(1 Usar Principal) FINALREPORTMod1rb1MEC

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/304628371
DEVELOPING NEW STORED GRAIN PACK FACTORS WITH KNOWN ACCURACY

FOR THE COMMON GRAINS IN TRADE UNDER A COMPLETE RANGE OF FIELD
CONDITIONS
Research · June 2016

DOI: 10.13140/RG.2.1.2033.3042
CITATIONS READS
0 1,371
7 authors, including:
Rumela Bhadra Josephine Boac

Kansas State University Kansas State University
52 PUBLICATIONS 147 CITATIONS 24 PUBLICATIONS 467 CITATIONS
SEE PROFILE SEE PROFILE
Samuel Mcneill Michael Montross

University of Kentucky University of Kentucky
19 PUBLICATIONS 34 CITATIONS 70 PUBLICATIONS 772 CITATIONS
SEE PROFILE SEE PROFILE
All content following this page was uploaded by Rumela Bhadra on 30 June 2016.
The user has requested enhancement of the downloaded file.

REPORT
on
DEVELOPING NEW STORED GRAIN PACK FACTORS

WITH KNOWN ACCURACY FOR THE COMMON GRAINS IN TRADE UNDER A
COMPLETE RANGE OF FIELD CONDITIONS
for the
INTER-AGENCY AGREEMENT
between the
AGRICULTURAL RESEARCH SERVICE (ARS)

and the
FEDERAL CROP INSURANCE CORPORATION (FCIC)
April 30, 2013
by
Rumela Bhadra, Josephine Boac, and Ronaldo Maghirang

Kansas State University, Manhattan
Mark Casada
USDA-ARS, Center for Grain and Animal Health Research
Manhattan, Kansas
Samuel McNeill, Michael Montross, and Aaron Turner

University of Kentucky, Princeton and Lexington
Sidney Thompson
University of Georgia, Athens
TABLE OF CONTENTS
EXECUTIVE SUMMARY ........................................................................................................... iv
Laboratory Compressibility Tests ........................................................................................... vii
Field Measurements of Packing Factor and Model Validation ................................................. ix
Model Development.................................................................................................................. xii
Summary of Conclusions ......................................................................................................... xiv
1. Introduction ............................................................................................................................. 1
1.5 Scientific Background ...................................................................................................... 1
1.6 Experimental Design ........................................................................................................ 3
1.6.1 Laboratory Investigations ......................................................................................... 3
1.6.2 Commercial Field Trials ........................................................................................... 3
2. Laboratory Compressibility Measurements ............................................................................ 5
2.1 Compressibility of Individual HRW Wheat Varieties ..................................................... 6
2.2 Composite Samples of HRW Wheat ................................................................................ 6
2.3 Compressibility Results for HRW Wheat ........................................................................ 6
2.4 Models for Laboratory Compressibility Curves............................................................. 18
2.4.1 Potential Model Equations ...................................................................................... 18
2.4.2 Evaluation of Compressibility Relationship at High Pressure ................................ 20
2.4.3 Selection of best Model .......................................................................................... 21
2.4.4 Summary of Model Equations .................................Error! Bookmark not defined.
3. Field Measurement of Compaction Factors .......................................................................... 26
3.1 Experimental Methods ................................................................................................... 26
3.1.1 Density and Compaction Relationships .................................................................. 27
3.1.2 Application of WPACKING Model ....................................................................... 28

3.1.3 Application of RMA Method .................................................................................. 29
3.1.4 Application of FSA-W Method .............................................................................. 30
3.2 HRW Wheat Field Data ................................................................................................. 31
3.2.1 Data Collection ....................................................................................................... 31
3.2.2 Results and Discussion ........................................................................................... 31
3.2.3 Comparison of WPACKING Model, RMA and FSA-W Methods ........................ 35
3.3 Test Weight and Dockage Relationships ....................................................................... 42
3.3.1 Background ............................................................................................................. 42
3.3.2 Materials and Methods ............................................................................................ 44
3.4 Corn Field Data .............................................................................................................. 58
3.4.1 Field Measurements and Data Collection ............................................................... 58
3.5 Time and Aeration Effects on Compaction Factors for Grain Bins ............................... 83
3.5.1 Background ............................................................................................................. 83
3.5.2 Data Collection ....................................................................................................... 84
3.6 Field data for other crops: soybeans, sorghum, oats, and barley ................................... 96
3.6.1 Background ............................................................................................................. 96
4. Compaction Factor Program ............................................................................................... 100
4.1 Using the Program ........................................................................................................ 100
4.2 "InputValues" Sheet ..................................................................................................... 101
4.2.1 Pre-Data ................................................................................................................ 102
4.2.2 GrainProperties ..................................................................................................... 102

4.2.3 Grain Storage ........................................................................................................ 103
4.2.4 Walltype&GrainHeight ......................................................................................... 104
4.2.5 Deductions ............................................................................................................ 105
4.2.6 GrainSurface ......................................................................................................... 106
4.2.7 HopperProperties .................................................................................................. 106
4.3 "ProductionWorksheet" Sheet ...................................................................................... 108
5. Conclusions ......................................................................................................................... 112
6. References ........................................................................................................................... 117
7. Acknowledgements ............................................................................................................. 122

NOTATION AND acronyms
AIC = Akaike Information Criterion

AoR = Angle of Repose
DON = Deoxynivalenol, also known as Vomitoxin
FGIS = Federal Grain Inspection Services
GAC = Grain Analysis Computer
HRW = Hard Red Winter (a variety of wheat)
KD= kernel density, lb/bu
MC= moisture content
PROC NLIN = Procedure Non Linear (a programing tool in SAS® software)
RH= relative humidity
SAS® = Statistical Analysis Software
SE= standard error
SEC = Standard Error of Calibration
SEP = Standard Error of Prediction
TW = Test weight (lb/bu)
USDA ARS = United States Department of Agriculture – Agriculture Research Service
USDA-FSA-W or FSA-W= United States Department of Agriculture – Farm Service
Agency
USDA-GIPSA = United States Department of Agriculture – Grain Inspection, Packers &
Stockyards Administration
USDA-RMA = United States Department of Agriculture – Risk Management Agency
w.b.= wet basis moisture content
Di = density at given pressure, lb/bu (kg/m3) given in equation 1
P = pressure, psi (kPa) given in equation 1
M = moisture content, decimal dry basis given in equation 1
a, b, c, e, and f = regression parameters obtained from model fitting procedures
D∞ = asymptotic value of density
DA = average bulk density of the grain after compaction, kg/m3 (lb/bu)
D0 = initial bulk density, or un-compacted test weight (TW), of the grain from the
Winchester Bushel Test, kg/m3 (lb/bu)
DX = bulk density of grain in the bin at a given depth, x, kg/m3 (lb/bu)
iv
P = average compaction ratio of the grain
n = number of layers in the grain depth used in WPACKING model
x = depth in the grain, m (ft)
R = RMA pack factor including test weight, also referred as Combined TW and Pack
factor,
DS = standard bulk density (standard test weight) for a given grain type, kg/m3 (lb/bu)
fC,X = compaction factor of the grain determined by WPACKING at a given depth, x
fC,W = average compaction factor of the grain determined by WPACKING
PW = average compaction ratio of the grain determined by WPACKING
RW = pack factor with TW of the grain determined by WPACKING
fC,R = average compaction factor of the grain determined from RMA method
PR = average compaction ratio of the grain determined from RMA method
RR = RMA Pack Factor from the Loss Adjustment Manual and Handbook
CPI = converted packing index from Federal Warehouse Examiner’s Handbook WS-3
AF = FSA Warehouse Group method adjustment factor
fC,F = average compaction factor of the grain determined from FSA Warehouse Group
method
PF = average compaction ratio of the grain determined from FSA Warehouse Group method
RF = FSA Warehouse Group Pack Factor
v
EXECUTIVE SUMMARY
ARS was asked by RMA to develop new stored grain packing factors for six grains; wheat,
corn, sorghum, soybean, oats, and barley. These values were measured in the lab with samples
from different geographic locations in the US production area to establish a representative range
of values and predictive packing equations for expected field conditions. Thus, our approach led
to the development of a science-based model based on measurable physical parameters for
determining the packing of grains in upright storage bins as well as round piles. The model was
then validated and calibrated with extensive field measurements and incorporated into a user-
friendly computer program to facilitate use by RMA, FSA, insurance providers, elevator
managers, farmers, and others working with stored grain inventory.
Stored-grain packing is defined as the increase in grain bulk density caused by the
cumulative weight of overbearing material on the compressible grain products. As material is
added, the vertical pressure increases in an exponential manner with grain height. Bin geometry,
material properties, and numerous other variables influence packing, therefore, these factors
were considered when developing the new packing model and conducting subsequent model
validation exercises. One goal of this scientific approach was to reduce the total amount of data
required to achieve accurate packing estimates over the range of bin sizes and various storage
conditions encountered in the grain industry. Accurately predicting pressure at all depths in a bin
is the initial step for predicting grain packing. An example of predicted vertical pressures for 30-,
60-, and 90-ft diameter bins is shown in Figure 1, which illustrates the relative effect of both
grain depth and bin diameter on increasing grain packing. The plotted pressures were predicted
using a modified version of Janssen’s equation with both density and pressure varying with
changing depth. This model is believed to better reflect the true physics of the packed grain
particles.
25
20
Pressure, psi
15
90 ft
10
60 ft
5 30 ft
0
0 20 40 60 80 100
Depth, ft
Figure 1-1 Internal grain pressure for 30 ft, 60 ft, and 90 ft diameter bins for varying depths.
vi
Laboratory Compressibility Tests
In order to predict particle compaction using Janssen’s equation, the relationship between
packing and pressure needs to be defined for the specific type and condition of grain in the bin.
Packing can then be calculated throughout the bin and the average packing for the whole bin can
be determined. To obtain the required relationships between packing and pressure, uniaxial
compression tests were performed for various whole grains using established methods and
equipment. Numerous uniaxial laboratory compression tests were performed on the grains using
a steel box equipped with a flexible rubber diaphragm. The box constrained the grain in a
chamber 12 12-inches in cross-section and 4-inches tall. The flexible rubber diaphragm in the
base of the box was used to produce an equivalent overbearing pressure using compressed air
matching conditions found in grain storage bins. Compaction of the grain was determined by
measuring the resulting deflection of the sample. Pressure was applied to the apparatus in seven
increments up to 20 psi, which is the range of pressures normally found in storage conditions
within the grain industry.
The preliminary packing factor model, WPACKING, used a fixed-polynomial equation
from Thompson et al. (1987) to describe the amount of compaction that occurs within the grain
because of the overbearing pressure. Four other potential models were evaluated to see if they
could describe the compressibility data more accurately. Two of the other models (the MMF
equation and the modified Page equation) were developed with a more theoretical basis than the
fixed-polynomial equation. All four of the new numerical models described the laboratory
compressibility data well and showed improvement over the previous fixed polynomial equation.
The modified Page equation fit the data slightly better than the other new models, although it
occasionally exhibited problems obtaining the best fit of the asymptotic coefficient contained
within the equation. The Farazdaghi-Harris equation fit nearly as well as the modified Page
equation and was more stable when fitting to the data:
1/ ∙ ∙ ∙ (1)
where Di is the density at a given pressure as predicted by the equation; TW is the initial test
weight; P is the overbearing pressure; M is the moisture content (wet basis); and a, b, c, and f are
the parameters determined by least squares best fit. This model was selected as the best for
representing the laboratory compressibility data.
In this study Hard Red Winter (HRW) wheat was intensively studied to evaluate the effects
of blending, growing season, growing location, and dockage using 27 untreated HRW wheat
samples from Idaho, Montana, South Dakota, Colorado, Oklahoma, North Dakota, Texas,
Kansas, Nebraska, Washington, and Oregon. These were primarily from the 2007 through the
2010 crop years with a test weight (TW) that ranged between 52.9 and 64.3 lb/bu. The samples
were tested at two moisture content (MC) levels, nominally 10% and 13%, that would represent
the range in expected moisture contents for HRW wheat. Moisture content and test weight
vii
accounted for the majority of the variation in the compacted bulk density (r2 > 0.96) over the 27
samples and two moisture content levels.
Although the packing behavior of the wheat samples was accurately described by test
weight and moisture content there were significant differences in the compressibility of
individual samples. For example, one sample with a test weight 54.4 lb/bu and a moisture
content of 9.3% increased to 59.9 lb/bu when subjected to an overburden pressure of 10 psi. This
is an increase of 10.1% in bulk density. Compared to a sample with a high test weight (63.3 lb/bu
at a moisture content of 10.2%), the bulk density increased to 67.8 lb/bu at an overburden
pressure of 10 psi, an increase of 7.1% in bulk density.
The composites were combined to represent possible blending conditions that might be
applicable to the grain industry according to location, growing season, and variety. There were a
total of ten composite samples made from the 27 single variety samples. The packing of the
composite samples could be predicted based on the regressions derived for the 27 individual
varieties (r2 of 0.95) using moisture content and test weight as the only variables (Figure 2).
72
70
68
Predicted Density (lb/bu)
66
64
62
60
58
56
54
52
52 54 56 58 60 62 64 66 68 70 72
Measured Density (lb/bu)
Figure 1-2 Densities of composite samples predicted from only TW and MC (R2 = 0.95)
The effect of dockage on the compressibility of the HRW wheat composites was
determined at levels of 0%, 1%, and 5% and two moisture content levels. Dockage reduced the
test weight. For one representative sample, the test weight was 61.7, 58.7, and 52.9 lb/bu when
the dockage was 0%, 1%, and 5%, respectively. Packing increased when the dockage level
increased. At an overburden pressure of 10 psi, the packing was 6.5%, 8.6%, and 13.2% for
dockage levels of 0%, 1%, and 5%, respectively. It was found that the majority of the effects of
dockage can be accounted for by using the measured test weight with dockage, thus no separate
accounting for effects of dockage was used.
viii
Knowledge of the test weight, moisture content, and dockage are required to accurately
determine packing. Obtaining representative samples from grain in storage can be problematic
due to bin entry and worker safety. Representative samples can more readily be obtained while
loading a grain bin by following recommended sampling procedures for trucks or, if possible, by
using a diverter-type sampler on the stream of grain being loaded.
Field Measurements of Packing Factor and Model Validation
The computer model was validated by comparing the predicted pack factor and mass of the
stored grain to actual values as determined through field measurements of grain inventory in bins
and piles. Comparing the model output to field measurements allowed other physical parameters
used in the model to be further assessed, such as wall friction, grain-on-grain friction, and the
lateral to vertical pressure ratio. Over 300 bins were measured in the field.
The actual mass of grain in each elevator and farm bin was determined from scale tickets
provided by cooperators. Additionally, the cooperators provided data on grain material
properties, such as TW and MC. The details of the bin geometry and grain surface profile were
measured by the investigators using a LEICO DISTO D8 laser distance and angle meter (Leica
Geosystems AG, St. Gallen, Switzerland), occasionally assisted by standard tape measures. On
regularly-shaped conic surfaces seven or more profile points were taken to determine the grain
surface profile. A more irregularly-shaped grain surfaces was measured using more points to
determine the surface profile. The irregular-shaped grain surfaces occur in larger diameter bins
that are emptied using a side draw. In the Eastern US, the LEICO DISTO D8 laser was used to
map the grain surface to account for any irregularities. The surface maps consisted of hundreds
of points taken along the grain to get an average angle of repose, as well as, determine the exact
height of the grain. To find the total amount of grain stored in a bin, the integral of the mapped
surface was taken, to find the volume, and that volume was multiplied by the TW of the grain.
The total amount of grain stored within the bin was determined by multiplying the average TW
by the total volume of grain. A “calculated packing factor” was determined as the ratio of total
mass from scale measurements to mass calculated from the grain volume and TW.
Field measurements were taken for bins containing either: wheat, corn, sorghum, soybeans,
oats, or barley. The packing factors of these six grains in vertical storage bins were compared to
predicted packing factors using three different methods: the computer packing model
(WPACKING), the USDA Risk Management Agency (RMA) method, and the USDA Farm
Service Agency warehouse group (FSA-W) method. Concrete and steel bins were measured in
Kansas, Oklahoma, and Texas storing HRW wheat, and in Kentucky, Michigan, and Indiana
storing soft red winter (SRW) wheat. Packing was measured in corrugated steel bins and
reinforced concrete silos with diameters ranging from 11 to 113 ft (3.4 to 34 m) and grain heights
ranging from 6 to 137 ft (1.8 to 42 m) for both HRW and SRW wheat. Predicted mass of
compacted stored wheat from these three methods were compared to the reported mass from
scale tickets.
ix
For the HRW wheat bins, pack factors predicted by the WPACKING model ranged from
0.929 to 1.073 for steel bins and 0.986 to 1.077 for concrete bins. Those predicted by the RMA
method ranged from 0.991 to 1.157 for steel bins and 0.993 to 1.099 for concrete bins. Pack
factors predicted by the FSA-W method ranged from 0.985 to 1.126 for steel bins and 1.012 to
1.101 for concrete bins. The absolute mean and median differences between the WPACKING
model-predicted mass and reported mass were 1.64% and -1.26%, respectively, and for
corrugated steel bins, 3.75% and 2.16% for reinforced concrete silos. In most cases, the model
under-predicted the mass in corrugated steel bins and over-predicted mass in concrete silos.
Comparison of the difference between RMA predicted values and reported values showed
an absolute mean of 4.41%, with a median difference of 1.91% for HRW wheat in steel bins, and
absolute mean of 3.25%, with a median difference of 1.03% for concrete bins. Most of the bins
were over-predicted by both the RMA and the FSA-W methods. Some of the large differences
observed in this case can be attributed to the unique geometry in these bins and the difficulty in
describing these bins using the program. Overall, WPACKING better predicted the mass of grain
in the bins than the current RMA and FSA-W procedures. Some of the differences may be
because the RMA and FSA-W methods do not include the effect of grain moisture content, or the
bin wall type on pack factors.
For bins measured in the Eastern U.S. containing SRW wheat, the scale ticket data were
often only available on a whole facility basis, or based on a differential between multiple
measurements, because many cooperators did not track scale tickets by individual bins or they
blended multiple bins when selling the grain. Figure 1-3 shows the WPACKING model-
predicted mass plotted against the actual mass based on scale ticket data. The model tended to
under-predict packing for SRW wheat with an average error between the observed and predicted
masses of –3.25% (standard deviation = 0.06).
SRW Wheat
Figure 1-3 Predicted versus actual inventory for SRW wheat for combined bins.
x
When determining the packing of HRW wheat, issues were raised with respect to
obtaining the required TW with dockage included in the sample. Most grain elevators follow the
USDA–GIPSA guidelines on official grain inspection in which dockage is removed prior to
measuring TW. This creates a challenge in predicting grain packing and conducting inventory
studies because the average TW of the grain in a bin should include the TW with dockage to
accurately determine grain packing. Hence, regression models between TW without dockage and
TW with dockage were obtained based on the reported scale data from three elevators located in
Kansas and Oklahoma. The correlation R2 value was found to be between 0.703 and 0.776 for
grain samples with dockage levels of 0% to 0.39%, 0.40% to 0.59%, 0.6% to 0.9%, and 1% and
above. In comparison, the R2 for a regression model that included the global data set, irrespective
of dockage levels, was found to be 0.736. While dockage level had a significant effect on the
TW, elevator location did not have any influence on the regression models. The Linear equation
and Power Model C equation (R2 = 0.745 and 0.748, respectively) were selected for predicting
TW with dockage as a function of dockage levels and TW without dockage. Residual plots for
these models yielded a random distribution.
For corn, packing factors were measured in steel bins with diameters ranging from 11.7
to 104.6 ft (3.57 to 31.9 m) and eave heights ranging from 8.31 to 102.5 ft (2.53 to 31.2 m) for
those bins. For corn bins, the maximum difference between the WPACKING program predicted
mass and the actual mass was from -4.54% to +4.53% with absolute average of 0.86%. For corn
steel bins, WPACKING model under-predicted 80% of the cases. But this trend was depended
slightly on grain height/bin diameter ratio (H/D ratio) of the bins. Concrete bins over-predicted,
irrespective of its H/D ratio. The maximum percent differences between RMA predicted and
reported mass is -4.54 to 4.53% with an absolute average of 1.46%, and for FSA-W is was -
3.33% to 5.67% with an absolute average of 1.60%. The variability of % differences was more in
RMA method than WPACKING model prediction, where most of the data points were close to
zero percent difference line. Commercial bins (both from farm sites and elevators) were
measured in Central, Midwest, and Southern regions of the U.S. to improve the overall
robustness of the compaction factor prediction results. Figure 1-4 shows the WPACKING
model-predicted mass plotted against the actual mass based on scale ticket data.
xi
Figure 1-4 Predicted versus actual inventory for corn for combined bins.
The long and short term time effects of grain settling on grain compaction were studied.
Changes in grain heights and packing factors were analyzed for one year in four steel bins
containing barley measuring 88.5 ft diameter and 66.0 ft in eave height. Three corrugated steel
bins of HRW wheat in Kansas, measuring 18.6 ft diameter and 13.3 ft in eave height, were
monitored over 20 days after three months of storage. One corrugated steel bin containing corn
in Kansas, measuring 24ft in diameter and eave height of 21.0 ft was monitored for 6 months.
Decrease in grain height for barley bins were seen after 4 months of storage and for aerated corn
bins, change was seen after 4 month of storage. For short term storage there was no noticeable
change in the Kansas HRW wheat bins. As the grain height decreases the packing was increased,
as expected. However, the amount of bins monitored for time and aeration effect studies and
their results are inconclusive at this stage. More bins needs to be tracked over time, which is in
turn challenging due to the market and dynamic nature of grain elevator and storage operations.
Still more bins are to be measured to study these effects.
Model Development
A computer program was developed to predict the packing of whole grains in storage bins.
The program was written using Visual Basic and is accessed using Microsoft Excel. The
program utilizes the differential form of Janssen’s equation to predict the variation in bulk
density of the stored material
dP/dy = D – (kPµ/R) (2)
where D is the bulk density of the grain, k is the lateral to vertical pressure ratio, µ is the
coefficient of friction of the grain on the bin wall, R is the hydraulic radius of the bin, P is the
vertical pressure, and y is the depth of material in the bin. In Janssen’s equation the bulk density,
xii
D, of the stored material is a function that varies with internal grain pressure and moisture
content. These functions were formulated from experimental data for each different grain, taking
into account variety and location effects. Utilizing this information, the estimated bin capacity in
standard bushels and the packing factor for eight different whole grains can be estimated as a
function of MC and TW of the stored material for storage structures of any height and diameter.
The computer program requires input from the user based on the type and properties of the stored
product and the geometry of the storage structure. A list of the required inputs for the program is
shown below.
Grain Properties: a) type of grain: Soft Red Winter Wheat; Hard Red Winter Wheat; Corn;
Soybean; Sorghum; Rice; Barley; Oats, b) Moisture content of the grain (% w.b.), c) dockage
(%), d) bulk density of the grain (lb/Bu).
Dimensions of the Storage Structure: a) Bin Diameter, b) Grain Height, c) Bin Cross-Section:
Circular; Square; Rectangular; Polygon; Round Piles, d) Bin Wall Type: Corrugated Steel;
Smooth Steel; Concrete, e) Top Grain Surface: Coned Upward; Coned Downward; Level, f)
Top Grain Height (ft.), g) Bottom Discharge Condition: Flat; Hopper Bottom, h) Hopper Bottom
Angle (degrees).
xiii
Summary of Conclusions
Laboratory Compressibility Measurements
After combining HRW wheat varieties to form composites representing possible blending
conditions applicable to the grain industry, the composite compressibility relationships were
compared to predictions from the original individual varieties. There were ten composite samples
prepared from the 27 single variety samples. The packing of the composite samples could be
predicted, with an R2 of 0.95, based on the regressions derived for the 27 individual varieties
using moisture content and test weight as the only variables.
The majority of the effects of dockage can be accounted for by using the measured test
weight with dockage, thus no separate accounting for effects of dockage was used.
A fixed-polynomial equation was used in the preliminary packing factor model,
WPACKING, to describe the fundamental compaction of grain as a function of overbearing
pressure. Four new models were evaluated in this study to improve on the prediction of
compaction from overbearing pressure. All four of the new numerical models described the
laboratory compressibility data well and showed improvement over the previous fixed
polynomial equation. While the modified Page equation fit the data slightly better than the other
new models, the Farazdaghi-Harris equation was selected as best model use in the packing factor
prediction program because of difficulties with properly fitting the asymptotic coefficient in the
modified Page equation with some data sets.
New compressibility relationships for corn, soybeans, wheat, milo, barley, oats, and rice
were developed with laboratory measurements. The coefficients for the Farazdaghi-Harris
equation were determined for all crops in the laboratory data and the results were incorporated
into the packing factor prediction model.
Field Compaction Measurements
Commercial and on-farm vertical storage bins made of concrete and corrugated steel, with
varying bin diameter, eave height, and grain properties were measured in 14 U.S. states to
determine the compaction factor of corn, soybeans, wheat, milo, barley, and oats. Bin diameters
ranged from 12 to 105 ft (3.6 to 32 m) and grain heights ranged from 7 to 140 ft (2.1 to 42 m).
Calculated compaction factors (based on the mass of grain in the bin determined from scale
measurements), WPACKING compaction factors (based on the computer model), and RMA
compaction factor were calculated for all measured bins and the resulting predictions of grain
mass compared. The major findings from the field measurements were:
 Calculated compaction factors for HRW wheat ranged from 1.4% to 9.5% in corrugated steel
bins; and from 2.4% to 10.1% in reinforced concrete bins. The preliminary model,
WPACKING mostly under-predicted the grain mass for steel bins but had a tendency to over-
predict mass for concrete bins with regular hopper bottoms. The RMA method tended to over
predict mass of HRW wheat except in some small bins.
xiv
 Corn quality properties like test weight, moisture content, damage, broken kernel content were
found to be in normal ranges for this crop. The average angle of repose was 20.40° for corn
bins. The absolute average difference between predicted mass and reported mass was 0.86% (-
4.54% to +4.53%) for the WPACKING model, 1.46% (-2.92% to 4.97%) for the RMA
method, and 1.60 % (-3.33% to + 5.67%) for the FSA-W method. The variability of these
differences was greater for the RMA method than for the WPACKING model, where most of
the data points were close to the zero percent difference line. For corrugated steel bins with
diameter above 15 m (49.2 ft), the WPACKING model over predicted for H/D ratios from 0
to 0.68 and under predicted for H/D ratios from 0.69 to 1.0 compared to reported mass values.
For steel bins with diameters less than 15 m, irrespective of H/D ratios, the differences were
randomly distributed. For corrugated steel bins with H/D ratios from 0 to 0.68 the RMA
method showed random differences for diameters less than 13 m (42.7 ft) and over predicted
for diameters greater than13 m. For other ratios and bin diameter less than 15 m the RMA
method predicted almost evenly on either side of zero percent difference line. For corrugated
steel bins with H/D ratios from 0 to 0.68 and bin diameter above 15 m, the FSA-W method
showed over prediction but below 15 m diameter, it was evenly distributed along zero percent
difference line. Concrete corn bins were mostly under predicted by all three methods, and the
differences were higher for the FSA-W method than for the RMA method and the
WPACKING model.
 For soybeans the WPACKING model predicted differences were from -3.04% to +2.26%
(average of +1.02%) and the RMA method was from -4.40% to +6.87% (absolute average of
+3.30%) where for both methods 95% of the bins showed over-prediction. For sorghum the
WPACKING model predicted differences from -2.34% to +2.51% (absolute average of
+0.84%) and using the RMA method yielded -7.37% to +9.32% (absolute average of
+5.10%), 85% of bins (only concrete) showed over-prediction from the RMA method but for
the WPACKING model it was mostly under-predictions close to zero percent difference line.
For oats the WPACKING model predicted differences from -6.45% to 8.81% (absolute
average of 4.17%), but using the RMA method it was from -21.64% to +9.07% (absolute
average of +12.18%) including both steel and concrete bins. For two concrete interstice oat
bins, RMA method gave around -31.5% differences when compared to scale ticket mass data.
For barley the WPACKING model gave +0.66% to +6.59% differences (absolute average of
+3.53%), but the RMA method yielded -15.18% to +2.71% (absolute average of +10.02%).
Clearly, for oats and barley in some cases the differences where large. This could be logical
because oats and barley are light weight, i.e., low bulk density, and have the tendency to
compact more than other crops. Thus, above average compaction behavior in oats and barley
can be the most difficult to predict.
 For soft white wheat, twelve soft white wheat bins from Idaho and Washington, a mixture of
both smooth and corrugated walled steel bins and concrete bins, have been measured with bin
diameters ranging from 13.25 ft to 78.04 ft and eave height from 64.30 ft to 112.7 ft, the
average AoR was found to be 20.34°. Because the WPACKING model and the RMA method
xv
do not have compaction factors to predict mass in the bins we could not calculated
differences, unlike the other crops. For durum wheat, four durum wheat steel corrugated bins
were measured with diameter of 36 ft and eave height of 25.52 ft and average AoR of 20.94°.
 Overall, WPACKING better predicted the mass of grain in the bins than the current RMA
procedure, depending on the crop type and in some cases H/D ratios. Some of the differences
may be because the RMA compaction factor method does not include the effect of grain
moisture content, grain height, or bin wall type on compaction factors.
Other Studies:
TW with dockage and without relationships
Regression models for predicting TW with dockage values (when TW without dockage for
HRW wheat samples are given) were obtained for four dockage classes with R2 values ranging
from 0.7033 to 0.776. The global plot did not show any bias towards the location of the TW
values. The Linear and Power Model C 3D regression models were selected as the two most
favorable models to present TW with dockage = f(dockage level, TW without dockage) with
reasonably high values of R2 and low values of standard error and AIC. The developed model is
valid and suitable over a range of moisture contents from 10 to 12.9% (wb). These models are
valid with dockage levels less than 2%, and usually that is the practical situation in the U.S grain
industry.
Time and aeration effects
The measured effects of time on compaction factor were small in limited testing. For corn
steel bins for both aerated and non-aerated significant change in grain height was observed after
4 months of storage, however, the decrease for aerated bin was 0.54% and for non-aerated was
0.13%. Steel bins with barley grain started showing change in grain height after 5 to 8 months of
storage (average 6 months of storage). Change in grain height vs time showed reciprocal
logarithm curve (from exponential family of curves) fit with R2 of 0.83 and standard error of
0.034. Literature review confirmed that effect of vibration from nearby railway tracks on packing
and settlement of grain depth is very unlikely in commercial and practical scenarios. However,
because of limited number of grain bins that have been studied for time and aeration effects,
these results are not conclusive. More studies on tracking grain depth for other commercial bins
with respect to aeration and time are still being carried out.
Compaction Factor Program
A computer program was also developed to make the predicted packing factors readily
available to the grain industry. In current form, the program is written using Visual Basic and is
accessed using Microsoft Excel. The computer program requires input from the user based on the
type and basic properties of the stored product and the dimensions, geometry, and construction of
the storage structure.
xvi
1. INTRODUCTION
Existing compaction factor data are of unknown reliability and their accuracy is often
questioned by the industry (Bhadra, 2013). Accurate data is required for government-mandated
inventory control and is a crucial component of new quality management systems being
developed to enable source verification in the grain handling industry. Because of the immense
variety of storage scenarios, the Risk Management Agency (RMA) is not able to do a thorough
assessment of bin inventory and compaction factors for their customers, as a result USDA-ARS
was asked to develop and calibrate new compaction factors for major crops in the U.S.
1.1 Scientific Background
Grain is somewhat compressible when subjected to the cumulative weight exerted from the
material above in a storage unit. The degree of compression depends on a number of variables
regarding grain type and condition, bin materials, and size and geometry of the storage structure.
Compression causes compaction that increases the bulk density of the material and, thus,
increases storage unit capacity. Accurate compaction factors are required to determine the mass
of grain in storage from bin dimensions and test weights. Inventory control is critical for stored
grain managers due to financial aspects (auditing by state agencies) and for the future utilization
of quality management systems. Most state warehouse agencies currently utilize compaction
factors from the Federal Warehouse Examiner’s Handbook (USDA, No Date), which has been
updated slightly over the years but many of these updates are not widely known among grain
industry stakeholders.
Several studies in the literature have attempted to develop a simple and convenient method
for estimating the amount of compaction in grain storage structures. One of the first was Bates
(1925) who suggested that the number of standard bushels of wheat (60 lb/bu) in a storage unit
was dependent on the test weight, depth of material, and dimension and shape of the structure.
Measurements revealed the average amount of compaction for wheat was 4.85% of the test
weight. Bates suggested different compaction factors would probably exist for other grains.
Malm and Backer (1985) attempted on-site measurement of compaction factors for six crops
(oats, barley, two types of sunflower, and two classes of wheat). Compaction factors were
estimated by measuring the total settlement of the stored material in bins 7 to 14 days after filling
and again at 23 to 40 days after filling. A statistical model was tested to correlate selected
physical properties such as moisture content, percent dockage, test weight, crop depth, and the
dimensions of the structure to the compaction factor, but only barley and oil sunflower yielded
significant terms in the model development. They attributed the difficulties in developing
predictive models to disturbances to the grain in the commercial setting. They also reported it
was common to have reports of existing compaction factors over-predicting the mass in storage.
There have also been several studies on the effect of fines and filling method on the compaction
factor (Chang et al., 1981, 1983; Stephens and Foster, 1976). These studies found that grain
spreaders distributed fines throughout the bin and increased the compaction factor of the grain.
1
In an effort to simplify the procedures for estimating stored grain inventories, a computer
model was developed by Thompson et al. (1987). This program employed the differential form
of Janssen's equation to estimate the pressure and in-bin bulk density for a given depth of grain
in a bin. Janssen’s (1895) equation is commonly used in the design of storage bins and utilizes
the properties of the stored material and the geometry of the storage structure to estimate in-bin
pressures. Janssen’s equation assumes the properties of the stored material to be constant.
However, in this program the differential form of this same equation is used which assumes the
properties of the stored materials to be mathematical functions, which vary with various grain
parameters (i.e. moisture content, vertical pressure etc.). By computing these values using the
differential form of Jannsen’s equation for each depth in a bin, the capacity for that structure and
the amount of compaction can be estimated more accurately by taking into account the variation
in grain properties.
This computer program was validated in a limited manner by Thompson et al. (1989) using
data from full size bins and has become the current ASAE Standard for estimating the storage
capacities of cylindrical grain bins (ASAE S413.1, 1999). Uniaxial compression tests were
conducted using clean, whole grains by Thompson et al. (1987) to determine the variation in
bulk density with respect to pressure and moisture content. Multiple varieties from six different
grains were tested with an apparatus used by Thompson and Ross (1983) to determine the
changes in bulk density at different overburden pressures between 0 and 172 kPa and moisture
contents between 10% and 16%. Mathematical expressions were developed which described the
change in bulk density with respect to changes in pressures and moisture content. These
expressions were incorporated into the WPACKING computer program to predict the amount of
compaction in bins of a given height and diameter. For example, an 11.9 m diameter bin level
filled with corn to a height of 11.9 m, was estimated to have a compaction factor of 3.8% (ASAE
S413.1, 1993). However, for compaction factor results from the model (Thompson et al., 1987)
to be accurate and accepted nationwide, it was proposed that they must be calibrated over the
range of planned usage. However, it was believed because the model itself is based on the true
physics of the grain compaction rather than a purely empirical approach the required number of
experimental tests would be able to be greatly reduced and, thus, does not need to be tested for
every possible combination of variables.
The preliminary model currently being developed for determining compaction factors will
be calibrated and validated by measuring compaction for wheat, corn, grain sorghum, soybeans,
oats, and barley in bins located over the major grain producing regions of the U.S. The new data
and refined model developed in this research will improve the scientific basis for predicting
compaction factors in stored grain.
The objective of the project is to refine and validate a procedure with known accuracy,
based on measurable physical parameters, for determining the compaction of grains within
upright storage structures and make the results available in a user-friendly software tool that can
be used by producers, elevator managers, and government officials. Factors identified for the
2
study are: (1) structural shape and size, (2) bin wall type, (3) type of grain, (4) bulk density (test
weight) of the incoming grain, and (5) moisture content of the grain, (6) additional factors such
as broken material and fines in the grain, time in storage and impact of facility aeration systems
will also be studied.
1.2 Experimental Design
1.2.1 Laboratory Investigations

Different varieties of each whole grain were tested to determine the range in the
compressibility at different moisture levels. This data is important to quantify the variability that
will exist in storage structure when numerous varieties are placed into storage. A standard device
to measure the compressibility of grain was used in our laboratories. Measurements were
replicated four times and equations governing the change in the compressibility of the bulk grain
as a function of moisture content, grain type, and vertical pressure were evaluated. The
laboratory work was divided into four sub-units: (1) wheat, (2) corn, (3) soybeans and sorghum,
and (4) barley and oats. Validation of the model consisted of laboratory and field investigations.
Wheat was evaluated first because of the high bulk density that creates the largest vertical
pressures in bins. Part of quality management systems is determining the error in measurements
and acceptable ranges of accuracy. Quantifying this variability and determining the accuracy of
the model and measurement techniques is important for developing control charts. In addition, it
is important to understand sources of errors in measurements and what can be changed to
improve the accuracy of the system.
1.2.2 Commercial Field Trials

Data was gathered from commercial bins to validate the model. Samples of grain from a
limited number of the bins were collected to measure the compressibility of the field samples.
The measured compaction was compared to the results from the model. A standard error was
calculated and the model accuracy determined based on the standard error. This accurate
confidence interval information alone will be a major improvement over the current methods for
which the errors are not known. In addition, the new model has considerable potential to have
better accuracy than the old methods because it accounts for the many important variables in
grain and bin properties that affect the final bin compaction but were not taken into account by
the old methods.
Data from commercial bins that are typical were needed for model validation. Bins
constructed of steel (welded smooth walled and corrugated) and reinforced concrete were
measured because they represent the range in coefficient of friction found in the industry.
Circular, interstice, square, and other shaped bins were measured. Data was collected from a
wide range of locations within the U.S. Data required for the major variables affecting stored
grain compaction are grain type, moisture content, test weight, and bin geometry and
dimensions, which are shown in Table 1. Variation across different regions of the U.S. was also
3
investigated. Other factors included bin wall material, dockage and fines content, loading
method, and variations within grain type, such as different varieties, or wheat class, especially
when those factors significantly affect grain size and morphological characteristics. The
predominate variety of wheat present in each tested bin was identified to the extent possible with
information from the cooperating elevators. In order to analyze statistically the effects of each
major variable properly experimental measurements of all combinations of just the major
variables, using only three levels of each variable and three replications in a total of six U.S.
regions, would have required 3,000 to 4,000 bins to be measured just for wheat. Using an
improved range with five or more levels of each variable and more replications would have
increased this total exponentially.
Table 1-1 Required bin measurements for pack factor study.
Bin measurements for calibration and validation study.
– bin wall type

– type of grain
– weight of grain loaded
– bulk density (test weight) of the incoming grain
– moisture content of the grain
– height when full (top of cone)
– hopper bottom angle, if not flat
– height and grain weight at intermediate points when appropriate for time effects and aeration
– complete height data for randomly loaded storages
– dockage, shrunken & broken, and FM for wheat; BCFM for corn; etc. for other crops
In an effort to avoid excessive cost from experimentally determining compaction factors

for all grains under all conditions, a science-based model was used to reduce the total amount of
data required to achieve valid results. A model for determining compaction factor for a wide
range of grains and bins was developed and had been previously partially calibrated in limited
experiments (e.g., Thompson et al., 1991). With this science-based model, a range of conditions
can be selected from all possible storage conditions, but the model itself is based on the true
physics of the grain compaction. Thus, experimental measurements were not needed at every
possible combination of variables rather only at enough conditions to cover the entire range of
the variables. Compaction was measured in a limited number of bins spread over three regions
of the U.S to determine if geographic effects, because of different varieties of grain grown in
various parts of the grain growing regions of the United States played a major role. The full
range of all variables was covered but only in an appropriate, limited set of combinations.
Sufficient data was obtained in this way to calibrate and validate the model over the range of
variables needed by industry. The model predictions were evaluated against the new data.
Existing compaction tools — the Federal Warehouse Examiner’s Handbook (USDA, No Date)
and USDA RMA pack factor tables were also evaluated based on data from the measured bins.
4
2. LABORATORY COMPRESSIBILITY MEASUREMENTS
Grain undergoes compression when subjected to the cumulative weight exerted by the
overlying material in the storage unit. Research associated with grain compressibility has shown
that the bulk density of compressed grain samples — corresponding to local bulk density in deep
storage bins — is primarily a function of grain type, initial test weight, overburden pressure, and
grain moisture content when there are low amounts of dockage and foreign material. A typical
density-pressure curve for a bulk grain sample is shown in Figure 2-1.
In order to predict the amount of packing of the bulk materials in a bin, equations
describing the changes in density with respect similar to those shown in Figure 2-1 are needed
for the stored materials. Analytical equations for each type of stored material must be developed
which describes the changes in bulk density with respect to grain type, initial test weight,
overburden pressure and grain moisture content
0.08
0.07
0.06
Density Increase, %
0.05
0.04
0.03
0.02
0.01
0.00
0 5 10 15 20
Pressure, psi
Figure 2-1 Typical laboratory data for increase in grain density (compaction) as a
function of applied uniaxial pressure at constant moisture content.
Uniaxial laboratory compression tests describing the changes in bulk density of grain and
feed ingredients have been performed and are reported in the literature. Some of the original
work was performed by Clower et al. (1973) and this procedure was further refined by
Thompson and Ross (1983). In this project the procedure of Thompson and Ross (1983), was
used in which grain was contained within the box in a chamber 12-  12-inches in cross-section
and 4-inches tall. A removable metal steel plate was used to seal the test chamber. A flexible
rubber diaphragm in the base of the box was used that allowed compressed air to apply vertical
pressures uniaxially. This produced an equivalent overbearing pressure via the compressed air
similar to conditions found in grain storage bins. Pressure was applied to the apparatus in seven
increments up to 20 psi, which is the range of pressures normally found in most all storage
conditions within the grain industry Compaction of the grain was determined by measuring the
5
resulting deflection of the sample. Deflection of the sample was measured using a dial gage
actuated using a metal rod inserted through a hole in the middle of the removable top plate. A
dial reading recorded after a waiting period of 8 min. Tests were conducted for each bulk
material at two different moisture contents. Moisture contents were selected to represent the
range of storage conditions normally used in the grain industry. For each sample of wheat 3
replications were performed.
The effects of blending, growing season and dockage were investigated with Hard Red
Winter (HRW) wheat. Two bags (approximate 50 lbs. each) each of 27 untreated hard red winter
wheat varieties were purchased from seed dealers in ID, MT, SD, CO, OK, ND, TX, KS, NE,
WA, and OR. These were primarily from the 2007 through the 2010 crop years. For each variety
one bag of wheat was placed in a plastic tub and randomly assigned to an environmental
chamber. The environmental chambers were set at 32°C/45% RH and 21°C/90% RH. This
resulted in an equilibrium moisture content of approximately 10 and 13% w.b. The wheat was
mixed by hand approximately weekly while the samples equilibrated.
2.1 Compressibility of Individual HRW Wheat Varieties
Based upon the uniaxial compression tests moisture content and test weight could be used
to account for the majority of the variation in the compacted bulk density (r2 = 0.95). There were
differences between growing season, location, and variety, but these were described by the test
weight and moisture content. The samples tested had a test weight between 52.9 and 64.3 lb/bu
(681 and 828 kg/m3). The sample with the lowest initial test weight and an average moisture
content of 11.8% had a packing factor of 1.39 at an overburden pressure of 10 psi. This was also
the sample with the largest error in predicted packing, but was still less than 4%. The sample
with the highest test weight and an average moisture content of 9.9% had a packing factor of
1.31 at an overburden pressure of 10 psi. The error in the predicted compacted density was much
lower though at 1.1%.
2.2 Composite Samples of HRW Wheat
Composites of the individual varieties were made to evaluate the effect of blending. The
composites were picked to represent possible blending conditions according to location, growing
season and variety. There were a total of ten composite samples made from the 27 single variety
samples. The packing of the composite samples could be predicted based on the regressions
derived for the 27 individual varieties (r2 of 0.95).
2.3 Compressibility Results for HRW Wheat
Figure 2-2 to Figure 2-5 give test weight (bulk density) and packing % with varying
pressure for numerous HRW wheat varieties at low and high moisture ranges, collected from
different geographic location of the U.S. TAM 111 and ART wheat varieties were only from
6
year 2009 while for ENDURANCE and JAGALENE the samples were collected from 2008,
2009, and 2010. (Note: all moisture contents reported herein are wet basis).
Figure 2-2 shows TAM 111 (2009) wheat samples from KS, TX, OK, NE, and CO. The
TAM 111 wheat (low moisture 9.5% to 10.1%) from KS, CO, TX had higher initial test weights
and consequently showed significantly lower packing with an increase in pressure. This trend
was true throughout many of the tests. The exception was the TAM 111 from OK. The TAM
111 sample at (10.1% m.c.) and 61.2 lb/bu had an high initial test weight (61.2 lb/bu) bu also had
higher packing % for an increase in pressure. For the high moisture group (12.4% to 12.8%), as
excepted KS, TX, and OK samples had higher initial test weights and lower packing %. Figure
2-3 shows the trend for ART 2009 wheat from OK, KS, and NE. With the increase in pressure
the compacted bulk density increased significantly and also the packing %. For both low (9.6%
to 9.9 %) and high moisture (12.4% to 12.7 %) scenarios, we observed that samples from NE had
larger initial test weights than those from either KS and OK but likewise had the lowest packing
%. The lower the test weights (bulk density), the greater should be the void space within the
sample, which should lead to more compaction and packing of grain. For low moisture content
KS and OK test weights were not significantly different with increase in pressure. However, in
high moisture samples the test weights from KS and OK were significantly different. Test weight
ranged from 61 to 68 lb/bu for low moisture ART 2009 wheat samples, but ranged from 59 to 67
lb/bu for the high moisture samples. High moisture ART 2009 samples had slightly higher
packing % (more significant in OK samples) than low moisture wheat.
Similar trends were observed for low moisture (9.7% to 10.2%) ENDURANCE wheat
samples from 2008 to 2010 (Figure 2-4). Samples with higher initial test weights (26A and 8A)
had lower packing % than samples with low initial test weight (24A). However, for high
moisture ENDURANCE (11.8% to 13.1%), the high initial test weight sample (8B) also showed
high packing. This trend was not expected, and perhaps high moisture contributed more to higher
packing than test weight effects. For ENDURANCE (including low and high moisture) the
maximum test weights (~67 lb/bu) was slightly lower than ART samples. ENDURANCE (24A)
samples had higher variability in test weights compared to the rest of the samples.
7
Test Weights: Tam 111 (Low Moisture) KS09 Test Weights: Tam 111 (High Moisture) KS09
9.8/62.4, TX09 9.8/61.9, OK09 10.1/61.2, 12.7/61.2, TX09 12.7/60.4, OK09 12.8/60.2, NE09
NE09 9.5/58.4, CO09 9.8/63.5 12.6/56.7, CO09 12.4/55.6
69 69
KS KS
64 64
TX TX
OK OK
59 59
NE NE
CO CO
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Tam 111 (Low Moisture)‐ Packing % Tam 111 (High Moisture)‐ Packing %
12 12
10 10
KS KS
8 8
TX TX
6 6
OK OK
4 4
NE NE
2 2
CO CO
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Figure 2-2* a) Comparison of test weight (lb/bu) for TAM 111 (low moisture wheat) among KS, TX, OK,
NE, and CO states. b) Comparison of test weight (lb/bu) for TAM 111 (high moisture wheat) among KS, TX,
OK, NE, and CO states. c) Comparison of packing (%) for TAM 111 (low moisture wheat) among KS, TX, OK,
NE, and CO states. d) Comparison of packing (%) for TAM 111 (high moisture wheat) among KS, TX, OK,
NE, and CO states. The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
*The TAM 111 wheat is from crop year 2009 . The average test weight and moisture values are indicated after the two letter
abbreviation of the state.
8
Test Weights: Art (High Moisture )
Test Weights: Art (Low Moisture) OK09 12.4/58.2, KS09 12.6/59.3, NE09 12.7/62.1
OK09 9.6/59.6, KS09 9.8/60.9, NE09 9.9/63.9
68 68
66 66
64 64
OK OK
62 62
KS 60 KS
60
58 NE 58 NE
56 56
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Art (Low Moisture) ‐ Packing % Art (High Moisture) ‐ Packing %
12 12
10 10
8 8
OK OK
6 6
KS KS
4 NE 4 NE
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Figure 2-3* a) Comparison of test weight (lb/bu) for ART (low moisture wheat) among KS, TX, OK, NE,
and CO states. b) Comparison of test weight (lb/bu) for ART (high moisture wheat) among KS, TX, OK, NE,
and CO states. c) Comparison of packing (%) for ART (low moisture wheat) among KS, TX, OK, NE, and CO
states. d) Comparison of packing (%) for ART (high moisture wheat) among KS, TX, OK, NE, and CO states.
The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
*The ART wheat is from crop year 2009 . The average test weight and moisture values are indicated after the two letter
abbreviation of the state.
9
Test Weights: Endurance (Low Moisture) Test Weights: Endurance (High Moisture)
OK09 10.1/61.0, TX09 10.2/60.5, OK08 9.7/54.3, OK10 OK09 12.8/59.5, TX09 13.1/58.3, OK08 11.8/53.3,
9.8/62.7 OK10 12.7/60.1
69 69
64 8A 64 8B
17A 17B
59 24A 59 24B
26A 26B
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Endurance (Low Moisture) ‐ Packing % Endurance (High Moisture) ‐ Packing %
12 12
10 10
8 8A 8 8B
6 17A 6 17B
24A 24B
4 4
26A 26B
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Figure 2-4* a) Comparison of test weight (lb/bu) for ENDURANCE (low moisture wheat) among KS, TX,
OK, NE, and CO states. b) Comparison of test weight (lb/bu) for ENDURANCE (high moisture wheat)
among KS, TX, OK, NE, and CO states. c) Comparison of packing (%) for ENDURANCE (low moisture wheat)
among KS, TX, OK, NE, and CO states. d) Comparison of packing (%) for ENDURANCE (high moisture
wheat) among KS, TX, OK, NE, and CO states. The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15,
and 20 psi.
*The ENDURANCE wheat is from crop year 2008,2009, and 2010. The crop year, average test weight, and moisture values are
indicated after the two letter abbreviation of the state.
10
Test Weights: Jagalene (Low Moisture) CO09 Test Weights: Jagalene (High Moisture) CO09
9.3/59.1, CO08 9.6/63.2, TX09 9.5/61.6, KS08 12.2/56.4, CO08, 11.9/56.6, TX09 12.7/59.9, KS08
10.1/63.5, KS09 10.0/62.7, KS10 10.1/61.7 12.3/61.2, KS09 12.8/61.4, KS10 13.0/59.3
69 69
4A 4B
64 6A 64 6B
13A 13B
59 18A 59 18B
21A 21B
54 27A 54 27B
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Jagalene (Low Moisture) ‐ Packing % Jagalene (High Moisture) ‐Packing %
12 12
10 10
4A 4B
8 6A 8 6B
6 13A 6 13B
18A 18B
4 4
21A 21B
2 2
27A 27B
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c D
Figure 2-5* a) Comparison of test weight (lb/bu) for JAGALENE (low moisture wheat) among KS, TX, OK,
NE, and CO states. b) Comparison of test weight (lb/bu) for JAGALENE (high moisture wheat) among KS,
TX, OK, NE, and CO states. c) Comparison of packing (%) for JAGALENE (low moisture wheat) among KS,
TX, OK, NE, and CO states. d) Comparison of packing (%) for JAGALENE (high moisture wheat) among KS,
TX, OK, NE, and CO states. The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
*The JAGALENE wheat is from crop year 2008,2009, and 2010. The crop year, average test weight, and moisture values are
indicated after the two letter abbreviation of the state.
11
In Figure 2-5 are shown the trends of initial test weight and packing % with change in
pressure for JAGALENE samples from 2008 to 2010. In the low moisture group (9.5% to
10.1%), 18A, 21A, 27A samples had higher initial test weights and consequently showed lower
packing %. However, for 27A the packing % was not as low as the 18A and 21A samples. This
is likely caused by a higher moisture content (10.1%) combined with slightly lower test weight
(61.7 lb/bu) in 27A compare to 18A and 21A. Similarly, in the high moisture group (11.9% to
13.0%), samples that had higher initial test weights (13B, 18B, and 21B) had significantly lower
packing % than other samples. However, for 27B even though there were high initial densities
this sample still yielded high packing %. Perhaps, this was caused by the high moisture (13.0%)
combined with a comparatively lower initial test weights (59.3 lb/bu).
Figure 2-6 to Figure 2-10 presents the initial test weights and packing % with increase in
pressure from 0 to 20 psi for all the available varieties (irrespective of crop year), from
Oklahoma, Colorado, Idaho, Kansas, and California & Oregon. Similar trends were found in
each state; the higher the initial test weight samples showed lower packing % as expected. For
these plots the wheat from the same varieties were pooled together to form composite samples
including all the available crop years for that particular variety. However, when this was done
there were some unexpected differences in initial test weight and packing. For example, in
Figure 2-8 (Idaho), the high moisture MORELAND wheat variety, sample 1B (12.8% and 61.6
lb/bu) had higher initial test weights but also higher packing % than 25B (12.4% and 58.5 lb/bu).
Again, in Figure 2-9 (Kansas), sample 27A (JAGALENE 10.1% and 61.7 lb/bu) had
comparatively higher bulk densities with varying pressure but showed significantly higher
packing %. It was expected that 27A would show low ranges of packing %. For Idaho, only low
moisture samples MORELAND (9.4 % and 61.9 lb/bu) variety was available. Hence, there was
no comparison available with samples with different moisture and average test weights. The
differences that did not follow the expected trends for test weight and moisture content suggest
that differences between variety samples sometimes dominates over the usual test weight and
moisture trends.
12
Test Weights: Oklahoma (Low Moisture) Test Weights: Oklahoma (High Moisture)
Art 9.6/59.6, End 10.1/61.0, Tam 10.1/61.2, End 9.7/54.3, End Art 12.4/58.2, End 12.8/59.5, Tam 12.8/60.2, End 11.8/53.3, End
9.8/62.7 12.7/60.1
68 68
66 66
5A 64 5B
64
8A 8B
62 62
11A 11B
60 60
24A 24B
58 58
26A 26B
56 56
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Okahoma (Low Moisture) ‐ Packing % Oklahoma (High Moisture) ‐ Packing %
12 12
10 10
8 5A 8 5B
8A 8B
6 6
11A 11B
4 24A 4 24B
26A 26B
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Figure 2-6* a) Comparison of test weight (lb/bu) among all wheat varieties (low moisture) in
Oklahoma. b) Comparison of test weight (lb/bu) among all wheat varieties (high moisture) in
Oklahoma. c) Comparison of packing (%) among all wheat varieties (low moisture) among in
Oklahoma. d) Comparison of packing (%) among all wheat varieties (high moisture) among in
Oklahoma. The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
*The wheat varieties include ART, ENDURANCE, and TAM 111. The variety, average test weight, and moisture values are abbreviated in the
graphs.
13
Test Weights: Colorado (Low Moisture) Test Weights: Colorado (High Moisture)
Jag9.3/59.1, Jag 9.6/63.2, Tam 9.8/63.5 Jag 12.2/56.4, Jag 11.9/56.6, Tam 12.4/55.6
68 68
66 66
64 64
4A 4B
62 62
6A 6B
60 60
16A 16B
58
58
56
56
54
54
1 2 3 4 5 6 7 8
1 2 3 4 5 6 7 8
a b
Colorado (Low Moisture) ‐ Packing % Colorado (High Moisture) ‐ Packing %
12 12
10 10
8 8
4A 4B
6 6
6A 6B
4 16A 4 16B
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Colorado. b) Comparison of test weight (lb/bu) among all wheat varieties (high moisture) in
Colorado. c) Comparison of packing (%) among all wheat varieties (low moisture) among in
Colorado. d) Comparison of packing (%) among all wheat varieties (high moisture) among in
Colorado. The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
The wheat varieties include JAGALENE and TAM 111. The variety, average test weight, and moisture values are abbreviated in the graphs.
14
Test Weights: Idaho (Low Moisture) Test Weights: Idaho (High Moisture)
Mor 9.4/61.9 Mor 12.8/61.6, Mor 12.4/58.5
68 68
66 66
64 64
62 62 1B
25A
60 60 25B
58 58
56 56
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Idaho (Low Moisture) ‐ Packing % Idaho (High Moisture) ‐ Packing %
12 12
10 10
8 8
6 6 1B
25A
25B
4 4
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Idaho. b) Comparison of test weight (lb/bu) among all wheat varieties (high moisture) in Idaho.
c) Comparison of packing (%) among all wheat varieties (low moisture) among in Idaho. d)
Comparison of packing (%) among all wheat varieties (high moisture) among Idaho. The x‐ axis
indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
The wheat varieties include MORELAND. The variety, average test weight, and moisture values are abbreviated in the graphs.
15
Test Weights: Kansas (Low Moisture) Test Weights: Kansas (High Moisture)
Tam 9.8/62.4, Art 9.8/60.9, Jag 10.1/63.5, Jag 10.0/62.7, Jag Tam 12.7/61.2, Art 12.6/59.3, Jag 12.3/61.2, Jag 12.8/61.4, Jag
10.1/61.7 13.0/59.3
68 68
66 66
64 9A 64 9B
12A 12B
62 62
18A 18B
60 60
21A 21B
58 58
27A 27B
56 56
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Kansas (High Moisture) ‐ Packing %

Kansas (Low Moisture) ‐ Packing %
12 12
10 10
8 9A 8 9B
12A 12B
6 6
18A 18B
4 21A 4 21B
27A 27B
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Kansas. b) Comparison of test weight (lb/bu) among all wheat varieties (high moisture) in
Kansas. c) Comparison of packing (%) among all wheat varieties (low moisture) among in
Kansas. d) Comparison of packing (%) among all wheat varieties (high moisture) among Kansas.
The x‐ axis indicates nominal pressures of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
The wheat varieties include TAM 111, ART, and JAGALENE. The variety, average test weight, and moisture values are abbreviated in the
graphs.
16
Test Weights: Oregon & California (Low Moisture) Test Weights: Oregon & California (High Moisture)
YR 8.9/58.0, YR 9.6/63.3 YR 11.6/57.6, YR 12.7/61.8
68 68
66 66
64 64
62 22A 62 22B
60 23A 60 23B
58 58
56 56
54 54
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
a b
Oregon & California ( Low Moisture) ‐ Packing % Oregon & California (High Moisture) ‐ Packing %
12 12
10 10
8 8
6 22A 6 22B
23A 23B
4 4
2 2
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
c d
Oregon and California. b) Comparison of test weight (lb/bu) among all wheat varieties (high
moisture) in Oregon and California. c) Comparison of packing (%) among all wheat varieties (low
moisture) among in Oregon and California. d) Comparison of packing (%) among all wheat
varieties (high moisture) among Oregon and California. The x‐ axis indicates nominal pressures
of 0, 1, 3, 5, 7, 10, 15, and 20 psi.
The wheat varieties include YECORA ROJO. The variety, average test weight, and moisture values are abbreviated in the graphs.
17
2.4 Models for Laboratory Compressibility Curves
As discussed above, the computer model requires the laboratory data for bulk density to be
specified as a function of grain moisture content and overburden pressure using an analytical
equation. Specific equations are needed for each grain type. An example density difference vs.
pressure curve is shown in Figure 2-11 for TAM111 wheat samples. The curve shown represents
an average of the data collected for three TAM 111 samples grown in different geographic
locations with an average moisture content of 12.7% (wb). The best fit line for two candidate
equations are also illustrated in the figure.
4.5
4.0
3.5
Density Difference ( di ‐ do ), lb/bu
3.0 Avg.
2.5 di‐do
2.0 MMF
(di‐do)
1.5
Page
1.0 (di‐do)
0.5
0.0
0 5 10 15 20
Overburden Pressure, psi
Figure 2-11 Bulk density difference data for average of TAM 111 varieties (MC = 12.7 % wb)
with least-squares best fit MMF and original Page models.
2.4.1 Potential Model Equations

The original polynomial equation (Thompson et al., 1987) used to model the density
change was:
∙ ∙ ∙ ∙ (2.1)
where:
Di = density at given pressure, lb/bu (kg/m3),
TW = test weight, lb/bu (kg/m3),
P = pressure, psi (kPa),
M = moisture content, decimal dry basis, and
a, b, and f are parameters determined by least squares best fit.
This equation is shown in the form of (Di – TW), which represents the amount of compaction that
occurs within the grain because of the overbearing weight of grain above. Other potential models
were evaluated that might describe the compressibility data more accurately. Density difference,
(Di – TW), was again used for all models because this takes into account only the effects of grain
compression. First, equation 1 was modified from the original fixed-polynomial equation by
allowing the coefficient of second-order term to vary, yielding:
18
∙ ∙ ∙ ∙ (2.2)
where c is a new parameter determined by least squares best fit.
The standard Page model (ASAE-S448, 2012) was also evaluated. The form of the standard
Page model used here followed the form presented by Overhults et al. (1973). After adding the
usual moisture term, “f  M  P,” this equation became:
∙
1 ∙ ∙ (2.3a)
where k, n, and f are parameters determined by least squares best fit. This equation usually fits
grain drying data well, where the moisture content is approaching an equilibrium value. Thus, it
was hypothesized that it might fit this data equally well because the grain density data at high
pressure should approach an equilibrium value of density. It is believed that eventually the
compressed grain must approach the kernel density — presumably in an asymptotic manner. For
example, in Figure 2-11, it appears that above 20 psi the density difference may approach an
asymptote. However, this model didn't fit the data in Figure 2-11 well below 5 psi, while the
MMF model fit well in this case, which was typical of several data sets evaluated using these
equations. This form of the Page equation did not allow the fitting routine to select the asymptote
that fit the data best, but rather, assumed an asymptote equal to the kernel density. A modified
form of the page equation was used so the asymptote could be fit as well:
∙
∙ 1 ∙ ∙ (2.3)
where D∞ is the asymptotic value of density, an additional parameter determined by least squares
best fit
The Morgan-Mercer-Flodin (MMF) model (Morgan et al., 1975) is a growth model,
developed using the Michaelis-Menten growth model as a starting point, which also approaches
an asymptote at high values of the independent variable:
∙ ∙ / ∙ ∙ (2.4)
where a, b, c, e, and f are parameters determined by least squares best fit. The Farazdaghi-Harris
model (Farazdaghi and Harris, 1968) is another growth model that fit the data very well in
preliminary tests. Including a moisture term, this model is:
1/ ∙ ∙ ∙ (2.5)
where a, b, c, and f are parameters determined by least squares best fit. Although this model does
not exhibit a true horizontal asymptote it worked well with these data that did not show strongly
asymptotic behavior.
Statistical analyses were also performed separately using the laboratory compressibility
data for each grain moisture level. In this case, the moisture term, “f  M  P,” was removed. Five
models were investigated as outlined below in a study looking at compressibility data from 27
19
samples of HRW wheat. The models were evaluated with data taken at pressures up to 20 psi
(0.14 MPa). In addition compressibility at higher pressures (up to 1600 psi; 1.1 MPa) was
evaluated to gain additional insight into the compression process.
2.4.2 Evaluation of Compressibility Relationship at High Pressure

While 20 psi is the highest pressure expected in most grain bins currently used in the
industry, it was desirable to investigate higher pressures to clarify the shape of the density-
pressure curve to aid with model selection. Therefore a modified compressibility box was used
for testing wheat compressibility in an Instron® Universal Testing Instrument to pressures above
100 psi. Figure 2-12 shows results from compressibility tests conducted over a range from 0 to
130 psi. In this test the curve did not appear to be approaching a horizontal asymptote at the high
pressure levels (although it appears it could to be approaching a diagonal asymptote).
6
5
Density Difference, lb/bu
0
0 20 40 60 80 100 120 140
Overburden Pressure, psi
Figure 2-12 Bulk density difference data for modified compressibility box with pressure
supplied by the Instron® Universal Testing Instrument.
Additional data for HRW wheat compressibility were obtained from Dr. Stuart Nelson that
were taken at pressures exceeding 3600 psi. The very high-pressure test cell used to obtain this
data was notably different than the compressibility boxes. The test cell was constructed of brass
and was 2 in. (5 cm) high with a 0.4  0.9 in. (1.0  2.3 cm) cross-section. With this small cross-
sectional area it is virtually certain that wall friction is supporting significant portions of the load
being applied such that the grain samples are subjected to lower pressures than those shown on
the horizontal axis and labeled the applied pressures. Therefore, it is expected that the sigmoidal
shapes seen in the pressure-density curve is somewhat dampened from what would be seen for a
curve with similar pressures where the vertical pressure is entirely imparted on the grain. This
effect would have stretched the curve horizontally — showing the same density changes at
pressures higher than would have occurred with no wall friction. Thus, rescaling the x-axis
would yield an identical, non-stretched shape; so, it appears there would be no significant change
20
in the shape of the curve due to the wall friction, except the horizontal stretching, in this high-
pressure test cell and the results shown (Figure 2-13) give a useful look at the changing density
effects as pressure increases to high levels.
30
Bulk density Difference, lb/bu
25
20
15
10
0
0 500 1000 1500 2000 2500 3000 3500
Applied Pressure, psi
Figure 2-13 Bulk density difference data for very high pressure test cell with wall friction
effects.
In this very high-pressure plot, there is a sigmoidal shape with two inflection points. The
first inflection point at around 500 psi coincides with the hypothesis that the apparent asymptote
from the low-pressure compressibility box results (Figure 2-11) is a not a true asymptote and we
should expect this inflection point followed by another inflection point at higher pressure. At the
second expected inflection point the curve turns back down and eventually approaches a true
asymptote (corresponding in this hypothesis to actual kernel density, since crushing would
eventually remove all void space between kernels). The second inflection point, occurring
roughly between 2000 and 2500 psi in these data, precedes the data approaching the true
asymptote. Kernel density, which assumes no void space between kernels, is approximately 110
lb/bu, while the highest density in the test cell was 90 lb/bu — thus, the data does appear to be
approaching the kernel density as the true asymptote, but even this data didn’t achieve a high
enough compression value to be certain about this final asymptote. However it is clear that the
apparent asymptote seen in the low-pressure data a little above 20 psi in the compressibility box
is not a true asymptote as there is an inflection point at much higher pressures where density
again increases more rapidly.
2.4.3 Selection of best Model

Table 1 summarizes the results of least squares best fits of density difference vs. pressure
for three moisture contents (MC #1 = 9.7% wb; MC #2 = 12.7% wb; MC #3 = 14.4% wb) for the
27 HRW wheat varieties. The modified Page equation was a better fit on average than any of the
other models as seen by the lower standard errors. In the individual variety data, the modified
21
Page equation fit better for all moisture levels for each of the individual samples as well. The
only problem seen with the modified Page equation was occasional data sets (13 of the 81 sets)
where it was difficult for the curve-fitting software to fit a precise value of the asymptote.
Apparently, in these cases the moderately asymptotic behavior seen up to 20 psi was so weak it
was difficult to find an asymptotic value in the modified Page equation that fit better than many
other values. Even in those cases there were values for the asymptote that allowed the modified
Page equation to fit better than the other equations. The MMF equation did not exhibit a problem
with fitting it’s asymptote in those cases.
Table 2.1 Standard error summary for five models for density vs. pressure on HRW wheat samples.
Average and Maximum Standard Errors (lb/bu)
Number Average SE (from 27 varieties) Maximum SE (27 varieties)

Model of Coef. MC #1 MC #2 MC #3 Average MC #1 MC #2 MC #3 Maximum
Fixed Polyn. Eqn. 3 0.0309 0.0290 0.0373 0.0324 0.0541 0.0490 0.0680 0.0680
Modified Polyn. Eqn. 4 0.0248 0.0261 0.0212 0.0240 0.0398 0.0439 0.0426 0.0439
Modified Page Eqn. 3 0.0113 0.0136 0.0120 0.0123 0.0194 0.0229 0.0276 0.0276
MMF Eqn. 4 0.0163 0.0186 0.0155 0.0168 0.0254 0.0289 0.0300 0.0300
Farazdaghi‐Harris Eqn. 3 0.0164 0.0186 0.0155 0.0168 0.0255 0.0289 0.0301 0.0301
Averages: 0.0200 0.0212 0.0203

0.0205
MC # 1= Moisture content of 9.7% wb; MC # 2= Moisture content of 12.7% wb; MC #3= Moisture content of 14.4% (wb). TW
at MC #1 = 63.3325 lb/bu, TW at MC # 2 = 62.1974 lb/bu, TW at MC#3 = 61.1136 lb/bu.
The other four models had higher standard errors than the modified Page equation and were
readily ranked by the average standard errors that mirrored the ranking for each individual data
set also. The MMF and Farazdaghi-Harris Equations ranked essentially equal slightly behind
modified Page, followed by the modified polynomial equation ranking fourth and the original
fixed polynomial equation with a little higher standard errors ranking fifth. Standard error is
generally the best indicator of how well a predicted model fits the experimental data as long as
the plot of the residuals doesn't indicate a problematic pattern. A potential issue is the maximum
standard error seen for the worst case for each model (Table 1) — this value may suggest the
model that is best overall, since it indicates which model best handled the hardest-to-fit cases.
However, when using maximum standard error the same ranking of the five models was seen as
with the average standard errors.
In the modified Page equation the asymptote value, d∞, is determined from the curve fitting
program rather than setting the asymptote at the kernel density value for that case. This
modification produced a better fit, i.e., lower standard error values, since the data was found not
asymptotically approaching the kernel density in this pressure range. The asymptotes (or the
maximum density value, D∞) obtained from fitting the modified Page equation ranged from
approximately 70 lb/bu up to 110 lb/bu with two cases a little higher than that. These asymptotes
22
from the results of curve fitting are called “apparent asymptote” because, theoretically, the
asymptotic value (D∞) for each case should be the kernel density (KD), which is about 110 lb/bu.
This phenomenon of the apparent asymptote is believed to be caused by the initial mechanisms
for compaction at low pressures in which large amounts of particle rearrangement occurs. This
results in a steep density-pressure curve at those low pressures. At higher pressures, much less
particle rearrangement occurs and much more particle to particle compaction occurs. At higher
pressure the dominant mechanism is particle to particle compaction with only small amounts of
particle rearrangement, thus a smaller slope that looks much like an asymptote occurs. The
models with asymptotes respond to this shape and fit the apparent asymptote based on this
lessening slope as the density increases near 20 psi. When much higher pressures were evaluated,
it was apparent that this slope change was not due to a true asymptote, however.
Figure 2-12 and Figure 2-13 indicate that the current range of pressure, 0 to 20 psi, was
well below pressure values where the kernel density and a true asymptotic value, would be
reached. Those figures indicate that the density curve increases further, from the level of the
apparent asymptote to the kernel density only at much higher pressure values (greater than 150
psi). The apparent asymptote is likely associated with the end of much of the kernel re-
arrangement within the compression box. This rearrangement phenomenon likely reaches a
maximum level, which would result in asymptotic behavior of the curve, but other phenomena
begin to allow additional compression and density increases so the density increase did not cease
when the kernel rearrangement reached its maximum. The asymptotes obtained from fitting the
MMF model were near the low end of the range obtained with the modified Page equation, from
about 66 to 78 lb/bu. The other three models have no asymptotic terms.
Figure 2-14 shows the residual plots of density difference vs. pressure for variety TAM 111
(average data from three geographic locations) for three moisture contents. For all models (5), all
varieties (27), and all moisture contents (3) the residual plots showed a similar pattern —
generally nearly identical patterns for the different equations, with different magnitudes leading
to the different standard errors seen for the equations. However, in a typical plot of density
change versus pressure such as for the MMF model in Figure 2-11, these deviations are too
small to be noticed. The residual pattern was a result of how kernel rearrangement and
compression behaved. These residuals are very small (the maximum residual anywhere for the
modified Page equation was about 0.023 lb/bu, or about 0.04% of the average density in the
tests). Apparently, there are subtle, but consistent changes in the way kernel rearrangement and
compression change within this pressure range that are not perfectly described by these smooth
curves, even though the deviation is too small to see at the full scale and range of pressure in the
standard compressibility tests.
2.4.1 Summary of Model Equations

This analysis indicates that all of the new equations showed improvement over the previous
fixed polynomial equation with the modified Page being slightly better than the others for
characterizing the laboratory density-pressure data. With the modified Page equation it was
23
occasional difficult for the best-fit nonlinear regression software to fit the asymptotic term
precisely. The MMF and Farazdaghi-Harris models fit nearly as well as the modified Page model
and did not exhibit any adverse curve-fitting characteristics. The changing mechanisms causing
compression at low pressures seem to cause small fluctuations in the density-pressure curves that
cannot be described by smooth curves like these. However, these new models are still extremely
accurate at representing the laboratory compressibility data as a function of overburden pressure.
24
Modified Poly Modified Poly Modified Poly
Figure 2-14 Residual plots of density difference (di – TW) vs. pressure of TAM 111 for four candidate models with wheat samples at
MC #1 (9.7% wb), MC #2 (12.7% wb), and MC #3 (14.7% wb
25
3. FIELD MEASUREMENT OF COMPACTION FACTORS
3.1 Experimental Methods
Field data for the project was collected from storage bins containing wheat, corn, soybeans,
sorghum, oats, or barley from 2010 to 2013 in 14 U.S. states. Bins were measured in Arkansas,
Colorado, Idaho, Iowa, Kansas, Kentucky, Minnesota, Michigan, Montana, Nebraska, North
Dakota, Oklahoma, South Dakota, and Texas. Farmer- and elevator-cooperators were contacted
and those who were willing to cooperate and provide data on grain quantities and properties of
their stored crops were visited and their vertical storage bins measured. The dimensions of the
storage structure and the grain volume were measured using a Leica Disto D8 laser distance
meter (Leica Geosystems AG, St. Gallen, Switzerland). The angle of repose (AoR) of the grain
surface was determined by taking an average of seven data points between the bin sidewall and
the top of the grain cone (if coned up) or to the bottom of the cone (if coned down). For flat
surface profiles, seven data points, on average, were also taken to determine the average grain
height. For complex surface profiles, up to 50 surface data points were measured. This data was
used to calculate the total volume of grain inside the bin, which included the cone volume,
cylinder volume, and hopper volume (if present).
In some cases the volume of the grain was calculated using volume per unit depth
(provided by the warehouse examiner’s audit sheets) as well as the laser depth profiling.
Conventional measurement techniques using a weighted tape measure were also used to
supplement the laser distance meter depth measurements, and to verify the circumference of
round metal bins.
In this study grain height was defined as the height from the permanent structural floor of
the bin to the point where the grain intersects the bin side wall. To differentiate, the eave height
was defined as the height from the permanent structural floor to the point where the sidewall of
the bin touches the roof of the bin. The eave height is a fixed value irrespective of grain volume
and grain slope inside the bin. Where direct volume calculations were difficult to determine
because of an irregular grain surface a computer aided design (CAD) software package (DS
SolidWorks Corp., Concord, Mass.) was used along with the height and dimension data to draw
the grain bin configuration. In some cases (e.g., hopper bottom bins), the empty-bin
measurements were also determined, either before filling the bin or after emptying. Information
such as method of filling, i.e., centered or off-centered, the use of spreader, test weight (bulk
density), moisture content, and percent dockage of the incoming load were in all cases provided
by farmers and elevator managers. Table 3-1 lists parameters that were collected for each bin.
The mass of grain inside the bin and corresponding grain properties were reported by
farmers after the grain was sold. For commercial bins the mass and corresponding grain
properties (i.e., test weight, moisture content, and dockage) were tracked and the volume was
measured when filling was complete and, when convenient, at intermediate points during filling.
Depending on the practice of the elevator, the general methodology was adjusted to obtain
26
accurate mass measurements and grain properties, while at the same time measuring grain
volume.
Table 3-1. Parameters from on-site bin measurement for compaction factor determination.
No. Parameters
1. Bin diameter (ft)
2. Eave height (ft)
3. Plenum height (ft)
4. Hopper bottom angle (degrees)
5. Grain height (ft) (calculated from bin measurements)
6. Angle of repose (degrees) (calculated from bin measurements)
7. Total volume of grain inside the bin (bu) (calculated from bin measurements and angle of repose)
8. Mass of grain from grain scale measurement (lb)
9. Method of filling the grain (centered or off-centered fill point)
10. Average test weight (lb/bu) (with or without dockage)
11. Moisture content (w.b.)
12. Dockage amount (%)
3.1.1 Density and Compaction Relationships

The fundamental concept of grain compaction is a ratio of the bulk density after
compaction in the bin to the initial bulk density of the grain before compaction. The initial bulk
density is equivalent to the grain test weight (TW). Thus compaction ratio is given by:
DA
P (3.1)
D0
where:
n
D X
DA  x 1
(3.2)
n
and:
DA = average bulk density of the grain after compaction, kg/m3 (lb/bu)
D0 = initial bulk density, or un-compacted test weight (TW), of the grain from the
Winchester Bushel Test, kg/m3 (lb/bu)
DX= bulk density of grain in the bin at a given depth, x, kg/m3 (lb/bu)
P = average compaction ratio of the grain
n = number of layers in the grain depth, and
x = depth in the grain, m (ft)
27
The compaction factor is defined as the reduction in volume of the stored grain due to
compaction (a factor usually expressed as a percent) and this term was used in the ASABE
Standards EP413.2 (ASABE Standards, 2010):
f C  P 1 (3.3)
where:
fC = average compaction factor of the grain
The mass of grain in a bin can be calculated from the measured volume of grain using:
M  D0  P  V (3.4)
where:
M = mass of grain in the bin, t (lb), and
V = measured volume of grain in the bin, m3 (ft3 or volumetric bu)
The common grain industry term, used by both RMA and FSA, is pack factor, R. This is also
referred to as “Combined Test Weight and Pack Factor” by RMA and is defined as:
D 
R  P  0  (3.5)
 DS 
where:
R = pack factor including test weight, and
DS = standard bulk density (standard test weight) for a given grain type, kg/m3 (lb/bu)
Then mass of grain measured in standard bushels is calculated as:

Standard Bu  R  V (3.6)
The mass of grain in lb is:

M  DS  Standard bu  DS  R  V (3.7)
3.1.2 Application of WPACKING Model

Data from field measurements were used as input in the WPACKING model. Input
variables included type of grain, grain properties (i.e. moisture content and test weight), grain
height, bin diameter, type of bin wall, hopper dimensions (if applicable) and method of bin
28
filling (resulting in top grain surface of coned upward, level, or coned downward). Based on
these values the model yielded the WPACKING compaction factor, fC,W, the WPACKING pack
factor, RW, and the predicted mass of the measured grain. The compaction factor, fC,W, and the
WPACKING pack factor, RW, were determined as an average of the values calculated at specific
grain depths by the model using the following equations.
D 
f C, X   X   1 (3.8)
 D0 
n
f C, X
f C, W  x 1
(3.9)
n
PW  f C,W  1 (3.10)
D 
R W  PW  0  (3.11)
 DS 
where:
fC,X = compaction factor of the grain determined by WPACKING at a given depth, x
fC,W = average compaction factor of the grain determined by WPACKING
n = number of depth layers used in WPACKING
PW = average compaction ratio of the grain determined by WPACKING
RW = pack factor with TW of the grain determined by WPACKING
The total mass of grain in the bin was predicted using Eq. (3.6). Pack factors for concrete
and steel bins (corrugated and smooth walled) containing HRW wheat were predicted by the
model. Values of the reported mass determined from scale measurements for each bin load were
compared to the predicted mass from the model.
3.1.3 Application of RMA Method

The Federal Crop Insurance Corporation of the USDA Risk Management Agency (RMA) is
the primary crop insurance provider in the U.S. Their Loss Adjustment Manual and Handbook
for small grains (USDA-RMA, 2012a and 2012b) specifies the use of pack factors for adjusting
grain bin measurements based on grain test weights and bin diameters. The RMA Pack Factor,
RR, is somewhat similar to compaction ratio defined above, but includes the TW measured for
the bin that was not in the definition of compaction ratio. The RMA method uses conventional
bin measurements and the RMA Pack Factor (obtained from RMA loss adjustment manual and
29
handbook) along with volume calculations to arrive at the final mass of grain in the bin. The
RMA Pack Factor simplifies the calculation by eliminating the test weight adjustment step.
To compare the RMA Pack Factor with other pack factor terms used in this study, the
following equations were used:
f C, R  P R - 1 (3.12)
D 
R R  PR  0  (3.13)
 DS 
where:
DS = standard bulk density (test weight) for a given grain type

fC,R = average compaction factor of the grain determined from RMA method
PR = average compaction ratio of the grain determined from RMA method
RR = RMA Pack Factor from the Loss Adjustment Manual and Handbook
RR represents the amount of packing predicted using the RMA method. The grain mass
calculated using RR were compared to the reported mass in the bins based on scale
measurements.
3.1.4 Application of FSA-W Method

The USDA Farm Service Agency (FSA) warehouse group uses the Federal Warehouse
Examiner’s Handbook WS-3 (USDA, no date). This handbook provides a detailed empirical
method for estimating the compaction factor for six varieties of grain: wheat, corn, milo, rye,
soybean, and oats. This method is based on the initial test weight of the material and the volume
per unit depth in the storage structure modified by an empirical correction value. The method
determines a pack factor, RF, which includes the test weight adjustment similar to the RMA
method.
To compare the FSA Warehouse Group Pack Factor with other pack factor terms used in
this study, the following equations were used:
CPI AF
f C, F   D o  D S  (3.14)
100 100
CPI AF
PF  1   D o  D S  (3.15)
100 100
 CPI  
R F  1  
AF
D o  D S   D o  (3.16)
 100 100   DS 
30
where:
CPI = converted packing index from Federal Warehouse Examiner’s Handbook WS-3
AF = FSA Warehouse Group method adjustment factor
fC,F = average compaction factor of the grain determined from FSA Warehouse Group method
PF = average compaction ratio of the grain determined from FSA Warehouse Group method
RF = FSA Warehouse Group Pack Factor
The grain mass calculated using RF was compared to the reported mass in the bins based on
scale measurements.
3.2 HRW Wheat Field Data
3.2.1 Data Collection

Data from this portion of the project was collected from storage bins containing HRW wheat
from 2010 to 2013 in Kansas, Oklahoma, and Texas.
3.2.2 Results and Discussion

Table 3-2 shows the summary of storage bin and grain characteristics for steel bins
containing HRW wheat measured from Kansas and Oklahoma. All of the steel bins measured
were corrugated steel bins. The bin diameters ranged from 15 to 104.6 ft (4.6 to 31.9 m) with
equivalent level grain heights from 13.4 to 84.1 ft (4.1 to 25.6 m).
The measured AoR for HRW wheat in steel bins ranged from 18.0 to 38.6 degrees with a
mean value of 24.5 degrees. Typically, the AoR for wheat crops ranged from 16.0 to 41.0
degrees (for filling or piling) and from 23.8 to 45.5 degrees (for emptying or funneling) (Stahl,
1950; Lorenzen, 1957, 1959; Mohsenin, 1986; Midwest Plan Service, 1987; Molenda and
Horabik, 2005). The measured AoR from the field are within the range of published AoR for
filling but did not necessarily follow the literature trend of increasing AoR with increasing
moisture content, probably due to the variability in test weight, amount of dockage, and kernel
size distribution in the wheat. The wide range in the measured AoR (18.0 to 38.6 degrees) may
also have been affected by the variation in the method of filling.
Ten of the steel bins monitored in this study had flat bottoms, while 6 bins were equipped
with hopper bottoms. The mean hopper bottom angle was 34.1 degrees and it ranged from 27.4
to 38.0 degrees. The lower range of angle seems to be a bit shallow for normal steel bin. Usually,
hopper bottom angle more than 30 degrees is expected. However, we believe due to dry HRW
wheat storage, may be 27.4 degrees would provide a reasonable shallowness to empty the bin.
31
Some steel bins were loaded off-center, i.e., the peak of the cone was offset from the center of
the bin (Table 2). The HRW wheat in steel bins had an average test weight of 59.6 lb/bu (767.7
kg/m3), average moisture content of 11.6% (w.b.), and average dockage of 0.57%.
The range of WPACKING pack factor, RW, was from 0.929 to 1.073, with the mean of
1.036. The RMA pack factor, RR, range was higher at 0.991 to 1.157. The RMA mean pack
factor was also the highest at 1.087. The FSA-W pack factor, RF, range was in between the
former two methods (0.985 to 1.126), with a mean of 1.082.
32
Table 3-2 Summary of bin characteristics and HRW wheat properties for the measured steel bins.[a]
Average Equivalent WPACKING

Type of Bin Type of Wall Method of Moisture Average Test Average Equivalent Level Height Pack Factor, RMA Pack FSAPack
Location Bin Configuration Bottom Finish Filling Content (%, wb) Weight (lb/bu) Dockage (%) Diameter (ft) (ft) RW Factor, RR Factor, RF
Kansas Round flat corrugated off-centered 12.3 60.9 0.44 15.0 20.7 1.050 1.048 1.068
Kansas Round flat corrugated centered 11.4 62.4 0.41 30.0 15.9 1.073 1.099 1.126
Kansas Round hopper corrugated off-centered 11.1 59.8 0.57 30.0 21.6 1.032 1.062 1.087
Kansas Round hopper corrugated centered 11.7 60.7 0.79 30.0 19.6 1.047 1.075 1.100
Kansas Round hopper corrugated centered 12.0 59.0 0.64 104.6 73.7 1.040 1.126 1.084
Kansas Round hopper corrugated off-centered 11.9 60.2 0.38 30.0 20.2 1.039 1.068 1.093
Oklahoma Round hopper corrugated centered 10.7 52.7 0.60 89.5 78.7 0.929 1.021 0.985
Oklahoma Round hopper corrugated centered 10.6 57.9 0.70 89.5 71.0 1.016 1.107 1.068
MIN 10.2 52.7 0.20 15.0 13.4 0.929 0.991 0.985
MAX 13.1 62.4 0.96 104.6 84.1 1.073 1.157 1.126
AVG 11.6 59.6 0.57 48.4 38.6 1.036 1.087 1.082
SD 0.8 2.3 0.24 30.0 26.2 0.036 0.049 0.037
MEDIAN 11.6 60.4 0.55 36.0 22.2 1.048 1.087 1.089
[a] Minimum values = MIN, Maximum values = MAX, Average values = AVG, Standard deviation values = S.D.
Table 3-3 shows storage bin and grain characteristics for HRW wheat measured in concrete bins
in Kansas, Oklahoma, and Texas. The diameter of the bins ranged from 15 to 33.7 ft (4.6 to 10.3 m) with
equivalent level grain heights ranging from 17.9 to 136.6 ft (5.5 to 41.6 m). As is typical of concrete
bins at commercial elevators, these bins were tall and narrow. The tallest grain height was 136.6 ft (41.6
m), which was higher than all of the steel bins measured (84.1 ft or 25.6 m).
The measured AoR for HRW wheat in concrete bins ranged from 15.7 to 45.1 degrees with a mean
value of 26.1 degrees. The AoR for HRW wheat in concrete bins were within the range of those found in
the literature (16.0 to 41.0 degrees) for piling or filling (Stahl, 1950; Lorenzen, 1957, 1959; Mohsenin,
1986; Midwest Plan Service, 1987; Molenda and Horabik, 2005), except for the three extreme values of
15.7, 43.4 and 45.1 degrees. The mean AoR value of those in concrete bins was steeper than those
measured in the steel bins (24.5 degrees). Some of the bins have cones with AoR ranging from 1.8 to 9.6
degrees. These were considered as apparent AoR, since these were representation of remnants of cones
and were not included in the calculation of the mean AoR.
All concrete bins had hopper bottom with either center-draw or side-draw. The mean hopper
bottom angle was 35.4 degrees and ranged from 33.8 to 47.3 degrees. The hopper bottom angles for
concrete bins were higher than those for the steel bins (27.4 to 38.0 degrees). Except for few cases, most
of the bins were centered filled. The grain characteristics were similar to those found in the steel bins
with a mean test weight of 59.9 lb/bu (771.5 kg/m3), mean moisture content of 11.5% (w.b.), and mean
dockage of 0.66%. None of the concrete bins measured used spreaders during filling. In most cases the
cross-section of the bin was circular. However, in the concrete bins located in Oklahoma and some in
Texas, the cross-section was a regular hexagon. In calculating the capacity of the bin using WPACKING
the equivalent diameters were used for these non-circular bins.
WPACKING pack factors, RW, for concrete were slightly higher than those for steel, ranging from
0.986 to 1.077, with the mean of 1.040. However, the RMA pack factor, RR, for concrete was slightly
lower, with mean of 1.036 and ranged from 0.993 to 1.099. The FSA-W pack factor, RF, range was
higher than the former two methods (1.012 to 1.101), with a mean of 1.056.
33
Table 3-3 Summary of bin characteristics and HRW wheat properties for the measured
concrete bins.[a]
Average
Moisture Average Equivalent WPACKING
Bin Type of Bin Method of Content (%, Test Weight Average Equivalent Level Pack Factor, RMA Pack FSAPack
Location Configuration Bottom Filling wb) (lb/bu) Dockage (%) Diameter (ft) Height(ft) RW Factor, RR Factor, RF
Kansas Round hopper off-centered 11.9 57.4 0.39 18.0 91.6 1.000 0.998 1.022
[c]
Kansas Round hopper off-centered 11.5 57.3 0.39 15.0 74.6 0.994 0.996 1.015
[c]
Kansas Round hopper centered 12.5 60.0 n.a. 15.0 68.2 1.041 1.035 1.055
Oklahoma Hexagon hopper centered 12.0 61.9 0.40 15.7 136.6 1.076 1.062 1.083
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[c]
[b]
[b]
[b]
[b] [c]
[b] [c]
[b] [c]
[b] [c]
[b] [c]
[b] [c]
[b] [c]
Texas Round hopper centered 11.5 60.2 0.80 33.7 129.7 1.055 1.099 1.101
Texas Interstice hopper centered 11.3 60.2 0.78 16.7 123.3 1.045 1.037 1.060
Texas Hexagon hopper centered 11.3 59.6 0.68 17.1 118.0 1.037 1.030 1.052
Texas Hexagon hopper off-centered 11.7 59.3 0.64 17.1 116.6 1.032 1.026 1.048
MIN 10.0 57.1 0.30 15.0 17.9 0.986 0.993 1.012
MAX 12.5 62.1 1.60 33.7 136.6 1.077 1.099 1.101
AVG 11.5 59.9 0.66 16.5 91.7 1.040 1.036 1.056
SD 0.6 1.5 0.24 3.1 40.0 0.026 0.023 0.022
MEDIAN 11.5 60.1 0.60 15.7 111.3 1.044 1.036 1.057
[a] Minimum values = MIN, Maximum values = MAX, Average values = AVG, Standard deviation values = S.D.
[b] These bins have been measured from a benchmark point indicating that they were not empty.
[c]
side-discharge hopper bottom bins
34
3.2.3 Comparison of WPACKING Model, RMA and FSA-W Methods
Figure 3-1 shows the difference between the reported mass (from scale measurements) and
the predicted mass from the WPACKING model, and RMA and FSA-W methods for corrugated
steel bins. The range of difference between the reported mass and the predicted mass by
WPACKING was from -4.70% to 1.17, with a median difference of -1.26%, indicating that the
model was under-predicting by 1.26% compared to the reported mass. There were only three
cases with an over prediction by WPACKING, two flat-bottom and one hopper-bottom steel
bins. WPACKING predicted mass from the large steel bins were always within -2% of the
reported mass.
The difference between the reported mass and RMA-predicted mass for steel bins ranged
from -3.71% to 11.0% (Figure 3-1). The median was 1.91%, which indicated that the RMA
method was over-predicting by 1.91% compared to the reported mass. The median difference
was higher than the WPACKING model (-1.26%). Seventy-five percent of measure steel bins
were over-predicted by the RMA method.
The range of the difference between the predicted mass from the FSA-W method and the
reported mass was from -1.40% to 7.63%, of which 12.5% of all the values was over-predicting
the reported mass of the steel bins (Figure 3-1). The median difference was 3.86%, which was
higher than both WPACKING (-1.26%) and RMA (1.91%) median differences.
35
Reported Mass (bu)
0 100,000 200,000 300,000 400,000 500,000

12%
10% WPACKING
RMA
8%
FSA
6%
% Difference
4%
2%
0%
-2%
-4%
-6%
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
Reported Mass (t)
Figure 3-1 Percent difference between WPACKING-, RMA-, and FSA-W-predicted and
reported mass for HRW wheat from corrugated steel bins.
Reported Mass (bu)

0 200,000 400,000 600,000
16,000 600,000
WPACKING Predicted Mass (bu)

WPACKING Predicted Mass (t)
12,000
400,000
8,000
200,000
4,000
0 0
0 4,000 8,000 12,000 16,000
Reported Mass (t)
Figure 3-2 WPACKING-predicted mass plotted against the reported mass for HRW
wheat from corrugated steel bins.
36
Reported Mass (bu)
0 200,000 400,000 600,000
16,000 600,000
RMA Predicted Mass (bu)

RMA Predicted Mass (t)
12,000
400,000
8,000
200,000
4,000
0 0
0 4,000 8,000 12,000 16,000
Reported Mass (t)
Figure 3-3 RMA-predicted mass plotted against the reported mass for HRW wheat
from corrugated steel bins.
Reported Mass (bu)

0 200,000 400,000 600,000
16,000 600,000
FSA Predicted Mass (bu)

FSA Predicted Mass (t)
12,000
400,000
8,000
200,000
4,000
0 0
0 4,000 8,000 12,000 16,000
Reported Mass (t)
Figure 3-4 FSA-W-predicted mass plotted against the reported mass for HRW wheat from
corrugated steel bins.
37
Figure 3-5 shows the comparison of the difference between WPACKING-, RMA-, and FSA-W-
predicted and scale mass for reinforced concrete bins. The range of difference of WPACKING-
predicted and reported mass for concrete bins was between -5.89% and 9.74%, which is higher
than that of corrugated steel bins. Measurements from concrete bins are often difficult due to its
construction, tall and narrow bin shape. However, the median difference was 2.16%, with the
majority of the bins (23 out of 35) being over-predicted by WPACKING.
The range of difference between RMA-predicted mass and the actual mass of grain reported
from scale measurements for concrete bins was from -7.19% to 7.75% (Figure 3-7). The median
difference was 1.03%, which was lower than the WPACKING model (2.16%). Most of the data
for concrete bins (21 out of 35) were over-predicted.
The range of the difference between the predicted mass from the FSA-W method and the
reported mass was from -4.70% to 10.43%, of which 30 out of 35 bins were over-predicted
(Figure 3-5). The median difference was 3.50%, which was higher than both WPACKING
(2.16%) and RMA (1.03%) median differences.
The reported mass and the predicted mass from WPACKING model, the RMA method,
and the FSA-W method were plotted against a 45-degree line to show how closely the measured
values were predicted. Figure 3-6, Figure 3-7, Figure 3-8 illustrate that the WPACKING model
closely predicted reported mass than the other two methods.
38
Reported Mass (bu)
0 20,000 40,000 60,000 80,000 100,000
12%
10%
WPACKING
8%
RMA
6% FSA
% Difference
4%
2%
0%
-2%
-4%
-6%
-8%
0 500 1,000 1,500 2,000 2,500
Reported Mass (t)
Figure 3-5 Percent difference between WPACKING-, RMA-, and FSA-W-predicted
and reported mass for HRW wheat from reinforced concrete bins.
Reported Mass (bu)

0 50,000 100,000 150,000
3,000 150,000
WPACKING Predicted Mass (bu)

WPACKING Predicted Mass (t)
2,500
2,000 100,000
1,500
1,000 50,000
500
0 0
0 500 1,000 1,500 2,000 2,500 3,000
Reported Mass (t)
Figure 3-6 WPACKING-predicted mass plotted against the reported mass for HRW
wheat from reinforced concrete bins.
39
Reported Mass (bu)
0 50,000 100,000 150,000
3,000 150,000
RMA Predicted Mass (bu)

RMA Predicted Mass (t)
2,500
2,000 100,000
1,500
1,000 50,000
500
0 0
0 500 1,000 1,500 2,000 2,500 3,000
Reported Mass (t)
Figure 3-7 RMA-predicted mass plotted against the reported mass for HRW wheat
from reinforced concrete bins.
Reported Mass (bu)
0 50,000 100,000 150,000

3,000 150,000
2,500
FSA Predicted Mass (bu)
FSA Predicted Mass (t)
2,000 100,000
1,500
1,000 50,000
500
0 0
0 500 1,000 1,500 2,000 2,500 3,000
Reported Mass (t)
Figure 3-8 FSA-W-predicted mass plotted against the reported mass for HRW wheat
from reinforced concrete bins.
40
From the WPACKING model, the absolute mean differences for steel (1.64%) and concrete
(3.75%) vary more from each other than their median differences (-1.26% and 2.16%,
respectively). The RMA absolute mean difference for steel (4.41%) was higher than
WPACKING model (1.64%) and the FSA-W method (3.40%). But the RMA absolute mean
difference for concrete bins (3.25%) was lower than the WPACKING model (3.75%) and the
FSA-W method (4.34%).
Overall, it is clear from Figure 3-1 and Figure 3-5 that the percent difference range for concrete
bins (-5.89% to 9.74%) is greater than steel bins (-4.70% to 1.17%) from the WPACKING
model. Perhaps one of the primary reasons could be that WPACKING model (ASABE
Standards, 2010) was mainly developed using model constant values generated from steel bins.
This may explain why the difference between WPACKING predicted and actual reported mass
of grain is higher for concrete bins compared to steel bins. Data in Figure 3-1 and Figure 3-5
also indicate that the model predictions can be improved by adjusting values in the model to
produce more accurate values than those currently being predicted using the current
compressibility relationship in the model for HRW wheat. Adjustment in the model can mean
getting more accurate values for k and μ by fine tuning within the limits of regression analyses.
The RMA predicted mass values deviated further from the measured mass in the steel bins
compared to the WPACKING predictions. Some of the possible reasons could be: the RMA pack
factors are based on empirical regression equations on few limited size bins and then for larger
capacities, the pack factors were extrapolated. RMA pack factors were taken from county FSA
pack factor database, however, source of those factors are unknown. Thus, for larger steel bins (≥
78.3 ft or 23.9 m in diameter) predicted mass of grain were calculated to be a lot different
(6.75% to 11.0% difference) from the reported (actual) mass of grain in the bin. Also, the effect
of bin wall type was not accounted in RMA pack factor numbers. Bin wall type and material
could affect the coefficient of friction and lateral to vertical pressure inputs as stated in the
Janssen’s (1895) model. However, the RMA predicted mass were closer to the reported mass for
concrete bins than those values from the WPACKING model. It may be because years ago we
had more concrete storage than steel bins, the RMA method may have been calibrated their data
for much older storage conditions.
On the other hand, the FSA-W method was better in predicting mass values from steel bins
than the RMA method, but not the concrete bins. Perhaps FSA-W data was calibrated more with
steel bins than concrete bins.
Regarding all three prediction; WPACKING model, RMA, and FSA-W; any error in the volume
measurement propagates through the determinations of predicted mass by WPACKING and by
the RMA and FSA-W methods. For example, a grain height difference of 2.5 cm would change
the predicted mass of grain by 3400kg (0.07%), and a difference of 5 cm would be 0.21% higher
than the original predicted mass. Thus, some of the errors in the predicted mass calculations can
be attributed to volume measurement errors. These measurement errors should be random errors
that don't affect the overall assessment of errors in the three prediction methods.
41
3.3 Test Weight and Dockage Relationships
3.3.1 Background
Agricultural grains such as wheat, corn, and soybean are compressible materials and, in
storage, they are affected by pressures from overbearing loads. However, the degree of
compressibility of stored grain varies with grain type, grain properties, and the geometry of the
bin in which the grain is stored. Several studies have investigated the compressibility of a variety
of food crops, such as ground shelled corn, wheat, corn, soybean, corn meal, sugar beet pulp,
cotton seed meal, and distillers grains without solubles (Loewer et al. 1977; Clower et al., 1973;
Malm and Backer, 1985;). Additional studies have been conducted related to the effect of grain
spreaders on the bulk density of stored wheat, yellow corn, and sorghum (Chang et al., 1981) and
different methods of transfer, such as choke fed and non-choke fed through an orifice (Change et
al., 1983).
Janssen’s (1895) equation is commonly used to predict the vertical and lateral pressures
in bins and is based on the bulk density of the stored material, coefficient of friction, lateral to
vertical pressure coefficient and the bin geometry. Studies have also been conducted in which
the degree of compressibility or packing of grain in bins has been estimated using a form of
Janssen’s equation (Thompson and Ross, 1983; Thompson et al., 1987; McNeill et al., 2008).
Grain packing models based on Janssen’s equation were adopted by ASAE standards in 1992
and later revised in 2010 (ASAE S413.2, 2010). However, based on a review of the literature,
little validation and calibration of any of these packing models has been performed with field
data from commercial bins.
Inventory control and proper documentation of stored grain is extremely important for
farmers, elevator managers, and in designing of grain elevators and is crucial for the grain bin
managers, who must track the quality of the crop and meet federal and state regulatory
obligations. For each truck load sample of grain stored in a bin, extrinsic and intrinsic
parameters of the stored grain are measured. Some of the important extrinsic grain quality
parameters are moisture content, test weight, dockage, foreign material, amount of damaged
kernels (mold damaged, purple pigmentation for sorghum kernels, and scab-damaged kernels for
wheat described as having a dull and chalky appearance), and broken and shrunken kernels
(mainly for corn), whereas the typical intrinsic parameters are protein content for wheat and
soybeans; oil content for canola, flaxseed, and sunflower seed; and determination of toxins, such
as Aflatoxin and DON (Deoxynivalenol, also referred to as Vomitoxin) for corn grain.
Guidelines on how to measure and analyze these parameters are given in the United States
Department of Agriculture – Grain Inspection, Packers and Stockyards Administration (USDA-
GIPSA) Handbook (USDA-GIPSA, 2009). Official inspection of grain by a state or federal
regulatory body can only be conducted using Federal Grain Inspection Service (FGIS) approved
equipment and procedures.
42
Wheat is grain that contains 50% or more whole kernels (after removal of the dockage) of
one of the following classes: Hard Red Spring (HRS), Hard Red Winter (HRW), Soft Red
Winter (SRW), Durum, Hard White (HW), Soft White (SW), mixed wheat, and unclassified
wheat (USDA GIPSA, 2009). HRW is the most widely planted and available wheat variety in
the U.S. It is a general norm that whenever possible (Matz, 1991), HRW wheat should be planted
preferentially over spring wheat. For HRW wheat, the moisture content, dockage, test weight,
and % of shrunken and broken kernels (SHBN) are the most important extrinsic parameters
measured as per FGIS guidelines. The protein content of wheat in the U.S. ranges from 7.0% to
18.0% for all classes and varieties. HRW wheat normally has higher protein content than other
classes of wheat and is used for making breads, rolls, pastries, pizzas, and other products because
of its elastic and extensible dough characteristics. HRW wheat normally has a slightly higher
market price than other varieties of wheat (Matz, 1991).
When grain is delivered to elevators, samples of HRW wheat are taken from incoming
trucks or trailers using mechanical probes. These samples are evaluated for moisture content
using a Grain Analysis Computer (GAC) meter before the removal of dockage (USDA-GIPSA,
2009). Dockage is the material other than the predominant grain that can easily be removed with
sieves and cleaning devices. Removal of dockage can be performed using an FGIS-approved
dockage tester or with hand sieves. For wheat samples, FGIS guidelines require that the sample
be placed on a series of round-hole sieves (12/64 in. on the top, 5/64 or 4.5/64 in. on the bottom)
in a mechanical sizer for 20 minutes. All the coarse material that remains on top of the top sieve
(12/64 in.) and all the materials that pass through the bottom sieve (5/64 or 4.5/64 in.) are termed
as dockage for all wheat varieties (USDA-GIPSA, 2009).
Test weight is defined as the weight of the volume of grain that is required to fill the
Winchester bushel cup (2,1540.42 in.3) to capacity (USDA-GIPSA, 2009). The unit of test
weight is then given in lb/bu. Thus, the test weight measures the initial bulk density under
specific conditions (Bern and Brumm, 2009). As per FGIS guidelines, the test weight for wheat
samples is determined after removal of dockage. As previously discussed, this standard
procedure of removing the dockage before test weight evaluation brings a new challenge for
research attempting to validate and calibrate grain packing models. To estimate the amount of
HRW wheat in a commercial bin, the test weight with dockage is required and is an important
step before proceeding with any calibration of grain packing models. However, no study has
been performed to correlate and analyze the relationship of test weight with and without dockage
for HRW wheat samples. Thus, the objectives of this paper are:
1. To correlate and develop a regression model for the test weight (TW) both with and
without dockage (lb/bu) for 4 different dockage levels: 0% to 0.39%, 0.4% to 0.59%,
0.6% to 0.9%, and 1% and above.
2. To determine if location has any effect on the global data (irrespective of dockage levels)
for TW with and without dockage plot.
43
3. To develop regression models for predicting the TW with dockage = f(dockage level, TW
without dockage).
3.3.2 Materials and Methods

3.3.2.1 Test Weight Data Collection
The FGIS-approved test weight apparatus consists of a hopper discharge container with a
slide gate and one quart container called a kettle (Seedburo Equipment, Chicago, IL), as
illustrated in the wheat and flour book (Wheat Marketing Center, 2008) . The grain sample is
placed in the hopper container and the slide gate is opened. The kettle is then filled such that the
grain overflows the kettle. The operator then strikes the top of the kettle with a stoker to level the
grain using three full length zigzag motions. The weight of the grain in the kettle is then
measured in pounds and is multiplied by 32 (number of quarts in a volume, i.e., the size of the
Winchester bushel) to obtain the test weight in pounds per bushel (lb/bu). For HRW wheat, the
standard bushel weight of one bushel is 60 lb. According to FGIS policy, the minimum test
weight per bushel for HRW should range from 51.0 to 60.0 lb/bu for all U.S. wheat grades No. 1
to No. 5 (Matz, 1991).
Test weight data were collected from 3 different elevators located in northeast Kansas
(Manhattan, KS), eastern Oklahoma (Enid, OK), and western Kansas (Goodland, KS) during the
wheat harvest season in 2011 and 2013. Test weights of HRW wheat samples without dockage
were obtained directly from scale data reports given by elevator managers. Samples were
collected using an FGIS-approved probing method immediately after a truck or trailer arrived at
the scales. A moisture content analysis was performed after the sample was collected and then
the sample was passed through the dockage tester. After cleaning (removal of the dockage), the
test weight of the sample was measured using the approved FGIS method as described in the
previous paragraph. Thus, the test weight without dockage for HRW wheat samples was
obtained. Additional samples were collected from the incoming trucks, and the TW with
dockage was determined using the same equipment.
3.3.2.2 Statistical Analyses
A statistical analysis was performed using both 2D and 3D regression modeling of test
weights with and without dockage values. The 2D analysis was performed using Microsoft
Excel 2007 (Redmond, WA), and samples were classified based on 4 different dockage levels:
0% to 0.39%, 0.4% to 0.59%, 0.6% to 0.9%, and 1% and above. A 3D surface model of the data
was constructed using CurveExpert Professional software (version 2.0.3, Informer Technologies,
Inc. 2013). A response surface model was used to perform the 3D curve fitting to predict the test
weight with dockage = f(dockage level, test weight without dockage). The 3D response surface
modeling was performed at a 95% significance level.
44
Figure 3-8 shows the plots of TW with dockage versus TW without dockage at the 4
different dockage levels: 0% to 0.39%, 0.4% to 0.59%, 0.6% to 0.9% and 1% and above. The
dockage data were obtained from the incoming load scale ticket information provided by the 3
different elevators. The best fit linear regression lines yielded R2 values from 0.703 to 0.776.
Overall, the R2 values were fairly high (~0.75), which indicated a significant positive correlation
between TW with and without dockage, as expected. The highest R2 (0.776) was determined for
the lowest dockage level (0% to 0.39%); at this dockage level, additional TW data were
collected. The amount of data collected at the higher dockage levels decreased with increases in
dockage. This was expected because in the U.S., dockage levels for HRW wheat samples are
typically below 0.4%. It is believed that a higher better regression fit (i.e., higher R2 values)
would have been possible if additional data were collected at higher dockage levels. However,
correlation coefficients and R2 must be interpreted carefully because a lower R2 value (<0.3) does
not simply mean that those correlations can be ignored. In most cases, emphasis must be given to
the experimental design and the two variables in the study for field data and agricultural sciences
(Kozak et al., 2008). For example, for dockage levels of 0% to 0.39%, the R2 value was 0.776,
which corresponds to an r value of 0.881. No outliers were observed in the dataset (> 2 standard
deviations away from the mean); therefore, our R2 values range from 0.703 to 0.776, which
represent a fairly positive correlation between TW without dockage vs. TW with dockage and
indicate that the best fit linear regression lines in Figure 2 are accurate prediction models for
HRW wheat samples.
45
66
66 Dockage from 0‐ 0.39 %. Dockage from 0.4‐0.59 %.
64
64
y = 0.8852x + 6.3277
y = 0.9021x + 4.9895
R² = 0.776
TW with Dockage (lb/bu)
62 62 R² = 0.7033

60 60
58 58
56
56
54
54
52
52 54 56 58 60 62 64 66 52
TW without Dockage (lb/bu) 52 54 56 58 60 62 64 66
TW without Dockage (lb/bu)
66
66 Dockage from 0.6‐0.9%. Dockage at 1 and above (%)
64 64
y = 0.9487x + 2.0906
R² = 0.717 y = 0.8735x + 6.5483
62 R² = 0.7662
62
TW wiith Dockage (lb/bu)
60 60
58 58
56 56
54 54
52 52
52 54 56 58 60 62 64 66 52 54 56 58 60 62 64 66
TW without Dockage (lb/bu) TW without Dockage (lb/bu)
Figure 3-8 Plot of TW with dockage vs. TW without dockage (lb/bu), classified based on the 4 dockage
levels of the HRW wheat samples. At dockage =0%, TW without dockage (x axis) = TW with dockage (y
axis).
46
The R2 value for the global dataset (all collected data) of TW (Figure 3-9A) with and without
dockage was determined to be 0.736, which means a reasonable degree of correlation between
the two variables. For laboratory-based experiments, R2 values above 0.95 are highly desired,
and they are acceptable for scientific research. However, field data contain inherent variability;
therefore, correlations with R2 values of 0.95 are rare. Thus, for this dataset, R2 values of 0.736
are fairly reasonable and acceptable for scientific purposes (Kozak, 2012). However, it was also
important to determine the effect that location had on the randomness of the global TW dataset
for HRW wheat samples. Figure 3-9B shows the random distribution of the TW data for each
location (Enid, OK, Goodland, KS, and Manhattan, KS) in this study. If the location in the
global data set produces a bias, then the regression model would have limited practical use. In
this study, bias was not found for the location, which is depicted in Figure 3-9B. In addition, a
portion of the TW data (Figure 3-9B) from Goodland, KS resulted from tracking a 105 ft. bin;
thus, there was additional data from Goodland, KS than the other two locations. However, the
Goodland, KS data (red color, Figure 3-9B) showed a random distribution, as expected. The best
fit linear regression lines for all dockage levels and the global dataset are shown in Figure 3-10.
Figure 3-10 clearly shows that TWs without dockage were higher than TWs with dockage for
each dockage level as well as the global data set. For example, a TW without dockage of 58
lb/bu is equivalent to a TW with dockage of 57 lb/bu in the 0.60% to 0.9% dockage class. Wheat
dockage wheat is made up of chaff, straw, or broken kernel or fines (USDA-GIPSA, 2009) and is
lighter than wheat kernels; therefore, TWs with dockage will be lower (lighter) than TWs
without dockage.
There was crossover of the 0.6% to 0.9% dockage level best fit regression line (dotted green)
over other dockage level lines after 59 lb/bu TW without dockage, indicating a possible overlap
of predicted TW values (Y axis) with other dockage levels.
47
66
Global dataset
64
y = 0.8761x + 6.6681

62 R² = 0.7355
60
58
56
54
52
52 54 56 58 60 62 64 66
A
66.00 Global Data (Location‐Wise)
64.00 Enid, OK, 2011
Goodland, KS, 2011

62.00
Manhattan Coop, KS,

60.00 2013
58.00
56.00
54.00
52.00
52 54 56 58 60 62 64 66
B
Figure 3-9A Plot of TW with dockage vs. TW without dockage (lb/bu) for HRW wheat samples for
the global dataset irrespective of dockage levels
Figure 3-9B Plot of TW with dockage vs. TW without dockage (lb/bu) indicating the location of the
HRW wheat samples.
48
66
All Best‐Fit Lines
64
62
60 Global
0 to 0.39% Dk
0.40 to 0.59% Dk
58
0.60 to 0.9 % Dk
> 1% Dk
56
54
54 56 58 60 62 64 66
Figure 3-10 Best fit regression lines for TW with dockage vs. TW without dockage (lb/bu) plot for each
dockage level of the HRW wheat samples. Dk is Dockage.
A 3D regression model for predicted TW with dockage (lb/bu) (Y, dependent variable) =
f(dockage level (%) (X1, independent variable), TW without dockage (lb/bu) (X2, independent
variable)) was performed using the CurveExpert Professional software tool. Table 3-4A
represents the seven 3D non-linear and linear regression models that were selected and found to
have a reasonable fit, with R2 values from 0.745 to 0.748 and standard error values from 0.877 to
0.880. The corresponding parameter estimates at the 95% confidence level for each model are
given in Table 3-4B. To select the best fit model, the lowest standard error and highest R2 value
are typically used as tools to compare candidate models; in this study, however, the differences
between the standard errors are very small. Therefore, to select the best model based solely on
the standard error is challenging, so the lowest standard error (0.877) was determined by the
Simplified Quadratic model (Model No. 6) and the highest R2 value (0.748) was determined by
the Full Cubic model (Model No. 1). However, the differences between R2 for many of our
models were insignificant. Therefore, a comparison of the Akaike Information Criterion (AIC)
was also used to select the best model from the pool of candidate models. AIC is a statistical
parameter that strikes a balance between the goodness of fit of a model and the complexity of the
model. The AIC is represented as:
AIC = 2k-2ln(L) (3.17)
49
Where, k is the number of parameters in the statistical model and L is the maximized value of the
likelihood function of the estimated model (Akaike, 1974). The preferred model will have the
lowest AIC value, and this technique includes a penalty prediction that discourages any increase
in the number of parameters that can lead to overfitting and a higher goodness of fit (Fang,
2011). For example, the Full Cubic model (Model No. 1) had the highest R2 value (0.748) but
not the lowest AIC value. This is because Model No. 1 has excess parameter estimates that
helped in overfitting the model. The Power Model C (Model No. 3) yielded the lowest AIC value
(-253.794) of all the other candidate models and had an R2 value of 0.747 and a standard error of
0.876. Hence, the Power Model C was selected as the best model to predict TW with dockage =
f(dockage level, TW without dockage). Another model that produced an accurate response
surface plot and has a simple mathematical form is the Linear model (Model No. 4). Although it
did not have the lowest values of AIC and standard error and highest value of R2, it was favored
over many other tested models because of its simple model structure and suitable response
surface plot (Figure 5A). Model No. 4 (AICi) is 0.03 ((exp (AICmin-AICi/2))=0.03) times as
probable as Model No. 3 (AICmin) to minimize the information loss while predicting the TW with
dockage values. Information loss in statistics in the measure of divergence between two
probability functions (Kullback and Leibler, 1951) and (exp (AICmin-AICi/2) is referred to as the
relative likelihood function of the ith model in a group of selected candidate models (Burnham
and Anderson, 2002; 2004). If the probability is from 0 to 1, the second selected model (i.e.,
Model No.4 in this case) is also an eligible model. Figure 3-11A and 3-11B are the 3D response
surface plot and residual plot for predicting TW with dockage with varying dockage levels and
TW without dockage values based on the Power Model C (Model No.3). Similarly, Figure 3-
12A and 3-12B are the response surface and residual plots for the Linear Model (Model 4).
Residual plots (Figure 3-11B and 3-12B) for both predictive models showed a random
distribution when plotted against the X1 axis (dockage level, %). The residual plot for both the
Power Model C and Linear model indicated good model prediction. Because there were no
previous data on the TW of HRW wheat with and without dockage levels, the data herein could
not be compared with other published results. TW is dependent on moisture content (USDA-
GIPSA, 2009), so it is imperative that the range of moisture content for all samples was known.
For the 3 locations used in this study, the moisture content of all samples ranged from 10.0% to
12.9% (wb), with an average of 11.50% (wb). Thus, the results and models for predicting TW
with dockage = f(dockage level, TW without dockage) given in this paper will be valid over this
range of moisture contents.
3.3.3.1 Cross Validation Analyses
The best fit linear model along with its parameter estimates was obtained using PROC NLIN
SAS (2002) version 9.1 with 95 % significant level (α =0.05 level of significance) with the same
dataset given in Figure 2A. The PROC NLIN model structure is same as model no. 4 in Table 3-
4A. Based on parameter estimates obtained for the coefficients of the given model in PROC
NLIN, we had calculated the standard errors for SEC (standard error or calibration) and SEP
50
(standard error for prediction) for several scenarios as described later. The cross validation
studies are based on the procedure developed by Casada and Armstrong (2009). SEC is defined
as the standard error that is calculated using TW with and without dockage dataset from two
location and leave the one location out. For our case, there were three locations – Enid, OK,
Manhattan, KS, and Goodland, KS. Hence, we iterated the process 3 times with each time we
considered either of two locations and leaving the third one out. SEP is defined as the standard
error that is calculated using TW with and without dockage dataset from the left out location
when fitted into the parameter estimates given by other two locations dataset. For example, SEP
of Enid, OK means that the TW with and without dockage dataset of Enid, OK has been fitted
with the model parameter estimates obtained from Manhattan, KS and Goodland, KS combined
dataset, using PROC NLIN. Hence, the SEC and SEP is given as:
∑ , ,
(3.18)
∑ , ,
(3.19)
Where,
SEC = standard error of calibration for set with combined data of only 2 locations (i.e., any one
location left out)
TWi, observed = Observed TW with Dockage (y- axis) for the ith observation in the set with 2
location data combined.
TWi, observed = Predicted TW with Dockage (y- axis) for the ith observation in the set with 2
location data combined, using parameter estimates obtained from PROC NLIN procedure.
m = no. of. observations/data points for the combined dataset of two locations.
c = number of coefficients used in the model.
n = no. of observations/data points in the location that was left out
Table 3-5is the summary of the cross validation analyses using SEP and SEC as discussed above.
For the second scenario where dataset of Goodland, KS was fitted with model parameter
estimates results from dataset of Enid, OK and Manhattan, KS, the SEP is 1.12 and it is higher
than average SEP (0.807) but SEC (0.544) was lower than average SEC (0.812). This means that
there was variability in the TW with and without dockage dataset for Goodland, KS location and
51
this variability was higher than other two locations (SEP 0.684 for predicting Manhattan, KS
dataset and SEP 0.612 for predicting Enid, OK dataset). One of the reasons for higher than
average SEP for Goodland, KS could be due to the fact that high dockage levels where reported
from Goodland, KS location causing more variation in the TW measurement. Also, in 2011
during wheat harvest there was unusual rainfall in Goodland, KS region which can cause unseen
variations in HRW crop quality parameters. Similar, inferences can also be seen in Figure 3-9B
However, SEP and SEC was within normal ranges for all the scenarios.
The Standard error for the entire dataset (including all the three locations), using linear model
structure like model no. 4 in Table IA, was obtained as 0.90 using PROC NLIN procedure. This
is slightly higher than the standard error obtained from CurveExpert (Standard error = 0.88)
software results given in Table IA. Thus, it can be concluded that the standard error obtained
from CurveExpert Software is not same as usual standard error formula, i.e., equation 2.
Special Consideration
Model 4 (Table 3-4A) is one of the selected favorable model for predicting TW with dockage
when dockage level and TW without dockage is given for HRW wheat samples. However, this
linear is not valid when dockage is 0%. At dockage 0%, TW with and without dockage will be
same. A global model to avoid the discontinuity in the linear function and power model C is
being developed.
52
Table 3-4A Selected 3D regression models for predicting TW with dockage (lb/bu, Y) = ƒ( dockage (%, X1) and TW without
dockage (lbbu, X2))*
Model Coefficient of
Model Determination
No.
Name Family (R2) AIC Std error Score Model Structure
1 Full Cubic Polynomial 0.749 -244.632 0.878 741 Y= a+b*X1+c*X2+d*X1^2+eX2^2+f*X1^3+g*X2^3+h*X1*X2+i*X1^2*X2+j*X1*X2^2
2 Full
Quadratic Polynomial 0.748 -248.683 0.878 740 Y= a+b*X1+c*X2+d*X1^2+e*X2^2+f*X1*X2
3 Power
Model C Power 0.748 -253.794 0.877 740 Y= a*X1^b+c*X2^d
4 Linear
Linear Regression 0.745 -246.696 0.880 738 Y= a+b*X1+c*X2+…..
5 Simplified
Cubic Polynomial 0.748 -247.186 0.878 740 Y= a+b*X1+c*X2+d*X1^2+e*X2^2+f*X1^3+g*X2^3
6 Simplified
Quadratic Polynomial 0.748 -250.704 0.878 740 Y= a+b*X1+c*X2+d*X1^2+e*X2^2
7 Paraboloid Polynomial 0.747 -247.695 0.879 740 Y= a*((b*X1+c)^2+(d*X2+e)^2)
* Score represents a weight given by the CurveExpert Professional software to rate the model performance; AIC is the Akaike Information Criterion; Std error is the statistical standard error; a to j are
the model parameter estimates given in Table IB; X1 (dockage level, %) and X2 (TW without dockage, lb/bu) are the two independent variables in the listed models; Y (TW with dockage (lb/bu)) is the
only dependent variable in the given models.; At dockage =0%, TW without dockage (x axis) = TW with dockage (y axis).
53
Table 3-4B Parameter estimates of the selected response surface 3D regression models given in Table 3-4A*
Parameter Estimates
Model
No. a b c d e f g h i j
1 1.57E+02 5.48E+01 -7.82E+00 2.02E+00 1.65E-01 -1.41E-02 -1.03E-03 -1.95E+00 -3.05E-02 1.70E-02
2 -2.40E+01 -6.24E-01 1.88E+00 1.22E-01 -8.19E-03 -3.59E-04
3 -6.80E-01 4.43E-01 1.47E+00 9.04E-01
4 6.10E+00 -3.65E-01 8.89E-01
5 2.19E+02 -7.82E-01 -1.04E+01 2.51E-01 1.97E-01 -2.77E-02 -1.15E-03
6 -2.39E+01 -6.47E-01 1.88E+00 1.22E-01 -8.17E-03
7 2.08E+00 2.47E-01 -6.51E-01 -4.06E-02 -2.87E+00
* For details on model structure and performance parameters, refer to Table IA and the corresponding model no.
54
A
B
Figure 3-11A Surface 3D plot for Power Model C (Table 1A and B) used for modeling TW with
dockage (lb/bu, Y) = f(dockage level, TW without dockage) for HRW wheat samples. Dockage level
is measured in % (X1 axis) and TW without dockage is measured in lb/bu (X2 axis), as per FGIS
guidelines. Figure 3-11B: Residual plot for Power Model C () predictions for TW with dockage
(lb/bu, Y) = f(dockage level, TW without dockage) for HRW wheat samples.
55
A
Figure 3-12A Surface 3D plot for Linear Model (Table 1A and B) used for modeling TW with
dockage (lb/bu, Y) = f(dockage level, TW without dockage) for HRW wheat samples. Dockage level
is measured in % (X1 axis) and TW without dockage is measured in lb/bu (X2 axis), as per FGIS
guidelines. Figure 5B: Residual plot for the Linear Model (Table 1A and B) predictions for TW with
dockage (lb/bu, Y) = f(dockage level, TW without dockage) for HRW wheat samples.
56
Table 3-5 Cross Validation result showing Standard errors for TW collected from 3 locations
SEC SEP
Sl Scenario
No.
Accepted Location Left out
Location
1 Enid & Goodland Manhattan 0.925 0.684
2 Enid & Goodland 0.544 1.124
Manhattan
3 Goodland & Enid 0.968 0.611
Manhattan
Average 0.812 0.807
57
3.4 Corn Field Data
3.4.1 Field Measurements and Data Collection

Detailed data on bin geometry and packing factors are based on corn storage bins measured
from 2011 to 2013. Elevator cooperators and individual farmers were contacted for their cooperation
in providing data on grain material properties and bin geometry. The in depth details on bin
geometry, i.e., bin diameter, eave height (side wall height of the bin), plenum height (some bins has
false floor in bin), angle of the roof if needed, bin wall material (smooth steel, corrugated steel, or
concrete), and type of bin bottom (hopper or flat bottom) are carefully measured using a LEICO
DISTO D8 laser distance and angle meter (Leica Geosystems AG, St. Gallen, Switzerland). The corn
grain surface profile was measured using laser device. Multiple points on the grain surface heap
were measured to describe the grain profile and measure the angle of repose of the stored grain. To
determine the “calculated” packing factor (%) and the mass of the grain in the bin, the measured
grain volumes was multiplied by the average test weight to yield the calculated mass of grain in the
bin. Average test weight (lb/bu), foreign material, damage (%), and moisture content (%, wb) values
were taken from “ticket” data collected by elevator managers or farmer based on their grain samples.
For more details on using laser method for measuring grain volume and obtaining grain quality bin
geometry parameters (on-site) please refer to our previous work on HRW wheat grain bins in the
previous field measurements of HRW bins section.
WPACKING program estimated values based on given user inputs on bin geometry and
grain material. Based on these values “WPACKING compaction factor (Cw%)” for given corn bins
is given . The federal crop insurance group of USDA RMA (Risk Management Agency) specifies
tabular values of pack factors for seven crops in US based on bushel capacity of the structures.
These tabular values were used to determine “RMA compaction factor (CR%)”. Finally, the three
kind masses were calculated – WPACKING predicted mass, RMA predicted mass, and reported
mass. For further information on comparison procedures for compaction factors (%) please refer to
HRW section.

3.4.2.1 Summary of Field Data Collection for Corn Bins
Table 3-6A summarizes the commercial corn bins that were measured in regions primarily
covering the Midwest, Central, and South Central regions of the U.S. (Arkansas, Kansas, Colorado,
Minnesota, Michigan, Iowa, Kentucky, Texas, and North Dakota). Except for two concrete bins
from Wakeeney, KS (Rows 42 and 43), the corn bins were all made of corrugated steel. This
highlights a recent shift in the grain industry in which large steel bins are commonly used rather than
concrete bins. Except for one steel bin (Row no. 2), all of the bins were center filled. Only one bin
was filled eccentrically, (Row no. 2) with an offset filling location of 0.5 ft from the center of the
58
bin. The average test weight (average of all the test weights from individual truck loads at the
elevator) ranged from 701.52 Kg/m3 (54.50 lb/bu) to 772.31 Kg/m3 (60.00 lb/bu), with a median test
weight of 743.99 Kg/m3 (57.60 lb/bu). Similarly, moisture contents ranged from 13.00 to 17.22%
(wb), with a median of 14.57% (wb). Foreign material (FM) ranged from 0.21 to 1.00%, with a
median of 0.50%. Clearly, these material properties of commercial corn bins were within the normal
ranges for corn found in the U.S. However, a moisture content of 17.22% (wb) may be slightly
higher than found normally throughout U.S. Table 3-6A and B also provides some details about the
corn grain piles inside the bins. Normally if a spreader is used, then the corn pile will be much flatter
with little or no angle of repose measured. For example, the steel bin from Rockwell, IA (Row No.
8) indicated that a spreader was used to flatten the grain pile inside the bin angle of repose is zero
(Table Ib, Row no. 8). Based on test weight data from Table 3-6A&Bthe test weights for bins that
used spreader (median value of 743.99 kg/m3 or 57.80 lb/bu) to flatten the piles was not
significantly different from the rest of the corn bins, in which spreaders were not used (741.42 kg/m3
or 57.60 lb/bu). This is in contrast to the results of Chang et al. (1983), who observed a change in
bulk density with the use of different types of spreader for corn grain. However, the average test
weight reported in Table IA represents the uncompacted bulk density measured from grain samples
collected using a USDA-GIPSA used prior to loading grain in the bin. Hence, it is not expected that
the type of spreader will affect the test weight values.
Table 3-6B is an extension of Table 3-6A, and provides information about the bin geometry,
angles of repose of the corn piles, and compaction factors. Table Ib indicates that bin diameters
ranged from 3.59 (11.76 ft) to 31.87 m (104.54 ft) (median of 9.30 m, 30.51 ft) and eave height
ranged from 4.43 (14.54 ft) to 31.38 m (102.93 ft) (median of 10 m, 32.80 ft). The cylinder height or
grain height is defined as the distance between the bottom of the sidewall (where the hopper/chute
part begins) and the points on the side wall where the grain surface touches the wall. It is calculated
by subtracting the airspace at the sidewall from the entire eave height. However, there are two corn
bins (Row No. 25 and 26) that had hopper bottom bins with a hopper bottom angle of 46.84°, the
effective eave heights for those bins would be eave height plus (1/3) of hopper height, i.e., 16.58 m
(54.38 ft) (Row No. 29) and 13.82 m (45.33 ft) (Row No. 30). Additionally, two concrete corn bins
in the list (Rows 42 and 43) had chute bottom angle of 42.06° instead of hopper bottoms, thus, the
effective eave height will be eave height plus 1/3 of chute height, 33.26 m (109.10 ft) for Row 42
and 43. A chute bottom has a chisel shaped cone in which the hopper is skewed to one side where
the grain discharge orifice is located. Hopper bottom or chute bottom angles of more than 34° are
desirable for the free flow of grain from the bin. Bins with hopper angle less than 34° are desirable
for free flow of grain from the bin.
When grain is loaded in the bin, it typically heaps or forms a cone upward shaped pile. The angle
of repose is the angle relative to the horizontal plane of the piled grain (Pierce and Bodman, 1987).
As shown in Table 3-6b, the angle of repose ranged from 2.0° to 30.23°, with a median value of
20.40°. Angle of repose (AoR) around 2° for corn is extremely low. However, in a commercial bin
setting there could be scenarios of partial unloading and use of spreader that could create an apparent
AoR. Hence, we think this case of extreme low AoR can be called as apparent AoR. The angle of
59
repose could also be subcategorized into two groups: static (for emptying or funneling) and dynamic
(for filling or piling). In Table 3-6b if it is angle of repose is indicated with letter “e” than the bin
was being emptied otherwise all the other angle of repose. The static angle of repose is the angle
caused by friction of the granular material when it is sliding against itself. The dynamic angle of
repose is the angle arising when the material is in motion, such as when grain is discharging from
bins or hoppers (Pierce and Bodman, 1987). Hunter (1984) used the angle of repose as the key
parameter to calculate the volume of wheat in storage facility. The angle of repose for corn is
reported to be approximately 16° for filling and piling and approximately 23.1 to 34.7° for emptying
or funneling (Stahl, 1950; Mohsenin, 1986; Molenda and Horabik, 2005). The median angle of
repose (20.18°) for the current study of corn bins was higher than suggested (AoR ~ 16°) for filling
corn bin according to Stahl (1950). However, the range for angle of response for the given corn bins
was within the limits suggested in the literature for the emptying angle of repose. Thompson et al.
(1990) showed that increasing the moisture content for corn samples from 12% to 16% (wb) does
increase the angle of repose of the stored grain from 22° to 28°. In this study, only a limited number
of moisture levels and bins were studied, so the results are not conclusive. For this paper, based on
angles of repose in commercial bins, we did not see any correlation between moisture content and
the angle of repose. Holligan et al. (1982) showed that the angle of repose can increase in contained
wheat (wheat stored in an enclosed storage facility) more than in unconfined piles. Such details are
beyond the scope of this current paper.
In McNeill et al. (2004), shelled yellow dent corn samples were conditioned at three moisture
levels (12%, 15%, and 18%, wb), and a calculated amount of fines or broken corn was added (with
particle size 1.0, 1.4, 2.0, or 2.8 mm) at a concentration of 2.5%, 5.0%, or 7.5% concentration levels.
The initial uncompacted bulk density for clean corn in McNeill et al. (2004) was found to be in the
range of 702.4 to 733.0 kg/m3, slightly lower than the test weight (or bulk density) found in this
study (701.52 to 772.31 kg/m3, with median of 743.99 kg/m3 given in Table IA). Test weight is
measured in the U.S. using FGIS guidelines and a Winchester bushel cup, and it is often confused
for bulk density (Bhadra et al., 2014; Wheat Marketing Center, 2008). Other corn quality
parameters like moisture content, foreign material (FM), and damage were determined Federal Grain
Inspection Service (FGIS) guidelines for yellow corn and is required by every elevators and grading
laboratories in U.S to be followed. Table IB also listed the pack factors for each corn bins calculated
using 3 methods- 1) the WPACKING software, 2) the USDA-RMA method and 3) the FSA-W
method. The WPACKING software predicts pack factors based on the bin diameter, grain height,
bin wall type, angle of repose, moisture, test weights, and bin bottom geometry while USDA-RMA
method predicts pack factors based on varying bin diameter and test weights, and FSA-W method
predicts pack factors based on bin diameter, bin shape, and test weights. The detailed relationship
between these 3 methods and derivation of pack factors in these 3 methods has been discussed in our
previous paper Boac et al. (2014). These pack factors were used later on to calculate bin capacities.
60
Table 3-6A Location details, crop quality characteristics, and bin type for corn bins measured for packing
factor analyses.*
Use of
Row. No. Location County Type TW (kg/m3) TW ( lb/bu) MC (%, wb) Avg FM (%) Damage (%) Spreader?
1 Manhattan, KS Riley Steel 707.95 55.00 15.5 No
2 Manhattan, KSϮ Riley Steel 707.95 55.00 15.5 No
3 Manhattan, KS Riley Steel 707.95 55.00 15.5 No
4 Holyoke, CO Phillips Steel 744.89 57.87 16.24 0.24 No
5 Goodland, KS Sherman Steel 746.57 58.00 17.22 ‐ No
6 Goodland, KS Sherman Steel 746.57 58.00 17.22 ‐ No
7 Hayfield, MN Dodge Steel 752.88 58.49 15.25 0.99 No
8 Rockwell, IA Cerro Gordo Steel 724.69 56.30 15.3 1 0.5 Yes
9 Rockwell, IA Cerro Gordo Steel 701.52 54.50 14.2 0.9 0.5 No
10 Rippey, IA Greene Steel 740.13 57.50 13.7 ‐ ‐ No
12 Rippey, IA Greene Steel 772.31 60.00 14 ‐ ‐ No
14 Rippey, IA Greene Steel 753 58.50 13.4 ‐ ‐ No
15 Rippey, IA Greene Steel 753 58.50 13.4 ‐ ‐ No
16 Britt, IA Hancock Steel 711.56 55.28 14.53 ‐ ‐ No
17 Hillje, TX Wharton Steel 741.42 57.60 14.6 0.5 0.23 No
23 Danevang, TX Wharton Steel 741.42 57.60 14.6 0.5 0.35 No
30 Clay Center, KS Clay Steel 743.99 57.80 14.5 0.23 ‐ Yes
31 Clay Center, KS Clay Steel 743.99 57.80 15 0.25 ‐ Yes
32 Clay Center, KS Clay Steel 743.99 57.80 14.4 0.3 ‐ Yes
33 Severance, KS Serverance Steel 739.62 57.46 15 0.3 0.11 No
34 Severance, KS Serverance Steel 766.91 59.58 15 0.52 0.11 No
35 Severance, KS Serverance Steel 733.95 57.02 14.5 0.63 0.11 No
61
36 Severance, KS Serverance Steel 733.95 57.02 14.5 0.5 0.11 No
37 Central City, IA Linn Steel 759.44 59.00 14 0.45 0.27 No
38 Waterloo, IA Black Hawk Steel 761.5 59.16 14.1 0.21 0.31 Yes
39 Litchville, ND Barnes Steel 765.88 59.50 14.5 0.21 No
40 Litchville, ND Barnes Steel 765.88 59.50 14.5 0.21 No
41 Hazen, ND Mercer Steel 733.18 56.96 13.03 0.3 No
42 Wakeeney, ks Trego concrete 745.15 57.89 13.5 0.5 No
43 Wakeeney, ks Trego concrete 746.57 58.00 13 0.52 No
44 Pocahontas, AR Randolph Steel 772.31 60.00 15.3 No
47 Borwnn City, MI Sanilac Steel 749.14 58.20 14.9 No
48 Borwnn City, MI Sanilac Steel 749.14 58.20 14.9 No
49 Versailles, KY Woodford Steel 718.51 55.82 14.5 No
50 Russellville, KY Logan Steel 753 58.50 14.5 No
* All the steel bins are corrugated steel wall; MC is moisture content (%, wb); FM is foreign material (%); TW is test weight usually measured by FGIS
guidelines in lb/bu; blank FM and Damage cells indicate that those data was not received from grain elevators or farmers.
Ϯ this bin has off‐centered filling and the cone peak was offset by 0.5 ft or 0.15 m towards north.
62
Table 3-6B Summary of bin geometry and compaction factors for commercial corn bins*
Cylinder Cylinder Angle of Hopper WPACKING RMA FSA‐W

Row. Diameter Diameter Eave height Eave height height height Repose bottom Angle Pack Factor Pack Factor Pack Factor
No. (m) (ft) (m) (ft) (m) (ft) (AoR) (°) (‐) (‐) (‐)
1 3.59 11.78 4.8 15.75 4.55 14.93 20.18 0 1.0124 1.003 1.015
2 3.59 11.78 4.8 15.75 4.45 14.60 27.29 0 1.0124 1.003 1.015
3 5.45 17.88 4.43 14.53 2.62 8.60 27.17 0 1.0086 1.003 1.029
4 27.36 89.76 17.74 58.20 16.96 55.64 19.23 0 1.0977 1.134 1.1
5 23.85 78.25 18.32 60.10 18.03 59.15 16.17 0 1.1052 1.134 1.102
6 14.38 47.18 23.87 78.31 23.7 77.76 26.6 0 1.1082 1.113 1.102
7 22.8 74.80 14.72 48.29 13.67 44.85 22.45 0 1.0994 1.143 1.11
8 9.95 32.64 6.64 21.78 4.83 15.85 0 0 1.0384 1.076 1.073
9 8.7 28.54 5.94 19.49 5.61 18.41 25.02 0 1.0075 1.024 1.038
10 14.63 48.00 6.38 20.93 6.09 19.98 30.23 0 1.0642 1.105 1.094
11 9.3 30.51 6.58 21.59 6.36 20.87 25.63 0 1.0907 1.095 1.113
12 9.75 31.99 7.39 24.25 7.16 23.49 25.63 0 1.1101 1.131 1.131
13 8.11 26.61 4.63 15.19 3.96 12.99 18 0 1.0481 1.064 1.076
14 8.23 27.00 4.68 15.35 4.36 14.30 26.53 0 1.076 1.087 1.102
15 6.38 20.93 4.57 14.99 3.96 12.99 18 0 1.0738 1.068 1.089
16 31.87 104.56 27.69 90.85 27.34 89.70 22.89 0 1.051 1.094 1.058
17 9.14 29.99 12.5 41.01 12.02 39.44 22.95e 45 1.0785 1.071 1.09
18 14.63 48.00 14.63 48.00 14.63 48.00 22.96 0 1.0795 1.105 1.096
19 14.63 48.00 14.63 48.00 14.37 47.15 20.45 0 1.0791 1.105 1.096
20 14.63 48.00 14.63 48.00 14.39 47.21 18.29 0 1.079 1.105 1.096
21 14.63 48.00 14.63 48.00 14.31 46.95 17.06 0 1.0789 1.105 1.096
22 23.77 77.99 25.91 85.01 25.91 85.01 18.91 0 1.0911 1.126 1.096
23 10.97 35.99 14.63 48.00 14.22 46.65 21.25 0 1.0775 1.092 1.096
24 9.14 29.99 12.19 39.99 11.8 38.71 25.12 0 1.074 1.071 1.09
25 9.14 29.99 12.19 39.99 11.89 39.01 21.85 46.84 1.0785 1.071 1.09
26 10.97 35.99 14.63 48.00 8.05 26.41 24.16 46.84 1.0761 1.092 1.096
27 10.97 35.99 14.63 48.00 14.63 48.00 19.83 0 1.0778 1.092 1.096
28 10.97 35.99 14.63 48.00 14.63 48.00 17.54 0 1.0777 1.092 1.096
29 21.95 72.01 18.29 60.01 18.17 59.61 19.56 0 1.0822 1.126 1.096
30 7.32 24.02 6.54 21.46 6.54 21.46 0 0 1.0678 1.061 1.086
31 7.32 24.02 6.43 21.10 6.17 20.24 0 0 1.0679 1.061 1.086
32 7.32 24.02 6.42 21.06 6.21 20.37 0 0 1.067 1.061 1.086
33 10.87 35.66 7.66 25.13 7.28 23.88 18.53 0 1.0662 1.092 1.093
34 7.47 24.51 5.18 16.99 5.18 16.99 17.84 0 1.0988 1.102 1.114
35 7.35 24.11 5.19 17.03 4.95 16.24 20.15 0 1.0513 1.045 1.074
63
36 9.14 29.99 7.62 25.00 6.55 21.49 15.7 0 1.0551 1.064 1.081
37 9.14 29.99 5.18 16.99 5.03 16.50 22.3 0 1.0874 1.095 1.113
38 14.63 48.00 15.61 51.21 11.13 36.52 0 0 1.1008 1.13 1.12
39 10.36 33.99 9.92 32.55 9.57 31.40 17.46 0 1.1057 1.123 1.124
40 7.32 24.02 5.5 18.04 5.35 17.55 20.87 0 1.0971 1.083 1.113
41 14.63 48.00 10 32.81 9.07 29.76 18.82 0 1.0565 1.097 1.085
42 6.25 20.51 31.38 102.95 29.04 95.28 18.09 42.06Δ 1.0828 1.061 1.08
43 6.25 20.51 31.38 102.95 28.53 93.60 17.48 42.06Δ 1.0827 1.061 1.081
44 9.14 29.99 7.36 24.15 5.51 18.08 0 0 1.1072 1.11 1.128
45 9.14 29.99 7.36 24.15 6.47 21.23 25.47 0 1.0548 1.064 1.081
46 9.14 29.99 7.36 24.15 5.86 19.23 20.1 0 1.0722 1.079 1.097
47 14.63 48.00 21.4 70.21 21.24 69.69 23.56 0 1.1 1.11 1.11
48 21.95 72.01 22.25 73.00 19.56 64.17 20.4 0 1.1 1.13 1.11
49 8.23 27.00 7.44 24.41 1.68 5.51 28.4 0 1.07 1.11 1.11
50 18.23 59.81 17.95 58.89 16.46 54.00 15.7 0 1.02 1.05 1.06
51 18.2 59.71 17.95 58.89 15.86 52.03 0 0 1.1 1.14 1.11
52 27.37 89.80 27.74 91.01 6.65 21.82 1.9Ϯ 0 1.1 1.14 1.11
*
zero angle of repose for Row 30, 31, 32, and 38 indicates that spreader were used and grain was leveled and it correlate with Table IA; e represents
angle of repose while emptying and the cone was inverse and rest of the bins had filling angle of repose; WPACKING is WPACKING software
developed by Thompson et al. (1987) and in ASABE Standards EP413.2 (2010); RMA is USDA-Risk Management Agency and pack factors are
reported from USDA-RMA Loss Adjustment Manual (LAM) standard handbook (2012a&b); FSA-W is Farm Service Agency and pack factor are
reported from USDA-Warehouse examiner’s handbook grain Pack Data WS-3 (no date).
Ϯ
Apparent angle of repose (AoR)
Δ
This bin have hopper bottom shape with side grain draw design, sometimes also known as chute bottom bin.
64
3.4.2.2 Predicted Mass from WPACKING Model and Comparisons of Bin Capacities
Figure 3-13 (a and b) represents the correlation between reported mass (t) and the mass predicted
by the WPACKING model (t) for all of the commercial corn bins given in the previous two tables.
The reported mass is the total mass grain in the bin determined by the elevator and is also known as
the actual ticket sale mass data. It was obtained along with other crop quality properties for each bin
as given in the previous two corn tables. Using the bin geometry data (bin diameter, grain height,
eave height, hopper bottom angle, shape of the bin, and type of bin) and the crop quality data
(average moisture, average test weight, and angle of repose) the mass and the predicted number of
bushels in the bin were calculated using the WPACKING model software. Figure 3-13 confirms the
high correlation (R2 = 0.9997) between actual reported mass (t) and WPACKING model-predicted
mass (t), as desired. For corn bins dataset, the WPACKING predicted mass (t) after accounting for
grain packing is very close to the actual storage data, with slope of 0.9998 for entire dataset,
indicating an almost identical match for the WPACKING model-predicted mass and the reported
mass (t). Identical match between predicted and reported mass in Figure 3-13 is represented by the
45 degree line where slope is 1. For RMA method the slope was 1.0206 and for FSA-W method is
was 0.9977. Overall, all the three methods almost predicted identical mass when compared to
reported values. However, based on the slopes, it is clear that WPACKING model was slightly better
than the other two model because the slope was 0.9998, closer to 1 than RMA –slope of 1.0206 and
FSA-W –slope of 0.9977. This is a very positive sign for further development and calibration of the
WPACKING model, and it indicates that the prediction of masses in commercial corn bins appears
to be reliable and accurate.
To proceed further with analysis of the predictive capabilities of the WPACKING model, we took
the % difference between WPACKING predicted mass and reported mass (y-axis), and we plotted
that difference against each bin capacity /reported mass (x-axis), as illustrated in Figure 2 ( a and b).
The % differences for corn bins ranged from -4.54% to +4.53%, with a median value of -0.28% and
absolute average of 0.86%. Thus, % differences range, it was within +/- 5.0%, conforming to the
idea that the current WPACKING model software by Thompson et al. (1990 and 1987) did a good
job of predicting the mass for commercial corn bins (mostly made up of corrugated steel, only 2
concrete bins). In Figure 3-14 (a and b), it can be seen that for most all corn bins the WPACKING
software “under-predicted” (i.e., negative % differences), indicating that the current WPACKING
model predicts masses that are slightly less than the reported mass values. The two concrete bins in
the bin list (Table 3-6A) showed “over-prediction” (i.e., positive % differences) for the WPACKING
model-predicted mass when compared to the reported masses. Thus, it is evident that the
construction material (concrete or steel) does have an effect on over- or under-prediction. However,
more concrete bin data needed in order to determine if this is indeed a trend with the WPACKING
software or just a special case. These data and the % differences will help us to improve, calibrate,
and validate the WPACKING model for corn crops with greater accuracy and reliability.
65
The USDA Risk Management Agency (USDA-RMA) is one of the governmental agencies
that has a pack factor procedure and has data concerning the predicted masses of commercial bins.
However, uncertainty in pack factor data, outdated bin structures, and the lack of reliability of pack
factor data has motivated the USDA-RMA to fund this current research study and to further develop
WPACKING software and use it with current commercial field data (Bhadra, 2013). Thus, it was
natural to compare the RMA-predicted mass and the reported mass for these corn bins. The %
differences ranged from -2.92% to +4.97%, with a median value of +0.31% and absolute average of
1.46%, shown in Figure 3-15. This range is very similar to the % differences of WPACKING
indicated in Figure 3-14. However, the % differences in Figure 3a (RMA predicted mass vs. reported
mass) is has more variability and spread than Figure 2a (WPACKING predicted mass vs. reported
mass), where majority of the data points (blue diamond symbol) were close to 0% difference line
with more under-prediction of WPACKING model for bin capacities less than 1,500 t and all the
bins with more than 1,500 t showed under-prediction. Almost reverse trends were observed for
Figure 3-15b when compared to Figure 3-14b. That is, for bin capacities >1,500 t (Figure 3-15b)
RMA method over-predicted in estimating grain mass in the corn and with increase in bin sizes
compare to Figure 2b where WPACKING model under predicted.
Figure 3-16a and b represents the % differences between predicted vs. reported mass using
three different methods- WPACKING model, RMA, and FSA-W. In the current commercial corn
bin data set, the FSA-W method (using FSA handbook pack factors) is mostly over-predicting
estimated mass values for corn bins <1,500 t but under-predicting for majority of the corn bins >
1,500 t. The % differences for FSA-W vs. reported mass are shown using green color triangle
symbol ranged from -3.33 % to +5.67% with a median value of + 1.45% and absolute average of
1.60%. The trend is different than % differences obtained by using RMA method (red color square
symbol) where there were almost majority of the corn bins evenly distributed on either side of zero
percent difference line for all corn bins listed in Table 3-6A. On comparing the 3 methods,
WPACKING model differences (black diamond symbol) were most close to zero percent difference
line, especially for bins <1,500 t, indicating WPACKING model did the better prediction of mass for
corn bins compare to RMA and FSA-W method of mass prediction, if the grain quality and bin
parameters are known.
Figure 3-17 the plot of % differences between predicted mass and reported mass for
WPACKING model with respect bin diameter. Figure 3-17 shows that in WPACKING model under
predicted when the bin diameter size was more than 15 m (about 50 ft). However, this graph is not
conclusive because WPACKING model prediction depends on bin diameter and grain height and
compaction changes with grain height. Hence, we have classified the % differences based on
effective grain height to bin diameter ratio (H/D, -) and plotted with respect to bin diameter in x axis
as given in Figure 3-18a. Effective grain height is defined as total grain height including the cone
(1/3 of the cone height calculated from angle of repose and diameter of the bin), in the cylinder part
where grain is parallel to the bin wall, and hopper (1/3 of the hopper height calculated from hopper
bottom angle and bin diameter). Subsequently, similar plots for RMA (Figure 3-18b) and FSA-W
(Figure 3-18c) methods % differences have been illustrated. Figure 6a shows that the 2 concrete bins
66
over predicted for % differences between WPACKING model predicted mass and reported mass.
For H/D ratios 0.69 to 1.0 the WPACKING model under predicted when bin diameters were more
than 15 m (~ 50 ft). Below 15 m diameter, the % differences were randomly distributed irrespective
for H/D ratio from 0 to 2.0. Figure 6b is showed random distribution of % differences between RMA
predicted and reported mass for all bin between H/D ratio 0 to 2.0. The only 2 concrete bins with
H/D ratio >2.0 showed over prediction of RMA method compared to actual reported values. Figure
3-18c is the % differences between FSA-W method predicted mass and reported mass for each corn
bin in Table IA. Like Figure 6a and b, FSA-W method also over predicted for 2 concrete bins
compared to reported mass. For bin diameter above 15 m (~ 50ft) the FSA-W method under
predicted for H/D ratios 0 to 2.0 and over predicted for all the H/D ratios is diameter is less than 15
m. Figure 3-18a, b, and c is more depth analysis of 3-16 a and b where % differences are only
plotted again reported mass but effect of bin diameter and H/D ratio is not shown.
More details on calculation procedures on these three methods can be found in Boac et al.
(2014) and origin and background of these methods are found in Bhadra et al. (2013).
67
1,400
45 degree line
1,200
1,000
Predicted mass (t)
WPACKING model
800
RMA method
600
FSA method
400
200
0
0 200 400 600 800 1,000 1,200 1,400
Reported mass (t)
Corn bins with capacity less than 1,500 t or 56, 000 bu
19,500
17,500 45 degree line
15,500
13,500
Predicted mass (t)
WPACKING model
11,500
RMA method
9,500
FSA method
7,500
5,500
3,500
1,500
1,500 3,500 5,500 7,500 9,500 11,500 13,500 15,500 17,500 19,500
Reported mass (t)
Corn bins with capacity more than 1,500 t or 56,000 bu

Figure 3-13 Correlation between predicted (WPACKING model, RMA, and FSA-W methods) and reported
mass for all corn bins.
68
Reported mass (Standard bushel)
0 8,000 16,000 24,000 32,000 40,000 48,000 56,000
6.0%
4.0%
Concrete bins
% diff between WPACKING predicted and Reported mass
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
0 200 400 600 800 1,000 1,200 1,400
Reported mass (t)
Corn bins with capacity less than 1,500 t or 56,000 bu

Reported mass (Standard bushel )
56,000 156,000 256,000 356,000 456,000 556,000 656,000 756,000
6.0%
4.0%
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
1,500 3,500 5,500 7,500 9,500 11,500 13,500 15,500 17,500 19,500
Reported mass (t)
Corn bins with capacity greater than 1,500 t or 56,000 bu

Figure 3-14 Percent difference between WPACKING predicted and Reported mass for all corn bins.
69
Reported mass (standard bushel)
0 8,000 16,000 24,000 32,000 40,000 48,000 56,000
6.0%
Concrete bins
4.0%
% diff between RMA predicted and Reported mass
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
0 200 400 600 800 1,000 1,200 1,400
Reported mass (t)

56,000 156,000 256,000 356,000 456,000 556,000 656,000 756,000
6.0%
4.0%
2.0%
0.0%
‐2.0%
‐4.0%
1,500 3,500 5,500 7,500 9,500 11,500 13,500 15,500 17,500 19,500
Reported mass (t)

Figure 3-15 Percent difference between RMA predicted and Reported mass for all corn bins.
70
0 8,000 16,000 24,000 32,000 40,000 48,000 56,000
6.0%
4.0%
% diff between Predicted and Reported mass
2.0%
0.0%
‐2.0% FSA
WPACK
RMA
‐4.0%
‐6.0%
0 200 400 600 800 1,000 1,200 1,400
Reported mass (t)

56,000 156,000 256,000 356,000 456,000 556,000 656,000 756,000
6.0%
4.0%
2.0%
0.0%
‐2.0%
WPACK
RMA
FSA
‐4.0%
‐6.0%
1,500 3,500 5,500 7,500 9,500 11,500 13,500 15,500 17,500 19,500
Reported mass (t)

Figure 3-16 Percent difference between Predicted and Reported mass for all corn bins using 3 methods –
WPACKING model, RMA, and FSA-W.
71
Bin diameter (m)
0 5 10 15 20 25 30 35 40 45
6.0%
% diff between WPACKING predicted and Reported mass 4.0%
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
0 20 40 60 80 100 120 140 160
Bin diameter (ft)
Figure 3-17 Percent difference between WPACKING model predicted and reported mass, with respect to bin
diameter sizes for corn bins.
Bin diamater (ft)
0 20 40 60 80 100 120
6.0%
WPACKING model
Concrete Bins
H/D ratio 1 to 2.0
4.0%
H/D 0 to 0.68
H/D ratio 0.69 to 1.0
2.0% H/D ratio > 2.0
0.0%
‐2.0%
‐4.0%
‐6.0%
0 5 10 15 20 25 30 35
Bin diameter (m)
Figure 3-18a Percent difference between WPACKING model predicted and reported mass, with respect to
bin diameter sizes and classified based on grain height/diameter ratio (H/D).
72
Bin diameter (ft)
0 20 40 60 80 100 120
6.0%
RMA method
Concrete Bins
4.0%
2.0%
0.0%
H/D ratio 1 to 2.0

‐2.0%
H/D 0 to 0.68
H/D ratio > 2.0

‐4.0%
‐6.0%
0 5 10 15 20 25 30 35
Bin diameter (m)
Figure 3-18b Percent difference between RMA predicted and reported mass, with respect to bin diameter
sizes and classified based on grain height/diameter ratio (H/D).
Bin diameter (ft)
0 20 40 60 80 100 120
8.0%
Concrete Bins
FSA method
6.0%
H/D ratio 1 to 2.0
H/D 0 to 0.68
% diff between FSA predicted and Reported mass
4.0%
H/D ratio > 2.0
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
0 5 10 15 20 25 30 35
Bin diameter (m)
Figure 3-18c Percent difference between FSA-W predicted and reported mass, with respect to bin diameter
sizes and classified based on grain height/diameter ratio (H/D).
73
Predicted Mass from WPACKING Model and Comparisons of Bin Capacities for Grouped Reported
Scales Bins
Table 3-7 is representation of all corn bins measured from eastern side of Mississippi river
locations (states like KY, MI, and IN). However, these corn bins lot has classified in a separate lot
because it was not possible to get the reported mass values from grain managers/farmers for each bin
rather reported scale mass values were given to us based on each location. For example, we had total
reported mass value from Brown City, MI but not scale values for the three bins measured from
Brown City, MI individually. Hence, we estimated the individual bin reported mass value based on
% difference between grouped reported scale value and WPACKING model predicted mass value
for a particular location. Obtaining individual reported mass values from commercial bins is usually
a challenge in an industry that is constantly adapting to the high volume market need. Blending and
moving of corn grain between two or multiple bins is a common scenario and unfortunately we have
some corn bins that could not be reported mass based on individual bins (like Table 3-6A) but based
on facility locations. Within each location the bin diameters were varying as indicated in Table II,
hence, estimating reported mass based on % differences on grouped scale and WPACKING model
values may be debatable. However, we have indicated in Figure 3-19 (a and b) that there were no
trends in % differences for bin with similar diameters, the differences varies due to location and
grain height. Hence, perhaps the estimated reported mass for each individual bins may be a fair and
logical approximate.
Figure 3-20 (a and b) shows that for 3 prediction methods - WPACKING model, RMA, and
FSA-W, most of the corn bins over predicted if bin capacities were more than 1,500 t and was
randomly distributed when it was less than 1,500 t. % differences between WPACKING model
predicted and reported mass with respected to bin diameter was plotted (Figure 3-21) and it showed
over prediction for bin diameter more than 15 m. This is just reverse when compare to Figure 3-17.
However, it is logical and desired to have similar trends in Figure 3-17 and 3-19 because essentially
both the plots are for corn commercial bins in U.S. Actually, Figure 9 is not quite conclusive because
it has a mixture of steel and concrete bins (27 steel and 13 concrete bins) and it does not show the
effect of grain height or H/D ratio in the plot. Hence, we have showed % for WPACKING model
(Figure 3-22a), RMA (Figure 3-22b), and FSA-W (Figure 3-22c) method predicted mass and
reported mass (y-axis) vs. bin diameter (x- axis), classified based on H/D ratios. Figure 3-22a is
compared to Figure 3-18a and it was concluded concrete bins mostly under predicted (i.e., negative
% differences) and because there were only 2 concrete bins in Figure 3-18a that over predicted
around 2.0%, it will not be considered enough bins to validate over prediction. Thus, concrete bins
under predicted using WPACKING model for all H/D ratio from 0 to 2.0 and above. For bin
diameters above 15 m (~ 50ft), based on Figure 3-18a and 3-22a we can conclude that for H/D ratio
0.69 to 1.0 WPACKING model under predicted (i.e., negative % differences) but for H/D ratio 0 to
0.68 WPACKING model over predicted when compared to their reported masses. These are valid
only for steel bins because concrete bins were less than 15 m is diameter.
74
Also, Figure 3-22a may appear to have less data points compared to Figure 3-22b and c but
actually, each data point in Figure 3-22 a represents multiple bins. Since, we have estimated the
reported mass for those individual bins based on % differences between total reported mass of that
facility and total WPACKING model predicted mass. For example, in Figure 3-22a for diameter
above 15 m there seems to be 5 data points for H/D ratio 0 to 1.0 but in fact there are 11 data points
representing 11 individual bins. For bin diameter below 15 m, irrespective of H/D ratios, for all steel
bins (combining Figure 3-18a and 3-22a) the % differences were randomly distributed and in most
cases it was very close zero percent difference line (Figure 3-18a). These results are very promising
and it indicates that for corn bins WPACKING model is doing fairly well in predicting grain mass
for steel bin diameter less than 15 m and can be adjusted accordingly to fit better the current field
data set for diameter above 15 m. Similarly, corn concrete bins trends in Figure 6a and 10a can
provide vital information to fine tune and adjust the WPACKING model to be more robust based on
field trials.
Likewise, based on Figure 3-18b and 3-22b we can conclude that % differences between
RMA predicted and reported mass were negative for concrete bins (under prediction) and 2 cases of
over prediction (about + 2.5%) showed in Figure 3-18b is not conclusive than Figure 3-22 b where
13 concrete bins under predicted. For steel bins irrespective of H/D ratio, % differences were fairly
distributed for bin diameter less than 15 m. H/D ratio 0 to 0.68 showed similar pattern in %
differences in Figure 3-18b and 3-22b, i.e., random distribution at lower bin diameter and over
prediction for diameter above 13 m. Also, based on Figure 3-18c and 3-22c we can conclude that %
differences between FSA-W predicted and reported mass were negative (i.e., under prediction) for
concrete bins, 2 over prediction cases (about +5.0%) in Figure 6c is not enough to counter the under
prediction results of 13 concrete bins in Figure 3-22c. For bin diameter above 15 m and H/D ratio 0
to 0.68, FSA-W method over predicted mass values. However, for other H/D ratios and bin diameter
less than 15 m, the % differences were evenly distributed on either side of zero percent line. Also,
the range of % difference in concrete bins for FSA-W method was higher (4.2% absolute average)
than RMA (3.5% absolute average) and WPACKING model (2.4% absolute average). Hence, once
again WPACKING model did slightly better in predicting mass than RMA and FSA-W method for
concrete corn bins.
75
Table 3-7 Summary of bin geometry and compaction factors for commercial corn bins for grouped reported scale values based on location.*
Equivalent Equivalent RMA FSA‐W

Diameter Diameter Eave Height Eave Height Grain Height Grain Height TW TW MC WPACKING Pack Pack
Location County Type (m) (ft) (m) (ft) (m) (ft) (kg/m3) (lb/bu) (%, wb) Pack Factor Factor Factor
Brown City, MI Sanilac Steel 14.54 46.65 21.4 68.66 7.50 24.07 749.14 58.20 14.9 1.08 1.11 1.11
Brown City, MI Sanilac Steel 21.82 70.01 22.25 71.39 21.69 69.59 749.14 58.20 14.9 1.1 1.13 1.11
Steel‐
Brown City, MI Woodford hopper 7.28 23.36 14.49 46.49 13.75 44.13 749.14 58.20 14.9 1.09 1.06 1.09
Versailles, KY Woodford Steel 10.91 35.00 9.39 30.13 8.08 25.93 725.97 56.40 13.1 1.04 1.08 1.08
Versailles, KY Woodford Steel 10.91 35.00 9.39 30.13 7.51 24.11 725.97 56.40 14 1.04 1.08 1.08
Midland, MI Gladwin Steel 14.54 46.65 14.16 45.43 7.18 23.05 740.52 57.53 14.7 1.07 1.11 1.1
Midland, MI Gladwin Steel 14.54 46.65 14.16 45.43 10.04 32.21 740.52 57.53 14.7 1.07 1.11 1.1
Steel‐
Midland, MI Gladwin hopper 4.54 14.57 7.12 22.84 7.94 25.48 740.52 57.53 14.7 1.06 1.04 1.07
Caskey, KY Christian Steel 31.82 102.09 27.74 89.00 20.13 64.57 772.31 60.00 13.7 1.13 1.17 1.13
Dunkirk, IN Jay concrete 3.35 10.75 19.85 63.69 20.30 65.13 753.00 58.50 14.3 1.08 1.06 1.07
Dunkirk, IN Jay concrete 3.66 11.74 17.74 56.92 18.15 58.22 714.39 55.50 13 1.02 1.01 1.03
76
Dunkirk, IN Jay concrete 3.35 10.75 7.32 23.49 7.48 24.00 723.40 56.20 16 1.04 1.02 1.03
Biven, KY Larue Steel 10.97 35.20 10.91 35.00 5.65 18.14 761.11 59.13 14 1.09 1.12 1.12
Russellville, KY Logan Steel 27.37 87.81 27.74 89.00 26.32 84.43 753.00 58.50 14 1.11 1.14 1.11
* All the steel bins are corrugated and they are center filled; Steel hopper bins have hopper bottom angle of 45°; all the bins had coned up grain profile but individual AoR is not reported but equivalent
grain height = grain height of the cylinder part + (1/3) hopper bottom height + (1/3) of cone height from AoR is being reported in Table II; MC is moisture content (%, wb); TW is test weight
(Kg/m3) but usually reported from grain elevators as lb/bu; WPACKING is WPACKING software developed by Thompson et al. (1987) and in ASABE Standards EP413.2 (2010); RMA is USDA-
Risk Management Agency and pack factors are reported from USDA-RMA Loss Adjustment Manual (LAM) standard handbook (2012a&b); FSA-W is Farm Service Agency and pack factor are
reported from USDA-Warehouse examiner’s handbook grain Pack Data WS-3 (no date).
77
6.00%
4.00%
% Diff between WPACKING predicted and Reported mass
2.00%
Dia 47.70 ft
"Dia 104.40 ft"

0.00%
"Dia 35.8 ft"
"Dia 26.9 ft"
‐2.00% "Dia 11 ft"
"Dia 28 ft"
"Dia 17 ft"
‐4.00%
"Misc Dia"
‐6.00%
‐8.00%
0 500 1,000 1,500 2,000 2,500
Reported mass (t)
(a) Corn bins with capacity less than 2,500 t
6.00%
4.00%
% Diff between WPACKING predicted and Reported mass
2.00%
Dia 47.70 ft
"Dia 104.40 ft"

0.00%
"Dia 35.8 ft"
"Dia 26.9 ft"
‐2.00% "Dia 11 ft"
"Dia 28 ft"
"Dia 17 ft"
‐4.00%
"Misc Dia"
‐6.00%
‐8.00%
2,500 7,500 12,500 17,500 22,500
Reported bu
b) Corn bins with capacity greater than 2,500 t
Figure 3-19 Percent difference between WPACKING predicted and Reported mass for all corn
bins
78
0 8,000 16,000 24,000 32,000 40,000 48,000 56,000
12.0%
10.0%
WPACKING
8.0%
RMA
% dif between Predicted and Reported mass
6.0%
FSA
4.0%
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
‐8.0%
0 200 400 600 800 1,000 1,200 1,400
Reported mass (t)
a) Corn bin capacity less than 1,500 t or 56,000 bu

56,000 156,000 256,000 356,000 456,000 556,000 656,000 756,000
12.0%
10.0%
8.0%
WPACKING
6.0%
% dif between Predicted and Reported mass
RMA
4.0%
FSA
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
‐8.0%
1,500 3,500 5,500 7,500 9,500 11,500 13,500 15,500 17,500 19,500
Reported mass (t)
b) Corn bin capacity more than 1,500 t or 56,000 bu
Figure 3-20 Percent difference between Predicted and Reported mass for all grouped corn bins
using 3 methods – WPACKING model, RMA, and FSA-W.
79
Bin diamter (m)
0 5 10 15 20 25 30 35
8.0%
6.0%
% diff between WPACKING Predicted and Reported mass

4.0%
2.0%
0.0%
‐2.0%
‐4.0%
‐6.0%
‐8.0%
0 20 40 60 80 100 120
Bin diameter (ft)
Figure 3-21 Percent difference between WPACKING model predicted and reported mass, with
respect to bin diameter sizes for grouped corn b ins in Table II.
Bin diameter (ft)
0 20 40 60 80 100 120
8.0%
6.0%
4.0%
2.0%
0.0%
SteelH/D ratio 0 to 0.68
Steel H/D ratio 0.69 to 1.0

‐2.0%
Steel H/D ratio 1 to 2.0
Steel H/D ratio >2.0

‐4.0%
concrete H/D ratio 0 to 0.68
Concrete H/D ratio 0.69 to 1.0

‐6.0%
Concrete H/D ratio 1 to 2.0
Concrete H/D ratio >2.0
‐8.0%
0 5 10 15 20 25 30 35
Bin diameter (m)
Figure 3-22a Percent difference between WPACKING model predicted and reported mass, with
respect to bin diameter sizes and classified based on grain height/diameter ratio (H/D); for the
grouped bins in Table II.
80
Bin diameter (ft)
0 20 40 60 80 100 120
12.0%
10.0%
8.0%
6.0%
4.0%
2.0%
0.0% Steel H/D ratio 0.69 to 1.0

‐2.0%
concrete H/D ratio 0 to 0.68

‐4.0%

‐6.0%
‐8.0%
0 5 10 15 Bin diameter (m) 20 25 30 35
Figure 3-22b Percent difference between RMA predicted and reported mass, with respect to bin
diameter sizes and classified based on grain height/diameter ratio (H/D); for the grouped bins in
Table II.
81
Bin diameter (ft)
0 20 40 60 80 100 120
12.0%
10.0%
8.0%
% diff between FSA predicted and Reported mass
6.0%
4.0%
2.0%

0.0%
Steel H/D ratio 0.69 to 1.0

‐2.0%
‐4.0% concrete H/D ratio 0 to 0.68

‐6.0%

‐8.0%
0 5 10 15 20 25 30 35
Bin diameter (m)
F
Figure 3-22c Percent difference between FSA-W predicted and reported mass, with respect to bin
diameter sizes and classified based on grain height/diameter ratio (H/D); for the grouped bins in
Table II.
82
3.5 Time and Aeration Effects on Compaction Factors for Grain Bins
3.5.1 Background
Compressibility in food grain materials is both a common and natural phenomenon that
occurs in commercial grain bin storage facilities. The grain particles behave diversely when
compressed and bulk density changes with increasing grain depth in the bin. Variation of
pressure in a typical grain bin can be modeled and studied based on Janssen’s classic grain
pressure equation correlating unit vertical pressure at any depth, bulk density, bin dimensions,
and bin wall friction (Janssen, 1895). Further studies by Loewer et al. (1977), Clower et al.
(1973), Thompson and Ross (1983), Thompson et al., (1987; 1990) have showed that bulk
density of corn is related to moisture and pressure, modeled coefficients of friction and lateral to
vertical pressure ratio (two constants of Janssen’s model) as a function of moisture and pressure,
and finally developed the amount of compaction in grain bin for any grain type, bin geometry,
discharge type, grain profile shape, given moisture and test weight values of the crops, and bin
material type, respectively. The works of Thompson et al. were later adopted by the American
Society of Agricultural and Biological Engineers in 1992 and was later revised and readopted on
2010 (ASABE Standards, 2010). This standard enables prediction of the amount of compaction
in variety of grain bin for any given crop type with known property values.
The loading effects and resultant moments of force for wheat and corn in corrugated steel
bins with height to diameter ratio of 4 were studied by Horabik et al. (1995). The vertical wall
load to total grain load ratio was affected by difference in the wall friction due to different bin
wall type, grain type, and position of eccentric discharge. During eccentric discharge of a 0.61 m
diameter model grain bin, floor loads were smaller on the side nearer to discharge orifice than on
the opposite side. Eccentric discharge results in unequal horizontal pressure distribution along
the circumference of a bin wall (Horabik et al., 1992). Thompson et al., (1986) showed that on a
model wheat bin when the discharge point was semi eccentric rather than eccentric, the discharge
point is near the wall. Molenda et al. (1993) provided information on spout filling method,
sprinkle filling, and uniformly filling through an annular ring near the bin wall for varying bin
wall types, and showed effects on static wall load to total grain load and dynamic to static wall
load ratio. Moysey (1984) studied the effect of grain spreaders on grain friction and bin wall
pressures and showed that method of filling the shear test box had a significant effect on bulk
density changes. It also showed that wall and hopper pressures during filling were lower when a
grain spreader was used and if there was stream filling. Chang et al. (1983) studied the effect of
4-trough, 2 trough and 1 trough self-propelled rotational grain spreaders for a 6.4 m dia bin for
wheat, corn, and grain sorghum. They also showed that bulk density and airflow resistance of
grain that was transferred with a spreader was higher than without spreaders, and that airflow
rate through grain was higher for situations where grain spreaders were used. Similarly, a
research study by Stephens and Foster (1976) showed that there was an increase in bulk density
from 766 to 871 kg/m3, improved the uniformity of distribution, and increased airflow resistance
83
from 122 Pa/m to 379 Pa/m at a given airflow rate for shelled corn bins when three mechanical
grain spreaders were used.
The effect of vibration on hopper flow of granular material (non-food grain) caused a slight
increase in discharge rate from vibrating hoppers, reaching a maximum value when the velocity
amplitude is greater than 1, but beyond that the discharge rate decreased slightly (Wassgren,
2002). A study by Ge et al. (2000) showed that maximum consolidation (density) for food grain
storage was independent of the vibration frequency given to the modeled grain bins However,
lower amplitudes and higher vibrations caused increased consolidation for red hard spring wheat.
Also, the rate of consolidation increased with grain depth in a non-linear fashion, and both
frequency and amplitude affected variation in grain consolidation. Hao et al. (1994) showed a
39% increase in lateral pressure on wheat bin walls due to vibration. Cheng and Zhang (2006)
used 30 Hz vibration frequency for 20 minutes on a model corrugated steel wheat bin of 5 m
diameter by 10 m - and predicted that the lateral pressure near the bottom of the bin increased by
17% and the bulk density increased by 8%. However, the principal factor for vibration from
studies in grain bins is the practicality. Ge et al. (2000) clearly stated that the evidence of about
10% maximum settlement or consolidation does not occur in field conditions. Under field
conditions where vibrations are caused mainly by railway traffic, the vibration parameters do not
seem to cause any added compaction in the grain bin. A former issue recognized by FSA from
vibration due to an unusual type of drive on the elevator leg does not occur in the modern grain
industry and elevators because those types of leg drives are no longer used. The FSA warehouse
group currently believes that equipment vibration in elevators does not cause significant
additional packing or consolidation in bins (Unpublished interview, 2011). Malm and Backer
(1985) showed that for six major crops in United States the proximity to the railroad, differences
in crop year, and availability of aeration had no significant effect on compaction factors of grain
bins. However, they also stated that there were problems in studying settling time effects on
grain bins because the change was very small between readings (23 to 40 days after filling of the
bin) and the storage bin was frequently disturbed. Thus, it is necessary to look further into grain
settlement and compaction changes for grain bins stored for long periods of time. The objective
to this paper is to study the effects of time and aeration on grain depth and compaction values for
grain crops for up to a year.
3.5.2 Data Collection

Three steel corrugated bins (Bin A, Bin B, and Bin C) at the USDA-ARS Center for Grain
& Animal Health Research (CGAHR), measuring 18.6 ft in diameter and 13.3 ft in eave height,
were studied for three days, from 6/25/2012 to 6/28/2012. The same bins were measured again
in September when two readings were taken 20 day apart. The aeration fans were activated at the
time of the first September measurement and ran for 155.5 hours before the second September
reading was taken. For bins A, B, and C the average moisture contents were 11.9%, 11.7 %, and
12.0% (w.b.), respectively; the average test weights were 58.8, 58.3, and 57.8 lb/bu, respectively.
84
Four barley bins in Idaho were tracked for a period of 12 months (November 2012 to
November 2013). All four bins measured 88.5 ft in diameter and 66.0 ft in eave height. For these
four bins the average test weight was 57.5 lb/bu and he average moisture content was 9.5%
(w.b.). One sorghum steel corrugated bin of 41.42 ft diameter and 32.74 ft eave height and corn
bin steel corrugated bin of 17.53 ft diameter and 20.47 ft diameter was tracked for 6 months with
no aeration in north eastern part of KS. However, a single aerated corn bin (24-ft diameter and
21.0-ft eave height) was monitored in Kansas for a period 6 months where it experienced a total
of 852 h of aeration. For KS corn aeration bin average test weight was reported as 59.85 lb/bu
and moisture 13.51% (wb).
Calculated packing factor was determined as the ratio of reported grain mass (from scale
tickets) to the calculated grain mass from volume measurements. This ratio was reported as a
percentage for the calculated packing factor. The packing factor from WPACKING (ASABE,
2010) was calculated based on grain height, moisture, test weight, angle of repose, and hopper
bottom angle of the bin and then compared graphically with the calculated packing factor for the
given bins.

For the USDA-ARS, CGAHR corrugated steel bins with HRW wheat, there was no
significant change in grain height when stored for less than a week, but compaction increased
between the first and second readings — for Bin B the change was 1.25% and for Bin C it was
0.64% (Fig 3-23). After the first reading the grain profile was flattened manually and a slight
increase in compaction may be expected due to disturbing the grain and walking on the surface.
However, such slight changes are negligible (< 1.0%) in a practical storage context. The
WPACKING-predicted compaction remained constant for the three bins. During the 20 days of
aeration, the grain height and compaction remained constant (Figure 3-24).
12.0 USDA time effect bins

11.8
11.6
11.4
Grain Height (ft)
11.2
Bin A
11.0
bin B
10.8
bin C
10.6
10.4
10.2
10.0
6/23/2012 6/25/2012 6/26/2012 6/28/2012 6/29/2012
Time
85
3.3 USDA time effect bins
3.2
3.1
Bin A
Compaction ( %)
3.0 bin B
2.9 bin C
Wpack A
2.8
Wpack B
2.7
Wpack C
2.6
2.5
6/23/2012 6/25/2012 6/26/2012 6/28/2012 6/29/2012
Time
Figure 3-23 Plot of Grain height vs. time and compaction factor (from WPACKING software)
vs. time for wheat in corrugated steel bins in Kansas.
86
12.0 USDA Aeration bins ( 155.5 hours)
11.8
11.6
11.4
11.2
Grain Height (ft)
11.0 Bin A
bin B
10.8
bin C
10.6
10.4
10.2
10.0
9/8/2012 9/13/2012 9/18/2012 9/23/2012 9/28/2012 10/3/2012
Time
14.0 USDA Aeration bins ( 155.5 hours)
12.0
10.0
Bin A
Compaction ( %)
8.0 bin B
bin C
6.0 Wpack Bin A
Wpack Bin B
4.0 Wpack Bin C
2.0
0.0
9/8/2012 9/13/2012 9/18/2012 9/23/2012 9/28/2012 10/3/2012
Time
Figure 3-24 Plot of Grain height vs. time and compaction factor (from WPACKING software)
vs. time for wheat in steel bins stored 1 month in Kansas. The bin was aerated for 155.5
hours between the two readings.
87
Figure 3-25 shows time effects for barley stored in corrugated steel bins for one year –
November 2012 to November 2013. In Figure 3-25 for Bins 1, 2, and 5, a significant decrease in
grain height is noticed after 8 months of storage, i.e., from July 2013, point #6. However, it is
clear that from June 2013 (point # 5) onwards, which is after 7 months of storage, there was a
slight decrease of grain height of 0.2%, and final decrease was 1.42% (on Nov 2013), when
compared to April 2013 data (point #5). This is apparently due to settling of the grain in the bin
when stored for long periods of time. Up to the April 2013 data (i.e., 5 months of storage) there
was no change in grain height observed in all three bins. Similarly, the calculated compaction
factor increased by 0.41% in July 2013 (point # 6, 8 months of storage) and by 1.41% in
November 2013, after 12 months of storage, compared to the compaction calculated for June
2013 (data point # 5). This behavior was only for only Bins 1 and 2, while Bin 5 only showed an
increase in compaction starting in April 2013, after 5 months of storage.
Figure 3-26 depicts the change in grain height for one corrugated steel bin (Bin 6)
containing barley with the same dimensions as bins 1, 2, and 5, but monitored only 6 months
(Nov 2012 to July 2013). A decrease in grain height of 0.46% after 5 months of storage was
observed when compared to the previous reading (data point # 2) on Jan 2013. However, this
change is very small and tracking of grain height ended after 6 months of storage. Also, the
WPACKING model predicted packing factors of 11% (for Bin 2 and 6) and 11.1 % (for Bin 1
and 5). The slight difference in packing percent was due to slight differences in initial grain
height and angle of repose. The very similar predicted packing factor for these very similar bin
conditions was expected according to the principles of the model.
88
Time effect in barley bin (ID)
66.0
65.0
64.0
Grain Height (ft)
63.0 Bin 1
Bin 2
62.0 bin 5
61.0
60.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Time of Storage (months)

9.5
8.5
7.5
Compaction ( %)
6.5
Bin 1
5.5 bin 2
bin5
4.5
3.5
2.5
0 1 2 3 4 5 6 7 8 9 10 11 12 13
Figure 3-25 Plot of grain height and calculated compaction factor vs. time of storage in months
for barley stored in corrugated steel bins in ID. Storage was from November 2012 to
November 2013.
89
66.0
65.0
64.0
Grain Height (ft)
63.0
Bin 6
62.0
61.0
60.0
0 1 2 3 4 5 6

9.5
8.5
7.5
Compaction ( %)
6.5
5.5 Bin 6
4.5
3.5
2.5
0 1 2 3 4 5 6
Figure 3-26 Plot of grain height and calculated compaction factor vs. storage time (months) for
barley crops in a corrugated steel bin in Idaho stored from Nov 2012 to July 2013.
90
The aerated steel bin with corn in Kansas exhibited a change in grain height after 4 months
of storage (Figure 3-27). This bin showed a 0.54% decrease in grain height in January 2013 and
an increase calculated compaction factor from 3.5% to almost 4%. The WPACKING predicted
compaction remained unchanged at 3.2% for the bin, since WPACKING does not use any time
or aeration effect. It was a 12 inch diameter and 1 HP vane axial fan used to aerate the corn bin
discussed in Figure 3-27. The grain bin was 24 ft in diameter and 21 ft in eave height and it was
corrugated steel bin. Based on the inputs of bin size and fan type, there was about 1,458 cfm
amount of airflow at the given grain depth, calculated from University of Minnesota web based
program for grain bin fan control and usage (http://webapps.bbe.umn.edu/fans/).
Corn Aeration Bin At KS (852 hours)

21.0
20.9
20.8
20.7
20.6
Grain Height (ft)
20.5
20.4
20.3
20.2
20.1
20.0
0 1 2 3 4 5 6 7 8
Storage time (months)
Corn Aeration Bin at Kansas

4.4
4.2
4.0
Compaction (%)
3.8
3.6
3.4
3.2
3.0
0 1 2 3 4 5 6 7 8
Storage time (months)
Figure 3-27 Plot of grain height and calculated compaction factor vs. storage time (months) for corn in an aerated
steel bin in Kansas. The bin was aerated for a total of 852 hours for the entire monitoring period.
91
Figure 3-28 represents storage and compaction behavior of two corrugated steel bins – one
for milo with 41.42 ft in diameter and 32.74 ft in eave height and one for corn with 17.53 ft in
diameter and 20.97 ft in eave height. However, the initial grain height, i.e., the height of the grain
from the floor of the bin to the sidewall where the cone profile of the grain starts, is 7.49 ft and
15.83 ft for milo and corn bin, respectively. There was some noticeable decrease in the grain
height due to long term storage. For milo crop, the grain height decrease after 4 months of
storage and there was of 0.67% and 2.80 % decrease in grain height at 4 and 6 months of storage
respectively, compare to the first month storage data point. Consequently, the calculated packing
increased to 0.29% (for 4 month storage) and 0.49% (for 6 month storage), compared to first
month storage compaction. Similar trends where seen for the corn steel bin; the grain height
decreased about 0.13% after 4 months and 0.63% after 6 months of storage, compared to first
month’s grain height. The compaction increased 0.16% after 4 months and 0.63% after 6 months
of storage. These two bins are non-aerated. WPACKING model predicted 2.3% packing percent
for milo bin and 2.9% for corn bin. WPACKING model predicted packing percent were constant
with change in grain height as reported in Figure 3-28. This was quite reasonable as because
WPACKING model does not change packing percent to such minor grain height changes but the
predicted mass (data not shown here) does changes according to the given grain height, angle of
repose, and other bin geometry parameters.
92
17.0
16.0
15.0
14.0
13.0
Grain Height(ft)
Bin 5(Milo)
12.0
Bin 2 (Corn)
11.0
10.0
9.0
8.0
7.0
6.0
5.0
0 1 2 3 4 5 6 7
3.5
3.0
2.5
Compaction (%)
2.0
Bin 5(Milo)
1.5
Bin 2 (Corn)
1.0
0.5
0.0
0 1 2 3 4 5 6 7
Figure 3-28 Plot of grain height and calculated compaction factor vs. storage time (months) for
corn and milo crop in non-aerated steel bin in Kansas.
93
HRW concrete round bins (located in USDA facility of Manhattan, KS) of 15 ft diameter ,
eave height (side wall height) of 83 ft, and hopper bottom angle of 33.76° was tracked for a
period of about a month (~ 800 hrs) and change in grain height was measured and plotted against
the time in Figure 3-28. Furthermore, the change in grain height (delta height) vs. time was
modeled using CurveExpert Software (version 2.0.3, Informer Technologies, Inc. 2013) with non
linear mathematical models. The best fit model and its predicted plot is given in Figure 7. The
average initial grain height in the bin was 78.55 ft and then change in grain height was measured
using laser device as discussed in HRW wheat commercial field data section previously. The
grain was transferred between two identical concrete bins in the USDA facility in Manhattan,
KS. Hence, the initial grain height was varying slightly and hence, average grain height is
reported here. However, the delta height is the difference in the grain height based on its initial
height. Figure 3-29 reveals those three non-linear models – reciprocal logarithm, exponential
association 3 and 2. The details of the model structure and parameter estimates are given in
Table 3-8.
Figure 3-29 Non-linear fit of change in grain height (delta height, ft) vs. time (hrs) for HRW wheat
in round concrete bins located in Manhattan, Kansas.
Based on the results of nonlinear modeling with CurveExpert it is evident that the exponential
model (model no. 1) is the best model that would predict delta as a function of time due to its highest R2
of 0.83, lowest Akaike Information Criterion (AIC) of -141.809 and standard error of 0.034, compared to
other two models in Table 3-8. AIC is a statistical parameter that strikes a balance between the goodness
of fit of a model and the complexity of the model. The preferred model will have the lowest AIC value,
and this technique includes a penalty prediction that discourages any increase in the number of parameters
that can lead to over-fitting and a higher goodness of fit (Akaike, 1974; Fang, 2011). The change in the
grain height was very low, ranging from 0 to 0.25 ft over a period of a month, and probably not
significant to conclude for grain height for long time storage. However, previous grain height examples
for HRW wheat (Kansas), barley (Idaho), and corn (Kansas with aeration) were all corrugated steel bins.
Hence, the data shown in Figure 3-28 for HRW wheat in concrete bins provides a short term look at grain
height change for tall and narrow bins, which are usually made of concrete as these were. Also, short term
storage effects for HRW wheat concrete bins will be valuable for future large scale time and aeration
effect storage studies relevant to grain compaction.
94
Table 3-8 Selected non-linear regression models for modeling delta height (ft, Y) = f( time, (hrs, X)*
Model Model
No. Name Family R2 AIC Std error Score Model Structure
1
Reciprocal logarithm Exponential 0.832 -141.809 0.034 769 Y= 1/(a + b*ln(X))
a = 1.27E+01, b = -1.29E +00

2
Exponential Association 2 Growth 0.763 -132.049 0.042 696 Y= a* (1- exp (-b*X))
a = 2.05E -01, b = 3.19E -02

3
Exponential Association 3 Growth 0.74 -132.56 0.042 677 Y= a* (b- exp (-c*X))
a = 1.96E -01, b = 1.08E + 00, c = 2.41E - 02
*Score represents a weight given by the CurveExpert Professional software to rate the model performance; AIC is the Akaike Information Criterion; Std error is the statistical standard error; X (time,
hrs) and Y (delta height, ft);R2 is coefficient of determination; Delta height is the change in the grain height in the concrete bins. Y is the dependent variable and X is independent variable.
95
3.6 Field data for other crops: soybeans, sorghum, oats, and barley
3.6.1 Background
As with the HRW wheat and corn field data chapters (sections 3.2 and 3.4) we have used
the LIECA DISTO laser meter and similar methods to measure the grain profile, angle of repose,
grain height, bin geometry and material data for other crops: soybeans, sorghum, oats, and
barley. Simultaneously, we also received crop quality data like moisture, test weight, damage,
and foreign material from the cooperators. Hence, using those data, predicted mass for each bin
was calculated using the WPACKING model and the RMA method. Differences between
predicted mass and reported mass from ticket sale data is again plotted against bin capacities.
Field data collection from commercial elevators and farmers has also been done for
soybeans, sorghum, oats and barley. Commercial soybean bins were all corrugated steel. The
diameters ranged from 14 ft to 75 ft and the eave height was from 13.01 ft to 74.87 ft. The
difference between predicted mass using both the WPACKING model and the RMA procedure
has been calculated for these steel bins and it was found to range from -3.04 % to +2.26% and -
4.40% to +6.87% for the WPACKING model and RMA method, respectively. The absolute
average was +1.02% for WPACKING and +3.30% for RMA. Based on this absolute percentage
and Figure 3-30, it is clear that WPACKING model did better than the RMA method in terms of
prediction. For 95% of the bins, both the WPACKING and RMA methods over-predicted.
Similarly, sorghum bins were found to have diameters from 14.95 ft to 89 ft and eave heights
from 40.96 ft to 140.23 ft with mostly concrete bins except four steel corrugated bins that are
marked in Figure 3-31. Differences were found to range from -2.34% to +2.51% (absolute
average of 0.84%) for the WPACKING model, whereas, it ranged from -7.37% to +9.32%
(absolute average of +5.10%) for the RMA method. Again, the RMA method had more
variations in the differences than the WPACKING model, indicating WPACKING model
provided better predictions compared to the RMA method. The four steel bins in Figure 3-31
showed under-prediction but the difference for RMA method (around -7.0%) was much larger
than for the WPACKING model (around -0.1%).
For oats, most of the field data was collected from concrete bins including a few concrete
interstice bins. The bin diameters ranged from 13.42 ft to 89.50 ft and eave height 84.33 ft to
124.07 ft. The steel bins and concrete interstice bins are marked in Figure 3-32. Differences
ranged from -6.45% to +8.81% (absolute average of +4.17%) for WPACKING model and -
96
31.59% to +9.07% (absolute average of 12.18%) for the RMA method. Clearly, concrete
interstice prediction using RMA method deviated a lot from actual ticket sale data. Likely, this
method was never calibrated for use with interstice bins. Also we found steel bins yielding
around +7.0% difference using both the WPACKING model and the RMA method, whereas,
concrete bins were randomly distributed across the zero percent line for both the methods, with
RMA differences higher than WPACKING.
Barley bins were all corrugated steel except one concrete bin, which showed much less
difference. Bin diameters ranged from 20.01 ft to 105.0 ft and eave height from 60 ft to 112.73
ft. The WPACKING model yielded differences from 0.66% to +6.59% (absolute average of
+3.53%) and the RMA method yielded -15.18% to +2.71% (absolute average 10.02%). Once
again, the RMA method showed higher values of differences and it mostly under–predicted. It
did better in predicting the one concrete bin, but this is not a conclusive result with only one
concrete bin. On the other hand, the WPACKING model showed only over-prediction for all the
bins as shown in Figure 3-33.
97
8.0%
0 50,000 100,000 150,000 200,000 250,000 300,000
7.0%
6.0%
5.0%
4.0%
3.0%
Soybean
2.0%
1.0%
0.0%
‐1.0%
RMA
‐2.0%
WPACKING
‐3.0%
‐4.0%
‐5.0%
‐6.0%
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Reported mass (t)
Figure 3-30 Percent differences between predicted mass and reported mass using
WPACKING model and RMA method for soybeans, all steel bins.
Reported mas (standard bushel)
0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000
10.0%
8.0%
Sorghum
6.0% WPACK
RMA
4.0%
2.0%
Steel bins
0.0%
‐2.0%
Steel bins
‐4.0%
‐6.0%
‐8.0%
‐10.0%
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000
Reported mass (t)
WPACKING model and RMA method for sorghum, all concrete bins unless indicated.
98
Reportee mass (standard bushel)
0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 80,000
15.0%
10.0%
5.0%
% dif between Prredicted and Reported capacity
0.0%
‐5.0%
Steel bins WPACK

‐10.0%
RMA
Oats
‐15.0%
‐20.0%
‐25.0%
‐30.0%
Concrete interstice
‐35.0%
0 2,000 4,000 6,000 8,000 10,000 12,000
Reported Capacity (t)
WPACKING model and RMA method for oats, mostly concrete bins unless indicated.

0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 400,000 450,000
10.0%
5.0%
% diff between Predicted and Reported mas
0.0%
Concrete bin
‐5.0% WPACK
Barley RMA
‐10.0%
‐15.0%
‐20.0%
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000
Reported mass (t)
WPACKING model and RMA method for barley, mostly steel bins unless indicated
99
4. COMPACTION FACTOR PROGRAM
A new user friendly software product was developed to predict pack factors based on the
laboratory and field data. The software runs within MS Excel 2007 and newer where the user can
enter relevant data and display results. Bin configuration diagrams are included so that needed
grain and bin geometry inputs to the model are clear for the user. The model comes conveniently
embedded inside an MS Excel 2007 spreadsheet file.
The RMA Excel version has three separate tabs (worksheets) across the bottom. The first sheet,
“Information,” describes the basic layout and operation of the software and other basic
information. The second sheet, “Input values,” shows required input material, grain and bin
properties, which is entered with the help of interactive diagrams on a separate User Form. The
third sheet, “Production Worksheet,” provides the main calculated values, e.g. combined test
weight and pack factor along with showing the key input values, all in a format very similar to
previous RMA Production Worksheets used for pack factor calculations.
4.1 Using the Program
Typical MS Excel security settings have macros, such as this program, disabled by default and
require the user to specifically enable the macro for the program to run. When the Excel file with
the program is opened in Excel 2007 and later a "Security Warning" notification bar should
display just below the menu bar. Instructions for enabling macros are included on the
"Introduction" sheet.
Figure 4-9 The "Security Warning" notification bar that should be displayed when the
program is opened.
100
If the "Security Warning" notification bar is not displayed under the menu tab, the needed
settings in Excel can be changed as follows:
1. Click the "Developer" menu tab (see below if it is not already displayed).
2. Click "Macro Security" under the Code section.
3. Click the "Disable all macros with notification" check box.
If the "Developer" menu tab is not shown, enable it as follows:
1. Click the Microsoft Office Button (top left).

2. Click "Excel Options".
3. Click "Popular" if it is not highlighted.
4. Select the "Show developer tab in the ribbon" check box.
4.2 "InputValues" Sheet
The InputValues sheet shows the grain and bin properties necessary to calculate compaction
factor. When the tab at the bottom for this sheet is selected it will display the input values the
model will use to calculate the compaction factor. Default values will be displayed until the
properties are modified in the user form, using [Ctrl + g]. Default values are not designed to be a
valid case, so you would not get a false result if you forget to enter your values; if you try
execute the calculations without putting in proper values the program will not run.
To modify grain and bin properties:
1. Select the InputValues tab by clicking that tab at the bottom of the worksheet (this is the
preferred starting point, but the keyboard shortcut can be pressed from any location in the
spreadsheet).
2. Enter keyboard shortcut [Ctrl + g].
This screen is a User Form that contains seven different tabs across the top that should be
carefully completed to insure that all of the required data is input into the model. After all of the
data is entered, the CLOSE button in the seventh tab, HopperProperties, must be clicked to
save the input data and proceed. The sections below explain the seven data input tabs, and
provide a tutorial on how to correctly supply all of the necessary information for the grain
compaction program.
101
4.2.1 Pre-Data
The first tab allows for the Date and bin Location to be input into the model. This information is
not necessary, but allows for convenient organization of data. The compaction factor model will
still operate correctly if this tab is left blank.
To modify information in the Pre-Data tab:
1. Select the Pre-Data tab at the top of the window.

2. Enter the Date information by-
a. Select the Month from the drop down menu.
b. Select the Day by using the arrows, or by inputting the numbers manually in
the box.
c. Select the Year by using the arrows, or by inputting the numbers manually in
the box.
3. Enter the Location information by-
a. Entering the Name of the location.
b. Entering the City that the location is in.
c. Selecting the State from the drop down menu.
Figure 4-10 The Pre-Data tab completed with example information.
4.2.2 GrainProperties
The GrainProperties tab allows for the physical grain property information to be input. The
Type of Grain, Moisture Content, and Test Weight are important grain properties that are
included in the calculation of the grain compaction factor. Values must be put into each box in
this tab or the program will show an error dialogue box when the ‘Input Value’ screen is closed.
To modify information in the GrainProperties tab:

102
1. Select the Grain Properties tab at the top of the window.
2. Select the Type of Grain from the provided list.
3. Click the OK button below the provided list once the grain type is selected.
4. Select the Moisture Content of the grain by using the arrows. This value should be in
percentage form.
5. Enter the Test Weight value of the grain into the appropriate box. This value should be
in pounds per bushel (lbs/bu.)
Figure 4-11 The GrainProperties tab completed with example data.
4.2.3 Grain Storage
The Grain Storage tab allows input of the bin geometry. The Type of Storage and the
corresponding dimension(s) necessary for grain compaction factor calculation. The geometry of
the storage bins cross sectional area should be used to determine the type of storage that should
be selected. The figure illustrates the necessary input dimensions for the different storage types.
To modify information in the Grain Storage tab:
1. Select the Grain Storage tab at the top of the window.

2. Determine the Type of Storage using the figure provided, select the corresponding Type
of Storage in the provided list.
3. Click the OK button below the provided list once the storage type is selected.
4. Enter the necessary geometric dimensions depending on the particular type of storage
selected. The program will not use any dimensions entered for the other types of storage
that were not selected under Type of Storage.
a. For Circular, Square, or Polygonal Storage Bins:
103
i. Enter the Bin Diameter (d) for circular storage bins, or the Bin
Dimension (a) for square and polygonal bins in the appropriate box. This
value should be in feet (ft.).
b. For Rectangular Storage Bins:
i. Enter the Bin Width (a), and the Bin Length (b) in the appropriate boxes.
These values should be in feet (ft.).
c. For Round Pile Storage:
i. Enter the Pile Base (diameter) in the appropriate box. This value should
be in feet (ft.).
Figure 4-12 The Grain Storage tab completed with the necessary information for a
circular storage bin with a 22-ft. diameter.
4.2.4 Walltype&GrainHeight
The Walltype&GrainHeight tab allows for the inclusion of the bin wall surface and the height
of grain mass in the compaction factor model. These two properties will be used to determine the
frictional forces and the lateral to vertical pressure ratio for the model. It is important to note that
the grain height is the height of grain at the side wall of the bin. If the grain mass does not
entirely fill the bottom hopper of a bin, or is a round pile, the grain height is assumed to be zero
for proper calculation of those cases.
To modify information in the Walltype&GrainHeight tab:
1. Select the Walltype&GrainHeight tab at the top of the window.

2. Select the Bin Wall Type from the provided list.
3. Click the OK button below the provided list once the wall type is selected.
4. Enter the Grain Height into the appropriate box. This value should be in feet (ft.).
5. If the grain mass does not entirely fill the hopper of a bin, or is in pile storage, leave the
Grain Height value as zero.
104
Figure 4-13 The Walltype&GrainHeight tab completed for a reinforced concrete bin
with a grain height of 52-ft.
4.2.5 Deductions
The Deductions tab allows the model to account for any object, or structure located inside of the
storage bin that will decrease its capacity. For example, an aeration tube in the bottom of a
storage bin will occupy some volume of the bin, and therefore reduce the capacity of grain that
can be stored. Accounting for these capacity deductions is required for accuracy of the
compaction factor program. To modify the information in the Deductions tab:
1. Select the Deductions tab at the top of the window.
2. Enter the total volume of Deductions present in the storage bin. This value should be in
cubic feet (cu. ft.).
3. If there are no deductions present in the storage bin enter the Deductions value as zero.
Figure 4-14 The Deductions tab completed for a bin with a 3-ft diameter x 11-ft long aeration pipe.
105
4.2.6 GrainSurface
The GrainSurface tab allows for inclusion of the grain mass surface condition in the compaction
factor program. Depending on the filling or emptying conditions that the bin has experienced, the
Top Grain Surface can be coned up, coned down, or level. If the grain mass is coned up or
down then the Cone Height must be determined for use in the model.
To modify the information in the GrainSurface tab:
1. Select the GrainSurface tab at the top of the window.

2. Select the type of Top Grain Surface present in the storage bin from the provided list.
3. Click the OK button located below the provided list once the Top Grain Surface type
has been selected.
4. If a coned up or coned down Top Grain Surface is selected, then the Cone Height must
be entered into the appropriate box. The Cone Height should be entered as a positive
value regardless of whether the surface is coned up or down. This value should be in feet
(ft.).
5. If the Top Grain Surface is level, the grain is being stored in a round pile, or the grain
does not completely fill the hopper of the storage been then the Cone Height box should
be left blank.
Figure 4-15 The GrainSurface tab completed for a level grain surface.
4.2.7 HopperProperties
The HopperProperties tab allows for hopper bottom bin geometries to be included in the
compaction factor program calculations. For hopper bottom bins, the Hopper Height and
106
Hopper Grain Height must be determined for use in the model. If the grain mass is above the
hopper, then the Hopper Grain Height is equal to the Hopper Height.
To modify the information in the HopperProperties tab:
1. Select the HopperProperties tab at the top of the window.

2. Select the type of Discharge Conditions present in the storage bin from the provided list.
3. Click the OK button located below the provided list once the Discharge Conditions
have been selected.
4. If the hopper bottom bin Discharge Conditions is selected, then the Hopper Height and
the Hopper Grain Height must be entered into the appropriate boxes. These values
should be in feet (ft.).
5. If the flat bottom bin or round pile Discharge Conditions are selected then the Hopper
Height is assumed to be zero, and the Hopper Grain Height should be left blank.
Figure 4-16 The HopperProperties tab completed for a hopper bottom bin with a
hopper height of 12-ft and a grain level above the hopper.
After all of the input data has been correctly added to the program, you may return to the
worksheet by clicking the CLOSE button at the bottom of the HopperProperties tab. If the
program is closed before all values are correctly entered into the program the model will not
operate properly and an error message will be shown.
107
Figure 4-17 The error message displayed after the incorrect input of the grain and bin
properties.
Once the input value window is closed the information is no longer stored in the input tabs. All
data must be re-entered if the input window is brought back up. Make sure that all information is
correct and properly entered into the program before closing the input value window.
4.3 "ProductionWorksheet" Sheet
The ProductionWorksheet sheet provides a convenient and organized summary of the

properties modified in the previous section, gross and adjusted production information, as well as
the combined test weight and pack factor. The top of the sheet shows the location information,
the type of construction (storage bin wall and bottom type), and the type of commodity being
stored.
108
Figure 4-18 The top of the ProductionSheet where the location information, type of
construction, and commodity type are listed.
Farther down the ProductionWorksheet, in Table A, a summary of the different bin

measurements can be found. These measurements include the bin geometries (length, diameter,
or width depending on the type of bin cross-sectional area), the depth of the cylindrical portion
of the storage bin, the hopper height, and any deductions to the capacity of the bin.
Figure 4-19 Table A of the ProductionWorksheet. The values of the different bin
measurements correspond to examples shown in the previos section.
Table B of the ProductionWorksheet shows the calculation of the gross production of the
storage bin being evaluated. The volume of grain contained in the bin (cubic feet) is determined
by adding the volume of grain contained in the cylindrical portion of the bin and the hopper. This
value is then converted into bushels using the proper unit conversion factor.
109
Figure 4-20 Table B of the ProductionWorksheet. This table shows the calculation of
the gross production of the example used in the previous section.
At the end of the ProductionWorksheet, Table C shows the calculation of the adjusted
production of the storage bin being evaluated. This table displays physical properties of the grain
(including dockage, moisture, and test weight) along with the Combined Test Weight and Pack
Factor determined by the model to determine the amount of grain in the storage bin.
Figure 4-21 Table C of the ProductionWorksheet. This shows the physical properties of
the example grain used in the previous section being combined with the
determined pack factor to calculate the adjusted production of the storage bin.
4.3.1 General Output Tab
In addition to the Production Worksheet tab discussed previously, there is also provision in
WPACKING program to get all the output data in a separate excel tab called as Output tab. The
screenshot of the output tab below shows the details on bin capacity in lbs and total volume (ft3),
further precise information about capacities related to cylinder part and hopper part of the bin is
also obtained. If the bin flat then capacity is automatically zero as given this below screen shot.
Hence, the total capacity of the bin is given directly in lbs. which is the most favored unit for
academic researchers. However, the total capacity in standard bushel unit (Total Standard
110
Bushels, bu.) which is the usual unit in the U. S. grain industry is also obtained. Other vital
information like compacted TW known as Packed Bushel Weight (lb/bu), Pack Factor
(obtained from WPACKING models), % Packing (given by the WPACKING models), and
Combined TW & Pack Factor (obtained by adjusting the pack factor with given initial TW).
Furthermore, instructions on how to proceed to calculate the next bin capacity and packing factor
is given, i.e., by pressing the Ctrl (control) and ‘g’ keys simultaneously will help to refresh all
the tabs and open the grain property form described in section 4.2.2.
Calculated Values
Bin Dimensions - (Bu/ft.) 567.99 To perform additional

calculations
Equivalent Height of Grain (ft.) use the short key
Cylinder: Round Pile: 45.00 Ctrl + g
3
Total Volume of Grain -(ft ) 31808.60
Cylinder: Round Pile: 31808.60
Hopper: 0.00
Bin Capacity:
Total Weight of Grain (lbs.) 1588085.42
Hopper: 0.00
Standard Bushel Weight (lb/bu) 56.00
Total Standard Bushels 28359.00

(Based on Standard
Bushel Weight)
Hopper: 0.00
Packed Bushel Weight (lb/bu) 62.13
%Packing 3.6
Pack Factor 1.036
Combined Test Weight 1.110

& Pack Factor
Figure 4-4 Output tab of WPACKING program showing bin capacity (lbs and standard
bushels), pack factor, % packing, and combined TW and pack factor.
111
5. CONCLUSIONS
Laboratory Compressibility Measurements
HRW Wheat, Variety, and Dockage Effects

After combining HRW wheat varieties from composites representing possible blending
conditions applicable to the grain industry, the composite compressibility relationships were
compared to predictions from the original individual varieties. There were ten composite samples
prepared from the 27 single variety samples. The packing of the composite samples could be
predicted, with an R2 of 0.95, based on the regressions derived for the 27 individual varieties
using moisture content and test weight as the only variables. The majority of the effects of
dockage were accounted for by using the measured test weight with dockage, thus no separate
accounting for effects of dockage was needed.
Compressibility Equations
The preliminary model, WPACKING, used a fixed-polynomial equation from Thompson et
al. (1987) to describe the amount of compaction that occurs within the grain because of the
overbearing grain. Four other potential models were evaluated to see if they could describe the
compressibility data more accurately. Some of the other models (the MMF equation and the
modified Page equation) were developed with a more theoretical basis than the fixed-polynomial
equation. All four of the new numerical models described the laboratory compressibility data
well and showed improvement over the previous fixed polynomial equation. The modified Page
equation fit the data slightly better than the other new models, although it occasionally could not
provide a best fit of the asymptotic coefficient contained within the model. As a result, the
Farazdaghi-Harris equation was selected as the best model for representing the laboratory
compressibility data.
Field Compaction Measurements

varying bin diameter, eave height, and grain properties were measured in 14 U.S. states to
determine the compaction factor of corn, soybeans, wheat, milo, barley, and oats. Bin diameters
ranged from 12 to 105 ft (3.6 to 32 m) and grain heights ranged from 7 to 140 ft (2.1 to 42 m).
Calculated compaction factors (based on the mass of grain in the bin determined from scale
measurements), WPACKING compaction factors (based on the computer model), and RMA
compaction factor were calculated for all measured bins and the resulting predictions of grain
mass compared. Specific conclusions for each section of the field measurements follow below:
HRW Wheat Field Data

varying bin diameter, eave height, and grain properties were measured in Kansas, Oklahoma, and
112
Texas to determine the compaction factor of hard red winter wheat. Bin diameters ranged from
15 to 104.6 ft (4.6 to 31.9 m) for steel corrugated bins and from 15 to 33.7 ft (4.6 to 10.3 m) for
reinforced concrete bins. Equivalent level grain heights ranged from 13.4 to 84.1 ft (4.1 to 25.6
m) for steel bins and from 17.9 to 136.6 ft (5.5 to 41.6 m) for concrete bins. These bins consisted
of flat bottom, regular hopper bottom, and side-discharge hopper bottom bins. The regular
hopper bottom bins had right conic hoppers with horizontal discharge openings while the side-
discharge hopper bottom bins had oblique conic hoppers with vertical discharge openings. The
WPACKING pack factor, RMA pack factor, and FSA-W pack factor were calculated for all
measured bins and compared to the reported mass of grain in the bin. The pack factor included
both the compaction and test weight effects on the grain. The reported mass was determined
from scale tickets provided by the operator. The major findings were:
 Pack factors predicted by the WPACKING model ranged from 0.929 to 1.073 for
steel bins and 0.986 to 1.077 for concrete bins. Those predicted by the RMA method
ranged from 0.991 to 1.157 for steel bins and 0.993 to 1.099 for concrete bins. Pack
factors predicted by the FSA-W method ranged from 0.985 to 1.126 for steel bins and
1.012 to 1.101 for concrete bins.
 For corrugated steel bins, the absolute mean difference between the model-predicted
mass and the reported mass was +1.64%, with a median difference of -1.26%. In most
of the bins, the computer model under-predicted the actual grain mass (13 out of 16).
The RMA absolute mean difference for steel (+4.41%) was higher than model
(+1.64%) and the FSA-W method (+3.40%). Most of the bins were over-predicted by
the RMA (12 out of 16) and the FSA-W methods (14 out of 16).
 For reinforced concrete bins, the absolute mean difference between the model-
predicted mass and the reported mass was +3.75%. WPACKING slightly over-
predicted the mass of grain in these bins. The RMA absolute mean difference for
concrete bins (+3.25%) was slightly lower than the model (+3.75%) and much lower
than the FSA-W method (+4.34%). The model-predicted median difference was also
much higher at +2.16% than those from steel bins. In most cases, the model and the
two methods over-predicted the actual reported value. Some of the large differences
observed in this case can be attributed to the unique geometry in these bins and the
difficulty in describing these bins using the program.
 Overall, WPACKING better predicted the mass of grain in the bins than the current
RMA and FSA-W procedures. Some of the differences may be because the RMA and
FSA-W methods do not include the effect of grain moisture content, or the bin wall
type on pack factors.
Corn Field Data

The major conclusions for the corn field date were the following:
113
 Commercial corn bins across U.S. were measured and grain mass in each bin were
calculated based on 3 methods – WPACKING model prediction, RMA method and
FSA-W method. All the 3 methods use the principles of grain compaction in bins.
 The corn quality properties like test weight, moisture content, damage, broken kernel
content were found to be in normal ranges. The average angle of repose was 20.40°
for corn bins. As far as the compaction factor and above methods are concerned
moisture and test weight are the two main factors that can affect grain compaction in
bin, apart from bin geometry and bin wall type. Most of the corn bins were made of
corrugated steel and while only a handful were concrete bins.
 The absolute average % difference between predicted mass and reported mass was
0.86% (-4.54% to +4.53%) for WPACKING model, 1.46% (-2.92% to 4.97%) for
RMA method, and 1.60 % (-3.33% to + 5.67%) for FSA-W method. The variability
of % differences was more in RMA method than WPACKING model prediction,
where most of the data points were close to zero percent difference line.
 For corrugated steel bins with diameter above 15 m (49.21 ft), WPACKING model
over predicted for H/D ratio 0 to 0.68 and under predicted for H/D ratio 0.69 to 1.0
compare to reported mass values. And for steel bins with diameter below 15 m,
irrespective of H/D ratios, % differences were randomly distributed.
 For corrugated steel bins with H/D ratio 0 to 0.68, RMA method showed random %
differences for diameter less than 13 m (42.653 ft) and over prediction for diameter
above 13 m. For other ratios and bin diameter less than 15 m RMA predicted almost
evenly on either side of zero percent difference line.
 For corrugated steel bins with H/D ratio 0 to 0.68 and bin diameter above 15 m,
FSA-W method showed over prediction but below 15 m diameter, it was evenly
distributed along zero percent difference line.
 Concrete corn bins showed mostly under prediction for all 3 methods, and the %
differences were higher for FSA-W method than RMA and WPACKING model.
Thus, conclusions from this study is very valuable to calibrate and fine tune WPACKING
model that will be used to predict grain mass in a bin based on volume measurement and
compaction factor. Such studies were not done previously by any group. This method of field
data calibration approach on WPACKING grain compaction model is vital even though such
studies comes added issues of cost, getting reported mass from grain managers, and geographical
scale of the project. However, we are able to provide this unique opportunity comparing
WPACKING model and other two known methods (RMA and FSA-W) of grain compaction
U.S. grain industry for commercial corn bins. This study not only will improve the prediction
capability of WPACKING model but also shows the deficiencies in RMA and FSA-W prediction
methods that are still being used today.
114
Other Issues
Test Weight and Dockage Relationships
The key conclusions of this research are the following:
 Regression models for predicting TW with dockage values (when TW without
dockage for HRW wheat samples are given) were obtained for four dockage classes
with R2 values ranging from 0.703 to 0.776.
 The global plot of TW with dockage vs. TW without dockage did not show any bias
towards the location of the TW values.
 The Linear and Power Model C 3D regression models were selected as the two most
favorable models to present TW with dockage = f(dockage level, TW without
dockage) with reasonably high values of R2 and low values of standard error and
AIC.
 The developed model is valid and suitable over a range of moisture contents from
10.0% to 12.9% (wb).
 Cross validation results showed reasonable average results for standard error of
prediction and calibration.
 Selected models are most effective for cases where the dockage levels are less than
2%. At dockage =0% TW with and without dockage are same. Hence, linear model
and power model C should be not be used when dockage =0%.
Grain compaction studies or any grain inventory-related research requires basic data from
elevator managers that include the TW with dockage (lb/bu). However, based on FGIS
guidelines, the TW for wheat is measured without dockage. These research findings and
regression models allow the prediction of TW with dockage when dockage levels and TW
without dockage are known. These results provide a crucial step towards field trial bulk HRW
wheat inventory validation studies and other silo and storage studies of HRW wheat.
Time Effect and Aeration Effects Bins

The measured effects of time on compaction factor were small in limited testing. For corn
steel bins for both aerated and non-aerated significant change in grain height was observed after
4 months of storage, however, the decrease for aerated bin was 0.54% and for non-aerated was
0.13%. Steel bins with barley grain started showing change in grain height after 5 to 8 months of
storage (average 6 months of storage). Change in grain height vs time showed reciprocal
logarithm curve (from exponential family of curves) fit with R2 of 0.83 and standard error of
0.034. Literature review confirmed that effect of vibration from nearby railway tracks on packing
and settlement of grain depth is very unlikely in commercial and practical scenarios. However,
because of limited number of grain bins that have been studied for time and aeration effects; we
would not take these results as conclusive. More studies on tracking grain depth for other
commercial bins with respect to aeration and time are still being carried out.
115
Minor Crops: Soybeans, Soft wheat, Barley, Sorghum, and Oat Field Compaction Measurements
 Soybeans the WPACKING model predicted % differences were from -3.04% to
+2.26% (absolute average of +1.02%) and RMA method was from -4.40% to +6.87%
(absolute average of +3.30%) where for both methods 95% of the bins showed over-
prediction by both RMA and WPACKING model prediction method..
 Sorghum the WPACKING model predicted differences from -2.34% to +2.51%
(absolute average of +0.84%) and using the RMA method yielded -7.37% to +9.32%
(absolute average of +5.10%), 85% of bins (only concrete) showed over-prediction
when RMA method was used but for WPACKING model it was mostly under-
predictions and close to zero percent difference line.
 Oats the WPACKING model predicted differences from -6.45% to +8.81% (absolute
average of +4.17%) but using RMA method it was from -21.64% to +9.07% (absolute
average of +12.18%). For two concrete interstice oat bins, RMA method gave around
-31.5% differences when compared to scale ticket mass data, which is higher than
normal ranges.
 Barley the WPACKING model gave +0.66% to +6.59% (absolute average of
+3.53%) differences but RMA method yielded -15.18% to +2.71% (absolute average
of +10.02%). Clearly, for oats and barley in some cases the % differences where
large. This could be logical because oats and barley are light weight, i.e., low bulk
density, and have the tendency to compact more than other crops. Thus, above normal
compaction sometimes makes it difficult to validate models with field data.
 Soft White Wheat- Twelve soft white wheat bins from ID and WA states, a mixture
of both smooth and corrugated walled steel bins and concrete bins have been
measured with bin diameter ranging from 13.25 ft to 78.04 ft and eave height from
64.30 ft to 112.73 ft , the average AoR was found to be 20.34°.
 Durum Wheat- Four durum wheat steel corrugated bins were measured with diameter
of 36 ft and eave height of 25.52 ft and average AoR of 20.94°.
Compaction Factor Program

A computer program was also developed to make the predicted packing factors readily
available to the grain industry. In current form, the program is written using Visual Basic and is
accessed using Microsoft Excel. The computer program requires input from the user based on the
type and basic properties of the stored product and the dimensions, geometry, and construction of
the storage structure.
116
6. REFERENCES
Akaike, H. 1974. A new look at the statistical model indentification. IEEE Transactions of
Automatic Control 19(6): 716-723.
Amundson, L. R. 1945. Determination of band stresses and lateral wheat pressures of a
cylindrical grain bin. Agricultural Engineering 26: 321-324.
ASABE Standards. 2010. EP413.2. Procedure for establishing volumetric capacities of
cylindrical bins. St. Joseph, Mich.: ASABE.
ASAE S448.1 Jul 2001(R2012). 2012. Thin layer drying of agricultural crops. American Society
of Agricultural and Biological Engineers. St. Joseph, Mich.
Bates, E.N. 1925. Estimating the quantity of grain in bins. USDA, Misc. Cir. No.41.
Bern, C., and T. J. Brumm. 2009. Grain test weight deception. University Extension, Iowa State
University. Applied Online available at :
http://www.extension.iastate.edu/publications/pmr1005.pdf Accessed on October 29, 2013.
Bhadra R., Boac, J. M., Casada, M.E., Thompson, S.A., Montross, M.D., McNeill, S.,
Maghirang, R.G. 2014. Technical Note: Comparison of test weights with and without
dockage for HRW wheat samples. Transactions of ASABE (in review).
Bhadra, R. 2013. Stored Grain Compaction Factor for Commercial Bins in U.S. World Grain
Magazine December 2013 Issue. Sosland Publishing Co., Kansas City, MO.
Bickert, W. G. and F. H. Buelow. 1966. Kinetic Friction of Grains on Surfaces. Transactions of
the ASAE 9(1): 129-131.
Boac, J., R. Bhadra, M. E. Casada, S. A. Thompson, M. D. Montross, S. G. McNeill, and R. G.
Maghirang. 2013. Stored grain compaction factors for wheat: comparison of methods.
Transactions of the ASABE. (in review).
Brubaker, J. E. and J. Pos. 1965. Determining static coefficients of friction of grains on structural
surfaces. Transactions of the ASAE 8: 53-55.
Burnham, K.P. and D. R. Anderson. 2002. Model selection and multimodel inference: a practical
information –theoretic approach (2nd ed.), Springer-Verlag, ISBN 0-387-95364-7
Burnham, K.P. and D. R. Anderson. 2004. Multimodal inference: understanding AIC and BIC in
model selection. Sociological Methods and Research 33: 261-304.
Casada, M. E. and P.R. Armstrong. 2009. Wheat moisture measurement with a fringing field
capacitive sensor. Transactions of ASABE. 52(5): 1785-1791
Chang, C. S., H. H. Converse, and C. R. Martin. 1983. Bulk properties of grain as affect by self-
propelled rotational type grain spreader. Transactions of the ASAE 26(5): 1543-1550.
117
Chang, C. S., L. E. Shackelford, F. S. Lai, C. R. Martin, and B. S. Miller. 1981. Bulk properties
of corn as affect by multiple-point grain spreaders. Transactions of the ASAE 26(6):
1632-1636.
Cheng, X. and Q. Zhang. 2006. A particulate model for predicting vibration-induced loads in
bulk solids storage bins. Transactions of the ASABE 49(3): 759-765.
Clower, R. E., I. J. Ross, G. M. White. 1973. Properties of compressible granular materials as
related to forces in bulk storage structures. Transaction of the ASAE 16(3): 478-481.
Cooper, A. R. and L. E. Eaton. 1962. Compaction behavior of several ceramic powders. Journal
of the American Ceramic Society 45 (3):97-101
Fang, Y. 2011. Asymptotic Equivalence between cross validations and Akaike information
criteria in mixed-effects models. Journal of Data Science 9(1): 15-21.
Ge, T., Q. Zhang, and M. G. Britton. 2000. Predicting grain consolidation caused by vertical
vibration. Transactions of ASAE 43(6): 1747-1753.
Graton, L. C. and H. J. Fraser. 1935. Systematic packing of spheres –with particular relation to
porosity and permeability. The Journal of Geology 43(8) Part I: 785-909.
Hao, D., Q. Zhang, and M. G. Britton. 1994. Effect of vibration on loads in a corrugated model
grain bin. Canadian Agric. Eng. 36(1): 29-35.
Holligan, P. J., R. Wingate-Hill, and H. Q. Danh. 1982. Capacities of rectangular containers
filled with free flowing materials. Journal of Agricultural Engineering Resources 27:
363-367.
Horabik, J. M. Molenda, and I. J. Ross. 1995. Comparison of load distribution in two similar
grain bins. Transactions of ASAE 38(6): 1875-1879.
Horabik, J., C. V. Schwab, and I. J. Ross. 1992. Non-symmetrical loads in a model grain bin
during eccentric discharge. Transaction of the ASAE 35(3): 987-992.
Hunter, A. J. 1985. Formulae for the volume of stored granular materials. Journal of Agricultural
Engineering Resources 32: 31-46.
Hunter, A. J. 1985. Formulae for the volume of stored granular materials. Journal of Agricultural
Engineering Resources 32: 31-46.
Illinois Agricultural Auditing Association. 1980. Test Pack Factor Tables. Bloomington, Ill.:
IAAA.
Janssen, H. A. 1895. Versuche uber getreidedruck in silozellen. Zeitschrift, Verin Deutscher
Ingenieure 39:1045-1049.
Kozak, M. 2008. Correlation and regression: similar or different concepts? Stat Transit –new
series 9(1): 159-162.
118
Kozak, M., W. Krzanowski, and M. Tartanus. 2012. Use of correlation coefficient in agricultural
sciences: problems, pitfalls and how to deal with them. Annals of the Brazilian Academy
of Sciences 84(4): 1147-1156.
Kullback, S. and R. A. Leibler. 1951. On information and sufficiency. The Annals of
Mathematical Statistics 22(1): 79-86.
Lambe, T. W. and R. V. Whitman. 1969. Soil Mechanics. New York: John Wiley & Sons, Inc.
Loewer, Jr. O. J., I. J. Ross, D. D. Kratzer, J. N. Walker. 1977. Properties of ground shelled corn
as related to forces in bulk storage structures. Transactions of the ASAE 20(1): 155-156.
Lorenzen, R. T. 1957. Effect of moisture content on mechanical properties of small grains. M.S.
thesis. Davis, California: University of California-Davis.
Lorenzen, R. T. 1959. Moisture effect on friction coeffecients of small grain. ASAE paper 59-
416, ASAE. St. Joseph, Mich.
Malm, J.K., and L.F Backer. 1985. Compaction factors for six crops. Transactions of the ASAE
28(5): 1634-1636.
Matz, S. 1991. The chemistry and technology of cereals as food and feed. 2nd Edition. Van
Nostrand Reinhold/AVI: New York, NY.
Mc.Neill, S. G., S. A. Thompson, and M. D. Montross. 2004. Effect of moisture content and
broken kernels on the bulk density and packing of corn. Applied Engineering in
Agriculture 20(4): 475-480.
McNeill, S. G., M. D. Montross, S. A. Thompson, I. J. Ross, and T. C. Bridges. 2008. Packing
factors of feed products in storage structures (Technical Note). Applied Engineering in
Midwest Plan Service. 1987. Storage. In Grain Drying, Handling and Storage Handbook.
(MWPS-13). 2nd ed. Ames, Iowa: Midwest Plan Service.
Mohsenin, N. N. 1986. Physical Properties of Plant and Animal Materials. 2nd edition. New
York, N. Y. : Gordon and Breach Science.
Molenda, M., and J. Horabik. 2005. Part 1. Characterization of mechanical properties of
particulate solids for storage and handling. In Mechanical Properties of Granular Agro-
Materials and Food Powders for Industrial Practice. J. Horabik and K. Laskowski. Eds.
Lublin, Poland: Institute of Agrophysics Polish Academy of Sciences.
Molenda, M., J. Horabik, and I. J. Ross. 1993. Loads in model grain bins as affected by filling
methods. Transactions of ASAE 36(3): 915-919.
Moore, D. W., G. M. White, and I. J. Ross. 1984. Friction of wheat on corrugated metal surfaces.
Transaction of the ASAE 27(6): 1842-1847
Morgan, P.H., L.P. Mercer, and N.N. Flodin. 1975. General model for nutritional responses of
higher organisms. Proc Nat Acad Sci 72(11): 4327-4331.
119
Moysey, E. B. 1984. The effect of grain spreaders on grain friction and bin wall pressures.
Journal of Agricultural Engineering Resources 30: 149-156.
O. J. Loewer, Jr., I. J. Ross, D. D. Kratzer, and J. N. Walker. 1977. Properties of ground shelled
corn as related to forces in bulk storage structures. Transactions of the ASAE 20(1): 155-
156.
Overhults, D.G., G.H. White, H.E. Hamilton, I.J. Ross. 1973. Drying soybeans with heated air.
Transactions of the ASAE 16(1): 112-113.
Pierce, R. O. and G. R. Bodman. 1987. Piling angles of corn and milo. American Society of
Agricultural Engineers Summer Meeting, Baltimore, MD. Paper No. 87-4058. St. Joseph,
MI.
R. E. Clower, I. J. Ross, and G. M. White. 1973. Properties of compressible granular materials as
related to forces in bulk storage structures. Transactions of the ASAE 16(3): 478-481.
Ross, I.J., T.C. Bridges, O.J. Loewer, and J.N. Walker. 1979. Grain bin loads as affected by grain
moisture content and vertical pressure. Transactions of the ASAE 22(3): 592-597.
S. G. McNeill, M. D. Montross, S. A. Thompson, I. J., Ross, and T. C. Bridges. 2008. Technical
Note: Packing Factors of Feed Products in Storage Structures. Applied Engineering in
Smalley, I. J. 1970. Variations on the particle packing theme of Graton and Fraser. Powder
Technology. 4(71): 97-101.
Stahl, B. M. 1950. Grain bin requirements. Circular 835. Washington, D. C.: USDA.
Stephens, L. E., and G. H. Foster. 1976. Grain bulk properties as affected by mechanical grain
spreader. Transactions of the ASAE 19(2): 354-358
Thompson, S. A., and I. J. Ross. 1983. Compressibility and frictional properties coefficients of
wheat. Transactions of the ASAE 26(4): 1171-1176, 1180.
Thompson, S. A., C. V. Schwab and I. J. Ross, 1989. Packing of whole grains in full-sized bins.
ASAE Paper No. 89-4547. St. Joseph, Mich.: ASAE.
Thompson, S. A., S. G. McNeill, I. J. Ross, and T. C. Bridges. 1987. Pack factors of whole
grains in Storage Structures. Applied Engineering in Agriculture 3(2): 215-221.
Thompson, S.A., C.V. Schwab, and I.J. Ross. 1991. Calibration of a model for packing whole
grains. Applied Engineering in Agriculture 7(4): 450-456.
Thompson, S.A., S.G. McNeill, I.J. Ross, and T.C. Bridges. 1990. Computer model for
predicting the packing factors of whole grains in flat storage structures. Applied
Engineering in Agriculture 6(4): 465-470.
USDA Warehouse examiner’s handbook grain Pack Data WS-3. No Date. Warehouse
Examiner’s Handbook. Washington, D.C.: USDA, Agricultural Stabilization and
Conservation Service, Warehouse Division (Now USDA, Farm Service Agency,
Commodity Services, Procurement and Warehouse Branch).
120
USDA. No Date. USDA Warehouse Examiner’s Handbook Grain Pack Data WS-3. Washington,
D.C.: USDA, Agricultural Stabilization and Conservation Service, Warehouse Division
(Now USDA, Farm Service Agency, Commodity Services, Procurement and Warehouse
Branch).
USDA-GIPSA. 2009. Inspecting Grain: Practical Procedures for Grain Handlers. Washington,
D.C.: USDA Grain Inspection, Packers and Stockyards Administration. Federal Grain
Inspection Service. Online available at:
http://www.gipsa.usda.gov/Publications/pub_fgis.html#hb Accessed on October 29,
2013.
USDA-RMA. 2012a. Loss Adjustment Manual (LAM) Standard Handbook: Directive No.
25010-1H. Washington, D.C.: USDA Risk Management Agency. Online available at :
http://www.rma.usda.gov/data/25000/2012.html Accessed on January 28, 2013.
USDA-RMA. 2012b. Small Grains Loss Adjustment Standards Handbook: Directive nN. 25430-
1H. Washington, D.C.: USDA Risk Management Agency. Online available at :
http://www.rma.usda.gov/data/25000/2012.html Accessed on January 28, 2013.
Wassgren, C. R. 2002. Effects of vertical vibration on hopper flows of granular material. Physics
of Fluids 14(10): 3439-3448.
Wheat Marketing Center. 2008. Wheat and Flour Testing Methods: A guide to understanding
wheat and flour quality: version 2. (pp 14-15), Kansas State University, Manhattan, KS.
Online available at: http://www.wheatflourbook.org/doc.aspx?Id=201 Accessed on Jan
23, 2014
121
7. ACKNOWLEDGEMENTS
The research was supported by USDA (CRIS No. 5430-43440-007-00D) and by the Kansas
Agricultural Experiment Station (Contribution No. _________). The assistance provided by
Dennis Tilley (USDA-ARS CGAHR), Dr. Ekramul Haque (KSU), Kevin Hamm (KSU), Chris
Weston (USDA-ARS CGAHR), and Howell Gonzales (KSU) in conducting the field tests are
highly appreciated.
We want to thank Dalton Henry (Kansas Wheat Commission), Dr. Tom Herald (USDA-ARS
CGAHR), Brian Linin and Richard Bauman (Frontier Ag, Inc.), Marvin Schlatter (ADM), Dr.
Danny Rogers, (KSU), Greg Mclure (Riley County Extension Office), Mike Schulte (Oklahoma
Wheat Commission), Ben Boerner (Texas Grain & Feed Association), Ned Bergman and Steve
Becker (Farm Service Agency), and Dr. Charles R. Hurburgh, Jr. (Iowa State University), and
other extension specialists, university professors, and grain researchers for assistance in locating
cooperators. Our thanks to elevator-cooperators with HRW wheat bins: Joey Meibergen and
Montie Walton of W.B. Johnston Terminal Elevator; Joshua Dechant of Frontier Ag, Inc.; and
Doug Biswell of Manhattan Coop, KS.
We would like thank our farmer-cooperators with HRW wheat bins who participated in this
project: Steve Clanton, Jay Cook, Terry Kastens (Kastens, Inc.), Brent Linin, Roger May (May
Family Farms), Bruce Otte (in memoriam), Lee and Margaret Scheufler (Scheufler Farms, Inc),
David Schemm, ND corn bin farmers, Larry Jons and his fellow farmers (IA) and Alan
Townsend.
122
123
View publication stats

(1 Usar Principal) FINALREPORTMod1rb1MEC

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(1 Usar Principal) FINALREPORTMod1rb1MEC

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

DEVELOPING NEW STORED GRAIN PACK FACTORS WITH KNOWN ACCURACY

Research · June 2016

Rumela Bhadra Josephine Boac

SEE PROFILE SEE PROFILE

Samuel Mcneill Michael Montross

SEE PROFILE SEE PROFILE

The user has requested enhancement of the downloaded file.

DEVELOPING NEW STORED GRAIN PACK FACTORS

AGRICULTURAL RESEARCH SERVICE (ARS)

Rumela Bhadra, Josephine Boac, and Ronaldo Maghirang

Samuel McNeill, Michael Montross, and Aaron Turner

Laboratory Compressibility Tests ........................................................................................... vii

Field Measurements of Packing Factor and Model Validation ................................................. ix

Model Development.................................................................................................................. xii

Summary of Conclusions ......................................................................................................... xiv

1.5 Scientific Background ...................................................................................................... 1

1.6 Experimental Design ........................................................................................................ 3

1.6.1 Laboratory Investigations ......................................................................................... 3

1.6.2 Commercial Field Trials ........................................................................................... 3

2. Laboratory Compressibility Measurements ............................................................................ 5

2.1 Compressibility of Individual HRW Wheat Varieties ..................................................... 6

2.2 Composite Samples of HRW Wheat ................................................................................ 6

2.3 Compressibility Results for HRW Wheat ........................................................................ 6

2.4 Models for Laboratory Compressibility Curves............................................................. 18

2.4.1 Potential Model Equations ...................................................................................... 18

2.4.2 Evaluation of Compressibility Relationship at High Pressure ................................ 20

2.4.3 Selection of best Model .......................................................................................... 21

2.4.4 Summary of Model Equations .................................Error! Bookmark not defined.

3. Field Measurement of Compaction Factors .......................................................................... 26

3.1 Experimental Methods ................................................................................................... 26

3.1.1 Density and Compaction Relationships .................................................................. 27

3.1.2 Application of WPACKING Model ....................................................................... 28

3.1.4 Application of FSA-W Method .............................................................................. 30

3.2 HRW Wheat Field Data ................................................................................................. 31

3.2.1 Data Collection ....................................................................................................... 31

3.2.2 Results and Discussion ........................................................................................... 31

3.2.3 Comparison of WPACKING Model, RMA and FSA-W Methods ........................ 35

3.3 Test Weight and Dockage Relationships ....................................................................... 42

3.3.1 Background ............................................................................................................. 42

3.3.2 Materials and Methods ............................................................................................ 44

3.3.3 Results and Discussion ........................................................................................... 45

3.4 Corn Field Data .............................................................................................................. 58

3.4.1 Field Measurements and Data Collection ............................................................... 58

3.4.2 Results and Discussion ........................................................................................... 58

3.5.1 Background ............................................................................................................. 83

3.5.2 Data Collection ....................................................................................................... 84

3.5.3 Results and Discussion ........................................................................................... 85

3.6.1 Background ............................................................................................................. 96

3.6.2 Results and Discussion ........................................................................................... 96

4. Compaction Factor Program ............................................................................................... 100

4.1 Using the Program ........................................................................................................ 100

4.2 "InputValues" Sheet ..................................................................................................... 101

4.2.1 Pre-Data ................................................................................................................ 102

4.2.2 GrainProperties ..................................................................................................... 102

4.2.4 Walltype&GrainHeight ......................................................................................... 104

4.2.5 Deductions ............................................................................................................ 105

4.2.6 GrainSurface ......................................................................................................... 106

4.2.7 HopperProperties .................................................................................................. 106

4.3 "ProductionWorksheet" Sheet ...................................................................................... 108

5. Conclusions ......................................................................................................................... 112

6. References ........................................................................................................................... 117

7. Acknowledgements ............................................................................................................. 122

1 Full Cubic Polynomial 0.749 -244.632 0.878 741 Y= a+bX1+cX2+dX1^2+eX2^2+fX1^3+gX2^3+hX1X2+iX1^2X2+jX1*X2^2

7 Paraboloid Polynomial 0.747 -247.695 0.879 740 Y= a((bX1+c)^2+(d*X2+e)^2)