ABCDEFG

UNIVERSITY OF OULU P .O. B 00 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

PUBLICATIONS EDITOR

Professor Esa Hohtola

University Lecturer Santeri Palviainen

Postdoctoral research fellow Sanna Taskila

Professor Olli Vuolteenaho

University Lecturer Veli-Matti Ulvinen

Director Sinikka Eskelinen

Professor Jari Juga

Professor Olli Vuolteenaho

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-0718-6 (Paperback)ISBN 978-952-62-0719-3 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2014

C 516

Pasi Karinkanta

DRY FINE GRINDING OF NORWAY SPRUCE (PICEA ABIES) WOOD IN IMPACT-BASED FINE GRINDING MILLS

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF TECHNOLOGY

C 516

ACTA

Pasi Karinkanta

C516etukansi.kesken.fm Page 1 Tuesday, December 2, 2014 11:22 AM

A C T A U N I V E R S I T A T I S O U L U E N S I SC Te c h n i c a 5 1 6

PASI KARINKANTA

DRY FINE GRINDING OF NORWAY SPRUCE (PICEA ABIES) WOOD IN IMPACT-BASED FINE GRINDING MILLS

Academic dissertation to be presented with the assent ofthe Doctoral Training Committee of Technology andNatural Sciences of the University of Oulu for publicdefence in the Arina auditorium (TA105), Linnanmaa, on23 January 2015, at 12 noon

UNIVERSITY OF OULU, OULU 2014

Copyright © 2014Acta Univ. Oul. C 516, 2014

Supervised byProfessor Jouko NiinimäkiDoctor Mirja Illikainen

Reviewed byDoctor Junyong ZhuDoctor Michaela Eder

ISBN 978-952-62-0718-6 (Paperback)ISBN 978-952-62-0719-3 (PDF)

ISSN 0355-3213 (Printed)ISSN 1796-2226 (Online)

Cover DesignRaimo Ahonen

JUVENES PRINTTAMPERE 2014

OpponentProfessor Kari Heiskanen

Karinkanta, Pasi, Dry fine grinding of Norway spruce (Picea abies) wood inimpact-based fine grinding mills. University of Oulu Graduate School; University of Oulu, Faculty of TechnologyActa Univ. Oul. C 516, 2014University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland

Abstract

Wood powders are used in numerous applications such as thermoplastics and filters, and a lot ofresearch effort has been put into developing novel ways of utilising them. The mechanicalprocessing of wood powders, especially at particle sizes below 100 µm, has been reported inseveral studies, but they lack information on the effect of fine grinding conditions on the particlemorphology and cellulose crystallinity, both of which are important parameters in the furtherprocessing of wood powders and in their various applications. This makes it very difficult todesign and optimise fine grinding processes with different applications in mind. The aim of thisthesis was to study the dry fine grinding of wood in several impact-based fine grinding mills inorder to find out their effect on the properties of the wood and to study the energy required for themechanical processing of the resulting powders.

The effect of the main operational parameters on the properties of dried Norway spruce woodand the energy consumption was studied using three impact-based fine grinding mills that werecapable of pulverising the wood down to a median particle size of less than 25 µm. It was foundthat the impact events occurring in media mills can be used for the production of very fine woodpowders with lower cellulose crystallinity and rounder shaped particles having more uniformshape distribution than powders pulverised to a similar size range by means of impact events innon-media mills. A practical estimate was obtained for the minimum specific energy consumptionin fine grinding in mills involving grinding media that could be utilised as a target for optimisation.Impact-based media milling under cryogenic conditions can be used to obtain different Norwayspruce wood powders from those produced under ambient grinding conditions, i.e. without thefreezing effect of nitrogen liquid. The energy efficiency of fine grinding can be enhanced bychoosing cryogenic rather than ambient conditions. The moisture content of the wood has greaterinfluence on the size and shape of the particles when milling is accomplished under ambientconditions. Torrefaction can reduce the energy consumption in impact-based media mills formedian particle sizes over 17.4 µm (± 0.2 µm), while the shape and cellulose crystallinity of theparticles are not significantly affected by torrefaction pretreatment as a function of energyconsumption.

Keywords: aspect ratio, cellulose crystallinity, comminution, dry fine grinding, energyconsumption, Norway spruce, particle size, processing, wood flour, wood powder

Karinkanta, Pasi, Metsäkuusen (Picea abies) kuivahienojauhatus iskuihinperustuvilla hienojauhatusmyllyillä. Oulun yliopiston tutkijakoulu; Oulun yliopisto, Teknillinen tiedekuntaActa Univ. Oul. C 516, 2014Oulun yliopisto, PL 8000, 90014 Oulun yliopisto

Tiivistelmä

Puujauheita käytetään laajalti erilaisissa sovelluksissa, kuten esimerkiksi biokomposiiteissa jasuodattimissa. Tämän lisäksi on olemassa paljon tutkimustietoa siitä, kuinka puujauheita voitai-siin hyödyntää laajemminkin. Puu voidaan mekaanisesti prosessoida alle 100 µm:n kokoluok-kaan, mutta yksityiskohtaista tietoa kuivahienojauhatuksen olosuhteiden vaikutuksesta jauhei-den morfologiaan ja selluloosan kiteisyyteen ei ole saatavilla. Puujauheen morfologialla ja sellu-loosan kiteisyydellä on kuitenkin merkittävä vaikutus sovelluksia ja jatkojalostusta ajatellen.Puun kuivahienojauhatuksen tiedon puute hankaloittaa merkittävästi prosessin suunnittelua jaoptimointia erilaisia sovelluksia varten. Tämän väitöskirjan tavoitteena on selvittää iskuihinperustuvien hienojauhimien vaikutukset puun ominaisuuksiin ja tutkia mekaanisen prosessoin-nin energiatehokkuutta hienojauhatuksessa.

Tutkimuksessa selvitettiin kolmen erilaisen iskuun perustuvan hienojauhatusmyllyn pääasial-listen operointiparametrien vaikutusta kuivatun metsäkuusen ominaisuuksiin ja energiankulutuk-seen. Jokaisella hienojauhimella onnistuttiin tuottamaan puujauhoja, joiden mediaanikoko olialle 25 µm. Iskuihin perustuvalla jauhinkappalemyllyllä saatiin tuotettua puujauhoa, jonka sellu-loosan kiteisyys on alhaisempi ja partikkelimuodot pyöreämpiä verrattuna samankokoisiin puu-jauhoihin, jotka on tuotettu iskuihin perustuvilla jauhinkappaleettomilla hienojauhatusmyllyillä.Työssä saatiin käytännöllinen arvio kuivatun metsäkuusen hienojauhatuksen minimienergianku-lutukselle iskuihin perustuville jauhinkappalemyllyille, mitä voidaan käyttää kyseisten mylly-tyyppien optimoinnin tavoitteena. Työssä havaittiin lisäksi, että kryogeenisiä jauhatusolosuhtei-ta käyttämällä voidaan tuottaa erilaisia puujauhoja verrattuna puujauhoihin, jotka prosessoidaanilman nestetyppijäädytystä, kun jauhatus suoritetaan iskuihin perustuvalla jauhinkappalemyllyl-lä. Ilman nestetyppijäädytystä puun kosteuspitoisuudella on merkittävämpi vaikutus puujauhojenominaisuuksiin kuin kryogeenisissä olosuhteissa jauhetuilla. Kryogeenisillä jauhatusolosuhteillavoidaan parantaa myös jauhatuksen energiatehokkuutta. Torrefioinnilla voidaan vähentää hieno-jauhatuksen energiankulutusta iskuihin perustuvilla jauhinkappalemyllyillä, kun tavoitekoonmediaani on yli 17,4 µm (± 0,2 µm). Torrefioinnilla ei ole vaikutusta selluloosan kiteisyyteen taipartikkeleiden muotoon energiankulutuksen funktiona.

Asiasanat: aspektisuhde, energiankulutus, hienonnus, kuivahienojauhatus, metsäkuusi,partikkelikoko, prosessointi, puujauho, selluloosan kiteisyys

Dedicated to my beloved brother, Marko Karinkanta

8

9

Acknowledgements

The work reported in this thesis was carried out at the University of Oulu Fibre

and Particle Engineering Laboratory during the years 2008–2014 in co-operation

with the Graduate School in Chemical Engineering (GSCE) and UPM. Funding

from the KAUTE Foundation, Emil Aaltonen Foundation, Tauno Tönning

Foundation, Jenny and Antti Wihuri Foundation, Finnish Cultural Foundation and

Riitta and Jorma J. Takanen Foundation is greatly appreciated.

My sincere thanks go to Professor Jouko Niinimäki for giving me an

opportunity to work on a new and interesting technology and for supervising my

thesis, and to Dr. Mirja Illikainen, for her efforts and personal guidance. In

addition, thanks are also due to Mrs. Tarja Sinkko, who coordinated the co-

operation with UPM.

I highly appreciate the reviewing work done by Dr. Michaela Eder and Dr.

Junyong Zhu, and I would also like to express my gratitude to Mr. Malcolm Hicks

for revising the English language of this thesis.

I would like to thank my fellow researchers Nina Pykäläinen, Sami Turunen,

Lauri Talikka and Jouko Lehto at the UPM for discussion and conversation

sessions that were very valuable for me, and many thanks also go to my

colleagues at the University of Oulu, especially Katja Ohenoja, Kalle

Kemppainen, Henrikki Liimatainen, Terhi Suopajärvi, Olli Taikina-aho and Antti

Haapala for their assistance in the research and their interesting conversations. I

would also like to thank to my fellow researchers in the Graduate School in

Chemical Engineering (GSCE), especially Riina Salmimies, Jari Heinonen, Juho

Lehmusto, Kalle Salonen and Jerzy Jasielec. The laboratory assistants Jarno

Karvonen and Jani Österlund deserve especial thanks for their assistance with the

experiments and laboratory work.

Finally, I would like to thank my wife Paula for keeping me going and being

on my side, and also my sons Niko and Oliver. My warm thanks go to my

relatives and siblings for their help and for believing in me, especially Heidi,

Kristiina, Markku, Pekka and Kirsti. I would also express my very special thanks

to my mother Seija and brother Marko, both of whom passed away early in 2012

– this would not have been possible without them. Lastly, I would like to thank all

my friends, especially Simo Ikonen, Lassi Urpilainen, Juho Syrjä and Antti Korpi.

Oulu, 2014 Pasi Karinkanta

10

11

List of abbreviations and symbols

CS Coefficients of classifier speed in the empirical models for air

classifier milling

CML Compound middle lamella of the wood cell wall

EW Earlywood

FSP Fibre saturation point

High GP Jet milling experiments in which the average overpressure of the

grinding air was approximately 10 bar

I(long) Coefficients of the rotational direction of a rotor towards a long

edge in the empirical models for air classifier milling

I(short) Coefficients of the rotational direction of a rotor towards a short

edge in the empirical models for air classifier milling

L Direction towards the longitudinal axis of wood

Low GP Jet milling experiments in which the average overpressure of the

grinding air was approximately 6 bar

LW Latewood

MS Coefficients of mill speed in the empirical models for air classifier

milling

MC Moisture content (%)

MFA Microfibrillar angle

ML Middle lamella of the wood cell wall

P Primary layer of the wood cell wall

R Direction towards the radial axis of wood

S1, S2, S3 Secondary wall layers 1, 2 and 3 in the cell wall of tracheids

T Direction towards the tangential axis of wood

WR Wood ray

XRD X-ray diffraction

A End-to-end amplitude of the oscillation movement in oscillatory ball

milling (m)

AR10 Aspect ratio corresponding to a value of 10% in the cumulative

projected area-based aspect ratio distribution

AR50 Median aspect ratio in the projected area-based aspect ratio

distribution

AR90 Aspect ratio corresponding to a value of 90% in the cumulative

projected area-based aspect ratio distribution

12

CrI Crystallinity index according to Segal et al. (1959) (%)

dS,X Sieve aperture with X% material pass (m)

d10 Particle diameter corresponding to a value of 10% in the cumulative

volume-based particle size distribution according to the laser

diffraction method (µm)

d50 Median particle size in the volume-based particle size distribution

according to the laser diffraction method (µm)

d90 Particle diameter corresponding to a value of 90% in the cumulative

volume-based particle size distribution according to the laser

diffraction method (µm)

Ek Estimated kinetic energy of the grinding ball during one end-wall

collision period (J)

ET Estimated total kinetic energy of the grinding ball during oscillatory

ball milling, also referred to as the total available impact energy (J)

f Oscillation frequency in oscillatory ball milling (Hz)

Mair Molar mass of air (g mol-1)

m Mass of the feed (g)

mball Mass of the grinding ball (g)

mP,dry Dry mass of the product in opposed jet milling experiments (g)

ML Mass loss due to torrefaction (%)

n Molar mass of air (mol)

p Pressure (bar)

pC Pressure of compressed grinding air (bar)

pUC Pressure of grinding air before compression (bar)

Q2 Cross-validated R2

R2 Coefficient of determination

R2(Adj.) Coefficient of determination adjusted for degrees of freedom

Rg Gas constant = 8.314 J mol-1 K´-1

SEC Specific electrical energy consumption of the system (J g-1)

SECJET Specific energy consumption based on the energy needed for

compression of the grinding air in jet milling (J g-1)

SECOSC Specific energy consumption based on the estimated total available

impact energy of the grinding ball in oscillatory ball milling (J g-1)

T Temperature of air (K)

t Milling time (s)

V Volume of air (m3)

VC Volume of compressed grinding air (m3)

13

VUC Volume of grinding air before compression (m3)

Δ* Width of particle size distribution

ΔAR Width of aspect ratio distribution

ΔWP Work done by a piston in isothermal and quasistatic compression of

an ideal gas (J)

ρ Density of air (g m-3)

σ Stress (MPa)

σC Crushing strength or peak stress in compression (MPa)

σu Ultimate strength (MPa)

14

15

List of original publications

This thesis is based on the following publications, which are referred throughout

the text by their Roman numerals:

I Karinkanta P, Illikainen M, Niinimäki J (2012) Pulverisation of dried and screened Norway spruce (Picea abies) sawdust in an air classifier mill. Biomass and Bioenergy 44: 96–106.

II Karinkanta P, Illikainen M, Niinimäki J (2014) Effect of different impact events in fine grinding mills on the development of the physical properties of dried Norway spruce (Picea abies) wood in pulverisation. Powder Technology 253: 352–359.

III Karinkanta P, Illikainen M, Niinimäki J (2013) Impact-based pulverisation of dried and screened Norway spruce (Picea abies) sawdust in an oscillatory ball mill. Powder Technology 233: 286–294.

IV Karinkanta P, Illikainen M, Niinimäki J (2013) Effect of grinding conditions in oscillatory ball milling on the morphology of particles and cellulose crystallinity of Norway spruce (Picea abies). Holzforschung 67(3): 277–283.

V Karinkanta P, Illikainen M, Niinimäki J (2014) Effect of mild torrefaction on pulverization of Norway spruce (Picea abies) by oscillatory ball milling: particle morphology and cellulose crystallinity. Holzforschung 68(3):337–343.

All the manuscripts were written by the primary author of this thesis, whose main

responsibilities were experimental design, data analysing, modelling and

reporting of the results. The co-authors participated in the design, analyses and

writing of the publications.

16

17

Contents

Abstract

Tiivistelmä

Acknowledgements 9 List of abbreviations and symbols 11 List of original publications 15 Contents 17 1 Introduction 19

1.1 Background ............................................................................................. 19 1.2 Outline of the thesis ................................................................................ 20

2 Present understanding 23 2.1 Norway spruce wood .............................................................................. 23

2.1.1 Chemical composition and microstructure ................................... 23 2.1.2 Mechanical properties and breakage behaviour ........................... 27

2.2 Dry fine grinding ..................................................................................... 32 2.2.1 Stressing conditions and material properties ................................ 33 2.2.2 Single and double impacts ............................................................ 36 2.2.3 Fine grinding techniques .............................................................. 38

2.3 Impact milling of wood and other lignocelluloses .................................. 40 2.3.1 Mills involving single impacts ..................................................... 42 2.3.2 Mills involving double impacts .................................................... 44 2.3.3 Pretreatments and milling conditions ........................................... 45

3 Aims of the study 49 4 Materials and methods 51

4.1 Raw material ........................................................................................... 51 4.2 Pretreatments prior to fine grinding ........................................................ 51 4.3 Fine grinding experiments....................................................................... 52 4.4 Wood powder sampling and analyses ..................................................... 55

4.4.1 Sampling ....................................................................................... 55 4.4.2 Analyses performed on diluted wood powder samples ................ 55 4.4.3 Analyses performed on dry wood powder samples ...................... 56

4.5 Modelling ................................................................................................ 57 5 Results and discussion 59

5.1 Fine grinding of dried Norway spruce wood in impact mills .................. 59 5.1.1 Air classifier mill .......................................................................... 59 5.1.2 Fluidised bed opposed jet mill ...................................................... 65

18

5.1.3 Oscillatory ball mill ...................................................................... 69 5.2 Comparison of impact milling techniques ............................................... 76 5.3 Effects of pretreatments and milling conditions on the properties

of Norway spruce wood in impact-based media milling ......................... 80 5.3.1 Drying and cryogenic grinding conditions ................................... 80 5.3.2 Mild torrefaction ........................................................................... 84

6 Conclusions 89 References 93 Original Publications 107

19

1 Introduction

1.1 Background

Wood is a renewable, abundant, non-food and carbon dioxide-neutral raw

material that can be used to replace non-renewable materials. Wood is widely

used as a fuel, construction material or raw material in cellulose and

lignocellulose-based products. Today there are various applications in which

wood is used in powdered form, such as thermoplastics and filters, and a lot of

research effort has been put into developing novel ways of utilising wood

powders (Kobayashi et al. 2008, Mori 2009).

The particle morphology and cellulose crystallinity of wood are crucial

factors in various processing steps and applications to value chains of different

kinds. These affect the digestibility of wood (Millett et al. 1975, 1976, 1979,

Fukazawa et al. 1982, Rivers & Emert 1987, Mooney et al. 1998, Mansfield et al.

1999, Chang VS & Holtzapple 2000, Chandra et al. 2007, Dasari & Benson 2007,

Zhu L et al. 2008, 2010, Kumar et al. 2009, Hendriks & Zeeman 2009, Zhu JY et

al. 2009a, Zhang M et al. 2012a, 2012b, Leu & Zhu JY 2013, Agarwal et al.

2013), and particle morphology is also important in its thermochemical

conversion (Reuther et al. 1985, Adams et al. 1988, Thunman et al. 2002, Lu et

al. 2010). When wood powder is employed as a filler or reinforcement material in

composites its particle morphology has an important role in the mechanical

properties of those composites (Zaini et al. 1996, Takatani et al. 2000, Stark &

Rowlands 2003, Khalil et al. 2006, Chen HC et al. 2006, Migneault et al. 2008,

2009, Bouafif et al. 2009, Renner et al. 2010, Kociszewski et al. 2012). Particle

morphology and cellulose crystallinity can be modified by mechanical treatment.

Since this mechanical treatment consumes a lot of energy (see Table 1), the profits

obtained through improved processing or from the final product need to exceed

the expense of the treatment for the latter to be economically feasible.

Although the fine grinding of wood has been studied in various ways, see

Millett et al. (1979), Fukazawa et al. (1982) and Agarwal et al. (2013), the main

subject under study have been the application of wood powder, not the

mechanical processing itself. Thus our present understanding of the process does

not extend to the effects of fine grinding conditions on the particle morphology

and cellulose crystallinity, which makes it very difficult to design and optimise

processes for the fine grinding of wood for different applications. The general aim

20

of this thesis is to obtain a knowledge of how the physical properties of Norway

spruce (Picea abies) wood develop during dry fine grinding with impact-based

mills under a variety of grinding conditions, and what influence those conditions

have on energy consumption.

1.2 Outline of the thesis

This thesis is organised into six chapters. Chapter 1 presents a brief background to

the topic. Chapter 2 summarises our present understanding of the structure and

properties of Norway spruce wood and the milling of wood, especially dry fine

grinding. The aims of the present research are presented in Chapter 3 and the

materials and methods used for testing the hypotheses in Chapter 4. The results

are summarised and discussed in Chapter 5 and the conclusions are presented in

Chapter 6.

21

Table 1. Energy requirements of different milling techniques in the dry grinding of

wood to a product size of less than 2 mm. Pilot-scale milling experiments are marked

(pilot). The energy requirement can be specified either for the milling equipment (mill)

or for the whole process (process).

Milling

technique

Raw material Feed size Product size Energy

requirement

Reference

Rotating knife

mill

Unspecified hardwood

(MC* = 4%–7%)

dS,100 =

22.40 mm

dS,100 = 1.60 mm 130.0 kWh t-1

(mill)

Gadoche &

López 1989

Hammer mill Unspecified hardwood

(MC* = 4%–7%)

dS,100 =

22.40 mm

dS,100 = 1.60 mm 130.0 kWh t-1

(mill)

Gadoche &

López 1989

Hammer mill

(pilot)

Poplar chips

(MC = 10%–15%)

dS,50 = 8a

mm

dS,99 = 1 mm 89.1 kWh t-1

(process)

Esteban &

Carrasco 2006

Hammer mill

(pilot)

Pine chips

(MC = 10%–15%)

dS,44 = 8a

mm

dS,99 = 1 mm 113.2 kWh t-1

(process)

Esteban &

Carrasco 2006

Hammer mill

(pilot)

Hybrid poplar wood chips

(MC* approximately 60%)

- dS,46 = 1a mm 102 kWh t-1

(process)

Schell &

Harwood 1994

Vibration mill Norway spruce

(MC* = 11%)

d = 22 mm d = 150 µm 800 kWh t-1

(process)

Kobayashi et al.

2008

Szego mill

(pilot)

Dry aspen wood chips

(MC = 13.5%)

dS,80 = 1.4

mm

dS,80 = 870 µm 310a kWh t-1

(mill)

Gravelsins 1998

Szego mill

(pilot)

Moist aspen wood chips

(MC = 45%–53%)

dS,80 = 1.4

mmc

dS,80 = 1.54 mm 200a kWh t-1

(mill)

Gravelsins 1998

Szego mill

(pilot)

Dry jack pine sawdust

(MC= 5.8%)

dS,80 = 2.2

mm

dS,80 = 970 µm 78 kWh t-1

(mill)

Gravelsins 1998

Szego mill

(pilot)

Moist jack pine sawdust

(MC= 37.4%)

dS,80 = 2.5

mm

dS,80 = 1.76 mm 100a kWh t-1

(mill)

Gravelsins 1998

Disc mill

(pilot)

Hybrid poplar wood chips

(MC* approximately 60%)

- dS,59 = 1.2a mm 122 kWh t-1

(process)

Schell &

Harwood 1994

Ultracentri-

fugal mill Spruce Sieved to

(2–4) mm

d50 = 197 µm 750 kWh t-1

(mill)

Repellin et al.

2010

Ultracentri-

fugal mill

Torrefied spruce Sieved to

(2–4) mm

d50 = 59 µm 90a kWh t-1

(mill)

Repellin et al.

2010

dS,X = sieve aperture with X% material pass, d50 = median particle diameter obtained by the laser

diffraction technique, d = particle diameter without no precise information provided in the source, MC =

moisture content in relation to total mass, MC* = moisture content with the relation to total mass or dry

mass not specified , a value read off from a graph and therefore not precise, b measuring method not

mentioned, and c obtained by sieving of dry wood chips

22

23

2 Present understanding

2.1 Norway spruce wood

Among the various wood species, Norway spruce (Picea abies [L.] H. Karst) was

chosen for this work because its properties and structure are well known and it is

an important natural resource in Germany, Austria, Slovakia, the Czech Republic,

Romania and the Scandinavian countries (Surmiński 2007). This section focuses

on the structure and mechanical properties of mature Norway spruce wood,

whereas the differences between trees grown in different surroundings or under

different nutritional conditions are not considered in detail.

2.1.1 Chemical composition and microstructure

The chemical composition of Norway spruce wood includes primarily cellulose

(> 39%) and lignin (< 36%) but also hemicelluloses and pectins (< 28%)

(Bertraud & Holmbom 2004, Surmiński 2007). Its microstructure comprises

mainly longitudinal cells, known as tracheids (up to 90%), the remainder being

mainly the cells forming wood rays (Hejnowicz 2007). An illustration of a cross-

section of a Norway spruce tree is provided in Fig. 1. Norway spruce tracheids are

elongated cells and have sometimes been described as hollow tubes of length

varying between 1.0 mm and 7.6 mm (Buksnowitz et al. 2010). Tracheid length

varies depending on characteristics such as the age of the tree and the radial

distance from the pith (Herman et al. 1998, Sarén et al. 2001, Buksnowitz et al.

2010). The cross-section of a tracheid is approximately rectangular, including an

open space known as the lumen (Sarén et al. 2001, 2006, Derome et al. 2012).

The length of the side of a tracheid seen in cross-section is less than 50 µm

(Havimo et al. 2008), and the cell wall thickness can vary between 1 µm and 9

µm (Havimo et al. 2008, Derome et al. 2012). The average cell wall thickness is

2.1 µm in earlywood and 3.9 µm in latewood (Havimo et al. 2008). The latewood

occupies from 2% to 30% of the total growth ring thickness (Hejnowicz 2007).

The walls of the tracheids contain bordered pits between 11 µm and 22 µm in

diameter (Hejnowicz 2007). These are mainly located in the radial walls of the

earlywood and are abundant at the overlapping ends of the tracheids (Hejnowicz

2007).

24

Fig. 1. Illustration of a cross-section of a tree. L denotes longitudinal axis, R radial axis

and T tangential axis.

Structure of tracheids

Tracheids are composed of a compound middle lamella (CML) and secondary cell

wall layers S1, S2 and S3, where the S2 layer comprises 80% of the entire width of

the secondary wall (Hejnowicz 2007). The CML, which contains a relatively high

amount of lignin but low amounts of cellulose and hemicelluloses (Jääskeläinen

& Sundqvist 2007: 50–51), can be separated into the middle lamella (ML) and

primary wall layer (P) (Jääskeläinen & Sundqvist 2007: 50–51). The ML binds

the tracheids to each other (Jääskeläinen & Sundqvist 2007: 50–51). The cell wall

of a tracheid is made up of a primary wall layer and the secondary wall layers,

where P is the outermost layer and S3 the innermost (Jääskeläinen & Sundqvist

2007: 50–51). The wall layers differ from others by virtue of the orientation of the

cellulose fibrils, as shown in Fig. 2. The angle between the cellulose fibrils and

the longitudinal axis is called the microfibrillar angle (MFA). Earlywood

tracheids exhibit an asymmetrical MFA distribution with values between -20° and

90° (Peura et al. 2008a). There are certain differences in MFA between the cell

wall layers, and also within a single cell wall layer, as shown in the cell wall

models (see Fig. 2). The average MFA tends to be between 20° and 35° near the

pith and decreases to below 20° towards the bark (Sarén et al. 2001, 2006, Peura

et al. 2008a, 2008b). It also tends to decrease within a single growth ring when

moving from the earlywood towards the latewood (Sarén et al. 2006, Lanvermann

et al. 2013). Likewise the relative distribution of cellulose, hemicelluloses and

lignin is known to vary between the cell wall layers (Panshin & de Zeeuw 1980:

106–107). According to Fengel and Stoll (1973), the thicknesses of the cell wall

layers lie in the range of 0.04–0.16 µm for CML, 0.12–0.71 µm for S1, 0.91–5.60

µm for S2 and 0.01–0.36 µm for S3.

25

Fig. 2. Cell wall models for Norway spruce: a) earlywood tracheid, b) latewood

tracheid from the mature wood and c) latewood tracheid from the juvenile wood, after

Brändström (2002). Black lines represent the orientation of cellulose fibrils in the cell

wall. Reproduced by permission of Jonas Brändström.

Cell wall structure of tracheids

In the cell walls of tracheids the cellulose molecules form the helically wound

reinforcing fibrils whereas the hemicelluloses and lignin provide a gluing and

stiffening matrix (Booker & Sell 1998). Besides the orientation of microfibrils in

a cell wall of wood also hemicelluloses and lignin can form orientated molecular

structures (Stevanic & Salmén 2009, Salmén et al. 2012). An elementary cellulose

fibril of native wood consists of 10 000 glucose units of length around 5.2 µm,

arranged so that they form a well-ordered crystalline structure or less ordered

amorphous mass (Jääskeläinen & Sundqvist 2007: 67–71, Fengel & Wegener

1989: 66–105). In Norway spruce earlywood 52% (± 3%) of the cellulose by

mass is in crystalline form (Andersson et al. 2004) and 30% (± 4%) of the wood

by mass is composed of crystalline cellulose (Andersson et al. 2004), or an even

lower proportion near the pith (Andersson et al. 2003). Jakob et al. (1995) report

that an elementary cellulose fibril in the S2 layer of Norway spruce wood is

uniformly 2.5 nm (± 0.2 nm) in diameter and is made up entirely of cellulose

crystallites. The crystallites oriented along the fibre axis in the S2 layer have been

reported to be 11 nm in length (Jakob et al. 1995). On the other hand, the average

length and diameter of cellulose crystallites in the S2 layer of Norway spruce

wood have been reported by Pirkkalainen et al. (2012) to be 33 nm (± 1 nm) and

26

2.7 nm (± 0.1 nm), respectively, values that are fairly close to those reported by

Andersson et al. (2003) and Peura et al. (2008a). Peura et al. (2008a) concluded

that the length of the cellulose crystallites can vary from 19.2 nm to 28.4 nm.

Some authors have suggested that there are no systematic variations in crystallite

dimensions as a function of the number of annual rings from the pith (Andersson

et al. 2003, Pirkkalainen et al. 2012). In contrast to this, Peura et al. (2007,

2008a) reported that there are small differences in the thickness and length of

crystallites between juvenile and mature wood. Individual elementary cellulose

fibrils in the S2 layer of Norway spruce wood together with some hemicellulose

glucomannan form cellulose fibril aggregates called macrofibrils with sides

ranging in length from 5 nm to 25 nm, averaging 15–16 nm, assuming a square

cross-section (Fahlén 2005). The centres of the elementary fibrils have been

estimated to be 4 nm apart (Jakob et al. 1996). The macrofibrils are surrounded

by lignin and hemicelluloses known as xylan and glucomannan (Fahlén 2005:

10). An illustration of an elementary cellulose fibril and macrofibrils in the S2

wall layer of Norway spruce wood is presented in Fig. 3.

Fig. 3. Illustrations of an elementary cellulose fibril and macrofibrils of various sizes in

the S2 cell wall layer of Norway spruce.

27

2.1.2 Mechanical properties and breakage behaviour

In mechanical terms, Norway spruce wood is a viscoelastic material (Havimo

2010), the viscous behaviour causes internal friction which converts any

mechanical energy imposed on it into heat (Eskelinen et al. 1982, Havimo 2009).

It also exhibits linear stress-strain behaviour when exposed to low magnitude

stresses, but when exposed to stresses of magnitudes close to its ultimate strength,

i.e. its breaking strength, non-linear stress-strain behaviour prevails (Dahl 2009).

The mechanical properties of Norway spruce wood vary depending on the loading

direction (Dahl 2009) and on structural variations such as the presence or absence

of early- and latewood and differences in MFA (Reiterer et al. 1999, Reiterer et

al. 2001a, Eder et al. 2008, 2009). Wood can be generally simplified as

orthotropic, i.e. by considering the mechanical differences in its radial,

longitudinal and tangential directions.

Fracture mechanics and breakage behaviour

In fracture mechanics of wood it is typical to distinguish three modes of loading

that lead to different forms of failure behaviour (Fig. 4). Frühmann et al. (2002)

have shown that pure tension failure can be more favoured energetically than pure

shear failure in Norway spruce wood. One reason for this may be energy losses

caused by friction between the fractured surfaces brought about by shear loads

(Frühmann et al. 2002). On the other hand, mixed mode loading (modes I and II)

can lead to failure that is energetically more favourable than pure tension failure

(Tschegg et al. 2001), since in this type the initial crack propagation always takes

place along the longitudinal axis irrespective of the starting notch or the degree of

mixity (Jernkvist 2001).

28

Fig. 4. Illustrations of different loading modes in fracture mechanics: a) mode I, tensile

stress leads to tension failure, b) mode II leading to in-plane or forward shear failure,

and c) mode III, leading to out-of-plane or transverse shear failure.

The breakage of wood can occur due to the separation of cells from each other by

peeling (intercellular failure), in which case the crack propagates mainly via the

CML (Boatright & Garrett 1983, Ashby et al. 1985). It is also possible that

breakage can occur within the cell wall layer (intracellular failure) (Boatright &

Garrett 1983, Ashby et al. 1985), leading to either intrawall failure, i.e. failure

within the secondary cell wall, or transwall failure, in which the fracture path

intersects the cell wall (Côté & Hanna 1983). Tensile stressing experiments with

Norway spruce wood have shown that fracturing mainly takes place in an

intercellular or intrawall manner when the stresses are perpendicular to the

longitudinal axis (Persson 2000: 46–51, Dill-Langer et al. 2002, Frühmann et al.

2003a, Wittel et al. 2005, Keunecke et al. 2007, Lanvermann et al. 2014),

whereas transwall failure takes place when there is tension along the longitudinal

axis (Bodner et al. 1997, 1998, Frühmann et al. 2003b, Müller et al. 2003). When

tension is perpendicular to the longitudinal axis stresses in excess of 1 MPa but

significantly less than 10 MPa are needed for breakage (Dill-Langer et al. 2002,

Wittel et al. 2005, Dahl 2009, Lanvermann et al. 2014), whereas if the tension is

parallel to the longitudinal axis over 30 MPa is needed for breakage of the wood

(Dahl 2009) or of a single tracheid (Eder et al. 2009). The stressed area in

experiments with wood samples is typically calculated as the whole area of the

cross-section in which the failure takes place. This results in lower ultimate

strengths by comparison with calculations that consider only the cross-section

area of the cell wall, so that the open space represented by the lumen is excluded.

When the latter calculation method is used local longitudinal ultimate tensile

strengths may be obtained that are significantly over 200 MPa (Burgert et al.

2003, 2005, Eder et al. 2008, 2009), i.e. strengths of this magnitude need to be

exceeded locally to cause transwall failure in single tracheids. There is no

29

information about the tensile loading of single tracheids in a tangential or radial

direction as their dimensions are too small for the tensile testing equipment (Eder

et al. 2013). In the case of shear experiments, stresses over 0.8 MPa but less than

9 MPa are sufficient to cause in-plane shear failures in different loading directions

(Dahl 2009). Dahl (2009) reported that in the case of samples where the shear

loading was initially directed to cause crack propagation across the longitudinal

axis the crack did not propagate in this direction but along the longitudinal axis.

Thus the shear stresses reported by Dahl (2009) do not consider the ultimate shear

strength required for transwall failure in Norway spruce wood. By considering

only the cross-section area of the cell wall in the calculations Gindl & Teischinger

(2002) estimated the shear strength of the cell wall parallel to the longitudinal

axis to be 27.5 MPa, i.e. this figure needs to be exceeded locally for intrawall

shear failure.

Compression failure

During compression the breakage of wood takes place due to tension and shear

failures but is accompanied by plastic deformation of its cells, i.e. buckling and

collapsing (Dumail & Salmén 1996, Poulsen et al. 1997, Persson 2000: 46–51,

Reiterer & Stanzl-Tschegg 2001b, Gindl and Teischinger 2002, De Magistris &

Salmén 2005, Benabou 2008, 2010). Typical stress-strain curves for Norway

spruce wood in the case of radial, tangential and longitudinal compression are

illustrated in Fig. 5. The stress-strain curves shown in Fig. 5 have some general

characteristics that have been distinguished for wood under compression (see

Gibson and Ashby 1988: 283–290, 309–311). At low stresses linear-elastic

deformation prevails, but when the stress increases to come close to the peak

stress plastic deformation and failures take place (Gibson and Ashby 1988: 283–

290, 309–311). The plateaux of the stress-strain curves correspond to a

progressive crushing of the wood, involving plastic deformation, during which a

significant increase in density is observed due to collapsing of the cells, allowing

the denser wood to resist compression, as can be observed in the stress-strain

curve in the form of strain toughening (Gibson and Ashby 1988: 283–290, 309–

311). The increase in strength due to densification by compression nevertheless

remains lower than could be expected from the increased density alone

(Blomberg et al. 2005). The mechanical strength involved in compression is

typically reported as taking the form of crushing strength, or peak stress, as

illustrated in Fig. 5 for the longitudinal, radial and tangential compression curves.

30

The crushing strengths of spruce wood are listed in Table 2, where it can be seen

that the crushing strength is much higher in a longitudinal direction than in any

other direction and that in addition to density, the compression rate and moisture

content have a significant influence on the crushing strength.

Apart from single static loading, cyclic compressive loading in tangential and

radial directions with low applied stresses (0.5–0.7 MPa) can cause local

collapsing of the earlywood and is particularly concentrated on the surface of the

wood that is in contact with the compressive force (Salmi et al. 2009, 2012a,

2012b). When considering cyclic compression in a longitudinal direction, Clorius

et al. (2000) reported that the fatigue life in wood is highly dependent on the

loading frequency, in that the total time to failure decreases with increasing

frequency. Visual failure was observed in the form of either a fine-meshed pattern

of compression zones scattered over a larger section or else localised failure

(Clorius et al. 2000). The fine-meshed failures increased with decreasing loading

frequency and increasing moisture content (Clorius et al. 2000).

Fig. 5. Illustration of a typical stress-strain curves for Norway spruce wood under

tangential, radial and longitudinal compression.

31

Table 2. Crushing strength of spruce wood under single compression. Values

including an asterisk (*) were read off from a graph and are therefore not precise.

Otherwise crushing strength is reported in terms of mean values. Bolded values apply

to Norway spruce wood, whereas Widehammar (2004) did not specify the species of

wood.

Loading

direction

MC Density

(g cm-3)

Compression rate Crushing strength/Peak

stress σC (MPa)

Reference

Longitudinal

‘Oven-dry’ low/medium/high 90* / 130* / 160* (Widehammar 2004)

8% 0.40 5.0 mm min-1 47.4±2.4 (Benabou 2008)

12% 0.41 0.3 mm min-1 45 (Poulsen et al. 1997)

12% 0.4*–0.9* 0.5 mm min-1 30*–110*

(Gindl & Teischinger

2002)

12–13% 0.40–

0.43

5.0 mm min-1 49* (Reiterer & Stanzl-

Tschegg 2001b)

10–15% 0.42–

0.48

(1.2 / 12000 /

180000) mm/min-1

40.2a / 51.8a / 56.0a (Eisenacher et al.

2013)

‘Fiber S’ low/medium/high 30* / 39* / 56* (Widehammar 2004)

‘Fully S’ low/medium/high 26* / 37* / 52* (Widehammar 2004)

Radial

‘Oven-dry’ low/medium/high 9* / 10* / 15* (Widehammar 2004)

12–13 0.40–

0.43

5.0 mm min-1 3.5* (Reiterer & Stanzl-

Tschegg 2001b)

‘Fiber S’ low/medium/high 3.6* / 5.0* / 7.5* (Widehammar 2004)

‘Fiber S’ 0.38–

0.48

0.0025 s-1/ 25 s-1 2.4* / 3.8* (Uhmeier & Salmén

1996)

‘Fully S’ low/medium/high 2.8* / 6.0* / 11* (Widehammar 2004)

‘Fully S’ 0.38–

0.48

0.0025 s-1/ 25 s-1 2.0* / 3.8* (Uhmeier & Salmén

1996)

Tangential

‘Oven-dry’ low/medium/high 19* / 24* / 30* (Widehammar 2004)

‘Fiber S’ low/medium/high 4.5* / 7.5* / 10* (Widehammar 2004)

‘Fully S’ low/medium/high 4.0* / 11* / 16* (Widehammar 2004)

‘Fiber S’ means fibre-saturated conditions in the wood, which in the case of Norway spruce is 30–34%

on a dry basis and 23–25% on a wet basis. ‘Fully S’ means a condition of the wood in which it carries as

much water as possible. Low/medium/high represent different strain rates in the split Hopkinson pressure

bar (SHPB) experiments, although the strain rate cannot be prescribed accurately due to the testing

technique. a Lateral dilation of the samples was constrained during compression.

32

Influence of moisture content

Moisture content has a significant influence on the mechanical and fracture

properties of wood (Gerhards 1982, Eskelinen et al. 1982, Salmén 1982,

Bengtsson 2000, Vasic & Stanzl-Tschegg 2007). When the moisture of wood

increases beyond the fibre saturation point (FSP), which varies between 30% and

34% on a dry basis and between 23% and 25% on a wet basis, the cell walls cease

swelling because they are totally saturated and the timber strength no longer

varies with moisture content (Bosshard 1974: 214, Berry & Roderick 2005). The

energy loss due internal friction is significantly higher for water-saturated

Norway spruce wood than for air-dried samples, even at temperatures below 30°C

(Eskelinen et al. 1982).

Moisture content also has a significant effect on the glass transition point of

amorphous cellulose, hemicelluloses and lignin (Jääskeläinen & Sundqvist 2007:

127–130). The polymer changes its physical state from a glass-like brittle material

to a rubber-like viscous material when the temperature increases beyond the glass

transition point (Salmén 1982, Jääskeläinen & Sundqvist 2007: 127–130). This

point occurs around 200°C in hemicelluloses, amorphous cellulose and lignin in a

dry state (Salmén 1982, Jääskeläinen & Sundqvist 2007: 127–130), but when the

moisture content of the wood increases hydrophilic hemicelluloses can be

softened at room temperatures (approximately 20°C) (Salmén 1982, 2004). A

higher moisture content will also lower the glass transition point of lignin and

amorphous cellulose, but these do not soften at room temperature (Salmén 2004,

Jääskeläinen & Sundqvist 2007: 127–130). In mechanical pulping the stressing

frequency also influences the softening temperature of the wood (Salmén et al.

1999).

2.2 Dry fine grinding

In mineral processing, grinding is the last stage in mechanical size reduction and

is performed using mills of various kinds after the crushing stage (Bernotat &

Schönert 2000, Wills 2006). It is sometimes referred as milling and is also

commonly separated into two processes, wet and dry grinding. According to

Lowrison (1974: 103–108) wet grinding typically means the grinding of a

material containing about 50% of liquid by volume in the uncombined state. The

present work will consider only dry grinding processes. Jankovic (2003) separates

grinding as used in mineral processing into four subcategories based on the

33

particle size of the product: conventional grinding, regrinding, fine grinding and

very fine grinding. Other subcategories to be found in the literature include

coarse, superfine and ultrafine grinding (Lowrison 1974: 60, 115, Orumwense &

Forssberg 1992, Jankovic 2003, Wang & Forssberg 2007, Zhao et al. 2009,

Barakat et al. 2013). There are some differences between authors, however, in the

product size ranges implied by these terms, so that fine grinding, for example, has

been classified as reduction to a particle size range of 125–1000 µm by Lowrison

(1974: 60), 100–1000 µm by Hukki (Lowrison 1974: 115), 200–600 µm by

Taggart (Lowrison 1974: 115) and 10–30 µm by Jankovic (2003). Barakat et al.

(2013) categorised the fine grinding of lignocellulosic biomasses to product size

range less than 100 µm, which is used as the targeted median particle size of the

product in this work.

2.2.1 Stressing conditions and material properties

To describe the material breakage achieved by different forms of size reduction

equipment it is convenient to evaluate our knowledge of this breakage in terms of

stressing conditions (Rumpf 1990: 103–114). This would make it possible to

evaluate the breakage mechanism involved when changing the size reduction

equipment, the physical conditions or the nature of the material. Peukert and

Vogel (2001) developed a processing model which distinguishes between material

properties and stressing conditions. Material function includes the information

about breakage rate and breakage function, which are in turn dependent on the

stressing conditions and stressing history that the particle has experienced

(Peukert 2004). The stressing conditions as such are considered in the role of a

machine function which comprises the type of size reduction equipment and the

conditions of operation (Peukert and Vogel 2001).

Material function

When considering a material function it must be noted that there are various

differences between the mechanical properties of materials. Even a single particle

may contain regions with different ultimate strengths, and crack propagation

during loading will probably take place in a region where the ultimate strength is

the same or lower than the stress applied (Fig. 6). Crack propagation fatigue can

take place even when the stress is insignificant, and this can weaken the material

so that the maximum stress or energy that is required for failure in the next

34

stressing event is lowered (Scott-Emuakpor 2007, Campbell 2012). Breakage of

the material does not necessarily happen at the strain rate equivalent to the

ultimate strength (Fig. 7a), as it only takes place when the strain energy density is

sufficient for the breakage (Fig. 7b) (ASM International 2003: 205, Jenkins &

Khanna 2005: 209–211). Chemical and physical treatments can be used prior to

mechanical size reduction to weaken the particle structure.

Machine function

A machine function involves three main parameters: stress intensity, stress

number and stress type (Peukert and Vogel 2001, Peukert 2003, 2004). According

to Peukert (2013), stress intensity should properly be termed stress energy, the

energy transferred to the particle upon stressing. The stress number, or frequency,

then indicates how many stress events take place during a certain period of time

in the course of mechanical size reduction (Bernotat & Schönert 2000, Peukert

and Vogel 2001, Peukert 2003, 2004). The third parameter, stress type, refers to

the type of applied stress (Bernotat & Schönert 2000, Peukert and Vogel 2001,

Peukert 2003, 2004). Compressive stresses are applied by compression or impact,

which can cause breakdown of the initial particles to smaller ones (Wills 2006,

Ennis et al. 2008), while shear stresses are typically applied by abrasive or cutting

techniques (Ennis et al. 2008, Lowrison 1974: 255–260). Other machine

functions include physical conditions such as the processing temperature and

pressure, which can have a significant influence on the breakage behaviour of the

particles.

35

Fig. 6. Illustration of a heterogeneous particle breakage under compressive loading at

varying compressive stresses. It is assumed that the stress is maintained for as long

as is needed for breakage to occur independently of the strain.

36

Fig. 7. Illustrations of different stress-strain curves a) with breaking points and

ultimate strengths and b) the blue (Fig 7a) stress-strain curve under different

stressing conditions.

Mill and product related stressing models

Kwade (2003) make a distinction between mill and product related stressing

models, in which the stressing conditions are considered from different points of

view. In the mill related stressing model the mechanical size reduction behaviour

of a mill is characterised by the type of the stress events, the number of such

events produced in the mill per unit time and the stress energy, which is the

magnitude of the energy that can be supplied to the product particle by the mill in

each stress event (Kwade 2003). In the product related stressing model, the result

in terms of the mechanical size reduction is determined by the type of stress, the

absolute number of stress events acting on one feed particle and the magnitude of

the specific energy or specific force during each stress event (Kwade 2003).

2.2.2 Single and double impacts

Single-particle breakage tests are used to study material behaviour under

conditions of slow compression and impact (Rumpf 1973, 1990: 108–114,

Tavares 2007), whereas abrasive failure has its own branch of science, called

tribology (see ASM International 2003: 259–266). Single-particle breakage tests

involve three main loading methods that result in outcomes of different kinds:

single impact, double impact and slow compression (Tavares 2007). In the first

case the particle is loaded by impact with one contact point (Fig. 8a), in the

second there are two or more contact points (Fig. 8b) and in slow compression the

particle is compressed between two planes (Fig. 8c). Special kinds of impact

testers can also be used in relation to mechanical pulping, e.g. where movement is

37

restricted at one end of the particle and the impacting device hits the other end

(Fig. 8d) (Eskelinen et al. 1982, Marton & Eskelinen 1982, Berg 2001). In this

case the directions of the forces applied to the particle during impact are similar to

the directions of those applied in shear failure (see Section 2.1.2), so that this

impact type will be referred to as shearing impact in the present work. Single-

particle breakage tests have been used to obtain knowledge about material

breakage behaviour under different loading conditions which can be used to

predict milling behaviour (Vogel & Peukert 2002, 2003, 2004, 2005, Meier et al.

2008, 2009). Stressing conditions can also be utilised directly in milling, for

example the mathematical expression for stressing conditions in stirred media

milling evaluated by Kwade et al. (1996) is nowadays being used in many studies

aimed at the optimisation of stirred media milling.

Fig. 8. Illustrations of different stress types: a) single impact, b) double impact, c) slow

compression and d) shearing impact. In the figure vPARTICLE is the particle velocity, vBALL

the grinding ball velocity, ω1 the angular velocity of the roll 1, ω2 the angular velocity

of roll 2 and vHAMMER the velocity of the impacting hammer.

Single particle tests have been used to obtain knowledge in relation to wood

construction and mechanical pulping. The crushing strength of wood is

significantly higher when it is exposed to a single or double impact than in typical

compression tests (Reid & Peng 1997, Pierre et al. 2013). With double impacts,

for instance, the longitudinal crushing strength of Norway spruce wood increases

approximately from 50 MPa to 70 MPa when the impact speed is increased from

0.001 m s-1 to 3.0 m s-1 (Neumann et al. 2011). Pierre et al. (2013) have reported

that fibre and fully saturated spruce (Picea excelsea) and poplar (Populus

euramericana) exhibit less fractured zones and a lower crushing strength than air-

dried samples when imposed to double impacts at a speed of 1.7 m s-1. As in

double impact tests, the impact speed in the case of single impacts increases the

crushing strength, but this effect has been found to be even more marked when

the impact is imposed across the grain rather than along the grain (Reid & Peng

1997). The energy losses in a single impact are higher than in typical compression

38

tests (Reid & Peng 1997). Some of the kinetic energy carried by the specimens is

converted to thermal energy and the work needed for the inelastic deformation

during the impact event, but some is transferred back due to rebounding (Reid &

Peng 1997). The deformation mechanism is more localised in single impact cases

than in typical compression tests (Reid & Peng 1997), and disintegration of the

specimen takes place near the targeted surface when the impact is along the grain,

whereas when it is perpendicular to the grain the deformation propagates into the

undeformed part of the specimen leaving the characteristic mushroom profile in

the deformed part (Reid & Peng 1997). Shearing impact studies consider only the

specimen orientations that cause fractures along the grain, because these are the

most closely related to the behaviour of wood in mechanical pulping (Eskelinen et

al. 1982, Marton & Eskelinen 1982, Berg 2001). Eskelinen et al. (1982)

concluded that there is no significant difference in impact energy with respect to

the direction of the fracture. The fracturing energy of Norway spruce wood in the

case of radial shearing impacts can be up to four times larger than that in pure

cleavage (Marton & Eskelinen 1982). Fracturing energy in shearing impact can

also be affected by the impact velocity (Berg 2001).

2.2.3 Fine grinding techniques

Yokoyama & Inoue (2007) classified fine grinding mills into five groups: impact

mills, ball media mills, air jet mills, roller mills and those of some other type. The

mills classified in this scheme as impact mills could more conveniently be named

rotor impact mills, as in the definition proposed by Nied (2007). In rotor impact

mills the size reduction is achieved by single impacts, where the kinetic or impact

energy being applied through the rotary movement of rotors (Nied 2007). Rotor

impact mills with integrated classifiers are frequently referred to as classifier

mills and can be used for producing the finest material of all the types of rotor

impact mills, whereas typical hammer mills are used for the production of coarser

material (Nied 2007). Classifiers are used to control the residence time

distribution of the particle inside the milling zone, which correlates with the

number of stress events (Peukert & Vogel 2001). Ball media mills are mills in

which grinding media such as balls or beads are driven by movement of the mill

casing or by an agitator (Yokoyama & Inoue 2007). These mills involve single

impacts in which the grinding balls collide with particles, and also double impacts

in which particles are caught between colliding grinding balls or between

grinding balls and the mill casing. There are generally no restrictions on the use

39

of non-spherical grinding media in these mills, which is a good reason why for

referring to them as media mills or grinding media mills, as preferred by Ennis et

al. (2008), Bernotat & Schönert (2000) and Orumwense & Forssberg (1992). An

air jet mill is a fluid energy mill, or jet mill, that uses high velocity jets of air to

impart energy to particles for size reduction (Chamayou & Dodds 2007). Besides

air, nitrogen gas and steam can also be used as the fluid in such mills (Camayou

& Dodds 2007). Jet mills can be further categorised as spiral, opposed, oval

chamber and target jet mills (Camayou & Dodds 2007, Ennis et al. 2008).

Yokoyama & Inoue (2007) have classified spiral and oval chamber jet mills as

attrition type jet mills. Spiral jet mills are also known as pancake mills (Camayou

& Dodds 2007). Fluidised bed opposed jet mills are ones in which air jets are

used to provide high-energy single impacts between particles in a fluidised bed

(Chamayou & Dodds 2007). Here the kinetic energy of the grinding air can be

evaluated from the grinding pressure, which has a significant influence on the

particle breakage mechanism (De Vegt 2007, Palaniandy et al. 2008, Palaniandy

& Azizli 2009). Jet mills are typically employed in the last stage of the dry fine

grinding process, to obtain the finest particles (Camayou & Dodds 2007,

Yokoyama & Inoue 2007). Roller mills can be divided into mills that use a roller

tumbling on a table or in a vessel (roller tumbling type) and those in which the

feed is ground between cylindrical rolls (roll type) (Yokoyama & Inoue 2007).

Roller mills with smooth roll profiles can be used to apply a fine grinding

technique based on slow compression. The “mills of other types” include those

which cannot be classified as rotor impact mills, grinding media mills, jet mills or

roller mills, including those in which cutting tools such as knives, shears, saws

and spikes are employed (Lowrison 1974: 255–260). Disc refiners or disc mills

are the ones that are most commonly used for refining wood and cellulose, i.e. in

mechanical pulping (Sundholm 1999, Tienvieri et al. 1999, Zhu JY 2011). In

these a rotor provides the mechanical energy needed for size reduction, as in rotor

impact mills, but due to their specific breakage mechanism designed to cause

defibration and the fibrillation of fibres (Salmén et al. 1999), they are classified as

“other mills” in the present scheme. Apart from disc refiners, grinding wheels

operating under varying physical conditions are also used for mechanical pulping

(Liimatainen et al. 1999, Sundholm 1999). Since the disc refiners and grinding

wheels used in mechanical pulping have been optimised for the production of

long fibres, their potential for fine grinding is poorly known. A summary of the

classification of fine grinding mills is presented in Table 3.

40

Table 3. Classification of fine grinding mills based on suggestions by Orumwense &

Forssberg (1992) and Yokoyama & Inoue (2007).

Category Technique Mill types/models

Media mills

Case driven Tumbling type, vibration type, planetary type, centrifugal fluidised-

bed type

Agitator driven Tower type, agitation vessel type, tubular type, annular type

Non-media mills

Rotor impact mills High-speed rotation disc type, hammer type, axial flow type, annular

type

Jet mills Target collision type, opposed jet type, attrition type

Roller mills Roller tumbling type, roll type

Other mills Cutting type, grinding wheel type, disc refiners, mortar and pestle

2.3 Impact milling of wood and other lignocelluloses

Impact milling is preferred for the mechanical size reduction of viscoelastic

materials, and there are in principal three impact milling techniques that can be

used for fine grinding purposes: opposed jet mills, high-speed rotor impact mills,

and grinding media mills operated in the impacting mode. The impact events

involved in these techniques are illustrated in Fig. 9.

Fig. 9. Schematic illustration of the impact events involved in a) an opposed jet mill, b)

a high-speed rotor impact mill, and c) an oscillatory ball mill, which is one kind of

vibration mill. The mills are not illustrated to scale with respect to size. Paper II,

published by permission of Elsevier.

One way of distinguishing the breakage mechanisms in milling is to consider the

developments in particle size, shape and cellulose crystallinity that occur during

milling, all of which can be attributed to the mechanical properties of the wood.

Changes in particle size, shape and cellulose crystallinity in Norway spruce wood

subjected to different impact loading intensities are listed in Table 4 in relation to

41

breakage behaviour (see Section 2.1.2). Intensity I in Table 4 represents loading

with stresses that are unable to cause fracturing in any direction by means of a

single stressing event. Intensity II represents a single loading situation in Norway

spruce wood in which interwall and intrawall failures can take place but transwall

failures do not occur. Cracks propagate by means of the energetically most

favoured path, which is along the grain, leaving the particles rather elongated in

shape. The breakage behaviour of the wood particles at Intensity II is rather

similar to the Class I breakage in the classification suggested by Leu & Zhu JY

(2013). Intensity III can cause severe intracellular- and intercellular damage in

early- and latewood independently of the loading direction, which leads to less

elongated shapes than with Intensity II. Intensity IV represents an impact which is

sufficient for the amorphisation of cellulose. According to Vincent (1990), the

theoretical strength of crystalline cellulose is 25 GPa, which is many magnitudes

higher than the stresses needed for transwall failures in tracheids (see Burgert et

al. 2003, 2005, Eder et al. 2008, 2009). The breakage behaviour of wood particles

at Intensities III and IV is rather similar to Class II breakage in the classification

suggested by Leu & Zhu JY (2013). As the stressed area during particle size

reduction may decrease as the particle size decreases, possibly resulting in higher

local stress, the occurrence of Intensities I–IV in milling is influenced by the

particle size range.

42

Table 4. Possible breakage behaviour due to differences in ultimate strength in

Norway spruce wood when exposed to a single impact of different intensities, and its

influence on the size, shape and relative degree of cellulose crystallinity.

Impact intensity Breakage behaviour Changes in size, shape and cellulose crystallinity

I

(No failures) - No breakage

- Fatigues are possible

- No changes

II

(Intercellular and

intrawall failures)

- Fractures depend on the

orientation of the wood

- Intercellular and intrawall

failures in tracheids

- No transwall failures in

tracheids

- Fatigues

- Particle size decreases but tracheids can remain

1–7 mm in size in a longitudinal direction

- Shape changes but tracheids can remain

elongated

- No changes in relative degree of cellulose

crystallinity

III

(Transwall

failures)

- Fractures independent of the

orientation of the wood

- Intracellular and intercellular

failures in tracheids

- Fatigues

- Particle size decreases

- Shape changes

- No changes in relative degree of cellulose

crystallinity

IV

(Amorphisation

of cellulose)

- Fractures independent of the

orientation of the wood

- Intracellular and intercellular

failures in tracheids

- Amorphisation of cellulose

- Fatigues

- Particle size decreases

- Shape changes

- Relative degree of cellulose crystallinity decreases

2.3.1 Mills involving single impacts

Rotor impact milling

Rotor impact mills can be used to grind wood to a median particle size well over

100 µm (Himmel et al. 1985, Gadoche & López 1989, Schell & Harwood 1994,

Paulrud et al. 2002, Esteban & Carrasco 2006, Zhu JY et al. 2009a, Repellin et al.

2010). Integrated sieves or classifiers are used in rotor impact milling to control

the fineness of the product (Nied 2007), but Schell & Harwood (1994) report that

these are not recommended by hammer mill vendors for the milling of moist

wood chips due to low production rates, excessive heat build-up, screen blinding

43

and equipment damage. Due to these problems in the processing of moist

lignocelluloses, hammer mills are recommended only for the size reduction of

lignocelluloses of a moisture content up to 10–15% on a wet basis (Kratky &

Jirout 2011).

Rotor impact milled wood particles are considered to be more cylinder-like

than spherical in shape (Zhu et al. 2009a). Paulrud et al. (2002) have shown that

the size and shape of the wood particles are affected by the type of rotor impact

mill, but more distinct differences in the properties of wood powders are evident

when rotor impact milled wood is compared with cutting milled wood. Rotor

impact milled wood particles are more elongated than cutting milled particles

(Paulrud et al. 2002), but less elongated than those milled with a disc refiner with

or without steam preheating (Schell & Harwood 1994).

There is no information on whether rotor impact milling can provide

sufficient stress intensities to cause a decrease in the relative degree of cellulose

crystallinity, nor is it known whether wood can be ground to a median particle

size below 100 µm, which would involve significant destruction of the cell walls.

Paper I in the present thesis investigates the dry fine grinding of Norway spruce

wood with a high-speed rotor impact mill equipped with an integrated air

classifier and considers the effect of the rotor and classifier speeds on the

properties of the wood powders and specific energy consumption during the

processing. Rotor speed is associated with the energy and intensity of the single

impacts, whereas it is the classifier that controls the time spent in the milling

zone, i.e. the number of impact events. The higher this number, the finer the

product will be if the impact energy is sufficient to cause breakage.

Opposed jet milling

It has been shown for wheat straw that opposed jet milling can be used to produce

very fine particles without any significant decrease in cellulose crystallinity (Silva

et al. 2012). Significant differences in the degree of cellulose crystallinity were

observed between jet milled and grinding media milled samples when the median

particle size of the wheat straw was less than 100 µm (Silva et al. 2012).

There is no information on how opposed jet milling influences the physical

properties of dry wood, especially when ground to a median particle size below

100 µm. This was investigated in Paper II using a fluidised opposed jet mill with

an integrated classifier and varying the grinding pressure in order to obtain

44

different kinetic energies in the impacts and the classifier speed in order to alter

the number of stress events.

2.3.2 Mills involving double impacts

Grinding media milling

By comparison with non-media mills, media mills are known for their ability to

cause efficient destruction of the cell walls, leading to disappearance of the

fibrous structure in the wood (Fukazawa et al. 1982, Maurer & Fengel 1992,

Mikushina et al. 2003, Agarwal et al. 2013). According to Mikushina et al. (2003)

the fibrous structure of air-dried aspen sawdust is destroyed more rapidly in

planetary ball milling than in vibration or tumbling ball milling, where the latter is

the slowest technique for this purpose. Besides efficient destruction of the fibrous

structure, media mills are well known for their ability to reduce the cellulose

crystallinity of the particles in order to obtain total amorphisation of the cellulose

when milling wood and other lignocellulosic substances with a low moisture

content (Pew & Weyna 1962, Millett et al. 1979, Rivers & Emert 1987, Hon

1987, Maurer & Fengel 1992, Stubičar et al. 1998, Kobayashi et al. 2007, da

Silva et al. 2010, Zhang Q et al. 2011, Silva et al. 2012). Some authors have

reported that a low moisture content in the wood is extremely important when it is

necessary to reduce the relative degree of crystallinity of cellulose in grinding

media milling (Millett et al. 1976, Kobayashi et al. 2007). Grinding media

milling can also influence the macromolecular structure of the wood components,

which can be seen as an increase in the water-soluble oligomer content, the

formation of mechanoradicals and reductions in the degree of polymerisation of

cellulose and lignin (Chang H et al. 1975, Hon & Glasser 1979, Hon 1987,

Mikushina et al. 2002, Ikeda et al. 2002, Guerra et al. 2006). Mechanoradicals are

formed on account of the cleavage of covalent bonds in lignin and cellulose (Hon

1987, Guerra et al. 2006).

Kobayashi et al. (2008) used vibration milling to reduce Norway spruce

wood to a median particle size of 24 µm, producing rounded broken fibres with

smooth surfaces and a significant decrease in the relative degree of cellulose

crystallinity. According to Schwanninger et al. (2004) the decrease in cellulose

crystallinity achieved during the vibration milling of Norway spruce wood is

mainly due to the mechanical treatment, whereas temperature and chemical

45

effects are of only minor importance. The cell corners and CML of Norway

spruce wood are the most resistant elements in vibration milling, while in the

secondary cell wall layer it is the S1 layer that is particularly sensitive and is

loosened at an early stage, whereas the S2 layer is split into lamellae of varying

diameters from 10 nm to 500 nm (Maurer & Fengel 1992). Maurer & Fengel

(1992) observed the presence of globular particles in powdered Norway spruce

wood that tended to aggregate during vibration milling and increased in number

with the prolongation of milling. In view of the proposal put forward by Nichols

et al. (2002), the term “agglomeration” is used rather than “aggregation” for an

assemblage of powder particles in this work. This agglomeration of globular

particles may be the reason why Kobayashi et al. (2008) found that a prolongation

of vibration milling from 30 min to 120 min increased the median particle size of

ground Norway spruce wood from 30 µm to 45 µm.

The agglomeration that takes place in Norway spruce wood during vibration

milling means that there is a certain median particle size range that imposes

limitations on particle size reduction. This has not been confirmed, however, and

it is not clear whether higher stress energies will help to reduce the particle size

any further or will merely intensify the agglomeration. This was investigated in

Paper III, where dried Norway spruce wood was pulverised with a simple

vibration mill in which double impacts played an important role. The stress

energy was varied by changing the mass of the grinding ball or its oscillation

frequency. Milling time and oscillation frequency were used to adjust the number

of stress events. It also remains unknown how high the moisture content can be

without exercising any significant influence on the crystallinity of cellulose in

grinding media milling. This was studied in Paper IV.

2.3.3 Pretreatments and milling conditions

There are various chemical, enzymatic and thermal pretreatments that have been

applied prior to the mechanical size reduction of wood (Lindholm & Kurdin

1999, Kenealy & Jeffries 2003, Rapp et al. 2006, Laxman & Lachke 2009, Zhu

JY et al. 2009a, 2009b, 2010a, Zhu JY & Pan 2010b, Zhu W et al. 2010). Thermal

pretreatment temperatures from 50°C to 150°C are considered to lead to non-

reactive drying, in which lignocelluloses lose moisture and shrink (Tumuluru et

al. 2011). This dehydration can cause defects and cracks in the wood (Repellin et

al. 2010). Temperature range from 150°C to 200°C is considered as the reactive

drying range, which results in structural damage due to cell wall collapse, and

46

correspondingly temperatures from 200°C to 300°C are considered to constitute a

destructive drying range, involving depolymerisation and devolatilisation

(Tumuluru et al. 2011). Drying also affects cellulose crystallinity (Esteves &

Pereira 2009, Leppänen et al. 2011). In terms of changes in Norway spruce wood

at the cellular level, drying in a reduced oxygen atmosphere at temperatures up to

220°C has been found to cause deformation in the earlywood and tangential

intracellular cracks in the latewood (Welzbacher et al. 2011).

Torrefaction is a thermal pretreatment applied at temperatures in the range

from 200°C to 300°C in an inert atmosphere that alters the physical and chemical

composition of a lignocellulose (Tumuluru et al. 2011, van der Stelt et al. 2011,

Acharya et al. 2012, Broström et al. 2012). It has been shown that torrefaction

can be used to reduce the specific energy requirement in the grinding of wood

with various designs of mill in order to obtain the desired particle size (Arias et

al. 2008, Repellin et al. 2010, Almendros et al. 2011, Phanphanich & Mani 2011,

Chen W-H et al. 2011, Kokko et al. 2012, van Essendelft et al. 2013). The

decrease in energy consumption correlates with the anhydrous weight loss due to

torrefaction (Repellin et al. 2010, Kokko et al. 2012). The reason for this

behaviour has been said to lie in the increased brittleness of the lignocellulose

during torrefaction (Tumuluru et al. 2011, Acharya et al. 2012), but we have no

information on whether torrefaction actually has any influence on grindability and

cellulose crystallinity in grinding media milling to a median particle size below

100 µm. Another important question is whether torrefaction pretreatment weakens

the structure so as to allow smaller particle sizes to be achieved. This was studied

in Paper V using impact-based grinding media mill.

In the case of viscoelastic materials cryogenic grinding is considered to be a

more effective method for fine grinding purposes (Wilczek et al. 2004). In

addition, cooling affects the movements of the molecules, so that plastic

deformation by viscous flows is reduced (Wilczek et al. 2004). Cryogenic

grinding conditions have been employed in the grinding media milling of wood

(Hon & Glasser 1979, Hon 1987, Maurer & Fengel 1992), leading Maurer &

Fengel (1992) to conclude that these conditions have no appreciable influence on

milling intensity by comparison with atmospheric conditions. However, a

reduction in temperature down to -196°C will promote a decrease in the ultimate

strength of wood (Gerhards 1982, Jiang et al. 2014). Moisture content is also

important, because a higher moisture content provides greater changes in ultimate

strength when lowering the temperature (Gerhards 1982). On the other hand,

when considering material with a high moisture content the use of processing

47

temperatures that are below the glass transition point of hemicelluloses will

prevent softening of these (Jääskeläinen & Sundqvist 2007: 127–130). The

moisture content of living trees is around 40–50%, and brittle behaviour is

typically observed at temperatures below -40°C (Jääskeläinen & Sundqvist 2007:

16, 127–130). We have no information on how the size, shape and crystallinity of

particles are affected by cryogenic processing conditions in fine grinding of wood

samples of varying moisture content. This was studied in Paper IV for moisture

content values between 1% and 50% on a wet basis. The fine grinding was

performed with an impact-based grinding media mill that was tailored for

cryogenic grinding.

48

49

3 Aims of the study

Wood powders have many interesting possible applications, but due to a lack of

knowledge about the processes involved, their industrial production is looked on

as ecologically unfeasible. The technology needed for dry fine grinding already

exists, but we do not know how the properties of the wood develop during fine

grinding and how the processing should be optimised to minimise the specific

energy consumption. The general aim was to obtain information on how the

physical properties of Norway spruce (Picea abies) wood develop during dry fine

grinding with impact-based mills under varying sets of grinding conditions and

what influence these have on energy consumption. The specific aims were to gain

an understanding of:

1. developments in the size and shape of dried wood particles in fine grinding

with impact mills (Papers I–III),

2. the influence of grinding conditions on specific energy consumption in the

fine grinding of dried wood with impact mills (Papers I–III),

3. the influence of cryogenic milling conditions and the drying of fresh wood on

the size, shape and cellulose crystallinity of wood particles in fine grinding

with double impact mills (Paper IV), and

4. the influence of mild torrefaction on the grindability, particle shape and

cellulose crystallinity of dried wood in fine grinding with double impact mills

(Paper V).

50

51

4 Materials and methods

4.1 Raw material

Norway spruce Picea abies (L.) H. Karst trees were harvested mainly from the

southern parts of Finland. The harvested trees were debarked and processed with

a circular saw at the Seikun Saha sawmill (Pori, Finland). The fresh sawdust was

gathered from the circular saw process and sent to the University of Oulu, where

it was stored in a freezer prior to the fine grinding experiments.

4.2 Pretreatments prior to fine grinding

The sawdust was screened and usually dried before the fine grinding experiments.

Fresh sawdust was used only in the experiments presented in Paper IV. Drying

took place in a large oven at non-reactive drying temperatures of 80°C (Papers III

and IV) and 105°C (Papers I, II, V). The dry matter content after drying was

measured with an MA100 moisture analyser (Sartorius AG, Germany). The

moisture content values listed in Table 5 were calculated by deducting the

measured dry matter content from 100%, and therefore represent the moisture

content on a wet basis. Screening was performed with a vibratory sieve equipped

with a screen having 4 mm circular holes. The particle size distribution of the

screened and dried sawdust was measured with an Analysette 3 (Fritsch,

Germany) vibratory sieve shaker in Paper I and with an e200LS (Hosokawa

Alpine, Germany) air jet sieve in Paper III. The median particle size was between

1 mm and 2 mm, and less than 4% of the particles by weight passed through a

sieve with an aperture of 250 µm in both cases. In Paper II the dried and screened

Norway spruce was preground with the same air classifier mill as was used for the

fine grinding in Paper I. This was a high-speed rotor impact mill 50 ZPS

(Hosokawa Alpine, Germany) integrated in a closed grinding circuit with a 50

ATP air classifier (Hosokawa Alpine, Germany). After pregrinding, the finest

particles were removed with the 50 ATP air classifier and the coarser particles

were used as a feed in the fine grinding experiments. For Paper V several samples

of dried and screened Norway spruce sawdust were torrefied in a nitrogen

atmosphere at temperatures below 230°C before fine grinding.

52

Table 5. Drying temperatures and moisture content after drying.

Paper Drying temperature (°C) Moisture content (%)

I 105 < 2

II 105 < 4

III 80 < 3

IV 80 1–50

V 105 < 1

4.3 Fine grinding experiments

The fine grinding equipment, operational parameters, milling conditions and

pretreatments used in the work reported in Papers I–V are listed in Table 6, and

the differences in the impact angles used in the air classifier milling are illustrated

in Fig. 10. The fine grinding experiments were performed under ambient milling

conditions unless otherwise stated. Detailed experimental plans are presented in

the original Papers I–V. For Paper I the air classifier mill was used for fine

grinding and its specific energy consumption was calculated from the overall

electricity consumption as presented in Paper I. For Paper II a fluidised bed

opposed jet mill, 100 AFG (Hosokawa Alpine, Germany), and an oscillatory ball

mill, Cryomill (Retsch, Germany), were used for fine grinding purposes. Like the

air classifier mill, the jet mill was also integrated into a closed grinding circuit

with the 50 ATP air classifier (Hosokawa Alpine, Germany).

The work done in compressing the grinding air in the jet milling experiments

was calculated on the assumption that air behaves as an ideal gas and that

compression is an isothermal process. The air was compressed by means of a

GA11FF screw compressor (Atlas Copco, Italy), but for the purpose of the

calculations the compression work was regarded as quasistatic and isothermal

work done by a piston ΔWP and calculated using the equation

=

=Δ

UC

C

air

gUC

C

UCgP lnln

p

p

M

TRV

V

VTnRW

ρ , (1)

where n is the amount of the substance (air), Rg the gas constant (8.314 J mol-1 K-

1), T the air temperature, V the volume of air, Mair the molar mass of air, ρ the

density of air and p the pressure of air. The subscript “C” in Equation 1 stands for

the compressed air and “UC” for the uncompressed air.

The 100 AFG fluidised bed opposed jet mill contains built-in devices for

measuring the classifier speed, pressure of the grinding air and total air flow rate,

53

data on which were collected during milling. The volume of the grinding air was

measured with an SS 30.301 thermal inline flow sensor (Schmidt technology,

Germany) connected to the pipe used to transfer the grinding air from the

compressor to the jet mill. The temperature during the fine grinding experiments

was approximately 293 K. The air volume in the SS 30.301 sensor was displayed

and recorded as a normal volume (Nm3) which in this particular device is the

volume at 20°C and 101,325 Pa according to the manufacturer. The dry mass of

product, mP,dry, was calculated by subtracting the dry mass of the wood that did

not get past the classifier from the input dry mass. The specific energy

consumption in jet milling SECJET was calculated as

P,dry

PJET SEC

m

WΔ= , (2)

Further information on the experiments is provided in Paper II.

In Paper III the fine grinding was studied with the oscillatory ball mill. For

calculation of specific energy consumption in oscillatory ball milling there was

estimated the total available impact energy of a single grinding ball ET (see Paper

III). The total available impact energy was calculated as

keT SN EE ⋅= , (3)

where

tf ⋅⋅= 2SN e, (4)

and

22ballk 8 fAmE ⋅⋅≈ , (5)

where in Equations 4 and 5 f is oscillation frequency, t milling time, mball the mass

of the grinding ball and A the end-to-end amplitude of the oscillation. The specific

energy in oscillatory ball milling SECOSC was calculated as

m

ETOSC SEC = , (6)

where m is the mass of the feed.

For the work described in Paper IV, fine grinding of screened Norway spruce

sawdust samples of varying moisture content was performed using the oscillatory

ball mill operated under ambient and cryogenic grinding conditions, using

nitrogen liquid (boiling point -196°C) as a coolant. Similarly, the fine grinding of

54

dried, screened and torrefied Norway spruce sawdust reported in Paper V took

place in the oscillatory ball mill. Torrefaction was performed in a nitrogen

atmosphere and the mass losses due to torrefaction (ML) were evaluated according

to Almeida et al. (2010), the result being used to represent the intensity of

torrefaction. The specific energy consumption as indicated in Paper V was

calculated using Equation 6. .

Table 6. Milling equipment and studied operational parameters, milling conditions and

pretreatments.

Paper Milling equipment Mill type(s) Studied operational parameters,

milling conditions and pretreatments

I Air classifier mill: a Rotor impact mill 50

ZPS integrated with an air classifier 50

ATP (Hosokawa Alpine, Germany)

High-speed

rotation disc type

Rotor/mill speed (50 ZPS), classifier

speed (50 ATP), impact angle (50

ZPS)

II Air classifier mill, a fluidised bed opposed

jet mill 100 AFG integrated with the air

classifier 50 ATP (Hosokawa Alpine,

Germany) and an oscillatory ball mill

CryoMill (Retsch, Germany)

High-speed

rotation disc type,

opposed jet type

and vibration type

Grinding pressure (100 AFG),

classifier speed (50 ATP)

III Oscillatory ball mill CryoMill (Retsch,

Germany)

Vibration type Grinding ball mass, mass of the

feed, oscillation frequency, milling

time

IV Oscillatory ball mill CryoMill (Retsch,

Germany)

Vibration type Cryogenic milling conditions,

Moisture content of the feed

V Oscillatory ball mill CryoMill (Retsch,

Germany)

Vibration type Torrefaction

Fig. 10. Impact angles studied in the air classifier milling experiments, referred to as

“long edge” or I(long) and “short edge” or I(short). The black arrow indicates the

rotation direction of the rotor.

55

The fine grinding experiments were replicated in most cases in order to obtain

information about the uncertainty affecting the processing and analysing. In these

cases the mean values were used for plotting and error bars were used to represent

95% confidence intervals when assuming a normal distribution. In the tables

presented in Chapter 5 the results are expressed in the following form: mean

value ± confidence interval.

4.4 Wood powder sampling and analyses

The measured properties of the pulverised Norway spruce wood samples as

indicated in Papers I–V are listed in Table 7. Since their X-ray diffraction (XRD)

profiles were measured in order to evaluate the crystallinity index, the measured

XRD profiles are not shown in Papers I, II and V.

Table 7. Properties of the wood powders as measured for Papers I–V

Measured property Paper

I II III IV V

Particle size distribution x x x x x

Aspect ratio distribution x x x x x

XRD-profile and crystallinity index x x x x

4.4.1 Sampling

The wood powders produced with the air classifier mill and fluidised bed opposed

jet mill as described in Papers I and II were sampled with a PT-type sample

divider (Retsch, Germany) to obtain representative samples for analyses.

4.4.2 Analyses performed on diluted wood powder samples

Preparation of diluted wood powder samples

The diluted wood powder samples were prepared as described in Papers I–V.

Particle size distribution

The volumetric particle size distribution of diluted wood powder samples was

measured with an LS 13320 laser diffraction-based particle size analyser

56

(Beckmann Coulter, U.S.A). The median size, d50, was used to represent the

particle size and the dimensionless parameter Δ* to represent the width of particle

size distribution. The latter was calculated as

µm 10* 1090 dd −=Δ , (7)

where d90 and d10 are the particle sizes in micrometres corresponding to the 90%

and 10% values in the cumulative size distribution, respectively. The size ratios

(d90-d10)/d50 and (d90-d10)/(2 d50) are more commonly used to provide information

about the width of particle size distribution (Nakach et al. 2004, Adi et al. 2007,

Palaniandy & Azizli 2009, Toraman 2011, Kotake et al. 2012) but these ratios

depend on the median particle size of the product. To make comparison between

samples with different median particle size easier the divider in the size ratios was

replaced by a constant value of 10 µm.

Aspect ratio distribution

Aspect ratio distributions were measured from photos taken with a charged

coupled device (CCD) as described in Papers I–V. The minimum aspect ratio (=1)

represents a perfect filled circle. In the case of wood particles, a higher aspect

ratio typically represents a more elongated shape. The cumulative aspect ratio

distribution was used to calculate the median aspect ratio AR50 and the width of

the aspect ratio distribution ΔAR, the latter being defined as the difference between

the aspect ratios corresponding to the 90% and 10% values in the distribution. An

Ultra Plus field emission scanning electron microscope (Carl Zeiss, Germany)

was used for the imaging of the finest particles to gain insight of their shape, but

these images were not used to calculate the aspect ratio distribution.

4.4.3 Analyses performed on dry wood powder samples

Preparation of the dry wood powder samples

The wood powders produced as described in Papers I–II and V were used for the

measurement of XRD profiles without drying, but all those produced for Paper IV

were vacuum-dried before measuring the XRD profile.

57

XRD profile and crystallinity index

Tablets were prepared from the dry wood powders using a hydraulic press and X-

ray diffraction profiles were measured with a D5000 diffractometer (Siemens,

Germany) as described in Papers I, II, IV and V. The degree of crystallinity of the

cellulose in terms of the crystallinity index (CrI) was calculated from the XRD

profile in each case by the method proposed by Segal et al. (1959).

4.5 Modelling

Empirical models for specific energy consumption (SEC) and various wood

powder properties were developed as a function of the operational parameters in

Paper I. A Modde 7.0.0.1 software (Umetrics, Sweden) was used in the statistical

analyses of the experimental data. In these statistical analyses, responses such as

SEC and the various wood powder properties were modified with a mathematical

operator to gain normally distributed data and good statistical significance for the

models. The effects can be derived unambiguously from the coefficients of the

empirical models and are therefore used as synonyms for them in the discussion.

In Paper III the total available impact energy of the grinding ball in oscillatory

ball milling was estimated (Eq. 3) and used to model the effects of the grinding

ball mass, oscillation frequency and milling time on the properties of the wood

powders and to estimate the specific energy consumption in fine grinding.

58

59

5 Results and discussion

5.1 Fine grinding of dried Norway spruce wood in impact mills

5.1.1 Air classifier mill

The SECs and the properties of the wood powders obtained from the air classifier

mill are listed in Table 8, which indicates that the wood was ground to a median

particle size of around 23 µm with a Δ* of over 5. The finest powders had AR50

2.3 and ΔAR 3.3 or higher (Table 8). To provide an insight into the shapes of the

particles, CCD photos of sample ZPS 17 are presented in Fig. 11. The largest

particles shown in Fig. 11a can be described as elongated and rectangular in shape

rather than roundish, while the finest ones (Fig. 11c) resemble roundish particles

but it is difficult to distinguish their shapes accurately because the pixel resolution

of the CCD photos is around 1.6 µm. The FESEM image shown in Fig. 11d

nevertheless suggests that they, too, are rectangular rather than roundish. The

lowest CrI (approximately 41%) was obtained at the highest classifier and mill

speeds, whereas reducing the mill speed from 20,000 min-1 to 8000 min-1 seemed

to increase CrI. This means that air classifier milling had only a minor influence

on the CrI of the wood by comparison with grinding media milling, where total

amorphisation has proved possible (see Section 2.3.2). There were large

variations in cellulose crystallinity between samples ground with similar

processing parameters (see Table 8), possibly due to uneven drying conditions.

Thermal treatment (Esteves & Pereira 2009), and also air-drying (Leppänen et al.

2011), may affect the cellulose crystallinity of wood. In this case drying was

performed in large containers without any mixing, and therefore there were local

variations in drying conditions depending on the position of the wood in the

container.

60

Table 8. Specific energy consumption (SEC) in air classifier milling and the properties

of the resulting wood powders. ‘Short’ and ‘long’ impact angles refers to the rotational

direction of the rotor towards the short or long edge. Experiments ZPS10–ZPS18 are

replicates for ZPS1–ZPS9. Modified from Paper I, published by permission of Elsevier.

Experiment (run

order)

Mill speed

(min-1)

Classifier

speed (min-1)

Impact

angle

SEC

(kWh t-1)

d50

(µm)

∆* AR50 ∆AR CrI (%)

ZPS1 (4) 8000 2000 Short 2505 208.5 76.8 2.8 6.3 47.2

ZPS2 (14)1 8000 8000 Short 30,994 32.8 11.9 2.5 4.7 44.8

ZPS3 (11) 20,000 2000 Short 1712 223.0 70.7 3.5 6.6 49.6

ZPS4 (8)1 20,000 8000 Short 4098 25.4 7.0 2.3 3.4 41.8

ZPS5 (5) 8000 2000 Long 2991 189.6 70.0 2.8 6.0 43.4

ZPS6 (3) 8000 8000 Long 38,160 33.0 12.1 2.5 4.7 45.9

ZPS7 (10) 20,000 2000 Long 1664 207.6 68.3 3.4 6.6 49.6

ZPS8 (1)1 20,000 8000 Long 7879 25.9 7.5 2.4 3.7 41.2

ZPS9 (6) 14,000 5000 Short 2593 49.1 15.8 2.8 5.2 43.0

ZPS10 (2) 8000 2000 Short 2549 198.4 71.7 2.8 6.3 49.3

ZPS11 (15) 8000 8000 Short 17,589 32.7 12.3 2.4 4.8 45.5

ZPS12 (9) 20,000 2000 Short 1723 249.5 80.0 3.5 6.6 50.2

ZPS13 (13)1 20,000 8000 Short 6648 23.0 5.6 2.3 3.3 41.5

ZPS14 (18) 8000 2000 Long 2431 189.9 69.6 2.7 5.9 48.2

ZPS15 (17) 8000 8000 Long 17,496 31.6 11.8 2.4 4.6 46.3

ZPS16 (7) 20,000 2000 Long 1844 232.9 72.8 3.5 6.8 48.0

ZPS17 (16) 20,000 8000 Long 4310 23.8 6.2 2.3 3.4 44.5

ZPS18 (12) 14,000 5000 Short 2588 44.6 14.5 2.7 4.8 45.7

ZPS19 (19) 20,000 8000 Long 4245 - - - - -

1 Grinding time was less than 40 min

61

Fig. 11. Three CCD photos showing some of the a) largest b) intermediate-sized, c)

finest observable wood particles, and d) FESEM image showing some small

rectangular-shaped particles in the rotor impact milled sample ZPS17.

Development of wood powder properties

The results of experiments ZPS1–ZPS18 (Table 8) were used for modelling of the

wood powder properties as a function of the processing parameters. The

coefficients used in these empirical models for the wood powder properties are

shown in Fig. 12 and the main statistical parameters for the models in Table 9.

The high values for R2, R2(Adj.), Q2 and reproducibility in Table 9 indicate that

the empirical models are statistically valid and the results reproducible, except in

the case of the crystallinity index. The model validity is over 0.25 for all the

properties studied (see Table 9) and therefore there is no lack of fit in the models.

In the case of ΔAR the low model validity can be considered an artefact, since the

other statistical parameters are well over 0.9 (see Table 9). The reason for poor

statistical values in empirical model for CrI was the large variations in CrI values,

62

even when similar processing parameters were used (Table 8). The reason for the

poor statistical values in the empirical model for CrI lay in the large variations in

CrI values even when similar processing parameters were used (Table 8).

According to Fig. 12, the classifier speed, mill speed and their interaction had a

significant influence on the wood powder properties, but the choice between long

and short edges had a negligible influence, because the respective coefficients

with error bars cross the zero line (Fig. 12). The classifier speed alone had the

most significant effect on the properties of the wood powders (Fig. 12). Contour

plots for the empirical models for wood powder properties as a function of mill

and classifier speeds in the speed ranges studied here are shown in Fig. 13. The

effect of the impact angle is ignored in the models due its insignificant influence

on the properties of the wood. Fig. 13 suggests that there is a certain classifier

speed for every wood powder property at which changes in mill speed do not

influence the property and above which the influence of mill speed is the reverse

of what it was below that speed.

Fig. 12. Scaled and centred coefficients in the empirical models for the development

of wood powder properties in air classifier milling. The empirical models depict

responses that have been modified with the mathematical operator mentioned on the

y-axis. Modified from Paper I, published by permission of Elsevier.

63

Table 9. Main statistical parameters of the empirical models for the wood powder

properties (95% confidence level and normal distribution). Modified from Paper I,

published by permission of Elsevier.

Property R2 R2 (Adj.) Q2 Model validity Reproducibility Cond. no.

d50 0.998 0.997 0.995 0.857 0.997 6.214

∆* 0.996 0.994 0.991 0.853 0.994 6.214

AR50 0.986 0.982 0.976 0.912 0.979 1.118

∆AR 0.985 0.980 0.973 0.442 0.987 1.118

CrI 0.728 0.644 0.480 0.750 0.666 1.118

Fig. 13. Contour-plotted empirical models for given wood powder properties as a

function of the classifier and mill speeds. The effect of the impact angle is ignored.

Modified from Paper I, published by permission of Elsevier.

Specific energy consumption

An SEC of over 5.8 kJ g-1 (1600 kWh t-1) was needed to obtain a d50 of around

190 µm in the fine grinding of Norway spruce wood, and over 15.5 kJ g-1 (4300

kWh t-1) to obtain a d50 of around 23 µm (Table 8). When a low mill speed was

used with the highest classifier speed, d50 was approximately 10 µm higher and

the specific energy consumption more than doubled relative to processing at a

high mill speed (Table 8). The increased specific energy consumption indicates

that a mill speed of 8000 min-1 provides insufficient stress energy when grinding

64

wood to a median particle size range of around 30 µm, so that higher mill speeds

should be used for fine grinding purposes.

The results shown in Table 8 were used for modelling SEC as a function of

the processing parameters. The specific energy consumption figures obtained in

experiments ZPS2, ZPS4, ZPS8 and ZPS13 were not included in the modelling

because the grinding time was less than 40 minutes, which will have had a

significant influence on the SEC (see Paper I for details). The coefficients in the

empirical model for the specific energy consumption in which the response is the

inverse value of SEC are shown in Fig. 14, as also are the main statistical

parameters of the empirical model. The high values for R2, R2(Adj.), Q2 and

reproducibility in Fig. 14 indicate that the model is statistically valid and the

results are reproducible. Since the validity of the model is above 0.25, the error in

the model is in the same range as pure error. Fig. 14 also shows that the classifier

speed and mill speed had an influence on the average specific energy

consumption in fine grinding, the classifier speed being the most significant

single factor (Fig. 14). The use of a higher classifier speed led to a higher specific

energy consumption, because the classifier wheel needed more power from the

motor to revolve faster. Another reason is that as the classifier speed increases the

particles need to be finer to get past the classifier and therefore their residence

time in the grinding and classifying section increases. This reduces the production

rate and increases the energy demand of the rotor in the grinding zone. The

specific energy consumption in air classifier milling can be reduced by using a

higher rotor speed (Figs. 14 and 15). The choice between the long and short

impact edge had a negligible effect, since their coefficients with error bars cross

the zero line (Fig. 14). A contour plot of the modelled specific energy

consumption as a function of the classifier and mill speeds in the speed ranges

studied here is shown in Fig. 15. Since the start-up of the process is inefficient in

terms of energy and the feeding speed was not optimised, the specific energy

consumption can be expected to be lower in an industrial process that is

continuous and well optimised.

65

Fig. 14. Scaled and centred coefficients of the operational parameters in the empirical

model for average specific energy consumption. The symbols for the coefficients are

explained in Fig. 12. Modified from Paper I, published by permission of Elsevier.

Fig. 15. Contour plot of the modelled average specific energy consumption in kWh t-1

when fine grinding lasts for 40 minutes. The effect of the impact angle is ignored.

Paper I, published by permission of Elsevier.

5.1.2 Fluidised bed opposed jet mill

As seen in Fig. 16, dried Norway spruce wood was ground in an opposed jet mill

to a median particle size of around 18 µm, giving a Δ* of around 5. The finest

powders produced had AR50 2.5 and ΔAR 3.6 or higher. The aspect ratios were

slightly higher than those for the finest powders produced with the air classifier

mill (see Table 8 and Fig. 16), and therefore the jet milled particles are smaller

and more elongated. This difference may, however, be due to the removal of

66

finest particles after pregrinding rather than to the different milling technique. To

provide an insight into the shapes of the particles, CCD photos of particles in jet

milled sample AFG-P8, with a median particle size of around 20 µm, are shown

in Fig. 17 (see Paper II for detailed information on the sample). The largest and

intermediate-sized wood particles, shown in Fig. 17a, can be described as

elongated, while the finest particles resemble roundish ones, as in air classifier

sample ZPS 17 (Figs 11c and 17c). When a grinding pressure of 10 bar was

applied at a classifier speed of 20,000 min-1 CrI was reduced from 47.5% to

around 44%, which means that opposed jet milling had only a minor influence on

CrI.

Fig. 16. Development of (a) d50, (b) ∆*, (c) AR50, (d) ∆AR and (e) CrI as a function of

classifier speed in opposed jet milling. Low GP stands for the grinding pressure

approximately 6 bar and High GP stands for the grinding pressure approximately 10

bar.

67

Fig. 17. Three CCD photos showing some of the a) largest b) intermediate-sized, c)

finest observable wood particles, and d) FESEM image showing some small

rectangular-shaped particles in opposed jet milled sample AFG-P8.

Development of wood powder properties

According to Fig. 16 the grinding pressure had substantial influence on particle

size and shape when the classifier speed was around 8000 min-1. This effect may

be due to the higher air flow rate in the case of a higher grinding pressure (see

Table 10), thus causing larger particles to pass through the classifier than with a

lower flow rate at a similar classifier speed. The air flow exerts a drag force on

the particles, driving them past the classifier wheel and into the product container

(see Benz et al. 1996). This drag force will depend on the particle size, with a

higher air flow rate increasing the particle size of the product. The increase in

aspect ratio is probably connected with the increased particle size. The influence

of the air flow rate on the particle properties became negligible at a classifier

speed of 20,000 min-1 (Fig. 16), even though a significant decrease in cellulose

crystallinity still took place at this classifier speed when a high grinding pressure

68

was used (Fig. 16). This means that a grinding pressure of approximately 10 bar

(overpressure) was enough to cause reduced cellulose crystallinity when d50 was

less than 70 µm (Fig. 16), whereas the use of a low grinding pressure caused a

very small or negligible decrease in CrI even though the wood was ground down

to a median particle size of 20 µm (Fig. 16).

Table 10. Averaged processing parameters, specific energy consumption and product

information on the jet milling experiments.

Experiment Classifier Grinding air Total air Feed or Product Energy

Speed

(min-1)

Pressure1

(bar)

Volume

(Nm3)

Compression

work2 (MJ)

Flow

rate

(Nm3 h-1)

Dry mass

mP,dry (g)

Median size3

(µm)

SECJET

(kJ g-1)

FEED - - - - - - 199 ± 6 -

AFG-P1 8040 5.92 7.23 1.42 86.18 25.92 52.1 54.71

AFG-P2 20,010 5.93 7.00 1.37 85.65 13.34 18.2 103.00

AFG-P3 8040 9.89 11.63 2.81 96.64 59.02 69.2 47.67

AFG-P4 19,980 9.88 11.38 2.75 94.70 29.71 20.0 92.66

AFG-P5 8010 5.93 7.20 1.41 86.37 40.05 48.2 32.28

AFG-P6 20,010 5.93 6.99 1.37 85.50 12.82 18.4 106.93

AFG-P7 8040 9.89 11.58 2.80 96.34 59.97 67.4 46.73

AFG-P8 20,010 9.83 11.87 2.87 95.98 32.34 19.9 88.59

1 Excess pressure compared with normal air pressure (NTP), 2 Compression work calculated using

Equation 2 with Mair = 28.96 g mol-1, pUC = 101,325 Pa, ρ = 1204 g m-3, T = 293.15 K and the averaged

pressure of the grinding air. 3 Measured in wood powder collected from the container after the classifier,

thus not including particles conducted to the filtering stage.

Specific energy consumption

Since most of the energy needed for processing in a jet mill is consumed in

compressing the pressurised air, it was the work needed for this purpose that was

considered here. The averaged processing parameters for the fine grinding of

dried Norway spruce wood in a fluidised bed opposed jet mill are listed in Table

10. Only dry air was considered when calculating the compression work. An

increase in classifier speed reduced the median particle size and the dry mass of

the product (Table 10), while an increase in grinding pressure increased the

median size and mass of the product, i.e. the production rate (Table 10). A small

decrease in the total average flow rate of air was observed on every occasion

when a higher classifier speed was used rather than a lower one, given that the

grinding pressure was kept constant (Table 10). On the other hand, when the

69

classifier speed was kept constant the total average flow rate of air, the volume of

the grinding air and the work needed for compressing the grinding air increased

on every occasion when the grinding pressure was increased (Table 10). The

specific energy consumption in milling was mainly affected by the classifier

speed (Table 10). At high classifier speeds the use of a higher grinding pressure

reduced the specific energy consumption but increased the median particle size of

the product (Table 10). As seen in Table 10, over 32 kJ g-1 (approx. 9 MWh t-1)

but less than 55 kJ g-1 (approx. 15 MWh t-1) was needed to obtain wood powders

with a median particle size between 48.2 µm and 69.2 µm from a feed with d50 =

199 µm (±6 µm), and over 88 kJ g-1 (approx. 25 MWh t-1) to obtain a median

particle size of around 20 µm (Table 10). Although the compression work

considers the ideal minimum work needed for compression, the fine grinding was

not fully optimised because the feeding rate was kept constant. The experiments

also involved an energy inefficient start-up period. Thus the energy consumption

may be even lower in an optimised continuous process.

5.1.3 Oscillatory ball mill

Dried Norway spruce wood was ground to a median particle size of less than 18

µm with a Δ* of around 5 with the oscillatory ball mill, and the finest powders

produced had AR50 and ΔAR significantly lower than 2.0 (Fig. 18). These values

are much lower than for the finest powders produced with the air classifier mill or

the opposed jet mill (see Table 8 and Fig. 16), and therefore oscillatory ball

milling can be said to produce particles that are smaller and rounder than in either

of the above. To provide an insight into the shapes of the particles, CCD photos of

the powder produced in the oscillatory ball mill, with a median particle size of

around 20 µm, are presented in Fig. 19. All of the wood particles shown in this

figure can be described as rectangular or roundish. FESEM images of some of the

finest particles obtained by this method are shown in Fig. 20. It was also possible

to produce wood powders by oscillatory ball milling that had a median particle

size over 100 µm with an AR50 of less than 1.8 and a ΔAR less than 2.4 (Fig. 18).

70

Fig. 18. Effect of grinding ball mass and oscillation frequency on the a) d50, b) ∆*, c)

AR50, d) ∆AR and e) temperature change as a function of the total available impact

energy of the grinding ball, ET. The mass of the feed in each setup was 1 g. Data

points containing an open space indicate the large end cut in the particle size

distribution. Modified from Paper III, published by permission of Elsevier.

71

Fig. 19. Three CCD photos showing some of the a) largest, b) intermediate-sized and

c) finest observable wood particles in an oscillatory ball milled sample with d50 = 21.5

µm, ∆* = 5.8, AR50 = 1.6 and ∆AR = 1.5.

Fig. 20. Eight different FESEM images a)-h) taken from one oscillatory ball milled

wood powder sample. Modified from Paper II, published by permission of Elsevier.

72

Development of wood powder properties

According to Fig. 18, the total kinetic energy of the grinding ball ET (see Paper

III) predicts the development in the size and shape properties of dried Norway

spruce wood in oscillatory ball milling upon a change in the mass of the grinding

ball or its oscillation frequency. It is therefore proposed that the mathematical

model evaluated here (Eq. 3) can be applied as a simplified mill based stressing

model to oscillatory ball milling in impact mode (for further information about

the impact mode, see Chen Y et al. 2004) for predicting milling behaviour when

changing the oscillation frequency, mass of the grinding ball (but not its volume)

and milling time. In this model the stressing energy is Ek (Eq. 4), the stress

number is SNe (Eq. 5) and stressing type is double impact (see Section 2.2.2). The

median particle size was inversely linearly dependent on ET in the ET range

considered before achieving a median particle size of 20 µm (Fig. 18a). The

mathematical model plotted in Fig. 18a is the same as Rittinger’s model, which is

commonly used in mineral processing to model the energy consumption when

considering particle sizes in the range 10–1000 µm (Lowrison 1974: 54–55).

Rittinger’s model has also been found to apply to the fine grinding behaviour of

α-lactose monohydrate in oscillatory ball milling in impact mode (Chen Y et al.

2004). When processing was continued after a median particle size of 20 µm had

been obtained, which occurred at ET = 1360 J according to the model derived

here, the resistance of the wood to further size reduction increased considerably

and its particle size properties (d50 and Δ*) flattened out at a certain level or even

increased (Figs. 18a and b). The smallest median particle size was found to be

between 13 µm and 20 µm for all the studied setups (Fig. 18a). In the case of

setups B–D the temperature change increased significantly when the particle size

properties evened out at a certain level (Figs. 18a, b and e). The rise in

temperature and sudden increase in comminution resistance indicate the presence

of viscoelastic behaviour in wood. This viscoelasticity is most likely the main

reason why the size properties of Norway spruce wood evened out at a certain

particle size range even though higher impact energies were used in processing

and pulverisation was continued after the minimum size had been attained. On the

other hand, AR50 and ΔAR did not flatten out at the same point as the median

particle size as a function of ET (Fig. 18a–d). This means that the particles do not

break due to fragmentation, because their volume-based particle size properties

remained unchanged. Presumably there may be a surface breakage mechanism,

attrition, involved. Attrition causes particles to become rounder but has an

73

insignificant effect on their volume-based size properties. In addition, the changes

in aspect ratio may be also due to deformation of the wood particles under

repeated stress that together with temperature change suggests viscoplastic

behaviour of Norway spruce wood once the minimum particle size has been

reached (Fig. 18). This observation suggests the presence of globular-shaped

particles that tend to agglomerate with prolonged milling times, as first described

for Norway spruce wood by Maurer and Fengel (1992). FESEM images of

particles found in an oscillatory ball milled sample are shown in Fig. 20, where

the roundish particles in Figs. 20g and h could be the globular shaped particles

that tend to agglomerate. Fig. 21 shows how varying the mass of the feed (setups

D–G) can affect the development of inverse values for d50 and Δ* as a function of

ET. According to Fig. 21a, median size follows an inverse linear trend as a

function of ET (Rittinger’s model) in the particle size range between 20 µm and

100 µm when the mass of the feed is between 1 g and 3 g. In the case of setup E

the trend is non-linear (Fig. 21a) and therefore cannot be modelled with

Rittinger’s model.

74

Fig. 21. Effect of changes in the amount of feed on the inverse of a) median size and b)

width of the particle size distribution as a function of total available impact energy of

the grinding ball, ET. The black line represents the model for setups B–D introduced in

Fig. 18. Data points containing an open space indicate the large end cut in the particle

size distribution. Modified from Paper III, published by permission of Elsevier.

Specific energy consumption

The specific energy consumption for different amounts of feed is shown in Fig.

22. Setup E is the least energy efficient when the desired median particle size is

less than 100 µm. The reason for the different comminution behaviour in the case

of setup E is probably the low amount of feed, which makes it possible for some

collisions to occur between the grinding ball and the grinding jar with little or no

wood present in the impact zone. According to Fig. 22, dried and screened

Norway spruce sawdust can be pulverised down to d50 = 100 µm with a minimum

specific energy consumption of 0.36 kJ g-1 (100 kWh t-1) and down to d50 = 20 µm

with 1.4 kJ g-1 (380 kWh t-1) when considering the total available impact energy

of the grinding ball in fine grinding. Kobayashi et al. (2008) has reported that 800

75

kWh t-1 was consumed for the whole process when Norway spruce wood chips

were pulverised from 22 mm down to 150 µm with a vibration mill. The specific

energy consumption calculated from the estimated total available impact energy

of the grinding ball (SECOSC) considers only the energy available for particle size

reduction, which is lower than the total electrical energy consumed by the

equipment, due to energy losses in driving the mill and in the electronics. On the

other hand, the theoretical energy needed for the generation of new surfaces can

be considered to be lower than the estimated SECOSC. Only in the ideal case,

where the total available impact energy in the process is applied for the

generation of new surfaces, should the estimated SECOSC be the same as the

theoretical energy needed for the generation of the new surface area. Since this

ideal situation will never exist in practical comminution, the estimated SECOSC

will always be higher than the theoretical energy needed for the generation of new

surfaces. Therefore the estimated SECOSC provides a better practical value for the

energy consumption than can be gained by calculating the theoretical energy for

the generation of new surfaces and can be regarded as a good target for the

optimisation of impact-based mills involving grinding media.

Fig. 22. Effect of the feed amount of feed on median particle size as a function of the

specific energy consumption. The data point containing an open space indicates a

large end cut in the size distribution. Modified from Paper III, published by permission

of Elsevier.

76

5.2 Comparison of impact milling techniques

Influence of double impacts on the development of wood properties in

milling

Figs. 23–25 show the development of Δ*, AR50, ΔAR and CrI in opposed jet

milling and oscillatory ball milling as a function of median particle size. In

oscillatory ball milling Δ* followed a linear trend as a function of d50 in fine

grinding from a d50 of 199 µm (± 6 µm) down to around 20 µm (Fig. 23), while in

the case of the jet milled samples Δ* decreased as a function of d50 and was

approximately identical to the particle size distribution of the oscillatory ball

milled samples when considering a median particle size around 20 µm (Fig. 23).

Median aspect ratio and ΔAR decreased more as a function of d50 in oscillatory

ball milling than in opposed jet milling (Fig. 24). The CCD photos shown in Fig.

26 provide an insight into the differences between the milling techniques in terms

of the aspect ratio distribution when d50 is around 20 µm. Opposed jet milling had

a significantly lower effect on the crystallinity index than did oscillatory ball

milling when the wood was ground down to a d50 around 20 µm (Fig. 25).

Fig. 23. Effect of oscillatory ball milling and opposed jet milling on the development of

the width of the particle size distribution ∆* as a function of the median particle size

d50. High GP and low GP mean average excess pressures of the grinding air of

approximately 10 bar and 6 bar, respectively. Paper II, published by permission of

Elsevier.

77

Fig. 24. Effect of oscillatory ball milling and opposed jet milling on the development of

a) the median aspect ratio AR50 and b) the width of the aspect ratio distribution ∆AR as

a function of the median particle size d50. High GP and low GP are as in Fig. 23.

Modified from Paper II, published by permission of Elsevier.

Fig. 25. Effect of the milling technique on development of the crystallinity index CrI as

a function of the median particle size d50. High GP and low GP are as in Fig. 23. Paper

II, published by permission of Elsevier.

78

Fig. 26. CCD photos showing some of the largest wood particles from a) the feed, b)

the product of opposed jet milling and c) the product of oscillatory ball milling. The jet

milled and oscillatory ball milled particles were of a median size of around 20 µm.

Modified from Paper II, published by permission of Elsevier.

The feed used in the jet milling and oscillatory ball milling experiments was

produced from a similarly treated raw material to that used in the air classifier

milling experiments. Also the feed was preground with the same air classifier mill

as was used in the air classifier milling experiments, but the finest particles were

removed from the feed prior to jet and oscillatory ball milling. Although a d50 of

around 23 µm was obtained in air classifier milling, CrI (%) could not be reduced

below 41% and the minimum AR50 and ΔAR were 2.3 and 3.3, respectively (Table

8). Thus the greatest differences in the properties of the wood powders were

observed between the media and non-media impact mill experiments (Figs. 23–25

and Table 8). In fine grinding of dried Norway spruce wood the presence of

double impacts favoured the production of rounder shaped particles with low

cellulose crystallinity by comparison with single impacts (Figs. 23–25 and Table

8).

Specific energy consumption

When comparing specific energy consumption it is important first to consider the

desired properties of the wood powder. When high cellulose crystallinity and

aspect ratio are required in dried Norway spruce the suitable impact-based fine

grinding techniques would be opposed jet mills and rotor impact mills. In the

present experiments it was possible to reduce Norway spruce wood from a

median particle size between 1 mm and 2 mm down to 23.8 µm with a specific

energy consumption of 15.5 kJ g-1 by air classifier milling (Table 8), whereas

opposed jet milling consumed over 32 kJ g-1 in producing wood powder with a

median particle size of 48.2 µm from a significantly smaller feed than that was

79

used in air classifier milling (Table 10). It should therefore be suggested that an

air classifier mill is more energy efficient technique than a fluidised bed opposed

jet mill for pulverising dried Norway spruce wood. In fact, the electric power

demand in the case of air classifier milling can be even lower in an industrial-

scale production plant. In the case of jet milling, the milling chamber in the

laboratory model has been built by Hosokawa Alpine with a similar architecture

to that of the larger-scale equipment of the AFG product family, so that the use of

compression energy for particle breakage in the laboratory equipment can be

considered to be very close to that observed in an industrial-scale milling plant

with a similar jet mill. It might be possible to optimise both types even further by

increasing the feed rate.

The specific energy consumption in the impact milling of dried Norway

spruce wood with a median particle size between 1 mm to 2 mm with media mills

can be as low as 0.36 kJ g-1 for a targeted d50 of 100 µm and 1.40 kJ g-1 for a d50

of 20 µm (Fig. 22). This specific energy consumption does not consider the

energy losses in driving the machine or any additional equipment, which are

naturally included in the specific energy consumption when considering the

electric energy demand of the mill. It therefore seems possible to obtain a lower

energy consumption with media mills than with an air classifier mill, but this

depends on the consumption of energy that is not directed at the comminution of

particles. This energy is significantly affected by the architecture of a media mill.

Kobayashi et al. (2012), studying the significance of the architecture of vibration

mills for energy efficient wood powder production, concluded that a multiple tube

vibration mill is more energy efficient than a single tube vibration mill and

estimated the specific energy consumption in the fine grinding of wood to be 1.0

kJ g-1 when using a multiple tube vibration mill. This figure is very close to the

values obtained in the present oscillatory ball milling experiments (Fig. 22), so

that media milling appears to be well suited for wood pulverisation.

80

5.3 Effects of pretreatments and milling conditions on the properties of Norway spruce wood in impact-based media

milling

5.3.1 Drying and cryogenic grinding conditions

Development of d50, Δ*, AR50 and ΔAR under ambient and cryogenic grinding

conditions is shown as a function of moisture content in Fig. 27, which suggests

that while moisture content has a significant influence on wood powder properties

under ambient grinding conditions, these properties are less dependent on

moisture content under cryogenic conditions (Fig. 27). When comparing grinding

conditions given the same moisture content (Fig 27a), it can be seen that a smaller

median particle size was obtainable under cryogenic conditions. When

considering the moisture contents of 1%, 27% and 50% in Fig 27a, increasing the

grinding time from 10 min to 30 min under ambient conditions did not reduce the

median particle size any further than had been obtained by cryogenic grinding for

10 min. It therefore seems that the use of cryogenic grinding conditions rather

than ambient conditions provides better energy efficiency in the fine grinding of

Norway spruce wood. Wood powder with a median particle size of less than 13

µm was obtained only under cryogenic conditions (Fig. 27a), when the finest

wood powder had a median size of 7.4, Δ* = 2.04, AR50 = 1.67 and ΔAR = 1.78. It

has not been possible to produce wood powders with a median particle size

significantly less than 13 µm in any other way (Figs 18, 27 and Tables 8 and 10).

It was possible to produce fine powders with a median particle size below 100 µm

an elongated particle shape (AR50 ≈ 3.0 and ΔAR ≈ 6.0) under ambient

conditions by pulverising Norway spruce wood with a moisture content of 44%

(Fig. 27). The CCD photos and FESEM images of various wood powders in Figs.

28 and 29 illustrate the differences in particle shape between the powders with

different AR50 and ΔAR values obtained by oscillatory ball milling. Vibration mills

are generally regarded in the literature as representing a technique for producing

roundish particles from wood (see Section 2.3.2), but milling moist wood under

ambient conditions does in effect produce elongated particles with AR50 and ΔAR

values very close to those produced with a rotor impact mill and jet mill (Figs. 24,

25, 27 and Table 8). The reason suggested for this is softening of the

hemicelluloses due to the plasticising effect of water at room temperature, which

favours cleavage in the primary and secondary cell wall layers (Jääskeläinen &

Sundqvist 2007: 127–130). Under cryogenic grinding conditions the temperature

81

is too low for any softening of the hemicelluloses and thus the particle shapes are

similar to those obtained from dried wood. In fact, the oscillatory ball milling

experiments with dried Norway spruce wood yielded some particles with a scaly

surface (see Figs. 20h and g), as were also found when wood powders were

milled under cryogenic conditions (Fig. 30).

Fig. 27. The differences between ambient and cryogenic grinding conditions in a) d50,

b) ∆*, c) AR50 and d) ∆AR as a function of moisture content in oscillatory ball milling.

The black data points represent experiments with replicates and the empty data points

indicate a large-end cut in the particle size distribution. Modified from Paper IV,

published by permission of De Gruyter.

82

Fig. 28. CCD photos from three wood powder samples with different aspect ratio

properties. In the figure there is shown three photos from each sample to illustrate

differences in different particle size ranges. Paper IV, published by permission of De

Gruyter.

Fig. 29. FESEM images of two wood powder samples. Paper IV, published by

permission of De Gruyter.

Fig. 30. FESEM images of a wood powder sample milled under cryogenic conditions.

83

The crystallinity indices of oscillatory ball milled samples pulverised under

ambient and cryogenic conditions for different periods of time are listed in Table

11. Segal et al. (1959) developed the method for estimating the relative degree of

cellulose crystallinity in cotton celluloses and therefore one explanation for the

negative CrI values shown in Table 11 can be due the differences in the X-ray

diffraction patterns between cotton cellulose and Norway spruce wood. The CrI

of the fresh feed was 39% and that of the oven-dried feed was also 39%, so there

were no observable changes in CrI due to oven-drying of the wood. It may be

seen from Table 11 that oscillatory ball milling of Norway spruce wood for at

least 10 min lowered the CrI by at least 4%-units. The fine grinding of wood a

with moisture content over 27%, which is slightly above the fibre saturation point

of Norway spruce wood (see Section 2.1.2), under ambient conditions had little

effect on CrI (Table 11), but when considering the milling of feeds with moisture

contents between 1% and 27% under ambient conditions, a lower CrI was

obtained with the samples that had a lower initial moisture content when milling

for 10 min (Table 11). In the case of the feed with a moisture content of 1%, a

longer grinding time reduced the CrI even more (Table 11), but when wood with a

moisture content between 8% and 50% was milled for 10 min under cryogenic

conditions the crystallinity indices varied between 24% and 30% (Table 11).

When samples with an initial moisture content of 1% were milled under

cryogenic conditions for 10 min the CrI was significantly lower than in the other

samples milled under similar conditions (Table 11). Under cryogenic conditions a

longer grinding time reduced CrI when the moisture content was between 1% and

50% (Table 11). Thus cryogenic grinding offers a good way of reducing the

cellulose crystallinity of wood with a moisture content well above the fibre

saturation point, which seems not to be possible under ambient conditions.

Consequently, it seems that the presence of bound, and more especially unbound,

water in Norway spruce wood plays a crucial role in protecting the crystalline

cellulose from destruction when exposed to double impacts. This protective

capability seems to be lost when the water is in a solid state during milling.

84

Table 11. Crystallinity indices of oscillatory ball milled wood samples. Modified from

Paper IV, published by permission of De Gruyter.

Conditions Milling time

(min)

CrI (%) when the moisture content of the feed was

50% 27% 17% 8% 1%

01 39 39

Ambient 10 35 32 20 10 -5

30 34 33 -21

Cryogenic 10 24 21 30 26 10

30 10 -1 -9

1 Milling time of 0 min refers to the feed

Fine grinding of moist Norway spruce wood was also tested with a high-speed

rotor impact mill and a fluidised bed opposed jet mill, but it was found that the

wood agglomerated in front of the feeding screw until it totally obstructed the

feed. Agglomeration was also observed in oscillatory ball milling, where the

samples of wood with a moisture content above 44% formed pancake-shaped

agglomerations at the end wall of the grinding chamber during milling. It was

possible to disperse these agglomerates, however, when preparing diluted samples

for the analyses, but this had to be helped along mechanically with a glass stick

prior to agitation with a magnetic stirrer. Agglomeration of moist wood in dry

grinding have also been reported by Gravelsins (1998) in Szego milling

experiments, where it was observed that moist wood particles of size below 150

µm agglomerated to form larger particles.

5.3.2 Mild torrefaction

The correlations between Δ* and d50 for dried and torrefied samples ground for

different periods of time are shown in Fig. 31. Mild torrefaction had an

insignificant influence on Δ* as a function of d50, and when the mass loss (ML)

was less than 0.2% it had no influence on d50 as a function of ET. Torrefaction

involving an ML over 1.4%, however, reduced the ET needed for obtaining a given

median particle size over 17.4 µm (± 0.2 µm) (Fig. 32). Specific energy

consumptions for the setups REF, TOR B and TOR C are listed in Table 12,

which indicates that a smaller median particle size was obtained with a similar

SECOSC when torrefaction pretreatment was more intensive and the median

particle size was over 17.4 µm. In the TOR C experiments where d50 was similar

or smaller than in the REF experiments the specific energy consumption

decreased by approximately 21–33% following TOR C pretreatment when the

85

targeted d50 was between 18.7 µm (± 0.5 µm) and 79 µm (± 3 µm) (Table 12), but

once ET = 3422 J was reached the differences in d50 between the REF, TOR B and

TOR C experiments became insignificant. When ET increased above 3422 J no

significant decreases in d50 were observed (Fig. 32). There is therefore a certain

particle size range in which torrefaction pretreatment cannot improve the energy

efficiency, and this coincides with the smallest particle size range obtained with

dried samples, i.e. 13–20 µm (see Section 5.1.3). Also the minimum median

particle size obtained with torrefied samples, 12.9 µm (± 0.6 µm) (Table 12), is

not significantly lower than the minimum median particle size of 13 µm reported

in Section 5.1.3. The reason for this is thought to be agglomeration of the globular

shaped particles and increased viscoelasticity of the wood when milling down to

d50 values below 20 µm, as reported and discussed in Section 5.1.3. Mild

torrefaction did not influence the median aspect ratio when ET was over 1000 J

(Fig. 33a), but when it was below 1000 J the differences in AR50 between the REF

and torrefied samples were fairly small (Fig. 33a). According to Fig. 33b, the

differences between the REF and torrefied samples were insignificant when

considering ΔAR as a function of d50. Mild torrefaction with an ML up to 2.8% had

a negligible effect on CrI when considering between REF and TOR C results

obtained with similar ET values or milling times (Table 13).

Media milling is considered a good pretreatment of wood for enzymatic

hydrolysis because of its ability to destroy the cell walls and reduce cellulose

crystallinity (see Section 2.3.2). Torrefaction provides a good way of lowering the

energy consumption required for obtaining a certain particle size, although special

care has to be taken not to process the material for longer than is needed to obtain

the limiting particle size, because the energy benefits due to torrefaction can be

lost.

86

Fig. 31. Effect of mild torrefaction on the median particle size d50 as a function of ∆*.

Modified from Paper V, published by permission of De Gruyter.

Fig. 32. Effect of mild torrefaction on the median particle size d50 as a function of the

total available impact energy of the grinding ball, ET. Modified from Paper V, published

by permission of De Gruyter.

Table 12. Specific energy consumption and median particle sizes in the REF, TOR B

and TOR C experiments. Paper V, published by permission of De Gruyter.

SECOSC (kJ g-1) d50 (µm)

REF TOR B TOR C

0.26 - 115 ± 6 79 ± 3

0.34 94.0 ± 1.1 53 ± 8 41 ± 5

0.51 44 ± 3 35 ± 3 32.6 ± 0.6

0.68 30 ± 3 26.6 ± 1.4 24.6 ± 0.7

0.86 23.3 ± 1.1 19.9 ± 0.8 18.7 ± 0.5

1.20 19.8 ± 1.0 17.0 ± 0.2 17.4 ± 0.2

1.71 13.8 ± 0.4 13.6 ± 0.4 12.9 ± 0.6

2.57 16.0 ± 0.0 14.5 ± 1.2 14 ± 2

87

Fig. 33. Effect of mild torrefaction on a) AR50 and b) ∆AR as a function of the total

available impact energy of the grinding ball. Modified from Paper V, published by

permission of De Gruyter.

Table 13. Comparison of CrI and d50 between REF and TOR C pretreated samples when

pulverised for various lengths of time. Paper V, published by permission of De

Gruyter.

Milling time (min) ET (J) REF (ML = 0.0%) TOR C (ML = 2.8%)

CrI (%) d50 (µm) CrI (%) d50 (µm)

0 0 44 ± 2 < 4000a 42.3 ± 0.3 < 4000a

2 684 28 ± 2 94.0 ± 1.1 27 ± 3 41 ± 5

3 1027 17 ± 3 44 ± 3 15.5 ± 1.2 32.6 ± 0.6

5 1711 11 ± 4 23.3 ± 1.1 9 ± 4 18.7 ± 0.5

a Feed was screened with a 4 mm sieve

88

89

6 Conclusions

Fine grinding of dry Norway spruce wood was studied using three types of fine

grinding mills in which particle breakage takes place due powerful impacts: two

types of non-media mill, known as the air classifier mill and the fluidised bed

opposed jet mill, and one vibration-type media mill known as the oscillatory ball

mill. These milling techniques proved capable of reducing the median particle

size of dried Norway spruce wood to below 25 µm.

The impact events occurring in the media mill were capable of producing

very fine wood powders from dried Norway spruce that had a lower cellulose

crystallinity and rounder shaped particles with a more uniform shape distribution

than in dried Norway spruce wood pulverised to a similar particle size range by

impact events occurring in non-media mills. In the case of the air classifier mill

and fluidised bed opposed jet mill the speed of the classifier had the strongest

effect on the physical properties of the resulting wood powders, whereas with the

oscillatory ball mill the amount of feed, the weight of the grinding ball and the

oscillation frequency had a significant influence on the physical properties of the

wood. When Norway spruce wood was ground to a median particle size of less

than 20 µm the use of more energy intensive impacts resulted in agglomeration

and increased the temperature of the wood during further processing. Although

the prolongation of media milling led to no further decrease in particle size, the

wood particles continued to become rounder and more homogeneous in shape.

The classifier speed had the greatest effect on the energy consumption in air

classifier and opposed jet milling, with a higher rotor speed in the air classifier

mill and a higher grinding pressure in the opposed jet mill yielding an increase in

the production rate when all the other operational parameters were kept constant.

In the air classifier mill the specific electrical energy consumption for attaining a

certain median particle size could be lowered by increasing the rotor speed, while

in the opposed jet mill an increase in grinding pressure reduced the specific

energy consumption but increased the eventual fineness when milling to a median

particle size around 20 µm. The energy consumption in oscillatory ball milling

was affected by the amount of feed, in that a very low feed volume resulted in

poor energy efficiency and an alteration in particle size reduction behaviour

during milling. In the fluidised opposed jet mill the energy requirement for

compression of the grinding air was over 32 kJ g-1 (approx. 9000 kWh t-1) when

pulverising dried Norway spruce wood with a median particle size of 199 µm (± 6

µm) down to a size between 48.2 µm and 69.2 µm. In the air classifier mill dried

90

Norway spruce wood with a median particle size between 1 mm and 2 mm was

pulverised down to 23.8 µm with a specific energy consumption of 15.5 kJ g-1

(4310 kWh t-1). Further optimisation would be possible in both mill types. A

practical estimate for the minimum specific energy consumption in the impact-

based media milling of dried Norway spruce wood, ignoring energy losses in

driving the machine and additional equipment, would be that the median particle

size could be reduced from 1–2 mm down to 100 µm with 0.36 kJ g-1 (100 kWh t-

1) and down to 20 µm with 1.4 kJ g-1 (380 kWh t-1). The total energy consumption

in media mills is higher than this, however, due to the consumption of energy that

is not directed at the particles, a source of energy loss that is significantly

influenced by the architecture of the mill.

Cryogenic grinding conditions can be used in impact-based media milling to

obtain different powders from Norway spruce wood from those produced under

ambient grinding conditions. Also, the energy efficiency of fine grinding can be

enhanced by choosing cryogenic instead of ambient grinding conditions. Wood

powders with a median particle size significantly lower than 13 µm were obtained

when milling under cryogenic conditions, whereas this was not possible under

ambient grinding conditions, and the reduction of cellulose crystallinity in an

impact-based media mill exhibits a different behaviour pattern as a function of

milling time under cryogenic in comparison to ambient conditions. On the other

hand, moisture content has a stronger influence on the size and shape of the

resulting particles when grinding is performed under ambient conditions. Thermal

pretreatment by torrefaction can reduce the energy consumption required in

impact-based media mills for attaining a certain particle size, provided that the

targeted median particle size is not below 17.4 µm (± 0.2 µm). The specific

energy consumption was typically reduced by more than 21% when the mass loss

due to torrefaction was 2.8%. The shape of the particles and their cellulose

crystallinity are not significantly affected by torrefaction pretreatment when

considering similar levels of energy consumption and torrefied samples having a

mass loss due to torrefaction of 2.8% or less.

The present findings suggest that the impact-based fine grinding of dried

wood should be divided into media and non-media milling processes, depending

on the final product. Media milling involves double impacts, which favour

lowered cellulose crystallinity and rounder shaped particles by comparison with

non-media mills. On the other hand, media mills can be used to obtain similar

kinds of wood powders as produced with non-media mills, but then the raw

material has to be moist rather than dry, which can cause different problems in

91

processing. These processing problems and the energy consumption required for

drying could be avoided by preferring cryogenic grinding conditions, but this also

has a specific effect on the properties of the product. The fine grinding of Norway

spruce wood with a median particle size in the millimetre region down to around

20 µm should be possible with an energy consumption of several hundred

kilowatt-hours per tonne, but further optimisation work would be needed to

achieve this specific energy consumption on an industrial scale. The fine grinding

of wood to a median particle size below 20 µm should be considered with

extreme caution, because the resistance of Norway spruce wood, for example, to

particle size reduction increases quite suddenly in this particle size region and can

lead to the formation of agglomerations.

92

93

References

Acharya B, Sule I & Dutta A (2012) A review on advances of torrefaction technologies for biomass processing. Biomass Convers Biorefin 2(4): 349–369.

Adams TN, Raymond DR & Schmid C (1988) Optimization of a swirl burner for pulverized-wood fuels. Tappi J 71(5): 91–96.

Adi H, Larson I & Stewart P (2007) Use of milling and wet sieving to produce narrow particle size distributions of lactose monohydrate in the sub-sieve range. Powder Technol 179(1–2): 95–99.

Agarwal UP, Zhu JY & Ralph SA (2013) Enzymatic hydrolysis of loblolly pine: effects of cellulose crystallinity and delignification. Holzforschung 67(4): 371–377.

Almeida G, Brito JO & Perré P (2010) Alterations in energy properties of eucalyptus wood and bark subjected to torrefaction: the potential of mass loss as a synthetic indicator. Bioresour Technol 101(24): 9778–9784.

Almendros M, Bonnefoy O, Govin A, Nastoll W, Sanz E, Anreux R & Guyonnet R (2011) Changes on wood powder morphology and flowability due to thermal pretreatment. In: Fatah N (ed) Proceedings du XIII° Congrès de la SFGP 2011 Société Française de Génie des Procédés: Des procédés au service du produit au coeur de l'Europe. Société Française de Génie des Procédés. URI: http://hal.archives-ouvertes.fr/docs/00/65/82/52/PDF/MA-SFGP-207.pdf. Cited 2013/03/21.

Andersson S, Serimaa R, Paakkari T, Saranpää P & Pesonen E (2003) Crystallinity of wood and the size of cellulose crystallites in Norway spruce (Picea abies). J Wood Sci 49(6): 531–537.

Andersson S, Wikberg H, Pesonen E, Maunu S-L & Serimaa R (2004) Studies of crystallinity of Scots pine and Norway spruce cellulose. Trees 18(3): 346–353.

Arias B, Pevida C, Fermoso J, Plaza MG, Rubiera F & Pis JJ (2008) Influence of torrefaction on the grindability and reactivity of woody biomass. Fuel Process Technol 89(2): 169–175.

Ashby MF, Easterling KE, Harrysson R & Maiti SK (1985) The fracture and toughness of woods. Proc R Soc London, Ser A 398: 261–280.

ASM International (2003) Characterization and Failure Analysis of Plastics. Materials Park, OH: ASM International, eBook Collection (EBSCOhost), EBSCOhost. Cited 2014/04/25.

Barakat A, de Vries H & Rouau X (2013) Dry fractionation process as an important step in current and future lignocellulose biorefineries: a review. Bioresour Technol 134: 362–373.

Benabou L (2008) Kink band formation in wood species under compressive loading. Exp Mech 48(5): 647–656.

Benabou L (2010) Predictions of compressive strength and kink band orientation for wood species. Mech Mater 42(3): 335–343.

Bengtsson C (2000) Stiffness of spruce wood – influence of moisture conditions. Holz Roh- Werkst 58(5): 344–352.

94

Benz M, Herold H, Ulfik B (1996) Performance of a fluidized bed jet mill as a function of operating parameters. Int J of Miner Process 44–45: 507–519.

Berg J-E (2001) Effect of impact velocity on the fracture of wood as related to the mechanical pulping process. Wood Sci Technol 35(4): 343–351.

Bernotat S & Schönert K (2000) Size reduction. In: Ullmann’s Encyclopedia of Industrial Chemistry. Wiley-VCH Verlag GmbH & Co. KGaA. URI: http://dx.doi.org/10.1002/14356007.b02_05. Cited 2013/05/06.

Berry SL & Roderick ML (2005) Plant-water relations and the fibre saturation point. New Phytol 168(1): 25–37.

Bertraud F & Holmbom B (2004) Chemical composition of earlywood and latewood in Norway spruce heartwood, sapwood and transition zone wood. Wood Sci Technol 38(4): 245–256.

Blomberg J, Persson B & Blomberg A (2005) Effects of semi-isostatic densification of wood on the variation in strength properties with density. Wood Sci Technol 39(5): 339–350.

Boatright SW, Garrett GG (1983) The effect of microstructure and stress state on the fracture behaviour of wood. J Mat Sci 18(7): 2181–2199.

Bodner J, Schlag MG & Grüll G (1997) Fracture initiation and progress in wood specimens stressed in tension, part I: clear wood specimens stressed parallel to the grain. Holzforschung 51(5): 479–484.

Bodner J, Schlag MG & Grüll G (1998) Fracture initiation and progress in wood specimens stressed in tension, part III: clear wood specimens with various slopes of grain. Holzforshung 52(1): 95–101.

Booker RE & Sell J (1998) The nanostructure of the cell wall of softwoods and its functions in a living tree. Holz Roh- Werkst 56(1): 1–8.

Bosshard HH (1974) Holzkunde, band 2: zur biologie, physik und chemie des holzes. Basel and Stuttgart, Birkhäuser-Verlag.

Bouafif H, Koubaa A, Perré P & Cloutier A (2009) Effects of fiber characteristics on the physical and mechanical properties of wood plastic composites. Composites, Part A 40(12): 1975–1981.

Broström M, Nordin A, Pommer L, Branca C & Di Blasi C (2012) Influence of torrefaction on the devolatilization and oxidation kinetics of wood. J Anal Appl Pyrolysis 96: 100–109.

Brändström J (2002) Morphology of Norway spruce tracheids with emphasis on cell wall organisation. PhD thesis. Swedish University of Agricultural Sciences, Department of Wood Science.

Buksnowitz C, Teischinger A, Grabner M, Müller U & Mahn L (2010) Tracheid length in Norway spruce (Picea abies [L.] Karst.) analysis of three databases regarding tree age, cambial age, tree height, inter-annual variation, radial distance to pith and log qualities. Wood Res (Bratislava, Slovakia) 55(4): 1–14.

Burgert I, Frühmann K, Keckes J, Fratzl P & Stanzl-Tschegg SE (2003) Microtensile testing of wood fibers combined with video extensometry for efficient strain detection. Holzforschung 57(6): 661–664.

95

Burgert I, Eder M, Frühmann K, Keckes J, Fratzl P & Stanzl-Tschegg S (2005) Properties of chemically and mechanically isolated fibres of spruce (Picea abies [L.] Karst.). Part 3: Mechanical characterisation. Holzforschung 59(3): 354–357.

Campbell FC (2012) Fatigue and fracture – understanding the basics, Materials park, Ohio: ASM International, eBook Collection (EBSCOhost), EBSCOhost. Cited 2014/05/02.

Chamayou A & Dodds JA (2007) Air jet milling. In: Salman AD, Ghadiri M & Hounslow MJ (eds) Handbook of powder technology, volume 12, particle breakage. 1st edition, Amsterdam, Boston, Heidelberg, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokio, Elsevier: 421–435.

Chandra RP, Bura R, Mabee WE, Berlin A, Pan X & Saddler JN (2007) Substrate pretreatment: the key to effective enzymatic hydrolysis of lignocellulosics. Adv Biochem Eng/Biotechnol 108: 67–93.

Chang H, Cowling EB, Brown W, Adler E & Miksche G (1975) Comparative studies on cellulolytic enzyme lignin and milled wood lignin of sweetgum and spruce. Holzforschung 29(5): 153–159.

Chang VS & Holtzapple MT (2000) Fundamental factors affecting biomass enzymatic reactivity. Appl Biochem Biotechnol 84–86(1–9): 5–37.

Chen HC, Chen TY & Hsu CH (2006) Effects of wood particle size and mixing ratios of HDPE on the properties of the composites. Holz Roh- Werkst 64(3): 172–177.

Chen Y, Ding Y, Papadopoulos DG & Ghadiri M (2004) Energy-based analysis of milling α-lactose monohydrate. J Pharm Sci 93(4): 886–895.

Chen W-H, Hsu H-C, Lu K-M, Lee W-J & Lin T-C (2011) Thermal pretreatment of wood (Lauan) block by torrefaction and its influence on the properties of the biomass. Energy 36(5): 3012–3021.

Clorius CO, Pedersen MU, Hoffmeyer P & Damkilde L (2000) Compressive fatigue of wood. Wood Sci Technol 34(1): 21–37.

Côté WA & Hanna RB (1983) Ultrastructural characteristics of wood fracture surfaces. Wood Fiber Sci 15(2): 135–163.

Da Silva ASA, Inoue H, Endo I, Yano S & Bon EPS (2010) Milling pretreatment of sugarcane bagasse and straw for enzymatic hydrolysis and ethanol fermentation. Bioresour Technol 101(19): 7402–7409.

Dahl KB (2009) Mechanical properties of clear wood from Norway spruce. PhD thesis. Norwegian University of Science and Technology, Department of Structural Engineering.

Dasari RK & Berson RE (2007) The effect of particle size on hydrolysis reaction rates and rheological properties in cellulosic slurries. Appl Biochem Biotechnol 137–140(1–12): 289–299.

De Magistris F & Salmén L (2005) Deformation of wet wood under combined shear and compression. Wood Sci Technol 39(6): 460–471.

De Vegt OM (2007) Jet milling from a particle perspective: Predicting particle fracture based on mechanical material properties. PhD thesis. University of Groningen, Faculty of Mathematics and Natural Sciences.

96

Derome D, Zillig W & Carmeliet J (2012) Variation of measured cross-sectional cell dimensions and calculated water vapor permeability across a single growth ring of spruce wood. Wood Sci Technol 46(5): 827–840.

Dill-Langer G, Lütze S & Aicher S (2002) Microfracture in wood monitored by confocal laser scanning microscopy. Wood Sci Tech 36(6): 487–499.

Dumail J-F & Salmén (1996) Compression behaviour of spruce wood under large plastic deformations. Nord Pulp Pap Res J 11(4): 239–242.

Eder M, Stanzl-Tschegg S & Burgert I (2008) The fracture behaviour of single wood fibres is governed by geometrical constraints: in situ ESEM studies on three fibre types. Wood Sci Technol 42(8): 679–689.

Eder M, Jungnikl K & Burgert I (2009) A close-up view of wood structure and properties across a growth ring of Norway spruce (Picea abies [L] Karst.). Trees 23(1): 79–84.

Eder M, Arnould O, Dunlop JWC, Hornatowska J & Salmén L (2013) Experimental micromechanical characterisation of wood cell walls. Wood Sci Technol 47(1): 163–182.

Eisenacher G, Scheidemann R, Neumann M, Droste B & Völzke H (2013) Dynamic crushing characteristics of spruce wood under large deformations. Wood Sci Technol 47(2): 369–380.

Ennis BJ, Witt W, Weinekötter R, Sphar D, Gommeran E, Allen T, Raymus GJ & Litster JD (2008) Solid-solid operations and processing. In: Green DW & Perry RH (eds) Perry’s chemical engineers’ handbook. 8th edition, McGraw-Hill, New York: 21-61–21-68.

Eskelinen E, Hu SH & Marton R (1982) Wood mechanics and mechanical pulping. Appita 36(1): 32–38.

Esteban LS & Carrasco JE (2006) Evaluation of different strategies for pulverization of forest biomasses. Powder Technol 166(3): 139–151.

Esteves BM & Pereira HM (2009) Wood modification by heat treatment: a review. BioResources 4(1): 370–404.

Fahlén J (2005) The cell wall ultrastructure of wood fibres – effects of the chemical pulp fibre line. PhD thesis. The Royal Institute of Technology, Department of Fibre and Polymer Technology.

Fengel D & Stoll M (1973) Über die veränderungen des zellquerschnitts, der dicke der zellwand und der wandschichten von fichtenholz-tracheiden innerhalb eines jahrringes. Holzforschung 27(1): 1–7.

Fengel D & Wegener G (1989) Wood: chemistry, ultrastructure and reactions. De Gruyter, Berlin, New York.

Frühmann K, Reiteter A, Tschegg EK & Stanzl-Tschegg S (2002) Fracture characteristics of wood under mode I, mode II and mode III loading. Philos Mag A 82(17–18): 3289–3298.

Frühmann K, Burgert I, Stanzl-Tschegg SE & Tschegg EK (2003a) Mode I fracture behaviour on the growth ring scale and cellular level of spruce (Picea abies [L.] Karst.) and beech (Fagus sylvatica L.) loaded in the TR crack propagation system. Holzforschung 57(6): 653–660.

97

Frühmann K, Burgert I & Stanzl-Tschegg SE (2003b) Detection of the fracture path under tensile loads through in situ tests in an ESEM chamber. Holzforschung 57(3): 326–332.

Fukazawa K, Revol J-F, Jurasek J & Goring DAI (1982) Relationship between ball milling and the susceptibility of wood to digestion by cellulase. Wood Sci Technol 16(4): 279–285.

Gadoche L & López GD (1989) Assessment of size reduction as a preliminary step in the production of ethanol from lignocellulosic wastes. Biol Wastes 30(2): 153–157.

Gerhards CC (1982) Effect of moisture content and temperature on the mechanical properties of wood: an analysis of immediate effects. Wood Fiber Sci 14(1): 4–34.

Gibson LJ & Ashby MF (1988) Cellular solids: Structure & properties. Oxford, New York, Beijing, Frankfurt, São Paulo, Sydney, Tokyo, Toronto, Pergamon Press.

Gindl W & Teischinger A (2002) Axial compression strength of Norway spruce related to structural variability and lignin content. Composites, Part A 33(12): 1623–1628.

Gravelsins RJ (1998) Studies of grinding of wood and bark-wood mixtures with the Szego mill. PhD thesis, University of Toronto, Department of Chemical Engineering and Applied Chemistry.

Guerra A, Filpponen I, Lucia LA, Saquing C, Baumberger S & Argyropoulos DS (2006) Toward a better understanding of the lignin isolation process from wood. J Agric Food Chem 54(16): 5939–5947.

Havimo M, Rikala J, Sirviö J & Sipi M (2008) Distributions of tracheid cross-sectional dimensions in different parts of Norway spruce stems. Silva Fennica 42(1): 89–99.

Havimo M (2009) A literature-based study on the loss tangent of wood in connection with mechanical pulping. Wood Sci Technol 43(7–8): 627–642.

Havimo M (2010) Variation in tracheid cross-sectional dimensions and wood viscoelasticity - extent and control methods. PhD thesis. University of Helsinki, Department of Forest Sciences.

Hejnowicz A (2007) Anatomy, embryology and karyology. Bud structure and shoot development. In: Tjoelker MG, Boratyński A & Bugała W (eds) Biology and ecology of Norway spruce, Forestry sciences 78. Dordrecht, Springer: 49–70.

Hendriks ATWM & Zeeman G (2009) Pretreatments to enhance the digestibility of lignocellulosic biomass. Bioresour Technol 100(1): 10–18.

Herman M, Dutilleul P & Avella-Shaw T (1998) Intra-ring and inter-ring variations of tracheid length in fast-grown versus slow-grown Norway spruces (Picea abies). IAWA journal 19(1): 3–23.

Himmel M, Tucker M, Baker J, Rivard C, Oh K & Grohmann K (1985) Comminution of biomass: hammer and knife mills. Biotechnol Bioeng Symp 15: 39–58.

Hon DN-S & Glasser WG (1979) The effect of mechanical action on wood and fiber components. Tappi 62(10): 107–110.

Hon DN-S (1987) Mechanochemistry of lignocellulosic materials. In: Grassie N (ed) Developments in polymer degradation-7. London & New York, Elsevier Applied Science Publishers Ltd: 165–191.

98

Ikeda T, Holtman K, Kadla JF, Chang H & Jameel H (2002) Studies on the effect of ball milling on lignin structure using a modified DFRC method. J Agric Food Chem 50: 129–135.

Jakob HF, Fengel D, Tschegg SE & Fratzl P (1995) The elementary cellulose fibril in Picea abies: comparison of transmission electron microscopy, small-angle x-ray scattering, and wide-angle x-ray scattering results. Macromolecules (Washington, DC, US) 28(26): 8782–8787.

Jakob HF, Tschegg SE & Fratzl P (1996) Hydration dependence of the wood-cell wall structure in Picea abies. A small-angle x-ray scattering study. Macromolecules (Washington, DC, US) 29(26): 8435–8440.

Jankovic A (2003) Variables affecting the fine grinding of minerals using stirred mills. Miner Eng 16(4): 337–345.

Jenkins C & Khanna S (2005) Mechanics of material: A modern integration of mechanics and materials in structural design. Burlington, Mass: Elsevier Academic Press, eBook Collection (EBSCOhost), EBSCOhost. Cited 2014/04/25.

Jernkvist LO (2001) Fracture of wood under mixed mode loading II. Experimental investigation of Picea abies. Eng Fract Mech 68(5): 565–576.

Jiang J, Lu J, Zhou Y, Zhao Y & Zhao L (2014) Compression strength and modulus of elasticity parallel to the grain of oak wood at ultra-low and high temperatures. BioResources 9(2): 3571–3579.

Jääskeläinen A-S & Sundqvist H (2007) Puun rakenne ja kemia. Helsinki, Otatieto. Kenealy WR & Jeffries TW (2003) Enzyme processes for pulp and paper: a review of

recent developments. In: Goodell B, Nicholas DD & Schultz TP (eds) Wood deterioration and preservation: advances in our changing world. Washington DC, American Chemical Society: 210–239.

Keunecke D, Stanzl-Tschegg S & Niemz P (2007) Fracture characterisation of yew (Taxus baccata L.) and spruce (Picea abies [L.] Karst.) in the radial-tangential and tangential-radial crack propagation system by a micro wedge splitting test. Holzforschung 61(5): 582–588.

Khalil HPSA, Shahnaz SBS, Ratnam MM, Ahmad F & Fuaad NAN (2006) Recycle polypropylene (RPP) - wood saw dust (WSD) composites - Part 1: the effect of different filler size and filler loading on mechanical and water absorption properties. J Reinf Plast Compos 25(12): 1291–1303.

Kobayashi N, Sato T, Okada N, Kobayashi J, Hatano S, Itaya Y & Mori S (2007) Evaluation of wood powder property pulverized by a vibration mill. J Jpn Inst Energy 86(9): 730–735.

Kobayashi N, Guilin P, Kobayashi J, Hatano S, Itaya Y & Mori S (2008) A new pulverized biomass utilization technology. Powder Technol 180(3): 272–283.

Kobayashi N, Tanabe Y & Itaya Y (2012) Pulverization technique of biomass by multiple tube vibration mill. In: Proceedings of the 5th Asian Particle Technology Symposium–APT 2012, 2012 July 2–5, Singapore: 155–156. URI: http://dx.doi.org/10.3850/978-981-07-2518-1_052. Cited 2013/09/05.

99

Kociszewski M, Gozdecki C, Wilczyński A, Zajchowski S & Mirowski J (2012) Effect of industrial wood particle size on mechanical properties of wood-polyvinyl chloride composites. Eur J Wood Wood Prod 70(1–3): 113–118.

Kokko L, Tolvanen H, Hämäläinen K & Raiko R (2012) Comparing the energy required for fine grinding torrefied and fast heat treated pine. Biomass Bioenergy 42: 219–223.

Kotake N, Kuboki M, Kiya S & Kanda Y (2011) Influence of dry and wet grinding conditions on fineness and shape of particle size distribution of product in a ball mill. Adv Powder Technol 22(1): 86–92.

Kumar P, Barrett DM, Delwiche MJ & Stroeve P (2009) Methods for pretreatment of lignocellulosic biomass for efficient hydrolysis and biofuel production. Ind Eng Chem Res 48(8): 3713–3729.

Kratky L & Jirout T (2011) Biomass size reduction machines for enhancing biogas production. Chem Eng Technol 34(3): 391–399.

Kwade A, Blecher L & Schwedes J (1996) Motion and stress intensity of grinding beads in a stirred media mill. Part 2: Stress intensity and its effect on comminution. Powder Technol 86(1): 69–76.

Kwade A (2003) A stressing model for the description and optimization of grinding processes. Chem Eng Technol 26(2): 199–205.

Laxman RS & Lachke AH (2009) Bioethanol from lignocellulosic biomass. Part I Pretreatment of the substrates. In: Pandey AD (ed) Handbook of plant-based biofuels, Boca Raton, London, New York, CRC Press: 121–139.

Lanvermann C, Evans R, Schmitt U, Hering S & Niemz P (2013) Distribution of structure and lignin within growth rings of Norway spruce. Wood Sci Technol 47(3): 627–641.

Lanvermann C, Hass P, Wittel FK & Niemz P (2014) Mechanical properties of Norway spruce: intra-ring variation and generic behavior of earlywood and latewood until failure. Bioresources 9(1): 105–119.

Leppänen K, Bjurhager I, Peura M, Kallonen A, Suuronen J-P, Penttilä P, Love J, Fagerstedt K & Serimaa R (2011) X-ray scattering and microtomography study on the structural changes of never-dried silver birch, European aspen and hybrid aspen during drying. Holzforschung 65(6): 865–873.

Leu S-Y & Zhu JY (2013) Substrate-related factors affecting enzymatic saccharification of lignocelluloses: our recent understanding. BioEnergy Res 6(2): 405–415.

Liimatainen H, Haikkala P, Lucander M, Karojärvi R & Tuovinen O (1999) Grinding and pressure grinding. In: Sundholm J, Gullichsen J & Paulapuro H (eds) Papermaking science and technology, book 5, mechanical pulping. Helsinki, Fapet Oy: 107–158.

Lindholm C-A & Kurdin JA (1999) Chemimechanical pulping. In: Sundholm J, Gullichsen J & Paulapuro H (eds) Papermaking science and technology, book 5, mechanical pulping. Helsinki, Fapet Oy: 222–249.

Lowrison GC (1974) Crushing and grinding: Size reduction of solid materials. London, Butterworths.

Lu H, Ip E, Scott J, Foster P, Vickers M & Baxter LL (2010) Effects of particle shape and size on devolatilization of biomass particle. Fuel 89(5): 1156–1168.

100

Mansfield SD, Mooney C & Saddler JN (1999) Substrate and enzyme characteristics that limit cellulose hydrolysis. Biotechnol Prog 15(5): 804–816.

Marton R & Eskelinen E (1982) Impact testing in the study of chip refining. Tappi J 65(12): 85–89.

Maurer A & Fengel D (1992) On the origin of milled wood lignin, part 1: the influence of ball-milling on the ultrastructure of wood cell walls and the solubility of lignin. Holzforschung 46(5): 417–423.

Meier M, John E, Wieckhusen D, Wirth W & Peukert W (2008) Characterization of the grinding behaviour in a single particle impact device: studies on pharmaceutical powders. Eur J Pharm Sci 34(1): 45–55.

Meier M, John E, Wieckhusen D, Wirth W & Peukert W (2009) Influence of mechanical properties on impact fracture: prediction of the milling behaviour of pharmaceutical powders by nanoindentation. Powder Technol 188(3): 301–313.

Migneault S, Koubaa A, Erchiqui F, Chaala A, Englund K, Krause C & Wolcott M (2008) Effect of fiber length on processing and properties of extruded wood-fiber/HDPE composites. J Appl Polym Sci 110(2): 1085–1092.

Migneault S, Koubaa A, Erchiqui F, Chaala A, Englund K & Wolcott MP (2009) Effects of processing method and fiber size on the structure and properties of wood-plastic composites. Composites, Part A 40(1): 80–85.

Millett MA, Baker AJ & Satter LD (1975) Pretreatments to enhance chemical, enzymatic and microbiological attack of cellulosic materials. Biotechnol Bioeng Symp 5: 193–219.

Millett MA, Baker AJ & Satter LD (1976) Physical and chemical pretreatments for enhancing cellulose saccharification. Biotechnol Bioeng Symp 6: 125–153.

Millett MA, Effland MJ & Caulfield DF (1979) Influence of fine grinding on the hydrolysis of cellulosic materials - acid vs. enzymatic. In: Brown Jr RD & Jurasek L (eds) Hydrolysis of cellulose: mechanisms of enzymatic and acid catalysis. Washington DC, American Chemical Society: 71–89.

Mikushina IV, Troitskaya IB, Dushkin AV & Bazarnova NG (2002) Changes in the chemical composition of wood during mechanochemical treatment. Chem Sust Dev 10(4): 441–445.

Mikushina IV, Troitskaya IB, Dushkin AV, Ol’khov YA & Bazarnova NG (2003) Transformations of wood structure under mechanochemical treatment. Chem Sust Dev 11(2): 363–370.

Mooney CA, Mansfield SD, Touhy MG & Saddler JN (1998) The effect of initial pore volume and lignin content on the enzymatic hydrolysis of softwoods. Bioresour Technol 64(2): 113–119.

Mori S (2009) Development of utilization technologies of biomass energy. J Environ Eng Manage 19(2): 67–72.

Müller U, Jošcák T & Teischinger A (2003) Strength of dried and re-moistened spruce wood compared to native wood. Holz Roh- Werkst 61(6): 439–443.

101

Nakach M, Authelin J-R, Chamayou A & Dodds J (2004) Comparison of various milling technologies for grinding pharmaceutical powders. Int J Miner Process 74S: S173–S181.

Nichols G, Byard S, Bloxham MJ, Botterill J, Dawson NJ, Dennis A, Diart V, North NC & Sherwood JD (2002) A review of the terms agglomerate and aggregate with a recommendation for nomenclature used in powder and particle characterization. J Pharm Sci 91(10): 2103–2109.

Neumann M, Herter J, Droste BO & Hartwig S (2011) Compressive behaviour of axially loaded spruce wood under large deformations at different strain rates. Eur J Wood Wood Prod 69(3): 345–357.

Nied R (2007) Rotor impact mills. In: Salman AD, Ghadiri M & Hounslow MJ (eds) Handbook of powder technology, volume 12, particle breakage. 1st edition, Amsterdam, Boston, Heidelberg, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokio, Elsevier: 230–249.

Orumwense OA & Forssberg E (1992) Superfine and ultrafine grinding – A literature survey. Miner Process Extr Metall Rev 11(1): 107–127.

Palaniandy S, Azizli KAM, Hussin H & Hashim SFS (2008) Effect of operational parameters on the breakage mechanism of silica in a jet mill. Miner Eng 21(5): 380–388.

Palaniandy S & Azizli KAM (2009) Mechanochemical effects on talc during fine grinding process in a jet mill. Int J Miner Process 92(1–2): 22–33.

Panshin AJ and de Zeeuw C (1980) Textbook of wood technology: structure, identification, properties and uses of the commercial woods of the United States and Canada. 4th edition, New York, McGraw-Hill.

Paulrud S, Mattsson JE & Nilsson C (2002) Particle and handling characteristics of wood fuel powder: effects of different mills. Fuel Process Technol 76(1): 23–39.

Phanphanich M & Mani S (2011) Impact of torrefaction on the grindability and fuel characteristics of forest biomass. Bioresour Technol 102(2): 1246–1253.

Persson K (2000) Micromechanical modelling of wood and fibre properties. PhD thesis. Lund University, Department of Mechanics and Materials.

Peukert W & Vogel L (2001) Comminution of polymers - an example of product engineering. Chem Eng Technol 24(9): 945–950.

Peukert W (2003) Particle properties – promise and neglect. China Particuol. 1(1): 13–19. Peukert W (2004) Material properties in fine grinding. Int. J. Miner. Process. 74S: S3–S17. Peukert W (2013) Personal communication. Peura M, Kölln K, Grotkopp I, Saranpää P, Müller M & Serimaa R (2007) The effect of

axial strain on crystalline cellulose in Norway spruce. Wood Sci Technol 41(7): 565–583.

Peura M, Müller M, Vainio U, Sarén M-P, Saranpaa P & Serimaa R (2008a) X-ray microdiffraction reveals the orientation of cellulose microfibrils and the size of cellulose crystallites in single Norway spruce tracheids. Trees 22(1): 49–61.

102

Peura M, Sarén M-P, Laukkanen J, Nygård K, Andersson S, Saranpää P, Paakkari T, Hämäläinen K & Serimaa R (2008b) The elemental composition, the microfibril angle distribution and the shape of the cell cross-section in Norway spruce xylem. Trees 22(4): 499–510.

Pew JC & Weyna P (1962) Fine grinding, enzyme digestion, and the lignin-cellulose bond in wood. Tappi 45(3): 247–256.

Pierre F, Almeida G, Huber F, Jacquin P & Pierré P (2013) An original impact device for biomass characterisation: results obtained for spruce and poplar at different moisture contents. Wood Sci Technol 47(3): 537–555.

Pirkkalainen K, Peura M, Leppänen K, Salmi A, Meriläinen A, Saranpää P & Serimaa R (2012) Simultaneous x-ray diffraction and x-ray fluorescence microanalysis on secondary xylem of Norway spruce. Wood Sci Technol 46(6): 1113–1125.

Poulsen JS, Moran PM, Shih CF & Byskov E (1997) Kink band initiation and band broadening in clear wood under compressive loading. Mech Mater 25(1): 67–77.

Rapp AO, Brischke C & Welzbacher CR (2006) Interrelationship between the severity of heat treatments and sieve fractions after impact ball milling: a mechanical test for quality control of thermally modified wood. Holzforschung 60(1): 64–70.

Reid SR & Peng C (1997) Dynamic uniaxial crushing of wood. Int J Impact Eng 19(5–6): 531–570.

Reiterer A, Lichtenegger H, Tschegg S & Fratzl P (1999) Experimental evidence for a mechanical function of the cellulose microfibril angle in wood cell walls. Philos Mag A 79(9): 2173–2184.

Reiterer A, Lichtenegger H, Fratzl P & Stanzl-Tschegg SE (2001a) Deformation and energy absorption of wood cell walls with different nanostructure under tensile loading. J Mat Sci 36(19): 4681–4686.

Reiterer A & Stanzl-Tschegg SE (2001b) Compressive behaviour of softwood under uniaxial loading at different orientations to the grain. Mech Mater 33(12): 705–715.

Renner K, Kenyó C, Móczó J & Pukánszky B (2010) Micromechanical deformation processes in PP/wood composites: particle characteristics, adhesion, mechanisms. Composites, Part A 41(11): 1653–1661.

Repellin V, Govin A, Rolland M & Guyonnet R (2010) Energy requirement for fine grinding of torrefied wood. Biomass Bioenergy 34(7): 923–930.

Reuther JJ, Karsner GG & Jack ST (1985) Plane flame furnace combustion studies of pulverized wood. In: Overend RP, Milne TA & Mudge LK (eds) Fundamentals of thermochemical biomass conversion. London, New York, Elsevier Applied Science Publishers: 793–810.

Rivers DB & Emert GH (1987) Lignocellulose pretreatment: a comparison of wet and dry ball attrition. Biotechnol Lett 9(5): 365–368.

Rumpf H (1973) Physical aspects of comminution and new formulation of a law of comminution. Powder Technol 7(3):145–159.

Rumpf H (1990) Particle technology. London, New York, Tokyo, Melbourne, Madras, Chapman and Hall.

103

Salmén L (1982) Temperature and water induced softening behaviour of wood fiber based materials. PhD thesis. The Royal Institute of Technology, Department of Paper Technology.

Salmén L, Lucander M, Härkönen E, Sundholm J (1999) Fundamentals of mechanical pulping. In: Sundholm J, Gullichsen J & Paulapuro H (eds) Papermaking science and technology, book 5, mechanical pulping. Helsinki, Fapet Oy: 34–65.

Salmén L (2004) Micromechanical understanding of the cell-wall structure. C R Biol 327(9–10): 873–880.

Salmén L, Olsson A-M, Stevanic JS, Simonivić J & Radotić K (2012) Structural organisation of the wood polymers in the wood fibre structure. BioResources 7(1): 521–532.

Salmi A, Salminen L & Hæggström E (2009) Quantifying fatigue generated in high strain rate cyclic loading of Norway spruce. J Appl Phys 106: 104905-1–104905-5.

Salmi A, Montonen R, Salminen LI, Suuronen J-P, Serimaa R & Hæggström E (2012a) Cyclic impulsive compression loading along the radial and tangential wood directions causes localized fatigue. J Appl Phys 112: 124913-1–124913-1-8.

Salmi A, Salminen LI, Engberg BA, Björkqvist T & Hæggström E (2012b) Repetitive impact loading causes local plastic deformation in wood. J Appl Phys 111: 024901-1–024901-8.

Sarén M-P, Serimaa R, Andersson S, Paakkari T, Saranpää P & Pesonen E (2001) Structural variation of tracheids in Norway spruce (Picea abies [L.] Karst). J Struct Biol 136(2): 101–109.

Sarén M-P, Serimaa R & Tolonen Y (2006) Determination of fiber orientation in Norway spruce using x-ray diffraction and laser scattering. Holz Roh- Werkst 64(3): 183–188.

Schell DJ & Harwood C (1994) Milling of lignocellulosic biomass. Appl Biochem Biotechnol 45–46(1): 159–168.

Schwanninger M, Rodrigues JC, Pereira H & Hinterstoisser B (2004) Effects of short-time vibratory ball milling on the shape of FT-IR spectra of wood and cellulose. Vib Spectrosc 36(1): 23–40.

Scott-Emuakpor OE (2007) Development of a novel energy-based method for multi-axial fatigue strength assessment. PhD thesis. The Ohio State University, Graduate Program of Mechanical Engineering.

Segal L, Creely JJ, Martin Jr AE & Conrad CM (1959) An empirical method for estimating the degree of crystallinity of native cellulose using the x-ray diffractometer. Text Res J 29 (10): 786–794.

Silva GGD, Couturier M, Berrin J-G, Buléon A & Rouau X (2012) Effects of grinding processes on enzymatic degradation of wheat straw. Bioresour Technol 103(1): 192–200.

Stark N & Rowlands RE (2003) Effects of wood fiber characteristics on mechanical properties of wood/polypropylene composites. Wood Fiber Sci 35(2): 167–174.

Stevanic JS & Salmén L (2009) Orientation of the wood polymers in the cell wall of spruce wood fibres. Holzforschung 63(5): 497–503.

104

Stubičar N, Šmit I, Stubičar M, Tonejc A, Jánosi A, Schurz J & Zipper P (1998) An x-ray diffraction study of the crystalline to amorphous phase change in cellulose during high-energy dry ball milling. Holzforschung 52(5): 455–458.

Sundholm J (1999) What is mechanichal pulping? In: Sundholm J, Gullichsen J & Paulapuro H (eds) Papermaking science and technology, book 5, mechanical pulping. Helsinki, Fapet Oy: 17–21.

Surmiński J (2007) Wood properties and uses. In: Tjoelker MG, Boratyński A & Bugała W (eds) Biology and ecology of Norway spruce, Forestry sciences 78. Dordrecht, Springer: 333–342.

Takatani M, Ito H, Ohsugi S, Kitayama T, Saegusa M, Kawai S & Okamoto T (2000) Effect of lignocellulosic materials on the properties of thermoplastic polymer/wood composites. Holzforschung 54(2): 197–200.

Tavares LM (2007) Breakage of single particles: quasi-static. In: Salman AD, Ghadiri M & Hounslow MJ (eds) Handbook of powder technology, volume 12, particle breakage. 1st edition, Amsterdam, Boston, Heidelberg, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokio, Elsevier: 3–68.

Thunman H, Leckner B, Niklasson F & Johnsson F (2002) Combustion of wood particles – a particle model for eulerian calculations. Combust Flame 129(1–2): 30–46.

Tienvieri T, Huusari E, Sundholm J, Vuorio P, Kortelainen J, Nystedt H & Artamo A (1999) Thermomechanical pulping. In: Sundholm J, Gullichsen J & Paulapuro H (eds) Papermaking science and technology, book 5, mechanical pulping. Helsinki, Fapet Oy: 159–221.

Toraman OY (2011) Use of wet grinding and aging to produce narrow particle size distribution. Part Sci Technol 29(5): 475–480.

Tschegg EK, Reiterer A, Pleschberger T & Stanzl-Tschegg SE (2001) Mixed mode fracture energy of sprucewood. J Mater Sci 36(14): 3531–3537.

Tumuluru JS, Sokhansanj S, Hess RJ, Wright CT & Boardman RD (2011) A review on biomass torrefaction process and product properties for energy applications. Ind Biotechnol 7(5): 384–401.

Uhmeier A & Salmén L (1996) Influence of strain rate and temperature on the radial compression behaviour of wet spruce. J Eng Mater Technol 118(3): 289–294.

Van der Stelt MJC, Gerhauser H, Kiel JHA & Ptasinski KJ (2011) Biomass upgrading by torrefaction for the production of biofuels: a review. Biomass Bioenergy 35: 3748–3762.

Van Essendelft DT, Zhou X & Kang BS-J (2013) Grindability determination of torrefied biomass materials using the Hybrid Work Index. Fuel 105: 103–111.

Vasic S & Stanzl-Tschegg S (2007) Experimental and numerical investigation of wood fracture mechanisms at different humidity levels. Holzforschung 61(4): 367–374.

Vincent JFV (1990) Fracture properties of plants. Adv Bot Res 17: 235–287. Vogel L & Peukert W (2002) Characterisation of grinding-relevant particle properties by

inverting a population balance model. Part Part Syst Charact 19(3): 149–157. Vogel L & Peukert W (2003) Breakage behaviour of different materials - construction of a

mastercurve for the breakage probability. Powder Technol 129(1): 101–110.

105

Vogel L & Peukert W (2004) Determination of material properties relevant to grinding by practicable labscale milling tests. Int J Miner Process 74S: S329–S338.

Vogel L & Peukert W (2005) From single particle impact behaviour to modelling of impact mills. Chem Eng Sci 60(18): 5164–5176.

Wang Y & Forssberg E (2007) Enhancement of energy efficiency for mechanical production of fine and ultra-fine particles in comminution. China Particuol 5(3): 193–201.

Welzbacher CR, Rassam G, Talaei A & Brischke C (2011) Microstructure, strength and structural integrity of heat-treated beech and spruce wood. Wood Mater Sci Eng 6(4): 219–227.

Widehammar S (2004) Stress-strain relationships for Spruce wood: Influence of strain rate, moisture content and loading direction. Exp Mech 44(1): 44–48.

Wilczek M, Bertling J & Hintemann D (2004) Optimised technologies for cryogenic grinding. Int J Miner Process 74S: S425–S434.

Wills BA (2006) Grinding mills. In: Napier-Munn TJ (ed) Wills’ mineral processing technology: an introduction to the practical aspects of ore treatment and mineral recovery. 7th edition, Amsterdam, Boston, Heidelberg, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokio, Butterworth-Heinemann: 146–185.

Wittel FK, Dill-Langer G & Kröplin B-H (2005) Modeling of damage evolution in soft-wood perpendicular to grain by means of a discrete element approach. Comput Mater Sci 32(3–4): 594–603.

Yokoyama T & Inoue Y (2007) Selection of fine grinding mills. In: Salman AD, Ghadiri M & Hounslow MJ (eds) Handbook of powder technology, volume 12, particle breakage. 1st edition, Amsterdam, Boston, Heidelberg, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokio, Elsevier: 487–508.

Zaini MJ, Fuad MYA, Ismail Z, Mansor MS & Mustafah J (1996) The effect of filler content and size on the mechanical properties of polypropylene/oil palm wood flour composites. Polym Int 40(1): 51–55.

Zhang M, Song X, Deines TW, Pei ZJ & Wang D (2012a) Biofuel manufacturing from woody biomass: effects of sieve size used in biomass size reduction. J Biomed Biotechnol. URI: http://doi.org/10.1155/2012/581039. Cited 2013/05/07.

Zhang M, Song X, Zhang P, Pei ZJ & Deines TW (2012b) Size reduction of cellulosic biomass in biofuel manufacturing: a study on confounding effects of particle size and biomass crystallinity. J Manuf Sci Eng 134(1): 011009-1–011009-9.

Zhang Q, Zhang P, Pei ZJ & Pei L (2011) Effects of treatments on cellulosic biomass structure in ethanol manufacturing: a literature review. In: ASME 2011 international mechanical engineering congress, volume 3, design and manufacturing. Denver, Colorado, ASME: 287–294. URI: http://doi.org/10.1115/IMECE2011-64304. Cited 2013/05/07.

Zhao X-Y, Ao Q, Yang L-W, Yang Y-F, Sun J-C & Gai G-S (2009) Application of superfine pulverization technology in biomaterial industry. J Taiwan Inst Chem Eng 40(3): 337–343.

106

Zhu JY, Wang GS, Pan X & Gleisner R (2009a) Specific surface to evaluate the efficiencies of milling and pretreatment of wood for enzymatic saccharification. Chem Eng Sci 64(3): 474–485.

Zhu JY, Pan XJ, Wang GS & Gleisner R (2009b) Sulfite pretreatment (SPORL) for robust enzymatic saccharification of spruce and red pine. Bioresour Technol 100(8): 2411–2418.

Zhu JY, Pan XJ & Zalesny Jr RS (2010a) Pretreatment of woody biomass for biofuel production: energy efficiency, technologies and recalcitrance. Appl Microbiol Biotechnol 87(3): 847–857.

Zhu JY & Pan XJ (2010b) Woody biomass pretreatment for cellulosic ethanol production: Technology and energy consumption evaluation. Bioresour Technol 101(13): 4992–5002.

Zhu JY (2011) Physical pretreatment - woody biomass size reduction - for forest biorefinery. In: Zhu JY, Zhang X & Pan XJ (eds) Sustainable production of fuels, chemicals, and fibers from forest biomass. Washington DC American Chemical Society: 89–107.

Zhu L, O’Dwyer JP, Chang VS, Granda CB & Holtzapple MT (2008) Structural features affecting biomass enzymatic digestibility. Bioresour Technol 99(9): 3817–3828.

Zhu L, O’Dwyer JP, Chang VS, Granda CB & Holtzapple MT (2010) Multiple linear regression model for predicting biomass digestibility from structural features. Bioresour Technol 101(13): 4971–4979.

Zhu W, Zhu JY, Gleisner R & Pan XJ (2010) On energy consumption for size-reduction and yields from subsequent enzymatic saccharification of pretreated lodgepole pine. Bioresour Technol 101(8): 2782–2792.

107

Original Publications

I Karinkanta P, Illikainen M, Niinimäki J (2012) Pulverisation of dried and screened Norway spruce (Picea abies) sawdust in an air classifier mill. Biomass and Bioenergy 44: 96–106.

II Karinkanta P, Illikainen M, Niinimäki J (2014) Effect of different impact events in fine grinding mills on the development of the physical properties of dried Norway spruce (Picea abies) wood in pulverisation. Powder Technology 253: 352–359.

III Karinkanta P, Illikainen M, Niinimäki J (2013) Impact-based pulverisation of dried and screened Norway spruce (Picea abies) sawdust in an oscillatory ball mill. Powder Technology 233: 286–294.

IV Karinkanta P, Illikainen M, Niinimäki J (2013) Effect of grinding conditions in oscillatory ball milling on the morphology of particles and cellulose crystallinity of Norway spruce (Picea abies). Holzforschung 67(3): 277–283.

V Karinkanta P, Illikainen M, Niinimäki J (2014) Effect of mild torrefaction on pulverization of Norway spruce (Picea abies) by oscillatory ball milling: particle morphology and cellulose crystallinity. Holzforschung 68(3):337–343.

Reprinted with permission from Elsevier (I–III) and De Gruyter (IV–V).

Original publications are not included in the electronic version of this thesis.

108

A C T A U N I V E R S I T A T I S O U L U E N S I S

Book orders:Granum: Virtual book storehttp://granum.uta.fi/granum/

S E R I E S C T E C H N I C A

500. Hosio, Simo (2014) Leveraging Social Networking Services on MultipurposePublic Displays

501. Ohenoja, Katja (2014) Particle size distribution and suspension stability inaqueous submicron grinding of CaCO3 and TiO2

502. Puustinen, Jarkko (2014) Phase structure and surface morphology effects on theoptical properties of nanocrystalline PZT thin films

503. Tuhkala, Marko (2014) Dielectric characterization of powdery substances usingan indirectly coupled open-ended coaxial cavity resonator

504. Rezazadegan Tavakoli, Hamed (2014) Visual saliency and eye movement :modeling and applications

505. Tuovinen, Tommi (2014) Operation of IR-UWB WBAN antennas close to humantissues

506. Vasikainen, Soili (2014) Performance management of the university educationprocess

507. Jurmu, Marko (2014) Towards engaging multipurpose public displays : designspace and case studies

508. Namal, Suneth (2014) Enhanced communication security and mobilitymanagement in small-cell networks

509. Huang, Xiaohua (2014) Methods for facial expression recognition withapplications in challenging situations

510. Ala-aho, Pertti (2014) Groundwater-surface water interactions in esker aquifers :from field measurements to fully integrated numerical modelling

511. Torabi Haghighi, Ali (2014) Analysis of lake and river flow regime alteration toassess impacts of hydraulic structures

512. Bordallo López, Miguel (2014) Designing for energy-efficient vision-basedinteractivity on mobile devices

513. Suopajärvi, Hannu (2014) Bioreducer use in blast furnace ironmaking in Finland :techno-economic assessment and CO2 emission reduction potential

514. Sobocinski, Maciej (2014) Embedding of bulk piezoelectric structures in LowTemperature Co-fired Ceramic

515. Kulju, Timo (2014) Utilization of phenomena-based modeling in unit operationdesign

C516etukansi.kesken.fm Page 2 Tuesday, December 2, 2014 11:22 AM

ABCDEFG

UNIVERSITY OF OULU P .O. B 00 F I -90014 UNIVERSITY OF OULU FINLAND

A C T A U N I V E R S I T A T I S O U L U E N S I S

S E R I E S E D I T O R S

SCIENTIAE RERUM NATURALIUM

HUMANIORA

TECHNICA

MEDICA

SCIENTIAE RERUM SOCIALIUM

SCRIPTA ACADEMICA

OECONOMICA

EDITOR IN CHIEF

PUBLICATIONS EDITOR

Professor Esa Hohtola

University Lecturer Santeri Palviainen

Postdoctoral research fellow Sanna Taskila

Professor Olli Vuolteenaho

University Lecturer Veli-Matti Ulvinen

Director Sinikka Eskelinen

Professor Jari Juga

Professor Olli Vuolteenaho

Publications Editor Kirsti Nurkkala

ISBN 978-952-62-0718-6 (Paperback)ISBN 978-952-62-0719-3 (PDF)ISSN 0355-3213 (Print)ISSN 1796-2226 (Online)

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

U N I V E R S I TAT I S O U L U E N S I SACTAC

TECHNICA

OULU 2014

C 516

Pasi Karinkanta

DRY FINE GRINDING OF NORWAY SPRUCE (PICEA ABIES) WOOD IN IMPACT-BASED FINE GRINDING MILLS

UNIVERSITY OF OULU GRADUATE SCHOOL;UNIVERSITY OF OULU,FACULTY OF TECHNOLOGY

C 516

ACTA

Pasi Karinkanta

C516etukansi.kesken.fm Page 1 Tuesday, December 2, 2014 11:22 AM