Purdue UniversityPurdue e-Pubs

Open Access Dissertations Theses and Dissertations

Fall 2013

An Analysis Of The Impact Of StorageTemperature, Moisture Content & Duration UponThe Chemical Components & Bioprocessing OfLignocellulosic BiomassArun AthmanathanPurdue University

Follow this and additional works at: https://docs.lib.purdue.edu/open_access_dissertations

Part of the Biochemistry Commons, Chemical Engineering Commons, and the Oil, Gas, andEnergy Commons

This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact epubs@purdue.edu foradditional information.

Recommended CitationAthmanathan, Arun, "An Analysis Of The Impact Of Storage Temperature, Moisture Content & Duration Upon The ChemicalComponents & Bioprocessing Of Lignocellulosic Biomass" (2013). Open Access Dissertations. 202.https://docs.lib.purdue.edu/open_access_dissertations/202

i

i

AN ANALYSIS OF THE IMPACT OF STORAGE TEMPERATURE, MOISTURE CONTENT & DURATION UPON THE CHEMICAL COMPONENTS &

BIOPROCESSING OF LIGNOCELLULOSIC BIOMASS

A Dissertation

Submitted to the Faculty

of

Purdue University

by

Arun Athmanathan

In Partial Fulfillment of the

Requirements for the Degree

of

Doctor of Philosophy

December 2013

Purdue University

West Lafayette, Indiana

ii

ii

Dedicated to my family: Appa, Amma and Kailash. This work would not have been

possible without your help, support and constant prodding.

iii

iii

ACKNOWLEDGEMENTS

I would like to thank my family for all their support through my academic life. My

parents and brother are responsible in no small part for any achievement I can claim.

My admission into Purdue University was due in part to efforts taken by Prof. Aiyagiri

Ramesh at the Indian Institute of Technology, Guwahati, and I would like to thank him

for it.

The present study was funded by the Midwest Centre for Bioenergy Grasses, through

USDA-NIFA Grant No. 2010-34328-21047, and I am grateful to the Centre for giving

me this opportunity to pursue graduate study.

I would like to thank my major professor, Dr. Nathan Mosier for his mentorship

throughout my graduate studies. Working with him has taught me about the principles of

scientific research, the means of effective scientific communication and the nitty -gritty of

academia and daily research, and is an experience I am deeply grateful for.

iv

iv

I am also grateful to Dr. Michael Ladisch, the director of LORRE, for his inputs at

critical junctures on scientific communication, data presentation and organization and the

professional conduct.

The present study is an interdisciplinary one, requiring an understanding of agronomy

and agricultural engineering. I am very grateful to Dr. Dennis Buckmaster and Dr. Keith

Johnson for their guidance and insights into these aspects.

I would like to thank the following people for their inputs on this project:

Dr. T. Kuczek at the Dept. of Statistics for his help analyzing the data

Dr. Okos and Dr. Weil for providing storage environments

Dr. K. Ileleji and Dr. R. Stroshine for helping with feedstock pre-processing

My work at LORRE has been alongside several wonderful people, whose help and

presence I am grateful for. In particular, Rick Hendrickson, Linda (Xingya) Liu, Barron

Hewetson and Tommy Kreke have all been great and helpful people to work along side.

Additionally, the help of Sarah DeFlora, Shuai Wang, Joshua Huetteman and Swetha

Vinjimoor has been invaluable. The present work would not have been possible without

their help.

v

v

TABLE OF CONTENTS

Page

LIST OF TABLES…… ........................................................................................................ vii

LIST OF FIGURES….. ...........................................................................................................ix

ABSTRACT…….……. ...........................................................................................................x

CHAPTER 1. LITERATURE REVIEW ......................................................................... 1

1.1 The Cellulosic Biorefinery ................................................................................ 1

1.1.1 Biomass: Components of Interest ......................................................... 5

1.2 Biomass Supply Systems ................................................................................. 11

1.2.1 Harvest operations ............................................................................... 12

1.3 Biomass Storage ............................................................................................... 17

1.3.1 Storage Background: Forage and Animal Feed ................................. 18

1.3.2 Storage Losses and Biomass Quality Changes .................................. 28

1.3.3 Storage Hypotheses ............................................................................. 33

CHAPTER 2. MATERIALS AND METHODS ........................................................... 36

2.1 Biomass Storage: Feedstock Preparation, Experimental Design & Storage Setup…………................................................................................................................... 36

2.1.1 Sweet Sorghum Bagasse: Aerobic and Anaerobic Storage............... 36

2.1.2 Corn Stover Storage............................................................................. 39

2.1.3 Switchgrass: Aerobic Storage ............................................................. 41

2.2 Sample Analysis:.............................................................................................. 49

2.2.1 Composition Analysis: ........................................................................ 50

2.2.2 Bioprocessing:...................................................................................... 54

2.2.3 High Performance Liquid Chromatography: ..................................... 58

vi

vi

Page

CHAPTER 3. RESULTS AND DISCUSSION ............................................................ 60

3.1 Effect of Storage Parameters on Biomass Losses during Aerobic Storage .. 60

3.1.1 Aerobic and Anaerobic Storage: Sweet Sorghum Bagasse............... 61

3.1.2 Aerobic Storage of Corn Stover.......................................................... 68

3.1.3 Aerobic Storage of Switchgrass.......................................................... 72

3.1.4 Conclusions: Storage Parameters and Dry Matter Loss .................... 85

3.2 Feedstock Bioprocessing: Impact of Storage Losses ..................................... 86

3.2.1 Pretreatment: Comparison of Stored and Pre-Storage Biomass ....... 87

3.2.2 Cellulose Hydrolysis............................................................................ 95

3.2.3 Conclusions: Bioprocessing .............................................................. 100

CHAPTER 4. CONCLUSIONS AND RECOMMENDATIONS ............................. 102

4.1 Conclusions .................................................................................................... 102

4.2 Recommendations: ......................................................................................... 107

BIBLIOGRAPHY……. ...................................................................................................... 111

APPENDICES

Appendix A Water Activity......................................................................................... 129

Appendix B Validation of Moisture Content & Glucose Concentration Determination Methods…………… .............................................................................. 142

Appendix C Switchgrass Wetting for Storage Under Dynamic Moisture Equilibrium ……………………………………………………………………………………….151

Appendix D Standardization of Liquid Chromatography Measurements ................ 155

VITA…………………........................................................................................................ 158

vii

vii

LIST OF TABLES

Table................................................................................................................................... Page

Table 2.1. Switchgrass Storage Moisture Concentrations ................................................... 44

Table 2.2. Solutions used in controlling water activity ....................................................... 48

Table 3.1. Sweet Sorghum Bagasse Storage: Dry Matter and Moisture Content Changes upon 24 Weeks of Aerobic and Anaerobic Storage ............................................................. 63

Table 3.2. Total Dry Matter and Component Losses upon 24 Weeks of Aerobic Sweet Sorghum Bagasse Storage at Low Moisture (12%) ............................................................. 64

Table 3.3. Total Dry Matter and Component Losses upon 24 Weeks of Aerobic Sweet Sorghum Bagasse Storage at High (26%) Moisture ............................................................ 65

Table 3.4. Total Dry Matter and Component Losses upon 24 Weeks of Anaerobic Sweet Sorghum Bagasse Storage ..................................................................................................... 66

Table 3.5. Total Dry Matter and Component Losses upon 8 Weeks of Aerobic Corn Stover Storage at High Moisture ........................................................................................... 70

Table 3.6. Total Dry matter and Component Losses upon Aerobic Storage of 34% moisture Switchgrass for 8 Weeks ........................................................................................ 78

Table 3.7. Total Dry Matter and Component Losses after Aerobic storage of 34% moisture Switchgrass for 16 Weeks...................................................................................... 78

Table 3.8. Statistical Analysis: Switchgrass Dry Matter Loss ............................................ 81

Table 3.9. Statistical Analysis: High Moisture Dry Matter Loss ........................................ 82

Table 3.10. Composition of Pre-storage Corn Stover: Before and After Liquid Hot-Water Pretreatment............................................................................................................................ 88

Table 3.11. Fractionation of Pre-Storage Stover Components upon Liquid Hot-Water Pretreatment............................................................................................................................ 89

Table 3.12. Composition of Post-storage Corn Stover (8 Weeks, 35 °C, 0.97 aw): Before and After Liquid Hot-Water Pretreatment ............................................................................ 90

Table 3.13. Fractionation of Stored Stover Components upon Liquid Hot-Water Pretreatment............................................................................................................................ 90

viii

viii

Table Page

Table 3.14. Composition of Pre-Storage Switchgrass Before and After Liquid Hot-Water

Pretreatment............................................................................................................................ 91

Table 3.15. Fractionation of Pre-Storage Switchgrass Components between Solids and Liquid during Liquid Hot-Water Pretreatment..................................................................... 92

Table 3.16. Composition of Stored Switchgrass Before and After Liquid Hot-Water Pretreatment............................................................................................................................ 93

Table 3.17. Fractionation of Stored Switchgrass Components between Solids and Liquid during Liquid Hot-Water Pretreatment................................................................................. 94

Table 3.18. 48-Hour Hydrolysis Yield of Pretreated Corn Stover...................................... 96

Table 3.19. 48-Hour Hydrolysis of Pretreated Switchgrass ................................................ 97

Table 3.20. Feedstock Particle-Size Distribution................................................................. 99

Appendix Table

Table A.1. Water Activity Ranges & Microbial Proliferation.........................................136

Table A.2. Water activities of some common foodstuffs .................................................. 137

Table B.1. Moisture Reading Comparison: ASABE Standard vs. Moisture Balance.... 144

Table B.2. T-test Comparison: Difference in Readings..................................................... 146

Table B.3. Regression Statistics: Moisture Content (ASABE) vs. Moisture Reading

(Balance)............................................................................................................................... 146

Table B.4. T-test comparison: Glucose Reading vs. Concentration ................................. 148

Table B.5. Regression Statistics: Glucose concentration vs. Glucose reading ................ 149

Table C.1. Pre-Storage Switchgrass Moisture Concentrations....................................... 153

Table D.1. Injection Variability during HPLC Analysis (LAP analysis)……………....156

Table D.2. Injection Variability during HPLC Analysis (Pretreatment analysis)............ 157

ix

ix

LIST OF FIGURES

Figure ................................................................................................................................. Page

Figure 1.1. Lignocellulose Components of Interest ............................................................... 4

Figure 1.2. Lignin Monomers................................................................................................ 10

Figure 2.1. Switchgrass Storage Experimental Design........................................................ 42

Figure 2.2. Humidity Control ................................................................................................ 47

Figure 2.3. Laboratory Analytical Procedure sequence developed by NREL ................... 53

Figure 3.1. Dry Matter Loss from Switchgrass Storage at <10 °C and varying water activities. Error bars indicate standard deviations................................................................ 75

Figure 3.2. Dry Matter Loss: Switchgrass Storage at a. 20 °C and b. 35 °C. Error bars indicate standard deviations................................................................................................... 76

Appendix Figure.........................................................................................................................

Figure A.1. Species Equilibrium between Phases……………………………………...133

Figure A.2. Water movement across Cell Membranes ...................................................... 135

Figure A.3. Equilibrium Vapor Pressures for a. Solution and b. Pure Water .................. 139

Figure B.1. Linear Regression: ASABE S358.2-determined moisture content vs. Moisture balance readings……………...........................................................................................147

Figure B.2. Linear Regression: Glucose Concentrations vs. Glucose Meter Reading .... 149

Figure C.1. Switchgrass Water Activity Isotherms........................................................ 152

x

x

ABSTRACT

Athmanathan, Arun. Ph.D., Purdue University, December 2013. An Analysis of the Impact of Storage Temperature, Moisture Content & Duration upon Chemical

Components & Bioprocessing of Lignocellulosic Biomass. Major Professor: Nathan Mosier.

The successful utilization of lignocellulosic biomass as a feedstock for fuels and

chemicals necessitates storage for 2-6 months. It is correspondingly important to

understand the impact of storage parameters – moisture concentration, temperature and

duration – on biomass quality.

As aerobic storage is the most viable large-scale solution, aerobic storage experiments

were carried out with three projected bioenergy feedstocks – sweet sorghum (Sorghum

bicolor) bagasse, corn (Zea mays) stover and switchgrass (Panicum virgatum). Stored

samples of each were examined for dry matter loss and composition change to develop a

material balance around carbohydrates and lignin.

A mean dry matter loss of 24% was observed at 8 weeks in sweet sorghum stored at high

moisture content (26% w/w). Soluble sugars predominated the dry matter loss in high-

moisture sorghum, causing an increase in the mass fraction of lignin in the biomass.

xi

xi

In comparison, low-moisture (12% w/w) samples showed negligible loss. High-moisture

sorghum dried from 26% to 20% in 8 weeks, and further to 8% in 24 weeks.

To control moisture loss in subsequent experiments, switchgrass and corn stover were

wetted to specific water activities (aw) and stored under equivalent controlled humidities.

Switchgrass stored at water activities of 0.65-0.85 showed no dry matter loss, irrespective

of storage temperature or duration. Switchgrass stored at a water activity of 0.99 (~33%

w/w moisture) showed significant dry matter loss at temperatures of 20 and 35 °C,

although substantial sample-to-sample variation was observed at 20 °C compared to

35 °C. At 8 and 16 weeks, switchgrass stored at 0.99 aw and 35 °C lost 7% dry matter on

average. Corn stover was stored under similar conditions (35 °C, 0.97 aw) for 8 weeks.

The stored samples showed 10% dry matter loss on average. The dry matter losses

predominantly consisted of cellulose and hemicellulose, resulting in the mass fraction of

lignin increasing in switchgrass from 26.6% to 27.0% in 8 weeks and 28.8% in 16 weeks,

and in stover from 23% to 25% in 8 weeks.

Samples of switchgrass and corn stover that had undergone significant (>5%) dry mater

loss were subjected to liquid hot-water pretreatment and washed solids subjected to

enzymatic hydrolysis. The glucose yield was calculated and compared to hydrolysis

results from samples that had not experienced significant loss to examine the impact of

storage losses on carbohydrate extractability. Both degraded and non-degraded

switchgrass achieved approximately 8% of theoretical yield from cellulose hydrolysis

before pretreatment, and 35% yield after pretreatment. Similarly, irrespective of storage

treatment, corn stover samples showed 16% hydrolysis yield before pretreatment and

xii

xii

approximately 55% yield after pretreatment. Due to the similar yields, storage losses in

the range observed were concluded to not significantly impact carbohydrate extractability

in terms of percent yield, although the amount of extractable carbohydrate per dry kg

biomass is reduced by 4-16% due to their preferential consumption.

1

1

CHAPTER 1. LITERATURE REVIEW

1.1 The Cellulosic Biorefinery

The convergence of biotechnology, nanotechnology, chemistry, agronomy and industrial

engineering has generated a new industrial paradigm, wherein these sciences can be used

in combination to manufacture - in the field or the factory - low cost, low value products

that are sold and used in bulk, from biological materials. These products are aimed to

supplement and eventually replace existing commodity products - foods, value-added

chemicals, fuels and material building blocks - and given their origin, have been termed

'bio'-commodities. The term 'biocommodity engineering' (Lynd et al., 1999) has been

coined to describe the paradigm.

The most notable thrust of biocommodity engineering has been towards the replacement

of fossil feedstocks - coal, oil and natural gas. These resources are required for the

manufacture of hydrocarbons for fuels and important value-added chemicals. The

worldwide demand for fossil feedstocks has been steadily increasing, giving rise to

political and economic issues. Moreover, the consumption of these feedstocks has led to

the atmospheric release of carbon formerly trapped within the earth, causing worldwide

climate change. A solution to this problem would be to use biological materials - termed

biomass - as renewable feedstocks that (as they are generated by trapping carbon from the

2

2

air) are also environmentally benign. The usage of ethanol generated from sugarcane

(Brazil) or corn starch (USA) as a fuel is a step in this direction, the two being termed

biofuels.

The effective replacement of fossil feedstocks requires the utilization of whole biomass,

rather than single components (starch, oils, free sugar etc.). Given the heterogeneous

composition of biomass and the applications that its components could have, conversion

of whole biomass into fuels and chemicals would require its deconstruction into specific

fractions and their subsequent transformation into the chemicals of interest, analogous to

crude oil transformation in a petroleum refinery. The term 'biorefinery' has been coined to

describe a facility envisioned as capable of transforming whole feed biomass into

multiple products (Lynd et al., 1999). The most abundant biomass available being

lignocellulose (Perlack et al., 2005), the constituent of plant structural tissues,

biorefineries must center their functioning on usage of lignocellulosic feedstocks.

The primary requirements of a biological feedstock are abundance and carbon-neutrality.

The biomass must be cultivable without significant environmental fallout from the

corresponding land-use change (Fargione et al., 2008; Popp et al., 2012; Searchinger et

al., 2008) and abundant enough to provide the necessary components for industrial scale

fuel and chemical manufacture. In the context of agriculture in the Midwestern United

States, grass-based forages are important lignocellulosic feedstocks. Three significant

forages are described below:

3

3

i. Sweet sorghum: Sorghum (S. bicolor) is an important cereal crop, grown across

the world for food grain, feed grain or forage. Sweet sorghums are specific variants

that accumulate substantial sucrose in the stalk, and have been cultivated as

alternatives to sugarcane or sugar beet. While its cultivation in the United States has

traditionally been for forage, sweet sorghum has gained substantial attention in

recent times as a feedstock for biofuels (Byrt et al., 2011). The key advantages of

using sweet sorghum (S. bicolor L. Moench) as a bio-based feedstock are its high

concentration of easily extractable and fermentable sugars, and its high tolerance of

drought-like conditions (Rooney et al., 2007). Sweet sorghum cultivation in the

United States has been reported as capable of producing 11-48 dry-Mg.hectare-1

(Erickson et al., 2011; Han et al., 2012; Miller & Ottman, 2010; Wortmann et al.,

2010) with corresponding sugar yields of 1.9 – 6.2 Mg.hectare-1

. Sweet sorghum

thus presents a viable source of sugars for ethanol manufacture.

ii. Corn stover: Corn (Z. mays) stover comprises the stalks, leaves, husks, and cobs

left over from corn grain production (Shinners & Binversie, 2007). Stover is

composed on average of 32% cellulose and 24% hemicellulose by dry weight

(Templeton et al., 2009). The United States has been estimated as capable of

producing 133-196 Teragrams of stover for conversion into biofuel (Graham et al.,

2007; Kim & Dale, 2004; Tan et al., 2012), making it an important source of

cellulosic sugars. Moreover, due to the established framework for the cultivation

and harvest of corn in the Midwestern United States, the potential exists for large-

scale stover collection for biofuel production. Corn stover thus constitutes an

important feedstock for manufacture of cellulosic biofuels.

4

4

iii. Switchgrass: Switchgrass (P. virgatum) is can be cultivated in many areas of the

United States and is capable of growing on marginal lands (Evanylo et al., 2005;

Varvel et al., 2008) or with lower nutrient inputs than more intensive crops like

corn (Sokhansanj et al., 2009). It is also a high-yielding perennial crop, McLaughlin

et al (2005) reporting the average dry matter yield of switchgrass across the United

States to be 11.2 Mg.hectare-1

and range from 4.5 to 23 Mg.hectare-1

depending

upon the region. Switchgrass contains 33.6% cellulose and 27% hemicellulose on

average (Based on composition data from Dien et al., 2006; Yanet al., 2010; Hu et

al., 2010; Liu et al., 2010; Woodson et al., 2013), making it an important source of

fermentable sugars for conversion into biofuels and biochemicals.

Given a conversion process, feedstock quality is determined through the extractible yield

(kg/dry-kg biomass) of the components of interest (Figure 1.1). It is thereby necessary to

review the common fractions of interest within lignocellulosic biomass, their potential

uses, and the ease of extracting them.

Figure 1.1. Lignocellulose Components of Interest

5

5

1.1.1 Biomass: Components of Interest

As described above, carbohydrates and lignin are the principle components of interest

within lignocellulosic biomass, being the fractions containing the most carbon and least

nitrogen. Due to their different chemical natures, their extraction and deconstruction

require different approaches. Presented below is a summary of the chemical properties,

uses and methods of extraction of each fraction.

1.1.1.1 Carbohydrates:

Carbohydrates are the primary platform for bio-based manufacture of fuels, where they

are metabolized by biocatalysts into fuel alkanols – ethanol and butanol. Carbohydrates

can also serve as feedstocks for the thermocatalytic manufacture of value added

chemicals and precursors (Bozell & Petersen, 2010). Plant carbohydrates can be

classified as structural or non-structural, depending on their location and function. Based

on this categorization, different strategies have to be adopted towards their extraction and

usage.

Non-structural carbohydrates serve largely as energy reserves for the plant. They

correspondingly tend to be monomers or easily degradable polymers. Their location

within the plant tissue depends strongly upon their size and polymerization. The major

carbohydrates of interest are free sugars and starch. The former comprise mono- and

disaccharides of glucose and fructose and are primarily located in plant sap. The sugars

are circulated as energy sources within the plant body, and so are highly soluble. Free

6

6

sugars can be extracted under mild conditions – mechanical treatment, water addition –

and subsequently fermented. They are important fractions on a dry weight basis of such

crops as sugarcane, sugar beet and sweet sorghum (Peters, 2007). Fuel ethanol is

commonly generated through the fermentation of either whole crop free sugar content, as

done in Brazil, or specific extracts such as molasses.

In contrast to free sugars, starches are glucose-based polysaccharides, made up of two

repeating units: amylose and amylopectin. Starches are synthesized for energy storage,

and can be found in fruits, seeds, growing tips and tubers or rhizomes. They are important

components of cereal crops such as rice, wheat, maize and oats, as well as tuber crops

such as potato and cassava. Corn starch is the feedstock for the manufacture of ethanol in

the United States. Starch extraction requires physical treatment to make the fraction more

accessible to subsequent separation processes. Extracted starch must then be

depolymerized chemically or enzymatically into glucose, upon which it can be fermented.

While not as vulnerable as free sugars, starches are easier to access and break down than

structural carbohydrates.

As the name suggests, structural carbohydrates make up and provide mechanical support

for the plant structure. Located primarily in plant cell walls, they are sugars with a very

high degree of polymerization. They are bound with lignin, which provides both

structural integrity and shields them from chemical and biological action. Structural

carbohydrates are present in all plants. While the precise composition varies across

7

7

species, family and age, they can be in turn be categorized into two particular families,

namely cellulose and hemicellulose.

Cellulose is a glucose polymer, made up of ß-(1,4)-linked D-glucopyranose molecules.

Cellulose makes up 15-30% of primary plant cell walls and up to 50-60% of the

secondary cell wall (Vermerris, 2008). The linked molecules are aggregated into micro-

and eventually macrofibrils. Cellulose is the most abundant biological polymer on Earth

and constitutes 37.5-45% by dry weight of grass-based feedstocks as corn stover, wheat

straw, switchgrass and cane bagasse and 46-49% by dry weight of woody feedstocks such

as pine wood and poplar (Mosier et al., 2005c; Saha, 2003).

Hemicellulose is the second carbohydrate that makes up plant cell walls. Hemicellulose

has a significantly different structure from cellulose, and shows more variation in

composition and structure across plant classes and species. While cellulose is a

homogeneous polymer of glucose, hemicellulose consists of several different monomeric

units. Hemicellulose contains hexoses (glucose, galactose and mannose), pentoses

(xylose and arabinose) and sugar acids such as glucuronic acid. Listed below are some

examples of hemicellulosic polysaccharides:

a. Xyloglucans: These are linear chains of ß-(1,4)-linked D-glucopyranose residues with

various side chains, xylose and arabinose being the most common.

b. Arabinoxylans: These are polymers with a backbone of xylopyranosyl molecules,

linked through ß-(1,3) or ß-(1,4)-linkages. The backbone can contain arabinose,

8

8

glucuronic acid and 4-oxy methyl glucuronic acid substitutions (arabinoxylans,

glucuronoarabinoxylans and 4-O-methyl-glucuronoxylans)

c. Mannans: Mannans have a backbone containing mannose. The backbone residues can

be substituted with glucose (glucomannans), galactose (galactomannans) or both

(glucogalactomannans)

Hemicelluloses comprise 25+% by dry weight of grass-based feedstocks such as corn

stover (25-28%), switchgrass (30-36%) and wheat straw (50%) (Byrt et al., 2011; Saha,

2003). They are also important components of woody biomass, comprising 14% by dry

weight of willow and poplar (Byrt et al., 2011).

Structural carbohydrate extraction is difficult, as the polysaccharides are bound to each

other and to lignin. A severe thermal and/or chemical treatment is required to fractionate

them from lignin and each other, and render them vulnerable to depolymerization. This

step, termed pretreatment is a critical step in biofuel manufacture (Yang & Wyman,

2008), and must be optimized for maximum carbohydrate recovery (Mosier et al., 2005b).

Cellulose shows the most resistance to depolymerization, termed recalcitrance.

Hemicelluloses are easier to extract and break down in comparison to cellulose, although

severe treatment is still required to separate them from lignin (Vermerris, 2008). The

rate- and viability-determining step in bio-based conversion of lignocellulose is the

separation of structural polysaccharides from lignin (Yang & Wyman, 2008). Following

their separation, cellulose and hemicellulose are depolymerized through enzyme addition

– cellulase and xylanase respectively. The generated monomers can then be fermented or

9

9

fed into specific conversion processes. The principal factor complicating their extraction

is the presence of lignin, which is described below.

1.1.1.2 Lignin:

Lignin is a phenylpropanoid polymer that provides a hydrophobic surface for intercellular

water conduction, shields cellulose and hemicellulose from enzymatic or chemical

attacks, and together with the two provides structural support for the plant body as a

whole. Lignin is made up of monomers derived from phenylalanine, termed monolignols.

The three monolignols are sinapyl alcohol, coniferyl alcohol and paracoumaric acid

(Figure 1.2.). Lignin is significantly less oxygenated and thus denser in energy content

than biomass carbohydrates. It thus constitutes an important feedstock for combustion

and catalytic manufacture of drop-in biofuels (Jae et al., 2010; Mendu et al., 2011; 2012;

Vermerris, 2008). During bio-based fuel manufacture however, lignin is problematic as it

shields structural carbohydrates, preventing extraction and depolymerization. Its presence

necessitates a severe physicochemical treatment, termed pretreatment, to fractionate it

from structural carbohydrates (Chapple et al., 2007; Mosier et al. 2005c). Moreover, the

pretreatment process in several cases breaks it down into phenolic monomers and

oligomers, which are inhibitory to both the carbohydrate-depolymerizing enzymes (Kim

et al., 2011b; Ximenes et al., 2011) and the fermenting microorganisms (Palmqvist &

Hähn-Hägerdal, 2000).

10

10

Extensive research has been carried with an aim to optimizing the fractionation of

biomass and subsequent conversion into fuels and chemicals. However, for the

biorefinery to be commercially viable it is also necessary that the feedstock supply be

cost-effective. Feedstock costs account for 35 – 50% of the total production cost of

cellulosic ethanol, as reported by the NREL (Foust et al., 2007) and Hess et al (2007).

These include the costs of cultivating biomass (Ericsson et al., 2009), packaging and then

transporting it to the biorefinery (Brechbill et al., 2011), while also adjusting for quality

changes during the process (Hess et al., 2007; Inman et al., 2010; Richard, 2010).

Detailed knowledge is therefore required as to the cultivation, harvest, densification and

transportation of biomass, so as to optimize the sequence of operations for minimum cost

Sinapylalcohol

Coniferylalcohol

Paracoumaricacid

Figure 1.2. Lignin Monomers

11

11

and quality changes. It is therefore necessary to develop knowledge of feedstock logistics

and supply networks.

1.2 Biomass Supply Systems

The effective usage of a particular feedstock for biofuel manufacture requires the

development of detailed feedstock logistics, covering feedstock harvest and

transportation to the biorefinery. An effective feedstock supply system must include all

unit operations from the field to the first stage of chemical treatment in a conversion plant.

Such an integrated system allows the optimal arrangements and combinations of unit

operations to reduce feedstock costs and increase overall value to both the refinery and

the cultivation system. The objectives in developing an integrated supply system are

threefold (Sokhansanj & Hess, 2009):

1) To select or develop technologies to reduce the costs of specific unit operations.

2) To optimize the sequence of operations the involved equipment for minimal

cost

3) To deliver a higher value biomass feedstock that product that improves product

yields and efficiencies.

The first requirement of a supply system is an estimation of the amount of feedstock

available for conversion. Feedstocks consist of both energy crops grown for the purpose

of conversion (switchgrass, miscanthus, cane for ethanol and woody biomass) and crop

residues, remnants of plants grown for food (corn, wheat), animal feed (corn, alfalfa),

fiber (cotton) or paper and timber (wood). It is important to estimate the amount (dry

12

12

weight/hectare) of biomass produced, the specific sections of the crop gathered and

corresponding compositions (Akin et al., 2006; Li et al., 2012). In recent times, certain

residues have been observed to improve soil quality and sequester carbon, making their

removal and utilization unsustainable in the long run (Hammerbeck et al., 2012;

Kochsiek & Knops, 2011). Such information has to be taken into account when

estimating the amount of biomass removable and the components of interest within it.

The subsequent requirement is a detailed knowledge – both technical and economic – of

the operations involved in harvesting a crop. Presented below are the various unit

operations that comprise the harvesting process, together with their specific impacts on

process feasibility.

1.2.1 Harvest operations

Biomass harvest comprises the sequence of operations from crop removal to its storage.

Harvest operations are rate- and feasibility-determining steps for the successful

functioning of the biorefinery, as they determine net availability, quality and cost of the

feed biomass (Ma, 2012; Petrolia, 2008). While there are variations based on crop species

and biomass product (haylage, silage etc.), harvest is made of the operations listed below.

1.2.1.1 Mowing and conditioning:

Mowing is the process of the cutting the s tanding plant in the field. Living matter being

typically close to 70% moisture, substantial drying is required to successfully move and

13

13

store the crop. Field drying, utilizing sunlight and wind to dry the crop, is the

predominant means due to its cost and the scale involved. The machines used for the

cutting are called mowers, the most common types being sickle cutterbar and rotary disc

mowers (Rider et al., 1993).

Conditioning refers to a series of operations performed after mowing, to quicken the

drying process. Chemical conditioning uses sodium or potassium carbonate to speed the

drying process. Mechanical conditioning, which is more common, involves mechanically

bruising, cracking and crushing the cut crop, to increase accessibility of plant tissue to

open air and sunlight. Mechanical conditioners are of three types, namely crimpers,

crushers and impellers (Rider et al., 1993). Modern machinery can combine the two

operations, chopping biomass and bruising it to expedite drying. Following the two, the

cut crop is arranged into denser collections called windrows or swaths, wherein it is set to

dry in the field.

1.2.1.2 Gathering and densification:

Cut, dried biomass is typically of low density. For example, leading feedstocks such as

corn stover and switchgrass were reported to have densities of 71 kg/m3 and 60 – 80

kg/m3 respectively (Shinners et al., 2007b; Sokhansanj et al., 2009). The material must

therefore be collected and densified into units with higher mass and suitable dimensions

to expedite downstream processes. Biomass may be further cut or chopped to bring its

particle size within acceptable limits.

14

14

Following biomass collection, the most common farm-based densification technique is to

package the feedstock into square or round bales. The baling machines can also wrap the

compressed biomass in twine, plastic or net wrap, for additional protection. Alternately,

the collected feedstock can be packed into loaves or stacks, the procedure termed loafing

or stacking. Woody feedstocks such as wood chips, sawdust and logging residues are

densified through pelletization, a process that has since been carried out with grass based

feedstocks as well (Mani et al., 2006; Tumuluru et al., 2011). Depending upon feedstock

species, composition, moisture content and densification technique, different final

densities are achievable. Sokhansanj reported densities of 140 – 180 kg/m3 for baled

switchgrass, the latter a suitable density for transporting distances of up to 160 km

(Sokhansanj et al., 2009).

The yield or efficiency of biomass collection (Mass gathered/Mass present on field) is an

important parameter, determining feedstock cost and thereby cultivation feasibility.

Shinners et al reported harvest yields of 37 – 55% with corn stover, depending on

whether the stover was chopped, baled dry or baled wet (Shinners et al., 2007b). Shinners

also reported stover gathering as the limiting step, determining yield per area. Similarly,

Monti et al reported harvest losses of 36 – 46% of the potentially harvestable biomass

when harvesting switchgrass (Monti et al., 2009). The losses were attributed both to

fractions of the biomass being uncut and to cut biomass not being gathered by the

harvesting machinery, the latter made worse in case of adverse weather.

15

15

1.2.1.3 Transportation:

The packaged feedstock must then be transported to a location for storage, and

subsequently to the refinery for breakdown. Biomass can be transported by road (trucks),

rail, waterway (barge) or through pipeline (Kumar et al., 2005). Transportation costs can

account for up to 46% of total cost of biofuel usage (Morey et al. 2010), and are thus

influential in selecting the location, scale and overall profitability of biorefineries

(Brechbill et al., 2011; Stephen et al., 2010; Suh et al., 2011; Sultana & Kumar, 2012).

Several studies have been published with formulae for evaluating transportation costs on

a per mass and per distance basis (Kumar et al., 2005; Sokhansanj et al., 2002), as well as

selecting the optimal means for it (Kumar et al., 2006).

1.2.1.4 Storage:

The growth and harvesting processes for plant-based biomass must by necessity be

seasonal, similar to conventional agriculture or forestry. As many feedstocks are farm or

forestry residues, their cultivation and harvest follows that of the main crop. However,

successful biorefinery operation necessitates a continuous supply of feed biomass.

Harvested densified biomass must thus be stored for mid- to long-term periods either at

the farm or a designated storage site. The focus of this study is on the impact of storage

on biomass quality.

16

16

1.2.1.5 Optimization:

Each operation described above must be optimized so that the resulting sequence can

produce the most biomass at the least cost. For example, Shinners et al reported a trade-

off between stover yield, fuel consumption and productivity (on a per area basis) when

gathering corn stover in (Shinners et al., 2009a), an optimal yield achievable by

modifying harvester configuration and separating grain and stover harvests (Shinners et

al., 2012). Multiple studies have analyzed and attempted to optimize the harvest sequence

for specific feedstocks such as corn stover (Petrolia, 2008; Sokhansanj & Turhollow,

2002; Sokhansanj et al., 2010; Suh et al., 2011) and switchgrass (Mitchell et al., 2008;

Perrin et al., 2008; Schmer et al., 2010; Sokhansanj et al., 2009) so as to deliver biomass

to the biorefinery at minimal cost. The scale of the manufacturing must also be taken into

account when optimizing the process, as reported by Suh et al when comparing the means

for densification and transportation of corn stover in Minnesota (2010).

In addition to quantity, it is important that the feedstock being delivered retain the quality

originally identified. It is thereby important to understand the impact of each step on the

quality of the feedstock. As discussed earlier, quality is determined by the extractible

yield (g/g dry) of the various components of interest, which is in turn affected by

composition and composition change. It is thereby important to understand the

composition changes that the feedstock can undergo while moving through the supply

chain. This study analyzes the change in composition and correspondingly quality in

biomass upon storage.

17

17

1.3 Biomass Storage

As described earlier, successful biorefinery operation requires long-term storage solutions

that can preserve the crop and possibly improve its digestibility at minimal cost. The

principle problem of storage is that biomass, being organic, is susceptible to microbial

activity under the right conditions of temperature, moisture and exposure. Microbial

respiration consumes organic carbon – from carbohydrates and lignin – and generates

carbon dioxide, methane other metabolic end products and heat. The process can become

self-sustaining beyond a threshold, wherein the heat generated sustains further microbial

growth and activity, causing extensive loss of carbon (Wihersaari, 2005; Williams, 1997).

The loss of carbon reduces the amount of fuel that can be manufactured from the biomass,

reducing product volume and yield, and increasing production costs. Moreover, the

released carbon dioxide, methane and nitrous oxide (generated from biomass protein) are

greenhouse gases, and their emission reduces the environmental sustainability of biofuel

manufacture (Emery & Mosier, 2012; Wihersaari, 2005). The problem is amplified by

scale, which following cultivation can range from several hundreds of kilograms to

several metric tons. A biorefinery drawing feedstock from multiple sites of cultivation

must have a storage method that‘s effective and economical at a large scale. It is therefore

important to ascertain the impact of storage parameters – temperature, the presence of

moisture and oxygen, exposure to sunlight, wind, rain and soil – on the chemical

composition of biomass.

18

18

While studies have recently been published discussing storage effects on biomass (Xiao

He et al., 2012; Shinners et al., 2007b; 2010; 2011), the impact of storage on overall

feedstock quality represents a knowledge gap. It is therefore necessary to examine

knowledge existing on storage methods, and their impact on the quality of biomass.

1.3.1 Storage Background: Forage and Animal Feed

The storage of biomass on a large scale is a problem that has been tackled by the forage

and animal feed industry, which sought to harvest and store cellulose-rich feeds –

predominantly grass-based – at minimal cost, while still retaining nutritive value. Given

that the majority of feeds are grass-based and often agricultural residues, the methods and

issues of forage storage are likely to recur when harvesting and storing them for biofuel

manufacture. Moreover, structural carbohydrates are an important component of forages,

as ruminants are capable of breaking them down. Cellulose is broken down in the animal

rumen through the combined action of ruminal enzymes and the microbial flora present.

The aqueous environment, mild conditions of temperature and pH and dependence upon

enzyme and microbial action make ruminal cellulose digestion similar to enzymatic

cellulose digestion. Thus, the digestibility and nutrient content – specifically

carbohydrate content – of a forage/forage-like feedstock could indicate its suitability for

bio-based breakdown, the two correlated by Lorenz et al (2009) and Anderson et al

(2009). Another industry involving long-term biomass storage is the pulp and paper

industry, which necessitates similar storage of woody biomass. Pulp can also serve as a

background for biomass storage, especially as woody biomass is an important projected

feedstock.

19

19

Described below are the methods used in the forage industry to store biomass, and the

parameters affecting effective storage in each case. The principle factor dividing the

various means of biomass storage is exposure to the air. This exposure simultaneously

enables microbial respiration and biomass drying, making preservation harder to predict.

On the basis of exposure, two main forms of storage emerge, namely dry storage and

ensilage or anaerobic storage. A description of each is presented below.

1.3.1.1 Dry storage:

Dry storage, also referred to as dry baling or hay storage, establishes low moisture

concentrations to prevent microbial activity. Dry storage is commonly carried out with

forages such as alfalfa, ryegrass, bromegrass, orchardgrass and corn stover.

The critical step in dry storage is the crop drying process, commonly carried out in the

field. At the time of harvest crop is mowed and conditioned and moved into swaths to

begin field drying. The swath is subsequently manipulated to ensure uniform and speedy

drying. Three possible operations can be carried out to improved drying:

1. Tedding: Using rotating tines, the swath content can be stirred and spread,

facilitating drying.

2. Swath inversion: Swath inverting machines pick up and turn over the swath,

exposing the wetter base contents to sunlight and air.

3. Raking and merging: Using mechanized rakes, the swath contents can be shifted

and mixed, exposing wetter material for drying.

20

20

Final moisture contents at or below 20% are desirable for stability (Rotz & Muck, 1994;

Rotz & Shinners, 2007). Once these moisture levels are reached, the forage is then

packaged into bales, by gathering the dried swaths or windrows and compressing their

contents into individual units of specific dimensions. The machines carrying out the

operation are called balers. Most hay in North America is packaged into large round bales,

90 – 180 cm in diameter and 120 – 160 cm in height, containing 200 – 900 kg dry matter.

Round bales must be wrapped in twine or plastic mesh at the time of preparation, which

adds stability but slows down the baling process. Alternately, the forage can be converted

into rectangular bales. Rectangular bales can be made small (25-35 kg), midsize (500 kg)

or large (1000 kg), the size chosen based on the harvest size and availability (or lack

thereof) of corresponding equipment and manual labor. Large rectangular bales are most

common in hay markets, especially in the western United States (Rotz & Shinners, 2007).

Some dry matter loss occurs during the baling process, biomass either being left on the

ground (pickup loss) or in the baling chamber (chamber loss). Losses are dependent on

the baler design used, the bale shape and mass, and hay moisture content, chamber losses

increasing with reduction in moisture content. Aside from bales, hay can also be packed

into pellets, stacks or loaves (Tumuluru et al., 2011). The packed biomass is then stored.

The moisture content is the main parameter determining storage stability. Biomass has

been reported as stable when stored at moisture content below 15-20% by weight (Rotz &

Muck, 1994; Rotz & Shinners, 2007). Field-drying being the only cost-effective option

upon large-scale crop harvests, the efficacy of drying is weather dependent. In case of

21

21

insufficient or non-uniform drying, baling and storage may be carried out at moisture

contents higher than the feed threshold stability. Biomass stability is then dependent upon

the efficacy of moisture loss during storage. As moisture content increases beyond 20%,

microbial activity becomes significant. At sufficiently high moisture levels microbial

respiration can generate sufficient heat to sustain itself as well as further damage the

stored material, the subsequent impact dependent on the density of dry matter and

packing features (Coblentz & Hoffman, 2009a; 2009b; Coblentz et al., 1996; 1998;

Martinson et al., 2011; Scarbrough et al., 2005; Turner et al., 2002). Similar results have

been reported when storing pulps and woody biomass, high moisture content enabling the

generation of heat as well as degradation products such as methane (Xiao He et al., 2012;

Jirjis, 1995; Larsson et al., 2012; Nurmi, 1999; Wihersaari, 2005). While the moisture

concentration (g/g or % by weight) within biomass was considered the sole parameter,

recent studies by Igathinathane et al have focused on the impact of humidity during

storage and the water activity of the stored material (Igathinathane et al., 2005; 2007;

2008; 2009), which may be expected to affect feed drying rates and correspondingly

microbial activity.

The stored biomass can be wetted by precipitation, or exchange moisture with the air and

soil. Its exposure to the environment is thus another important parameter in storage

stability. Location is the prime factor. The packed biomass can be stored both indoors –

barns, storage sheds etc. – and outdoors on the field. Biomass stored outdoors is exposed

to greater extremes of temperatures and precipitation, making location important.

Multiple studies have reported the greater stability of biomass when stored indoors

22

22

(Coblentz, 2009; Collins et al., 1995; Shinners et al., 2007b; 2009b; 2010). Another

determinant of exposure is contact with the soil. Bales can be stored on the ground, being

in extended contact with the soil. Alternately, they can be stored on modified surfaces

such as crushed rock and gravel, or on elevated surfaces (pallets) built for the purpose of

keeping the biomass away from the soil. Depending on its composition, the soil can be a

reservoir of moisture, and biomass stored in contact with it can absorb the same, affecting

storage stability. When comparing storage surfaces, pallets have been found to be suitable

for biomass stability, followed by crushed rock or gravel and finally regular soil

(Coblentz, 2009; Russell et al., 1990; Shinners et al., 2007b; 2009b).

Exposure to weather and soil can be offset to different extents by wrapping the bale with

twine, plastic mesh or sheet plastics such as tarpaulin. The wrapping helps maintain bale

structure and limits its contact with the environment. Sheet plastics provide the best

coverage, plastic-wrapped bales consistently showing less dry matter loss than twine or

mesh-wrapped bales (Shinners et al., 2009b; 2009c) . As crop scale increases however,

the cost of the plastic reduces overall profitability.

While low moisture content is the chief agent behind the preservation of the forage

material, preservative compounds can be added in case of insufficient drying. Two kinds

of preservative agents can be added. Inhibitor compounds restrict microbial respirat ion

and correspondingly dry matter loss, allowing for safe storage even if the hay is wetted

after baling (Pitt, 1990). Inhibitors include propionic acid and its salts or mixtures with

acetic acid, ammonia and urea. Alternatively, preservatives can be drying agents. Sodium

23

23

and potassium carbonate are the most commonly used agents. These chemicals are

sprayed on forage at the time of mowing, to enhance its drying. Both act on the wax

cuticle on plant surfaces, reducing its resistance to drying.

Dry storage depends upon the absence of moisture for effective preservation. An

alternative is to limit the presence of oxygen, which is also vital for sustained microbial

activity. This is carried when carrying out ensilage.

1.3.1.2 Ensilage:

Aeration being necessary for continued microbial respiration, biomass can be stored wet

when sealed away from air, either by wrapping in plastic or by storing in sealed locations.

Ensilage or anaerobic storage is carried out at much higher moisture contents in

comparison to hay storage, and is carried out when environmental conditions make large-

scale forage field drying difficult. The stored product is referred to as silage, the name

derived from silos, the most commonly used storage locations for it. A combination of

anaerobic conditions and low pH is used to prepare silages from crops sufficiently rich in

soluble sugars, such as whole corn and stover, alfalfa, sorghum, timothy grass,

orchardgrass, ryegrass, sorghum and sweet sorghum and barley.

The harvest procedure is similar to that of forages stored dry. The crop is mowed,

conditioned and tedded to avoid effluent production in case it is excessively wet, and

arranged into windrows. The windrows are merged prior to harvest and densification

24

24

(Shinners, 2003). Dry matter loss during the harvest is much lower compared to dry

baling, as the crop is still substantially wet (Muck & Kung, 2007) and thus holds together

better.

Following densification, the crop is sealed from air. Under anaerobic conditions, bacterial

fermentation begins, lactic acid bacteria being the chief microbial agents. The free sugar

content within the wet crop is fermented into lactic and other carboxylic acids. The

circulating acids reduce the crop pH, inhibiting further microbial activity. Ensilage thus

preserves the majority of the crop at the cost of its free sugar content.

The location where the biomass is stored has to support the maintenance of anaerobic

conditions. The dedicated locations for storing it are termed silos. Three types of silos are

common:

1. Tower silos: These are steel and concrete vertical storage structures, into which

the harvested forage is piled. The material forms a column within the tower, its

weight compressing the base and establishing anaerobic conditions within. The

top surface, exposed to air, dries out and forms an insulating layer. The

compression at the base can lead to effluent production. The moisture levels of the

packed material must be checked and adjusted beforehand to prevent this.

2. Pressed bag silos: These are horizontal storage structures, consisting of large

plastic bags into which the harvested forage is packed and compressed. Bag sizes

and bagging machinery vary with the scale of production. The bags may be stored

25

25

on the field, or on concrete or asphalt surfaces. The plas tic material is susceptible

to puncturing by birds, animals or hailstones and so must be checked regularly.

3. Piles/Bunker silos: Silage is commonly stored on the ground (soil, concrete or

asphalt) and covered with plastic. A variant of this method is to pile it in an

enclosed space within two or three walls and then cover it. The resulting structure

is called a bunker silo. Piles and bunker silos carry the lowest capital cost.

However, due to the greater exposure, there is greater risk of spoilage and dry

matter loss. High-density packing and high rates of removal minimize losses.

On a smaller scale, silages can also be baled similar to hay, wrapped in plastic and then

stored on the field or in indoor facilities. The wrapping plastic must be thoroughly

airtight and of suitable thickness, and the bales must periodically be checked for holes

that could let in air and leak effluent material.

The crop being ensiled must be sufficiently rich in non-structural free sugars to ensure

quick, effective fermentation and generation of preservative carboxylic acids. These

sugars must be contained within the plant fluids so as to effectively circulate throughout

the crop mass. Structural carbohydrates such as cellulose and hemicellulose, and storage

carbohydrates such as starch cannot be broken down quickly (or at all) by lactic acid

bacteria. Thus, the content of water-soluble carbohydrates within a forage species is a key

indicator of how effectively it can be ensiled. The crop must be rich in monosaccharides

(glucose and fructose being the most readily consumed) and disaccharides, as they are the

most soluble (Buxton & O'Kiely, 2003; Muck & Kung, 2007). Certain polysaccharides

such as plant fructans, while not directly convertible, are broken down post-cutting by

26

26

plant enzymes into soluble monosaccharides and oligosaccharides which can then be

fermented into acids (Muck & Kung, 2007).

Silage moisture content significantly affects its stability and must be within a specified

range. Given the high concentrations of sugars, ionic species and other solutes, sufficient

water must be free for microbial utilization. The fraction of free water, termed water

activity, (described further in Appendix A) must not be lower than 0.93 – 0.945 for lactic

acid bacteria growth. This has been reported to correspond to water concentrations of 60%

by weight in ryegrass and 67% in alfalfa. The growth of lactic acid bacteria and

corresponding fermentation has been correlated to water content, reducing as the dry

matter content of the silage mixture increases beyond 35%, and ceasing when it is 70% or

higher (Muck & Kung, 2007). Low moisture concentrations are also reported to reduce

the acid tolerance of lactic acid bacteria, the final pH of the silage reported as increasing

with dry matter concentration (and thereby decreasing with moisture concentration).

Depending upon the forage species ensiled, the final silage pH values can reach ranges of

5.0- 5.5+ at moisture levels below 50% by weight. As a pH close to 4.0 is necessary for

stability, this is extremely undesirable. At the other extreme, high moisture content

enables the growth of competing microbes, particularly Clostridial species. Moreover,

upon compaction within the silo the excessive water is lost as effluent. Soluble organic

components are lost with the water, reducing both feed quality and local groundwater

quality (Muck & Kung, 2007).

27

27

Another factor affecting silage stability is the ability of the crop mass to resist a change in

pH, referred to as its buffering capacity. Of particular significance is the resistance to pH

changes in the range 6.0 – 4.0. This resistance has been attributed to chemical species

within the plant biomass such as organic acids – citric, malic and malonic – and their

salts, phosphates and orthophosphates, sulfates, nitrates and chlorides, and to a lesser

extent to forage protein content (Buxton & O'Kiely, 2003; Muck & Kung, 2007).

Buffering capacities vary across species, those of forage legumes in particular being very

high. They also decline with crop maturity, the buffering chemical species being slowly

replaced by insoluble tissue components as the plant matures. A low buffering capacity is

desirable, as other microbes can compete with lactic acid bacteria at higher pH values.

A major issue when preparing silage on a large scale is contamination by bacteria of the

Clostridia family. Clostridial bacteria are commonly present on harvested forages as

spores, and begin proliferation under anaerobic conditions. They can pose significant

competition to lactic acid bacteria unless the pH of the mixture is brought down rapidly,

as the optimal pH for their growth and functioning is 7.0 – 7.4, compared to 4.0 for lactic

acid bacteria (McDonald, 1981). Clostridia proliferate at water activities of 0.995 or

higher and are present when silages (or localized regions) are sufficiently wet (>70%

moisture by weight). They can also proliferate near sufficiently aerobic zones where

microbial respiration breaks down carboxylic acids – lactic and acetic – causing the local

pH to rise (Muck et al., 2003). Clostridial proliferation hinders silage preservation.

Clostridia compete with lactic acid bacteria for the available sugars, fermenting them into

a mixture of products. Butyric acid is the chief product, with butanol, ethanol, propionic,

28

28

acetic and lactic acids are formed in smaller amounts. Butyric acid is weaker in

comparison to lactic acid, as a result of which the pH drop arising from its formation is

not sufficient to preserve the forage material. Secondly, Clostridia ferment amino acids,

generating carbon dioxide, ammonia and degradation products (McDonald, 1981; Rooke

& Hatfield, 2003). The release of ammonia can increase the pH, enabling further

microbial proliferation and thus forage degradation. Thus, the consumption of amino

acids severely reduces the forage protein content and thus its value as a feed, while the

generation of carbon dioxide from the storage increases the silage‘s carbon footprint

should it be used as a fuel or chemical feedstock.

Similar to hay, propionic acid mixtures and ammonia can be added to silage to help

preserve it. Cellulases and hemicellulose enzymes can also be added so as to slowly

break down structural carbohydrates for further acid fermentation (Pitt, 1990).

Ensiling has gained some attention in the preservation of biomass for conversion to

biofuel, as the extended presence of acid helps break down lignocellulose, making it

more vulnerable to breakdown. The use of ensilage as an early pretreatment has been

reported with sweet sorghum by Henk et al (1994).

1.3.2 Storage Losses and Biomass Quality Changes

Following the identification of parameters impacting microbial activity and thus storage

loss, it is necessary to evaluate the impacts of storage loss on the quality of the feed.

Described below is a description of the precise impact dry matter loss has on the stored

29

29

feed, specifically the fraction remaining. Forage studies again provide a suitable

background to examine storage impacts.

1.3.2.1 Storage Losses: Forage Background

Carbohydrates are preferentially consumed by microbes upon respiration, nonstructural

sugars accounting for the maximal fraction of the dry matter lost during storage (Moore

& Hatfield, 1994; Rotz & Muck, 1994). Structural carbohydrates and proteins are

consumed subsequently. Lignin is consumed least preferentially, due to its hydrophobic

nature and phenolic composition. Forage studies report dry matter loss as increasing the

proportion of ‗fiber‘ components i.e. cellulose, hemicellulose and lignin, indicated

through fiber metrics – NDF (Neutral Detergent Fiber), ADF (Acid Detergent Fiber) and

ASL (Acid Soluble Lignin) – and decreasing the feed digestibility.

Coblentz et al and Shinners et al both reported increases in NDF and ADF concentrations

in alfalfa hay stored baled at high moisture, significantly also reporting a decrease in in -

vitro dry matter indigestibility (Coblentz et al., 1996; Shinners et al., 1996). The former

also reported losses in non-structural sugars (Coblentz et al., 1997) when storing alfalfa.

Similar results were reported for the storage of baled bermudagrass (Coblentz et al., 2000;

Turner et al., 2002) and orchardgrass (Martinson et al., 2011). The impact of dry matter

loss is reviewed by Rotz et al (Rotz & Muck, 1994).

30

30

The loss of free and non-structural sugars can be expected to increase the proportion of

cellulose and hemicellulose, structural carbohydrates that are chemically bound to lignin.

Lignin content has been identified as increasing indigestibility in forages (Akin, 1989;

Besle et al., 1994; Chen et al., 2003; Guo et al., 2001; He et al., 2003; Jung & Deetz,

1994) due to the steric impedance it provides to digestive enzymes, its hydrophobic

nature, and its generation of toxic phenolic compounds upon breakdown (Jung & Deetz,

1994). These issues are very similar to those faced when carrying out bio-based

breakdown of lignocellulose for fuels and chemicals. Indeed, Fahey et al proposed post -

harvest physical and chemical treatments of forage material to increase its digest ibility in

the target ruminant (Fahey et al., 1994).

In summation, forage data indicates high dry matter losses to primary cause high losses

of the more easily accessed carbohydrates, leaving behind the more resistant sections of

biomass i.e. cellulose, hemicellulose and lignin. The remaining biomass is likely to be

more recalcitrant towards biological breakdown.

1.3.2.2 Biomass Storage: Forage Data Limitations

The background established by the forage industry provides detailed knowledge of the

impact of different storage parameters, the likely composition change arising from

extensive microbial respiration and dry matter loss as well as the resistance of the

remaining material that is to be broken down. Several uncertainties however exist when

applying this knowledge to the storage of biomass for the biorefinery.

31

31

Forage characterization is largely based on fiber metrics – NDF, ADF and ASL – that

indicate the content of cellulose, hemicellulose and lignin (Moore & Hatfield, 1994).

Given the specific applications of biomass components in the biorefinery, it is necessary

to be aware of the exact composition of biomass on a dry mass basis. It is thereby

necessary to also ascertain the changes in composition – especially the content of

cellulose, hemicellulose and lignin – brought about by the storage process.

Secondly, the role of composition in causing biomass recalcitrance – the resistance to the

separation, extraction and subsequent enzymatic depolymerization of lignocellulosic

carbohydrates – is not entirely certain. Lignin is a major factor due to its binding

cellulose and hemicellulose and shielding both from enzymatic attack (Adani et al., 2011;

Kumar et al., 2012; Lynd et al., 1999; Zhao et al., 2012a), which necessitates biomass

pretreatment (Mosier et al., et al., 2005c). Moreover, several pretreatments, such as dilute

acid, steam and controlled-pH liquid hot water treatments break down lignin into

phenolic monomers and oligomers that are toxic to enzymes and biocatalysts (Klinke,

2004; Ximenes et al., 2011; Ximenes et al., 2010; Zhao et al., 2012a). Lignin content has

been held as the major impedance in the breakdown of cellulosic biomass (Chapple et al.,

2007; Lionetti et al., 2010; Santos, Lee, Jameel, Chang, & Lucia, 2012; Studer et al.,

2011; Zhao et al., 2012a). Recent studies however, indicate lignin content alone as

insufficient in predicting biomass resistance. Yu et al and Ishizawa et al reported

delignification as increasing cellulose hydrolysis yields until lignin content reached an

optimal minimum, beyond which further delignification had either no impact (Yu et al.,

32

32

2011) or a negative impact on hydrolysis (Ishizawa et al., 2009). Similarly, when

comparing multiple pretreatment methods for dead pine, Rio et al reported acidic

pretreatments as producing more digestible material than alkaline treatments, even

though the latter removed a larger fraction of the constituent lignin (Rio, 2010). Most

significantly, Rollin et al (2011) compared switchgrass pretreated by soaking in aqueous

ammonia and by the COSLIF method (Cellulose Solvent and Organic Solvent-based

Lignocellulose Fractionation). Ammonia soaking strongly delignified the switchgrass,

removing ~74% of acid-insoluble lignin present, while the COSLIF method removed

~34%. The latter however exhibited close to 80% glucan digestibility at low enzyme

loading, while the former showed approximately 45% digestibility under enhanced

hydrolytic conditions. Upon examination for cellulase accessibility, estimated through

enzyme adsorption, COSLIF-pretreated switchgrass was observed to have greater

accessibility, despite its higher lignin content. The findings indicate that it is the

particular structure of lignin, as opposed to its content, that affects its impedance of

hydrolysis. Thus, while composition changes provide an indication as to the likely

recalcitrance, the latter must be established empirically, and investigated on the basis of

biomass structure as well.

Thirdly, forage studies have largely lacked time-variant analysis, focusing on the end-

point quality of the stored biomass, and thereby the influence of storage parameters on it.

Biomass storage duration can be highly variable, based on technological and economic

factors, and thus time is an important storage parameter to evaluate. Moreover, the

inclusion of time as a storage parameter would require other variables to be maintained to

33

33

a greater degree than is present. Particularly in case of moisture content, a major

limitation is the absence of continuous data gathered throughout storage experiments.

Moisture is exchanged between biomass and the surrounding environment, air or soil, the

transfer affected by exposure, temperature, environmental moisture content (humidity or

soil water content), aeration and the mass-transfer dynamics resulting from biomass

packing density. When estimating the role of moisture, the dynamics of exchange must

also thus be accounted for, necessitating a time-dependent storage study with precise

control on biomass moisture content and exchange.

The above factors present the need for a time-dependent study tracking composition

changes as a result of storage parameters, as well as examining the structural changes in

stored biomass and investigating their connection to composition change and

correspondingly storage parameters. The objective of the present study is to address this

particular knowledge gap.

1.3.3 Storage Hypotheses

The objective of the present study is to examine the impact of storage temperature,

moisture and duration, under aerobic conditions, upon the quality of lignocellulosic

biomass. Quality is defined by the amount and ease of sugar release by enzymatic

hydrolysis of the biomass. In order to accomplish this, a material balance tracking

changes in chemically defined components (glucan, xylan, and lignin) was performed for

three types of biomass (sweet sorghum bagasse, corn stover, and switchgrass). The

biomass was subjected to enzymatic hydrolysis by cellulose enzymes for biomass before

34

34

and after storage and with or without liquid hot water pretreatment prior to enzyme

treatment. Based on the knowledge developed in the forage industry, three hypotheses

predicting how storage temperature, moisture, and time affect biomass quality were

developed.

The first hypothesis is that biomass stored aerobically will show significant dry matter

loss when its moisture content is greater than the threshold of 20-25% w/w. The moisture

content threshold can be expected to depend upon the biomass chemical composition,

being the moisture concentration establishing water activities ranging 0.8-0.9 or higher.

The second hypothesis is that carbohydrates will be preferentially consumed, and will

thus constitute a substantial fraction (<50%) of the dry matter loss observed. In

comparison, lignin consumption will be much lower. Dry matter loss can be expected to

result in a change in biomass composition, wherein the fraction represented by

carbohydrates (% w/wdry) will be unchanged or reduce, while the lignin fraction will

increase. Among the carbohydrate fractions, free sugars (mono- and disaccharides) can

be expected to be consumed the fastest due to their accessibility and ease of metabolism

by microorganisms, followed by hemicellulose and cellulose.

The final hypothesis is that the release of glucose upon pretreatment and hydrolysis will

be reduced for biomass that has undergone storage losses, due to the preferential

consumption of carbohydrates and enrichment of lignin. Lignin plays chemical and

structural roles in impeding cellulose hydrolysis, and its enrichment has been reported to

35

35

reduce forage ruminal digestibility. As the composition of the biomass affects the impact

pretreatment has on it, it is likely that biomass that has undergone significant storage

losses will undergo bioprocessing differently from the original material.

The remainder of this dissertation discusses the design of the experiments, the methods of

experimentation, data collection and analysis, and the results obtained to examine these

hypotheses.

36

36

CHAPTER 2. MATERIALS AND METHODS

2.1 Biomass Storage: Feedstock Preparation, Experimental Design & Storage Setup

The present study was based on the storage of three different feedstocks, each of which

was harvested, prepared and stored under specific conditions. Each storage experiment

tested specific parameters for their impact on the biomass. Presented below is a

description of the feedstock used, the pre-storage preparation, storage conditions and the

parameters tested.

2.1.1 Sweet Sorghum Bagasse: Aerobic and Anaerobic Storage

2.1.1.1 Experimental Design:

Moisture content and exposure to oxygen were the varied parameters when storing sweet

sorghum bagasse. Bagasse stored aerobically dry (12.3% w/w moisture) was contrasted

with aerobic wet bagasse (25.5% w/w moisture) as well as bagasse ensiled (68.7% w/w

moisture). All three were stored for a period of 24 weeks in the same location and at the

same temperature.

2.1.1.2 Feedstock preparation:

The biomass used in the experiment was derived from Keller and Sugardrip variants of

sweet sorghum (S. bicolor), cultivated at Purdue Agronomy Center for Research and

37

37

Extension (ACRE), and fertilized in the spring with a nitrogen treatment of 224 kg-

N/hectare.

Sorghum from both Keller and Sugardrip variants was used to generate biomass for the

aerobic treatments. The crop was harvested by hand in October 2010, by cutting 6 inches

above the soil removing grain heads. Stalks were pressed to remove the juice with a small

two-roller press and spread in a single lay on turf to dry in the sunlight. After drying on

the turf for three days, the bagasse remnant was moved to a pavement section and dried

for an additional three days to a final moisture content of 15.8% moisture by weight,

upon which it was shredded in a yard waste chipper. Sorghum silage (anaerobic

treatment) was prepared from Keller variants alone, which were harvested and crushed as

described above, and shredded at 68.7% moisture w/w.

2.1.1.3 Storage setup:

Following the shredding process, sorghum bagasse was either dried further or re-wetted

to the target moisture levels. Bagasse stored aerobically at low-moisture was prepared by

allowing the shredded material to dry further to 12.3% w/w moisture. Aerobic high-

moisture bagasse was prepared by re-wetting the shredded material (15.8% moisture).

Samples of shredded bagasse were gathered in buckets and mixed with water, 355 mL

water being added to 2.7 kilograms of bagasse. The mixing was calculated to bring the

moisture content to 25.5% w/w. No wetting or drying was carried out for

ensiled/anaerobic bagasse, which was 68.7% moisture after shredding.

38

38

Moisture content was measured using a custom-made portable drying tube that can be

fitted with a hairdryer. The assembly obtained from the laboratory of Prof. D.

Buckmaster at the Department of Agricultural & Biological Engineering, and was

developed by him (Buckmaster, 2005). Samples of biomass were weighed and transferred

into the tube, following which the hairdryer was fitted and hot air circulated through the

tube, drying the biomass for 5-10 minutes, following which the biomass was re-weighed.

The process was repeated until the measured biomass weight was unchanging. The

difference between this final weight and the initial provided the weight of moisture in the

biomass. By measuring moisture contents at the start and end of the storage process, the

dry matter loss was calculated.

2.1.1.3.1 Aerobic treatments

Biomass was packed into mini-bales using a modified Earthquake® W-1000 log-splitter

(Earthquake, Cumberland, WI) with an enclosed cylindrical baling chamber. A plastic

trash packer bag was placed inside the packing chamber, into which the wet sorghum

samples were packed. After being compressed twice for 5-10 seconds, bags were

extracted from the packing chamber and immediately wrapped with twine to maintain

density. Mini-bales were 33-35 cm in height and 17-24 cm in diameter, with initial dry

densities of 136-280 kg/m3. Four 6-inch slits were cut into each bag to allow air exchange,

and the bags placed in 5-gallon buckets for long-term storage.

39

39

2.1.1.3.2 Anaerobic treatments:

Sorghum bagasse was pressed into mini-bales as described above, which were then

loaded into silage bags, made from 18-inch (0.4572 m) diameter 6-Mil (0.1524 mm

thickness) polyethylene tubing, cut to 0.5-metre length and sealed at the ends using a

propane blowtorch. Mini-bales were double bagged to prevent leakage. Tedlar® bag gas

valves (Sigma Aldrich, St. Joseph, MO) were installed in each bag prior to sealing. The

outer layer was fitted with a gas valve and sealed, and air drawn out of the bags through

the valves, using a vacuum pump. Each mini-bale contained approximately 4 kilograms

of wet bagasse, the moisture content being 68% w/w at the time of packing. Silage mini-

bales were consistent, having dimensions of 28 cm (height) x 16 cm (diameter), and

initial dry matter densities of 222-233 kg/m3.

The bags were stored in the basement of Potter Engineering Center at Purdue University.

While not continuously monitored, room temperature was maintained nearly constant at

20 °C. Aerobic treatments were sampled at 8 and 24 weeks, while silage was sampled at

24 weeks alone. Dry matter content was calculated by measuring total bale weight and

moisture content, the latter in triplicate. At least three samples from each treatment were

analyzed for composition changes.

2.1.2 Corn Stover Storage

Corn stover was obtained from Isaac Emery at Purdue University, having been used in

his dissertation project (Emery, 2013). Specific samples showing high dry matter loss

were analyzed.

40

40

2.1.2.1 Feedstock preparation:

Stover was collected by hand on the 5th of November 2012, one day after grain harvest,

from the Throckmorton-Purdue Agricultural Center (TPAC) in Tippecanoe County,

Indiana. Stover moisture was measured on the 8th of November, and varied from 35.4 –

56.1%. Biomass was allowed to air dry indoors to 13.1%, and re-wetted for storage

treatments.

2.1.2.2 Storage setup:

A small steel baling column, based on designs by Coblentz et al (1993) and a hand crank

press were used to generate laboratory-scale samples, 10.4 x 10.4 cm depth x height and

11 to 14 cm in length. For each sample, roughly 350 g biomass (dry weight) was

sampled from bulk storage. From this, 3 subsamples of at least 25 g were bagged and

dried for moisture content determination, and 3 additional subsamples of 10 to 30 g were

wrapped in plastic mesh and placed with the main sample in a biomass reactor (See

Emery et al (Emery, 2013) for further detail).

The biomass was stored in 7.57-liter plastic containers fitted with airtight lids (Gamma

Squared, Carlsbad CA). Stover samples were placed on perforated shelves within the

containers. Saturated salt solutions with excess salt were placed below the shelves to

control the humidity within the container. The containers were placed in temperature-

controlled rooms and biomass stored therein for eight weeks. Stover samples were

removed in February 2013 and examined for dry matter los s. Moisture content was

41

41

measured through the ASABE S 358.2 protocol (ASABE, 2012). Samples showing high

dry matter loss were subjected to composition analysis and bioprocessing. The samples

used in the present study were stored at 35 °C and a water activity of 0.97.

Stover had not been milled prior to packing. Following dry matter loss estimation, the

samples analyzed were shredded for 30 seconds in a food processor and sieved through a

0.25-inch screen. For comparison, pre-storage stover was similarly treated. Shredded

samples were compared for composition and bioprocessing performance.

2.1.3 Switchgrass: Aerobic Storage

2.1.3.1 Experimental design:

The objective of the experiment was to investigate the impact of temperature and water

activity on switchgrass quality. At least three levels of each parameter were included in

the design, to as to establish with maximum clarity their role in storage and the resulting

degradation. Storage duration also governing the changes in quality, four levels of

duration (in weeks) were incorporated into the initial design, shown in Figure 2.1. The

experiment was carried out from July 2012 – February 2013. Samples of switchgrass

containing 100 dry grams each constituted the replicates for each cell.

The temperature levels, in degrees Celsius, were chosen to reflect seasonal averages at

different times of the year. Temperature has a non-linear impact on microbial growth; the

latter absent below a threshold (varying across microbial families) and increasing rapidly

42

42

with temperature till the latter reaches an optimal value (Huang et al., 2011). Because of

the non-linear effect, as well as the diurnal fluctuations in temperature, the levels were

separated by at least ten degrees Celsius.

Water activity was chosen as a parameter to control for the loss of moisture to the

atmosphere. To establish a dynamic moisture equilibrium, the water activity of the

storage atmosphere was maintained at the same level as the biomass. Water activity

levels were chosen to gradually increase the likelihood of microbial activity, as well as

the range of microbial families capable of surviving and growing on the biomass (See

Appendix A). Microbial activity being absent at water activities below 0.5, 0.65 was

selected as the lowest level, and 0.95-0.99 as the highest level.

Figure 2.1. Switchgrass Storage Experimental Design

43

43

2.1.3.2 Feedstock preparation:

The switchgrass (P. virgatium) ecotype known as ―Shawnee‖ was cultivated at the

Throckmorton-Purdue Agricultural Center in May 2007. No-till seeding was carried out,

at the rate of 18 kg.hectare-1

. From the second year of cultivation onwards, the crop was

treated annually with 84 kg.hectare-1

of nitrogen, in the form of AgrotainTM

-treated urea.

The crop was harvested each year in the fall, and allowed to grow again.

The biomass used in the storage experiment was harvested in November 2011, at the end

of the fourth growing season after establishment. The standing crop was cut with a Carter

flail-type chopper (Carter Manufacturing Co., Brookston, IN) chopper, piled on the side

of the field and collected by hand the same day. At the time of harvest, switchgrass

moisture content ranged from 25 – 32%. Harvested switchgrass was stored indoors in

Ace® 30-gallon yard waste paper bags (Ace Hardware, Oak Brook, IL). The switchgrass

moisture content declined to 10% by the first week of December. Switchgrass was stored

until February 2012, when it was sampled for composition analysis.

To standardize the particle size range, dried switchgrass was milled through a 0.25-inch

screen in a hammer mill. The hammer mill, belonging to Dr. K. Ileleji at Purdue

University Department of Agricultural & Biological Engineering, was a product of Glen

Mills Inc. (Clifton, NJ). It consisted of 11 swinging hammers and was fitted with a 2-

Horsepower (230 V, 3-phase) Speedmaster Adjustable Speed AC Motor (Leeson Electric

Corporation, Grafton WI) (Further details available in Probst et al, 2013). The milled

product was subsequently sieved in a Vibroscreen® circular vibratory screener (Kason

44

44

Corporation, Millburn NJ), and the material retained between meshes of opening size

0.20-inch (No. 4 mesh) and 0.034-inch (No. 20 mesh) used for the experiment.

Table 2.1. Switchgrass Storage Moisture Concentrations

Temperature

(°C)

Water activity

(aw)

Moisture Content

(% )

<10

0.65 10.1

0.75 12.8

0.85 16.1

0.99 34.1

20

0.65 10.5

0.75 12.0

0.85 15.8

0.99 32.4

35

0.65 12.1

0.75 15.6

0.85 19.1

0.99 34.1

2.1.3.2.1 Switchgrass Sample Preparation:

Milled and sieved switchgrass was wetted to the levels listed in Table 2.1 by equilibrating

with deionized water for 48 hours (Described in Appendix C). Moisture content was

measured through ASABE Standard 358.2. Wetted switchgrass was packaged into tubes

45

45

made from perforated sheet plastic. Polyethylene plastic mesh (Opening s ize 0.06‖x0.06‖)

was purchased from McMaster-Carr (Chicago IL), and cut into sheets of approximately

33 cm x 36 cm dimensions. Each sheet was rolled along its length into a tube and a

plastic cable tie attached 1-2 cm from one end to seal it. The resulting assembly was

labeled and weighed, together with a second plastic tie, on a Mettler PE 3600 Delta

Range scale (Mettler-Toledo, Columbus OH) and the weight noted. The scale was then

tared and wetted switchgrass loaded into the tube until the necessary weight was reached

(calculated from switchgrass moisture content). The tube was then sealed with the second

plastic tie and the weight of the entire assembly measured and noted. Following

preparation, samples were stored at 4 °C in sealed vessels or zipper storage bags before

being transferred into the buckets for the storage experiment.

2.1.3.3 Storage setup:

Biomass storage was carried out in 6.5-gallon plastic buckets with screw top lids,

purchased from Purdue University stores. Two 0.25-inch diameter holes were drilled into

each bucket lid to allow air exchange. Three buckets were used for each combination of

temperature and humidity, each bucket containing 6 switchgrass samples. Single samples

were randomly sampled from each bucket at the relevant time points, and analyzed for

dry matter loss and composition change.

The switchgrass samples were stored on shelves made from flexible perforated

polypropylene sheets and PVC tubing purchased from McMaster-Carr (Chicago, IL). The

46

46

shelves were fitted halfway down the height of the bucket, partitioning it in two and

allowing exchange of air between the partitions (Figure 2.2.). Saturated salt solutions

(0.75 – 1 liter) were loaded into the lower partition (Table 2.2.). Potassium iodide,

sodium bromide and strontium nitrate were purchased from Sigma Aldrich (St. Louis,

MO). The remaining salts were purchased from Purdue University Stores (Purdue

University, West Lafayette, IN)

The solutions establish moisture equilibrium with the air, controlling its relative humidity

(Greenspan, 1977; Igathinathane et al., 2005) and thereby controlling moisture migration

to and from the wetted biomass. Pure water (Reverse-osmosis or deionized) was used to

maintain the relative humidity of 99%.

To control temperature, the buckets were placed in temperature-controlled rooms at

Purdue University West Lafayette campus. Low temperature storage was carried out in a

cold-storage chamber set to 45 °F (7.2 °C) at the Philip E. Nelson Hall, housing the

Department of Food Science. A subsequent set of trials was carried out in refrigerators at

the Potter Engineering Center. Mid-temperature storage was carried out in a temperature-

controlled bay room at the Department of Agricultural and Biological Engineering

Building, set to 20 °C. High temperature storage was carried out at the Department of

Agronomy grain drying lockers set to 35 °C in Lilly Hall at Purdue University. Three

buckets were used for each combination of temperature and water-activity.

47

47

To monitor variation in temperature and humidity, Acu-Rite Home Comfort monitors

purchased from Amazon.com (Seattle, WA) were placed in each bucket. Readings from

the sensors were logged on a weekly/bi-weekly basis. In case of drops in humidity, the

buckets were opened and water added to the salt solution. Biomass samples were stored

under these conditions and sampled at 1, 2, 4 and 8 weeks.

2.1.3.4 Sampling:

Samples were randomly drawn at the time points described in the design, one per bucket.

Each sample was removed, and weighed on the same Mettler PE 3600 scale as when

being prepared. The contents were then transferred into a plastic zipper storage bag and

Figure 2.2. Humidity Control

48

48

moisture content measured using a HB43-S Halogen Moisture Analyzer (Mettler-Toledo

LLC, Columbus, OH) instead of the ASABE Standard (Method validation in Appendix

B.1). Using the recorded data from the start of the experiment, the change in dry matter

was calculated for each bale. The contents from samples gathered at a particular time

point were examined further if they showed a dry matter loss greater than or equal to 5%

since the previous sampling point, for that given combination of temperature and water

activity.

Table 2.2. Solutions used in controlling water activity

Temperature

(°C)

Relative humidity

(% )

Solution

(Saturated when applicable)

<10

65 Sodium bromide

75 Sodium chloride

85 Potassium chloride

99 Water

20

65 Potassium iodide

75 Sodium chloride

85 Potassium chloride

99 Water

35

65 Potassium iodide

75 Sodium chloride

85 Strontium nitrate

99 Water

49

49

2.1.3.4.1 Microbial Activity Estimation:

Samples of the stored switchgrass were analyzed through isolation plating and incubation

at the Purdue Plant & Pest Diagnostic Laboratory, and examined for the fungal species

growing. Two types of cultivation were carried out, as described below:

1. Cultivation: The samples were incubated at 24-25 °C in a near 100% humidity

chamber consisting of a clear airtight plastic box with a water soaked paper towel

at the bottom. Over this a layer of screen wire was placed to support the plant

tissue and prevent direct contact with the water. The samples were checked for

fungal growth at 2, 3 and 4 days after incubation.

2. Isolation plating: Approximately 1 teaspoon of non-surface sterilized plant tissue

was plated onto quarter-strength Potato Dextrose Agar loaded with antibiotics.

The plates were wrapped with ParafilmTM

and incubated under fluorescent lights

and checked for fungal growth at 2 and 3 days.

The fungi cultivated in this manner were identified through microscopic examination and

identification of fungal fruiting structures.

2.2 Sample Analysis:

Following the establishment of dry matter loss, samples were analyzed for composition

change, to establish a mass balance for the dry matter lost. They were subsequently

subjected to bio-based breakdown to examine if the dry matter loss impacted feed

bioprocessing quality. The procedures for each are described as follows.

50

50

2.2.1 Composition Analysis:

The analytical procedures developed by the National Renewable Energy Laboratories

(NREL), termed LAPs, were used to determine the composition of original switchgrass

and stored switchgrass (Sluiter et al., 2010). The sequence of operations is shown in

Figure 2.3. Key steps are described below:

1. Water/Ethanol wash (LAP 010 (Sluiter et al., 2008)): Biomass milled to pass

through a 40-mesh (0.42 mm) screen was loaded into WhatmanTM

cellulose

thimbles purchased from Sigma Aldrich (St. Louis, MO) and the weight and

moisture content of the switchgrass recorded. The thimble was then loaded into a

Soxhlet extractor apparatus including a pre-weighed round-bottom flask

containing 190 ml of deionized water. The Soxhlet wash was carried out for 8-12

hours following which the water was removed for analysis, and replaced with 190

ml of 95% ethanol. The wash was then continued for 24 hours, after which the

flask containing the ethanol wash was subjected to rotary evaporation to remove

the ethanol. The weight of the solubles dissolved in ethanol was determined

through weighing the dried flask and comparing it to the original recorded weight.

Using the recorded dry weight of loaded biomass, the fraction of ethanol soluble

compounds was determined. Similarly, high performance liquid chromatography

(HPLC) analysis was used to determine the concentration of concentration of

water solubles such as sugars and lactic acid, following which their mass and thus

fraction by dry weight was calculated. The washed biomass was then dried and

subjected to acid hydrolysis.

51

51

When carried out with original switchgrass, the above step revealed the

proportion of solubles to be low (<5% in total), and was not performed when

analyzing stored switchgrass samples.

2. Acid hydrolysis (LAP 002 (Sluiter et al., 2012)): 0.3 gram of washed biomass

was loaded into 100 ml Pyrex tubes and incubated at 30 °C with 3 ml of 75% w/w

sulfuric acid for 2 hours. 84 ml of deionized water was added to the tubes, which

were sealed and autoclaved at 121 °C for one hour. After cooling, each tube‘s

contents were filtered, using a Buchner flask fitted with a labeled, pre-weighed

suction crucible and connected to a vacuum pump. The filtrate was analyzed

through HPLC for glucose, xylose, arabinose and acetic acid. The liquid volume

being known (87 ml), the masses and subsequently fractions by dry weight of

glucan, xylan, arabinan and acetate were determined. By examining the

absorbance of the filtrate liquid at 320 nm, relative to 4% sulfuric acid as a blank,

the proportion of acid soluble lignin was determined.

3. Insoluble lignin determination (LAP 002): Following the filtration, the

crucibles (containing residue) were heated overnight in a 105 °C oven, cooled for

an hour in a desiccator and weighed on a Denver M220D balance (Denver

Instruments, Bohemia, NY) to the fourth decimal place, the process repeated until

three weights were obtained with a consistency of 0.001 grams. The crucibles

were then flamed until all present carbon was burned, heated in a 575 °C oven for

at least 6 hours, cooled subsequently for 4 hours in a desiccator and weighed to

the fourth decimal place. This process, termed ‗ashing‘ was repeated until two

weights consistent within 0.0003 grams are recorded. The ashing process had

52

52

been carried out with empty crucibles. By comparing the three sets of weights, the

insoluble lignin content was calculated.

Combining the compositions of original and stored switchgrass with the observed dry

matter losses, a mass balance was developed for biomass carbohydrate and lignin content.

53

53

Figure 2.3. Laboratory Analytical Procedure sequence developed by NREL

(Sluiter et al. 2012)

54

54

2.2.2 Bioprocessing:

Both the original and switchgrass were subjected to pretreatment – a thermochemical step

designed to improve cellulose accessibility to cellulase – and subsequently to enzymatic

hydrolysis. Both steps are described in further detail below.

2.2.2.1 Liquid Hot-water pretreatment:

Biomass was subjected to liquid hot water pretreatment, as developed by the Laboratory

of Renewable Resource Engineering (LORRE) at Purdue University (Kohlman et al.,

1998; Mosier et al., 2005a). Pretreatment was carried out in 316 stainless steel tubes with

an outer diameter of 1 inch (2.54 cm) and thickness of 0.083 inches (2.1 mm), capped at

either end with 1 inch (2.54 cm) Swagelok tube end fittings (Swagelok, Indianapolis, IN)

and having an internal volume of 45 ml. Biomass and water were loaded into the tubes so

as to occupy a volume of 33.75 ml, leaving 25% of the headspace for expansion during

the heat up. Assuming the pretreated slurry to have a density equal to water (1 g/ml), the

final weight of the mixture loaded was 33.75 gm. Biomass dry solids content having been

determined beforehand, water was added so as to give the mixture a dry solids content of

15% (w/w basis). Following the loading of the mixture, the tube was sealed, the cap

tightened with a wrench. The assembly was then immersed in a Techne SBS-4 Fluidized

Bath (sandbath) (Techne c/o Bibby Scientific US, Burlington NJ) and heated, the optimal

pretreatment temperature and time varying with solids loading, biomass species and

composition.

55

55

The optimal hot water pretreatment conditions for switchgrass were reported by Kim et al

(2011a) as 200 °C for 10 minutes. For comparison with the switchgrass, corn stover

stored and analyzed separately was pretreated as well. Based on the optima reported by

Mosier et al (2005b), corn stover was pretreated at 190 °C for 15 minutes. Tube

temperature profiles developed earlier were used to determine the time necessary for the

contents to reach the desired temperature. Accordingly, the tubes were immersed for 7

minutes and 40 seconds in addition to the pretreatment time noted above. Following the

pretreatment, tubes were cooled in cold water and opened to extract pretreated materials.

2.2.2.1.1 Pretreatment Fractionation Analysis:

Following the first pretreatments performed on original and stored biomass, a mass

balance was developed for the carbohydrates and lignin and their fractionation into the

solid and liquid phases. Samples were pretreated in triplicate when developing this mass

balance for original and stored biomass. The process is described in further detail below.

Vacuum filtration assemblies were set up, consisting of a vacuum flask fitted with a

Buchner funnel containing a Whatman® No. 5 filter paper and connected to a vacuum

pump. Following pretreatment, the tubes were opened and their contents emptied onto the

filtration assembly to separate pretreated solids and liquid.

Pretreatment liquid was analyzed through an NREL analytical procedure (LAP 14).

Liquid samples were loaded in 10 ml glass vials and 72% sulfuric acid added in a

56

56

liquid/acid ratio of 1:0.0335. Vials were sealed and autoclaved at 121 °C for one hour,

following which their contents were filtered and analyzed through HPLC, together with

filtered samples of the original pretreatment liquid. The concentrations of

monosaccharides – glucose, xylose and arabinose – and sugar degradation products –

furfural and hydroxymethyl furfural –within the autoclaved samples were used to

calculate the net carbohydrate content present the liquid fraction. By comparing

concentrations between autoclaved and original pretreatment liquid samples, the

oligosaccharide content within the liquid fraction was calculated.

Following the removal of pretreatment liquid samples, hot deionized water was added to

the solids to remove all solubles still present through the vacuum filtration process. Solid

samples were washed thrice, each wash with 100 ml boiling deionized water. Washed

solids were dried overnight at 45 °C and milled to pass through a 40 mesh screen. Milled

solids were then subjected to NREL LAP 002 described above, to determine the

carbohydrate and lignin content within the solid fraction. Combined with the liquid

concentration data, a mass balance was developed for the pretreatment process.

The mass balance was required to estimate the extent of carbohydrate – glucan and xylan

– solubilization, the generation of potential inhibitors through pretreatment, and the final

composition of washed pretreated solids. Following the mass balance calculations,

biomass samples – original and stored – were pretreated and washed solids subjected to

enzymatic hydrolysis.

57

57

2.2.2.2 Enzymatic hydrolysis:

Following composition analysis, the washed (non-dried) solids prepared after

pretreatment were hydrolyzed using cellulase and ß-glucosidase enzymes. Glucose

release was compared between original biomass (untreated), the washed pretreated solids

of original biomass, stored biomass (untreated) and the washed pretreated solids of stored

biomass.

Biomass was loaded into tared 250 ml NalgeneTM

screw-cap plastic bottles (Nalge Nunc

International Corporation, Rochester, NY), the mass loaded corresponding to 1 dry gram

either of original biomass or washed pretreated solids. The bottle content mass was then

made up to 100 gm by adding 50 mM sodium citrate buffer (pH 4.8, 0.2% w/v sodium

azide), generating a dry solids concentration of 1% w/w for hydrolysis. Enzymes were

then added to each bottle. All hydrolyses were carried out in triplicate.

The hydrolysis was carried out using NS22086 acquired from Novozymes Bioenergy

(Novozymes A/S, Denmark) as the cellulase, and Novozyme 188 (catalogue no. C6105)

purchased from Sigma Aldrich (St. Louis, MO) as the cellobiase. Cellulase was added to

a final concentration of 50 Filter Paper Units (FPU) gm-glucan-1

and cellobiase to a

concentration of 130 Cellobiase Units gm-glucan-1

, the final total protein concentration

being 109.6 mg-protein/g-glucan.

Following enzyme addition, bottles were closed and placed in a New Brunswick G24

Environmental Incubator Shaker (New Brunswick Scientific Co., Edison, NJ), set to

58

58

50 °C and 200 rpm. Liquid samples of 1.0 – 2.0 ml were taken from each bottle just

before incubation, and 24 and 48 hours after incubation. Samples were filtered using

syringes fitted with 0.2 µm filters and stored in labeled Eppendorf vials at -20 °C.

Hydrolysis samples were examined for glucose concentration using a ReliOn® Confirm

blood glucose meter (Walmart, Bentonville, AR), fitted with ReliOn® Confirm/Micro test

strips (Method validation in Appendix B.2). Upon significant glucose generation, they

were analyzed through HPLC for exact glucose and xylose concentrations. Using the

composition data developed earlier, hydrolysis glucan and xylan yields were calculated,

the former used as the metric of effective bioprocessing.

2.2.3 High Performance Liquid Chromatography:

Concentration data developed throughout the various experiments was generated through

HPLC analysis. A description of the equipment used and analytical methods involved is

presented below.

The HPLC assembly used consisted of a Waters 2414 Refractive Index Detector (Waters

Corporation, Milford, MA), an Aminex® HPX-87H 300 x 7.8 mm column (Bio-Rad

Laboratories, Hercules CA) and an Alliance Waters 2695 Separations Module (Waters

Corporation, Milford, MA). Column temperature was maintained at 65°C. 5 mM H2SO4

was used as the mobile phase, at a flow rate of 0.6 ml/min.

59

59

Samples were analyzed for two sets of compounds. Composition samples, generated from

the NREL LAP experiments, were examined for glucose, xylose, arabinose and acetic

acid. Pretreatment and hydrolysis samples however were examined for citric acid (buffer),

cellobiose, glucose, xylose, arabinose, acetic acid, hydroxymethyl furfural and furfural.

Standards were prepared for each set of compounds. Standard set readings were all within

5% of the actual concentration (Refer Appendix D). Using linear standard curves

generated from the standard sets, the sample concentrations were determined.

60

60

CHAPTER 3. RESULTS AND DISCUSSION

3.1 Effect of Storage Parameters on Biomass Losses during Aerobic Storage

The seasonal harvest of cellulosic biomass for the purpose of year-round conversion into

biofuels and chemicals necessitates medium- to long-term storage. The storage of

cellulose-rich biomass has been studied at length in the forage and animal feed industry,

which provides a suitable knowledge background. Stored biomass is vulnerable to

microbial decay, resulting in consumption of biomass components at different rates. Free

and non-structural carbohydrates are consumed preferentially, and have been reported to

account for the highest fractions of dry matter losses by Rotz et al (1994) and Moore et al

(1994). Proteins and structural carbohydrates are subsequently consumed. Due to its

chemical and physical recalcitrance, lignin is reported as consumed with least preference,

and is enriched upon storage loss.

As a result of the differential consumption rates, storage losses can be expected to reduce

both the concentrations of the components of interest (free and structural carbohydrates)

of biomass and their extractible yield (g/g-dry), due to the enrichment of the recalcitrant

fractions. This is indicated in forage analyses, which show stored feeds to have higher

concentrations of fiber components – Neutral Detergent Fiber (NDF), Acid Detergent

Fiber (ADF) and Acid Soluble Lignin (ASL), indicating concentrations of cellulose,

61

61

hemicellulose and lignin – and lower digestibility (measured in-vitro and in-vivo). As

forage composition metrics are not chemically well defined and non-additive – NDF,

ADF and ASL denote overlapping fractions, as do Crude Protein and Acid Detergent

Insoluble Nitrogen – it is necessary to estimate the precise losses of specific biomass

components upon storage. The objective of this study was to develop a material balance

around components of interest – free and structural carbohydrates and lignin – and track

their changes upon storage.

The present study utilized feedstocks generated from three grass species – sweet sorghum,

corn/maize and switchgrass. All three are agronomically important grasses that can be

cultivated seasonally in large amounts for conversion into biofuels. Due to their projected

importance, it is important to evaluate the impact of storage on their quality. The results

from the storage of the chosen feedstocks are presented below.

3.1.1 Aerobic and Anaerobic Storage: Sweet Sorghum Bagasse

Sweet sorghum is traditionally stored for short intervals between harvest and juice

extraction, as the soluble sugars are vulnerable to microbial fermentation. Eiland et al

(1983) reported the loss of up to 50% of the soluble sugar content in 4-6 days after crop

harvest. Bellmer et al (2008) and Lingle et al (2013) both reported rapid drops in pH in

sweet sorghum following harvest, attributed to fermentative acid generation. Due to the

high soluble sugar content, ensilage has been used to preserve harvested sweet sorghum

or bagasse remnants. When storing sorghum for biofuel feedstock, ensilage is

advantageous, as the acidic conditions have been reported as rendering cellulose more

62

62

digestible (Henk & Linden, 1994). While sorghum storage has been studied in more

depth recently, the quality changes arising from storage still present a knowledge gap.

The present study contrasted moisture content and exposure to air as storage parameters

during the storage of sweet sorghum bagasse. Sweet sorghum (Keller and Sugardrip

variants) was cultivated at Purdue Agronomy Center for Research and Extension (ACRE),

harvested by hand in October 2010 and harvested s talks pressed to remove the juice. The

pressed material, termed bagasse, was dried both on the field and indoors to a final

moisture content of 15.8% w/w and shredded through a waste-chipper. Shredded bagasse

was dried further to 12% w/w moisture, or re-wetted to 25% and 68% w/w respectively,

and packaged into mini-bales using a modified log-splitter. Bagasse at 12% and 25%

moisture was stored aerobically, while that wetted to 68% moisture was stored

anaerobically in sealed bags (Refer Sections 2.1.1 in Materials and Methods).

The three treatments were examined after 24 weeks of storage, additional samples taken

at 8 weeks. Samples were examined for dry matter loss and composition changes as a

result. The dry matter losses and final moisture contents for each treatment are presented

in Table 3.1.

Substantial dry matter loss (31%) was observed upon aerobic storage at high moisture

(25.6% w/w). Samples for the same treatment gathered at 8 weeks showed 25% loss on

average, indicating the bulk of the microbial activity to be complete by this period. Dry

matter loss was lower for anaerobically stored wet bagasse (68.7% w/w moisture),

63

63

averaging 7.4% at 24 weeks. Individual anaerobic replicates showed losses ranging

between 0.4 and 15%. Aerobic storage at low moisture (12.3%) resulted in the least dry

matter loss, which averaged 4.3% at 24 weeks. Samples of the treatment at 8 weeks

showed zero dry matter loss, indicating little microbial activity upon storage.

Aerobically stored samples lost moisture throughout the storage process. Both wet and

dry bagasse dried to final moisture contents of 6-8% by weight. Biomass stored indoors

has been reported to dry to 15-20% moisture w/w which ensures subsequent stability

(Rotz & Muck, 1994), and this finding is consistent with the same.

Table 3.1. Dry Matter and Moisture Content Changes in Sweet Sorghum Bagasse after 24

Weeks of Aerobic and Anaerobic Storage

Storage

Treatment

Initial

Moisture content

(% )

24 Weeks

Final Moisture

Content

(% )

Dry Matter

Loss

(% )

Mean ± S.D.

Aerobic Dry 12.3 ± 2.0 6.5 ± 0.6 4.3 ± 1.6

Aerobic Wet 25.6 7.7 ± 1.0 31.2 ± 2.7

Anaerobic/Ensiled 68.7 ± 0.7 69.8 ± 1.2 7.4 ± 4.6

Composition analysis was carried out for original and stored (24 weeks) bagasse using

NREL procedures described in Section 2.2.1, to examine changes in individual biomass

constituents. By combining the dry matter changes with composition data, a material

balance was subsequently developed, on the basis of a unit of bagasse consisting of 100

64

64

grams (dry matter) before storage. Through this analysis, the losses of individual

chemical constituents during storage were determined. The material balance for sorghum

bagasse stored aerobically is listed in Table 3.2 (Low moisture) and Table 3.3 (High

moisture). T-test comparisons were carried out between component masses to examine

the statistical significance of the losses at the 95% confidence level.

Table 3.2. Total Dry Matter and Component Losses upon 24 Weeks of Aerobic Sweet

Sorghum Bagasse Storage at Low Moisture (12%)

Biomass

Component

Original Bagasse Aerobic Dry

(12% w/w moisture)

Composition (% )

(Mean ± S.D.)

Mass

(g/100 dry g)

Composition (% )

(Mean ± S.D.)

Mass

(g/100 dry g)

Dry matter - 100 - 95.7

Glucan 24.6 ± 0.2 24.6a 24.1 ± 0.5 23.0

Xylan 10.0 ± 0.1 10.0b 10.5 ± 0.4 10.0

b

Arabinan 1.5 ± 0.0 1.5c 1.0 ± 0.1 0.9

Lignin 9.5 ± 0.1 9.5d 11.8 ± 1.0 11.3

Sol. Sugars 40.9 40.9e 44.2 ± 2.8 41.85

e

Values with the same superscript are not statistically distinct at the 95% confidence level

Free sugars, or soluble sugars, were the principal component lost during storage.

Comparing the compositions of original bagasse to that stored wet aerobically,

approximately 30 grams of free sugars were lost per 100 grams of bagasse. Small

changes were observed in structural carbohydrate content, which were not found to be

65

65

statistically significant. Examining the composition of bagasse stored dry aerobically,

very little composition change was observed from before storage. Small losses (≈1% or

less) were seen in structural sugars. Free sugar content however was found to be

unchanged.

Table 3.3. Total Dry Matter and Component Losses upon 24 Weeks of Aerobic Sweet Sorghum Bagasse Storage at High (26%) Moisture

Biomass

Component

Original Bagasse Aerobic Wet

(26% Moisture)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Dry matter - 100 - 68.8

Glucan 24.6 ± 0.2 24.6a 35.7 ± 4.2 24.9

a

Xylan 10.0 ± 0.1 10.0b 13.6 ± 5.0 9.5

b

Arabinan 1.5 ± 0.0 1.5c 1.9 ± 0.5 1.3

c

Lignin 9.5 ± 0.1 9.5d 17.6 ± 2.1 12.3

d

Sol. Sugars 40.9 40.9e 14.3 ± 4.2 10.5

Values with the same superscript are not statistically distinct at the 95% confidence level.

The material balance for bagasse stored anaerobically is listed in Table 3.4. As with

aerobic bagasse, the predominant change observed was in the free sugar content. In the

absence of air, sugars were seen fermented into a range of products, predominantly lactic

acid. Small amounts of ethanol, glycerol, butanediol and acetic acid were also observed

(data not shown). The free sugar content of the bagasse was largely consumed, similar to

66

66

wet aerobic storage. Structural carbohydrates and lignin were enriched; their mass

increasing compared to the original. This apparent increase may be attributed to the

extreme losses in bagasse soluble sugars, which caused extreme enrichment in the

structural components.

Table 3.4. Total Dry Matter and Component Losses upon 24 Weeks of Anaerobic Sweet Sorghum Bagasse Storage

Biomass

Component

Original Bagasse Anaerobic

(68% Moisture)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Dry matter - 100 - 92.6

Glucan 24.6 ± 0.2 24.6a 39.4 ± 1.2 35.9

f

Xylan 10.0 ± 0.1 10.0b 13.1 ± 0.8 11.9

g

Arabinan 1.5 ± 0.0 1.5c 2.4 ± 0.1 2.2

h

Lignin 9.5 ± 0.1 9.5d 19.0 ± 1.0 17.2

i

Sol. Sugars 40.9 40.9e 5.5 ± 1.6 5.0

j

Lactic Acid - - 5.8 ± 0.8 5.2

Values with the same superscript are not statistically distinct at the 95% confidence level.

The results show free sugars to be the only component of interest significantly lost during

sorghum bagasse storage. In case of the bagasse stored wet aerobically, free sugar losses

account for approximately 94% of the total dry matter loss. As respiration is limited

during anaerobic storage, sugar loss thus corresponds to the generation of a variety of

67

67

fermentation products and possibly microbial biomass. The loss of soluble sugars is

concordant with forage research, as they are most accessible and easily metabolized

(Moore & Hatfield, 1994). While there are no studies on the aerobic storage of sweet

sorghum, the results of anaerobic storage corroborate the present observations. Schmidt

et al (Schmidt et al., 1997) compared ensilage treatments of sweet sorghum bagasse.

Bagasse ensiled with formic acid (0.5% w/w) was contrasted with that ensiled with

cellulase and hemicellulase enzymes. Formic acid treatment was observed to reduce pH

and preserve the bagasse with complete recovery of free sugars, while in case of the latter,

almost 50% of the free sugar content was lost due to conversion into lactic acid and other

fermentation products. Similarly, when examining anaerobic storage of whole sweet

sorghum, Williams et al (2012) compared storage treatments for recoveries of cellulose

and hemicellulose, with no assessment of free sugar losses. The effective preservation of

the free sugar fraction when storing sweet sorghum is an ongoing problem, although high

recoveries (≥ 90%) have been reported for structural carbohydrates upon ensiling

(Williams, 2012).

The results also demonstrate the need to control for moisture losses when carrying out

laboratory scale storage. Bagasse stored aerobically lost moisture through the storage

process, both low (12%) and high (25%) moisture samples drying to <10%. The moistu re

treatment thus becomes the initial moisture content, rather than the content throughout

storage. The role of moisture is hard to assess when it is being lost to the air. Moisture

loss is limited in commercial scale bales by the mass transfer rates from bale depths to

surface. In case of laboratory-scale samples however, the loss is significant and must be

68

68

prevented or accounted for. This was carried out when storing corn stover and

switchgrass.

3.1.2 Aerobic Storage of Corn Stover

As it is an important step in the overall supply chain, stover storage has been analyzed in

further depth recently. Several studies have been published by Shinners et al examining

dry matter loss under common storage conditions at the farm- and commercial-scale

(Shinners, et al., 2007a; 2007b; 2009a; 2011). Similarly, Igathinathane et al (2008)

reported the role of storage parameters in promoting fungal infestation, specifically

investigating the role of water activity to account for moisture equilibrium. While these

studies have examined changes as a result of storage conditions, the material balance for

the components lost during storage continues to present a knowledge gap, which the

present study aimed to address.

The corn stover used was originally gathered and stored by Isaac Emery for his

dissertation project (Emery, 2013). Stover was dried indoors to 13.1%, re-wetted for

storage treatments and packed using a custom-made column. The biomass was stored in

sealed containers under controlled temperature (35 °C) and water activity (0.97) for eight

weeks, as described in Section 2.1.2.2 of Materials and Methods. Unlike the switchgrass,

stover had not been processed to control particle size during storage. Samples of both

original and stored biomass were correspondingly shredded in a food processer for 30

seconds and the shredded material sieved through a 0.25-inch screen. Material retained

69

69

on the screen was shredded and sieved again. Sieved material was used in composition

and bioprocessing analyses.

Three stover samples were obtained that showed 10% dry matter loss on average.

Composition analyses were then used to determine the components lost. As with sweet

sorghum, a material balance was developed, using 100 grams (dry weight) of stover

before storage as a basis. The material balance for stover components is listed in Table

3.5. While the free sugar concentrations were not measured, they could be assumed low

(3-5% by dry weight), stover collected after grain harvest generally containing low

amounts of free/soluble sugars (Chen et al., 2007; Elander et al., 2009). A loss of

approximately 10% was seen in glucan, and 7% in xylan. Arabinan changes were not

monitored, as its initial concentration was extremely low. A small loss was observed in

the lignin content, which was found to lack statistical significance.

The carbohydrate content of the stored material is not different from that of the original,

indicating that all structural sugars were consumed with equal preference. Lignin content

is seen to increase upon storage, and its losses were the smallest, indicating little or no

consumption. Given the relative absence of free/soluble sugars, these findings are

concordant with forage studies, which predict the preferential consumption of the

structural carbohydrate fraction (Moore & Hatfield, 1994; Rotz & Muck, 1994).

70

70

Table 3.5. Total Dry Matter and Component Losses upon 8 Weeks of Aerobic Corn Stover Storage at High Moisture

Component

Original Stover Stored Stover

Composition (% )

(Mean ± SD)

Mass

(g/100 dry g)

Composition (% )

(Mean ± SD)

Mass

(g/100 dry g)

Dry Matter - 100 - 90

Glucan 31.1 ± 0.2 31.1a

31.4 ± 0.5 28.2b

Xylan 20.3 ± 0.1 20.3c

20.9 ± 0.1 18.8d

Arabinan 4.2 4.2e

3.3 ± 0.7 3.0f

Lignin 23.0 ± 0.4 23.0g 25.3 ± 1.4 22.7

g

Values with the same superscript are not statistically distinct at the 95% confidence level

The results establish the importance of the initial feedstock composition upon the storage

losses. In the absence of substantial free sugars, which are easily metabolizable and

provide energy for microbial proliferation, the dry matter losses are substantially lower

for aerobic high-moisture storage (10% loss instead of 30%), despite the controlled

humidity. These are similar to previously reported losses during aerobic indoors stover

storage, by Shinners et al (2007b) and Shah et al (2011), which ranged 1-8% upon 8-9

months of storage. The results were compared as the latter reported the bulk of the loss as

occurring early into storage, which was also observed with sorghum bagasse earlier. The

results are concordant with previous forage data, as biomass stored indoors is reported to

rapidly dry to stability (Rotz & Muck, 1994), irrespective of initial moisture. In case of

corn stover, losses are reported to depend more upon storage location and exposure than

the moisture content. Shah et al (2011) reported similar losses for low- (15-20% w/w)

71

71

and high-moisture (30-35% w/w) bales when stored outdoors wrapped in tarpaulin,

ranging 5-11%. In the same experiment, low-moisture bales stored outdoors with

breathable film cover showed 14-17% dry matter losses, indicating the exposure to be a

more important factor affecting dry matter loss. Similarly, Shinners et al reported losses

ranging 29-39% when storing stover bales wrapped in sisal twine outdoors in contact

with the ground (2007b), and 8-39% when storing stover piled on the ground without

cover (2011). Losses were reduced (<20%) by removing ground contact (storage on

pallets) or providing effective cover or wrapping (tarpaulin, net wrap, plastic twine).

When considering aerobic storage, exposure is thus more important a parameter than

moisture content. Significantly, the dry matter losses in the 2007 Shinners study included

pickup losses when gathering bale contents. These losses arise due to loss of structural

integrity in the bale, as well as breakage of bale wrap, both caused by weathering. Such

losses become significant when stover is gathered on a commercial scale.

As with the sweet sorghum, the results track for the first time the changes in the

components of interest in biomass. Shinners et al (2011) tracked changes in fiber

components in stover, calculating changes in cellulose and hemicellulose from the same

and corresponding changes in theoretical ethanol yield, using a model developed by the

NREL. The present stover compositions however are chemically defined, and can be used

to better calculate changes in ethanol yields from the biomass.

The results of sorghum and stover storage both involve sampling at a single time point.

Given that the maximal microbial activity occurs early into the storage process, frequent

72

72

sampling at this point would help better understand the progression of microbial growth

and dry matter loss. Secondly, the stover experiment used a single, high level of water

activity to examine the impact of moisture under controlled moisture conditions. The role

of water activity can be better understood by examining storage under multiple levels of

water activity. These changes were incorporated into the storage of switchgrass.

3.1.3 Aerobic Storage of Switchgrass

Previous studies on switchgrass storage have demonstrated its resistance to microbial

activity and dry matter loss. Early on, Wiselogel et al (1996) reported small (~2%)

compositional changes in switchgrass arising as a result of indoor storage losses. No

material balance was developed however, to correlate dry matter loss to composition

changes. When comparing switchgrass storage in indoor and outdoor locations (gravel,

grass sod), Sanderson et al (1997) subsequently reported losses ranging 4-6%. More

recently, Monti et al (2009) examined indoor storage to conclude switchgrass storage

losses as minor in comparison to those from harvest. The highest dry matter losses have

been reported by Shinners et al (2010), who examined the impact of bale wrapping

(sisal/plastic twine, mesh net wrap, plastic/breathable film), storage locations (indoors,

outdoors) and access to air (ensiled vs. aerobic) upon switchgrass storability. A loss of 15%

was reported with bales stored outdoors wrapped in sisal twine. Significantly, the loss

included pickup losses caused by sisal twine breakage and accompanying bale collapse.

Bales wrapped in plastic twine or net wrap showed 9% loss when stored outdoors, due to

the greater integrity. Film-wrapped bales showed statistically identical losses to those

stored indoors (~5%), indicating bale weathering to be the most significant factor in loss.

73

73

Shinners et al also examined switchgrass composition, noting drops in the concentration

of NDF and ADF in the biomass undergoing dry matter loss.

Of the studies cited above, Shinners et al alone correlated dry matter loss to composition

change, using non-additive fiber metrics. The above studies moreover did not control

moisture transfer or from the biomass. This study aims to address the knowledge gap by

establishing a material balance for the components of switchgrass stored under controlled

temperature and water activity.

Switchgrass harvested in November 2011 was dried indoors to 6% w/w moisture, milled

through a 0.25-inch screen and sieved between 0.20-inch and 0.034-inch screens. The

material retained was used for the storage experiment. Samples of this material were

analyzed through NREL procedures for composition, presented in Table 4.7 and 4.8. In

addition to the structural components presented, the switchgrass contained approximately

1% (dry basis) soluble sugars and 3% (dry basis) ethanol solubles.

The switchgrass was wetted to a range of moisture contents and packaged in plastic mesh.

The samples were stored under a range of temperatures and water activities, the storage

environments. The storage setup of switchgrass is described in Section 3.1.3 of Materials

and Methods. Stored switchgrass was sampled at 1,2,4 and 8 weeks for dry matter loss

and composition change. An extra sample, stored for redundancy, was gathered at 16

weeks.

74

74

Temperature and relative humidity were monitored using sensors, placed in the bucket.

Constant monitoring was carried out for samples stored at high-humidity. With the

exception of a single sharp rise in 35 °C bucket chambers, caused by a malfunction in the

building heating system, temperature remained within stable limits throughout. The rise

was spotted early on and the buckets moved to a room with better control. In contrast to

temperature, humidity was observed to drop at different time points. As the bucket

contents could exchange air with the outside, moisture loss over time was possible, which

led to the drops. Water was periodically added to the solutions in the bucket bases to

maintain humidity. The results of switchgrass storage under these conditions are

presented below.

3.1.3.1 Dry Matter Losses:

No significant dry matter loss was observed in switchgrass stored at water activities of

0.65-0.85 (10.6-19% w/w moisture), irrespective of temperature. At the highest water

activity (0.99, 32-34% moisture), dry matter loss was observed upon storage at 20 and

35 °C, but not at <10 °C (Figure 3.1). The dry matter losses at high-moisture treatments

are plotted in Figures 3.2.a (20 °C) and 3.2.b (35 °C), becoming significant at 2 weeks.

Losses are observed up to eight weeks, however no significant additional losses were

observed between 8 and 16 weeks.

In case of 20 °C storage (Figure 3.2.a), substantial variations were observed between

samples at each time point, resulting in standard deviations larger than the mean. At 2

75

75

and 16 weeks the mean percent losses were 0.5% and 3.6%, while the accompanying

standard deviations were 2.5 and 5.8%). The mean and standard deviation were closer in

value at 4 weeks (Mean = 2.1, Standard deviation = 2.2) and 8 weeks (Mean = 3.7,

Standard deviation = 3.79). The maximum individual dry matter loss was 9.6%, observed

after 16 weeks of storage at 20°C.

Figure 3.1. Dry Matter Loss from Switchgrass Storage at <10 °C and varying water activities. Error bars indicate standard deviations

76

76

a.

b.

Figure 3.2. Dry Matter Loss: Switchgrass Storage at a. 20 °C and b. 35 °C. Error bars indicate standard deviations

77

77

Dry matter loss was more consistent for storage at 35 °C, plotted in Figure 3.2b.

Switchgrass samples began to show dry matter loss at the second week of storage, which

became more consistent at 4 weeks, indicated by the standard deviations listed in Table

3.6. Similar to storage at 20 °C, the 8- and 16-week samples showed similar dry matter

losses, suggesting that microbial activity causing dry matter loss had ceased. The highest

loss observed was 9.5% dry matter, for one sample stored for 16 weeks at 35°C.

Composition analysis was completed for switchgrass samples that lost more than 5% of

their dry matter between sampling periods. Dry matter loss of this magnitude was reached

for samples stored for 8 and 16 weeks at both 20°C and 35°C. Correspondingly,

composition analysis was carried out for all samples stored at 35 °C for at least 8 weeks

and single samples stored at 20°C for 8 and 16 weeks. As dry matter losses were more

consistent in 35 °C samples, material balances were calculated for the three samples

stored for 8 weeks (Table 3.6) and three samples stored for 16 weeks (Table 3.7).

Similar to the corn stover, structural carbohydrates were consumed during storage, with

changes being observed in the glucan, xylan and arabinan present. At 8 weeks, glucan

changes were not found to be statistically significant at 95% confidence. At 16 weeks

however, approximately 16% of the initial glucan content was consumed. Similarly, 5%

of the initial xylan content was lost at 8 weeks, and 16% at 16 weeks. Arabinan losses

were negligible at 8 weeks, while 12% of the initial arabinan content was consumed at 16

weeks. The losses indicate that cellulose and hemicellulose are lost in similar proportion.

78

78

Table 3.6. Total Dry matter and Component Losses upon Aerobic Storage of 34% moisture Switchgrass for 8 Weeks

Component

Original Switchgrass 8 Weeks Stored

Switchgrass

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Dry Matter - 100 - 92.9

Glucan 35.9 ± 0.3 35.9a 37.1 ± 1.2 34.4

a

Xylan 22.3 ± 0.2 22.3b

22.7 ± 0.4 21.1c

Arabinan 2.6 ± 0.1 2.6d

3.0 ± 0.0 2.8e

Lignin 26.6 ± 0.3 26.6f

27.0 ± 0.4 25.0g

Values with the same superscript are not statistically distinct at the 95% confidence level

Table 3.7. Total Dry Matter and Component Losses after Aerobic storage of 34% moisture Switchgrass for 16 Weeks

Component

Original Switchgrass 16 Weeks Stored

Switchgrass

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Composition

(% )

(Mean ± SD)

Mass

(g/100 dry g)

Dry Matter - 100 - 92.4

Glucan 35.9 ± 0.3 35.9a

32.2 ± 1.8 29.8b

Xylan 22.3 ± 0.2 22.3c

20.3 ± 1.3 18.7d

Arabinan 2.6 ± 0.1 2.6e

2.5 ± 0.1 2.3f

Lignin 26.6 ± 0.3 26.6g 28.8 ± 0.6 26.6

g

Values with the same superscript are not statistically distinct at the 95% confidence level

79

79

Lignin concentrations were observed to increase over time and the 8-week samples

showed a small loss in lignin mass. At 16 weeks however, the net lignin mass was

unchanged in comparison to the original. The increase in concentration is consistent with

forage research, resulting from the preferential consumption of carbohydrates compared

to lignin (Jung & Deetz, 1994; Moore & Hatfield, 1994).

3.1.3.2 Switchgrass Microbial Activity:

The high-moisture (0.99 aw, 34%) switchgrass storage containers were observed to

develop a musty or fungal odor 2 weeks into storage. The samples drawn were seen to

have colored (black or dark green) spots, and significant amounts of dusty material

resembling fungal spores were released when the contents were transferred into storage

bags. Samples of the stored biomass were correspondingly taken to the Plant Pathology

and Diagnostic Lab at Purdue University, where they were subjected to incubation and

isolation plating, as described in Section 2.1.3.4.1 of Materials and Methods. The fungal

fruiting structures grown through these methods were identified through microscopic

examination.

The predominant fungi growing on the switchgrass were identified as belonging to Mucor,

Penicillium, Aspergillus and Fusarium genii. These fungi are known contaminants of

food-grains and forages (Pereyra et al., 2008; Mostrom & Jacobsen, 2011; Rasmussen et

al., 2011; Mobashar et al., 2012). They are also capable of degrading cellulose. In

particular, Aspergillus and Penicillium cellulases have been analyzed for use in

80

80

lignocellulose bioprocessing (Lynd et al., 2002). Similarly, Fusarium species have gained

attention as generating laccase enzymes, which degrade lignin (Kwon & Anderson, 2002;

Obruca et al., 2012; Wu et al., 2010). These fungi are thus capable of consuming the

structural carbohydrates of switchgrass. Additional fungi identified include Chaetomium,

Uromyces, Alternaria, Cladosporum and Mortierella.

3.1.3.3 Detailed Statistical Analysis: Switchgrass Storage Parameters

The data collected from the aerobic storage of switchgrass at three temperatures and four

moisture (water activity) levels was analyzed for statistical significance through

univariate split-plot analysis, using a repeated measures model.

The parameters used are listed in Table 3.8 below. The percentage dry matter was the

response variable, correlated to the storage parameters of temperature, water activity, and

storage time/duration. As temperature and water activity were established and not varied

from the start of the experiment, the two variables and their interactions were designated

whole-plot effects. For a given combination of temperature and water activity, the storage

duration or time was varied. Correspondingly, time and its interactions with the other

variables were designated split-plot effects. Three levels of temperature and four levels of

water activity were used, corresponding to the original design. As samples were taken

past 8 weeks for the highest level of water activity alone, four levels of t ime – 1, 2, 4 and

8 weeks – were used. As they were not compared beforehand, the buckets in which

storage was carried out could not be assumed identical – the three samples gathered at

81

81

each time point could not thus be treated as replicates – and were designated random

variables. As three buckets were used in every combination of temperature and water

activity, they were nested within the interaction factor.

Table 3.8. Statistical Analysis: Switchgrass Dry Matter Loss

Factor Levels

Degrees

of freedom

F-test

Compared

factor F-value P-value

Whole-plot effects

Temperature ‗T‘ 3 2

Bucket(T×aw)

13.65 0.0004

Water activity ‗aw‘ 4 3 18.72 <0.0001

T×aw interactions 12 6 3.76 0.055

Bucket(T×aw) 24 -

Split-plot effects

Time ‗t‘ 4 3

Error

17.75 <0.0001

t×aw interactions 16 9 5.9 <0.0001

t×T interactions 12 6 0.88 0.5104

T×aw×t interactions 48 18 0.99 0.4689

Error 71 -

The data was analyzed using PROC MIXED in SAS 9.3 (The SAS Institute, Cary, NC).

As seen in Table 3.8, temperature and water activity are significant at the 95% confidence

level, while their interaction is not, although its P-value is 0.055. Of the split-plot effects,

82

82

time and time×water activity interactions were were significant, having P-values smaller

than 0.0001.

Data from switchgrass stored at the highest water activity level was analyzed separately,

including the observations from 16 weeks, to inves tigate the impact of temperature and

time on dry matter loss. As with the previous analysis, buckets were nested, this time

within temperature. The data was again analyzed through PROC MIXED in SAS 9.3

(The SAS Institute, Cary, NC). As seen in Table 3.9, temperature was not significant (P-

value = 0.1239), time alone correlating strongly with the changes in percentage dry

matter.

Table 3.9. Statistical Analysis: High Moisture Dry Matter Loss

Factor Levels Degrees

of

freedom

F-test

Compared factor

F-value P-value

Whole-plot effects

Temperature ‗T‘ 3 2 Bucket(T)

13.65 0.1239

Bucket(T) 6 -

Split-plot effects

time ‗t‘ 5 4

Error

18.72 <0.0001

T×t interactions 15 8 3.76 0.1784

Error 23 -

83

83

3.1.3.4 Discussion:

Dry matter losses were observed at the highest switchgrass moisture level (0.99 aw, 34%

w/w) alone, the biomass staying stable at lower moisture levels. These observations are

consistent with the reported resilience of switchgrass to microbial infestation and dry

matter loss. Sanderson (1997), Monti (2009) and Shinners (2010) have all reported losses

averaging 5% or lower during indoor storage, the switchgrass rapidly drying to stability

before substantial dry matter loss can take place. Losses of the order of 10% or higher

were reported by Shinners upon outdoor storage, and were largely due to weathering and

bale breakage. The losses observed with the highest moisture level, ranging 5-10%, are

consistent with the above observations given that they took place under controlled

humidity. They are also consistent with the results of the wet sweet sorghum storage,

wherein the bulk of the dry matter loss was observed at 8 weeks, with little additional

loss observed at 24 weeks. In case of the high-moisture switchgrass, 8- and 16-week

samples showed very similar dry matter losses, indicating that microbial activity had

ceased between the two time points.

The present study used water activity in place of moisture concentration by weight alone,

to control for biomass drying. The objective was to examine if microbial activity could

occur at low moisture concentrations if dynamic moisture equilibrium were maintained.

The observed dry matter loss – negligible at lower moisture levels, significant at the

highest – and statistical analysis shows water activity to be a significant storage

parameter (in determining dry matter). However, they indicate a threshold level to be

required, for microbial activity to take place. These results are consistent with forage data,

84

84

which predicts low or no dry matter loss upon indoor storage at moisture concentrations

≤20% by weight. Similarly, the absence of dry matter loss upon storage at 7-9 °C

indicates a threshold temperature as necessary for microbial activity, which is below

20 °C as the latter temperature levels show dry matter loss. The greater consistency of dry

matter loss at 35 °C shows temperature to enhance microbial activity beyond the

threshold.

The material balance shows the preferential consumption of structural carbohydrates,

with similar losses (~15%) observed in glucan, xylan and arabinan. This is consistent

with the results of corn stover storage, as well as forage literature predicting the

preferential loss of carbohydrates. The results are strengthened by the predominance of

fungal genii capable of both cellulose (Aspergillus, Penicillium and Mucor) and lignin

(Fusarium). The time taken for dry matter loss to become significant might

correspondingly reflect the time taken for the fungi present to access and metabolize

sufficient carbohydrates to reach the critical mass required for growth and proliferation.

The results further illustrate the role of initial composition in affecting dry matter loss and

composition change. The losses observed with switchgrass are comparable to those

observed with corn stover (i.e. ~10%), and are much lower than those observed with

sweet sorghum. Both switchgrass and stover have low concentrations of soluble sugars,

which would provide microbes an easily accessible source of energy. Dry matter losses

are correspondingly lower for both.

85

85

3.1.4 Conclusions: Storage Parameters and Dry Matter Loss

To understand the impact of storage parameters on biomass quality, three separate

storage experiments were carried out. The experiments used different feedstocks, and

were successively refined. Presented below are the inferences drawn from them.

Both moisture content and temperature are seen to be important storage parameters, with

critical threshold levels. Biomass wetted above a specific threshold concentration and

stored above a threshold temperature is observed to undergo dry matter loss. From the

moisture levels at which dry matter loss was observed, as well as forage data (Rotz &

Muck, 1994), the threshold can be assumed in the neighborhood of 20% w/w or higher.

When below the moisture threshold, the material is observed to dry to <10% w/w, losing

insignificant amounts of dry matter en route. As established by the switchgrass results,

irrespective of humidity, biomass wetted below the threshold will be stable and lose little

or no dry matter. As the highest moisture levels alone were above the threshold, the

significance of moisture content above the threshold could not be investigated. Similarly,

a threshold temperature is required for microbial activity to occur, as seen in the absence

of dry matter loss upon storage at 7-9 °C. While both 20 and 35 °C are above this

threshold, the greater consistency in dry matter loss at the latter temperature indicates

temperature to be a significant factor above the threshold, consistent with general

microbiology.

Under temperatures and moisture levels conducive to microbial activity, aerobic storage

losses for biomass are observed to depend upon the species composition, as sweet

86

86

sorghum bagasse showed almost three times the dry matter loss (31%) as corn stover

(10%) and switchgrass (7.5%), even though it was wetted to a lower concentration (26%

w/w) than the latter (~34% w/w). This difference can be attributed to the substantially

higher concentration of free sugars in the sorghum (40% w/w), given that free sugars

accounted for 94% of the dry matter loss in high-moisture aerobic sorghum bagasse. This

is consistent with forage research, as free sugars are the easiest fraction of biomass for

microbes to consume.

Dry matter loss involves the loss of carbohydrates throughout all the experiments.

Cellulose and hemicellulose were consumed when storing corn stover and switchgrass. In

case of sweet sorghum, while some structural carbohydrate loss was observed during

aerobic dry storage, free sugars alone were consumed during the aerobic wet storage of

bagasse, possibly because the microbial species capable of rapidly metabolizing free

sugars dominated the biomass surface.

3.2 Feedstock Bioprocessing: Impact of Storage Losses

Following the precise quantification of storage losses, it was necessary to examine their

impact on the conversion efficacy and yield of the structural carbohydrates (cellulose and

hemicellulose) remaining. Due to the preferential consumption of free and cellulosic

sugars, and corresponding enrichment of biomass lignin, feedstocks that have undergone

significant storage losses may be expected to be more recalcitrant to bio-based

breakdown, as reported for stored forages (Jung, 1989). To examine this hypothesis,

samples of original and stored biomass were subjected to enzymatic hydrolysis, with and

87

87

without liquid hot-water pretreatment. The material balances upon pretreatment and

glucan yields upon enzyme hydrolysis were compared to determine what impact, if any,

dry matter loss had upon cellulose conversion.

Sweet sorghum bagasse was excluded from these analyses, as no significant change had

occurred in its cellulose and hemicellulose contents. As switchgrass and corn stover had

undergone losses in cellulose and hemicellulose during storage, they were analyzed for

sugar release as a result of liquid hot-water pretreatment and cellulase hydrolysis.

3.2.1 Pretreatment: Comparison of Stored and Pre-Storage Biomass

Corn stover and switchgrass were both subjected to liquid hot-water pretreatment at 15%

w/w solids loading in 35 ml sealed, stainless steel reactors. Pretreated solids and liquor

were separated and each analyzed for biomass components – carbohydrates and lignin –

and a material balance developed for their fractionation. The pretreatment and subsequent

analyses are described in further detail in Section 2.2.2.1 of Materials and Methods.

3.2.1.1 Corn Stover Pretreatment:

Shredded corn stover was pretreated at 190 °C for 15 minutes, with an additional 7.66

minutes for heat up. The conditions were based on the optimal pretreatment conditions

reported by Mosier et al (2005b) and Elander et al (2009). Following pretreatment, the

slurry was separated into solid and liquid fractions, each of which was analyzed for

biomass constituents.

88

88

As shown in Table 3.10, the glucan and lignin content were enriched to 47% and 33%

respectively, while the xylan content was reduced to 10%. Combined with chemical

analysis of the pretreatment liquid, a material balance was developed for the fractionation

of stover components using a 100 dry grams of stover as a basis, shown in Table 3.11.

Table 3.10. Composition of Pre-storage Corn Stover: Before and After Liquid Hot-Water Pretreatment

Component

Composition (% Dry Weight)

(Mean ± SD)

Original Stover Pretreated Solids

Glucan 31.1 ± 0.2 47.3 ± 2.6

Xylan 20.3 ± 0.1 10.7 ± 0.6

Arabinan 4.2 ± 0.0 0.6 ± 0.0

Lignin 23.0 ± 0.4 33.7 ± 0.7

Approximately 59% of the total dry matter loaded for pretreatment was accounted for.

The highest extent of closure was achieved with the glucan content, with 89% of the

original glucan accounted for within the pretreated solids, and another 9% solubilized

into gluco-oligomers, monomeric glucose and hydroxymethyl furfural (from glucose

dehydration). Substantially lower closures are obtained for xylan (69%) and arabinan

(39%), possibly due to the formation of degradation products – furan complexes – not

detectible through liquid chromatography. Similarly, the solubilized phenolic compounds

formed through lignin breakdown could not be detected through chromatography and

correspondingly only the fraction retained within the solids is accounted for. This

89

89

constitutes 86% of the original lignin content present. The low degree of solubilization of

cellulose and lignin reflects their recalcitrance to chemical breakdown.

Table 3.11. Fractionation of Pre-Storage Stover Components upon Liquid Hot-Water

Pretreatment

Component

Post-

pretreatment Medium

Percentage of

original (% )

(Mean ± SD)

Total

Closure

Dry Matter Solid 58.7

58.7 Liquid Not Determined

Glucan Solid 89.3 ± 5.0

97.8 Liquid 8.5 ± 2.3

Xylan Solid 30.7 ± 1.6

68.7 Liquid 38.0 ± 11.7

Arabinan Solid 8.9 ± 0.4

37.8 Liquid 28.8 ± 4.6

Lignin Solid 86.2 ± 1.7 86.2

The pretreatment of stored stover samples (that lost 10% of their dry matter content),

resulted in solids with composition listed in Table 3.12. Glucan, xylan and lignin contents

are all similar to those seen in solids generated by pretreatment of pre-storage stover. The

fractionation of the components into the liquid and solid phases during pretreatment using

100 dry grams of stored stover is shown in Table 3.13.

90

90

Table 3.12. Composition of Post-storage Corn Stover (8 Weeks, 35 °C, 0.97 aw): Before and After Liquid Hot-Water Pretreatment

Component

Composition (% Dry Weight)

(Mean ± SD)

Stored Stover Pretreated Solids

Glucan 31.4 ± 0.5 45.2 ± 1.1

Xylan 20.9 ± 0.1 11.1 ± 0.7

Arabinan 3.3 ± 0.7 0.7 ± 0.0

Lignin 25.3 ± 1.4 36.1 ± 0.2

Table 3.13. Fractionation of Stored Stover Components upon Liquid Hot-Water Pretreatment

Component

Post-

pretreatment Medium

Percentage of

original (% )

(Mean ± SD)

Total

Closure

Dry Matter Solid 62.4

62.4 Liquid Not Determined

Glucan Solid 89.9 ± 2.1

94.4 Liquid 4.5 ± 0.8

Xylan Solid 33.1 ± 2.0

64.6 Liquid 31.5 ± 7.7

Arabinan Solid 13.5 ± 0.0

45.2 Liquid 31.7 ± 2.9

Lignin Solid 88.9 ± 0.4 88.9

91

91

The fractionation is seen to be similar to that of pre-storage stover, with about 62% of the

loaded dry matter accounted for. Similar degrees of closure were established for glucan

(94%) and lignin (89%). Small differences (~5%) were seen in both the overall closure

and the solid and liquid fractions of the xylan and arabinan. The similarities indicate

fractionation characteristics to be unaffected by storage changes. To further examine this

possibility, switchgrass pretreatment was analyzed.

3.2.1.2 Switchgrass pretreatment:

Switchgrass was pretreated at 200 °C for 10 minutes, with an additional 7.66 minutes of

heat up time. The pretreatment conditions were based on the optimization studies

published by Kim et al (2011a) and Garlock et al (2011) for maximum glucose yields.

Similar to stover, material balances were developed for the fractionation of original and

stored switchgrass between the liquid and solid phases during pretreatment.

Table 3.14. Composition of Pre-Storage Switchgrass Before and After Liquid Hot-Water Pretreatment

Component

Composition (% Dry weight)

(Mean ± SD)

Original

Switchgrass

Pretreated

Solids

Glucan 35.9 ± 0.3 43.5 ± 3.9

Xylan 22.3 ± 0.2 11.3 ± 1.8

Arabinan 2.6 ± 0.1 1.0 ± 0.2

Lignin 26.6 ± 0.3 30.6 ± 0.9

92

92

The pretreatment of switchgrass generated solids with the chemical composition shown

in Table 3.14. Glucan content was enriched to from 36% to 44% and lignin from 27% to

31%. By contrast xylan content was reduced to from 22% to 10%. The enrichment of

glucan and lignin in the solids is consistent with the results of corn stover pretreatment.

The corresponding fractionation of these components between the pretreated solids and

pretreatment liquid is shown in Table 3.15.

Table 3.15. Fractionation of Pre-Storage Switchgrass Components between Solids and

Liquid during Liquid Hot-Water Pretreatment

Component Medium

Percentage of original

(% )

(Mean ± SD)

Closure

Dry Matter Solid 71.7 ± 0.6

71.7 Liquid Not Determined

Glucan Solid 86.9 ± 7.1

90.3 Liquid 3.4 ± 0.2

Xylan Solid 36.1 ± 5.4

80.0 Liquid 43.9 ± 3.7

Arabinan Solid 27.1 ± 6.5

73.2 Liquid 46.1 ± 1.0

Lignin Solid 82.3 ± 2.2 82.3

A lesser degree of solubilization was observed upon switchgrass pretreatment, as 72% of

the original dry matter was retained in the solid fraction. Approximately 87% of the

93

93

glucan was retained in the solids, while a small fraction was solubilized. In case of xylan,

36% of the original content was retained in the solids, and 44% solubilized into xylo -

oligomers, monomeric xylose or furfural. Arabinan was similarly solubilized, 27%

retained in the solids and 46% converted into arabino-oligomers or free arabinose. A

greater degree of closure was achieved for xylan (80%) and arabinan (73%), possibly as

less material was converted into undetectable furan degradation products.

Table 3.16. Composition of Stored Switchgrass Before and After Liquid Hot-Water

Pretreatment

Component

Composition (% )

(Mean ± SD)

Stored Switchgrass

Pretreated Solids

Glucan 37.1 ± 1.2 47.0 ± 1.9

Xylan 22.7 ± 0.4 12.8 ± 0.5

Arabinan 3.0 ± 0.0 0.8 ± 0.1

Lignin 27.0 ± 0.4 32.9 ± 1.0

For comparison, switchgrass stored for 8 weeks at 35 °C, 0.99 aw was pretreated

identically. The solids generated by stored switchgrass pretreatment have similar

composition (Table 3.16), with slightly higher concentrations of glucan and lignin

compared to pretreat switchgrass without storage. The corresponding material balance for

stored switchgrass pretreatment is listed in Table 3.17.

94

94

Table 3.17. Fractionation of Stored Switchgrass Components between Solids and Liquid during Liquid Hot-Water Pretreatment

Component Medium

Percentage of original (% )

(Mean ± SD)

Closure

Dry Matter Solid 71.9 ± 4.7

71.9 Liquid Not Determined

Glucan Solid 90.8 ± 7.8

92.6 Liquid 1.9 ± 0.3

Xylan Solid 41.3 ± 15.9

81.0 Liquid 39.7 ± 8.7

Arabinan Solid 20.3 ± 9.6

51.3 Liquid 31.0 ± 3.4

Lignin Solid 87.8 ± 3.9 87.8

The fractionation was seen to be similar to that of pre-storage switchgrass. As with the

original material, 72% of the dry matter loaded was retained in the pretreated solids.

Glucan and lignin fractionation were similar to the pre-storage material as well, although

a slightly higher (~5%) degree of closure was achieved with both. In case of

hemicellulosic sugars, xylan fractionation is identical to that of pre-storage switchgrass,

while lower amounts of arabinan were detected in both pretreated solids and liquid,

resulting in a lower degree of closure (51%) than pre-storage switchgrass (72%). Due to

the low concentration of arabinan in the biomass, this was considered insignificant.

95

95

When comparing feedstocks, significant differences are observed between s tover and

switchgrass pretreatment. A higher fraction of the loaded dry matter is solubilized in case

of the former, indicating lesser recalcitrance compared to switchgrass. Storage treatments

however were not observed to cause significant differences in biomass pretreatment

characteristics, indicating little structural changes as a result of carbohydrate loss. To

further examine storage impact on cellulose reactivity, enzyme hydrolysis was carried out,

and storage treatments compared.

3.2.2 Cellulose Hydrolysis

Cellulose hydrolysis was subsequently carried out with washed pretreated solids

generated from original and stored feedstocks (stover and switchgrass). Hydrolysis was

carried out in 250 ml NalgeneTM

bottles, into which pretreated solids and 50 mM citrate

buffer (pH 4.8) were added, to a final dry solid concentration of 1% w/w. Cellulase

(NS22086) and cellobiase (Novo 188) were added to concentrations of 50 FPU and 130

CBU per gram of glucan respectively, the total protein concentration being 109.6 mg-

protein/g-glucan. Hydrolysis was carried out at 50 °C for 48 hours, and samples taken at

0, 24 and 48 hours. The percent glucan yield was calculated from the glucose

concentration. For comparison, non-pretreated biomass – stover and switchgrass – was

also subjected to hydrolysis under identical conditions. Hydrolysis and accompanying

analyses are described in further detail in Section 2.2.2.2 of Materials and Methods.

The results of corn stover hydrolysis are presented in Table 3.18. No change was

observed in cellulose reactivity as a result of storage. Both original and stored stover

96

96

showed hydrolysis yields of approximately 16% of theoretical without pretreatment,

while the washed pretreated solids, upon hydrolysis, resulted in 55 – 58% yields. A T-test

comparison was carried out for the 48-hour glucan yields between original and stored

stover. The glucan yields from non-pretreated stover were statistically indistinct with a P-

value of 0.95 and those from pretreated stover were found to be statistically indistinct

with a P-value of 0.10.

Table 3.18. 48-Hour Hydrolysis Yield of Pretreated Corn Stover: Original vs. Stored

Biomass

Corn Stover

% Glucan Yield

(Mean ± SD)

Non-Pretreated

Biomass

Washed Pretreated

Solids

Original 16.1 ± 1.1 54.7 ± 2.9

Stored

(8 Weeks, 35 °C, 0.97 aw) 15.9 ± 5.3 58.5 ± 1.1

The glucan yields upon switchgrass hydrolysis are listed in Table 3.19. Switchgrass

samples were found have significantly lower digestibility compared to corn stover, as

indicated by the 48-hour glucan yields. As with stover, storage was not observed to cause

significant changes in cellulose reactivity. Non-pretreated switchgrass was observed to

have uniformly low digestibility, with 7-8% of the glucan present digested in case of both

original and stored switchgrass samples. Pretreated solids were more digestible, 34% of

the glucan being hydrolyzed on average in case of original switchgrass, and 34-43% on

average in case of the stored bales. Paired T-test comparisons showed the hydrolysis

97

97

yields of non-pretreated switchgrass to be statistically indistinct with a P-value of 0.77,

and of pretreated switchgrass to be indistinct with a P-value of 0.11.

Table 3.19. 48-Hour Hydrolysis of Pretreated Switchgrass: Original vs. Stored Biomass

Switchgrass

% Glucan Yield

(Mean ± SD)

Non-Pretreated

Biomass

Washed

Pretreated Solids

Original 8.4 ± 2.4 34.2 ± 3.2

Stored

(8 Weeks, 35 °C, 0.99 aw) 7.7 ± 2.9 37.7 ± 4.8

Note: Glucose release during stored switchgrass hydrolyses was measured through the

ReliOn® glucose meter alone (Ref. Appendix B.2)

The above results establish that storage losses, while reducing the amount of

carbohydrate that can be converted into biofuels or chemicals, do not alter the reactivity

or extractability of the s tructural carbohydrate content within lignocellulose.

3.2.2.1 Feedstock bioprocessing: Particle Size

While hydrolysis yields for switchgrass and corn stover were unaffected by storage losses,

they were significantly lower than those previously reported for both corn stover and

switchgrass. Under optimal pretreatment conditions, 90% glucan hydrolysis was reported

by Mosier et al (2005b) upon hydrolysis under conditions corresponding to the NREL

LAP 009 (50 °C, 200 rpm, 1% glucan w/v, 168 hours) and at a lower enzyme loading of

15 FPU/g-glucan. When investigating optimal pretreatment for switchgrass bioprocessing,

98

98

Garlock et al (2011) reported close to 95% glucan hydrolysis with pretreated switchgrass

solids upon hydrolysis corresponding to NREL LAP 009, and at enzyme loadings of 18-

20 FPU/g-glucan cellulase and 30-40 CBU/g-glucan cellobiase. Comparing switchgrass

ecotypes, Kim et al reported 83-87% glucan yields when hydrolyzing pretreated

switchgrass solids under identical conditions to Garlock et al. Despite the longer

hydrolysis times, the higher yields are significant as they were achieved at much lower

enzyme loadings, and because the bulk of the hydrolytic activity occurs within the first

24-48 hours. This difference bore investigating.

Upon examining feedstock pre-processing, the principal difference found was in

feedstock particle size. In case of switchgrass, Garlock et al utilized material that was

knife-milled to pass through a 2 mm (Corresponding to the ASTM No. 10 Mesh) screen,

while Kim et al utilized material that was milled to pass through an ASTM No. 40 Mesh

(0.42 mm) screen. Similarly, the results reported by Mosier et al used stover that had

been knife-milled to pass through a 0.25-inch screen and milled before pretreatment to

pass through an ASTM No. 40 Mesh screen. As particle size has been reported as

substantially impacting feedstock bioprocessing efficacy (Khullar et al., 2013; Pedersen

& Meyer, 2009; VanWalsum et al., 1996; Vidal et al., 2011; Yeh et al., 2010), the

difference in particle size may account for the differences observed.

99

99

Table 3.20. Feedstock Particle-Size Distribution

Particle Size

% By Weight

Current Feedstocks 2001 Corn

Stover Switchgrass Corn Stover

>20-Mesh 77.2 - 32.5

Between 20- & 40-

Mesh 22.8

12.2 16.1

<40-Mesh 26.7 50.2

Total 100 38.9 99.3

To examine the possible role of particle size, switchgrass and stover samples were sieved

through ASTM No. 20 (0.85 mm) and No. 40 mesh screens to examine particle size

distributions. Samples of knife-milled corn stover prepared in 2011 that had been used by

Mosier et al were also sieved for comparison. The results are listed in Table 4.20. The

older stover sample was observed to have a higher proportion of particles small enough to

pass through both screens. The older stover sample was subsequently pretreated and the

washed solids subjected to identical enzyme hydrolysis, resulting in a glucan yield of

83%. Pretreatment and hydrolysis were subsequently repeated with switchgrass that had

first been milled to pass through a 40-Mesh screen. Hydrolysis of the washed pretreated

solids resulted in a glucan yield of approximately 76%, substantially closer to the yields

reported in the literature. When repeated at one-third the enzyme loading, hydrolysis

resulted in a glucan yield of 42%, slightly higher than what had been achieved with

whole switchgrass at three times the enzyme loading. The results all strongly indicate

particle size to strongly affect glucan yield during hydrolysis.

100

10

0

3.2.3 Conclusions: Bioprocessing

Dry matter losses have been reported to reduce the digestibility of forage biomass, due to

the preferential consumption of carbohydrates and corresponding enrichment of lignin.

Correspondingly, the impact of storage losses upon biomass stored for use as biofuel bore

investigation.

The key finding of the bioprocessing analyses in the present study is that storage losses to

the observed extents (7-10%) do not alter the reactivity of cellulose to hydrolytic

enzymes or correspondingly, the extractible yield (gdry ,extracted/gdry,available) of the structural

carbohydrates. This is in contrast to forages. The predominant factor accounting for this

difference is (liquid hot-water) pretreatment, which opens up the lignocellulosic structure

and substantially increases the accessibility of the cellulose to enzyme action (Mosier et

al., 2005c; Yang & Wyman, 2008). Pretreatment has been proposed as a means to

improving forage digestibility before feeding to ruminants (Fahey et al., 1994), and can

thus compensate for any changes resulting from storage losses.

Due to their preferential consumption, carbohydrates are observed to make up 50-90% of

the total losses. Their loss represents a reduction in the amount of ethanol that can be

manufactured per dry ton of harvested feedstock. This has significant economic

implications for biofuel manufacturing given the various energy and chemical inputs

required for the cultivation and harvest of biomass feedstock. Moreover, as the

metabolized fractions are converted into chemical species with global warming potential

(Emery & Mosier, 2012), these losses also have the potential to reduce the environmental

101

10

1

gains of cellulosic biofuel utilization. Storage must thus be optimized so as to minimize

losses.

The subsequent finding has been the role of particle size in the efficacy of enzymatic

digestion. Of the three whole feedstocks – corn stover from 2001, switchgrass and corn

stover from 2012 – compared, the glucan yields upon hydrolysis have been directly

correlated to the fraction by mass of small particles (<No. 40 Mesh). Hydrolysis of

washed solids generated from milled switchgrass pretreatment resulted in 76% glucan

yield, compared to 34% yield upon hydrolysis of pretreated whole switchgrass. Even

when comparing non-pretreated biomass, corn stover from 2012, which had a

substantially higher fraction of small particles than switchgrass, showed a greater glucan

yield upon hydrolysis. In comparison to storage losses and composition changes, particle

size treatments were observed to have a significant impact on the extractable yield of the

cellulose. Cellulose accessibility being a critical factor in the efficacy of hydrolysis

(Arantes & Saddler, 2011; Jeoh et al., 2007), the improvements in yield may be attributed

to the increased accessibility of biomass particles to the enzymes, as a result of smaller

particle size. Further investigation is required, to establish the role of biomass particle

size on the efficacy of lignocellulose bioprocessing.

102

10

2

CHAPTER 4. CONCLUSIONS AND RECOMMENDATIONS

4.1 Conclusions

The current study presented conclusions on the impact of various storage parameters

upon biomass composition, and correspondingly its bioprocessing quality. Presented

below are the conclusions that could be drawn from the data presented previously relating

to the factors contributing to storage loss and the impact of these losses on bioprocessing

performance.

The objectives of this study had been coalesced into three hypotheses corresponding to:

1. The Impact of Moisture

2. The Preferential loss of Carbohydrates over other constituents

3. The Impact of Dry matter losses on Bioprocessing performance

Moisture content was hypothesized to strongly affect dry matter losses during aerobic

storage. In line with forage data, moisture contents greater than 20-25% w/w were

expected to show significant dry matter loss. This hypothesis was largely confirmed by

the data. A threshold moisture content was seen as necessary for significant microbial

activity and loss. In case of sweet sorghum, 26% w/w moisture was found sufficient to

cause substantial dry matter loss, compared to 33% for both corn stover and switchgrass,

103

10

3

indicating that the threshold is determined by the feedstock. Even when placed under

controlled humidity, moisture contents below the threshold resulted in insignificant dry

matter loss.

When water activity was used as a parameter in place of moisture concentration, no loss

was observed at levels below 0.97 for corn stover and switchgrass. The lower water

activity levels (0.65-0.85) were found to correspond to moisture contents of 20% w/w or

lower, corroborating results form forage research. While sweet sorghum water activity

was not measured, the substantial differences in composition (notably free sugar content)

could allow for higher water activity at lower moisture concentrations, the threshold for

microbial activity correspondingly being lower. The statistical analysis of switchgrass

data showed water activity to be a statistically significant contributor to dry matter loss,

establishing its importance as a parameter, and thereby that of moisture content.

The second initial hypothesis was that carbohydrates would be preferentially consumed

during storage, based on the background established by forage storage. This hypothesis

was confirmed by the material balance developed for storage losses.

Of the components measured during composition analysis, changes were observed in the

carbohydrate content alone upon storage losses. Lignin was not consumed during storage,

its concentration increasing as a result of the losses and significant changes in the total

dry mass. When examining structural carbohydrate loss, glucan (a cellulose component)

and xylan (a hemicellulose component) were consumed in approximately equal

104

10

4

proportion. Although hemicellulose is more accessible to microbes (Mosier et al., 2005c;

Zhao et al., 2012a), preferential consumption was not observed. It may be inferred from

the present data that structural carbohydrates are consumed with no specific preference

(cellulose vs. hemicellulose). Carbohydrate losses comprised 56-94% of the total dry

matter losses observed in the feedstock, indicating their preferential consumption.

Furthermore, the content of easily accessible carbohydrates (i.e. soluble or free sugars)

was observed to substantially affect storage losses, with sweet sorghum bagasse showing

close to 25% dry matter loss in 8 weeks, compared to 7-10% loss for corn stover and

switchgrass. On this basis, the preferential consumption of carbohydrates upon storage is

confirmed.

The final hypothesis tested was the impact of storage losses upon the hydrolytic

extractability of glucose from the remnant biomass, even after pretreatment. Dry matter

losses result in carbohydrate loss and the enrichment of lignin in stored biomass. When

stored for animal feed, the remnant has consistently been reported as less digestible by

livestock in comparison to the original (Jung, 1989; Rotz & Muck, 1994). Given the role

of lignin in impeding cellulose extraction and breakdown (Chapple et al., 2007; Mosier et

al., 2005c; Studer et al., 2011; Zhao et al., 2012a), its enrichment was hypothesized to

increase biomass recalcitrance. Upon examination, this hypothesis was disproved by the

data in the present study.

The identical (at 95% confidence level) glucan release from original and feedstocks for

both switchgrass and stover, established that dry matter losses in the range seen (5-10%)

105

10

5

do not alter the bioprocessing characteristics of the feedstock remnant. Non-pretreated

samples were seen to have equally low extractability, while pretreatment was observed to

increase the enzymatic digestibility of original and stored feedstocks alike. On a large

scale, while the carbohydrate losses represent a loss in fermentable ethanol per dry ton of

feedstock or per unit area of arable land, their bio-based extractability on a dry mass basis

can be expected to stay the same based upon the present study. This difference may be

attributed to pretreatment, which breaks open the lignocellulose structure and improves

digestibility, overcoming any structural impedances arising from lignin enrichment.

The data presented further conclusions apart from those corresponding to the initial

hypothesis.

Similar to water activity, temperature is also observed to require a threshold, in order for

microbial activity to occur. Dry matter losses being observed only at the highest water

activity level (0.99), the effect of temperature at these levels must be examined. Low dry

matter loss (~4%) was observed upon switchgrass storage at low temperatures, indicating

that low temperatures preserve the material. Switchgrass storage at 20 °C resulted in

losses ranging 1.4-8.7% in 8 week samples, and 0-9% in 16 week samples. Greater

consistency was observed upon 35 °C storage, dry matter losses ranging 5-9% at both 8

and 16 weeks. Statistical analysis of the switchgrass data established temperature as a

significant contributor to dry matter loss as a whole, although analysis of high moisture

data alone showed it to lack statistical significance. Further investigation is thus required

to confirm its significance.

106

10

6

Storage analyses have typically been for durations of 6 months or higher (Coblentz, 2009;

Khanchi et al. 2009; Martinson et al., 2011; Rigdon et al., 2013; Shinners et al., 2007b;

2011; Williams, 2012), to reflect the long storage periods in practice. Sweet sorghum

bagasse was carried out for 6 months. The dry matter loss observed in 6-month samples

(31%) however was not substantially higher than in subsamples that were taken at 8

weeks (25%), indicating that the maximum activity had transpired by that point. When

weekly samples were taken during switchgrass storage, individual losses ranging 3-5%

were seen at four weeks. Maximum individual dry matter losses of approximately 9%

were seen in 8-week and 16-week samples, and the averages of both time points were

close (7%), indicating little activity occurring after eight weeks. Thus, biomass stored

indoors under controlled aerobic conditions reaches stability in approximately eight

weeks, with little further activity seen afterwards. The changes observed in forages may

be due to environmental changes – temperature and precipitation – that the biomass is

exposed to during year-round outdoors storage. Additionally, while 8- and 16-week

samples showed similar overall dry matter loss, greater glucan and xylan (6%) losses

were observed in the latter compared to the former (1-2%). As storage was aerobic, the

data presents the possibility that microbial respiration is accompanied by biomass

generation, reducing the measured dry matter loss. Further study is required to confirm

this possibility.

Apart from the feedstock species, another factor influencing the free sugar concentrations

is the harvest date. Sorghum was harvested before senescence and was correspondingly

rich in free sugars. In comparison, switchgrass, which was harvested well after

107

10

7

senescence, had extremely low (<1%) of free sugars and can be assumed to contain lower

amounts of protein as well. Variation has been reported in soluble sugar content with

harvest date and crop growth stage, in switchgrass (Bélanger et al., 2012), sorghum (Han

et al., 2012; Zhao et al., 2012b) and maize (Lynch et al., 2012). Thus, both crop species

and time of harvest must be taken into consideration when designing storage

environments.

Finally, particle size was found to influence bioprocessing. In contrast to storage

treatments, altering the particle size by milling was observed to bring about substantial

changes in switchgrass pretreatment and hydrolysis characteristics. This observation

indicates particle size to be a key factor in cellulose reactivity during pretreatment and

hydrolysis, and is concordant with the findings of Zeng et al (2007) linking particle size

to the efficacy of cellulose hydrolysis.

In addition to the above conclusions, the study presented several areas of exploration,

which are described in further detail below.

4.2 Recommendations:

The present study used ‗mini-bales‘ containing 0.1-2.5 dry kilograms of biomass. As

commercial-scale baling typically packages 500+ dry kilograms per bale (Shinners et al.,

2009b; 2010), the role of the larger mass and density must be investigated. The higher

density limits the transfer of air and moisture at different depths. Farm scale bales must

be correspondingly sampled at different depths to accurately assess overall moisture

108

10

8

content, dry matter loss and composition change (Shinners et al., 2007b; 2009b; 2010).

The transfer limitations can lead to different mois ture and microbial regimes within the

same bale, significantly altering the role of moisture concentration by weight. At high

densities, bale heating due to microbial respiration is a significant possibility, as heat

transfer is limited alongside mass transfer. As the generated heat can enable further

microbial activity, the role of external temperature is also altered upon large-scale

baleage (Coblentz et al., 1997; 2000; Coblentz & Hoffman, 2009a; 2009b; Turner et al.,

2002). Lastly, when biomass is baled on a commercial scale, the losses resulting from

loss of bale integrity become significant, as gathering or picking up broken or scattered

bale contents is commercially unfeasible. Thus dry matter recovery can reduce

substantially without corresponding microbial activity, if bales collapse or break, as

reported by Shinners et al (2007b) and Monti et al (2009). For these reasons, future

research must attempt to incorporate bale size (dimensions) and density into storage

analysis.

Effective biomass storage must preserve the significant fractions of biomass at minimal

cost. It is therefore important to compare the potential loss in biomass value (dollars/dry

metric ton) with the cost (dollars/dry metric ton) of storing it.

While the overall cost of harvesting and supplying cellulosic biomass for conversion into

biofuel have been researched extensively (Sokhansanj et al., 2009; Suh et al., 2011;

Brechbill et al., 2011; Judd et al., 2012), fewer studies have been carried out on storage

costs specifically. Cundiff et al (1996) reported the costs of storing baled switchgrass,

109

10

9

depending upon bale shape, machine configurations and storage parameters (wrapping,

surface etc.). Subsequently, Thorsell et al (2004) and Mooney et al (2012) have modeled

the costs of switchgrass storage, the latter investigating changes in selling price arising

from dry matter loss. Nevertheless, the assignation of a specific cost per dry metric ton to

store a given feedstock under specific parameters represents a knowledge gap.

Conversely, multiple means have been used to estimate the value of stored biomass, and

changes as a result of storage loss. Shinners et al examined changes in Theoretical

Ethanol Yield as a result of dry matter loss and composition change (2011), while

Mooney (2012) developed loss models for selling price fluctuations as a result of storage

loss. Value assignation on a monetary basis requires analysis or assumptions on the yields

of products that can be manufactured from the feedstock, as well as their selling price, as

was recently reported by Humbird et al (2011). While empirically determined selling

prices have been developed, based on biomass moisture content and integrity (Project

Liberty, n.d.), further development is required to develop biomass value on a dry weight

basis, so as to compare to storage cost. The development of the above metrics, when

combined with storage data, can be used to develop a database for the outcomes arising

from the storage of specific feedstocks under specific conditions.

The improvement observed in cellulose hydrolysis resulting from particle size reduction

indicates particle size as a point of further investigation. When in the micrometre range,

decreasing the particle size of input feedstock has been reported to improve the efficacy

of pretreatment (Chundawat et al. 2006) and glucan yields upon enzyme hydrolysis

110

11

0

(Chundawat et al., 2006; Zeng et al., 2007; Yeh et al., 2010; Khullar et al., 2013). The

role of particle size is not clear, however, as when in the centimeter range, increasing

particle size is reported to improve both pretreatment fractionation and hydrolysis yield

(Harun et al., 2013; Liu et al., 2013). Zhang et al reported the impact of particle size

reduction as possibly confounded with the changes in cellulose crystallinity resulting

from its milling (Zhang et al., 2012), although particle size reduction improved

hydrolysis and fractionation when crystallinity was controlled. Similarly, while Zeng et al

reported lower particle size as improving the cellulose reactivity, the improvement was

negated upon pretreatment. Based on these observations, the precise role of particle size

in improving biomass fractionation upon pretreatment and subsequently cellulose

reactivity upon hydrolysis forms an important area of investigation.

11

1

BIBLIOGRAPHY

111

11

1

BIBLIOGRAPHY

Adani, F., Papa, G., Schievano, A., Cardinale, G., D‘Imporzano, G., & Tambone, F. (2011). Nanoscale Structure of the Cell Wall Protecting Cellulose from Enzyme Attack. Environmental Science & Technology, 45(3), 1107–1113. doi:10.1021/es1020263

Akin, D. E. (1989). Histological and Physical Factors Affecting Digestibility of Forages. Agronomy Journal, 81(1), 1–9.

Akin, D. E., Morrison, H. W., Rigsby, L. L., Barton, F. E., II, Himmelsbach, D. S., & Hicks, K. B. (2006). Corn Stover Fractions and Bioenergy. Appl Biochem Biotechnol, 129(1 - 3), 104–116.

Anderson, W. F., Dien, B. S., Jung, H.-J. G., Vogel, K. P., & Weimer, P. J. (2009).

Effects of Forage Quality and Cell Wall Constituents of Bermuda Grass on Biochemical Conversion to Ethanol. BioEnergy Research, 3(3), 225–237. doi:10.1007/s12155-009-9063-9

Arantes, V., & Saddler, J. N. (2011). Cellulose accessibility limits the effectiveness of

minimum cellulase loading on the efficient hydrolysis of pretreated lignocellulosic substrates. Biotechnology for Biofuels, 4(1), 3. doi:10.1186/1754-6834-4-3

ASABE. (2012). Moisture Measurement — Forages (No. ANSI/ASAE S358.3) (pp. 1–3). ASABE. Retrieved from https://elibrary.asabe.org/

Bellmer, D. D., Huhnke, R. L., & Kundiyana, D. (2008). Issues with In-Field

Fermentation of Sweet Sorghum Juice. Presented at the 2008 ASABE Annual International Meeting, Providence, Rhode Island.

Besle, J.-M., Cornu, A., & Jouany, J.-P. (1994). Roles of Structural Phenylpropanoids in Forage Cell Wall Digestion. Journal of the Science of Food and Agriculture, 64(2), 171–190. doi:10.1002/jsfa.2740640206

Bélanger, G., Savoie, P., Parent, G., Claessens, A., Bertrand, A., Tremblay, G. F., et al.

(2012). Switchgrass silage for methane production as affected by date of harvest. Canadian Journal of Plant Science, 92(6), 1187–1197. doi:10.4141/cjps2011-202

112

11

2

Biomass Producer Handbook . (n.d.). Biomass Producer Handbook . POET-DSM. Retrieved July 2013, from http://www.projectliberty.com/handbook/

Bozell, J. J., & Petersen, G. R. (2010). Technology development for the production of

biobased products from biorefinery carbohydrates—the US Department of Energy‘s ―Top 10‖ revisited. Green Chemistry, 12(4), 539. doi:10.1039/b922014c

Brechbill, S. C., Tyner, W. E., & Ileleji, K. E. (2011). The Economics of Biomass Collection and Transportation and Its Supply to Indiana Cellulosic and Electric Utility Facilities. BioEnergy Research, 4(2), 141–152. doi:10.1007/s12155-010-9108-0

Buckmaster, D. R. (2005). A Vortex Forage and Biomass Sample Dryer. Presented at the ASAE Annual International Meeting 2005. Retrieved from http://elibrary.asabe.org

Buxton, D. R., & O'Kiely, P. (2003). Preharvest Plant Factors Affecting Ensiling. In D. R. Buxton, R. E. Muck, & J. H. Harrison (Eds.), Silage Science and Technology (Vol. 42, pp. 199–250). American Society of Agronomy Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc.

Byrt, C. S., Grof, C. P. L., & Furbank, R. T. (2011). C4 Plants as Biofuel Feedstocks: Optimising Biomass Production and Feedstock Quality from a Lignocellulosic PerspectiveFree Access. Journal of Integrative Plant Biology, 53(2), 120–135. doi:10.1111/j.1744-7909.2010.01023.x

Chapple, C., Ladisch, M., & Meilan, R. (2007). Loosening lignin‘s grip on biofuel production. Nature Biotechnology, 25(7), 746–748.

Chen, L., Auh, C.-K., Dowling, P., Bell, J., Chen, F., Hopkins, A., et al. (2003). Improved forage digestibility of tall fescue (Festuca arundinacea) by transgenic down -

regulation of cinnamyl alcohol dehydrogenase. Plant Biotechnology Journal, 1(6), 437–449. doi:10.1046/j.1467-7652.2003.00040.x

Chen, S.-F., Mowery, R. A., Scarlata, C. J., & Chambliss, C. K. (2007). Compositional Analysis of Water-Soluble Materials in Corn Stover. Journal of Agricultural and Food Chemistry, 55(15), 5912–5918. doi:10.1021/jf0700327

Chundawat, S. P. S., Venkatesh, B., & Dale, B. E. (2006). Effect of particle size based separation of milled corn stover on AFEX pretreatment and enzymatic digestibility. Biotechnology and Bioengineering, 96(2), 219–231. doi:10.1002/bit.21132

Coblentz, W. K. (2009). Effects of Wrapping Method and Soil Contact on Hay Stored in

Large Round Bales in Central Wisconsin. Applied Engineering in Agriculture, 25(6), 835–850.

113

11

3

Coblentz, W. K., & Hoffman, P. C. (2009a). Effects of bale moisture and bale diameter on spontaneous heating, dry matter recovery, in vitro true digestibility, and in situ

disappearance kinetics of alfalfa-orchardgrass hays. Journal of Dairy Science, 92(6), 2853–2874. doi:10.3168/jds.2008-1920

Coblentz, W. K., & Hoffman, P. C. (2009b). Effects of spontaneous heating on fiber composition, fiber digestibility, and in situ disappearance kinetics of neutral detergent

fiber for alfalfa-orchardgrass hays. Journal of Dairy Science, 92(6), 2875–2895. doi:10.3168/jds.2008-1921

Coblentz, W. K., Fritz, J. O., & Bolsen, K. K. (1993). Baling System for Making Laboratory-Scale Hay Bales. Agronomy Journal, 85(4), 962–965.

Coblentz, W. K., Fritz, J. O., Bolsen, K. K., & Cochran, R. C. (1996). Quality Changes in

Alfalfa Hay During Storage in Bales. Journal of Dairy Science, 79(5), 873–885. doi:10.3168/jds.S0022-0302(96)76436-6

Coblentz, W. K., Fritz, J. O., Bolsen, K. K., Cochran, R. C., & Fu, L.-Q. (1997). Relating Sugar Fluxes during Bale Storage to Quality Changes in Alfalfa Hay. Agronomy Journal, 89, 800–807.

Coblentz, W. K., Fritz, J. O., Bolsen, K. K., King, C. W., & Cochran, R. C. (1998). The

effects of moisture concentration and type on quality characteristics of alfalfa hay baled under two density regimes in a model system. Animal Feed Science and Technology, 72(1–2), 53–69. doi:10.1016/S0377-8401(97)00179-X

Coblentz, W. K., Turner, J. E., Scarbrough, D. A., Lesmeister, K. E., Johnson, Z. B.,

Kellogg, D. W., et al. (2000). Storage Characteristics and Nutritive Value Changes in Bermudagrass Hay as Affected by Moisture Content and Density of Rectangular Bales. Crop Science, 40(5), 1375–1383. doi:10.2135/cropsci2000.4051375x

Collins, M., Swetnam, L. D., Turner, G. M., Hancock, J. N., & Shearer, S. A. (1995).

Storage Method Effects on Dry Matter and Quality Losses of Tall Fescue Round Bales. JOURNAL OF PRODUCTION AGRICULTURE, 8(4), 507–514.

Cundiff, J. S., & Marsh, L. S. (1996). Harvest and Storage Costs for Bales of Switchgrass in the Southeastern United States. Bioresource Technology, 56(1), 95–101. doi:http://dx.doi.org/10.1016/0960-8524(95)00166-2

Dien, B., Jung, H.-J. G., Vogel, K. P., Casler, M. D., Lamb, J., Iten, L., et al. (2006). Chemical composition and response to dilute-acid pretreatment and enzymatic saccharification of alfalfa, reed canarygrass, and switchgrass. Biomass and Bioenergy, 30(10), 880–891. doi:10.1016/j.biombioe.2006.02.004

Eiland, B. R., Clayton, J. E., & Bryan, W. L. (1983). Losses of Fermentable Sugars in Sweet Sorghum During Storage. Transactions of the ASAE, 26(5), 1596–1600.

114

11

4

Elander, R. T., Dale, B. E., Holtzapple, M., Ladisch, M. R., Lee, Y. Y., Mitchinson, C., et al. (2009). Summary of findings from the Biomass Refining Consortium for Applied

Fundamentals and Innovation (CAFI): corn stover pretreatment. Cellulose, 16(4), 649–659. doi:10.1007/s10570-009-9308-y

Emery, I. R. (2013). Direct and Indirect Greenhouse Emissions from Biomass Storage: Implications for Life-Cycle Assessment of Biofuels. (N. S. Mosier, Ed.). Purdue University, West Lafayette, IN.

Emery, I. R., & Mosier, N. S. (2012). The impact of dry matter loss during herbaceous biomass storage on net greenhouse gas emissions from biofuels production. Biomass and Bioenergy, 39(c), 237–246. doi:10.1016/j.biombioe.2012.01.004

Erickson, J. E., Helsel, Z. R., Woodard, K. R., Vendramini, J. M. B., Wang, Y.,

Sollenberger, L. E., & Gilbert, R. A. (2011). Planting Date Affects Biomass and Brix of Sweet Sorghum Grown for Biofuel across Florida. Agronomy Journal, 103(6), 1827. doi:10.2134/agronj2011.0176

Ericsson, K., Rosenqvist, H., & Nilsson, L. J. (2009). Energy crop production costs in the EU. Biomass and Bioenergy, 33(11), 1577–1586. doi:10.1016/j.biombioe.2009.08.002

Evanylo, G. K., Abaye, A. O., Dundas, C., Zipper, C. E., Lemus, R., Sukkariyah, B., &

Rockett, J. (2005). Herbaceous Vegetation Productivity, Persistence, and Metals Uptake on a Biosolids-Amended Mine Soil. Journal of Environment Quality, 34(5), 1811. doi:10.2134/jeq2004.0329

Fahey, G. C., Bourquin, E. C., & Titgemeyer, E. C. (1994). Postharvest Treatment of

Fibrous Feedstuffs to Improve Their Nutritive Value. In Forage Quality, Evaluation, and Utilization (pp. 715–766). American Society of Agronomy, Madison, WI.

Fargione, J., Hill, J., Tilman, D., Polasky, S., & Hawthorne, P. (2008). Land Clearing and the Biofuel Carbon Debt. Science, 319(5867), 1235–1238. doi:10.1126/science.1152747

Foust, T. D., Wooley, R., Sheehan, J., Wallace, R., Ibsen, K., Dayton, D., et al. (2007). A

National Laboratory Market and Technology Assessment of the 30x30 Scenario (No. NREL/TP-510-40942) (pp. 1–261). National Renewable Energy Laboratory. Retrieved from http://www.nrel.gov/docs/fy07osti/40942.pdf

Garlock, R. J., Balan, V., Dale, B. E., Pallapolu, V. R., Lee, Y. Y., Kim, Y., et al. (2011).

Comparative material balances around pretreatment technologies for the conversion of switchgrass to soluble sugars. Bioresource Technology, 102(24), 11063–11071. doi:10.1016/j.biortech.2011.04.002

Graham, R. L., Nelson, R., Sheehan, J., Perlack, R. D., & Wright, L. L. (2007). Current

and Potential U.S. Corn Stover Supplies. Agronomy Journal, 99(1), 1. doi:10.2134/agronj2005.0222

115

11

5

Greenspan. (1977). Humidity fixed points of binary saturated aqueous solutions. Journal of Research of the National Bureau of Standards A. Physics and Chemistry , 81A(1), 89–96.

Guo, D., Chen, F., Wheeler, J., Winder, J., Selman, S., Peterson, M., & Dixon, R. A. (2001). Improvement of in-rumen digestibility of alfalfa forage by genetic manipulation of lignin O-methyltransferases. Transgenic Research, 10(5), 457–464. doi:10.1023/A:1012278106147

Hammerbeck, A. L., Stetson, S. J., Osborne, S. L., Schumacher, T. E., & Pikul, J. L. (2012). Corn Residue Removal Impact on Soil Aggregates in a No-Till Corn/Soybean Rotation. Soil Science Society of America Journal, 76(4), 1390. doi:10.2136/sssaj2011.0421

Han, K. J., Alison, M. W., Pitman, W. D., Day, D. F., Kim, M., & Madsen, L. (2012). Planting Date and Harvest Maturity Impact on Biofuel Feedstock Productivity and Quality of Sweet Sorghum Grown under Temperate Louisiana Conditions. Agronomy Journal, 104(6), 1618. doi:10.2134/agronj2012.0213

Harun, S., Balan, V., Takriff, M. S., Hassan, O., Jahim, J., & Dale, B. (2013). Performance of AFEX

TM pretreated rice straw as source of fermentable sugars: the

influence of particle size. Biotechnology for Biofuels, 6(1), 1–17. Retrieved from http://www.biotechnologyforbiofuels.com/content/6/1/40

He, Xiao, Lau, A. K., Sokhansanj, S., Lim, C. J., Bi, X. T., & Melin, S. (2012). Dry matter losses in combination with gaseous emissions during the storage of forest residues. Fuel, 95(C), 662–664. doi:10.1016/j.fuel.2011.12.027

He, Xu, Hall, M. B., Gallo-Meagher, M., & Smith, R. L. (2003). Improvement of Forage

Quality by Downregulation of Maize O-Methyltransferase. Crop Sci., 43(6), 2240–2251. doi:10.2135/cropsci2003.2240

Henk, L. L., & Linden, J. C. (1994). Silage Processing of Forage Biomass to Alcohol Fuel. In M. E. Himmel, J. O. Baker, & R. P. Overend (Eds.), ACS Symposium Series (Vol. 566, pp. 391–410). Enzymatic Conversion Of Biomass For Fuels Production.

Hess, J. R., Wright, C. T., & Kenney, K. L. (2007). Cellulosic biomass feedstocks and

logistics for ethanol production. Biofuels, Bioproducts and Biorefining , 1(3), 181–190. doi:10.1002/bbb.26

Hu, Z., Sykes, R., Davis, M. F., Brummer, E. C., & Ragauskas, A. J. (2010). Bioresource Technology. Bioresource Technology, 101(9), 3253–3257. doi:10.1016/j.biortech.2009.12.033

116

11

6

Huang, L., Hwang, C.-A., & Phillips, J. (2011). Evaluating the Effect of Temperature on Microbial Growth Rate-The Ratkowsky and a Bělehrádek-Type Models. Journal of Food Science, 76(8), M547–M557. doi:10.1111/j.1750-3841.2011.02345.x

Humbird, D., Davis, R., Tao, L., Kinchin, C., Hsu, D., Aden, A., et al. (2011). Process Design and Economics for Biochemical Conversion of Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and Enzymatic Hydrolysis of Corn Stover (No.

NREL/TP-5100-47764) (pp. 1–147). Golden, CO: National Renewable Energy Laboratory. Retrieved from http://www.nrel.gov/docs/fy11osti/47764.pdf

Igathinathane, C., Pordesimo, L. O., Womac, A. R., & Sokhansanj, S. (2009). Hygroscopic Moisture Sorption Kinetics modeling of Corn stover and its Fractions. Applied Engineering in Agriculture, 25(1), 65–73.

Igathinathane, C., Womac, A. R., Sokhansanj, S., & Pordesimo, L. O. (2005). Sorption equilibrium moisture characteristics of selected corn stover components. Transactions of the ASABE, 48(4), 1449–1460.

Igathinathane, C., Womac, A. R., Sokhansanj, S., & Pordesimo, L. O. (2007). Moisture

sorption thermodynamic properties of corn stover fractions. Transactions of the ASABE, 50(6), 2151–2160.

Igathinathane, C., Womac, A., Pordesimo, L., & Sokhansanj, S. (2008). Mold appearance and modeling on selected corn stover components during moisture sorption. Bioresource Technology, 99(14), 6365–6371. doi:10.1016/j.biortech.2007.11.075

Inman, D., Nagle, N., Jacobson, J., Searcy, E., & Ray, A. E. (2010). Feedstock handling

and processing effects on biochemical conversion to biofuels. Biofuels, Bioproducts and Biorefining, 4(5), 562–573. doi:10.1002/bbb.241

Ishizawa, C. I., Jeoh, T., Adney, W. S., Himmel, M. E., Johnson, D. K., & Davis, M. F. (2009). Can delignification decrease cellulose digestibility in acid pretreated corn stover? Cellulose, 16(4), 677–686. doi:10.1007/s10570-009-9313-1

Jae, J., Tompsett, G. A., Lin, Y.-C., Carlson, T. R., Shen, J., Zhang, T., et al. (2010). Depolymerization of lignocellulosic biomass to fuel precursors: maximizing carbon efficiency by combining hydrolysis with pyrolysis. Energy & Environmental Science, 3(3), 358. doi:10.1039/b924621p

Jeoh, T., Ishizawa, C. I., Davis, M. F., Himmel, M. E., Adney, W. S., & Johnson, D. K. (2007). Cellulase digestibility of pretreated biomass is limited by cellulose accessibility. Biotechnology and Bioengineering, 98(1), 112–122. doi:10.1002/bit.21408

Jirjis, R. (1995). Storage and drying of wood fuel. Biomass and Bioenergy, 9(1 - 5), 181–190. doi:10.1016/0961-9534(95)00090-9

117

11

7

Judd, J. D., Sarin, S. C., & Cundiff, J. S. (2012). Design, modeling, and analysis of a feedstock logistics system. Bioresource Technology, 103(1), 209–218. doi:10.1016/j.biortech.2011.09.111

Jung, H. G. (1989). Forage Lignins and Their Effects on Fiber Digestibility. Agronomy Journal, 81, 33–38.

Jung, H. G., & Deetz, J. A. (1994). Cell Wall Lignification and Degradability. In Forage Quality, Evaluation and Utilization (pp. 315–346). American Society of Agronomy, Madison, WI.

Khanchi, A., Jones, C. L., & Sharma, B. (2009). Characteristics and compositional

variation in round and square sorghum bales under different storage conditions. Presented at the 2009 ASABE Annual International Meeting, Reno, Nevada.

Khullar, E., Dien, B. S., Rausch, K. D., Tumbleson, M. E., & Singh, V. (2013). Effect of particle size on enzymatic hydrolysis of pretreated Miscanthus. Industrial Crops And Products, 44, 11–17. doi:10.1016/j.indcrop.2012.10.015

Kim, S., & Dale, B. E. (2004). Global potential bioethanol production from wasted crops and crop residues. Biomass and Bioenergy, 26(4), 361–375. doi:10.1016/j.biombioe.2003.08.002

Kim, Y., Mosier, N. S., Ladisch, M. R., Pallapolu, V. R., Lee, Y. Y., Garlock, R., et al.

(2011a). Comparative Study on Enzymatic Digestibility of Switchgrass Varieties and Harvests Processed by Leading Pretreatment Technologies. Bioresource Technology, 102(24), 11089–11096. doi:10.1016/j.biortech.2011.06.054

Kim, Y., Ximenes, E., Mosier, N. S., & Ladisch, M. R. (2011b). Soluble

inhibitors/deactivators of cellulase enzymes from lignocellulosic biomass. Enzyme and Microbial Technology, 48(4-5), 408–415. doi:10.1016/j.enzmictec.2011.01.007

Klinke, H. B., Thomsen, A. B., & Ahring, B. K. (2004). Inhibition of ethanol-producing yeast and bacteria by degradation products produced during pre-treatment of biomass. Applied Microbiology and Biotechnology, 66(1), 10–26. doi:10.1007/s00253-004-1642-2

Kochsiek, A. E., & Knops, J. M. H. (2011). Maize cellulosic biofuels: soil carbon loss can be a hidden cost of residue removal. GCB Bioenergy, 4(2), 229–233. doi:10.1111/j.1757-1707.2011.01123.x

Kohlman, K. L., Ladisch, M. R., Weil, J. R., Westgate, P. L., & Yang, Y. (1998,

December). Processes for treating cellulosic material. (Office of Technology Transfer, Purdue Research Foundation, Ed.). United States Patent and Trademark Office.

Kumar, A., Cameron, J. B., & Flynn, P. C. (2005). Pipeline Transport and Simultaneous Saccharification of Corn Stover. Bioresource Technology, 96(7), 819–829. doi:10.1016/j.biortech.2004.07.007

118

11

8

Kumar, A., Sokhansanj, S., & Flynn, P. (2006). Development of a Multicriteria Assessment Model for Ranking Biomass Feedstock Collection and Transportation

Systems. Applied Biochemistry and Biotechnology, 129, 71–87. doi:10.1385/ABAB:129:1:71

Kumar, L., Arantes, V., Chandra, R., & Saddler, J. (2012). The lignin present in steam pretreated softwood binds enzymes and limits cellulose accessibility. Bioresource Technology, 103(1), 201–208. doi:10.1016/j.biortech.2011.09.091

Kwon, S.-I., & Anderson, A. J. (2002). Genes for multicopper proteins and laccase activity: common features in plant-associated Fusariumisolates. Canadian Journal of Botany, 80(5), 563–570. doi:10.1139/b02-035

Larsson, S. H., Lestander, T. A., Crompton, D., Melin, S., & Sokhansanj, S. (2012).

Temperature patterns in large scale wood pellet silo storage. Applied Energy, 92(C), 322–327. doi:10.1016/j.apenergy.2011.11.012

Li, Z., Zhai, H., Zhang, Y., & Yu, L. (2012). Cell morphology and chemical characteristics of corn stover fractions. Industrial Crops And Products, 37(1), 130–136. doi:10.1016/j.indcrop.2011.11.025

Lingle, S. E., Tew, T., Rukavina, H., & Boykin, D. (2013). Post-harvest Changes in

Sweet Sorghum II: pH, Acidity, Protein, Starch, and Mannitol. BioEnergy Research, 6(1), 178–187. doi:10.1007/s12155-012-9248-5

Lionetti, V., Francocci, F., Ferrari, S., Volpi, C., Bellincampi, D., Galletti, R., et al. (2010). From the Cover: Engineering the cell wall by reducing de-methyl-esterified

homogalacturonan improves saccharification of plant tissues for bioconversion. Proceedings of the National Academy of Sciences, 107(2), 616–621. doi:10.1073/pnas.0907549107

Liu, L., Ye, X. P., Womac, A. R., & Sokhansanj, S. (2010). Variability of biomass

chemical composition and rapid analysis using FT-NIR techniques. Carbohydrate Polymers, 81(4), 820–829. doi:10.1016/j.carbpol.2010.03.058

Liu, Z.-H., Qin, L., Pang, F., Jin, M.-J., Li, B.-Z., Kang, Y., et al. (2013). Effects of biomass particle size on steam explosion pretreatment performance for improving the

enzyme digestibility of corn stover. Industrial Crops And Products, 44, 176–184. doi:10.1016/j.indcrop.2012.11.009

Lorenz, A. J., Anex, R. P., Isci, A., Coors, J. G., de Leon, N., & Weimer, P. J. (2009). Forage quality and composition measurements as predictors of ethanol yield from maize (Zea mays L.) stover. Biotechnology for Biofuels, 2(1), 5. doi:10.1186/1754-6834-2-5

119

11

9

Lynch, J. P., O'Kiely, P., & Doyle, E. M. (2012). Yield, quality and ensilage characteristics of whole-crop maize and of the cob and stover components: harvest date

and hybrid effects. Grass and Forage Science, 67(4), 472–487. doi:10.1111/j.1365-2494.2012.00868.x

Lynd, L. R., Weimer, P. J., van Zyl, W. H., & Pretorius, I. S. (2002). Microbial Cellulose Utilization: Fundamentals and Biotechnology. Microbiology And Molecular Biology Reviews, 66(4), 506–577. doi:10.1128/MMBR.66.3.506–577.2002

Lynd, L. R., Wyman, C. E., & Gerngross, T. U. (1999). Biocommodity Engineering. Biotechnology Progress, 15(5), 777–793. doi:10.1021/bp990109e

Ma, S., & Eckhoff, S. R. (2012). Economy of Scale for Biomass Refineries: Commodity vs. Contract Pricing, Area Utilization Factor, and Energy Crops. Transactions of the ASABE, 55(2), 599–607.

Mani, S., Sokhansanj, S., Bi, X., & Turhollow, A. (2006). Economics of producing fuel pellets from biomass. Applied Engineering in Agriculture, 22(3), 421–426.

Martinson, K., Coblentz, W., & Sheaffer, C. (2011). The Effect of Harvest Moisture and Bale Wrapping on Forage Quality, Temperature, and Mold in Orchardgrass Hay. Journal of Equine Veterinary Science, 31(12), 711–716. doi:10.1016/j.jevs.2011.05.003

McDonald, P. (1981). Clostridia. In The Biochemistry of Silage (pp. 77–90). John Wiley & Sons, Ltd.

McLaughlin, S. B., & Adams Kszos, L. (2005). Development of switchgrass (Panicum virgatum) as a bioenergy feedstock in the United States. Biomass and Bioenergy, 28(6), 515–535. doi:10.1016/j.biombioe.2004.05.006

Mendu, V., Harman-Ware, A. E., Crocker, M., Jae, J., Stork, J., Morton, S., III, et al.

(2011). Identification and thermochemical analysis of high-lignin feedstocks for biofuel and biochemical production. Biotechnology for Biofuels, 4(1), 43. doi:10.1186/1754-6834-4-43

Mendu, V., Shearin, T., Campbell, J. H., Stork, J., Jae, J., Crocker, M., et al. (2012).

Global bioenergy potential from high-lignin agricultural residue. (C. R. Somerville, Ed.)Proceedings of the National Academy of Sciences, 109(10), 4014–4019. doi:10.1073/pnas.1112757109/-/DCSupplemental/pnas.201112757SI.pdf

Miller, A. N., & Ottman, M. J. (2010). Irrigation Frequency Effects on Growth and

Ethanol Yield in Sweet Sorghum. Agronomy Journal, 102(1), 60. doi:10.2134/agronj2009.0191

Mitchell, R., Vogel, K. P., & Sarath, G. (2008). Managing and enhancing switchgrass as a bioenergy feedstock. Biofuels, Bioproducts and Biorefining , 2(6), 530–539. doi:10.1002/bbb.106

120

12

0

Mobashar, M., Blank, R., Hummel, J., Westphal, A., Tholen, E., & Südekum, K. H. (2012). Ruminal ochratoxin A degradation—Contribution of the different microbial

populations and influence of diet. Animal Feed Science and Technology, 171(2-4), 85–97. doi:10.1016/j.anifeedsci.2011.10.002

Monti, A., Fazio, S., & Venturi, G. (2009). The discrepancy between plot and field yields: Harvest and storage losses of switchgrass. Biomass and Bioenergy, 33(5), 841–847. doi:10.1016/j.biombioe.2009.01.006

Mooney, D. F., Larson, J. A., English, B. C., & Tyler, D. D. (2012). Effect of dry matter loss on profitability of outdoor storage of switchgrass. Biomass and Bioenergy, 44(C), 33–41. doi:10.1016/j.biombioe.2012.04.008

Moore, K. J., & Hatfield, R. D. (1994). Carbohydrates and Forage Quality. In Forage

Quality, Evaluation and Utilization (pp. 229–280). American Society of Agronomy, Madison, WI.

Morey, R. V., Kaliyan, N., Tiffany, D. G., & Schmidt, D. R. (2010). A Corn Stover Supply Logistics System. Applied Engineering in Agriculture, 26(3), 455–461.

Mosier, N. S., Hendrickson, R., Brewer, M., Ho, N., Sedlak, M., Dreshel, R., et al. (2005a). Industrial scale-up of pH-controlled liquid hot water pretreatment of corn fiber

for fuel ethanol production. Applied Biochemistry and Biotechnology, 125(2), 77–97. doi:10.1385/ABAB:125:2:077

Mosier, N., Hendrickson, R., Ho, N., Sedlak, M., & Ladisch, M. R. (2005b). Optimization of pH Controlled Liquid Hot Water Pretreatment of Corn Stover. Bioresource Technology, 96(18), 1986–1993. doi:10.1016/j.biortech.2005.01.013

Mosier, N., Wyman, C. E., Dale, B., Elander, R. T., Lee, Y. Y., Holtzapple, M. T., & Ladisch, M. (2005c). Features of promising technologies for pretreatment of lignocellulosic biomass. Bioresource Technology, 96(6), 673–686. doi:10.1016/j.biortech.2004.06.025

Mostrom, M. S., & Jacobsen, B. J. (2011). Ruminant Mycotoxicosis. Veterinary Clinics of North America: Food Animal Practice, 27(2), 315–344. doi:10.1016/j.cvfa.2011.02.007

Muck, R. E., & Kung, L. J. (2007). Silage Production. In R. F. Barnes, C. J. Nelson, K. J.

Moore, & M. Collins (Eds.), Forages Vol. 2: The Science of Grassland Agriculture (6 ed., pp. 617–633). Blackwell Publishing.

121

12

1

Muck, R. E., Moser, L. E., & Pitt, R. E. (2003). Postharvest Factors Affecting Ensiling. In D. R. Buxton, R. E. Muck, & J. H. Harrison (Eds.), Silage Science and Technology (pp.

251–304). American Society of Agronomy Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc.

Nurmi, J. (1999). The storage of logging res idue for fuel. Biomass and Bioenergy, 17(1), 41–47.

Obruca, S., Marova, I., Matouskova, P., Haronikova, A., & Lichnova, A. (2012).

Production of lignocellulose-degrading enzymes employing Fusarium solani F-552. Folia Microbiologica, 57(3), 221–227. doi:10.1007/s12223-012-0098-5

Palmqvist, E., & Hähn-Hägerdal, B. (2000). Fermentation of Lignocellulosic Hydrolysates. II: Inhibitors and Mechanisms of Inhibition. Bioresource Technology, 74(1), 25–33. doi:10.1016/S0960-8524(99)00161-3

Pedersen, M., & Meyer, A. S. (2009). Influence of Substrate Particle Size and Wet

Oxidation on Physical Surface Structures and Enzymatic Hydrolysis of Wheat Straw. Biotechnology Progress, 25(2), 399–408. doi:10.1002/btpr.141

Pereyra, M. L. G., Alonso, V. A., Sager, R., Morlaco, M. B., Magnoli, C. E., Astoreca, A. L., et al. (2008). Fungi and selected mycotoxins from pre- and postfermented corn silage.

Journal of Applied Microbiology, 104(4), 1034–1041. doi:10.1111/j.1365-2672.2007.03634.x

Perlack, R. D., Wright, L. L., Turhollow, A. F., Graham, R. L., Stokes, B. J., & Erbach, D. C. (2005). Biomass as Feedstock for a Bioenergy and Bioproducts Industry: The

Technical Feasibility of a Billion-Ton Annual Supply (pp. 1–78). Oak Ridge National Laboratory.

Perrin, R., Vogel, K., Schmer, M., & Mitchell, R. (2008). Farm-Scale Production Cost of Switchgrass for Biomass. BioEnergy Research, 1(1), 91–97. doi:10.1007/s12155-008-9005-y

Peters, D. (2007). Raw Materials. In R. Ulber & D. Sell (Eds.), Advances in Biochemical Engineering/Biotechnology (Vol. 105, pp. 1–30). Berlin, Heidelberg: Springer Berlin Heidelberg. doi:10.1007/10_031

Petrolia, D. R. (2008). The economics of harvesting and transporting corn stover for

conversion to fuel ethanol: A case study for Minnesota. Biomass and Bioenergy, 32(7), 603–612. doi:10.1016/j.biombioe.2007.12.012

Pitt, R. E. (1990). Silage and Hay Preservation. (J. Kazmierczak, Ed.). Northeast Regional Agricultural Engineering Service.

122

12

2

Popp, A., Krause, M., Dietrich, J. P., Lotze-Campen, H., Leimbach, M., Beringer, T., & Bauer, N. (2012). Additional CO2 Emissions From Land Use Change - Forest

Conservation as a Precondition for Sustainable Production of Second Generation Bioenergy. Ecological Economics, 74, 64–70. doi:10.1016/j.ecolecon.2011.11.004

Probst, K., Kingsley, A., Pinto, R., Bali, R., Kumar, P., & Ileleji, K. E. (2013). The Effect of Moisture Content on the Grinding Performance of Corn and Corncobs by Hammer Milling. Transactions of the ASABE.

Rasmussen, R. R., Rasmussen, P. H., Larsen, T. O., Bladt, T. T., & Binderup, M. L. (2011). In vitro cytotoxicity of fungi spoiling maize silage. Food and Chemical Toxicology, 49(1), 31–44. doi:10.1016/j.fct.2010.09.007

Richard, T. L. (2010). Challenges in Scaling Up Biofuels Infrastructure. Science, 329(5993), 793–796. doi:10.1126/science.1189139

Rider, A. R., Barr, S. D., & Pauli, A. W. (1993). Hay and Forage Harvesting (4 ed.). Moline, IL: Deere and Company Service Publications.

Rigdon, A. R., Jumpponen, A., Vadlani, P. V., & Maier, D. E. (2013). Impact of various storage conditions on enzymatic activity, biomass components and conversion to ethanol

yields from sorghum biomass used as a bioenergy crop. Bioresource Technology, 132(C), 269–275. doi:10.1016/j.biortech.2013.01.055

Rio, L. F., Chandra, R. P., & Saddler, J. N. (2010). The Effect of Varying Organosolv Pretreatment Chemicals on the Physicochemical Properties and Cellulolytic Hydrolysis

of Mountain Pine Beetle-Killed Lodgepole Pine. Applied Biochemistry and Biotechnology, 161(1-8), 1–21. doi:10.1007/s12010-009-8786-6

Rollin, J. A., Zhu, Z., Sathitsuksanoh, N., & Zhang, Y. H. P. (2011). Increasing cellulose accessibility is more important than removing lignin: A comparison of cellulose solvent -

based lignocellulose fractionation and soaking in aqueous ammonia. Biotechnology and Bioengineering, 108(1), 22–30. doi:10.1002/bit.22919

Rooke, J. A., & Hatfield, R. D. (2003). Biochemistry of Ensiling. In D. R. Buxton, R. E. Muck, & J. H. Harrison (Eds.), Silage Science and Technology (pp. 95–141). American

Society of Agronomy Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc.

Rooney, W. L., Blumenthal, J., Bean, B., & Mullet, J. E. (2007). Designing sorghum as a dedicated bioenergy feedstock. Biofuels, Bioproducts and Biorefining, 1(2), 147–157. doi:10.1002/bbb.15

Rotz, C. A., & Muck, R. E. (1994). Changes in Forage Quality During Harvest and

Storage. In Forage Quality, Evaluation, and Utilization (pp. 828–868). American Society of Agronomy, Madison, WI.

123

12

3

Rotz, C. A., & Shinners, K. J. (2007). Hay Harvest and Storage. In R. F. Barnes, C. J. Nelson, K. J. Moore, & M. Collins (Eds.), Forages vol. 2: The Science of Grassland Agriculture (6 ed., pp. 601–616). Blackwell Publishing.

Russell, J. R., Yoder, S. J., & Marley, S. J. (1990). The Effects of Bale Density, Type of Binding and Storage Surface on the Chemical Composition, Nutrient Recovery and Digestibility of Large Round Hay Bales. Animal Feed Science and Technology, 29, 131–145.

Saha, B. C. (2003). Hemicellulose bioconversion. Journal of Industrial Microbiology & Biotechnology, 30(5), 279–291. doi:10.1007/s10295-003-0049-x

Sanderson, M. A., Egg, R. P., & Wiselogel, A. E. (1997). Biomass losses during harvest and storage of switchgrass. Biomass and Bioenergy, 12(2), 107–114. doi:10.1016/S0961-9534(96)00068-2

Santos, R. B., Lee, J. M., Jameel, H., Chang, H.-M., & Lucia, L. A. (2012). Effects of

hardwood structural and chemical characteristics on enzymatic hydrolysis for biofuel production. Bioresource Technology, 110(C), 232–238. doi:10.1016/j.biortech.2012.01.085

Scarbrough, D. A., Coblentz, W. K., Humphry, J. B., Coffey, K. P., Daniel, T. C., Sauer,

T. J., et al. (2005). Evaluation of Dry Matter Loss, Nutritive Value, and In Situ Dry Matter Disappearance for Wilting Orchardgrass and Bermudagrass Forages Damaged by Simulated Rainfall. Agronomy Journal, 97, 604–614.

Schmer, M. R., Mitchell, R. B., Vogel, K. P., Schacht, W. H., & Marx, D. B. (2010).

Efficient Methods of Estimating Switchgrass Biomass Supplies. BioEnergy Research, 3(3), 243–250. doi:10.1007/s12155-009-9070-x

Schmidt, J., Sipocz, J., Kaszas, I., Szakacs, G., Gyepes, A., & Tengerdy, R. P. (1997). Preservation of Sugar Content in Ensiled Sweet Sorghum. Bioresource Technology, 60(1), 9–13. doi:10.1016/S0960-8524(97)00003-5

Searchinger, T., Heimlich, R., Houghton, R. A., Dong, F., Elobeid, A., Fabiosa, J., et al. (2008). Use of U.S. Croplands for Biofuels Increases Greenhouse Gases Through Emissions from Land-Use Change. Science, 319(5867), 1238–1240. doi:10.1126/science.1151861

Shah, A., Darr, M. J., Webster, K., & Hoffman, C. (2011). Outdoor Storage Characteristics of Single-Pass Large Square Corn Stover Bales in Iowa. Energies, 4(12), 1687–1695. doi:10.3390/en4101687

124

12

4

Shinners, K. J. (2003). Engineering Principles of Silage Harvesting Equipment. In D. R. Buxton, R. E. Muck, & J. H. Harrison (Eds.), Silage Science and Technology (Vol. 42, pp.

361–404). American Society of Agronomy Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc.

Shinners, K. J., & Binversie, B. N. (2007). Fractional yield and moisture of corn stover biomass produced in the Northern US Corn Belt. Biomass and Bioenergy, 31(8), 576–584. doi:10.1016/j.biombioe.2007.02.002

Shinners, K. J., Adsit, G. S., Binversie, B. N., Digman, M. F., Muck, R. E., & Weimer, P. J. (2007a). Single-pass, split-stream harvest of corn grain and stover. Transactions of the ASABE, 50(2), 355–363.

Shinners, K. J., Bennett, R. G., & Hoffman, D. S. (2012). Single- and Two-Pass Corn Grain and Stover Harvesting. Transactions of the ASABE, 55(2), 341–350.

Shinners, K. J., Binversie, B. N., Muck, R. E., & Weimer, P. J. (2007b). Comparison of

wet and dry corn stover harvest and storage. Biomass and Bioenergy, 31(4), 211–221. doi:10.1016/j.biombioe.2006.04.007

Shinners, K. J., Boettcher, G. C., Hoffman, D. S., Munk, J. T., Muck, R. E., & Weimer, P. J. (2009a). Single-Pass Harvest of Corn Grain and Stover: Performance of Three Harvester Configurations. Transactions of the ASABE, 52(1), 341–350.

Shinners, K. J., Boettcher, G. C., Muck, R. E., Weimer, P. J., & Casler, M. D. (2010). Harvest and Storage of Two Perennial Grasses as Biomass Feedstocks. Transactions of the ASABE, 53(2), 359–370.

Shinners, K. J., Huenink, B. M., Muck, R. E., & Albrecht, K. A. (2009b). Storage

Characteristics of Large Round Alfalfa Bales: Dry Hay. Transactions of the ASABE, 52(2), 409–418.

Shinners, K. J., Huenink, B. M., Muck, R. E., & Albrecht, K. A. (2009c). Storage Characteristics of Large Round and Square Alfalfa Bales: Low‐ Moisture Wrapped Bales. Transactions of the ASABE, 52(2), 401–407.

Shinners, K. J., Straub, R. J., Huhnke, R. L., & Undersander, D. J. (1996). Harvest and Storage Losses Associated with Mid-Size Rectangular Bales. Applied Engineering in Agriculture, 12(2), 167–173.

Shinners, K. J., Wepner, A. D., Muck, R. E., & Weimer, P. J. (2011). Aerobic and

Anaerobic Storage of Single-pass, Chopped Corn Stover. BioEnergy Research, 4(1), 61–75. doi:10.1007/s12155-010-9101-7

125

12

5

Sluiter, A., Hames, B. R., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., & Crocker, D. (2012). Determination of Structural Carbohydrates and Lignin in Biomass: Laboratory

Analytical Procedure (LAP) (No. NREL/TP-510-42618) (pp. 1–18). National Renewable Energy Laboratory. Retrieved from: http://www.nrel.gov/biomass/analytical_procedures.html#lap-002

Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., & Templeton, D. (2008). Determination of

Extractives in Biomass: Laboratory Analytical Procedure (No. NREL/TP-510-42619). National Renewable Energy Laboratory. Retrieved from: http://www.nrel.gov/biomass/analytical_procedures.html#lap-010

Sluiter, J., & Sluiter, A. (2010). Summative Mass Closure: Laboratory Analytical

Procedure (LAP) Review and Integration (No. NREL/TP-510-48087) (pp. 1–13). National Renewable Energy Laboratory. Retrieved from: http://www.nrel.gov/biomass/analytical_procedures.html#smc

Sokhansanj, S., & Hess, J. R. (2009). Biomass Supply Logistics and Infrastructure. In J.

R. Mielenz (Ed.), Biofuels: Methods and Protocols (Vol. 581, pp. 1–25). Springer US. doi:10.1007/978-1-60761-214-8_1

Sokhansanj, S., & Turhollow, A. R. (2002). Baseline cost for corn stover collection. Applied Engineering in Agriculture, 18(5), 525–530.

Sokhansanj, S., Mani, S., Tagore, S., & Turhollow, A. F. (2010). Techno-economic

analysis of using corn stover to supply heat and power to a corn ethanol plant - Part 1: Cost of feedstock supply logistics. Biomass and Bioenergy, 34(1), 75–81. doi:10.1016/j.biombioe.2009.10.001

Sokhansanj, S., Mani, S., Turhollow, A., Kumar, A., Bransby, D., Lynd, L., & Laser, M.

(2009). Large-scale production, harvest and logistics of switchgrass ( Panicum virgatum L.) - current technology and envisioning a mature technology. Biofuels, Bioproducts and Biorefining, 3(2), 124–141. doi:10.1002/bbb.129

Sokhansanj, S., Turhollow, A., Cushman, J., & Cundiff, J. (2002). Engineering aspects of

collecting corn stover for bioenergy. Biomass and Bioenergy, 23(5), 347–355. doi:10.1016/S0961-9534(02)00063-6

Stephen, J. D., Sokhansanj, S., Bi, X., Sowlati, T., Kloeck, T., Townley-Smith, L., & Stumborg, M. A. (2010). The impact of agricultural residue yield range on the delivered

cost to a biorefinery in the Peace River region of Alberta, Canada. Biosystems Engineering, 105(3), 298–305. doi:10.1016/j.biosystemseng.2009.11.008

126

12

6

Studer, M. H., DeMartini, J. D., Davis, M. F., Sykes, R. W., Davison, B., Keller, M., et al. (2011). Lignin content in natural Populus variants affects sugar release. (R. R. Sederoff,

Ed.) Proceedings of the National Academy of Sciences, 108(15), 6300–6305. doi:10.1073/pnas.1009252108/-/DCSupplemental

Suh, K., & Suh, S. (2010). Economic and Environmental Implications of Corn Stover Densification Options for Biofuel in Minnesota. Transactions of the ASABE, 53(4), 1183–1192.

Suh, K., Suh, S., Walseth, B., Bae, J., & Barker, R. (2011). Optimal Corn Stover Logistics for Biofuel Production: a Case in Minnesota. Transactions of the ASABE, 54(1), 229–238.

Sultana, A., & Kumar, A. (2012). Optimal siting and size of bioenergy facilities using

geographic information system. Applied Energy, 94(C), 192–201. doi:10.1016/j.apenergy.2012.01.052

Tan, Z., Liu, S., Bliss, N., & Tieszen, L. L. (2012). Current and potential sustainable corn stover feedstock for biofuel production in the United States. Biomass and Bioenergy, 47(C), 372–386. doi:10.1016/j.biombioe.2012.09.022

Templeton, D. W., Sluiter, A. D., Hayward, T. K., Hames, B. R., & Thomas, S. R. (2009).

Assessing corn stover composition and sources of variability via NIRS. Cellulose, 16(4), 621–639. doi:10.1007/s10570-009-9325-x

Thorsell, S., Epplin, F. M., Huhnke, R. L., & Taliaferro, C. M. (2004). Economics of a coordinated biorefinery feedstock harvest system: lignocellulosic biomass harvest cost. Biomass and Bioenergy, 27(4), 327–337. doi:10.1016/j.biombioe.2004.03.001

Tumuluru, J. S., Wright, C. T., Hess, J. R., & Kenney, K. L. (2011). A review of biomass densification systems to develop uniform feedstock commodities for bioenergy application. Biofuels, Bioproducts and Biorefining, 5(6), 683–707. doi:10.1002/bbb.324

Turner, J. E., Coblentz, W. K., Scarbrough, D. A., Coffey, K. P., Kellogg, D. W., McBeth,

L. J., & Rhein, R. T. (2002). Change in nutritive Value of Bermudagrass Hay during Storage. Agronomy Journal, 94(1), 109–117.

VanWalsum, G. P., Allen, S. G., Spencer, M. J., Laser, M. S., Antal, M. J., & Lynd, L. R. (1996). Conversion of Lignocellulosics Pretreated with Liquid Hot Water to Ethanol. Applied Biochemistry and Biotechnology, 57-58, 157–170.

Varvel, G. E., Vogel, K. P., Mitchell, R. B., Follett, R. F., & Kimble, J. M. (2008). Comparison of corn and switchgrass on marginal soils for bioenergy. Biomass and Bioenergy, 32(1), 18–21. doi:10.1016/j.biombioe.2007.07.003

127

12

7

Vermerris, W. (2008). Composition and Biosynthesis of Lignocellulosic Biomass. In W. Vermerris (Ed.), Genetic Improvement of Bioenergy Crops (pp. 89–142). New York, NY: Springer New York. doi:10.1007/978-0-387-70805-8_4

Vidal, B. C., Dien, B. S., Ting, K. C., & Singh, V. (2011). Influence of Feedstock Particle Size on Lignocellulose Conversion—A Review. Applied Biochemistry and Biotechnology, 164(8), 1405–1421. doi:10.1007/s12010-011-9221-3

Wihersaari, M. (2005). Evaluation of greenhouse gas emission risks from storage of

wood residue. Biomass and Bioenergy, 28(5), 444–453. doi:10.1016/j.biombioe.2004.11.011

Williams, A. G., Hoxey, R. P., & Lowe, J. F. (1997). Changes in temperature and silo gas composition during ensiling, storage and feeding-out grass silage. Grass and Forage Science, 52(2), 176–189. doi:10.1046/j.1365-2494.1997.00066.x

Williams, S. D., & Shinners, K. J. (2012). Farm-scale anaerobic storage and aerobic

stability of high dry matter sorghum as a biomass feedstock. Biomass and Bioenergy, 46(C), 309–316. doi:10.1016/j.biombioe.2012.08.010

Wiselogel, A. E., Agblevor, F. A., Johnson, D. K., Deutch, S., Fennell, J. A., & Sanderson, M. A. (1996). Compositional changes during storage of large round

switchgrass bales. Bioresource Technology, 56(1), 103–109. doi:10.1016/0960-8524(95)00171-9

Woodson, P., Volenec, J. J., & Brouder, S. M. (2013). Field-scale potassium and phosphorus fluxes in the bioenergy crop switchgrass: Theoretical energy yields and

management implications. Journal of Plant Nutrition and Soil Science, 176(3), 387–399. doi:10.1002/jpln.201200294

Wortmann, C. S., Liska, A. J., Ferguson, R. B., Lyon, D. J., Klein, R. N., & Dweikat, I. (2010). Dryland Performance of Sweet Sorghum and Grain Crops for Biofuel in Nebraska. Agronomy Journal, 102(1), 319. doi:10.2134/agronj2009.0271

Wu, Y.-R., Luo, Z.-H., Chow, R. K.-K., & Vrijmoed, L. L. P. (2010). Bioresource Technology. Bioresource Technology, 101(24), 9772–9777. doi:10.1016/j.biortech.2010.07.091

Ximenes, E., Kim, Y., Mosier, N., Dien, B., & Ladisch, M. (2010). Inhibition of

cellulases by phenols. Enzyme and Microbial Technology, 46(3-4), 170–176. doi:10.1016/j.enzmictec.2009.11.001

Ximenes, E., Kim, Y., Mosier, N., Dien, B., & Ladisch, M. (2011). Deactivation of cellulases by phenols. Enzyme and Microbial Technology, 48(1), 54–60. doi:10.1016/j.enzmictec.2010.09.006

128

12

8

Yan, J., Hu, Z., Pu, Y., Brummer, E. C., & Ragauskas, A. J. (2010). Chemical compositions of four switchgrass populations. Biomass and Bioenergy, 34(1), 48–53. doi:10.1016/j.biombioe.2009.09.010

Yang, B., & Wyman, C. E. (2008). Pretreatment: the key to unlocking low-cost cellulosic ethanol. Biofuels, Bioproducts and Biorefining, 2(1), 26–40. doi:10.1002/bbb.49

Yeh, A.-I., Huang, Y.-C., & Chen, S. H. (2010). Effect of particle size on the rate of enzymatic hydrolysis of cellulose. Carbohydrate Polymers, 79(1), 192–199. doi:10.1016/j.carbpol.2009.07.049

Yu, Z., Jameel, H., Chang, H.-M., & Park, S. (2011). The effect of delignification of

forest biomass on enzymatic hydrolysis. Bioresource Technology, 102(19), 9083–9089. doi:10.1016/j.biortech.2011.07.001

Zeng, M., Mosier, N. S., Huang, C.-P., Sherman, D. M., & Ladisch, M. R. (2007). Microscopic examination of changes of plant cell structure in corn stover due to hot water

pretreatment and enzymatic hydrolysis. Biotechnology and Bioengineering, 97(2), 265–278. doi:10.1002/bit.21298

Zhang, M., Song, X., Zhang, P., Pei, Z. J., Deines, T. W., & Wang, D. (2012). Size Reduction of Cellulosic Biomass in Biofuel Manufacturing: A Study on Confounding

Effects of Particle Size and Biomass Crystallinity. Journal of Manufacturing Science and Engineering, 134(1), 011009. doi:10.1115/1.4005433

Zhao, X., Zhang, L., & Liu, D. (2012a). Biomass recalcitrance. Part I: the chemical compositions and physical structures affecting the enzymatic hydrolysis of lignocellulose. Biofuels, Bioproducts and Biorefining, 6(4), 465–482. doi:10.1002/bbb.1331

Zhao, Y. L., Steinberger, Y., Shi, M., Han, L. P., & Xie, G. H. (2012b). Changes in stem composition and harvested produce of sweet sorghum during the period from maturity to a sequence of delayed harvest dates. Biomass and Bioenergy, 39(c), 261–273. doi:10.1016/j.biombioe.2012.01.020

12

9

APPENDICES

129

12

9

Appendix A Water Activity

A major limitation in the existing literature on forage storage, specifically the impact of

moisture content is that the exchange of moisture is unaccounted for. Stored biomass

exchanges water with its surrounding environment, which affects biomass moisture

content and correspondingly microbial activity. The process is likely influenced by

temperature, the moisture content of the surrounding environment (soil, air), aeration and

the mass transfer dynamics caused by biomass packing density. To examine the impact of

moisture on biomass, it is necessary to use a parameter capturing both the physical

concentration of water present (% w/w) and the likelihood of the water escaping the

biomass. Moisture content alone does not indicate the likelihood and extent to which

water affects biomass or enables microbial growth on biomass. Water activity, designated

aw and expressed as a ratio between 0.0 and 1.0, provides a more complete and accurate

assessment of the role of water in biomass. This section defines water activity and

examines how it affects the impact of water on biomass.

A.1. Water Activity: Definition & Significance:

The activity of a chemical species at a given temperature is classically defined as the ratio

of its fugacity f in a given state to its fugacity f° in a selected standard state (Lewis &

Randall 1923) (Equation 1).

(1)

a=fT

fT°

130

13

0

Fugacity being a stand-in for pressure when correcting for deviations from ideality, it has

the same dimensions. Activity is correspondingly calculated as a ratio of pressures,

specifically vapor pressures, in practice. Applying the concept to water, and using pure

water at the same temperature as the standard state, water activity is obtained. The water

activity of a system or mixture at a given temperature becomes the ratio of the vapor

pressure of water pw in the system to that of pure water p°w at the same temperature

(Equation 2) (Reid 2007).

(2)

A.1.1. Thermodynamic Significance:

The Gibbs‘ free energy of a system (G) is the key thermodynamic parameter in

determining its phase and chemical equilibria, or lack thereof. An equivalent parameter

applied to each species in each phase is its chemical potential µi, the partial derivative of

net free energy with respect to the amount (moles) of the ith species present in that

particular phase. The chemical potential of a species, also known as partial molar Gibbs‘

free energy, represents the change in the system‘s free energy through the addition of a

differential amount of the species to a finite amount of solution at constant temperature

and pressure (Equation 3). As a partial property, at a particular temperature and pressure

can be calculated from the free energy of the pure species under the same conditions

(Equation 4). (Ref. Smith et al. 1996)

(3)

aw,T

=pw,T

pw,T0

æ

è

çç

ö

ø

÷÷

mi=G

i=

¶(Gsys)

¶ni

é

ë

êêê

ù

û

úúúP,T ,n

j¹i

131

13

1

(4)

Where,

Gsys is the Gibbs‘ free energy of the system,

Gi,pure is the Gibbs‘ free energy of pure species i under temperature T and pressure P

µi is the chemical potential of species i

ni and xi are the number of moles and mole fraction of species i.

Chemical potential is important in determining phase equilibria in multiphase systems. A

chemical species will be in dynamic equilibrium between two phases when its chemical

potential is the same in each one (Figure A.1.). Owing to its dependence on temperature,

pressure and moles present, chemical potential is often measured using a standard state as

a reference point, the temperature being constant. Under such situations, assuming ideal

gas behavior, the chemical potential of a species is calculated using equations 5-8 (Smith

et al. 1996).

(5)

Where Γ(T) is a temperature dependent function and P is the system pressure

Substituting Gi in equation 2, we get chemical potentials at a given and a

reference/standard state:

(6)

mi =Gi, pure + RT ln(xi )

Gi, pure = G(T )+ RT lnP

mi,T = G(T )+ RT ln(pi)

132

13

2

(7)

(8)

Where,

pi = P.xi is the partial pressure of species i

p°i is the partial pressure of species i in the reference state

T = Temperature (Kelvin)

R = Gas constant

The ratio (pi/po

i) is recognizable as the chemical activity of species i, and thus indicates

the fractional deviation of species ‗i‘ from the chemical potential of the reference state.

Applying the definition to water, and using pure water in equilibrium with water vapor at

a given temperature as the reference state, water activity is obtained.

A.2. Water Activity: Biological Implications

As described above, the chemical potential of a species governs the rate of its movement

from one phase to another, the phase boundaries being the sites of exchange (Figure A.1).

In case of living cells, the dynamic exchange of water in this manner is essential for cell

growth and function (Figure A.2.a.). A sharp drop in the chemical potential of water

outside the cell membrane would drive water outside the cell, causing osmotic stress that

could prove lethal (Figure A.2.b.). Accordingly, water activity is a parameter that affects

microbial viability.

mi,T = G(T )+RT ln(pi )

mi,T

= mi,T + RT ln(pi

pi

)

133

13

3

Water activity has been reported as a parameter affecting microbial growth and multiple

models have been developed for its impact (Gibson & Hocking 1997; Cuppers et al.

1997). Microbial activity has been observed to require a threshold water activity of

approximately 0.6, below which growth does not occur. Table A.1 contains the microbial

species capable of growth at various water activities, while Table A.2 lists the range of

water activities of common foodstuffs (Table contents reproduced from Appendices D

and E, Water Activity in Foods: Fundamentals and Applications by Barbosa-Cánovas et

al.). This knowledge has been important in the preservation of food grains, wherein

predictive models have been developed for the growth of molds and fungi such as

Fusarium (Torres et al., 2003; Mylona & Magan, 2011) and Aspergillus (Torres et al.

2003; Samapundo et al., 2007; Garcia et al., 2011; Mousa et al., 2011; Astoreca et al.,

Figure A.1. Species Equilibrium between Phases

134

13

4

2012;) and corresponding spoilage of stored gain. Moreover, the manipulation of water

activity is an important means towards preserving foodstuffs containing high amounts of

water without drying. The chemical potential of water decreases as it binds to such

solutes as salts, simple and complex carbohydrates, proteins, sugar alcohols etc., reducing

water activity in turn. As a result of the binding, the fraction of water with the necessary

free energy to act as a solvent and osmotic medium for microbial activity is decreased,

irrespective of the mass and concentration of water. Thus, wet foodstuffs can be

successfully preserved over a long period if their water activity is low.

During storage, wet biomass may be expected to undergo microbial decay, the extent of

activity governed by its water activity. The biomass will simultaneously exchange

moisture with its environment i.e. air and soil, which can reduce or exacerbate storage

losses. The exchange of water is once again driven by the difference in water activity,

this time between the biomass and the air or soil. It is therefore important to examine the

parameters governing moisture transfer into or out of the air and soil, and their

relationship to water activity.

135

13

5

Figure A.2. Water movement across Cell Membranes at a.

Equilibrium water activity and b. Low water activity

136

13

6

Table A.1. Water Activity Ranges and Microbial Proliferation

Water activity Microbial species/families

(Cumulative with increasing water activity)

<0.60 None

0.61 – 0.70 M. bisporus, S. rouxii, E. amstelodami

(Xerophilic fungi)

0.71 – 0.80 E. chevalieri, several Aspergillus molds, P.

citrinum, P. martensii, S. bailii

0.81 – 0.90

Several Penicillium molds, Aspergillus

fumigatus, Aspergillus parasiticus, Aspergillus

clavatus, Saccharomyces cerevisiae, Candida

yeasts, Staphylococcus aureus

0.91 – 0.95

B. subtilis, B. cereus, L. monocytogenes,

Clostridium botulinium A & B, E. coli, Vibrio

parahaemolyticus, Vibrio cholerae, Salmonella

family,

0.96 – 1.0 Clostridium botulinum E, Pseduomonas

fluorescens

137

13

7

Table A.2. Water activities of some common foodstuffs

Foodstuff Moisture Content

(% Weight basis)

Water activity

Range Temperature

(° C)

Bread 37 0.939 – 0.96 25

Rice 7 0.491 – 0.591 25

Fresh meats - 0.968 – 0.99+ 20 – 25

Whole milk 88 0.98 – 0.99 22

Condensed milk - 0.833 25

Jams 29 - 32 0.833 – 0.839 21 - 30

Honey 12.6 - 21 0.477 – 0.552 25

A.3. Water Activity and the Environment

While water activity is a thermodynamic ratio that denotes the likelihood of water

exchange between systems, various environmental parameters have been developed to do

the same. Significant to biomass storage are relative humidity, governing moisture

transfer from the air, and soil water potential governing moisture transfer from the soil. It

is therefore necessary to evaluate their thermodynamic basis and possible connection to

water activity.

The percent relative humidity is the factor governing moisture transfer from the air.

Defined at fixed temperature as the ratio of partial pressure of water vapor to the

equilibrium (with liquid phase) vapor pressure of water in the air, relative humidity is

138

13

8

analogous to water activity as an indicator of the chemical potential of water in the air

(Equation 9).

(9)

Given a closed system consisting of an aqueous solution in equilibrium with the enclosed

atmosphere, the partial pressure of water in the atmosphere becomes the equilibrium

vapor pressure of the solution (Fig. A.3.a.). Upon replacing the solution with pure water

at the same temperature, equilibrium would be obtained when the air is saturated (Figure

A.3.b). Thus, the equilibrium vapor pressure becomes the saturation vapor pressure for

the atmosphere at the given temperature. The resulting pressure ratio, the water activity

of the solution, also becomes relative humidity when expressed as a percentage. Thus, at

a fixed temperature, water will be in dynamic equilibrium between a mixture or solution

at a particular water activity, and air at the same relative humidity.

Similar to humidity, the movement of water through soil is governed through a parameter

termed the water potential. While similar to chemical potential in referring to free energy,

water potential does so on a volumetric rather than molar basis, and correspondingly has

the dimensions of pressure. Water potential is affected by temperature, gravity, solute

concentration, mechanical forces such as capillary action and surface tension and

localized pressure. It is of particular significance when examining the transfer of water

from soil into roots and thereon to extremities and in determining the efficiency of

RHT =pvapor ,T

pvapor ,Tsat

139

13

9

irrigation. The similarity of water potential to chemical potential is seen in Equation 10

below (Reproduced from (Rawlins & Campbell 1986):

(10)

Where,

Ψ = Water potential,

pvapor,T = Vapor pressure of water in equilibrium with the soil

psat

vapor,T = Saturated vapor pressure of the air at the same temperature

M = Molar mass/volume correction

Figure A.3. Equilibrium Vapor Pressures for a. Solution and b. Pure Water

Y =RT

Mln(pvapor ,T

pvapor ,Tsat

)

140

14

0

As seen from the above equation, the soil water potential correlates to relative humidity,

and thereon to water activity as well, Gee et al reporting the use of water activity meters

to measure it (1992). Water activity can thus be used to estimate the possibility of

moisture transfer from the soil in a given environment.

A.4. Conclusion: Water Activity and Biomass Storage

Water activity thus plays two important roles as a parameter in biomass storage. Firstly,

the water activity of biomass indicates the likelihood of microbial proliferation and

activity, better than moisture concentration (mass basis) alone. More significantly, it

indicates the likelihood of moisture exchange between the biomass and its surrounding

environment. Water activity therefore makes a more useful parameter when examining

the impact of moisture on biomass during storage. The present study focuses on biomass

and air alone, and fixes the parameters – relative humidity and water activity – so as to

keep water in a state of dynamic equilibrium between the two. This enables a time-

variant study on the effects of moisture presence and thereby microbial activity.

Appendix A. Bibliography:

Astoreca, A., Vaamonde, G., Dalcero, A., Ramos, A. J., & Marin, S. (2012). Modeling

the effect of temperature and water activity of Aspergillus flavus isolates from corn. International Journal of Food Microbiology, 156(1), 60–67. doi:10.1016/j.ijfoodmicro.2012.03.002

Barbosa-Cánovas, G. V., Fontana, A. J., Schmidt, S. J., & Labuza, T. P. (n.d.). Water Activity in Foods: Fundamentals and Applications. IFT Press: Blackwell Publishing.

Cuppers, H. G. A. M., Oomes, S., & Bruls, S. (1997). A Model for the Combined Effects of Temperature and Salt Concentration on Growth Rate of Food Spoilage Molds. Applied

and Environmental Microbiology, 63(10), 3764–3769.

141

14

1

Garcia, D., Ramos, A. J., Sanchis, V., & Marín, S. (2011). Modeling the effect of temperature and water activity in the growth boundaries of Aspergillus ochraceus and

Aspergillus parasiticus. Food Microbiology, 28(3), 406–417. doi:10.1016/j.fm.2010.10.004

Gee, G. W., Campbell, M. D., Campbell, G. S., & Campbell, J. H. (1992). Rapid Measurement of Low Soil Water Potentials Using a Water Activity Meter. Plant and Soil, 56, 1068–1070.

Gibson, A. M., & Hocking, A. D. (1997). Advances in the predictive modeling of fungal growth in food. Trends in Food Science & Technology, 8, 353–358.

Lewis, G. N., & Randall, M. (1923). A Useful Function, Called the Activity, and Its Application to Solutions. In Thermodynamics and The Free Energy of Chemical

Substances (First Edition. pp. 254–277). McGraw-Hill Book Company, Inc.

Mousa, W., Ghazali, F. M., Jinap, S., Ghazali, H. M., & Radu, S. (2011). Modelling the

effect of water activity and temperature on growth rate and aflatoxin production by two isolates of Aspergillus flavus on paddy. Journal of Applied Microbiology, 111(5), 1262–

1274. doi:10.1111/j.1365-2672.2011.05134.x

Mylona, K., & Magan, N. (2011). Fusarium langsethiae: Storage environment influences

dry matter losses and T2 and HT-2 toxin contamination of oats. Journal of Stored Products Research, 47(4), 321–327. doi:10.1016/j.jspr.2011.05.002

Rawlins, S. L., & Campbell, G. S. (1986). Water Potential: Thermocouple Psychrometry. In A. Klute, G. S. Campbell, R. D. Jackson, M. M. M, & D. R. Nielsen (Eds.), Methods of Soil Analysis, Part 1 (2nd ed., pp. 597–618). Madison: American Society of Agronomy

Inc., Soil Science Society of America Inc.

Reid, D. S. (2007). Water Activity: Fundamentals and Relationships. In T. P. Labuza, S. J.

Schmidt, A. J. Fontana Jr, & G. V. Barbosa-Canovas (Eds.), Water Activity in Foods: Fundamentals and Applications (pp. 15–28). Ames, Iowa: IFT Press: Blackwell Publishing.

Samapundo, S., Devlieghere, F., Geeraerd, A., Demeulenaer, B., Vanimpe, J., & Debevere, J. (2007). Modelling of the individual and combined effects of water activity

and temperature on the radial growth of Aspergillus flavus and A. parasiticus on corn. Food Microbiology, 24(5), 517–529. doi:10.1016/j.fm.2006.07.021

Smith, J. M., Van Ness, H. C., & Abbott, M. M. (1996). Solution Thermodynamics: Theory. In Introduction to Chemical Engineering Thermodynamics (Fifth edition. pp.

315–365). McGraw-Hill.

Torres, M. R., Ramos, A. J., Soler, J., Sanchis, V., & Marin, S. (2003). SEM study of

water activity and temperature effects on the initial growth of Aspergillus ochraceus, Alternaria alternata and Fusarium verticillioides on maize grain. International Journal of Food Microbiology, 81, 185–193.

142

14

2

Appendix B Validation of Moisture Content & Glucose Concentration Determination

Methods

The determination of the moisture content of biomass samples, and of the release of

glucose during bioprocessing were critical steps in assessing storage dry matter losses

and biomass processing efficacy respectively. Several methods have been developed for

each, with a wide range of sample sizes, accuracy, reproducibility and processing time

required. Due to time and cost constraints, two methods were used for biomass moisture

determination and glucose concentration in the present study. The results were compared

to validate the methods, and establish if data gathered through two different methods

could be treated as identical.

B.1. Moisture Content Measurement

Forage moisture content is calculated from the observed weight difference in samples

following drying. The drying conditions, specifically temperature, vary with the protocol

utilized. At present, a common standard method is ASABE Standard 358.2 (ASABE,

2012). The method accounts for heterogeneity in farm- and commercial-scale bales by

utilizing large samples (25 grams or higher). Moisture content is calculated from the

weight difference in the sample following drying in an oven at 105 °C for 24 hours.

The ASABE standard method was utilized to estimate switchgrass moisture content at the

time it was wetted for storage. It could not be used with stored switchgrass samples as

each sample consisted of a hundred dry grams, some of which would have been lost

during storage. The method used therefore had to minimize sampling amount. Moisture

143

14

3

content was correspondingly determined using an HB43-S Halogen Moisture Analyzer

(Mettler-Toledo LLC, Columbus, OH), which utilized 1.0-2.0 gram samples for moisture

analysis. The balance calculates moisture content from the change in sample mass upon

heating through an electrical resistance heater at 100-105 °C. Heating is carried out

several minutes, ceasing when the rate change in sample mass drops below a threshold.

As the sample size and heating period are different, the methods needed to be compared

to see if equivalent results would be produced from the same sample material.

To validate the method, samples of switchgrass wetted to different levels were gathered,

and moisture content measured through both methods. At least five measurements were

taken for each moisture level and the data analyzed statistically. The readings, mean

values and standard deviation are shown in Table B.1 below.

144

14

4

Table B.1. Moisture Reading Comparison: ASABE Standard vs. Moisture Balance

Target

Moisture Content

(% )

Measured Moisture Content (% )

ASABE S 358.2 Moisture Balance

Reading Mean

(± St. Dev.) Reading

Mean

(± St. Dev.)

10.6

9.90

10.1 ± 0.12

9.61

9.4 ± 0.22

10.16 9.26

10.20 9.28

10.05 9.26

10.05 9.71

16.8

15.65

15.0 ± 0.37

14.48

14.2 ± 0.16

14.74 14.32

15.06 14.10

14.80 14.16

14.86 14.11

18.9

18.09

18.1 ± 0.05

17.16

17.0 ± 0.17

18.18 16.92

18.08 16.89

18.06 17.18

18.15 16.79

(Continued on the next page)

145

14

5

Table B.1. (Continued)

Target Moisture

Content

(% )

Measured Moisture Content (% )

ASABE S 358.2 Moisture Balance

Reading Mean

(± St. Dev.) Reading

Mean

(± St. Dev.)

33.6

33.41

34.1 ± 0.59

33.73

34.4 ± 3.71 34.56 31.11

34.23 38.44

31.78

32.4 ± 0.66

29.35

30.9 ± 1.36 33.08 31.99

32.22 31.25

32.17

32.4 ± 0.24

31.95

32.9 ± 1.01 32.36 32.66

32.65 33.95

Errors in measurement were carried out for each moisture level, the error being defined

as follows:

A paired t-test comparison was carried out for the errors at each moisture level, using

PROC TTEST on SAS 9.3 (The SAS Institute, Cary, NC). The null hypothesis was that

the error was zero. The P-value of the test was found to be 0.16 (Table B.2), and

correspondingly the null hypothesis could not be rejected at confidence levels greater

than 84%.

146

14

6

Table B.2. T-test Comparison: Difference in Readings

Measurement Error T-statistic P-value

(H0: Error = 0)

Moisture Reading – ASABE Result -1.67 0.1564

To further analyze the data, linear regression was carried out with the mean moisture

content determined by the ASABE method as the dependent variable, using PROC REG

on SAS 9.3. T-tests were carried out on the regression parameters. For the two methods

to be identical, the null hypotheses were that the slope was equal to one and the intercept

was equal to zero. The T-test results are shown in Table B.3.

Table B.3. Regression Statistics: Moisture Content (ASABE) vs. Moisture Reading (Balance)

Parameter Estimate Standard

Error Test T-statistic P-value

Intercept 1.32 0.817 H0: I = 0

Ha: I ≠ 0 1.61 0.182

Slope 0.97 0.033 H0: S = 1

Ha: S ≠ 1 -1.038 0.352

The P-value was found to be 0.18 for the intercept test, and 0.35 for the slope test.

Correspondingly, the null hypotheses could not be rejected at confidence levels greater

than 81.8% for the intercept, and 64.8% for the slope. On this basis, mean moisture

content as measured through the Mettler-Toledo analyzer could safely be considered

equivalent to the value determined through ASABE Standard 358.2.

147

14

7

B.2. Glucose Concentration Measurement

When carrying out cellulose hydrolysis, liquid chromatography was used to establish

concentration data, by comparing sample chromatograms to those containing sugar

standards of known concentrations. The cost in time (1 hour per sample) and reagents

made it necessary to use a fast, cost-effective screening method to determine if glucose

concentrations were low or unchanged from control cases. Correspondingly, the method

described by Fitzgerald et al (Fitzgerald & Vermerris, 2005) utilizing commercial-grade

blood glucose meters was examined for this purpose. This study used a ReliOn® Confirm

blood glucose meter (Walmart, Bentonville, AR), fitted with ReliOn® Confirm/Micro test

strips, which measured sample concentrations within 10 seconds. Glucose concentrations

Figure B.1. Linear Regression: ASABE S358.2-determined moisture content (y) vs. Moisture balance readings (x)

148

14

8

were measured for solutions at known concentrations, prepared in the pH 4.8 citrate

buffer used for hydrolysis. The device readings were found to be consistent within a

range of 0.5 – 5.0 g/l range, outside of which extreme fluctuations were observed. The

readings within this range were compared through a paired t-test with the glucose

concentrations as prepared.

Measurement errors between glucose meter readings and the actual concentrations were

computed and a student t-test performed using PROC TTEST on SAS 9.3 with the null

hypothesis that the error was equal to zero (Table B.4). The test was found to have a P-

value of 0.056. Correspondingly, the null hypothesis could not be rejected for confidence

levels greater than 94.4%.

Table B.4. T-test comparison: Glucose Reading vs. Concentration

Error T-statistic P-value

(H0: Error = 0)

Glucose reading – Concentration 2.16 0.056

As with moisture determination method validation, regression was performed using

PROC REG on SAS (Figure B.2.). As described previously for the moisture

determination method validation, regression parameters were tested against the null

hypotheses that the intercept was equal to zero and the slope equal to one (Table B.5).

149

14

9

Table B.5. Regression Statistics: Glucose concentration vs. Glucose reading

Parameter Estimate Standard

Error Test T-statistic P-value

Intercept -0.04 0.202 H0: I = 0

Ha: I ≠ 0 -0.18 0.858

Slope 0.90 0.069 H0: S = 1

Ha: S ≠ 1 -1.038 0.17

The intercept test was found to have a P-value of 0.85 and the slope test a P-value of 0.17.

Correspondingly, the null hypotheses could not be rejected at confidence levels greater

than 14.2% for the intercept and 83% for the slope. It was therefore safe to assume that

the blood glucose meter provided accurate results, especially to test if samples warrant

more detailed and precise analysis through HPLC.

Figure B.2. Linear Regression: Glucose Concentrations (y) vs. Glucose

Meter Reading (x)

150

15

0

Appendix B. Bibliography:

ASABE. (2012). Moisture Measurement — Forages (No. ANSI/ASAE S358.3) (pp. 1–3). ASABE. Retrieved from https://elibrary.asabe.org/

Fitzgerald, J., & Vermerris, W. (2005). The utility of blood glucose meters in biotechnological applications. Biotechnology and Applied Biochemistry, 41(3), 233–239. doi:10.1042/BA20040133

151

15

1

Appendix C Switchgrass Wetting for Storage under Dynamic Moisture Equilibrium

The switchgrass had to be wetted so as to remain at the same water activity as its

environment. Water activity is correlated to moisture content, water sorption isotherms

having been developed for specific mixtures as a function of water concentration by mass

(Labuza & Altunakar, 2007). Similar isotherms were developed for plant tissues by

Igathinathane et al (2005; 2007; 2009) and Albert et al (1989) for corn stover and alfalfa

respectively. Switchgrass wetting is as described below.

C.1. Switchgrass Water Activity Determination

Dried switchgrass was milled so as to pass through a 0.016-inch (No. 40 Mesh) screen,

and equilibrated with deionized water for 48 hours so as to reach moisture contents

ranging from 8% to 35% (w/w), the content measured with a HB43-S Halogen Moisture

Analyzer (Mettler-Toledo LLC, Columbus, OH). Wetted switchgrass was then loaded in

an Aqualab Dewpoint 4TE Water Activity Meter (Decagon Devices, Pullman WA), and

its water activity measured in triplicate. Milled switchgrass was used, as the whole

material was difficult to load into the machine‘s sampling cups. Water activity curves

were developed for the switchgrass at 20, 25 and 35 °C, water activity being dependent

upon temperature. The curves are shown in Figure C.1. Using the curves, the moisture

content required to reach the desired water activities was determined. Due to device

limitations, it was not possible to measure water activities at temperatures below 15 °C.

The water activity of switchgrass at <10 °C was assumed to be the same as that of

switchgrass at 20 °C. Using the curves developed the moisture contents (% w/w)

152

15

2

necessary for the chosen water activity levels were determined, and are listed in Table

C.1.

Switchgrass (milled, sieved) was mixed with deionized water in 4- or 8-gallon zipper

storage bags, water volumes calculated to bring about the necessary moisture content by

weight. The gallon bags were stored at 4 °C for 48-72 hours, for equilibration to take

place. During this period, bag contents were mixed by hand daily to ensure uniformity.

Depending upon their size, multiple bags were used to contain the switchgrass wetted to a

given level. Following the equilibration period, the contents of these bags were mixed

together, and packaged into the samples used for the experiment. Three test samples of

Figure C.1. Switchgrass Water Activity Isotherms

153

15

3

approximately 50 grams were taken at this point and analyzed for moisture contents

through ASABE protocol S358.2, the results of which are listed in Table C.1.

Table C.1. Pre-Storage Switchgrass Moisture Concentrations

Temperature

(°C)

Water activity

(aw)

Required

% Moisture

(w/w)

Calculated

% Moisture (w/w)

Mean St. Dev

<10

0.65 10.6 10.1 0.601

0.75 13.0 12.78 0.193

0.85 16.8 16.07 1.01

0.99 33.6 34.07 0.593

20

0.65 10.6 10.51 0.247

0.75 13 11.95 0.012

0.85 16.8 15.75 0.238

0.99 33.6 32.36 0.662

35

0.65 13.0 12.11 0.100

0.75 15.7 15.58 0.118

0.85 18.9 19.12 0.497

0.99 33.6 34.07 0.593

154

15

4

Appendix C. Bibliography

Albert, R. A., Huebner, B., & Davis, L. W. (1989). Role of Water Activity in the Spoilage of Alfalfa Hay. Journal of Dairy Science, 72(10), 2573–2581.

doi:10.3168/jds.S0022-0302(89)79398-X

Igathinathane, C., Pordesimo, L. O., Womac, A. R., & Sokhansanj, S. (2009).

Hygroscopic Moisture Sorption Kinetics modeling of Corn stover and its Fractions. Applied Engineering in Agriculture, 25(1), 65–73.

Igathinathane, C., Womac, A. R., Sokhansanj, S., & Pordesimo, L. O. (2005). Sorption equilibrium moisture characteristics of selected corn stover components. Transactions of the ASABE, 48(4), 1449–1460.

Igathinathane, C., Womac, A. R., Sokhansanj, S., & Pordesimo, L. O. (2007). Moisture sorption thermodynamic properties of corn stover fractions. Transactions of the ASABE,

50(6), 2151–2160.

Labuza, T. P., & Altunakar, B. (2007). Water Activity Prediction and Moisture Sorption

Isotherms. In T. P. Labuza, S. J. Schmidt, A. J. Fontana Jr, & G. V. Barbosa-Canovas (Eds.), Water Activity in Foods: Fundamentals and Applications (pp. 109–154). IFT

Press: Blackwell Publishing.

155

15

5

Appendix D Standardization of Liquid Chromatography Measurements

To validate the concentration readings of the High Performance Liquid Chromatography

apparatus, standard solutions were run which contained the compounds to be detected –

sugars, sugar derivatives and carboxylic acids – at known concentrations. The solutions

were analyzed at 4 dilutions, with three injections run for each dilution. This was done to

detect possible errors as the concentrations reached extreme highs and lows. The

averages of the readings detected are listed in Table D.1 (Standard solution for analysis

of LAP data), and Table D.2 (Additional compounds present during analysis of

pretreatment and hydrolysis data).

156

15

6

Table D.1. Injection-to-Injection Variability during HPLC Analysis (Standards corresponding to compounds detected during LAP analysis)

Compound

Concentrations (g/L)

True Value

Reading

Mean Standard

Deviation

Glucose

0.5 0.5 0

1 0.992 0.0015

2 2.012 0.0025

4 3.996 0.0113

Xylose

0.25 0.245 0.0006

0.5 0.501 0.0017

1 1.008 0.0017

2 1.996 0.0021

Arabinose

0.25 0.237 0.0006

0.5 0.509 0.0025

1 1.009 0.0031

2 1.995 0.0026

Acetic Acid

0.125 0.126 0.002

0.25 0.247 0.003

0.5 0.502 0.001

1 1.000 0.0015

157

15

7

Table D.2. Injection-to-Injection Variability during HPLC Analysis (Standards corresponding to additional compounds detected during Pretreatment analysis)

Compound

Concentrations (g/L)

True Value

Reading

Mean Standard

Deviation

Cellobiose

0.5 0.512 0

1 1.014 0.0006

2 1.958 0

4 4.016 0.0087

Furfural

0.25 0.2533 0.0006

0.5 0.4913 0.0015

1 1.007 0.0031

2 1.999 0.0105

Hydroxymethyl

furfural

0.25 0.252 0.0006

0.5 0.494 0.0006

1 1.006 0.0006

2 1.998 0.0055

15

8

VITA

158

15

8

VITA

Arun Athmanathan

Graduate School, Purdue University

Education

B.Technology, Biotechnology, 2006, Indian Institute of Technology, Guwahati

M.S., Engineering, 2008, Purdue University, West Lafayette, Indiana

Ph.D., Agricultural and Biological Engineering, 2013, Purdue University, West Lafayette,

Indiana

Research Interests

- Identification of lignocellulosic sources for conversion into fuels and chemicals

- Bioprocessing

- Generation of advanced biofuels