. u-Iom ~. a.1‘l.n.‘-I1¥

.3 t ‘ " -'" “ 9"

THESIS L.....\....i-.... “a a
t“.

a ("6..A,

A. UTA-1'-‘VIIK "-.l‘¢.~‘B-J b ‘u-.’
w

a 0 ¢ 4‘
‘ f5 in .:“' fi‘ . «L!

-I!!- 'v Viv-‘9'."

 

 

 

This is to certify that the

thesis entitled

HEAT GENERATION AND DRY MATTER LOSS
DURING STORAGE OF RECTANGULARLY BALED ALFALFA HAY

presented by

Dennis R. Buckmaster

has been accepted towards fulfillment
of the requirements for

M.S. degree in A.E.

5: ”AW

“1/
t)

Major professor '

Date Feb. [32/] /§J’é

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

MSU

LIBRARIES

—_

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

 

 

HEAT GENERATION AND DRY MATTER LOSS
DURING STORAGE OF RECTANGULARLY BALED ALFALFA HAY

BY

Dennis R. Buckmaster

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

MASTER OF SCIENCE

Department of Agricultural Engineering

1986

ABSTRACT

HEAT GENERATION AND DRY MATTER LOSS
DURING STORAGE OF RECTANGULARLY BALED ALFALFA HAY
BY

Dennis R. Buckmaster

Alfalfa hay is commonly stored in rectangular bales
at a moisture content below 18 percent (w.b.). To properly
evaluate the benefits of preservatives or other alternative
management schemes used to increase this moisture limit, the
biological process of storage must be thoroughly understood.

Dry matter loss in rectangularly baled alfalfa hay was
empirically modeled as a function of ‘moisture content at
baling. Dry matter loss was increased 0.5 percent for each
percent increase in baling moisture above 11.5 percent.

A finite difference heat transfer model was applied to
stacks of baled alfalfa hay to determine heat generation
rates. Mean heat generation rates over the first thirty days
of storage ranged from 0.0 to 0.243 W/kg of hay material.
Heat generation rate varied as the square of moisture
content and the square root of density with a maximum rate

occurring after approximately 8 days in storage.

ACKNOWLEDGEMENTS

The author wishes to extend his appreciation to the
following persons for their assistance in completing the
research and analyzing the results. To my parents Mr. and
Mrs. Ivan Buckmaster and my wife Corinne for their constant
support and encouragement to attempt such a task. To my
friends, Mr. Randy Davis and Mr. Phil Noakes for their hard
work to collect data, and their encouragement along the way.

To my committee members, Dr. Ajit Srivastava, Dr. Robert
Wilkinson, and Dr. William Thomas, for their input to the
experimental procedures and course selection.

To my major professor, Dr. C. Alan Rotz, for providing
careful guidance and supportive criticism when it was needed.
Thank you for placing confidence in me to do this work and

helping me carry it through.

ii

TABLE OF CONTENTS

LI ST OF TABLES O O O O O I O O O O O O O O O O O O I O C v

LIST OF FIGURES O O O O O O O O O O O O O O O O O O O Vii

NOMENCLATURE O O O O O O O O O O O O O O O O O O O O O i x

1 Q I NTRODUCT I ON 0 O O O O O O O O O O I O O O O O O O O 1

2 Q OBJECT I VES O I O O O O O O O O O O O O O O O O O O O 3

3. LITERATURE REVIEW . . . . . . . . . . . . . . . . . 4
3.1 Storage of Alfalfa Hay . . . . . . . . . . . . . 4

3.2 Dry Matter Loss . . . . . . . . . . . . . . . . 5

3.3 Quality Changes . . . . . . . . . . . . . . . 7

3.4 Thermal Properties . . . . . . . . . . . . . . 8

3.5 Heating in Storage . . . . . . . . . . . . . . 12

3.6 Preservatives . . . . . . . . . . . . . . . . . 13
3.6.1 organic aCids O O O O O O O O O O O O O 14

3 Q 6 O 2 AnhYdrous Amenia O O O O O O O O O O O 15

3.6.3 Other Preservatives . . . . . . . . . . . 16

4. EXPERIMENTAL PROCEDURE . . . . . . . . . . . . . . 18
4.1 Harvesting of Hay Treatments . . . . . . . . . . 18

4.2 Initial sampling 0 O O O O O 0 O O O O O O O O 20

403 Storage 0 O O O 0 O O O O O O O O 0 O O O O I 20

4.4 Final samplin O O O O O O O O O O O O O O O O 21

5. DATA ANALYSIS 0 O O O O O O O O O O O O O O O O O O 23
5.1 Dry Matter Loss . . . . . . . . . . . . . . . . 23

5.2 Temperature . . . . . . . . . . . . . . . . . . 24

iii

TABLE or CONTENTS (cont.)

6 0 EXPERIMENTAL RESULTS 0 C O O O O O O O O O I O O O O 2 6

6.1 Treatments . . . . . . . . . . . . . . . . . . 26
6.2 Dry Matter Loss . . . . . . . . . . . . . . . . 26
6.3 Temperature . . . . . . . . . . . . . . . . . . 36

7. MODEL DEVELOPMENT . . . . . . . . . . . . . . . . . 42
7.1 Dry Matter Loss . . . . . . . . . . . . . . . . 42

7.2 Temperature . . . . . . . . . . . . . . . . . 50

7.3 Heat Generation . . . . . . . . . . . . . . . . 51
7.3.1 Finite Difference Model . . . . . . . . 52

7.3.2 Estimating Thermal Properties . . . . . 57

7.3.3 Estimating Heat Generation Rates . . . . 62

7.3.4 Heat Generation Model . . . . . . . . . 67

8 0 MODEL VAL I DAT I ON C O O O O O C O O O O O O O O O O 8 0

8.1 Dry Matter Loss . . . . . . . . . . . . . . . . 81
8.2 Temperature . . . . . . . . . . . . . . . . . . 86
8.3 Heat Generation . . . . . . . . . . . . . . . . 91
8.4 Model Sensitivity . . . . . . . . . . . . . . . 92

9 O SWRY AND CONCLUSIONS 0 O O O O O O O O O O O O O 97

10 0 REFERENCES 0 O O O O O O O O O O O I O O O O O O 10 1

iv

Table 3.1

Table 4.1

Table 6.1

Table 6.2

Table 6.3

Table 6.4

Table 6.5

Table 6.6

7.1

Table

Table 7.2

Effect

Target

LIST OF TABLES

of moisture
alfalfa hay on quality parameters.

content

treatments of moisture

of baled

content

and density desired for the hay storage

experiments.

Treatments

with

varying

moisture

contents and densities obtained in three

experiments.

Dry matter loss,
and heating for 5 bale

eratures

maximum and mean temp-

stacks

of alfalfa baled at varying moisture and

density levels (Experiment 1).

Dry matter loss,
and heating for 5 bale

eratures

maximum and mean temp-

stacks

of alfalfa baled at varying moisture and

density levels (Experiment 2).

Dry matter loss,
and heating for 5 bale

eratures

maximum and mean temp-

stacks

of alfalfa baled at varying moisture and

density levels (Experiment 3).

Pearson product moment correlation coef-

ficients for several

storage parameters

of non-chemically treated alfalfa hay.

Pearson product moment correlation coef-

ficients for several

storage parameters

of untreated and acid treated alfalfa

hay.

Regression models of dry matter
temperature,

maximum temperature,

and heating in degree days as

mean

loss

functions

of baling moisture and initial density.

Comparison
spec1f1c

of

known

and estimated

heat values of tobacco to

estimated specific heat values of hay.

I

19

27

29

30

31

33

34

49

61

LIST OF TABLES (cont.)

Table 7.3 Comparison of known and estimated

thermal conductivity values of
granulated cork to estimated thermal
conductivity values of baled hay. . . .

Table 7.4 Mean heat generation rates for alfalfa
baled at varying moisture and density
levels (Experiment 1). . . . . . . . .

Table 7.5 Mean heat generation rates for alfalfa
baled at varying moisture and density
levels (Experiment 2). . . . . . . . .

Table 7.6 Mean heat generation rates for alfalfa
baled at varying moisture and density
levels (Experiment 3). . . . . . . . .

Table 8.1 Data used for validation of models which

predict dry matter loss and storage
temperatures. 0 O O O O O O O O O O O 0

vi

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

LIST OF FIGURES

Stacking
experiments. . . . . . .

procedure for

storage

Dry matter loss vs baling moisture

for small stacks of rectangularly baled

alfalfa hay (Experimental data).

Maximum storage temperature vs baling
moisture for small stacks of
rectangularly baled alfalfa hay
(Experimental data). . . . . . . . . .
Mean storage temperature vs baling
moisture for small stacks of
rectangularly baled alfalfa hay
(Experimental data). . . . . . . . . .
Heating in degree days vs baling
moisture for small stacks of
rectangularly baled alfalfa hay

(Experimental data). . .

Dr matter loss as a function of baling
baled

m015ture for
alfalfa hay. . . . . .

Dry matter
maximum storage

rectangularly

rectangularly baled alfalfa hay.

loss as a function of
temperature

of

Dry matter loss as a function of mean
rectangularly

storage temperature of
baled alfalfa hay. . .

Thermocouple positions in
stack of five bales. . .

Finite difference model of
hay stack. . . . . . . .

Variation in moisture
time. 0 O I O O O O O 0

Heat generation rates of
baled alfalfa hay vs time.

vii

a

treatment

a five bale

content over

rectangularly

22

35

37

39

40

44

46

47

53

54

63

72

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

7.8

8.2

8.3

8.5

8.6

8.7

8.8

LIST or FIGURES (cont.)

Moisture and density effects on heat
generation rate in rectangularly baled
alfalfa hay. O O O O O O O O O O O I

Validation of a model which predicts
dry matter loss as a function of
moisture content at baling. . . . . .

Validation of a model which predicts
dry matter loss as a function of
maximum temperature reached in storage.

Validation of a model which predicts
dry matter loss as a function of mean
temperature during the first 30 days in
storage. . . . . . . . . . . . . . .

Validation of a model which predicts
maximum temperature as a function of
moisture content at baling and initial
density. . . . . . . . . . . . . . .

Validation of a model which predicts
mean temperature as a function of
moisture content at baling and initial
density. . . . . . . . . . . . . . .

Validation of a model which predicts
heating in degree days as a function of
moisture content at baling and initial
denSitYO O O O O O O O O O O O O O O

Hay temperature vs time in a small
stack of rectangularly baled alfalfa
hay (model validation). . . . . . . .

Predicted hay temperatures vs time in a

large stack of rectangularly baled
alfalfa hay. . . . . . . . . . . . .

viii

75

82

84

85

87

89

90

93

95

NOMENCLATURE

Symbol Definition Un1ts
A thermal diffusivity mZ/s
B simplifying constant (A*DT/K) m3°C/W
C specific heat J/KgoCC
Ca specific heat KJ/Kgoc
Cb specific heat Btu/lboF
Cc specific heat cal/gmo C
D density --
Dd dry matter density Kg/mg
D wet density Kg/m
DYw wet density at time of baling Kg/m3
DD heating in degree days >35o C °C*day
DT time increment s
DX grid increment m
DML dry matter loss (% of initial) %
F Fourier modulus -—
FDM final dry matter Kg
G heat generation rate W/m3
G2 heat generation rate W/Kg
vap heat required to evaporate water KJ
HI horizontal surface convective heat
transfer coefficient W/m2 oC
H2 vertical surface convective heat
transfer coefficient W/m2 oC
IDM initial dry matter Kg
K thermal conductivity (W/mOC)
L dry matter lost Kg
M m01sture content (wet basis) decimal
MI moisture content at time of
baling (wet basis) decimal
m moisture content (wet basis) %
M30 moisture content 30 days after
baling decimal
M60 moisture content 60 days after
baling decimal
P application rate of propionic
acid (% of wet weight) %
Qnet heat leaving hay stack KJ
r Pearson Product Moment correlation
coefficient --
R multiple correlation coefficient
t time from baling days
Tmax maximum hay temperature in storage °C

ix

NOMENCLATURE (cont.)

Symbol Definition Un1ts
Tmean mean temperature during the first
30 days of storage
Ta ambient temperature °C
Ti j p temperature at: °C
' ' x node = i

.y node = j

time = p*DT
THP total heat production Kg
V volume m
x density 1b5ft3
Y thermal diffusivity ft /h

1. INTRODUCTION

The ideal alfalfa hay handling system would 1) allow for
convenient crop handling, 2) be inexpensive, 3) not allow
material or dry matter loss, and 4) not allow quality
deterioration. Numerous methods of storing alfalfa hay do
exist, but the ideal storage system does not now exist.
Stacks, round and rectangular bales of all sizes, pellets,
and high density cubes are all used. In an effort to find an
optimum storage method, much research has been conducted in
the area of alfalfa hay storage systems. Because thermal and
physical properties may vary significantly within a unit of
stored hay, the study of changes during storage is far from
an exact science.

Alfalfa hay storage research is usually conducted as a
simultaneous comparison of two or more storage methods. As
examples: 1) inside vs. outside storage, 2) stacks vs.
rectangular bales, or 3) chemically treated bales vs. non-
treated bales. Primary considerations in comparing such
treatments have been dry matter loss and quality changes
during the storage period. Because this type of research is
usually performed as a comparison test, the results are
simply comparisons of two or more methods. Conclusions drawn
from experiments conducted in this manner are limited to
the experimental conditions, i.e., a given moisture level and
density, or fixed environmental conditions. In order for the

results to be more applicable, models describing the changes

in each storage method should be developed. Models of the
storage process which accurately simulate the real situation
would be valuable tools to use when evaluating alternative
methods for harvesting and storing alfalfa hay.

‘ One must remember that models are decision aids, not
decision makers. Decision aids in the area of storing
alfalfa hay would suggest correct answers to questions like:
"Will change occur in the hay during the storage period?";
"Is the change beneficial?"; "Will any deterioration
occur?"; or "Will the stored hay heat enough to cause a barn
fire?". If the answers to all questions were a clear "yes" or
"no", models would be unnecessary. It is the fact that the
answers are "sometimes yes" and "sometimes no” that provides
the motivation to describe the storage process. Useful
models will not only indicate yes or no answers to such
questions, but will also give quantitative information.

The research work of this study is not a comparison of
storage systems or chemical treatments. It is, rather, an
in-depth look at storage of alfalfa hay in standard
rectangular bales. Quantity of material taken out of storage
and heating during storage as affected by moisture level and
density at the time of baling are discussed and appropriate

models are developed.

2. OBJECTIVES

The objectives of the research were to describe the

changes

during

1.

which occur to alfalfa hay in rectangular bales

storage. Specific objectives were:

To develop an empirical model which predicts dry
matter loss during storage as a function of initial
moisture content and density of the hay as it
enters storage.

To develop an empirical model which predicts the
heat generation rate of baled alfalfa while in
storage as a function of moisture and density
levels.

To model the heat transfer process throughout a
stack of hay based upon assumed physical and
thermal properties in order to apply information
obtained from small hay stacks to stacks of any

size and shape.

3. LITERATURE REVIEW

3.1 STORAGE 95 ALFALFA HAY

There are many ways to store alfalfa hay. This
discussion concerns only baled alfalfa hay stored in a barn
in the conventional manner, i.e., without refrigeration or
forced ventilation. For alfalfa stored in this manner, the
term "safe storage" implies: 1) little heating of the stored
hay, 2) no molding, and 3) no degradation of nutrients in the
hay during the storage period. Safe storage of baled alfalfa
is normally assumed if the baling moisture is lower than 20%;
however, Hall (1980) reported that safe storage for 200 days
requires a maximum of 15% moisturel.

When alfalfa is cut, it contains 70 to 80% water. It
can easily take up to 4 or 5 days for the hay to dry down to
15 - 18% moisture in the field. During this field curing,
considerable respiration and leaching losses can occur. Hay
which dries slowly or becomes rewetted can have considerable
microbial growth on it causing nutrient losses. Mechanical
handling of dry hay also leads to considerable losses
(Savoie, et a1., 1982). Raising the baling moisture
decreases leaching, microbial and mechanical losses in the
field; however, hay baled too wet will heat severely,
causing of nutrients. Spontaneous combustion can occur with

1 All moisture levels in this thesis are percent wet basis
unless otherwise noted.

even more severe consequences (Hoffman and Bradshaw, 1937;

Bohstedt, 1944).

In an effort to increase the safe baling moisture limit,
preservatives such as salts, organic acids, anhydrous

ammonia, urea, and bacterial inoculants have been used with

varying degrees of success.

3.2 DRY MATTER LOSS

 

Baled alfalfa decreases in weight during storage due to
loss of moisture and loss of dry matter. Baled hay
approaches 14 - 15 % moisture in storage. When it reaches
moisture equilibrium with the environmental conditions, no
more moisture weight loss will occur. Dry matter loss in
storage is due to continued reSpiration and microbial
activity which may occur when there is sufficient moisture in
the environment for this activity.

Most researchers report a correlation between baling
moisture and dry matter loss (Rotz, et al., 1984; Nelson
1966; Nelson, 1968; Nelson, 1972; Jorgensen, et al.,l978);
however, no models for predicting dry matter loss have been
proposed. Martin (1980) suggested hay may lose 5 - 10 % dry
matter if baled with less than 20% moisture. Jorgensen, et
al., (1978) reported that nontreated hay baled at over 20%
moisture resulted in 14% dry matter loss.

Dry matter loss of baled alfalfa hay is a function of

several factors such as baling moisture, maturity, bale

density, and the type of storage facility. For a fixed
storage condition, the primary factor was moisture and the
secondary factor was maturity. Density was reported to have
no effect on dry matter loss (Nelson, 1966, 1968).

Storage loss data for non-chemically treated baled hay
from several researchers (Martin, 1980; Koegel, et al., 1983;
Shepherd, et al., 1966) was compiled, and a simple linear
regression model was developed from the data. Although
conditions were not identical for each researcher, a good
correlation between baling moisture and dry matter loss was
obtained. Fifteen (15) data points were used (3 remote points

were removed) to develop this relationship:

DML = 77.0*MI - 10.71 (3.1)
(r2 = 0.93 std. error = 1.3)
Where:

DML = dry matter loss (% of initial)
MI = moisture content at baling (decimal wet basis)
Because data from several tests were combined to get
this relationship, it should not be taken as an accurate
indicator, rather as a motivator for study in this area.
Waldo and Jorgensen (1981) suggested the following rule
of thumb: 1% loss in dry matter for each 1% decrease in
moisture content during storage. Since hay usually
approaches 15% moisture in storage, this rule indicates a 5%
loss at 20% moisture, 10% loss at 25% moisture, etc.
Equation (3.1) indicates nearly a 0.8% loss in dry matter for

each percentage point increase in baling moisture. With the

reasonable assumption that all hay approaches the same
moisture level in storage, equation (3.1) is in reasonable

agreement with that rule of thumb.

3.3 QUALITY CHANGES

If hay is baled at a low moisture level and stored
inside, few nutrient changes occur during storage (Moser,
1980). Weeks, et al.(l975) reported little chemical change
in loosely stacked hay harvested with up to 40% moisture.
However, other research indicates that as hay is baled with
moisture levels exceeding 20% and normal density levels, the
heat and mold occurring do affect nutrient retention (Miller
et al., 1967). Several researchers have reported significant
quality changes during storage as baling moisture was
increased (Jorgensen, et al., 1978; Miller, et al., 1967;
Nehrir, et al., 1978; Nelson, 1966; Nelson, 1968).

Miller, et a1. (1967) listed the effects of baling
moisture on several quality properties of baled alfalfa hay.
This information is summarized in Table 3.1

Nelson (1968) published numerous graphs of the effect of
moisture level on nutrient retention in non-chemically
treated high density bales. Retention of all chemical
constituents measured was significantly decreased by
increasing baling moisture. Maturity significantly affected
retention of carbohydrates, organic matter, crude fat, and

dry matter. Maturity did not significantly affect retention

of crude protein, crude fiber, or nitrogen free extract.

Table 3.1 Effect of moisture content of baled alfalfa hay on
quality parameters (Miller et al., 1980. ).

Property Effect of Increased Bale Mo1sture
Crude Protein Content no effect
Ash Content increased
Cell Wall Constituents increased
Cellulose Content increased
Acid Detergent Fiber Content increased
Lignin Content increased
Water Soluble Carbohydrates no effect
Dry Matter Digestibility decreased
Crude Protein Digestibility decreased
Digestibility of Water

Soluble Carbohydrates decreased
Gross Energy decreased

3.4 THERMAL PROPERTIES

 

 

The thermal properties of baled hay are difficult to
estimate because hay is porous, contains varying amounts of
water and may be composed of different types of hay
materials. Some work has been done to estimate thermal
conductivity, specific' heat and thermal diffusivity for
alfalfa silage (Jiang, et al., 1985), but this material is
very different from baled hay. Jiang, et al., (1985)
evaluated thermal properties for chopped hay varying in

moisture content from 50 to 80% and varying in wet density

from 400 to 800 Kg/m3. Hay stored in the form of bales is
not chopped, and varies from approximately 12 to 27% in
moisture and 100 to 250 Kg/m3 in wet densityz. Although the
results found by Jiang, et al. (1985) should not be used in
baled hay applications they, are listed here for comparison.
Specific heat, thermal conductivity, and thermal diffusivity
equations obtained through regression procedures for haylage

type materials were as follows:

ca = 2.2573 - 0.003237*Dw + 0.0001197*Dw*m (3.2)
A = 0.1329 - 9.22(1o)’5*nw + 0.6(lO)-7*Dw2 - (3.3)
1.oa(1o)‘5*m2
x = 0.2236 - o.ooo3o74*ow - 0.001061*m +
0.00000816*Dw*m (3.4)
Where:

m = moisture (% wet basis)

A = thermal diffusivity (mz/sec)
C3 = specific heat (KJ/KgOC)

K = thermal conductivisy (W/m°C)
Dw a wet density (Kg/m )

Mohsenin (1980) discusses procedures for evaluating
thermal ‘properties and gives results from research done to
evaluate thermal properties. A relationship for thermal
diffusivity of baled hay as a function of density was

presented by Ott and Horbut (1964). For a given moisture

level of 8.2%, the following relationship was found:

2 Wet density refers to density as is (wet basis). Dry
density (or dry matter density) refers to the equivalent
density if the material contained 0% moisture.

10

Where:

Y
X

thermal diffusivity (FtZ/h)
density (lb/ft )

Siebel (1892) proposed two equations for specific heat
for food materials, one based on temperatures above freezing,
the other based on temperatures below freezing. For

temperatures above freezing, the specific heat equation was:

Cb = 0.008*m + 0.20 (3.6)
Where:
Cb = specific heat (Btu/lboF)
0.2 = assumed specific heat of the dry solid
For materials with high moisture contents, Siebel's
equation gives a reasonable estimate of specific heat;
however, for low moisture material, more error occurs. Bern
(1964) conducted experiments to evaluate the specific heat of
ground alfalfa. As shown in Mohsenin (1980), Bern's results
indicated a good correlation between moisture content and
specific heat. No equation is given, but estimating from a
given figure (pg. 49, Mohsenin, 1980), equation (3.7) is
approximately true for moisture levels between 4 and 20

percent wet basis.

cc = 0.22 + 0.0142 . m (3.7)

Where:

Cc a specific heat (cal/gm°C)

ll

Conversion of Siebel's equation (3.6) into the units of

equation (3.7) results in equation (3.8):

CC = 0.199 + 0.00797 * m (3.8)

Equations (3.7) and (3.8) are in reasonable agreement
for predicting specific heat. In order to use either
equation, we need to consider the hay/air mixture to be one
solid with water as the second material. This is a
reasonable assumption since the baled hay is dense enough
that natural convective currents within the stored material
would be minimal. For forced ventilation drying models, this
assumption would need to be modified.

Thermal conductivity of baled alfalfa has not been
measured. A form of predicting thermal conductivity Of wet
solids is given by Andersen (1950). It is similar to

Siebel's equation for specific heat:

x = M*Kwater + (l—M)*Ksolid (3.9)
Where:
K = thermal conductivity of the wet hay/air mixture
(W/mOC)
Kwater = thermal conductivity of water (W/mOC)
Ksolid = thermal conductivity of the dry hay/air
mixture (W/mOC)
In order to use this equation for baled hay, the thermal
conductivity of dry baled hay is needed. Again, as in the
method for predicting specific heat, the hay/air mixture

should be considered as one solid with water as the second

material.

3.5 HEATING TE STORAGE

As moisture content increases when hay is baled, heat
development during storage increases. Temperatures of stored
hay up to 50°C (122°F) do not significantly affect quality,
but as the bale temperatures exceed 60°C (140°F), feed value
is decreased. Rotz, et al. (1983) reported the following
linear regression equation for maximum bale temperature (°C)
versus baling moisture for untreated hay in small (10 bale)

stacks:

T = 4.38 + (1.38 * m) (3.10)

max
(r = 0.90)

Equation (3.10) implies that in order to keep
temperatures below 50°C (122°F), the hay must be less than
33% in moisture. This figure should be used very
conservatively because it pertains to a very small stack.
Large stacks are known to attain temperatures above 50°C
(122°F) even though baling moisture may be less than 33%.
Heat generated within the hay cannot be dissipated as rapidly
from large stacks as from smaller stacks. Also, Jorgensen
et, a1. (1978) suggested not baling at over 30% moisture
because of shrink and stack movement. With few exceptions,
(e.g. Koegel, et al., 1983) baled alfalfa over the 30%
moisture level cannot be preserved adequately with any form
of preservation.

Several researchers have published time/temperature

13

curves for hay bales in storage (Nelson, 1968; Nelson 1966;
Weeks, et al., 1975; Hathaway, et al., 1984; Koegel, et al.,
1983; Miller, et al., 1967). Nelson (1966 ,1968, 1972) gives
curves for degree days of heating for varying moisture and
density levels. This indicates a total amount of heating,
but does not indicate when or how fast this internal heat
generation occurs. Models predicting heat generation rates

for baled alfalfa hay in storage have not yet been presented.
3 . 6 PRESERVAT IVES

Preservatives are used in baled alfalfa to raise the
upper limit on the safe baling moisture. Effects of
preservatives are various, but the ideal preservative should:

1. Increase nutrient retention and perhaps add
nutr1ents.
. Increase dry matter retention.

Suppress temperature rises.

Inhibit mold development.

. Be cost effective.

m 01 IF DJ M
0

Be safe and easy to apply.

14

3.6.1 ORGANIC ACIDS

 

Organic acids, primarily propionic3 or its salts, are
they most commonly used preservatives in baled alfalfa. The
effect of propionic acid on dry matter retention has been
debated. Several researchers have reported improved dry
matter retention in acid treated hays (Davies and Warboys,
1978; Jorgensen, et al., 1978; Nehrir, et al., 1978).
Johnson and McCormick (1976) treated hay with Hay Savor4 (a
commercial product) which did not affect dry matter
retention. Davies and Warboys (1978) reported that acid
treated hay dried to a lower level during storage. The
advantage of this may be a longer allowable storage period.

Jorgensen, et al. (1978) reported effective hay
preservation with moisture levels to 30-35% with treatments
of l) propionic acid, 2) Chemstor (a commercial product), and
3) propionic acid plus formaldehyde. Acid treatments reduced
heating and molding; however, in vitro dry matter
digestibility was lower for treated hay than for the dry
control hay. Davies and Warboys (1978) reported improved
nutrient retention due to treatment in only one experiment.
Nutrient retention was not improved by Hay Savor (Johnson and

McCormick, 1976).

3 Also known as propanoic acid (CH3CH2COOH).

4 Trade names are used solely to provide specific
information. Mention of a trade name does not constitute
a warranty of the product by Michigan State University,
nor an endorsement of the product to the exclusion of
other products not mentioned.

15
Suggested application rates for propionic acid are: 1,
1.5, and 2% of hay mass for 20-25, 25-30, and 30—35% moisture

hay, respectively (Schaeffer and Martin, 1979).

3 . 6 . 2 ANHYDROUS AMMONIA

 

Anhydrous ammonia (NH3) can be successfully used as a
preservative for high moisture hay. Applied at a rate of 1%
of dry matter, anhydrous ammonia preserved alfalfa hay with
up to 33% moisture (Knapp, et al., 1975). Koegel, et al.
(1983) successfully preserved alfalfa at the 50% moisture
level with ammonia treatment at a rate of 2.5% (wet basis).

Most often, application of anhydrous ammonia is done in
storage. Bales are first wrapped in plastic, then anhydrous
ammonia is slowly released into the hay. Some
investigations have been performed by injecting ammonia into
the bales prior to placement into storage (Koegel, et al.,
1983; Rotz, et al., 1984): however, wrapping the hay is still
necessary to prevent the ammonia from escaping (Hathaway, et
al., 1984).

Treating wet alfalfa with anhydrous ammonia
significantly reduces dry matter loss (Knapp, et al., 1975).
Rotz, et al. (1984) reported dry matter loss for ammoniated
(22.1-32.l% moisture) hay to be similar to that of non-
chemically treated dry (12.5-15.8% moisture) hay.

Quality improvement of alfalfa hay treated with

anhydrous ammonia is mainly an increase of crude protein

16

content. This is due to the presence of additional nitrogen
which can be utilized by rumen bacteria. Ammonia treatment
also inhibits mold development, (Koegel, et al., 1983; Knapp,
et al., 1975) improves physical appearance, (Rotz, et al.,
1984; Koegel., et al., 1983) and suppresses temperature rises
after the initial heat of solution (Hathaway, et al., 1984;
Rotz, et al., 1984; Knapp, et al., 1975).

For a preservative to be effective, it must stop
respiration (i.e., C02 production) in the harvested forage.
Hathaway, et al. (1984) reported that 1840 ppm of ammonia gas
is necessary to inhibit carbon dioxide production.

Anhydrous ammonia is not widely used as a hay
preservative. The primary reason is safety. Anhydrous
ammonia can cause severe burns and can be extremely irritable
to the eyes and skin. Handling of anhydrous ammonia and
application equipment must be done with proper precautions
and safety equipment such as gas masks and rubber gloves.
However, treated hay may be safely handled after the ammonia
has been absorbed by the moisture in the hay (Rotz et a1,
1984). Ammonia treatment can also cause toxicity to animals
when application rates exceed 3.0% of dry matter (Rotz et

al., 1984).

3.6.3 OTHER PRESERVATIVES

In addition to organic acids and anhydrous ammonia,

salts, bacterial inoculants, and urea have been the subject

 

17

of some research as preservatives for baled hay.

Using sodium chloride as a preservative is not a new
idea. It was used before the invention of the refrigerator
to cure meats. However, as a hay preservative, it has not
been as successful. To be effective, salt must be applied at
a rate of l to 2% of the hay weight. At this rate, feeding
problems have been experienced (Moser 1980).

Bacterial inoculants have been tried as preservatives
for baled hay, but not successfully (Rotz, et al., 1983).
Inoculants are added to ensiled products to improve
fermentation and promote fermentation at lower temperatures.
Usually inoculants help the lactic acid producing bacteria
gain control of the preservation over the spoilage bacteria.
Since fermentation in baled alfalfa is not desirable,
inoculants which aid fermentation will probably not act as
preservatives in baled alfalfa.' Also, effectiveness of
inoculants is dependent upon the moisture of the forage.
Moisture levels in baled alfalfa are usually too low for
effective growth of inoculating bacteria and preservation by
inoculants.

Urea, like bacterial inoculants, is better suited to
silages. It has been used for several years in corn silage
in order to increase the non-protein nitrogen (NPN) level. It
has been applied in granular form to baled hay, however,

preservation was not improved (Rotz, et al., 1983).

4. EXPERIMENTAL PROCEDURE

4.1 HARVESTING g: HAY TREATMENTS

 

Three experiments were conducted; one each from first,
second and third cutting alfalfa. In each experiment, the
same basic procedure was followed.

The standing crop was cut when between 10 and 50% bloom.
It was mown with a 2.7 m wide mower-conditioner with a
cutterbar and intermeshing rubber rolls for conditioning.
The alfalfa was laid into a full width swath approximately
2.1 m wide for faster and more uniform drying. Sufficient
hay was mown for 7 to 8 bales per treatment. Hay was then
tedded and raked at different moisture levels so drying would
take place at different rates. The moisture levels at which
the swaths were handled depended upon the weather conditions.
Some treatments required no tedding, while others were tedded
then raked twice. Handling the swath in this manner allowed
for baling of different moisture levels at nearly the same
time.

The target treatments were all possible combinations of
6 moisture levels and 2 density levels. Target moisture
levels were 45, 40, 35, 30, 25, 20, and 15% wet basis.
Density levels were set somewhat arbitrarily, one being high
density (>10Kg/m3), the other being low density (<8 Kg/m3).
Two treatments per experiment were treated with propionic

acid (Table 4.1). In test three, the six driest target

18

 

19

treatments were not baled due to lack of hay and poor
weather. The term treatment is used in this context as
'Niifferent bale conditions”, not as "chemical treatment” or

“handling pract ice" .

Table 4.1 Target treatments of moisture content and

density desired for the hay storage
experiments.
Moisture Low Density High Density
(% w.b.) (<8 kg/m ) (>10 kg/m )
40 X X
35 X X
30 X x
30a x
25 X X
25b x
20 X X
15 X X

a With propionic acid applied at 2.0 % of hay mass.
b With propionic acid applied at 1.5 % of hay mass.

When the windrow moisture level was near a target
moisture level, 7 or 8 bales of the treatment were baled with
a E"Sperry New Holland model 310 baler. No special features or
Ct‘éirlges were used to form the bales; however, the baler did
have a hydraulic bale tensioner rather than the 1 standard
spt‘ing type. Bale density was varied by adjusting the
preezsure to the hydraulic tensioner on the baler. Low density
“Elsa. (achieved by setting the gage pressure to 0 kPa (0 psig).
and high density was achieved by setting the gage pressure to

approximately 1750 kPa (250 psig). For each treatment, five

20

bales of consistent moisture and density were placed into

storage.

4 . 2 INITIAL SAMPLING

After the bales were formed, core samples of 2.5 cm by

approximately 40 cm were taken with a Penn State core

sampler. For determination of moisture content, three bales

per treatment were cored at least once each from an end.

These samples were dried in a 60°C oven for 2 to 3 days. The
moisture levels given by these three samples were averaged to

give a moisture level for the treatment. The mean size of the

samples used for initial moisture content determination was

38 . 8 grams.

After the core sampling was completed, each bale was

WEighed, and the length measured for determination of density

and dry matter content. Dry matter was computed using the

aVex-age moisture level for the treatment and individual bale

Weights. Density was computed using the bale weight and

dimensions given by the bale length and cross sectional area

°f the baler chamber (36.8 cm x 45.7 cm).

‘ - 3 STORAGE

Following all initial sampling, the treatment sets of

£i-"e bales each were placed side by Side in stacks for

stOrage inside a barn. No forced ventilation or auxiliary

21

heating was used. Stacking procedure is illustrated in
Figure 4.1. Styrofoam sheets of 5.0 cm thickness with a
thermal conductivity (1:) value of 0.0267W/mOK were placed
between each treatment to help isolate the treatments.

During the first 30 days of storage, ambient and bale
temperatures were recorded every 6 hours by a Campbell, model
CR5, data logger. Three thermocouples were placed per
treatment in select bales. They were located one each in the
three center bales of each treatment stack. The
thermocouples were placed in the bored hole created by the
core sample taken for moisture content determination (see
section 4.2). The approximate thermocouple position within
the bale was 30 cm from the end, on the centerline of the

bale. The bored holes were plugged after the thermocouples

we re in place .

4 - 4 FINAL SAMPLING

Because temperatures stabilized after approximately 30
631's in storage, any change or loss occurring in stored hay
was assumed to occur during the first 30 days. Final
San'lpling was done 60 days after baling to assure that all
treatments had stabilized.

Final sampling was performed in the same manner as the
i“itial sampling; however, bale lengths were not measured.

The mean size of the samples used for final (60 day) moisture

°°ntent determination was 40.0 grams.

22

Styrofoam insulation

 

 

 

 

 

Thermocouple
' "r%" positions
0

 

 

 

 

 

7

.92 m

_/

 

 

 

 

 

 

 

.05 m -—-o L—o46m—4J

Figure 4.1 Stacking procedure for storage experiments.

5. DATA ANALYSIS

All data was separated into two sets for analysis. One
set included data collected from all treatments described in
Table 4.1. The other set was only data collected from
treatments which were not treated with propionic acid.
Temperatures and dry matter loss values for these treatments
were generally more consistent because effects of the acid

treatment were removed.

5.1 DRY MATTER LOSS

 

Moisture contents and bale weights measured initially
and after 60 days in storage were used to estimate the total
amount of dry matter in each bale at these times. Dry matter
loss was the difference between initial and 60 day dry matter

values, expressed as a percent of the initial dry matter:

DML = 100*(IDM - FDM)/IDM (5.1)
Where:
DML = dry matter loss as a percent of initial dry
matter
IDM = initial dry matter
FDM = final dry matter

Dry matter loss values were calculated for each bale.
The dry matter loss value for a treatment was the mean of
the five dry matter loss values from the five bales of a
treatment. For each of the three experiments, one-way

analysis of variance was used to determine treatment effects

23

24

on dry matter loss. Duncan's Multiple Range Test was used to
determine which treatments within an experiment had
significantly different dry matter loss values.

Results from all experiments were combined so a two-way
analysis of variance could be used for dry matter loss
analysis. Moisture contents were divided into 5 ranges and
density was divided into 3 ranges. Dry matter loss was
broken down by moisture and density to determine significance
of each.

Data collected from the three experiments was also
combined for a correlation analysis. Pearson's product
moment correlation coefficients were used to determine which
variables were correlated. Correlations between dry matter
loss and other storage variables such as baling moisture and

density were observed.

5.2 TEMPERATURE

 

Individual bale temperature data was collected once
every six hours (see section 4.3). These temperatures were
averaged over a 24 hour period to give a mean daily
temperature. Since there were three thermocouples per
treatment, there were three mean daily temperature values per
treatment.

The mean daily temperatures were compared over the first
30 days of storage to find the maximum temperature reached by

each treatment stack. Thirty days of temperature data with

thre
tam;
tea;
var
rep
tern

dai

25

three thermocouples per treatment gave 90 mean daily
temperature values per treatment. These mean daily
temperatures were first analyzed using two-way analysis of
variance and a breakdown of temperature by treatment by
repetition to assure that there were no noticeable errors in
temperature measurements within a treatment. The 90 mean
daily temperature values for each treatment were averaged to
give a mean treatment temperature for the first 30 days in
storage. One-way analysis of variance was used to determine
treatment effects on mean temperature. Duncan's Multiple
Range Test was used to determine treatments within an
experiment which had significantly different mean
temperatures.

Days for which the hay stack temperature exceeded 35°C
were also summed to calculate the heating in the form of
degree days.

Correlations of temperatures and heating to moisture,
density, and dry matter loss were determined with the data
collected from the three experiments combined. Pearson's
product moment correlation coefficients were used to

determine which variables were significantly correlated.

5. EXPERIMENTAL RESULTS
6.1 TREATMENTS

As discussed in Experimental Procedure (section 4.1),
several target treatments were desired. When the treatments
were harvested, not all target treatments were represented.
Yet, others were obtained more than once per experiment.
Table 6.1 summarizes the actual treatments obtained.
Treatments in experiment 1 were as close to the target
treatments as could be expected. The second experiment did
not have the higher moisture levels and experiment 3 did not
have the lower moisture levels; however, a good range of
moisture and density levels was achieved by the combination

of the three experiments.
6.2 DRY MATTER LOSS

Dry matter loss measurements in the three experiments
ranged from 0.6 to 19.5%. Experiments 1 and 2 had a wider
range of moisture levels (Table 6.1) and thus, had a wider
range of resulting dry matter loss values. Dry matter loss
values from experiment 2 were lower because the target
treatments with higher moisture levels were not reached. The
mean standard deviation of the dry matter loss values for a
given treatment within experiments 1, 2, and 3 were 6.3, 3.8,

and 4.3 respectively. This indicates that dry matter loss

26

27

Table 6.1 Treatments with varying moisture contents and
densities obtained in three experiments.

Experiment Treatment Moisture Densig
Number Label (% w.b.) (Kg/m )
1 101 11.5 111
1 102 14.6 175
1 103 16.9 75
1 104 14.3 172
1 105 24.2 91
1 106 25.4 236
1 107 27.7 106
1 108 31.0 268
1 109 30.4 128
1 110 35.2 289
1 111 48.0 189
1 112 43.0 302
1 113 19.3a 233
1 114 27.2b 252
2 201 16.7 74
2 202 16.2 199
2 203 18.3 87
2 204 16.9 175
2 205 18.3 100
2 206 17.3 191
2 207 18.4 100
2 208 20.8 202
2 209 24.5 111
2 210 27.0 225
2 211 32.7 130
2 212 32.2 273
2 213 21.7a 228
2 214 30.2c 250
3 301 23.4 101
3 302 24.0 252
3 303 30.5 175
3 304 30.5 295
3 305 34.7 172
3 306 36.9 287
3 307 35.0 164
3 308 36.2 300
3 309 29.0b 244

of wet weight) propionic acid.
of wet weight) propionic acid.
of wet weight) propionic acid.

a Treated with 1.2
b Treated with 1.6
Treated with 1.7

dFdeP
AAA

28

was measured more accurately in the latter two experiments.

Hay which has been baled wet provides a damp environment
which is conducive to microbial growth. The microbes consume
hay dry matter and in turn, generate heat from their
activity. Respiration rate is also higher in wet hay than in
dry hay. Since both microbial activity and respiration are
causes of dry matter disappearance, a positive correlation
between moisture and dry matter loss was expected.

Results of one-way analyses of variance performed
separately for each of the three experiments indicated that
treatment effects of moisture and density on dry matter loss
were very significant. Two-way analysis of variance was used
to determine if moisture, density, or an interaction between
moisture and density were significant factors. Moisture
levels were divided into 5 ranges and density levels were
divided into 3 ranges. Significance levels of moisture,
density and interaction of the two were p=.01, p=.10, and
p=.02 respectively. From these results, moisture is the most
important factor affecting dry matter loss.

The dry matter loss values corresponding to given bale
conditions are included in Tables 6.2, 6.3, and 6.4 along
with statistical evaluations. Dry matter loss was

consistently increased with an increase in baling moisture.

M:

llllfl‘lll‘t.

29

Table 6.2 Dry matter loss, maximum and mean temperatures,
and heating for five-bale stacks of alfalfa baled
at varying moisture and density levels
(Experiment 1).

Baling Baled Dry Matter Temperatures2 Heating3
Moisture Density Loss Maximum Mean Deg.Days
(% w.b.) (Kg/cu m) (%) (°C) (°C)

11.5 111 2.1ab 25.9 18.6a o
14.6 175 2.8ab 24.0 18.48 0
16.9 75 0.6a 24.6 17.98 0
14.3 172 2.2ab 25.8 18.28 0
24.2 91 4.7abc 24.7 18.0a 0
25.4 236 9.9d 47.6 32.8 117
27.7 106 5.8bc 28.6 21.0bc 0
31.0 268 8.6C 42.2 28.6d 81
30.4 128 5.7bC 34.8 22.8C 17
35.2 289 12.1d 57.9 36.8 247
48.0 189 17.8e 59.7 30.6 169
43.0 302 19.58 62.4 41.5 354
19.34 233 3.3ab 26.2 18.0ab 0
27.25 252 11.3d 36.5 29.5d 18

1 Dry matter loss during a 60 day storage period.
2 Maximum temperature durin the first 30 days of storage.
Mean temperature over the first 30 days of storage.

3 Degree days that temperature measurements exceeded 35°C.

4 Propionic acid applied at a rate of 1.2% (wet basis).

5 gropionic acid applied at a rate of 1.6% (wet basis).

3°C e Superscript letters indicate values which were not
significantly different by Duncan's Multiple Range
Test (p < 0.05).

30

Table 6.3 Dry matter loss, maximum and mean temperatures,
and heating for five-bale stacks of alfalfa baled
at varying moisture and density levels
(Experiment 2).

Baling Baled Dry Matter Temperatures2 Heating3
Moisture Density Loss Maximum Mean Deg.Days
(% w.b.) (Kg/cu m) (%) (°C) (°C)

16.7 74 4.1ab 26.9 21.2a 0
16.2 199 1.86 33.1 22.66 0
18.3 87 5.0ade 27.1 21.86 0
16.9 175 3.5a 37.0 25.6b 14
18.3 100 2.43 26.3 21.98 0
17.3 191 3.8ab 39.9 27.6d 17
18.4 100 3.43 26.0 21.43 0
20.8 202 4.5abc 46.1 30.2 45
24.5 111 5.8ade 37.9 24.4bC 4
27.0 225 8.9Cd 46.5 34.6 103
32.7 130 9.2d 43.8 27.6d 32
32.2 273 9.0Cd 55.2 41.4 252
21.74 228 5.9ade 36.7 28.3d 4
30.25 250 8.3de 34.0 26.5Cd 1

1 Dry matter loss during a 60 day storage period.
2 Maximum temperature during the first 30 days of storage.
Mean temperature over the first 30 days of storage.

3 Degree days that temperature measurements exceeded 35°C.

4 Propionic acid applied at a rate of 1.2% (wet basis).

5 gropionic acid applied at a rate of 1.7% (wet basis).

3°C Superscript letters indicate values which were not
significantly different by Duncan's Multiple Range
Test (p < 0.05).

31

Table 6.4 Dry matter loss, maximum and mean temperatures,
and heating for five-bale stacks of alfalfa baled
at varying moisture and density levels
(Experiment 3).

Baling Baled Dry Matter Temperatures2 Heating3
Moisture Density Loss Maximum Mean Deg.Days
(% w.b.) (Kg/cu m) (%) (°C) (°C)

23.4 101 2.8a 24.3 11.6 0
24.0 252 4.4ab 39.1 26.8C 22
30.5 175 12.0de 42.9 25.8C 49
30.5 295 9.6Cd 46.5 32.0 112
34.7 172 10.0Cd 44.4 24.9bC 62
36.9 287 9.4cd 50.2 40.0d 218
35.0 164 11.4d 46.4 23.2ab 62
36.2 300 15.0e 52.7 41.3d 250
29.04 244 6.9bc 32.1 21.3a 0

1 Dry matter loss during a 60 day storage period.
2 Maximum temperature during the first 30 days of storage.
Mean temperature over the first 30 days of storage.

3 Degree days that temperature measurements exceeded 35°C.
4b gropionic acid applied at a rate of 1.6% (wet basis).
a C e Superscript letters indicate values which were not

significantly different by Duncan's Multiple Range
Test (p < 0.05).

 

‘ ‘
if):

9'!-

C01
st:

C
m
U

W
po:

00

de

3?.

H0.

no;

no:

32

Even with baling moisture as low as 11.5%, there were no
treatments which had a dry matter loss value of zero (0.0).
This indicates either: 1) the baling moisture must be lower
than approximately 11.5% to eliminate dry matter loss, or 2)
that there exists a threshold dry matter loss which occurs
regardless of bale conditions.

Tables 6.5 and 6.6 include Pearson product moment
correlation coefficients for dry matter loss related to other
storage parameters. Table 6.5 contains the coefficients
obtained using only data collected from non-chemically
treated stacks while results found in Table 6.6 were obtained
using all treatments. When data from stacks treated with
propionic acid were included, significance levels of
correlations were not affected but coefficients were lowered.

Two-way analysis of variance suggested that dry matter
loss was related to moisture. Correlation analysis also
indicated a relationship, as there was a very significant
positive correlation between dry matter loss and baling
moisture (r=.92)1. A scatter plot of the data illustrates
the relationship (Figure 6.1).

Effects of propionic acid treatment are difficult to
determine from the data. Treatments with propionic acid
applied did not have the same moisture and density levels as
nontreated stacks. Comparisons of acid treated stacks to
non-chemically treated stacks with somewhat similar moisture

1 Correlation coefficients mentioned in the text are for
non-chemically treated hay (Table 6.5).

33

Table 6.5 Pearson product moment correlation coefficients
for several storage parameters of non-chemically
treated alfalfa hay.

Moisture Density Max. Mean Degree
Content Temp. Temp. Days
Density .54
Max.
Temp. .83 .80
Mean
Temp. .65 .85 .90
Degree
Days .75 .80 .87 .90
D. M.
Loss .92 .61 .87 .73 .82

a All coefficients were significant at the p=.01 level.

34

Table 6.6 Pearson product moment correlation coefficients
for several storage parameters of untreated and
acid treated alfalfa hay.a

Moisture Density Max. Mean Degree
Content Temp. Temp. Days
Density .51
Max.
Temp. .81 .69
Mean
Temp. .64 .76 .89
Degree
Days .72 .67 .86 .87
D. M.
Loss .91 .58 .85 .73 .78

a All coefficients were significant at the p=.01 level.

35

on

.Amumv HmuamEHquxmv km: Mmammam vmamn haumaawcmuomu
mo mxumum HHmEm “Om musumfioa mafiamn m> mmoa umuums %Hn H.o muswfim

3.; NV 023%. 2 azfizoo ”$2902

0*

on

_

(D
F
C)

11
I)

j
e
(Iomuz 4o :4) 5501 aauvw A80

19

 

rON

36

and density levels show that propionic acid did not reduce
dry matter loss, as compared to no chemical treatment in
experiments 1 and 2. Experiment 3 gave some indication that

dry matter loss is reduced when propionic acid is applied.

6 . 3 TEMPERATURE

 

Wet hay continues to respire more than dry hay. It also
provides a more favorable environment for microbial growth.
With these sources of heat, temperatures should increase more
in wet hay than in dry hay. The experimental results support
this hypothesis.

The maximum temperatures of each treatment stack during
the first 30 days in storage are included in Tables 6.2, 6.3,
and 6.4. Correlation analysis (Tables 6.5 and 6.6) indicated
that both moisture (r=.83) and density (r=.80) are
significantly correlated to maximum temperature. Positive
correlations show that increased moisture and/or density
levels were related to an increase in maximum temperature
(p=.05). Figure 6.2 shows the relationship between moisture
and maximum temperature.

Results of the one-way analyses of variance performed
separately for each of the three experiments indicated that
treatment effects on mean temperature during the first 30
days in storage were significant (Tables 6.2, 6.3, and 6.4).
Therefore, moisture and/or density levels significantly

affect storage temperatures (p=.05).

37

.Amump Hausmefiummxmv hm: mmammam voamn haumaawcmuoou mo
mxomum HHmEm How manumHoE waHHmn m> musumumdewu mmeOuw abaaxmz N.o madman

3.; 5 oz_._<m 2 azEzoo $55.02

 

on 0* . on ON or
- b b b - l b I ”F
m
C. U C .l I
a. o o mN W
o n
e a . m
D O D 0 1mg” mm
o a 11m
00
o o I
O O O I O 9 m
o a
o 3
0 1mm
. (W
O (
o

 

8

38

Comparisons of treatments with similar densities and
different moisture levels indicate that an increase in
moisture content at the time of baling will increase mean
temperatures in storage (r=.65). Increased mean temperature
with increased moisture is in agreement with the concept that
respiration rate is higher in wetter hay and microbial
activity increases with increasing moisture. Figure 6.3
illustrates the relationship between baling moisture and mean
temperature.

Similarly, comparisons of treatments with similar
moistures and different density levels indicate that
increased bale density led to increased storage temperatures
(r=.85). This may be explained by the amount of heat
generated. Heat generation should be expressible in terms of
heat generated per unit mass (eg. W/Kg). Denser hay would
then generate more heat per unit volume. Each stack was of
approximately the same volume; therefore, if more heat were
generated in a dense stack, temperatures may be higher. This
would depend upon the specific heat of the material.

Days for which the bale temperatures exceeded 35°C were
summed to compute heating, in the form of degree days, for
each treatment. These values are also included in Tables
6.2, 6.3, and 6.4. Correlation coefficients (Tables 6.5 and
6.6) indicate that an increase in baling moisture and/or
density is related to an increase in heating. This was
expected since both mean and maximum temperatures were

correlated in a similar manner. Figure 6.4 illustrates

39

on

.Amump Hmucmsaumdxmv mm: mmammam vmamn %HumH=wGMuomu mo
mxomum HHmEm pom musumfioe mafiamn m> opsumumnamu mwmuoum cam: m.o musmah

3.; NV 023% 2 EEzoo $55.0:

 

— . ow . Own - o-N OP
. 0..
o

W

O D C ‘
-8 m
o . a . o 5. , 1
0 é. o o 3
o o o n o M
a o 19... 3
. . w
. m
0 Nu
3
o o o IQ? )
0
(\

 

Ion

.hm: mmamwam poamn hflumaswauowu mo
mxomum HHmEm pom musumeoe wGHHmn m> mmmp mmuwmp ca wawummm «.0 ouswfim

3.; “5 oz.._<m 2 5328 “$25.02

 

40

o... . on ON or
a . TITGLI‘I‘QLTIIIOJIO
on
C C D O C H
I o .. a
. m
C C . filOOF 0
\I
. D.
.9
m
can a
o m 9
. o m.
C O 1 IA
a:
noon <
C.
C.
o t 0
(

 

{00+

41

the relationship between baling moisture and heating in
degree days.

Hay dry matter serves as fuel for respiration and
microbial activity. As more of the fuel is ”burned", more
heat is generated. This heating causes temperatures to rise.
It is no surprise that the data indicated positive
correlations between dry matter loss and temperatures reached
during storage (Tables 6.5 and 6.6). Maximum temperature had
a higher correlation (r=.87) to dry matter loss than mean
temperature (r=.73) or heating in degree days (r=.82).

Results from all three experiments show that propionic
acid treatment reduces maximum and 30 day mean temperatures
in storage as compared to no chemical treatment. Heating in
degree days is also reduced through propionic acid treatment.
Lower temperatures may be the result of decreased microbial
activity. Since temperatures in acid treated hay were lower,
it would seem to follow that dry matter loss would be reduced
for acid treated hay (since it is correlated to temperatures
and heating); however, this was not evident from the data

collected in this study.

7. MODEL DEVELOPMENT
7.1 DRY MATTER LOSS

One objective of this research was to develop a
functional relationship which predicts dry matter loss from
baling moisture content and initial density. Since moisture
and density influenced dry matter loss (section 6.2), a model
of the relationship between dry matter loss and baling
moisture and density was feasible.

Data from Tables 6.2, 6.3, and 6.4, for which no
propionic acid treatment was involved, were used to develop
the dry matter loss models.

Dry matter density can be expressed in terms of moisture

level and wet density:

Dd = D, - (M * Dw) (7-1)
= (1 - n) * 0,, (7.2)
Where:

Dd = dry matter density

Dw = wet density (as is)

M = moisture content (decimal wet basis)

With this relationship, dry matter density was used as
an independent variable in some instances, rather than an
interaction term of moisture times wet density (M*Dw).
Stepwise multiple regression was used to develop the model.

Results from the analysis of variance indicated that

moisture content was the most important variable affecting

42

43

dry matter loss. A scatter plot of the dry matter loss as
related to baling moisture content (Figure 6.1) indicates a
linear relationship between dry matter loss and moisture
level. Therefore, a linear regression analysis was used.
Adding density, either wet or dry, as an independent variable
to the equation already including moisture, did not improve
the accuracy of the prediction (p=.05). The model which best
predicted dry matter loss as a function of baling moisture

was (Figure 7.1):

DML = —5.4 + (48.0 * MI) (7.3)
(r2 = .85 std. error = 1.9)
Where:

DML = dry matter loss (% of initial)
M1 = moisture content at baling (decimal wet basis)
The r2 value of 0.85 indicates that 85% of the variance
in the dry matter loss data was explained by baling moisture.
Dry matter loss increased approximately 0.5% for each
percentage point increase in baling moisture (p=.05). This
is approximately half the dry matter loss as predicted by a
rule of thumb given by Waldo and Jorgensen (1981). The
discrepancy may be due to the crudeness of a rule of thumb,
and the differences between large and small hay stacks.
Correlation coefficients shown in Tables 6.5 and 6.6
indicate that dry matter loss was also related to
temperatures during storage (p=.05). Regression models were
used to establish relationships between dry matter loss and

storage temperatures.

44

.zm: mmammam swamp
mammasmcmuomu pom musumwoe mafiamn mo :oauoasm m mm mmoa nouuma kun

3.; S uz_._<m E .358 ”mama:

H.n muswfim

 

 

 

 

 

on 0+ on ON P
F) b h b . — n - b o
1m
19
Am.H u uouum .vav

0 Ana. I NHV 1nF

35.3.8 + 6;..- u as .

O

l
(D
C“

(IDRW! :0 x) 5301 8311571 A80

45

A scatter plot of dry matter loss versus maximum
temperature (Figure 7.2) indicates a non-linear relationship.
Stepwise regression analysis was used to develop a quadratic
relationship between dry matter loss and maximum temperature.

The best fit curve for the data was:

DML = 0.00432 * (Tmax)2 (7.4)
(r2 = .81 std. error = 2.1)
Where:
Tmax = maximum temperature reached in storage (C)

A constant term did not improve the regression equation
(p=.05).

A plot of the mean temperature versus dry matter loss
indicates a linear relationship (Figure 7.3). Simple linear

regression was used to find the best fit model:

DML = -4.7 + (0.45 * Tmean) (7.5)
(r2 = .57 std. error = 3.3)
Where:
Tmean = mean temperature over the first 30 days in

storage (C)

The previous models were developed from data collected
from those treatments without propionic acid application. In
an effort to determine if propionic acid decreases dry matter
loss, further regression analysis was performed. For this
model, all data in tables 6.2, 6.3, and 6.4 were used.
Stepwise regression was used with dry matter loss as the

dependent variable. Independent variables were moisture, wet

46

.km: mmammam pmamn haumaswcmuoou mo
musumudeou mmmuouw Essaxme mo cowuocsm m mm mmoa youuma mun ~.n ouswam

8V $255.28 23232:
mm 64 on ‘nm

0 AH.N u Houum .pumv
o A . I My
Hm I N
x

NE
NA evrmmeoo.o u use

 

T I
v-

.5
0011!"! 40 :4) 5801 83mm A80

 

I
C)
Ci

47

.hms mmammam pmamn haumaswamuoou mo
manumuanou owmpoum some «0 cowuocsm m mm mmoH umuumE hum m.m muswwm

8V $2.595» 2%:

on 0.? on ON 0 _.

 

— b h p n p b r

 

Am.m u uouum .pumv 0

Ana. u Nuv

Gmwfi
A 94mq.ov + ~.¢- n 42¢

l 1
C3 “3
F

(Iomuz 40 z) 5301 aauvw A80

I
I!)
v-

 

l
C3
C9

48

density, dry density, and propionic acid application rate.
Again as in the previous model (7.1), density did not affect
dry matter loss (p=.05). Propionic acid application rate did
affect dry matter loss (p=.07). Dry matter loss as a

function of baling moisture and propionic acid application

rate was:
DML = -5.4 + (49.0 * MI) — (1.32 * P) (7.6)
(R2 = .62 std. error = 3.3)
Where:

P = propionic acid application rate (% of wet
weight)

From this model propionic acid treatment reduced dry
matter loss (p=.07). The loss reduction (retention gain) is
approximately 1.3% for each percentage point increase in
application rate. Equation (7.6) is in excellent agreement
with equation (7.3) when a propionic acid rate of zero (0.0)
is used. Data used for the model (7.6) was collected from
treatments having application rates of propionic acid ranging
from 0 to 1.7%. Use of this model beyond this range of
application rates may not yield correct conclusions.

The regression models involving dry matter loss are

summarized in Table 7.1.

Table

49

7.1 Regression models of dry matter loss, maximum
temperature, mean temperature, and heating in
degree days as functions of baling moisture and
initial density.a

Equation rZ/R2 std. error

Propionic treatments deleted:

1. DML = —5.4 + (48.0 * MI) .85 1.91
2. DML = 0.00432 * (Tmax)2 .81 2.14
3. DML = —4.7 + (0.45 * Tmean) .57 3.28
4. T = 6.2 + (72.0 * MI) + (0.078 * DI )
max w .89 4.60
5. T = 7.4 + (24.0 * MI) + (0.073 * DI )
mean ”.83 3.80
6. DD = —186 + (490 * MI) + (0.71 * DIw) .78 47.00
Propionic treatments included:
7. DML = -5.4 + (49.0 * MI) - (1.32 * P) .62 3.29
8. T = 7.4 + (78.0 * MI) + (0.062 * DI ) — (4.1 * P)
max " .84 5.00
9. T = 7.5 + (28.0 * MI) + (0.062 * DIV) - (2.7 * P)
.78 3.90
10. -l75 + (520 * MI) + (0.59 * DIV) - (50.0 * P)
.74 47.00
DIw = wet density at baling (Kg/m3.)
M1 = moisture content at baling (decimal wet basis)
Tmax maximum temperature reached in storage (C)
Tmean = mean temperature during the first 30 days

in storage (C)

P = application rate of propionic acid (% of wet
weight)

DD - degree days above 35°C

DML - dry matter loss (% of initial)

a All coefficients are over the p=.05 significance level with
an exception in equation 7 {1.32 (*P) is at p=0.07 level}.

50

7.2 TEMPERATURE

 

Stepwise regression procedures were used to develop
relationships between moisture, wet density, dry density, and
propionic acid application rate to mean temperature, maximum
temperature, and heating in degree days. The temperature and
heating models presented here are applicable only to small
stacks (approximately 5 bales). Two sets of data were used.
The first included only data from Tables 6.2, 6.3, and 6.4
corresponding to treatments without propionic acid applied.
With a significance level of 0.05 required for a variable to

be included in the equation, the following relationships were

obtained:

Tmax = 6.2 + (72.0* MI) + (0.078 * Dlw) (7.7)
(R2 = .89 std. error = 4.60)

Tmean 7.4 + (24.0 * MI) + (0.073 * DIV) (7.8)
(R2 = .83 std. error = 3.80)

DD = -186 + (490 * MI) + (0.71 * DIV) (7.9)
(R2 = .78 std. error = 47.0)

All data from Tables 6.2, 6.3, and 6.4 was used in the
second analysis to determine the effects of propionic acid
treatment on storage temperatures. Again with a significance
level of 0.05 required for a variable to be used in the

equation, the following relationships were obtained:

Tmax = 7.4 + (78.0 * MI) + (0.062 * DIV) - (4.1 * P) (7.10)

(R2 = .84 std. error = 5.00)

51

Tmean = 7.5 + (28.0 * MI) + (0.062 * Dlw) - (2.7 * P) (7.11)
(R2 = .78 std. error = 3.90)

DD = -175 + (520 * MI) + (0.59 * DIV) — (50.0 *P) (7.12)
(R2 = .74 std. error = 47.0)

These three equations are in reasonable agreement with
(7.7), (7.8), and (7.9) when a propionic acid rate of zero
(0.0) is used. Mean temperature, maximum temperature, and
heating in degree days during storage are each decreased
significantly by the application of propionic acid and
increased significantly by increases in moisture and/or
density levels. The regression models involving maximum
temperature, mean temperature, and heating in degree days are

summarized in Table 7.1.

7.3 HEAT GENERATION

 

As indicated by the temperature models developed in the
previous section, wetter alfalfa hay heats more in storage.
The models developed so far are for a small stack (5 bales).
To make the results useful for simulation of a larger stack,
a different modeling approach is required.

In order to predict temperatures in a large stack of
hay, the heating rate of the stored product must be known.
Degree days is not an adequate measure of heating since it
only quantifies the amount of heat generated above a chosen
base temperature and does not indicate the rate of heat

generation. A more useful form of energy units would be Watts

52

per kilogram of matter (W/Kg) or Watts per cubic meter
(W/m3). In order to develop some type of model to predict
heating in this way, a heat transfer model of the hay stack
was developed. Because the heating rate was unknown (most
likely being a function of time) and thermal properties would
change over time due to the drying process, an analytical
solution to the problem was impossible. Therefore, a finite

difference model of a hay stack was developed.

7.3.1 FINITE DIFFERENCE MODEL

Figure 7.4 illustrates an individual treatment stack of
five bales. Because treatment stacks were placed side by
side (Figure 4.1) and insulation was placed between them,
there were assumed to be no temperature variations in the z
direction (Fig 7.4). By symmetry, the bales were "cut" in
half in the x direction and considered to be insulated at the
center line. This simple step decreases the number of
necessary calculations by 1/2. The bottom of the stack was
assumed to be insulated as the hay was placed on a wooden
floor. The model is then a two dimensional model with two
insulated sides and two convective heat transfer sides
(Figure 7.5).

Numerical methods are used quite frequently to solve
heat transfer problems. Numerous textbooks and other
handbooks discuss the finite difference method. Myers (1971)

discusses heat transfer problems in depth.

 

 

 

 

 

 

 

 

 

1.82 m

 

 

 

 

 

1
f
I
I ————— )-— v-——————
Pr’r- 3
/ l
/
’ 1
I(.15,1.29,.23) )4
)— ————— —-1)-——_.
’l
/
/ I
/
T )
l(.15,.92,.23) )fi
’4— ————— )——‘I—-__-__
/
/ l
/
/ l
I
|(.15,.55,.23) X
/’1- ————— n—-——--——_
/ l
/ .L
l
' .
,m. .....
I
A, I

Figure 7.4 Thermocouple positions in a treatment

stack of five bales.

 

 

Vertical surface convection

 

Y 54
/——-—— Horizontal surface convection
/

 

 

 

 

/—
(.15,.92) Assumed position of
vii/rfik/////-- target temperatures

 

 

\\ ‘5. \4_ ‘1, \\

 

 

 

 

 

 

 

 

1.82 m
, Insulated surfaces
1/’/
,7
,/ .
/?
/
IJ /
DX
.153 m 1‘
A ________>_
I / / / / x
p Dx )
.153m .—

 

 

l.-——- .46 m -—-—.q

Figure 7.5 Finite difference model of a five
bale hay stack.

55

To simplify the equations, grid length increments on the
x and y axes (DX and DY) were taken to be equal. To develop
the finite difference equations, the first law of
thermodynamics was applied to small control volumes for all
positions in the grid. For interior points, the sources of
heat were conduction transfer from neighboring control
volumes and heat generation. For points along the outside
edge, an additional heat source (or sink) was the convection
to the ambient air. Applying energy conservation to control
volumes for different positions in the stack leads to the

following explicit equations:

interior points (x,y) x#0,w ; y#0,v :
Tm.n.p+1 = F*(Tm+1.n.P + Tm-1.n.P + Tm.n+1.P +

Tm'n_1’p + Tm'n'p) + B*G + Tm’n'p (7.13)

left edge points (0,y) y#0,v :
Tm,n,p+1 ‘ 2F"‘(Tm+l,n,p + '5Tm,n+l,p + °5Tm,n-1,p '

2Tm'n'p) + B*G + Tm,n,p (7.14)

right edge points (w,y) y#0,v :

Tm.n.p+1 = 2F*(T -1.n.P + 'STm.n+1.P + '5Tm.n-1.P '
2Tm,n,p) + B*G + Tm,n,p +
(2BH2/DX)*(Ta - Tm'n'p) (7.15)

bottom edge points (x,0) x#0,w

Tm'n'p+1 3 2F*(.5T _1’n’p + .5Tm+1'n’p + Tm'n+1'p ‘

2Tm'n'p) + B*G + Tm'n'p (7.16)

56

top edge points (x,v) x#0,w :

Tm,n,p+1 ‘ 2F*('5Tm—l,n,p I 'STm+l,n,p + Tm,n-l,p ’

2T + B*G + T +

mlnlp) min'p

(zanl/Dx)*(Ta - Tm'n'p) (7.17)

bottom left corner (0,0) :

Tm,n,p+l = 2F*(Tm+1,n,p + Tm,n+1,p ‘ Tm,n,p) +
B*G + Tm,n,p (7.18)
top left corner (0,v) :
Tm,n,p+l : 2F*(Tm+l,n,p + Tm,n—l,p ' 2Tm,n,p) I
B*G + Tm,n,p +
(ZBHl/DX)*(Ta - Tm’n'p) (7.19)
bottom right corner (w,0) :
Tm,n,p+l ' 2F*(Tm—l,n,p I Tm,n+1,p ’ 2Tm,n,p) I
B*G + Tm'n'p +
(ZBHZ/DX)*(Ta - Tm'n'p) (7.20)

top right corner (w,v) :

Tm,n,p+l ‘ 2FMTm-l,n,p + Tm,n-l,p ’ 2Tm,n,p) I

B*G + Tm'n'p +

(2A/DX)*(H1 + H2)*(Ta - Tm'n’p) (7.21)
Where:

B = A*DT/K

F = A*DT/(DX)2

Ti'j’t - temperature (c) at:x node i

y node j

time = t*DT

H2 - vertical surfacezconvective heat transfer
coefficient (W/m °C)

H1 - horizontal surfase convective heat transfer
coefficient (W/m °C)

57

G = heat generation rate (W/m
T a ambient temperature (C)
DT = time increment (s)

Dx = grid length increment (m)

K thermal conductivity (g/m°C)
A thermal diffusivity (m /s)

3)

In order for the solution to be stable, the time

increment (DT) was limited by the following equation:

or < ((DXZ/A) / (2*(H*DX/K + 1)) (7.22)

7.3.2 ESTIMATING THERMAL PROPERTIES

 

As discussed in the literature review (section 3.5),
thermal properties for baled alfalfa hay are not well known.
It was desirable to have expressions for thermal conductivity
and thermal diffusivity as functions of moisture and density.
The thermal properties may also vary with the actual content
of the hay (e.g., some grass vs. pure clover or alfalfa) but
these differences would likely be much less important than
moisture and density. The estimation equations used were
taken from references discussed in section 3.5.

Specific heat for alfalfa hay was given by Bern (1964).

Changing equation (3.7) to SI units yields:

C = 919 + 5933 * M (7.23)
Where:

C a specific heat (J/Kg)

According to Andersen (1950), the thermal conductivity of

a wet solid is given by:

58

K = M*Kwater + (1-M)*Ksolid (3.9)

If we consider the hay/air mixture to be the solid and
water as the other component, equation (3.9) is adequate
once we know the thermal conductivity of the hay/air mixture.

Ott and Horbut gave an equation for thermal diffusivity
for hay at 8.2 % moisture. This equation (3.5) converted

into SI units is:

A 082 = 6.01(10)‘7 + 1.3(10)‘9Dw (7.24)

Where:

A 082 = thsrmal diffusivity of hay at 8.2% moisture
' m /s
Dw = wet density (Kg/m3)

By definition of thermal diffusivity:

A = K/(D*C) (7.25)
Where:
A = thermal diffusivity
K = thermal conductivity
C = specific heat
D = density

For the given moisture level of 8.2 % and a given

density Dw' then:

A.082 = K.082/(Dw,.082*c.082) (7.26)

Where:

subscript .082 refers to the moisture content at
which the property is evaluated

If equations (7.23) and (7.24) were exact rather than

empirical, substitution of both into equation (7.26) would be

59

exact. Even though they are empirical, some liberty was
taken to combine these equations together since this is the
best data available. Since the temperature range of concern
in this problem is small (lo-70°C), the thermal properties
would not change considerably with temperature. Substituting
equations (3.9), (7.23), and (7.24) into equation (7.26)

leads to an expression for Ksolid:

“56118 = 9.2(10)'4*nw’.082 + 2‘10)'6*(Dw,.082)2 - .0536(7 27)

Where:
Ksolid = thermal conductivity of the hay/air

solid (W/m°C)

Dw' 082 = wet density at 8.2% moisture (Kg/m3)

The known density (Dw) is at a given moisture M. To
convert to equivalent density for a moisture level of 8.2%

the following relationship can be used:

DW,.082 = 1.09 * (I‘M) * Dw (7.28)

These equations, repeated for clarity, along with the

measured density, Dw, were used to estimate thermal

properties of baled hay:

c = 919 + 5933 * M (7.23)
DV,.082 = 1.09 * (I‘M) * Dw (7.28)

Ksolid = 9.2(10)"4')=1>‘,,“082 + 2(10)’5*(0w“082)2 - .0536(7 27)

x = M* + (1-M)*K - (3.9)
A = x/IBS‘E) _ 5°1‘d (7.25)

60

Where:
C - specific heat (J/Kg)
3)

D

w a wet density (Kg/m

M = moisture content (decimal wet basis)

K - = thermal conductivity of the hay/air
5°1‘d solid (W/m°C)

Kwater 3 thermal conductivity of water (W/moc)
K a thermal conductivity of wet hay (W/moc)

A = thermal diffusivity of wet hay (mZ/s)

To validate the use of the previous five equations for
thermal property estimation, known thermal properties of
similar substances were compared. The specific heat of hay
was assumed to be similar to that of tobacco. Table 7.2
contains the known specific heat of tobacco and estimated
specific heat of alfalfa and tobacco. Equation (3.9)
predicted the specific heat of tobacco quite well and was
thus used to estimate specific heat for alfalfa. Equation
(3.2) (Jiang et. a1, 1985) did not yield a comparable value
,for specific heat. This is because the equation is
applicable to alfalfa silage which is wetter and more dense
than baled alfalfa.

Thermal conductivity of tobacco has not been presented
in the literature. It is known for many substances but none
quite so similar to baled alfalfa. Granulated cork is used
here for comparison (Table 7.3). Estimated thermal
conductivity of cork is somewhat higher than the known value.

Therefore, the estimated thermal conductivity for alfalfa may

61

be higher than the true value. The thermal conductivity for

baled hay as estimated in the model is also higher than
estimated by extrapolating equation (3.4) (Jiang, et.

1985).

Table 7.2 Comparison of known and estimated specific
values of tobacco to estimated specific
values of hay.

Material Moisture Specific
(% w.b.) Heat
(J/Kg°C)
Tobacco
(known)a 16.6 1431
28.0 2598
(estimated)b 16.6 1904
28.0 2580
Baled
Alfalfa
(estimated)b 20.0 2105
25.0 2402
(estimated)C 20.0 3248
25.0 3354

a Known value for Tobacco taken from Chakrabarti
Johnson, 1972.

b Estimated by equation (7.23) as used

that

al.,

heat
heat

and

in the model3

c Estimated by equation (3.2) with a density of 176 Kg/m

for comparison only.

62

Table 7.3 Comparison of known and estimated thermal
conductivity values of granulated cork to
estimated thermal conductivity values of baled

hay.
Material Moisture Densigy Thermal Conductivity
(% w.b.) (Kg/m (W/m°C)
Cork - granulated
(known) ? 86 0.05
Cork
(estimated)b 5.0 86 0.07
Baled alfal a
(estimated) 20.0 176 0.23
25.0 176 0.24
(estimated)C 20.0 176 0.18
25.0 176 0.18

a Known value for granulated cork taken from ASHRAE
,1981.

b Estimated by equations (3.9), (7.27), and (7.28) as
used in the model.

C Estimated by equation (3.4) for comparison only.

7.3.3 ESTIMATING HEAT GENERATION RATES

 

To estimate thermal properties for the finite difference
model developed in section 7.3.1, variations in the moisture
and density levels of the hay over time must be known. The
data taken provided only initial moisture, 60 day moisture
and initial density levels Moisture was assumed to change
exponentially over time as illustrated in Figure 7.6. The
moisture variation is not known to follow such a curve;

however, the hay is drying and should follow a similar

63

.mefiu um>o ucmucoo unaumfioe ca coaumfium> 0.5 ouswfim

 

 

363 m2:
cm) on 0* on ON or o
a . . _ . . t . . opd
(med
loud
eke . Suarez»:
InNd
lend

(819°C) 19M IowPep) 1N31N00 sanISlow

64

pattern as set by grains in the drying process. A linear
change in moisture content over time was considered but this
would not allow continuity in the rate of change of moisture
at 60 days. The value of the exponential constant for each
treatment stack was determined using initial and 60 day
moisture levels. Equilibrium moisture content of the hay was
not taken into consideration, and perhaps should have been.
Estimating moisture content in this manner allows some error
in that it projects an equilibrium moisture content of the
hay at time infinity of 0.0%.

Density variations in the hay were assumed to be caused
solely by the loss in moisture. That is, dry matter density
was assumed to remain constant. This assumption is not
entirely correct as some stack settling occurs (especially in
wetter hay) and some dry matter loss occurs (see sections 6.2
and 7.1). However, the settling and dry matter loss changes
would tend to offset one another. With the moisture and
density changes over time estimated in this manner, the
finite difference model developed previously was used to
estimate heat generation rates.

As discussed in section 5.3, three daily mean treatment
temperatures were known. The position of these known daily
temperatures were approximately as shown in Figure 7.4. The
temperatures were averaged to give a mean daily treatment
temperature. The 30 mean daily treatment temperatures were
assumed to be the temperature at the location (x=0.92,

y-0.153) (Figure 7.5) and were called "target" temperatures.

65

The three temperature measurements were averaged to remove

fluctuations in the temperature data.

Equation (7.13) was solved for the heat generation rate,

G = (Tm.n.P+1 - Tmpnrp - F*(Tm+llnrp + Tm-l'n'p +

Tm,n+l,p + Tm'n_1'p + Tm'n'p))/B (7.29)

Where:

G = heat generation rate (W/m3)

T = target temperature at node (m n) given
m n +1 '

' 'p by data (c)
Tm n p = current temperature at node (m,n) (C)
Tother = current temperatures surrounding

node (m,n) (C)
B = A*DT/K 2
A = thermal diffusivity (m /s)
DT = time increment (s)
K = thermal canductivity (W/m°C)
F = A*DT/(DX)

The heat generation rate for a given day was predicted
using immediate past temperatures calculated using the finite
difference model and a target temperature for the next day.
This heat generation rate was then used to calculate new
temperatures throughout the stack. This procedure was
repeated for the equivalent of 30 days. Ambient and target
temperatures and moisture content were updated on a daily
basis.

The grid size used in the finite difference model for
predicting heat generation rates was determined by trying
different values for the grid increment (Dx) and evaluating

the difference in results. A grid increment of 0.153 m was

chosen. Variation in results from a grid this size as

 

66

compared to a very small grid (DX=0.046 m) were minimal.
Increasing the grid increment above this value, however, led
to sizable error.

This method for estimating heat generation rate allows
some error but the error is relatively small compared to the
variation in thermal properties within a bale. Estimating
heat generation rate from equation (7.29) requires the
assumption that nodal temperatures surrounding the "target"
node (m,n) remain constant over some time period. These
temperatures are actually changing over time. As an
indication of the error involved in this procedure, the
difference between target temperature and the temperature
calculated using the estimated heat generation rate reached a
maximum of approximately 5%.

There are more accurate methods of predicting unknown
thermal characteristics but the curve form of how that
property (heat generation rate in this case) changes over
time must be assumed (Beck, 1977). Since the form of the
heat generation curve is not known, the method described here
was used.

This procedure for estimating heat generation rate was
repeated for each treatment listed in Table 6.1. Heat
generation rate was converted from a per unit volume basis

to a per unit mass basis by:

G2 = G / Dw (7.30)

 

67

Where:
62 = heat generation rate (W/§?)

G = heat generation rate (W/m
Dw = density of the hay (Kg/m3)

7.3.4 HEAT GENERATION MODEL

Heat generation rate data was obtained for each of the
37 treatments. The mean heat generation rates over the 30
day period for treatments in each experiment were analyzed
using one way analysis of variance. Treatment effects on
mean heat generation rates were clearly significant.
Duncan's Multiple Range Test was used to determine which
treatments had significantly different mean heat generation
rates (Tables 7.4, 7.5, and 7.6).

Mean heat generation rates over the first 30 days in
storage increase as moisture increases. This supports the
hypothesis explained previously concerning temperatures in
storage. Two way analysis of variance was used to determine
if moisture and density effects were statistically
significant. Moisture contents were divided into 5 ranges
and density into 2 ranges. Moisture, density, and the two
way interaction of moisture and density were each related to
heat generation rate (p=.01). Therefore, any model used to
predict heat generation rates would need to include each of

these as potential independent variables.

68

Table 7.4 Mean heat generation rates for alfalfa hay baled
at varying moisture and density levels
(Experiment 1).

Baling Moisture Baled Density Heat Generation Rate1

(% w.b.) (Kg/cu m) (W/Kg)
11.5 111 0.0046
14.6 175 0.020ab
16.9 75 0.005a
14.3 172 0.017ab
24.2 91 0.009a
25.4 236 0.1489f9
27.7 106 0.047abc
31.0 268 0.097Cde
30.4 128 0.069de
35.2 289 0.1709h
48.0 189 0.161f9h
43.0 302 0.214h
19.32 233 0.009a
27.23 252 0.106def

1 Mean heat generation rate over first 30 days in storage.

g Propionic acid applied at a rate of 1.2% of hay weight.
grgggonic acid applied at a rate of 1.6% of hay weight.
3°C e Superscript letters indicate values which were not
significantly different by Duncan's Multiple Range

Test (p < 0.05).

 

69

Table 7.5 Mean heat generation rates for alfalfa hay baled
at varying moisture and density levels
(Experiment 2).

 

............................................................. p
Baling Moisture Baled Density Heat Generation Ratel
(% w.b.) (Kg/cu m) (W/Kg)

16.7 74 0.010ab

16.2 199 0.051ab )

18.3 87 0.006ab

16.9 175 0.087de

18.3 100 0.011ab

17.3 191 0.049abC

18.4 100 0.011ab

20.8 202 0.070abc

24.5 111 -0.0033

27.0 225 0.102Cd

32.7 130 0.06 abC

32.2 273 0.153

21.72 228 0.015ab

30.23 250 0.031bc

1 Mean heat generation rate over first 30 days in storage.

2 Propionic acid applied at a rate of 1.2% of hay weight.

3 gropionic acid applied at a rate of 1.7% of hay weight.

abc Superscript letters indicate values which were not

significantly different by Duncan's Multiple Range Test
p < 0.05 .

70

Table 7.6 Mean heat generation rates for alfalfa hay baled
at varying moisture and density levels
(Experiment 3).

Baling Moisture Baled Density Heat Generation Rate1
(% w.b.) (Kg/cu m) (W/Kg)
23.4 101 0.003
24.0 252 0.114ab
30.5 175 0.125ab
30.5 295 0.137b
34.7 172 0.123ab
36.9 287 0.219C
35.0 164 0.156b
36.2 300 0.243C
29.02 244 0.067a

1 Mean heat generation rate over first 30 days in storage.

2 Propionic acid applied at a rate of 1.6% of hay weight.

3°C Superscript letters indicate values which were not

significantly different by Duncan's Multiple Range Test
p < 0.05 .

71

Using the database of heat generation rates for varying
moistures, densities, and days from baling (1110 points
total), a model of heat generation rate was developed. Again,
as for the dry matter and temperature analyses, two sets of
data were used. First, that data corresponding to non-
chemically treated stacks and second, data from all
treatments including treatments with propionic acid applied.

It was desired to express the heat generation rate as a
function of moisture and density. If time from baling could
be discarded without loss of accuracy, the model would be
more versatile. However, a breakdown of heat generation
rates by moisture and time from baling indicated that time
from baling was an important factor.

Heat generation rates were averaged for each day over
all treatments; a plot of heat generation rates over time
illustrates the time effect on heating (Figure 7.7).
Moisture and density are both decreasing slowly during
storage; therefore, a model with* only these two as
independent variables would not allow for an increase in heat
generation rate over the first several days (Figure 7.7).

Because the heat generation rate peaks at approximately
8 days, the data was "split" for model development. There is
nothing magic about day 8 except that the breakdown of heat
generation by time from baling showed that on the average,
the maximum heat generation rate occurred on this day. Two
sets of data were used in the regression procedure. The

first, data set corresponded to time from baling less than 9

72

on

.oEHu m> has
mmammam poamn uaumaswcmuomu mo wmumu coaumumcmw and: n.n muswam

A863 023$ 28: us:

 

_ nu cm 2 o. n. a
— b p b — 00.0
C
o o o o o
(no.0
o o o o o
C
O .
o o o o IOPO
O O. O
O C
o Impd
C

 

IONd

(fix/M) 31w NouvaaNao 1V3H

73

days. The second data set corresponded to time from baling
greater than 7 days. Data from day 8 was used in both sets
to provide continuity.

Stepwise regression was used to develOp the models with
heat generation rate as the dependent variable. Independent
variables included moisture, density, an interaction term
(moisture times density), the square and square root of each
of these three, and time from baling. For non-chemically
treated hay, the best fit equations (p=.05) predicting heat

generation rates were:

For t g 8:
62 = 2.47m2 + 0.021*t + 0.0119*Dw'5 - 0.307 (7.31)
(R2 = .558 std. error = 0.120)
For t Z 8:
G2 = 0.0000256*(M*Dw)2 - 0.005*t + 0.018141%;5 -
0.00000185*Dw2 - 0.060 (7.32)
(R2 = .452 std. error = 0.080)
Where:

GZ = heat generation rate (W/Kg)
M = moisture (decimal get basis)
Dw = wet density (Kg/m )

t = time from baling (days)

The equations do not provide for continuity at t=8, so
to estimate heat generation rate on day 8, results from the
two equations were averaged.

The same procedure was repeated with data included from

all treatments with the exception that propionic acid

application rate was added in the list of independent

74

variables. The best fit equations (p=.05) predicting heat

generation rate with propionic acid treatments were:

For t 5 8:
62 = 2.39m2 + 0.020*t - 0.088*P + 0.0126*Dw'5 — 0.306 (7.33)
(R2 = .564 std. error = 0.115)
For t 3 8:
62 = 0.0000145*(M*Dw)2 - 0.004*t - 0.039*p +
0.0146*(M*Dw)'5 + 0.037 (7.34)
(R2 = .359 std. error = 0.089)
Where:

P = propionic acid application rate
(% of wet weight)

Comparisons of equations (7.31) and (7.32) to equations
(7.33) and (7.34) are difficult because of the difference in
form; however, all models indicate that heat generation rate
is increased by an increase in moisture and/or density level.
Figure 7.8 illustrates the effects of moisture and density on
heat generation rate for a fixed time from baling (t=6). For
a fixed density and time from baling, heat generation rate
increases as the square of moisture. Fixing moisture and
time from baling shows that heat generation rate increases
approximately as the square root of density in all equations
except (7.34). In (7.34) heat generation rate varies almost
linearly with density.

Heat generation rate increases linearly as time
progresses during the first few days, and decreases linearly

over time thereafter.

75

mm.

on

.hm: memmam umamn haumaswcmuumu CH
sump coaumuwcmw Dam: so muommww zuwmsop 06m musumfioz m.n muswfim

3.; .5 #5562
MN

cu

m.— r\. o.

 

E so\9_ on; I
E :o\uv_ 02 I
E ao\0v_ 0nN film
E :o\m¥ omN Ole

00.0

 

(Bx/M) 31w Nouvaz-INz-Io 1v:-)H

76

Propionic acid application decreases heat generation
rate. This decrease in heating is greatest during the first
several days of storage.

Heat generation can only occur with a material serving
as a source of energy. That energy source in baled hay is
the hay dry matter. Therefore, a relationship between dry
matter lost and total heat generated should be apparent.

In a hay stack, two processes are taking place which
affect the relationship between dry matter loss and heat
generation. One process is the drying of the hay; this
moisture removal requires heat or energy. The other process
is the oxidation of carbohydrates which reduces the available
dry matter in the hay. The chemical reaction for the

oxidation of carbohydrates is considered to be:

C6H1206(S) + 602(9) -’> 6C02(g) + 6H20(1) +
2820 KJ/mole of carbohydrate (7.35)

In this reaction, the carbohydrate (glucose) is a solid,
the oxygen and carbon dioxide are gases, and the water
produced is a liquid. The water produced as a liquid must be
evaporated to be removed from the stack. Taking the heat of

vaporization of water (at 25 °C) into consideration decreases

the amount of heat produced:

C6H1205(s) + 602(9) --> 6C02(g) + 6H20(g) +
2557 KJ/mole of carbohydrate (7.36)

77

The amount of hay dry matter consumed in a hay stack can

be expressed by:

L = Dw*V*DML*(l-MI) (7.37)
Where:
L = amount of dry matter lost gKg)
D = density of wet hay (Kg/m )

DML = percent of initial dSy matter lost
V = volume of the stack (m ) ‘
(l-MI) 2 Kg of dry matter per Kg of wet hay

The total heat produced in a stack of hay is the amount

of dry matter consumed times the energy content of the dry

matter:
THP = L * 14206 (7.38)
Where:

THP = total heat production (KJ/stack)
14206 = KJ of energy per Kg of dry matter

consumed
= (2557KJ/mole * lmole/lBOg * lOOOg/Kg)

As mentioned previously, water is being evaporated from
within the hay during storage. The amount of heat required to
dry the hay is a function of the amount of moisture removed.
With a storage time period of 30 days considered, the amount

of heat required for moisture evaporation is given by:

Hevap = (MI-M30)*Dw*v*2433 (7.39)
Where:
H = KJ of heat used to evaporate water

Mggap moisture content after 30 days in storage
(decimal wet basis)

2433 = KJ of heat required to evaporate 1 Kg of
water at 25°C

78

The net heat released from a stack is the difference
between total heat production and heat used for water

evaporation:

Qnet = 14206*Dw*V*DML*(l-MI) — 2433*(MI-M30)*Dw*v (7.40)

Where:

Qnet = heat leaving hay stack (KJ)

The net heat released can also be estimated as a

function of mean heat generat1on rate (szean) by:

Qnet = 2592*V*Dw*62mean (7.41)
Where:
szean = mean heat generation rate over 30 days

of storage (W/Kg of wet hay)
2592 = (30d) * (864005/d * 1KJ/1000J

Setting these two expressions equal to one another and
solving for dry matter loss results in a theoretical equation

relating heat generation rate to dry matter loss:
DML = 0.182*62mean/(l-MI) + 0.171*(Ml-M30)/(1-M1) (7.42)

This process could be used for any length of storage
period. A storage period of 30 days was used, as the mean
heat generation rate for this time was estimated.

Regression analysis was used to determine if
experimental dry matter loss data and estimated heat
generation rates agreed with the theoretical development.
The relationship between dry matter loss and heat generation

was forced to be that of equation (7.42).

 

79

Moisture content on day 30 was not measured in the
experiments. To estimate the moisture content on day 30, a
weighted average of initial and 60 day moisture content was
used. The moisture is considered to be decreasing
exponentially since the hay is drying; therefore, 30 day
moisture should be closer to 60 day moisture than initial
moisture. The coefficients of 1/3 and 2/3 have no

mathematical basis; they were chosen as estimates.

M30 = l/3*MI + 2/3*M60 (7.43)
Where:
MI = measured moisture content at time of baling
”60 = measured moisture content on day 60
Regression of experimental data was used to develop a
model for predicting dry matter loss as a function of
estimated heat generation rate, initial moisture content and

estimated 30 day moisture content. The resulting model was:

DML = 0.214*62mean/(1-MI) + 0.175*(MI-M30)/(l-MI) (7.44)
(R2 = .85 std. error = 1.83)

Neither coefficient was statistically different (p=.01)
than the corresponding coefficient in the theoretical model
(7.42). This close agreement between the theoretical and
experimentally determined relationships validates the
procedure used to estimate heat generation rate. With this
relationship and the heat generation model, dry matter loss

can be predicted for any size hay stack.

8. MODEL VALIDATION

The models developed in the previous chapter were
developed solely from data taken in three experiments during
the summer of 1985. Two sets of data were used to validate
the models. The first set of data was from experiments
performed during the summer of 1984. Ten different non-
chemically treated stacks of ten bales each were stored
during this season. Data were collected just as for this
study with the exceptions that bale density was not measured
and dry matter loss was evaluated after 30 days rather than
60. The second set of data was taken from an experiment which
had a relatively large (100 bales) stack of hay. The data
used for model validation is summarized in Table 8.1.

For each non-chemically treated stack in the first
validation data set, models were used to predict a dependent
parameter (eg. DML) from an independent parameter (eg. M). A
linear regression was then performed with the predicted value
as the dependent variable and the actual value from
validation data as the independent variable. For an exact
fit of a model, the intercept and slope of the resulting
equation would be 0.0 and 1.0 respectively. Deviations from

this indicated error in the model.

80

81

Table 8.1 Data used for validation of models which predict
dry matter loss and storage temperatures.

Baling Temperature Heating Dry Matter
Moisture Maximum Mean Deg. Days Loss
(% w.b.) (°C) (°C) >35°c (%)

15.6 30 20 0 1.0
22.7 47 32 68 5.1
37.2 55 39 208 10.8
15.8 28 22 0 0.7
28.1 45 32 73 8.7
12.5 21 15 0 0.0
30.6 44 36 77 9.1
24.1 42 28 37 6.9
20.6 38 18 15 4.4
19.0 27 24 0 0.7
25.83 35 25 1 1.0
25.98 33 26 5 3.2
30.78 42 37 84 11.4
33.26 32 27 14 2.8
25.4b 36 29 2 4.7

a Treated with propionic acid (1% of dry matter)

b Data taken from a stack of approximately 100 bales.
However, temperatures and dry matter loss were measured
on only 10 bales from the center. Density was

estimated to be 175 Kg/m .

8.1 DRY MATTER LOSS

Dry matter loss can be predicted from moisture content
at baling from equation (7.3). Comparisons of the model
predictions to actual dry matter loss indicate that this
model is quite accurate. Figure 8.1 illustrates the fit of
the model. Regression analysis with predicted dry matter loss
(as a function of initial moisture) as the dependent variable

and actual dry matter loss as the independent variable show,

82

 

.wcfiamn um ucouaoo ousumaoa mo cofiuocSM
m mm mmOH assume %u0 muuwpoua SUH£3 H6008 m 00 coaumpfiam> H.m madman

goes to 5 $0.. SEE can .222
0.6. 9 or m

— _ r _

C3

\
”24

((01181! 40 x) 550') HEIJJNW A80 0310(0388

fin

I
C)
F

l
a)
F

 

 

\ . ran

83

that the slope was not different (p=.05) from 1.0; nor was
the intercept different (p=.05) from 0.0. Dry matter loss
for the 100 bale stack as predicted from baling moisture is
6.8%. Actual dry matter loss was only 4.7%. Perhaps the
model predicting dry matter loss from baling moisture is not
applicable to larger stacks. Oxygen availability in large
stacks would most likely be lower than in small stacks. The
oxygen limitation may in turn affect microbial activity and
thus dry matter loss. More data from large stacks needs to
be compared to determine if this model can be applied to
large stacks.

Dry matter loss can also be predicted from maximum
storage temperature or mean temperature during the first 30
days of storage (equations 7.4 and 7.5). Dry matter loss as
predicted by either of these were generally higher than the
actual dry matter loss (Figures 8.2 and 8.3). Slopes of the
equations relating predicted dry matter loss (as a function
of temperatures) to actual dry matter loss were both less
than 1.0. This indicates that the models suggest more
dependency of dry matter loss on temperatures than may exist.
For the large stack, estimated dry matter loss values, given
mean temperature and maximum temperature were 8.3 and 5.6%
respectively. Actual dry matter loss was 4.7%. As for
smaller stacks, using maximum or mean temperatures to predict
dry matter loss led to an over estimation. The mean
temperature model yielded more accurate results than did the

maximum temperature model. Even though the models predicted

84

.omeODm :H pmnomou ouaumumaewu Essaxms mo :oauuasm
m we mmoH nouumE has muofipoua :oan3 H0008 a mo cowumwfiam> «.0 madman

92:5 to .5 $0.. $52 ta .229.

0W 9 0' m 0

_ (b _

CD

I
H)

j
C)
v-

1
I)
F

 

 

\ . fax“

(lolllul 4° %) SSO'I HELIVW A80 (JELLOIOEHd

85

.omeODm ca mmmp 0m umufim was wcfiu=0 musumumaewu amma mo
cofiuocsw m we mmoH umuuma >u0 muofipmua soaps Hmpoa m mo coHDMpHHm> m.w Tasman

92:5 to .5 $3 6.5: $5 .222
cu m. or m 0

_ _ p _

C3

.1
d)

10.

.3.
(IDIMUI )0 z) 550') 83.1.va A80 0310(0388

 

 

\
r,

(3

Ci

86

dry matter loss greater than the actual, they do indicate
correct trends.

Dry matter loss for the treatments stored in 1984 with
propionic acid applied were extremely variable. Influences
of propionic acid on dry matter loss (as estimated from the
models) were smaller than the fluctuations in the data;
therefore, no validation of the propionic acid treatment
models could be done. More data needs to be collected to
prove the effects of propionic acid, because the effects are

small.

8.2 TEMPERATURE

 

The temperature models developed in section 7.2 involve
moisture and density as independent variables. The
validation data from the 10 bale stacks did not include
density levels, so an estimated density of l60Kg/m3 was
assumed. This corresponds to a typical bale. Densities of
the bales in the large stack were not known either; however,
average bale weight was 23.8 Kg. With an average bale length
of 91 cm assumed, the estimated density was 175 Kg/m3.

Comparisons of predicted maximum temperatures to actual
maximum temperatures (Figure 8.4) indicated error in the
model. Over estimation of lower temperatures and under
estimation of higher temperatures suggests that maximum
temperature may increase more for increased initial moisture

content than the model shows. Density tends to increase as

 

.huamamp Hmauficfi was wswamn um ucoucoo ousumaoa mo cofiuo:=w
6 mm assumuooEou ESEmeE muofikua guess Hmpoa m 00 cowumvfiam> «.0 ouswam

80 22.5.22”: 2:222 232
00 on 0* on 0N

— b _ p — b _ e inN

 

87

(0) aanlvaadwal 'wnWvaw 031010388

 

 

88

moisture increases; if this were taken into account in the
validation, model fit would be improved. Application of this
model to the large stack predicted maximum temperature well.
Predicted maximum temperature was 36°C and the actual maximum
temperature was 38°C. More data should be compared to assure
the validity of using this model in large stacks. A model of
the heat transfer in a stack of hay including heat
generation (section 7.4) would be more applicable to large
stacks as stack size and structure need to be considered.

A plot of the predicted mean temperature versus actual
mean temperature indicated a poor model (Figure 8.5). From
this validation, mean temperature during the first 30 days in
storage is more dependent upon moisture than the model
suggests. Prediction of mean temperature from baling
moisture and density for the large stack were reasonable but
slightly low. Actual mean temperature was 29°C and
predicted mean temperature was 26°C. This mean temperature
regression model was developed from small stacks. Small
stacks have more surface area per unit volume and thus can
dissipate the heat more effectively. Mean temperatures in
large stacks should be predicted using a heat transfer model
which includes the heat generated by the hay.

The validation curve of heating in degree days (Figure
8.6) indicates large error in the model; however, with the
exception of one point, correlation between predicted heating
and actual heating is very high. The predicted value of

degree days (63) for the larger stack of 100 bales was not

89

 

.mufimaov Hmfiufisw paw wcaama um ucouaoo ousumaoa mo sofiuoasm
m mm assumuoaewu some muofipwud 50H33 Hopoa m 00 coaumpfiam> n.w muswwm

80 mm2<mma22 22m: .222
on 3 on ON 0.

n p. h b — p b - 40F
\ .

 

(0) aanlvaadwal Nvaw 0310(0388

 

 

\ than

9O

.zufimcmu Hmauaca vcm wcwamn um ucmuaoo unaumfios mo soauoaSM m
mm m%mv wmuwmv ca wcfiumwn muuwvmua Sowna Hmvoa m mo cowumvaam> o.m unawam

m><o mmmomo z. ozfifi: .232
emu omw cm. a»: 0% o

 

 

100..

\ ICON

SAVO 338930 NI ONIlV3H O310|03Hd

 

\ IonN

91

close to the observed value (2). This indicates the need for
a more comprehensive model. The large stack was stored
outside, unlike the small stacks used to develop the models.
Wind currents which were not an important factor for inside
storage would increase heat transfer from the stack to the
environment.

The models predicting storage temperatures which include
effects of propionic acid were difficult to validate because
of inconsistency in the validation data. Again as for dry
matter loss, the predicted effects of propionic acid
treatment were smaller than normal fluctuations in the data.
The models did predict correct trends in maximum temperature,
mean temperature and degree days, but observed values
deviated from the model predictions somewhat. The deviations
between observed and predicted values were less than the

deviations within the validation data.

8.3 HEAT GENERATION

 

Heat generation rate for non-chemically treated hay can
be estimated with equations (7.29) and (7.30). To be useful,
these estimated values must be used in a heat transfer model
of a hay stack. Validation of the heat generation model was
performed by using it, in combination with the finite
difference model, to predict hay temperatures in a 10 bale
stack. The stack chosen for validation of the model was

baled at 41 percent moisture and the density was estimated to

92

be 225 Kg/m3. Figure 8.7 illustrates the fit of the model.
Ambient temperatures from the experiment were used in the
model to eliminate any discrepancies which might be caused by
differing ambient temperatures. The largest difference
between actual temperatures and predicted temperatures occurs
approximately 20 days after baling. The drop in temperature
occurring during days 13-17 is often observed. The most
important result from a temperature model of hay in storage
is the maximum temperature. It is the maximum temperature
which indicates whether or not the hay can be stored safely
(i.e., no combustion). The model did predict maximum
temperature quite accurately but did so with a time lag of

approximately 3 days.

8 . 4 MODEL SENSITIVITY

 

Because the heat generation model was developed from a
heat transfer model of a hay stack, accuracy of the heat
transfer model directly affects accuracy of the heat
generation model. Estimation of the thermal conductivity of
hay was believed to be the largest source of error in the
finite difference heat transfer model.‘

' Sensitivity of the heat generation model to thermal
conductivity was determined by changing the value of thermal
conductivity used in the model by 10% and observing the
change in estimated heat generation rate. On the average, a

10% increase in thermal conductivity yielded an increase of

 

93

on

.Acofiumwfiam> Hovoev mm: undamam voamn
mammastMuomu mo Momum HHmEm m ca mafia m> ouSumquEwu kmm n.m muowfim

$33 oz...<m 28: m2:

“83:2

T. .\.\/ \/ L
.\ f \\\V/ / k <)(\ /\u

\‘

 

:3
F0
(0) aanlvaadwal

 

94

6.9% in estimated heat generation rate. Similarly, a 10%
decrease in thermal conductivity led to a 8.4% decrease in
estimated heat generation rate.

‘ Thermal conductivity of baled alfalfa hay was estimated
from empirical relationships (section 7.3.2). From this
estimation procedure, it was believed that the estimated
thermal conductivity is accurate to within 1 30%. With a
potential error in estimated thermal conductivity as high as
30%, the actual heat generation rate may deviate from the
estimated value by as much as 1 20%.

The close agreement between the experimental and
theoretical relationships between dry matter loss and heat
generation rate (section 7.4) supports the assumed values for
thermal conductivity. Heat generation rate was well within
120% of the value expected; therefore, the estimated values
of thermal conductivity were accurate.

To illustrate the usefulness of the hay stack heat
transfer model, several simulation trials of a large cubic
stack (1080 bales) were run. Simulation of a stack this size
yielded the same maximum internal temperatures as larger
(2000 bales) stacks. Therefore, a stack of 1080 bales was
used to minimize computer time.

Figure 8.8 gives the time temperature curves for three
different bale conditions. Ambient temperature was set to a
constant 20°C. The model presented does not consider
chemical reactions may which occur once the hay has reached

approximately 70°C. Therefore, any hay temperatures

 

95

 

.zm: mmamuam poamn >Humaswcmuomu
mo xomuw mwuma m c« mafia m> mwusumuooewu hm: wouoacmum m.w muowwm

A963 023%. 29: m2:

 

 

 

 

 

om mm o.“ m. 0.. .... m.
o
0..
on
non
AE :o\uv_ on; 83308 .02... R m— ..
:8
F -8
he .38. 9.5 23%.: .33 n 8 1o 5
10m
A5 8\9_ 85 23.35 Ba... x aw mom
loo.

(0) aanlvaadwal

96

exceeding 70°C indicate that combustion may occur later in
storage. With this in mind, the results in Figure 8.8 agree
quite well with discussion in the literature concerning safe
baling moisture limits. The model indicates that to assure
combustion will not occur, baling moisture must be limited to
approximately 22%.

Nutrient value decreases once temperatures exceed 60°C.
The model indicates that this happens at approximately 20%
moisture. Experience has shown that baling above 20%
moisture often leads to quality changes.

Finally, 18% moisture is generally considered to be the
limit if no quality changes or excessive dry matter losses
are to occur. With a maximum temperature of approximately
46°C predicted for a moisture level of 18%, this would be
true.

Heat generation rates for hay treated with propionic
acid can be estimated using equations (7.31) and (7.32).
Using this model to predict temperatures in a hay stack led
to results which were not expected. A simulation of hay
baled with a moisture level of 25% and a density of l60Kg/m3
indicated that propionic acid treatment at 1% of hay weight
reduced maximum temperature by 46°C. Propionic acid does
decrease temperatures but not by this magnitude. More work
should be done to evaluate the effects of propionic acid on
heat generation before a model of this type can be made

valid.

 

9. SUMMARY AND CONCLUSIONS

Hay baled wet provides an ‘environment conducive to
microbial growth. Wet hay also respires more than dry hay.
The combination of these two activities causes more dry
matter to be lost and temperatures to rise higher in wet hay
than in dry hay. As temperatures rise, nutrient degradation
occurs and the possibility of combustion increases. Baling
hay at moisture levels lower than 18% (wet basis) assures
safe storage and minimal nutrient change; but drying hay to
this level in the field results in considerable losses due to
leaching, respiration and mechanical handling.

Preservatives are added to baled hay to increase the
moisture limit for storage. Proven effective preservatives
are organic acids and anhydrous ammonia; acids being the most
common because they are safer and easier to apply.
Evaluation of preservatives is done by comparing material
coming out of storage which has been chemically treated with
the preservative to material coming out of storage which was
not treated. The results of these comparisons are limited to
the experimental conditions. To make the results applicable
to more situations, a more general approach is needed.
However, before preservatives Can be evaluated, we must have
a solid understanding of what happens during storage without
preservatives. .

This study was performed to model the effects of

moisture and density on dry matter losses and heating during

97

98

storage of rectangularly baled alfalfa hay. Three
experiments were performed in which a total of 37 treatments
of five bales each were placed in storage. Treatment
differences were in bale moisture and density levels.
Propionic acid was also applied to 5 treatments. Moisture was
varied from 11.5 to 48.0% wet basis; density was varied from
74 to 302 Kg/m3. Temperatures were measured every 6 hours
during the first 30 days in storage. Dry matter loss was
evaluated after a 60 day storage period. Effects of moisture
and density on temperatures reached in storage and dry matter
loss which occurs in storage were determined. Statistical
models of these effects were also developed. Models were
validated through comparison with hay storage data taken
previously.

Small stacks of five bales each were used in the
experiments. Temperature analyses of stacks this size cannot
be applied to larger stacks because small stacks can more
rapidly dissipate heat to their environment. A finite
difference heat transfer model was applied to small hay
stacks to predict heat generation rates. Regression
techniques were then used to develop a model which predicted
the heat generation rate of alfalfa hay based on moisture,
density, and time from baling. This model used in
combination with the finite difference heat transfer model
can be used to predict temperatures which occur in a hay
stack of any size. Modeling the storage process in this

manner removes some limitations in the process of comparing

 

99

alternative storage practices.

Conclusions regarding storage of rectangularly baled

alfalfa hay were:

1.

Dry matter loss was not affected significantly by bale
density (neither wet nor dry matter density) but was

significantly increased by increased moisture level. The
best fit model to predict dry matter loss from baling

moisture was:

DML = -5.4 + 48.0 * MI
Where:
DML = dry matter loss (% of initial)
MI = moisture content at baling (decimal wet

basis)

Storage temperatures were significantly increased by

increases in either moisture or wet density.

Dry matter loss was significantly related to storage
temperatures. Dry matter loss was proportional to mean

temperature and the square of maximum temperature.

Numerical methods (as opposed to an analytical solution)
must be used to model the heat transfer process for a
stack of hay because thermal properties change due to the
drying process. Also, solving an internal nodal point
finite difference equation for heat generation rate as a
function of current nodal temperatures and a known
“target" temperature is a reasonable method for

predicting heat generation rate in baled alfalfa.

__-..

 

 

100

Heat generation rate in rectangularly baled alfalfa hay
during storage was a function of moisture, density, and

time from baling. Heat generation rate reached a maximum

.approximately 8 days after baling and varied as the

square of moisture and the square root of density. The

best fit models for heat generation rate were:

For t 5 8:

62 = 2.47*M2 + 0.021*t + 0-0119*Dw‘5 — 0.307

For t 3 8:
62 = o.oooozse*(M*nw)2 - 0.005*t + 0.181*Dw'5 - 0.060
Where:

62 = heat generation rate (W/Kg)

M = moisture (decimal get basis)

Dw = wet density (Kg/m )

t = time from baling (days)
Propionic acid application at the time of baling
decreased dry matter loss approximately 1.3% for each
percent (of wet weight) of acid applied. The
application of propionic acid also significantly
decreased temperatures and heat generation rate during

storage. The decrease in heat generation rate was

greatest during the first several days of storage.

{‘1

 

10. REFERENCES

ASHRAE Handbook, Fundamentals. 1981. American Society of
Heating Refrigerating and Air Conditioning Engineers, Inc.
Atlanta, GA.

Andersen, S.A. 1950. Automatic Refrigeration. MacLaren and
Son Ltd. for Donfoss. Nordborg, Denmark.

Beck, J.V. 1977. "Sequential Estimation of Thermal
Parameters". Journal of Heat Transfer. 99(2)314-321

Bern, J.C. 1964. The Specific Heat of Alfalfa. M.S. Thesis.
University of Nebraska, Lincoln, Nebraska.

Bohstedt, G. 1944. "Nutrition Value of Hay and Silage as
Affected by Harvesting, Processing, and Storage".
Agricultural Engineering. 25(9)337-340.

Chakrabarti, S.M. and W.H. Johnson. 1972. "Specific Heat of
Flue Cured Tobacco by Differential Scanning Calorimetry".
Transaction of the ASAE. 15(5)928-931.

Davies, M.H. and 1.8. Warboys. 1978. "The Effect of
Proprionic Acid on the Storage Losses of Hay". Journal of the
British Grassland Society. 33:75-82.

Hall Carl W. 1980. Drying and Storage of Agricultural Crops.
pp 258-290. AVI Publishing Co. West Port, Connecticut.

Hathaway, John E. 1983. Preservatives for Baled Hay.
Unpublished research paper. Agricultural Engineering Dept.,
Michigan State University.

Hathaway, Koegel, and Straub. 1984. Requirements for Ammonia
Preservation of High Moisture Forages. ASAE Paper 84-
1536. Am. Soc. Agr. Engrs. St.Joseph, Michigan.

Hoffman, E.J. and M.A. Bradshaw. 1937. "Losses of Organic
Substances in the Spontaneous Heating of Alfalfa Hay".
Journal of Agricultural Research. 54(3)159-184.

Jiang, Shufen, J.C. Jofriet and 6.8. Mittal. 1985. Heat
Transfer Properties of Whole Plant Silage. ASAE Paper 85-
4005. Am. Soc. Agr. Engrs. St.Joseph, Michigan.

Johnson, J.C. Jr. and W.C. McCormick. 1976. "Influence of a
Preservative on Bermudagrass Hay Stored with Approximately 16
and 24% Moisture Content“. Journal of Animal Science.
42(1)175-179.

101

102

Jorgensen, N.A., P.R. Fritschel and G.P. Barrington. 1978.
Preliminary Report - Chemical Treatment of High-Moisture Hay.
Dept. of Dairy Science and Agricultural Engineering. Univ of
Wisconsin.

Knapp: W.H., D.A. Holt and V.L. Lechtenberg. 1975. "Hay
Preservation and Quality Improvement by Anhydrous Ammonia
Treatment". Agronomy Journal. 67(6)766-769.

Koegel, R.G., R.J. Straub and R.J. Shinners. 1983. Ammonia
Treatment of Large and Small Hay Packages. ASAE Paper 83-
1534. Am. Soc. Agr. Engrs. St.Joseph, Michigan.

Martin N.P. 1980. "Harvesting and Storage of Quality Hay".
New Developments in Forages. American Forage and Grassland
Council.

Miller, L.G., D.C. Clanton, L.F. Nelson and O.E. Hoehne.
1967. "Nutritive Value of Hay Baled at Various Moisture
Contents". Journal of Animal Science. 26(6)1369-1373.

Mohsenin, Nuri N. 1980. Thermal Properties of Foods and
Agricultural Materials. Gordon and Breach, New York.

Moser, L.E. 1980. "Quality of Forage as Affected by Post-
Harvest Storage and Processing". Crop Quality, Storage, and
Utilization. American Society of Agronomy.

Myers, Glen E. 1971. Analytical Methods in Conduction Heat
Transfer. McGraw-Hill. NY.

Nehrir, H., Kjelhard, P.M. Anderson, T.A. Long, L.D. Hoffman,
J.B. Washbo, L.L. Wilson and J.P. Mueller. 1978. "Chemical
Additives and Hay Attributes". Transactions of the ASAE.
21(2)217—221.

Nelson, Leon F. 1966. "Spontaneous Heating and Nutrient
Retention of Baled Alfalfa Hay During Storage". Transactions
of the ASAE. 9(4)509-512.

Nelson, Leon F. 1968. "Spontaneous Heating, Gross Energy
Retention and Nutrient Retention of Hi h-Density Alfalfa Hay
Bales". Transactions of the ASAE. 11(5 575-600,607.

Nelson, Leon F. 1972. "Storage Characteristics and Nutritive
Value of High-Density Native Hay Bales". Transactions of the
ASAE. 15(2)201-205,210.

Ott, L.E. and L.W. Horbut. 1964. Thermal Diffusivity of
Compressed Alfalfa Hay. ASAE Paper 64-817. Am. Soc. Agr.
Engrs. St.Joseph, Michigan.

 

103

Rotz, C.A., D.J. Sprott, J.W. Thomas and J.E. Hathaway. 1983.
Preservation of Baled High-Moisture Alfalfa Hay. ASAE Paper
83-1533. Am. Soc. Agr. Engrs. St.Joseph, Michigan.

Rotz, C.A., D.J. Sprott, R.J. Davis and J.W. Thomas. 1984.
Anhydrous Ammonia Injection Into Baled Forage. ASAE Paper 84-
1537. Am. Soc. Agr. Engrs. St.Joseph, Michigan.

Savoie P., C.A. Rotz, H.F. Bucholtz and R.C. Brook. 1982.
”Hay Harvesting System Losses and Drying Rates". Transactions
of the ASAE. 25(3 581-585,589.

Schaeffer, Craig C. and Neal P. Martin. 1979. Hay
Preservation. Ext. Folder 489. Agricultural Extension
Service. University of Minnesota. -

Shepherd, J.B., H.G. Wiseman, R.E. Ely, C.G. Melin, W.J.
Sweetman, C.H. Gordon, L.G. Schoenleber, R.E. Wagner, L.E.
Campbell, G.D. Roane and W.H. Hosterman. 1954. Experiments in
Harvesting and Preserving Alfalfa for Dairy Cattle Feed. USDA
Extension Bulletin #1079.

Siebel, E. 1892. "Specific Heats of Various Products". Ice
and Refrigeration. 2:256-257

Thomas, J.W. 1978. "Preservatives for Conserved Forage
Crops". Journal of Animal Science. 47(3)721-735.

Waldo, D.R. and N.A. Jorgensen. 1981. "Forages for High
Animal Production: Nutritional Factors and Effects of
Conservation". Journal of Dairy Science. 64(6)1207-1229.

Weeks, S.A., F.G. Owen and G.M. Petersen. 1975. "Storage
Characteristics and Feeding Value of Mechanically Stacked
Loose Hay". Transactions of the ASAE 18(6)1065-1069.