 + 40.0
Comparison of this equation with the
experimental curve yields a determination
coefficient of 0.991. In this case, also,
there was no significant difference
between the fitted and experimental values
at the 0.05 level.
To check the application further, this
method was applied to stress relaxation
data for bovine muscle, apple flesh and
cheese. The experimental data reported by
Morrow and Mohsenin (1969) for bovine
muscle, Morrow et al. (1971) for apple
flesh and Shama and Sherman (1973) for
cheese were used. In all cases the fitted
equation compared very well with the
experimental data. The lowest determina
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
2000 3000 4000
tion coefficient of 0.984 between the fitted
and experimental values was obtained for
(4)
bovine muscle. To test the overall perfor
mance of this method the comparison was
made between the pooled fitted and experi
mental values for all food materials
reported earlier. The results presented in
Fig. 3 show an excellent agreement with a
determination coefficient of 0.996. This
line was not significantly different from
the equal value line at the 0.05 level. Com
parison of this method with the graphical
technique of stress relaxation given by
Mohsenin (1978) also did not show any
significant difference at the 0.05 level.
CONCLUSIONS
On the basis ofthe comparison made for
different food materials, it was found that
the technique of fitting the computer curve
to the stress relaxation data, based on the
method of successive residuals developed
in this study, is an easy and reliable
method for determining the exponential
functions representing given stress-relaxa
tion data. This method can handle any
number of exponential terms depending
upon the precision and accuracy required.
This method can also be used to obtain the
best exponential curve fit by calculating
the linear determination coefficient
between the values obtained for a different
number of exponential terms and the
experimental data, comparing the fitted
values with the experimental data, taking
the one with the highest determination
coefficient and not significantly different
from the experimental data as the best fit
curve. The method is simple, accurate and
less time consuming.
CUNNINGHAM, J. R., I. HLYNKA, and J. A.
ANDERSON. 1953. An improved relax-
ometer for viscoelastic substances applied to
the study of wheat dough. Can. J. Technol.
31:98-108.
ERH, K. T. 1952. Application of spline func
tion to soil science. Soil Sci. 114(5):
333-336.
FERRY, J. 1961. Viscoelastic properties of
polymers. John Wiley and Sons, New York.
MOHSENIN, N. N. 1978. Physical properties
of plant and animal materials; structural,
physical characteristics and mechanical
properties. Gordon and Breach Science Pub
lishers, New York.
MOORE, E. 1974. Experimental fitting using
integral equations. Int. J. Num. Meth. Eng.
8: 271-276.
MORROW, C. T. and N. N. MOHSENIN.
1969. Transient mechanical response in
bovine muscle. ASAE Paper no. 69-562.
American Society of Agricultural Engineers,
St. Joseph, Mich.
MORROW, C. T, D. D. HAMANN, N. N.
MOHSENIN, and E. E. FINNEY, Jr. 1971.
Mechanical properties of Red Delicious
apple. ASAE Paper no. 71-32. American
Society of Agricultural Engineers, St.
Joseph, Mich.
SHAMA, F. and P. SHERMAN. 1973. Stress
relaxation during forced compression studies
on foods. J. Text. Stud. 4(3): 353-362.
211
TITLES AND ABSTRACTS OF PAPERS FOR PRESENTATION AT THE 30TH MEETING OF THE
CANADIAN SOCIETY OF AGRICULTURAL ENGINEERING, MONTREAL, MAY 18-22, 1987
87-101 CONTROLS AND ROBOTICS
GREENHOUSES
R. Kok, Dept. Agric, Eng., Macdonald College
of McGill University, Ste, Anne de Bellevue,
Quebec H9X ICO.
87-102 HIGH PRESSURE SODIUM LIGHTING
EFFECTS ON GREENHOUSE HEATING
REQUIREMENTS
D. Brault, R. Boily, Dept. Agric. Eng.,
A. Gosselin, Dept. Plant Sci., and C.
Gueymard, Centre de Specialisation en
Cultures abritees, Laval University, Ste. Foy,
Quebec G1K 7P4.
Greenhouse crops grown with high levels of
supplementary lighting require lighting systems
with electrical loads of up to 150 W/m2 of
floor area. The energy consumed by these
systems is converted to heat and light which
can provide a significant part of the
greenhouse heating requirements. A 6 by 30 m
bay of a gutter connected greenhouse,
equipped with a high pressure sodium lighting
system (60 400 W lamps) for the year-round
production of tomatoes, was monitored
throughout a complete heating season. The
lighting system had an electrical load of
2 7.8 kW and was operated during off-peak
hours in order to take advantage of lower
rates. Heating requirements were met with a
72 kW electrical furnace also operating during
off-peak hours, and a 72 kW oil-fired furnace
was used the rest of the time. Off-peak hours
were between 0:00 and 6:30 and between 6:30
to 24:00 when the outdoor temperature was
above -17C.
The heat balance of the greenhouse was
established in order to determine
contribution of the lamps to the heating
requirement of the building. The heat
generated by the lighting system was found to
supply a significant part of the structure's
heating requirements, while off-peak utility
rates reduced total heating costs.
87-103 CONCEPTION AND DEVELOPMENT
OF AN INTELLIGENT GREENHOUSE
CONTROL - OPERATING SYSTEM
R. Kok, G. Desmarais and J. Larouche, Dept.
Agric. Eng., Macdonald College of McGill
University, Ste. Anne de Bellevue, Quebec
H9X 1C0.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
IN
An intelligent greenhouse was conceived,
defined and partly designed to comprise
physical and virtual machines organized
heir ar chially to mimick biological
arrangements. To deal with the complexity of
the biological machine, its control structure
was analyzed in terms of three types of
functions; prevention, assurance and task
execution; and four major levels of control;
physical, instinctive, Pavlovian and intelligent.
The greenhouse control subsystem was then
created by defining the goals, the functions
and the mechanisms, and incorporating these
into the various virtual and physical
components. The virtual components are
resident in three linked computers which can
be instructed to evoke a number of control
modes of behaviour patterns of different
intelligence levels. The physical and virtual
components are described as well as the
structure of the control system.
87-104 SOLAR ENERGY STORAGE IN WET
SOIL FOR GREENHOUSE
APPLICATION
H. Bernier, Inst. Tech. agr.alim., La Pocatiere,
Quebec, G.S.V. Raghavan, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X 1C0, and
J. Paris, Dept. Chem. Eng., Ecole
polytechnique, Montreal, Quebec H3C 3A7.
By increasing the surface between air and
greenhouse soil, the latter can become a
relatively low cost heat storage material, and
temperatures can be increased in the plant
root zone. A Nordic greenhouse (Harnois
Industries, St. Thomas, Quebec) was built in
the an east-west orientation at La Pocatiere in
order to evaluate the thermal efficiency of a
soil heat exchanger and storage system. The
system was made of 26 102 mm unperforated
corrugated plastic drain tubes of 12 m length
and buried half each at depths of 450 and
7 50 mm along the greenhouse length at a
450 mm spacing. A 0.75 kW blower circulated
hot air collected in the greenhouse through
the tubes at a rate of 0.9 m3/s. The soil
storage had a seasonal temperature fluctuation
of 10 to 12C and the average soil
temperature correlated very closely with the
average greenhouse air temperature. The
performance of the system appeared to be
more influenced by greenhouse air temperature
213
rather than by incident solar radiation, and
improvements were observed in crop growth
and yields. Results indicate that solar energy
contributed 58% of total heating requirements
from February to June and from September to
December, thus representing approximately a
33% energy conservation. The system payback
period might be as low as two years,
depending on operating conditions and initial
cost.
87-105 DESIGN CONSIDERATIONS FOR AN
AGRICULTURAL ECO-CYBORG
R. Kok and L. Gauthier, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X ICO.
The term cyborg is usually applied to a
machine which has both biological and
mechanical or electrical components. An
eco-cyborg is an ecological system consisting
of biological, mechanical, electronic and
virtual components. The design of a food
producing eco-cyborg will be necessary to
support life in regions of the planet or
universe normally inhospitable to humans.
In order to have a dynamically stable,
compact and efficient eco-cyborg, it will be
necessary to rely on a number of machines or
species. In many instances it may not be
possible to accomodate a large number of
species, especially if the system is to be
production oriented. A small number of
species will produce an unstable system,
however the eco-cyborg can be stabilized by
the creation or addition of virtual species
which are resident on a computer, but of
which the effects are real. Data or knowledge
bases and artificial intelligence technologies
will be required to support such functions as
cognition, memory, strategy generation and
assessment, pattern recognition and adaptive
behaviour. The system will also rely heavily
on sensors, robotics, simulation, modelling and
the understanding of interrelations between
biological organisms and their environment.
The paper attempts to lay out a
conceptual framework in which to define and
describe this technology. It also examines a
number of design concepts applicable to
system implementation, and the possible
evolutionary value for humans of the
agricultural eco-cyborg.
87-106 DESIGN CRITERIA FOR
WATERPROOF CONCRETE
CYLINDRICAL TANKS AND SILOS
214 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
J.C. Jofriet, School of Eng., University of
Guelph, Guelph, Ontario NIG 2W1.
87-107 LAMELLAR BIN GRAIN PRESSURE
RESEARCH DEVICE
M.G. Britton, Dept. Agric. Eng., University of
Manitoba, Winnipeg, Manitoba R3T 2N2.
Model lamellar bins have been used in grain
pressure research at the University of
Manitoba for a number of years. The grain
mass within this unique construction system
reacts to changes in the state of stress during
emptying by permitting grain flow through the
slatted sides of the bin. Studies have been
conducted to consider the effect of bin
emptying rate, filling rate, emptying port
location, bin size, fraction on the bin slats
and variation with height above the bin floor.
A summary of the results to date is
presented, together with a speculation on their
implications for the design of grain storage
bins.
87-108 STRUCTURAL RELIABILITY OF
NAILED CONNECTIONS FOR
CANADA PLAN SERVICE TRUSSES
D.J. Masse, Eng. and Stat. Res. Centre,
Agriculture Canada, Ottawa, Ontario K1A 0C6,
and J.J. Salinas, Carleton University, Ottawa,
Ontario K1S 5B6.
Results of load tests on nailed joints carried
out previously by the Canada Plan Service
design center were analyzed with a new
technique to better establish the safety
(probability of failure) and serviceability of
nailed connections. This technique considered
the different failure modes that a nailed joint
can experience. It also considered the
uncertainties relating to the strength of
materials (plywood, lumber, nails) and to the
maximum lifetime applied load.
87-109 TESTING OF FIVE SYNTHETIC
FLEXIBLE BIOGAS STORAGE BAGS
M. Cournoyer, R. Racine Urgel Delisle Assoc,
St. Charles sur Richelieu, Quebec J0H 2GO,
and R. Chagnon, Agriculture Canada, St. Jean,
Quebec J3B 6Z8.
Five biogas storage bags of 50 m3 capacity
and made of CPE, XR5, PVC-Vintex,
PVC-Sarnafil and Hypalon were tested over a
14 month period. The PVC-Vintex and 0.91 mm
Hypalon did not prove to be suitable materials
as they required many major repairs at very
low gas storage pressures. However, their
performance could be greatly improved by
using a thicker membrane, thus increasing the
cost. Hypalon was the most problematic
material, and required a special cement for
gluing the repair material. PVC-Vintrex had a
multitude of pinholes at the end of the test,
while CPE was more reliable and could sustain
higher pressures. Hypalon tended to tear into
long holes and the seams tended to be weak
points.
XR5 and PVC-Sarnafil provided the best
results. They sustained pressures over 22 cm
water, showed very few holes and did not
develop tears. They were easily repairable
with rubber or PVC patches. Although the
XR5 material was half the price of the
Sarnafil, the finished bags had the same cost
due to fabrication techniques. Ice inside, or
snow outside the bags did not affect their
performance for storing biogas.
With an adequate selection of materials,
this type of gas storage is practical and cost
competitive for a farm digester. However, it
does require a minimum of protection against
heavy winds and some anchorage, or repairs
will be required each year no matter what the
material. The ease of repair techniques and
availability of repair material rank high in the
material selection criteria.
87-110 OPTIMIZATION OF SILAGE PLASTIC
FILM
P. Savoie, Agriculture Canada, Laval
University, Ste. Foy, Quebec G1K 7P4.
Because of the high cost of tower silos, a
number of alternate silage storage methods
have been developed, such as bunker silos,
unwalled stack silos, bagged chopped silage
and bagged round bale silage. The alternate
methods all use polyethylene film to ensure a
proper seal and the anaerobic conditions
necessary for good fermentation and
preservation. However, the plastic film is
vulnerable to weather conditions such as hail,
wind, frost and sunlight, as well as accidental
damage caused by rodents, birds or improper
handling. A low-cost film may be the cause of
important silage losses, whereas an expensive
high quality film is expected to minimize
losses. The objective is to minimize the total
costs of the plastic film and the silage losses.
Silage loss data from small 20 kg capacity
bags of 100, 150 and 200 />tm thicknesses is
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
presented. Losses were minimal from intact
bags, but increased considerably from
punctured or broken bags. A theoretical silage
loss model is proposed to take account of film
thickness, silage density, duration of storage
and the risk of film breakage.
Recommendations are made with respect to
the optimum film thickness according to the
silo type, the expected duration of storage
and the relative value of plastic and silage.
87-111 ENVIRONMENTAL CONTROL FOR
LIVESTOCK SHELTERS
E.M. Barber, Dept. Agric. Eng., Univ. of
Saskatchewan, Saskatoon, Sask. S7N 0W0.
87-112 THE PHYSIOLOGICAL RESPONSE OF
GROWING PIGS TO HIGH CYCLIC
AND CONSTANT TEMPERATURES
J.J.R. Feddes, Dept. Agric. Eng., University of
Alberta, Edmonton, Alberta T6G 2G6.
The physiological response of growing pigs to
high temperature conditions was investigated
with indirect calorimetry. One calorimeter was
maintained at a constant 33C air temperature
with an average dewpoint of 23C. The other
was varied sinusoidally between 26C at 0530h
and 40C at 1400h, for an average daily
temperature of 33C. Feed and water
consumption, oxygen, carbon dioxide and water
vapor concentrations, temperatures and air
mass flow rates were measured every four
minutes during the latter nine days of each of
the 12 day replicates. Animal physiological
parameters were observed also using a
portable data logger.
Differences were seen in pig respiratory
rates, rates of heat production, forehead
temperatures and other physiological
indicators.
87-113 COMPARISON OF MODULATED
VERSUS NON MODULATED CONTROL
SYSTEMS FOR SIDEWALL AIR INLETS
IN A NATURALLY VENTILATED
SWINE BARN
Y. Choiniere, F. Blais, Alfred College of Agr.
and Food Tech., Alfred, Ontario K0B 1A0, and
J.A. Munroe, Eng. and Stat. Res. Centre,
Agriculture Canada, Ottawa, Ontario K1A 0C6.
Two inlet control systems for a naturally
ventilated swine finishing barn were compared
on the basis of temperature regulation, C O2
215
and NH4 concentrations and electrical being critical to the design of a heat transfer
consumption. The barn was fitted with optimizing controller; the rate at which the
continuous above-centre pivot rotating doors in supply airflow is changed, the amount of heat
the sidewalls, and a continuous ridge opening. transfer degradation permitted before
The non modulated system used themrostats, triggering defrost and the maximum rate of
compressed air and air cylinders to open or heat transfer permitted immediately following
close the air inlets. During the winter, barn a defrost. Further development is warranted
temperature fluctuations of 8C were observed and needed to establish the critical control
within a 6 to 8 min period. C 02 parameters, as well as the installation and
concentrations fluctuated between 1500 and operation procedures.
3500 ppm depending on whether the inlets
were open or closed. NH4 remained rather 87-115 TEMPERING AIR USING AN AIR-SOIL
constant at 6 to 8 ppm. Electricity EXCHANGE SYSTEM
consumption was 157 kWh over a one year
period. The modulated control system used M.G. Britton, T.V. Murray, M.T. Burns, Dept.
thermostats, a gear motor and a time delay Agric. Eng., University of Manitoba, Winnipeg,
to step the inlets open or closed. During the Manitoba R3T 2N2, and M.A. Stumborg,
winter, barn temperature fluctuations of about Agriculture Canada Research Station, Swift
IC were noted at the level of the animals. Current, Saskatchewan S9H 3X2
C 0 2 concentrations varied between 2800 to
3200 ppm, while NH4 was 5 to 7 ppm. Research into the technical and economic
Electricity consumption was less than 1 kWh feasibility of tempering ventilation air for
during one year. controlled environment structures using soil at
shallow depths as an energy storage medium
87-114 OPTIMIZING HEAT RECOVERY IN AN has been underway since 1985. The research
AIR TO AIR HEAT EXCHANGER facility and experimental results are
OPERATING UNDER FROSTING summarized. The demonstrated air temperature
CONDITIONS tempering potential is outlined, and the
application to an energy intensive animal
M.R. Bantie and E.M. Barber, Dept. Agric. production building in both summer and winter
Eng., Univ. of Saskatchewan, Saskatoon, Sask. operation is discussed, along with a
S7N 0W0. preliminary design approach.

Air to air heat exchangers can be used to 87-116 HIGH PRESSURE VENTILATION
preheat ventilating air and hence reduce INLETS FOR ANIMAL HOUSING
heating costs for livestock barns. However,
frost accumulation is a major problem in this J.J. Leonard and F. Kloster, Dept. Agric.
application. Currently available frost control Eng., University of Alberta, Edmonton, Alberta
systems are based on some combination of T6G 2G6.
time, pressure loss, core temperature or
exhaust air temperature. They do not result in In order to obtain suitable trajectories of cold
an optimal rate of heat transfer, independent air jets for minimum ventilation during winter,
of barn temperature and relative humidity. high air velocities are required. Generally
The development and testing of a frost these velocities cannot be achieved using
control strategy is presented, based on a conventional exhaust fans and inlet designs.
measured instantaneous rate of heat transfer. One possible alternative is to use a high
The temperature rise of the supply air is pressure centrifugal fan to provide minimum
measured and the rate of heat transfer is ventilation during cold weather. Laboratory
controlled by positioning a damper to regulate measurements of air flows and distribution
the flow rate of cold supply air. Experiments using such a fan and a radially discharging
were conducted using a 470 1/s counterflow ceiling inlet are described. The results of
metal plate heat exchanger. Tests were these measurements are used to assess the
performed for a cold air temperature of -25C, feasibility of practical systems, and to suggest
and for return temperatures and relative possible designs.
humidities of 13 and 25C, and 40 and 75%,
respectively. The prototype controller was 87-117 PRODUCTION OF BIOGAS IN A 50 m3
confirmed to operate properly as predicted. DOWNFLOW FIXED FILM DIGESTOR
Three control parameters were identified as

216 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 R. Chagnon, Agriculture Canada, St. Jean,
Quebec J3B 6Z8, D. Morrissette, P. Toupin and
S. AUard, Centre de Recherche industrielle du
Quebec, Quebec.
Since 1982, Agriculture Canada has contracted
to le Centre de Recherche industrielle
Quebec the design, construction and testing of
a 50 m digestor for swine manure. It is an
anaerobic downflow fixed film reactor located
on a 1400 hog farm at St. Henri de Levis
near Quebec City. A new fixed film support
called 'digestubes1 was developed, made from
recycled plastic, and it provides a surface of
102 m2/m3 of digester. After 18 months of
operation, the digester was opened and a
digester bundle pulled out. The digestubes
were completely covered with a 2.3 mm thick
bio film, the interior of which comprised
mainly minerals, but the outside of which was
filled with bacteria.
The following average results
obtained: retention time was 5 days, COD
(inlet) 60,000 mg/1, COD removed between 45
and 50%, total solids 2.5%, biogas production
50m3/d and methane content in the biogas
70%. The results were similar to
obtained in a previous experiment using a
616 1 fixed film digester, except for COD
removal which was 5 to 10% higher in the
smaller digester. Since sampling errors may
have caused the difference, the results
obtained from the larger digester could have
been used to design the smaller one.
The biogas produced was utilised for both
electricity and heat generation after being
cooled to remove excess moisture and stored
in an 80 m3 hypalon bag. The hot water
produced was used to preheat incoming
manure, and some of the electricity was
returned to the utility grid. Some work on
hydrogen sulfide removal was also carried out.
The results indicated that anaerobic digestion
of swine manure in a bioreactor is a useful
tool for saving energy, while at the same
time reducing the manure pollutants. However,
this digester did not produce sufficient biogas
or electricity to render its operation cost
effective with respect to energy.
87-118 AN INTEGRATED SYSTEM
SWINE MANURE TREATMENT USING
SCREENING AND A SEQUENTIAL
BATCH REACTOR
L. Fernandes and E. McKyes, Dept. Agric.
Eng., Macdonald College of McGill University,
Ste. Anne de Bellevue, Quebec H9X 1C0.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Laboratory studies were conducted on a
sequencing batch reactor (SBR) for the
treatment of screened liquid swine manure.
The dry matter concentration in the raw
manure was in the order of 4%, and the SBR
was operated on a 24 h cycle. The influence
du of various parameters such as hydraulic
retention time, solid biological retention time
and anoxic/aerobic sequencing was tested. The
results showed that the SBR can treat swine
manure adequately under proper operating
conditions. The reduction of total suspended
solids and COD ranged between 90 and 98%,
and removal of NH4 was over 95%.
Furthermore, an alternating anoxic/aerobic
sequence during the fill and react phases
permitted a high rate of denitrification.
87-119 SEQUENCING BATCH REACTOR
TREATMENT FOR DAIRY MANURE
were K.V. Lo and P.H. Liao, Dept. Bio-Resource
Eng., University of British Columbia,
Vancouver, B.C. V6T 1W5.
Two 5 1 bench scale sequencing batch reactors
those (SBR) were used to treat dairy manure at
room temperature. The effect of the fillrreact
time ratio was studied with feed
concentrations maintained between 2000 and
3000 ppm of COD. Three operating cycles of
4, 3 and 2 h and three levels of feeding (3, 2
and 1 1) were tested. The best operating mode
was obtained in an operating cycle of 4 h
with a 2 1 fill, and over 90% B0D5 and 80%
COD removal were observed. The cycle of 2
h with a l l fill gave comparable results as
well, and both these modes could treat 12
1/day of wastewater. 24 1/day could be
handled by the 3 h cycle with 3 1 feed, or
the 2 h cycle with 2 1 feed, but the
treatment efficiency was lower. The results
indicate that the SBR can be an efficient and
reliable treatment method for dairy manure.
87-120 THE APPLICATIONS OF MEMBRANE
FILTRATION TO SILAGE EFFLUENT
A.P. Dunlea and V.A. Dodd, Dept. Agric. &
FOR Food Eng., University College Dublin, Dublin,
Ireland.
The range of silage effluent produced by grass
fermentation is 10 1 to 100 1 per tonne of
grass at 25 to 16% dry matter, including up
to 90,000 mg/1 BOD 5 and considerable
quantities of N, P, K, Mg, Ca, lactic acid and
217
carbohydrates. Current disposal methods are
normally land spreading or storage and
subsequent feeding. Neither of these is
practised widely in Ireland, thus causing silage
effluent to be the Country's largest
agricultural pollutant of surface and ground
waters.
Membrane filtration can be applied to
fractionate effluents into concentrates
feeding, requiring less storage, and a permeate
of lower polluting danger. A rig was built to
accomodate cellulose acetate and aromatic
polymer membranes using a DDS Lab 20 unit
and an operating pressure of 28-32 kPa.
Preliminary results indicate that the cellulose
acetate membrane gives a superior
performance, with an increase in dry matter
of 3.5 to 9.9% and better retention of the
components analysed.
87-121 BEEF FEEDLOT TREATMENT
VEGETATIVE FILTERS
R. Lagace, R. Adam, Dept. Agric. Eng., Laval
University, Ste. Foy, Quebec G1K 7P4, and
M. Vallieres, Min. de l'Env. du Quebec, Ste.
Foy, Quebec.
This paper describes the experimental site and
method for evaluation of beef feedlot runoff
treatment in the Quebec cold climate. A
settling basin and a vegetative filter were
designed for the treatment of a feedlot for 75
head. The pollution loads were measured at
the exits of the pen, the settling basin and
the vegetative filter. All input data such as
precipitation and flows, and the sampling
equipment were controlled by a microcomputer
and a data acquisition system. The results of
the fall of 1985 and 1986 show a removal of
more than 99% of most pollutants from the
feedlot runoff by this system.
87-122 INFLUENCE OF SEWAGE SLUDGE
APPLICATION ON HYDRAULIC
PHYSICAL PROPERTIES OF A SILTY
CLAY LOAM
E. Kodsi, S.T. Chieng and J. de Vries, Dept.
Bio-Resource Eng., University of British
Columbia, Vancouver, B.C. V6T 1W5.
The Ladner soil is a lowland soil which suffers
from a lack of hydrologic responsiveness, as is
reflected by ponding conditions in the fall and
winter. The possible improvement of the
structure of this poorly drained soil through
sewage sludge application could benefit
218
farming in the area. A greenhouse experiment
was conducted using columns of this soil with
application rates of 0, 33 and 100 t/ha of
composted sewage sludge. The soil columns
were subjected to periodic 24 hour simulated
ponding events. It was found that sewage
sludge increased the ponding tolerance of the
soil as reflected by significant differences in
for hydraulic conductivities of the columns. Fifty
days following the last ponding, the percent
stable aggregates averaged 13.7, 26.9 and
48.1% for the three sewage treatments,
respectively, and there wass a slight decrease
in soil bulk density with sludge application.
The effects of sewage sludge on essential
nutrient and heavy metal uptake by bermuda
grass were examined also, and there were no
differences in the foliage concentrations of P,
K, Cu, Ni or Pb. At the highest sledge rate,
however, N, Cd and Zn concentrations were
BY higher.
87-201 EFFECT OF TILLAGE REDUCTION
AND FERTILIZER TYPE ON CORN
YIELD
A. Weill and E. McKyes, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X 1C0.
A project was undertaken to evaluate the
feasibility of growing grain corn in Quebec
under reduced or zero tillage, together with
an organic source of fertilizer. Large plots
were established on sandy loam and clay soils
with conventional moldboard, reduced chisel
and zero tillage, and inorganic or dairy
manure fertilizer. The results obtained in 1985
and 1986 show that comparable yields can be
produced with reduced or zero tillage. The use
of manure reduced yields slightly compared to
those obtained with inorganic fertilizer.
Manure on zero tillage also caused difficult
weed control problems, seed toxicity and a
AND greater plant sensitivity to late spring frosts.
87-202 SOIL REACTING FORCES FOR A
REDESIGNED BENTLEG PLOW
H.P. Harrison and Z.J. Licsko, Dept. Agric.
Eng., University of Alberta, Edmonton, Alberta
T6G 2G6.
Using a six cell dynamometer, the soil
reaction forces in the longitudinal, lateral and
vertical directions were determined for three
half scale models of a bentleg plow rilling a
remolded laboratory soil. Two of the models
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
 had a 45 degree bentleg, with one having a
rake angle at the rilling interface and the
other not. The third model had a 30 degree
bentleg angle and an interface rake angle. The
design of the models was based on the
necessity of achieving considerable torsional
strength, and the models were subjected to a
static load test prior to use. The laboratory
experiment was conducted using a split plot
factorial design, the factors including two
depths and three spacings for each plow. In
addition to conducting several analyses of
variance of the data, the wrenches and power
spectral densities were determined and are
discussed. The penetration of the bentleg plow
was improved greatly by including a 15 degree
rake angle, and yet there was very little
change in the draft. One of the predominant
power spectral density frequencies was
attributed to the cyclic failure of the soil
ahead of the plow. The results were
sufficiently encouraging to recommend that
field tests should be conducted with full scale
models.
87-203 TRACTOR FUEL EFFICIENCY
DEMONSTRATOR
J.C. Chang and M. Green, Eng. Services Br.,
Alberta Agriculture, Edmonton, Alberta
T6H 5T6.
A Case-International 2394 farm tractor and a
25 ft field cultivator were instrumented with
sensors and an onboard data acquisition
system, including a computer and printer, in
order to determine drawbar power, fuel
efficiency and other variables during draft
test runs in the field. The unit was used to
demonstrate to farmers the savings in fuel
possible through gearing up - throttling down
on light loads, and proper tractor ballasting.
Improvements in energy output per unit of
fuel were an average 24% and up to 73%, and
savings in fuel per unit field area were an
average of 18% and a maximum of 44% by
gearing up. Ballsting up to 2200 kg resulted in
an average saving in fuel per field area of
17%, and up to 40 %. These de monstrations
have proved to be a very effective extension
method, and they will be used in the future
also to develop a data base of power
consumption and fuel consumption for tillage
operations at different locations across
Alberta.
87-204 PERFORMANCE STUDY OF THE BEN
BUCKET EXCAVATOR
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
H. Thibault, E. McKyes, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X 1C0, and
R. Theriault, Dept. Agric. Eng., Laval
University, Ste. Foy, Quebec G1K 7P4.
The Ben Bucket loader attachment for
agricultural purposes was evaluated, after
some design modifications, in its functions of
excavation, ditch digging, material spreading
and levelling and solid manure handling,
compared to other conventional machines. The
attache raent performed well, consistently
better than the backhoe, normal front end
loader or back blade depending on the task.
The modifications and evaluation of the Ben
Bucket included a design analysis of the
mechanical and hydraulic elements from the
static and dynamic points of view, the time
and energy efficiency during some common
operations compared to other machines and
the soil cutting abilities in different soil types
and consistencies.
87-205 A FARM SCALE WHOLE CROP
HARVESTING EXPERIMENT
R.P. Hoemsen, Inst, for Technological
Development, University of Manitoba,
Winnipeg, Manitoba R3T 2N2, and
R.P. Zentner, Agriculture Canada Research
Station, Swift Current, Saskatchewan S9H 3X2.
An experiment in whole crop harvesting was
conducted on a farm scale in 1985, following
two prior trials on a commercial scale in 1983
and 1984. Whole crop harvesting appeared to
be more technically feasible when practised on
the farm scale, but capacity was limited by
the drying of grain. If the whole grain and
straw crop is dried, economics become a
major concern. Successful application to
western Canada will therefore depend upon
the availability of a low cost crop residue
burner to take advantage of the available
straw as a fuel source, and the development
of applications for utilising the straw and
chaff for food and fibre as well as for fuel.
The 'thresh and dry' approach is most
technically and economically feasible. This will
require further development work in selecting
and modifying existing farm machinery.
87-206 EVALUATING THE TECHNICAL AND
ECONOMIC FEASIBILITY OF
MACHINE HARVESTING
STRAWBERRIES
219
E.M. Lauro, Horticultural Res. Inst, of
Ontario, Vineland Station, Ontario LOR 2E0,
and G.B. Hergert, Eng. and Stat. Res. Centre,
Agriculture Canada, Ottawa, Ontario K1A 0C6.
The performance of a prototype self propelled
combine style harvester was evaluated when
used to harvest strawberries. The harvester
was tested over a period of three years in
both a solid bed system and conventional
matted strawberry rows, each planted with
varieties Midway and V.6747R-6. Information
was collected to determine the pickup
efficiency, harvest losses, effective field
capacity and pickup yield. This information
was used for an economic assessment of the
harvester. A comparison of costs for the
prototype harvester, a harvest aid and
conventional hand harvest methods was made
to determine potential economic benefits of
using the mechanized harvest system.
87-207 A VERTICAL LIFT DIGGER FOR
HARVESTING POTATOES
CD. McLeod and G.C. Misener, Agriculture
Canada Research Station, Fredericton, N.B.
E3B 4Z7.
The concept of elevating potatoes, and
associated material such as soil, directly
behind the digging share improves the chances
of simplifying present day potato harvesters.
A prototype unit was built and evaluated, and
has proven to be successful for harvesting
potatoes damage free. The key component for
elevating potatoes at a 60 degree angle is a
weighted belt that operates on top of the
digger bed for the full height required to
convey potatoes into a bulk truck. The weight
of the belt holds the potatoes in place on the
digger bed while aiding soil separation from
the potatoes. The paper describes the vertical
lift digger, giving design criteria and
performance details of the machine.
87-208 DEVELOPMENT OF A TRACTOR PTO
TORQUE METER AND IMPLEMENT
CARRIAGE
C. Vigneault, G. St. Amour, D.J. Buckley,
Eng. and Stat. Res. Centre, Agriculture
Canada, Ottawa, Ontario K1A 0C6, P. Savoie,
Agriculture Canada, and D. Tremblay, Dept.
Agric. Eng., Laval University, Ste. Foy,
Quebec G1K 7P4.
A tractor PTO torque meter was developed to
220 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
measure the torque required by PTO driven
implements. A trailer was built with a three
point hitch to provide support for the
torque meter and to enable any three point
hitch implement to be monitored in situ. Tests
were conducted to compare the torque meter
data with that from two different
dynamometers and to calibrate the
torque meter. The system was designed to
gather data in the field using a digital
cassette recorder. This data can then be
transferred to a computer in the laboratory.
Programs were developed to analyse, graph
and present the data.
87-209 ELECTRONICS IN AGRICULTURE:
MORE PROBLEMS AND MORE
PRODUCTION
M. Potter, Varity Corp., Mississauga, Ontario.
87-210 AN INTELLIGENT HARVESTER FOR
STRAWBERRIES
M. Zellerer, D.J. Hebb and KJ. Wilkie, Dept.
Agric. Eng., Technical University of Nova
Scotia, Halifax, N.S. B3J 2M3.
Strawberries are a high value cash crop which
currently require intensive and expensive
manual labour to harvest. A system is under
development which employs robotic and
machine vision technologies to pick
strawberries, with the system operated by a
microcomputer having dual processing
capability. A digital camera vision system
performs the initial berry identification and
selection, and a special purpose manipulator is
used to harvest the berries. Digital control on
the manipulator positions the picking assembly
which orients berries and removes them from
the plant with minimum damage. Tests of the
system in the laboratory confirm the potential
for an intelligent strawberry harvesting system
87-211 DEVELOPMENT OF A DUAL FUEL
CONTROL SYSTEM FOR DIESEL
ENGINES
J.G. Davies, W.B. Reed and G.C. Zoerb, Dept.
Agric. Eng., Univ. of Saskatchewan, Saskatoon,
Sask. S7N 0W0.
A control system for dual fuel, using propane
or compressed natural gas was developed for
any four cycle diesel engine. A number of
engines have been converted to the system
and tested with a dynamometer to develop the
fuel control strategy, verify the control
system operation and in some cases by field
or highway teating. A map of maximum
gaseous fuel substitution over the operating
speed and load ranges was compiled for each
engine to develop the dual fuel control
strategies. Further development work is
required before this technology is proven and
ready for commercial use in diesel engines,
but it is important considering the number of
diesel engines which exist today.
87-301 PROBLEMS AND PROSPECTS IN
WATER QUALITY MODELLING
W.T. Dickinson, School of Engineering,
University of Guelph, Guelph, Ontario
NIG 2W1.
This review paper addresses nonpoint source
models which have been developed and used as
loading models for agricultural watersheds.
The following questions are explored with
reference to a number of specific models.
What is their stated purpose? What are the
input requirements? What outputs do they
provide? Of what use have they been to date?
The perspective of the group of models is
then presented including the identification of
achievements, gaps and possible new
directions.
87-302 HYDRAULIC CONSTRAINTS ON SIZE
DISTRIBUTIONS OF SOIL
TRANSPORTED IN INTERRILL FLOW
B.T. Guy, W.T. Dickinson, G.J. Wall and
R.P. Rudra, School of Engineering, University
of Guelph, Guelph, Ontario NIG 2W1.
Two groups of interrill sediment transport
capacity experiments were conducted using a
0.7 by 1.5 m long laboratory flume, artificially
roughened with a loamy sand soil glued onto
the inner surface. In the first group, uniform
laminar flows were produced from a constant
head tank, while in the second group, flows
were produced by a single nozzle rainfall
simulator centered over the flume. Three
slopes of 2, 9 and 20% and three discharges
or rainfall intensities of 0.5, 3.0 and 10.0
1/min were used in a three by three factorial
complete block design with two replications.
The loamy sand soil was injected into the
flow from an overhead hopper, and flow
velocity, water and sediment discharge and
transported sediment size distribution were
measured.
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
For both types of water flow, significant
differences were found between the size
distribution of available and transported
sediment in the cases of smaller slopes and
water flow rate. For the higher flows and
slopes the differences became insignificant.
For the uniform flow runs, a method was
developed to determine the flow competance
(diameter of the largest particle transported),
and these results compared favourably with
the Shields criterion. For the rainfall
simulations, the Shields criterion
underestimated the flow competance, and
alternate methods were examined.
87-303 NUMERICAL ESTIMATES OF THE
RAINFALL EROSIVITY FACTOR
CA. Madramootoo, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X 1C0.
The rainfall erosivity factor is a key element
in the universal soil loss equation. Estimates
of this index are very scanty in eastern
Canada. There are several empirical methods
for predicting the erosivity factor, but they
were developed in other climatic regions,
primarily in the United States. Thus there is
a need to test these methods locally in order
that recommendations on their reliability may
be made. Such results could lead to a better
prediction of rainfall erosivity factors for
different climatic locations, and eventually
produce the benefit of a more reliable use of
the universal soil loss equation.
87-304 LABORATORY VERIFICATION OF
GREEN-AMPT EQUATION
K.A. Tan, R.P. Rudra, W.T. Dickinson,
D.E. Elrick and G.J. Wall, School of
Engineering, University of Guelph, Guelph,
Ontario NIG 2W1.
Rainfall infiltration experiments into a three
layer soil system were conducted in the
laboratory under simulated rainfall conditions.
A low permeability layer was simulated using
a metal sheet having a known number of holes
to transmit water at a low rate. Tests were
performed using two rainfall rates of 16 and
39 mm/h, three depths to the low permeability
layer of 20, 50 and 100 mm and two types of
low permeability layers having conductivities
of 0.015 and 0.008 of the conductivity of the
overlying layer. A Green-Ampt infiltration
model was applied to this system, with the
221
hydraulic conductivity of the soil layers 87-308 SUBIRRIGATION AND S OIL
represented by their harmonic mean. Using the HYDRAULIC PROPERTIES
laboratory parameters, rainfall rate, wetting
front suction, thicknesses of layers, S.T. Chieng, Dept. Bio-Resource Eng.,
conductivity of layers and initial and University of British Columbia, Vancouver,
saturation water contents of layers as input B.C. V6T 1W5.
parameters to the model, the results compared
well to the experimental measurements. Subirrigation was conducted by using an
existing subsurface drainage system in a clay
87-305 SOIL CONSERVATION IN NEW having silty clay loam soil. Water table
BRUNSWICK fluctuations throughout the experimental
period were recorded by using automatic
J.L. Daigle, New Brunswick Dept. Agr. water table recorders. Soil hydraulic
properties, including hydraulic conductivity,
87-306 DESIGN CRITERIA FOR DRAINAGE drainable porosity, water retention and
OF IRRIGATED LANDS infiltration rate were measured in situ
periodically and in the laboratory. Results
R.S. Broughton, Dept. Agric. Eng., Macdonald from three years of field trials between 1983
College of McGill University, Ste. Anne de and 1985 indicate that subirrigation can be
Bellevue, Quebec H9X ICO. successfully practised through a subsurface
drainage system, but subirrigation may
Drainage facilities may be needed on irrigated adversely affect the hydraulic properties of
lands for one or more of the following the soil.
purposes. 1. To intercept leakage from
irrigation canals. 2. To leach excess salts 87-309 MOLE DRAINAGE IN EASTERN
from the soil. 3. To prevent waterlogging due CANADA
to non-uniform irrigation or excess rainfall
during a part of the year. 4. To drain excess C. Weill and S. Natho-Jina, Alfred College of
water from depressed areas, from the Agr. and Food Tech., Alfred, Ontario
downstream edges of surface irrigated fields KOB 1A0.
or from irrigation canals. Sample boundary
conditions and design criteria will be A study on mole drainage was conducted at
presented for each of these situations. Alfred College in 1986. A plow was used to
install mole drains at three sites having
87-307 COMBINED SUBSURFACE IRRIGATION different soil physical properties and
AND DRAINAGE SYSTEMS topographies. Hydro metric measurements taken
at one of the sites showed that the mole
B. von Hoyningen-Huene, R.S. Broughton, drainage system responded quickly to rainfall
S. Prasher, Dept. Agric. Eng., Macdonald events, maintaining the water table levels at
College of McGill University, Ste. Anne de approximately 60 cm below the soil surface.
Bellevue, Quebec H9X ICO, and R. Asselin, Repeated inspections of the mole channels
Ministere de 1'Agriculture, de 1'Alimentation et using a borescope permitted checks on the
des Pecheries du Quebec, Nicolet, Quebec stability of the drains after installation. Most
JOJ 1E0. of the mole channels retained their original
shape, but some of them were partly
An experiment was conducted on flat sandy destroyed after heavy rainfalls during the
soil in southern Quebec to study the fesibility sum mer of 1986.
of subirrigation in this region. The field
experiment was designed on 10 ha with eight 87-310 OBSERVATIONS ON THE CLOGGING
replicates of two treatments consisting of OF DRAINAGE SYSTEMS BY IRON
irrigated and non-irrigated maize plots. It was OXIDE
found that the subsurface irrigation raised the
water table satisfactorily to a predefined R. Asselin, Ministere de 1'Agriculture, de
depth, resulting in adequate water supply to 1'Alimentation et des Pcheries du Quebec,
the crop root zone. The subsurface irrigation Nicolet, Quebec JO J 1E0.
treatment resulted in grain maize yields which
were double those of the non-irrigated plots. Observations were made on 14 drainage
systems from one to ten years old in order to

222 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

evaluate (a) the magnitude of the clogging by
iron oxide as a function of time after
installation, and (b) a field method for
measuring the concentration of ferrous ions in
the water. The field method was found precise
and easy to use, and therefore can be a Drying of farm crops is an integral part of
practical tool for engineers involved in the
design of drainage systems. The observations
showed that drainage systems were in a
relatively advanced state of clogging six years
after installation. Clogging by iron oxide was
observed in sandy soils and was greatly
influenced by iron bacteria.
87-401 KINETICS OF MOISTURE
ADSORPTION BY SILICA GEL
DURING GRAIN DRYING
A.S. Mujumdar, Dept. Chem. Eng, G.S.V.
Raghavan and Z. Alikhani, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
Anne de Bellevue, Quebec H9X ICO.
The use of various solid dessicant materials
for grain drying has been reported in the
literature. The dessicant is mixed with the
grain and separated when the drying period is
over. Silica gel, due to its desirable physical
and chemical properties, is one of the best
dessicant materials for this purpose. Silica gel
is an inert material (about 99.7% Si02) which
is very easy to handle and regenerates at
relatively low temperatures (as low as 70C).
Because an air temperature of about 80C can
be obtained easily with a flate plate solar
collector, silica gel can be a good material
for the storage of solar energy.
The main objective of this study was to
investigate the kinetics of grain drying by
silica gel. The dessicant material and the
grain were kept in an intimate contact by
placing them in layers in air-tight jars. The
initial mass of grain for each experimental
unit was 400 g. Three dessicant to grain mass
ratios (DGMR) of 1:1, 1:2 and 1:4 as well as
four residence times of 1, 2, 4 and 8 hours
were used. The silica gel was regenerated in
an electric oven at 85C.
The results of these drying tests are
presented and discussed. Both freshly
harvested and re wetted corn were used in the
study. The difference between the drying
kinetics of the grains is discussed also.
87-402 PREDICTING PRESSURE PATTERNS
IN CANOLA BINS
D.S. Jayas, Dept. Agric. Eng., University of
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Manitoba, Winnipeg, Manitoba R3T 2N2,
S. Sokhansanj, E.B. Moysey and E.M. Barber,
Dept. Agric. Eng., University of Saskatchewan,
Saskatoon, Sask. S7N 0W0.
agricultural production systems. The current
trend is to dry farm crops, including canola,
using near ambient temperature air. Aeration
is the most commonly used method for cooling
stored grains and oilseeds. In both near
ambient drying and aeration systems, air in
quantities varying from 10-30 1/s-m3 for near
ambient drying and 1-2 1/s-m3 for aeration is
forced through a perforated area, and is then
distributed throughout the grain mass. The
success of the systems is highly dependent on
the uniformity of airflow within the stored
mass. The uniformity of airflow is affected by
the configuration of the perforated area used
for introduction of the air, bin geometry,
resistance to airflow of the product and the
amount, size and distribution of foreign
material in the stored bulk.
A simulation model which accounts for
anisotropy of canola properties and foreign
material, in addition to the factors considered
in earlier models, has been developed to
predict pressure patterns in bins filled with
canola. The predicted pressures have been
validated against measurements in a farm bin
with a 4.6 m diameter. Along with the details
of the model, a discussion of its utility is
presented.
87-403 PRESSURE DROP CHARACTERISTICS
OF TWO-DIMENSIONAL SPOUTED
BEDS OF GRAIN
MX Kalwar, G.S.V. Raghavan and E.R. Norris,
Dept. Agric Eng., Macdonald College of
McGill University, Ste. Anne de Bellevue,
Quebec H9X 1C0, and A.S. Mujumdar, Dept.
Chem. Eng., McGill University.
Two dimensional spouted beds for the drying
of grains offer a number of advantages over
conventional tubular spouted beds, the
principal one being the ease of scale-up from
laboratory to commercial size units. In the
design and operation of such a device, it is
essential to know the bed pressure drop
characteristics for various grains, and how
they are affected by the aspect ratio (length
over width) of the bed cross section, and the
presence of vertical partitions. The latter
provide control over the solids circulation rate
and eliminate the restriction of maximum
223
spoutable bed height. Experimental data are
presented for various grains and are compared
with the results of existing pressure drop
models of conventional round spouted beds
87-^404 DRYING OF LEUCAENA
C.K. Sankat, Dept. Mech. Eng., and
McGaw, Dept. Chem. Eng., The University of
the West Indies, St. Augustine, Trinidad and
Tobago.
Leucaena leucocephala, commonly called
leucaena or ipil-ipil, is a leguminous plant
with considerable practical potential for the
tropics. Giant leucaena varieties are grown in
densely populated stands principally for lumber
production. However, for many tropical
countries deficient in proteinaceous feed
resources, leucaena holds great promise. Dwarf
varieties that are regularly mowed can yield
large quantities of forage, up to 20 tonnes of
dry matter per hectare per annum with crude
protein contents of 25-30%. Leucaena is
sometimes called the 'alfalfa' of the tropics.
For leucaena forage to be more widely used
in the feed production industry, its moisture
content should be reduced from the initially
high 80% (wet basis) to a suitably low level
of 12-14%, which will allow for its storage
and subsequent processing.
Using a thin layer drying apparatus, the
rate of drying of leucaena leaves was
investigated. Drying temperatures ranged from
35-90C, and air velocities from 0.05-0.30 m/s.
The methodology and results of
experiments are presented, together with the
drying behaviour of leucaena compared to that
of alfalfa. A limitation to the extensive use
of leucaena, particularly in non-ruminant
rations, has been attributed to the toxic
nature of mimosine which is present largely in
the water soluble fractions of the leaves and
seeds. Reports indicate that mimosine may be
realtively unstable at elevated temperatures.
The effect of drying temperatures on the
mimosine levels of the leucaena forage is
reported.
87-405 DESIGN, CONSTRUCTION,
INSTALLATION, MODIFICATION AND
EVALUATION OF A HEAT
EXCHANGER FOR LOWERING
ENERGY CONSUMPTION OF
DRYERS
M. Levesque, Eng. and Stat. Res.
Agriculture Canada, Ottawa, Ontario K1A 0C6.
224
In 1982 an air-to-air heat exchanger was
designed, constructed and installed in a
continuous cross flow farm scale dryer. It was
designed to transfer heat from the hot wet
exhaust air to the cooler replacement air, and
the details of the design are presented. Tests
were run with approximately 25% of the warm
D.R. air leaving the bottom of the drying section
being recycled with the incoming cool air. The
test results showed average energy savings of
18.8% due to the heat exchanger. In addition,
the average temperature rise at the upper
dryer fan was 27.9C with the heat exchanger
operating, compared to 16.2C with exhaust air
recycling only.
Some operational problems related to
accumulations of red-dog, dust and water are
described, together with steps which were
taken to eliminate them in a modified design
in 1985. The new design performed well in
1985, and based on costs of approximately
$3000 in material and $3000 labour, the
investment should be recoverable in between
985 and 1240 hours of operation (which is
about the single harvest operation for a
400 ha corn farm in Ontario).
87-406 AN INTEGRATED GAS CONTROL
AND TEMPERATURE MONITORING
SYSTEM FOR CONTROLLED
ATMOSPHERE STORAGE
E.M. Lauro, CL. Chu and J.D. Wismer,
Horticultural Res. Inst, of Ontario, Vineland
Station, Ontario LOR 2E0.
these
An automatic system has been developed at
the Horticultural Research Institute of Ontario
to monitor temperatures and to control C02
and O2 levels in 23 controlled atmosphere
storage rooms and in 30 portable unit storage
containers. A microcomputer and data
acquisition and control unit are used to
operate the system continuously. Software has
been developed also to allow the user to
interact directly with the system. The system
regularly records controlled atmosphere room
temperatures and gas levels, and compares
them with set points. Gas levels are adjusted
automatically as required. A remote alarm
system is activated when excessive deviations
from the set points occur.
GRAIN 87-^407 QUANTIFICATION OF RESPIRATION
OF FRUITS AND VEGETABLES USING
PRESSURE SENSORS
Centre,
F. Forder, G.S.V. Raghavan and Y. Gariepy,
CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987
Dept. Agric. Eng., Macdonald College of carrots and potatoes, two major stored
McGill University, Ste. Anne de Bellevue, products in Quebec, were monitored in order
Quebec H9X ICO. to investigate their possible use in a storage
disease detection system. The experiments
The respiratory activity is the essential were conducted using carrot and potato
parameter in the design and development of diseases caused by fungi and bacteria. For
techniques for harvesting, handling, storing and carrots, the diseases were watery soft rot
transporting perishable commodities. There are (Sclerotina scelerotiorum) and gray mold
many methods available to measure respiration (Botrytis cinerea). Bacterial soft rot (Erwinia
rate (RR). However, most of them present caro to vora) and bacterial ring rot
numerous disadvantages which justifies the (Corynebacterium sepedoniecum) were selected
need' for an accurate, simple and inexpensive for the potato disease trials. The experimental
apparatus which could accomodate a wide setup was designed to simulate ventilated
range of fruit and vegetable sizes. storage bins. For that purpose, a strip-trap
An instrument was designed and developed headspace analysis technique was adopted with
to measure the RR of various plant materials, purified air as the purge gas to
following a pressure variation (PV) method. A preconcentrate the volatiles. Traps of
sensitive and reliable pressure transducer was adsorbent (Chromasorb 105) were located at
used to monitor pressure changes resulting air outlets of the small 8 1 containers in
from respiratory activity of plant products. which samples were placed. The trapped
Tests were conducted at room temperature on volatiles were extracted subsequently by a
four fruits (bananas, tomatoes, strawberries thermal desorptiLon method involving cryogenic
and apples) and five vegetables (asparagus, focussing. Gas chromatographic analyses were
brussel sprouts, potatoes, carrots and celery) performed by a modified Hewlett Packard
covering a wide range of respiration rate. The model 5890 A instrument fitted with a
values of RR in terras of CO2 production and wide-bore capillary glass column. The main
O2 consumed were determined simultaneously purpose of the study was to determine the
by gas chromatography. potential of volatile monitoring as a storage
The results obtained showed good management tool.
agreement between the two test methods,
since the values determined by the PV method 87-409 PERFORMANCE OF AN AIR
were not significantly different from those CONDITIONING HEAT PUMP USED IN
obtained by gas chromatography. The POTATO STORAGE
respiratometer is easy to fabricate, simple to
use and inexpensive. D. Lord, R. Boily, Dept. Agric. Eng., Laval
University, Quebec, Quebec G1K 7P4,
87-408 VOLATILE PROFILES FOR DISEASE C. Vigneault and D. Leblanc, Eng. and Stat.
DETECTION IN STORED CARROT Res. Centre, Agriculture Canada, Ottawa,
AND POTATO ENVIRONMENTS Ontario K1A 0C6.

E. Ouellette, G.S.V. Raghavan, Dept. Agric. The technical feasibility and economics of a
Eng., and R. Reeleder, Dept. Plant Sci., ground water heat pump system for a
Macdonald College of McGill University, Ste. temperature controlled potato storage facility
Anne de Bellevue, Quebec H9X ICO. was studied. A prototype heat pump system
was installed and tested. A mathematical
Given favorable conditions, some vegetable model was then developed and used to
storage diseases have the potential of simulate the proposed system, to obtain the
producing extensive losses in both the quality optimum performance parameters. The
and quantity of products, if they are not optimum economic heat pump power, defined
detected on time. Today, disease detection in terms of total cost to own and operate the
still relies on mostly visual and olefactory system, was determined using climatic and
signs, and on the ability of the storage economic data.
manager to recognize them. Monitoring
techniques that could identify and measure 87-410 C 0 M P A RIS 0 N BETWEEN A
abnormal crop stresses are needed so that CONVENTIONAL COOLING SOURCE
corrective measures can be applied to AND STORED WINTER COLDNESS
minimize losses. FOR PRECOOLING VEGETABLE
The volatile profiles emanating from stored PROCESSING

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 225

C. Vigneault, Eng. and Stat. Res. Centre, work done during cutting, yield force per
Agriculture Canada, Ottawa, Ontario K1A 0C6, blade area and maximum force per cross
L. Blouin and G. Jacob, Roche Associes Ltee., sectional area were affected by the type of
Ste. Foy, Quebec G1Z 4M3. storage. The cut performed approximately five
cm from the plant base showed the greatest
A 250 tonne block of ice was made from differences in mechanical properties among
winter coldness using the layer by layer the storage conditions.
technology developed by Agriculture Canada.
The block was stored in an insulated building 87-501 COOLING OF STRAWBERRIES WITH
and used to precool vegetables. The results LIQUID CARBON DIOXIDE
obtained were compared with those of a
conventional water based precooling system F. Gamache and D. DesHets, Dept. Agric.
utilizing a compressor. The water quality, Eng., Laval University, Ste. Foy, Quebec
energy efficiency and economics of both G1K 7P4.
systems were evaluated and compared. The
water recirculation system was optimized, and Strawberries were cooled with liquid carbon
the results demonstrated that the system using dioxide. The cooling rate and convective heat
the ice block as a heat sink had the transfer coefficient were measured, and design
advantages of greater energy efficiency and data suggested.
increased cooling capacity.
87-502 DETERMINING THE FREEZING TIME
87-411 QUALITY ASSESSMENT OF STORED OF A PARALLELEPIPED SHAPED
LEEKS (CA, RA, HRA) USING THE FOOD PRODUCT
WARNER-BRATZLER SHEAR
APPARATUS D. Leblanc, R. Kok, Dept. Agric. Eng.,
Macdonald College of McGill University, Ste.
J.M. Scazzosi, G.S.V. Raghavan, Y. Gariepy, Anne de Bellevue, Quebec H9X ICO, and.
E.R. N orris, Dept. Agric. Eng., Macdonald G.E. Timbers, Eng. and Stat. Res. Centre,
College of McGill University, Ste. Anne de Agriculture Canada, Ottawa, Ontario K1A 0C6.
Bellevue, Quebec H9X ICO, and J. Munroe,
Eng. and Stat. Res. Centre, Agriculture The freezing time of a parallelepiped shaped
Canada, Ottawa, Ontario K1A 0C6. food product was determined and compared to
results obtained by calculation with various
The quality assessment of food materials using existing analytical models. French fried
mechanical means is an area of study which potatoes were chosen as the food product.
lacks standardization. The need for practical Temperature distributions in the french fry as
and easily obtainable quality indices is great, a function of time were obtained by
notably in the vegetable storage area. The measuring temperatures at 1 ram intervals
problem is that of finding a measurable from the surface to the centre under constant
parameter related to some textural or other freezing conditions. The therm ophysical
sensory measure which is readily detected by properties of french fries at approximately
the consumer. A Warner-Bratzler shear 20C and -20C were determined for use in the
apparatus was used to measure the mechanical analytical models. The average moisture
properties of leeks (cultLvar Alaska) which content of the fries was 73.72%, the average
were stored for a period of 90 days in a density at 20C and -20C was 1069 and 1012
controlled atmosphere, regular atmosphere or kg/m3, the thermal conductivities were 0.50
high humidity atmosphere, and duplicate and 1.0 W/m-C, respectively, and the heat
samples were sprayed with a fungicide. capacities, using Siebel's formulae, 3310 and
Measurements of moisture content, work 1760 J/kg - K. The surface heat transfer
done per cross sectional area, yield stress coefficient under the freezing conditions was
over the cross section, yield force per unit measured to be 21 W/m2-K.
blade projected area, maximum forces, work
done to the yield point and the slope of the 87-503 COMBINED ULTRAFILTRATION AND
force displacement curve were conducted on ULTRAVIOLET TREATMENT FOR THE
leeks trimmed to market quality. RECOVERY OF CHILLING BRINE
Compensation for size differences was
introduced into the calculations to allow D. McGinnis, Agriculture Canada, Ottawa,
comparisons. The sample moisture content, H. Black, Diversified Research Lab, Toronto,

226 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

G.W. Whitby, Trojan Technologies Inc., London, unsaturated triglycerides in supercritical
B. Walczack, Zenon Environmental Inc., carbon dioxide was carried out. Extraction
Burlington, and L. Norrie, Knud Sim onsen experiments were performed on mixtures of
Industries Inc., Toronto, Ontario. pairs of pure triglycerides to investigate the
interactive behaviour on oil solubilities in CO2.
A prototype recovery and treatment system The solubilities of the simple triglycerides
for sodium chloride brine, used as a chilling were found to be strongly dependent on
medium in the processed meat industry, was extraction temperature and pressure. For a
designed, developed and tested. The system series of saturated triglycerides, their
uses ultrafiltration (UF) to remove particles solubilities were found to be directly
such as fat and protein fibre from proportional to the negative logarithm of their
contaminated brine, and ultraviolet radiation molecular weights. The addition of one double
(UV) to destroy microorganisms of public bond to each of the fatty acid chains for a
health significance. Initial pilot scale studies C18 triglyceride caused a significant increase
were used as a basis for further development in its solubility in C02. However, the addition
and construction of a prototype brine recovery of a second double bond to each of the fatty
system that was subsequently installed in a acid chains did not result in any additional
meat processing plant for testing and increase in solubility under the conditions
evaluation. The ability of various UF studied.
membranes to operate on a sustained basis
was evaluated in conjunction with 87-505 HOLDING TIME CALCULATION FOR
measurements to determine their operational STERILIZATION OF LOW ACID
characteristics, including particle and FOODS DURING ASCEPTIC
microorganism rejection capabilities. The PACKAGING
effectiveness of varied levels of UV
illumination to destroy bacteria in UF treated F. Castaigne and C. Lacroix, Dept. Food Sci.
brine permeate was investigated also. An & Tech., Laval University, Ste. Foy, Quebec
economic evaluation of various design G1K 7P4.
alternatives was carried out to determine a
cost-effective system which will allow meat A method of calculating holding times is
processors to recycle at least 90% of the presented for sterilizing low acid foods
brine which is presently discarded frequently containing particles during an asceptic
or daily. Results of this project indicate that packaging process. The method is based on the
the use of a brine recovery system is a viable estimation of heating parameters f and j,
alternative to frequent brine disposal, which are used in strerilization calculations
particularly in view of the environmental and following Ball's method. When this method is
waste handling considerations. not applicable, equations are provided which
allow the construction of the heating curve
87-504 EFFECTS OF MOLECULAR for the particles, and holding times are
STRUCTURE ON THE SOLUBILITY OF calculated by integration.
TRIGLYCERIDES IN SUPERCRITICAL
CARBON DIOXIDE 87-506 T R A N SIE NT METHOD FOR
DETERMINING THERMAL
N.R. Bulley, Dept. Agric. Eng., University of CONDUCTIVITY OF FOODS
Manitoba, Winnipeg, Manitoba R3T 2N2, and
N.T. Cheok, Kuantan, Pahang, West Malaysia. D.J. Jayas and D. Griffin, Dept. Agric. Eng.,
University of Manitoba, Winnipeg, Manitoba
Supercritical fluids are being investigated for R3T 2N2.
their potential use as solvents in the oil and
fats industry. In order to predict extraction Most foods are either warmed, cooked,
efficiencies and oil behaviour under different pasteurized, sterilized, cooled, chilled or
extraction conditions, it is necessary to know frozen at some point in their development for
how the solubilities of the oils having human consumption. To predict the rate at
different molecular structure respond to which these processes can be accomplished
changes in extraction temperature and requires the knowledge of the thermal
pressure. Research on the effects of conductivity, k, of foods. Changing
temperature and pressure on the equilibrium temperature within food products, which will
solubilities of pure simple saturated and affect food quality, can be predicted if k is

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987 227

known. But the thermal conductivity of all group. There are many advantages to insects
foods has not yet been standardized due to an as compared with conventional agricultural
infinite variety. A transient method of species. For example, high conversion
estimating k for foods is suggested. The efficiencies may be obtained on very low
method has been tested in an application to quality feedstocks, and high nutritional
potatoes, and tests with other products is upgrades can be achieved due to symbiotic
planned for the near future. The details of relationships between insects and protists. The
the method and its verification are presented. aim of this research project has been to
develop an insect farm technology which
87-507 AUTOMATED LINE HEAT SOURCE allows routine production of tonnes per day.
SYSTEM FOR THE MEASUREMENT The technology contains many elements from
OF THERMAL CONDUCTIVITY AND conventional agriculture as well as bio-process
DIFFUSIVITY engineering. Many reactor design concepts are
also applicable to this type of 'farm'. To test
D. McGinnis, Eng. and Stat. Res. Centre, the developments, a process incorporating a
Agriculture Canada, Ottawa, Ontario K1A 0C6. semicontinuous reactor was designed and
operated for a number of months. The process
A system was developed for the rapid, flow chart, reactor design and overall system
automatic and simultaneous measurement of performance are presented.
thermal conductivity and diffusivity of food
materials and, with a knowledge of the food 87-509 OPTIMIZATION OF THE NUT

density, the determination of its specific heat. ROASTING PROCESS

The system utilizes the line-heat source
method for conductivity and diffusivity J. Moss, Research Centre, Canada Packers
determination, and operates under control Inc., Toronto, Ontario M6N IK4.
from a microcomputer which is interfaced
with a data logger for temperature and In the manufacture of peanut butter the
voltage measurements. Components which exist roasting process is one of the most difficult
in many well equipped thermal process to control, and yet is one of the most critical
laboratories were used. Experimental in flavour development. The Proctor and
measurements are made automatically after Schwartz roaster presently being used is a
the user has entered the necessary continuous through circulation model with an
experimental conditions and desired inherently long dead time, and there is a lack
experimental precision to the microcomputer. of sensors capable of accurate measurement
The system was found to be simple to use, of moisture content and nut colour in real
and capable of providing accurate time. In addition, there is a large number of
measurements, together with statistical variables such as input temperature and
precision statements in less than 15 minutes moisture, type and size of nuts, air flow
for common food materials. The system rates, bed depth, roaster dry and wet bulb
software was programmed in advanced temperatures, nut residence time and the like.
IBM-BASIC (compiled, version 3.0). The Furthermore, there is a lack of research on
software was written to provide maximum the roasting of nuts, especially with respect
user flexibility, and to permit automatic to the engineering aspects.
calibration of the diffusivity measurement The objective of this project is to
apparatus using standard substances. determine the analytical model which best fits
the drying curves that occur during the
87-508 DESIGN AND PERFORMANCE OF AN roasting process. Variable residence times,
INSECT FARM CHEMICAL REACTOR temperatures, relative humidity and input nut
FOR HUMAN FOOD PRODUCTION temperature and moisture content are
considered in the analysis.
R. Kok, K. Lomaliza and U.S. Shivhare, Dept.
Agric. Eng., Macdonald College of McGill 87-510 THE ROLE OF ENGINEERS IN THE
University, Ste. Anne de Bellevue, Quebec CANADIAN FOOD PROCESSING
H9X ICO. INDUSTRY

Insects can be grown as livestock to produce P. Gitelman, President, U.F.L. Foods Inc.,
animal protein for human food. More than a Toronto, 0 ntario.
million species are available in this taxonomic

228 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2, SUMMER 1987

 NEWS HIGHLIGHTS

Events at the Annual CSAE Meeting together with the Engineering Institute of Canada
Canadian Engineering Centennial convention, May 18 - 22,1987,
le Palais de Congres, Montreal, Quebec

CSAE affiliates officially with the ASAE. CSAE and ASAE Eric Norris is Master of Ceremonies at the CSAE Annual
Presidents, Don Allen and Bill Johnson sign a letter of agree Banquet, McGill University Faculty Club, May 20.
ment at the CSAE Council meeting, May 19, 1987.

Ron Britton is the incoming President of CSAE. He also was Walter Bilanski of the University of Guelph was presented the
presented with the Canadian Sheet Metal Building Institute Maple Leaf Award.
Award.

Dennis Hodgkinson is named the CSAE Canadian Agricultural

Engineer of the Year.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 29, NO. 2. SUMMER 1987 229

Len Staley is presented as a CSAE Fellow by Don Allen. Luc Choiniere was given the Glen Downing CSAE Award for
his contributions in the agricultural machinery field.

ASAE President, Bill Johnson, gives a thorough speech on the Phillipe Savoie, formerly the Quebec Regional Director, be
needs and challenges for agricultural engineers in the future. comes Vice-President (Technical).

230 CANADIAN AGRICULTURAL ENGINEERING, VOL. 29. NO. 2. SUMMER 1987

 Call for papers

31st Canadian
Agricultural
Organised by the
Canadian Society of Engineering
Agricultural Engineering
as part of the
Conference
Agricultural Institute of
Calgary, Alberta
Canada Annual Conference
August 21 -24, 1988
Sessions are planned in the following areas
Soil and Water Structures and Environment
Coordinator: Dr. D. Chanasyk Coordinator: Mr. R. Borg
University of Alberta Alberta Agriculture
Department of Soil Science 4920 - 51 Street
Edmonton, Alberta T6G 2G6 Red Deer, Alberta T4N 6K8

Food Engineering Power and Machinery

Coordinator: Dr. K. V. Lo Coordinator: Dr. K. Domier
University of British Columbia University of Alberta
Bio.-Res. Eng. Dept. Agr. Eng. Dept.
Vancouver, B.C. V6T1W5 Edmonton, Alberta T6G 2H1

Energy and Processing Non-Divisional

Coordinator: Dr. R. J. Ford Coordinator: Dr. P. Savoie
University of Saskatchewan Agriculture Canada
Agr. Eng. Dept. Departement de genie rural
Saskatoon, Sask. S7N 0W0 Sainte-Foy, Quebec G1K 7P4

Papers presented by graduate students are eligible for the graduate student paper competition (if eligible, please
indicate specifically in a covering letter).
Authors will be notified of acceptance or rejection of papers by March 1988.
Send abstracts of approximately 250 words of the proposed papers, before November 1, 1987

to : Dr. P. Savoie
C.S.A.E. Vice-President, Technical
Agriculture Canada
Departement de genie rural
Universite Laval
Sainte-Foy, Quebec G1K7P4
 JOURNAL OF AGRICULTURAL
ENGINEERING RESEARCH
Editor: D. J. White
Ministry of Agriculture, Fisheries and Food, London, U.K.

The Journal ofAgricultural Engineering Research reflectsthe broad spectrum of interdisciplinary interests
inherent in this field. It publishes mainly original research papers but also includes in-depth review
articles and short research notes.

Research Areas Include

Crop production, harvesting, and storage
including establishment, drilling, trans
planting, and thinning; crop protection;
design, performance, and use of spraying
equipment; fertilizer application and hydro
ponics; harvesting methods, monitoring crops
for harvest, treatment, cleaning, drying, and
storage of produce; physical and biological
properties of crops and their monitoring and
manipulation
Cultivation systems and implementation
including soil mechanics, implement-soil
interaction, and wear of soil-working parts
Farm buildings including design, materials,
loading factors; environmental control in
livestock housing
Livestock feeding, housing, monitoring,
handling, and transport
Soil drainage and irrigation
Tractor and vehicle design, performance, and
use including power units, fuels, trans
missions, tyres, brakes, accessories, and power
take-off systems; vehicle-soil interaction,
tractive performance, handling, and soil com
paction; ergonomics, ride and cab design; use
in different field conditions and around
buildings
Waste engineering, handling, and treatment

Published for The British Society for Research in Agricultural Engineering by Academic Press London

Recent Contents: Ventilation of Livestock Buildings by Natural Convection, M.P. Foster, MJ. Down. A
Resistive Probe Moisture Sensor for Tropical Root Crops and Vegetables, G.O.I. Ezieke. The Thermodynamic
Performance of a Continuous-flow Fluidized Bed Grain Disinfestor and Drier, G.R. Thorpe. Modelling the
Performance of a Cross-flow Grain Drier, M.E. Nellist. Physical Changes in Stored Bulk Rice, M.C. Gough,
H.S. Cheigh, S.K. Kim, T.W. Kwon. The Compaction of a Partially Dried Lucerne, D. Nason.

Volumes 36-38 (1987), 12 issues (including subject indexes)

Annual Subscription Rates: UK: 105.00, Overseas: $204.00

Coupon JAER/87/E.60 To: Journals Marketing Department

Academic Press Inc. (London) Ltd
Journal of Agricultural Engineering Research 3rd Floor, 1 Vincent Square
London SW1P 2PN
Please tick Telephone: 01-630 7881
Please send me a sample copy and subscription details Fax: 01-828 5449
Telex: 28648 CASPEG G

Signature Name (Capitals)

Address

Date

1VINCENT SQUARE, LONDON SW1P 2PN

ACADEMIC PRESS 1250 SIXTH AVENUE, SAN DIEGO, CA 92101

The Editorial Board will assess suitability
and essential detail of papers submitted for pub
lication in Canadian Agricultural Engineering.
One or more reviewers will be used. Their
comments and suggestions will be compiled
and submitted to the author. The review will
ensure that:
1. A research paper presents a piece of re
search carried to a well-defined stage of
advancement and the conclusions are ade
quately supported by the experimental
results.
2. A technicalpaper presents a clear, concise,
and factual outline and interpretation ofthe
development, design, test, or analysis under
consideration and that it is a contribution in
the field of agricultural engineering.
3. A general paper on education, research, or
extension is pertinent to major changes in
curriculum, research, or extension or to for
ward-looking developments in these areas.
4. A technical note, of one journal page or
less, on equipment development, technique
of measurement, or method of analysis will
have an application for other workers in the
field of agricultural engineering.
MANUSCRIPT
The manuscript should be typed double-
spaced on paper 8Vi x 11 inches (21.6 x 27.9
cm) with margins not less than Wa inches (3.3
cm). The first page should contain only the
title, authors' names, addresses (including
postal codes), and contribution number where
applicable. Tables and captions for illustrations
should be on separate pages, placed after the
text. Manuscript paper with numbered lines is
preferred. The original and two copies are
required.
The title of the paper should be capitalized
and centered on the page; it should give an
accurate description of the article, using key
words that can be used for computer-indexing.
ORGANIZATION
This paper should be organized to conform
with present Journal practice. Research and
Technical papers must include a short abstract
section of about 200 words.
Major headings Center on the page with
all words in capital letters.
Subheadings Start at left-hand margin,
capitalize first letter of major words.
Sub-subheadings Start at left-hand mar
gin, in lower case except first letter of first
word, and underline.
Technical and detailed information should
be included only in the form of description,
NOTES TO CONTRIBUTORS
table, graph, chart or photograph. In general,
follow the Council of Biological Editors Style
Manual, 5th ed., published by the Council of
Biology Editors, Inc., Bethesda, MD 20814.
References
List references alphabetically by authors at
the end. Include year of publication, title in
lower case except first letter of first word, and
source, with volume and page numbers where
applicable. Names of periodicals should be
abbreviated in the form given in BIOSISList of
Serials with Title Abbreviations (Bioscience
Information for Service of Biological
Abstracts, 211 Arch Street, Philadelphia, Pa,
19102). Material in press, with the name ofthe
journal, may be used as a reference. Private
communications and unpublished reports
should be referred to in parentheses in the text.
Avoid the use of footnotes. Use the author-date
system in the manuscript when referring to arti
cles in the Reference section.
Tables
Designate tables at the top by table number
(Roman numerals) and title, all in capital let
ters. All headings and other information in
tables are to be in lower case except first letter
of first word. Keep the table compact and place
it across the page wherever possible. Do not use
vertical lines.
Measurements
Use the metric system (SI) in the text.
Tables, charts and graphs should be given only
in metric units.
Equations
Equations and formulas must be set up
clearly. Use capitals for symbols as much as
possible and lower case for superscripts and
subscripts. Greek and other characters should
be identified clearly. Equations should be num
bered on the right-hand margin in large num
bers and in line with the center ofthe equation.
Abbreviations
For commonly used terms, consult the CBE
Style Manual. Abbreviate units of measure
only when used with numerals. Do not use
abbreviations in the title. Normally, numbers
less than 10 should be spelled out, e.g. six.
Paragraphs
If paragraphs are to be numbered, designate
by Arabic numerals. Designate sub-paragraphs
by lower case letters in parentheses.
ILLUSTRATIONS
Either original drawings or clean, glossy
photographs are acceptable for illustrations. An
illustration or group of them should be planned
to fit, after reduction, onto a space 58 mm wide
(one column) or 122 mm wide (two columns).
The original should be not more than three
times the size of final figure. The number of
engravings required, and therefore the cost to
the author, can be reduced if two or more draw
ings can be grouped together. For identifica
tion, the figure number and author's name
should be written on the lower left corner with
soft pencil.
Line drawings should be carefully made in
India ink on white drawing paper or tracing
linen. Letters, numerals, labels and axis cap
tions should be made in capital size with a
stencil or lettering set, not typewritten. Letters
and numerals must be at least V/2 mm high and
preferably 2 mm high in final form. Curves on
graphs must be 0.3 mm wide after reduction.
Axes and grid lines should be clearly visible but
inconspicuous; a width of 0.2 mm after reduc
tion is suggested. Figure numbers and captions
should be typed on a separate page, not on the
original illustrations. When a paper is sub
mitted for publication, the original illustrations
need not be provided so long as the copies are of
such quality that reviewers can understand
them. Original drawings must be provided
when the paper is accepted for publication.
DISCUSSIONS
Discussions may be submitted on any paper
or technical note published in the Journal for a
period of not more than four months following
publication. Discussion of a paper or technical
note is open to anyone who has significant
comments or questions about the content of the
paper/technical note. A discussion will not be
accepted for publication if it contains material
readily found elsewhere, is purely speculative,
introduces personalities or otherwise falls be
low the standards of a technical paper in a
professional journal. Authors will be given an
opportunity to reply to discussions.
The format for discussions differs from those
of papers in that figures are to be identified by
capital letters to avoid confusion with the orig
inal paper. The discusser should refer to him
self as "the writer" or "I" and to the author of
the original paper as "the author." The first
page shows the title of the original paper with a
footnote to identify the author, volume, page
and date. Name and address ofthe writer ofthe
discussion follow the title.
Discussions will be reviewed by the Editorial
Board and possibly the reviewers ofthe original
paper. The length of a discussion is restricted to
one journal page. Lengthy discussions will be
returned for shortening, or the writer may be
encouraged to submit a paper or technical note.
Electrical Power and Processing

SIMULATION OF ADSORPTION DRYING OF CORN, WHEAT, BARLEY AND OATS USING SODIUM
BENTONITE
K.C. Watts, W.K. Bilanski, and D.R. Menzies 173

AUTOMATIC TIMER CONTROL FOR A BATCH-IN-BIN DRYER

R.B. Brown, L. Otten, and J.E. Brubaker 179

DISTRIBUTION OF FOREIGN MATERIAL IN CANOLA BINS FILLED USING A SPREADER OR SPOUT

D.S. Jayas, S. Sokhansanj, E.B. Moysey, and E.M. Barber 183
THE EFFECT OF AIRFLOW DIRECTION ON THE RESISTANCE OF CANOLA (RAPESEED) TO AIRFLOW
D.S. Jayas, S, Sokhansanj, E.B. Moysey, and E.M. Barber 189

Structures and Environment

DESIGN, COST AND PERFORMANCE OF A FREE-ACCESS, TWO-LEVEL PEN FOR GROWING-FINISHING

PIGS
P.A. Phillips and D. Fraser 193

USE OF PLANNED DRAFTS IN AN ATTEMPT TO REDUCE PREWEANING MORTALITY IN BABY PIGS

P.A. Thacker and E.M. Barber 197

Food Engineering

AUTOMATED LINE-HEAT SOURCE SYSTEM FOR THE MEASUREMENT OF THERMAL CONDUCTIVITY

AND DIFFUSIVITY
D.S. McGinnis 201

A CURVE-FITTING PROGRAM TO STRESS RELAXATION DATA

R.P. Rudra 209

ABSTRACTS OF TECHNICAL PAPERS 213

NEWS HIGHLIGHTS 229