Modelling the Flow of Cane Constituents through the Milling Process ...

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

Thesis submitted by

Omkar P Thaval B.Tech (Sugar Technologist)

for the degree of Master of Applied Science (Research) School of Chemistry, Physics & Mechanical Engineering

Science & Engineering Faculty

Queensland University of Technology

September 2012

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

Dedicated to my mother

Mrs. Bharati P. Thaval

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory I

Keywords

Bagasse, Brix Extraction, Dynamic Modelling, Extraction Model, Impurities,

Insoluble Solids, Juice, Mass Balance, Milling Efficiency, Mill Parameters, Process

Modelling, Simulation Packages, Sugarcane, Soluble Solids, Suspended Solids.

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory III

Abstract

This thesis reports on an investigation to develop an advanced and

comprehensive milling process model of the raw sugar factory. Although the new

model can be applied to both, the four-roller and six-roller milling units, it is

primarily developed for the six-roller mills which are widely used in the Australian

sugar industry. The approach taken was to gain an understanding of the previous

milling process simulation model “MILSIM” developed at the University of

Queensland nearly four decades ago. Although the MILSIM model was widely

adopted in the Australian sugar industry for simulating the milling process it did have

some incorrect assumptions. The study aimed to eliminate all the incorrect

assumptions of the previous model and develop an advanced model that represents

the milling process correctly and tracks the flow of other cane components in the

milling process which have not been considered in the previous models.

The development of the milling process model was done is three stages. Firstly,

an enhanced milling unit extraction model (MILEX) was developed to access the

mill performance parameters and predict the extraction performance of the milling

process. New definitions for the milling performance parameters were developed and

a complete milling train along with the juice screen was modelled. The MILEX

model was validated with factory data and the variation in the mill performance

parameters was observed and studied. Some case studies were undertaken to study

the effect of fibre in juice streams, juice in cush return and imbibition% fibre on

extraction performance of the milling process. It was concluded from the study that

the empirical relations developed for the mill performance parameters in the

MILSIM model were not applicable to the new model. New empirical relations have

to be developed before the model is applied with confidence.

Secondly, a soluble and insoluble solids model was developed using modelling

theory and experimental data to track the flow of sucrose (pol), reducing sugars

(glucose and fructose), soluble ash, true fibre and mud solids entering the milling

train through the cane supply and their distribution in juice and bagasse streams.. The

soluble impurities and mud solids in cane affect the performance of the milling train

and further processing of juice and bagasse. New mill performance parameters were

IV Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

developed in the model to track the flow of cane components. The developed model

is the first of its kind and provides some additional insight regarding the flow of

soluble and insoluble cane components and the factors affecting their distribution in

juice and bagasse. The model proved to be a good extension to the MILEX model to

study the overall performance of the milling train.

Thirdly, the developed models were incorporated in a proprietary software

package “SysCAD’ for advanced operational efficiency and for availability in the

‘whole of factory’ model. The MILEX model was developed in SysCAD software to

represent a single milling unit. Eventually the entire milling train and the juice screen

were developed in SysCAD using series of different controllers and features of the

software. The models developed in SysCAD can be run from macro enabled excel

file and reports can be generated in excel sheets. The flexibility of the software, ease

of use and other advantages are described broadly in the relevant chapter. The

MILEX model is developed in static mode and dynamic mode. The application of the

dynamic mode of the model is still under progress.

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory V

Table of Contents

Keywords ................................................................................................................. I

Abstract .................................................................................................................. III

Table of Contents ..................................................................................................... V

List of Figures .........................................................................................................XI

List of Tables ....................................................................................................... XIII

Symbols ................................................................................................................ XV

Statement of Original Authorship ....................................................................... XVII

Acknowledgements .............................................................................................. XIX

List of Publications .............................................................................................. XXI

Chapter 1: Introduction ........................................................................................ 1

1.1 Introductory Remarks ...................................................................................... 1

1.2 The Australian Sugar Industry ......................................................................... 1

1.3 Composition of Sugarcane ............................................................................... 1

1.3.1 Botany of sugarcane ............................................................................... 1

1.3.2 Composition of juice ............................................................................... 2

1.3.3 Composition of fibre ............................................................................... 3

1.4 The Milling Process ......................................................................................... 3

1.4.1 Overview ................................................................................................ 3

1.4.2 Cane preparation ..................................................................................... 3

1.4.3 The milling train ..................................................................................... 4

1.4.4 The six-roller mill ................................................................................... 6

1.4.5 Extraction theory .................................................................................... 7

1.4.6 Simulation and modelling of the sugar factory ........................................ 8

1.5 Scope of Research............................................................................................ 9

1.5.1 Research problem ................................................................................... 9

1.5.2 Objectives ............................................................................................. 10

1.5.3 Individual contribution to the research team .......................................... 10

VI Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

1.6 Overview of Thesis ....................................................................................... 11

Chapter 2: Literature Review ............................................................................ 13

2.1 Introductory Remarks .................................................................................... 13

2.2 The MILSIM Model ...................................................................................... 13

2.2.1 Introductory remarks ............................................................................ 13

2.2.2 Extraction performance parameters ...................................................... 14

2.2.3 Determining extraction of a single milling unit ..................................... 16

2.2.4 Calculating reabsorption factor & imbibition coefficient....................... 19

2.2.5 Reabsorption factor & imbibition coefficient multipliers ...................... 20

2.2.6 Concluding remarks ............................................................................. 21

2.3 The modified MILSIM extraction model ....................................................... 21


2.3.2 Development of new mill performance parameters ............................... 21

2.3.3 Analytical mode ................................................................................... 23

2.3.4 Predictive mode .................................................................................... 23


2.4 Wienese’s extraction model ........................................................................... 24


2.4.2 Mill performance parameters ................................................................ 24

2.4.3 Model application ................................................................................. 25


2.5 Other mass balance models ............................................................................ 26


2.5.2 The fibre flow model ............................................................................ 26

2.5.3 Wienese’s model .................................................................................. 30

2.5.4 Loubser’s model ................................................................................... 30


2.6 Soluble solids model...................................................................................... 31


2.6.2 Sucrose................................................................................................. 32

2.6.3 Reducing sugars ................................................................................... 33

2.6.4 Ash ...................................................................................................... 34

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory VII

2.6.5 Concluding remarks .............................................................................. 34

2.7 Insoluble solids model ................................................................................... 35

2.7.1 Introductory remarks............................................................................. 35

2.7.2 Bagasse production model .................................................................... 35

2.7.3 Wright’s model ..................................................................................... 35


2.8 Dynamic modelling of the milling process ..................................................... 36


2.8.2 McWhinney’s dynamic model .............................................................. 36


2.9 Concluding Remarks ...................................................................................... 37

Chapter 3: The MILEX Model ........................................................................... 39

3.1 Introductory remarks ...................................................................................... 39

3.2 The milling unit model ................................................................................... 39


3.2.2 Corrected filling ratio............................................................................ 39

3.2.3 Corrected reabsorption factor ................................................................ 41

3.2.4 Corrected imbibition coefficient............................................................ 41

3.2.5 Separation efficiency ............................................................................ 42

3.3 Juice screen model ......................................................................................... 43

3.4 The milling train model.................................................................................. 44

3.5 Solving the MILEX model ............................................................................. 45

3.5.1 Analytical mode.................................................................................... 45

3.5.2 Predictive mode .................................................................................... 48

3.6 Exploring the model....................................................................................... 50

3.6.1 A base case for testing the model .......................................................... 50

3.6.2 Effect of the constant fibre rate assumption on mill parameters ............. 51

3.6.3 Separation efficiency ............................................................................ 52

3.6.4 Effect of including juice screen in the model ......................................... 53

3.7 Exploring the model....................................................................................... 54

VIII Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory


3.7.2 Factory test at Isis sugar mill ................................................................ 54

3.7.3 Running the model ............................................................................... 57

3.7.4 Discussion of results ............................................................................. 61


3.8 Case studies ................................................................................................... 62

3.8.1 Effect of separation efficiency and juice in return stream from juice

screen on extraction performance ......................................................... 63

3.8.2 Effect of fibre in mixed juice on extraction performance ...................... 65

3.8.3 Effect of added water rate on extraction performance ........................... 66

3.9 Concluding remarks ...................................................................................... 76

Chapter 4: Modelling the Cane Components .................................................... 77

4.1 Introductory remarks ..................................................................................... 77

4.2 Model framework .......................................................................................... 77

4.3 Soluble solids model...................................................................................... 78


4.3.2 Sucrose................................................................................................. 78

4.3.3 Reducing sugars ................................................................................... 79

4.3.4 Soluble ash ........................................................................................... 88

4.3.5 Proteins ................................................................................................ 90


4.4 Insoluble solids.............................................................................................. 93


4.4.2 An overall mass balance model ............................................................ 93

4.4.3 The model for a milling unit ................................................................. 96

4.4.4 The model for the juice screen .............................................................. 97

4.4.5 Model application ................................................................................. 97


4.5 Exploring the model ...................................................................................... 98

4.5.1 The model calibration step .................................................................... 98

4.5.2 Results from the model ......................................................................... 99

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory IX

4.5.3 Effect of impurities on extraction ........................................................ 103

4.6 Concluding remarks ..................................................................................... 104

Chapter 5: SysCAD Model Development ......................................................... 107

5.1 Introductory remarks .................................................................................... 107

5.2 Modelling the milling process ...................................................................... 107

5.3 Process Modelling using SysCAD ............................................................... 108

5.4 Static milling train model ............................................................................. 108

5.4.1 Introductory remarks........................................................................... 108

5.4.2 Developing a milling unit model in SysCAD ...................................... 109

5.4.3 Developing a juice screen model in SysCAD ...................................... 112

5.4.4 Developing a milling train model in SysCAD ..................................... 114

5.4.5 Excel reports ....................................................................................... 116

5.4.6 Model application ............................................................................... 119

5.4.7 Concluding remarks ............................................................................ 121

5.5 Dynamic milling train model ....................................................................... 121


5.5.2 Developing the dynamic milling train model in SysCAD .................... 122

5.5.3 Factors affecting mill extraction performance ..................................... 122

5.5.4 Concluding remarks ............................................................................ 124

5.6 Concluding remarks ..................................................................................... 124

Chapter 6: General Discussions and Conclusions ............................................ 127

6.1 Introductory remarks .................................................................................... 127

6.2 Aim of the research ...................................................................................... 127

6.3 Summary and conclusions of research .......................................................... 127

6.3.1 The enhanced and comprehensive mill extraction model ..................... 127

6.3.2 The cane component model ................................................................ 128

6.3.3 SysCAD model ................................................................................... 128

6.4 Significance of the research ......................................................................... 129


6.4.2 Extraction benefits .............................................................................. 129

X Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

6.4.3 Inputs for downstream models ............................................................ 130

6.5 Recommendations for future research .......................................................... 130

6.6 Concluding remarks .................................................................................... 131

Bibliography ....................................................................................................... 133

Appendices ........................................................................................................ 139

Appendix A-Glossary of terms .................................................................... 139

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XI

List of Figures

Figure 1.1. Schematic diagram of milling train .......................................................... 5

Figure 1.2. Typical layout of a milling unit (Neill et al., 1996) .................................. 7

Figure 2.1. The inlet and outlet quantities of single mill ‘n’ ..................................... 16

Figure 2.2. Milling train fibre mass flows................................................................ 27

Figure 3.1 Percent juice expressed at the four nips of a six-roller mill ..................... 40

Figure 3.2 The inlet and outlet quantities of single milling unit ‘n’ (MILEX) .......... 46

Figure 3.3 (K/C) ratio results .................................................................................. 58

Figure 3.4 Imbibition coefficient results .................................................................. 59

Figure 3.5 Imbibition coefficient multiplier results .................................................. 60

Figure 3.6 Separation efficiency results ................................................................... 61

Figure 3.7 Effect of separation efficiency and juice in cush return on brix extraction .................................................................................................. 63

Figure 3.8 Effect of separation efficiency on brix extraction ................................... 65

Figure 3.9 Effect of fibre in mixed juice on screen efficiency and extraction ........... 66

Figure 3.10 Effect of imbibition% fibre on extraction performance ......................... 70

Figure 3.11 Brix extraction trends for #2, #3, #4 and #5 mills ................................. 71

Figure 3.12 Imbibition coefficient empirical equation functions trend ..................... 73

Figure 3.13 Imbibition coefficient empirical equation trend .................................... 74

Figure 3.14 Extraction figures for MILEX (Standard and Revised definitions) ........ 75

Figure 4.1 Exploration of invert ratio model ............................................................ 84

Figure 4.2 Effect of temperature on protein concentration ....................................... 92

Figure 4.3 Effect of true fibre factor on the ratio of mud solids in mixed juice to mud solids in cane ................................................................................. 99

Figure 4.4 Mill streams brix and fibre mass flows ................................................. 100

Figure 4.5 Mill streams soluble solids mass flow ................................................. 101

Figure 4.6 Mill streams insoluble solids mass flow ............................................... 102

Figure 4.7 Effect of true fibre factor on extraction ................................................. 104

Figure 5.1 Single milling unit in SysCAD ............................................................. 109

Figure 5.2 Milling unit access window .................................................................. 110

Figure 5.3 Analytical controller access window .................................................... 112

Figure 5.4-Juice screen access window ................................................................. 113

Figure 5.5 Milling train model in SysCAD ............................................................ 114

Figure 5.6 General controller- Mill tunning ........................................................... 116

XII Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

Figure 5.7- Model application-cush return to #2 mill ............................................ 120

Figure 5.8-Model application-cush return to #3 mill ............................................. 121

Figure 5.9 Dynamic milling process model ........................................................... 122

Figure 5.10 Extraction v/s time ............................................................................. 124

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XIII

List of Tables

Table 1.1. Chemical composition of sugarcane (Barnes, 1974) .................................. 2

Table 1.2. Soluble solids in juice (Walford, 1996) ..................................................... 3

Table 2.1. Standard values for milling unit fibre analysis (Kent, 2001) .................... 29

Table 2.2. Effect of fibre content in juice on milling unit fibre rate (Kent, 2001) ..... 29

Table 3.1 Input values for the model ....................................................................... 51

Table 3.2 Filling ratio ............................................................................................. 51

Table 3.3 Reabsorption factor ................................................................................. 52

Table 3.4 Imbibition coefficient .............................................................................. 52

Table 3.5 Separation efficiency input values for 2% fibre content of expressed juice .......................................................................................................... 53

Table 3.6 Fibre% juice values for separation efficiency of 95% .............................. 53

Table 3.7 Effect of juice screen on calculated performance parameters ................... 54

Table 3.8 Cane supply and imbibition details .......................................................... 55

Table 3.9 Analysis method ...................................................................................... 56

Table 3.10 Bagasse moisture content results (%) ..................................................... 56

Table 3.11 Bagasse brix content results (%) ............................................................ 56

Table 3.12 Juice brix content (%) ............................................................................ 56

Table 3.13 Juice fibre content (%) ........................................................................... 57

Table 3.14 Average flow measurements .................................................................. 57

Table 3.15 Bagasse analysis .................................................................................... 67

Table 3.16 Compaction and roller surface speed of the mills ................................... 67

Table 3.17 Mill performance parameters (MILSIM)................................................ 67

Table 3.18 Mill performance multipliers (MILSIM) ................................................ 67

Table 3.19 Corrected mill performance parameters (MILEX) .................................. 68

Table 3.20 Mill performance multipliers (MILEX) ................................................. 68

Table 3.20 Regression analysis for brix extraction trends ........................................ 69

Table 4.1 Parameters and their levels studied to determine the effect on invert ratio........................................................................................................... 80

Table 4.2 Rate of hydrolysis in mill products .......................................................... 85

Table 4.3 Sugar analysis results from Isis experiment ............................................. 87

Table 4.4 Analysis of variance for reducing sugars experiment ............................... 87

Table 4.5 Soluble ash in mill products .................................................................... 89

Table 4.6 Protein in mill products ........................................................................... 91

XIV Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

Table 4.7 Mill performance parameters corrected for juice fibre flows .................... 98

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XV

Symbols

푨

Ash (%)

푨풙

Insoluble Ash (%)

푨풚 Soluble Ash (%)

푩 Bagasse

푩 Brix (%)

푪 Cane

푪 Filling Ratio

푪푶 Corrected filling ratio

풅푭 Density of fibre (kg/m3)

푬 Extraction (%)

푬풌 Extraction (Theoretical) (%)

푭 Feed

푭 Fibre (%)

푭푻푭 True fibre factor

푮 Crushing factor

푰 Impurities (%)

푰 Imbibition

푰푪 Imbibition Coefficient

푰푪푶 Corrected imbibition coefficient

푰푹 Invert ratio

푱 Juice

푱 Expressed Juice

푱풔 Cush return from juice screen

XVI Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

푱푴 Mixed Juice

푲 Reabsorption Factor

푲푶 Corrected reabsorption factor

푴 Mud Solids

풎̇ Mass Flow Rate

풏 Mill number

(풏 − ퟏ) Preceding mill, in case of 푛 = 1 it refers to cane

푷 Mass percentage

푷 Pol (%)

풑 Proteins (%)

푹푱푭푱풔 Juice/Fibre ratio of the return stream

푹푺 Reducing sugars

푹푨푩 Ash/Brix ratio

푹푨풑 Protein/Brix ratio

푺푪 Screen efficiency (%)

푺풏 Separation efficiency (%)

푺푻푭풏 True fibre separation efficiency (%)

푻푭 True Fibre

푽̇ Volume flow rate (m3/s)

푽̇푬

Escribed volume (m3/s)

푾 Water (%)

풀

Brix fraction

풁 Purity (%)

풁푹 Purity ratio

휼 Mixing efficiency

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XVII

Statement of Original Authorship

The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To the

best of my knowledge and belief, the thesis contains no material previously

published or written by another person except where due reference is made.

Signature: _________________________

Date: _________________________

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XIX

Acknowledgements

Firstly, the author sincerely wishes to thank his principal supervisor A/Prof.

Geoff Kent of the Queensland University of Technology for his encouragement to

commence this project and the enormous support and advice throughout this

investigation. The author also wishes to thank his associate supervisor Prof. Ross

Broadfoot of the Queensland University of Technology for his ongoing assistance

throughout this investigation.

Secondly, the author wishes to thank the Sugar Research & Development

Corporation (SRDC), Sugar Research Limited (SRL) and Queensland University of

Technology (QUT) for their financial support of this project.

Thirdly, the author wishes to thank those who assisted in the experimental

investigations that formed part of this research project. The management and

production staff at Isis Central Mill Company Limited is acknowledged for providing

the juice samples and the assistance during the sampling. The author wishes to thank

Wanda Stolz, Danny Nguyen and Caroline Thai for their guidance and assistance

with the laboratory experiments. The author wishes to thank Robyn Lloyd for

conducting the proteins experiments. The author wishes to thank Prof. William

Doherty and Dr. Mark Harrison of the Queensland University of Technology for

their fruitful discussion and advice with the result analysis.

Fourthly, the author wishes to thank those who assisted in validating the

developed model. The Isis Central Mill Company Limited is acknowledged for

providing the cane analysis data. The author wishes to thank A/Prof. Geoff Kent, Mr.

Neil McKenzie and Dr. Floren Plaza for conducting the bagasse and juice analysis at

the Isis Central Mill Company Limited that formed part of the model validation. The

author wishes to thank New South Wales Sugar Milling Co-operative Limited for the

permission to use and publish the Condong and Broadwater factory data.

Fifthly, the author wishes to thank the staff of KWA company; in particular Dr.

John McFeaters and Merry Huang for the licensing, training and support of the

SysCAD process modelling software.

XX Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory

Sixthly, the author wishes to thank the staff and students of the Centre for

Tropical Crops & Biocommodities (CTCB); in particular the Sugar Research &

Innovation group for their ongoing guidance and support throughout this

investigation.

Seventhly, the author wishes to thank his former research guide Late Prof.

G.M.Jenekar, HOD, Dept of Sugar Technology, S.G.G.S Institute of Engineering and

Technology, Maharashtra India for his encouragement to conduct research in the

field of sugar engineering. The author wishes to thank the Dept of Sugar Technology

staff for the assistance provided in the research projects.

Lastly, the author wishes to thank his parents and friends for their support and

encouragement during this long investigation.

Modelling the Flow of Cane Constituents through the Milling Process of a Raw Sugar Factory XXI

List of Publications

1. Thaval, OP, Kent, GA (2012). Modelling the flow of juice through a mill.

International Sugar Journal, 1363 (114), 36-40.

2. Thaval OP, Kent GA (2012). An enhanced mill extraction model.

Proceedings of the 34th Annual Conference of the Australian Society of

Sugar Cane Technologists, Cairns, Australia. (CD-ROM), 34: 11p

3. Thaval OP, Kent GA (2012). Modelling the flow of cane constituents in

the milling process. Proceedings of the South African Sugar Technologists'

Association; 85th Annual Congress, Durban, South Africa. 85: 435-453.

4. Thaval OP, Kent GA (2013). Advanced computer simulation of the milling

process. Proceedings of the XXVIII International Society of Sugar Cane

Technologists Congress, Sau Paulo, Brazil. (Accepted)

5. Thaval OP, Kent GA (2013). Improving mill extraction performance

through dynamic simulation. (In progress)

Introduction 1

CHAPTER 1: INTRODUCTION

1.1 Introductory Remarks

Process modelling is an integral part of any process industry and is

undertaken to simulate how things are done. The process model gives a description

or a prediction of what the process looks like. Developing such models requires

meticulous knowledge of the process. The sugar industry is a process industry and

various models have been developed to represent the different unit operations used in

the industry. The milling process is primarily a unit operation used to extract juice

from sugarcane. Several models have been developed to simulate the process. This

thesis deals with the development and application of an advanced and comprehensive

milling process model for a raw sugar factory.

This chapter provides a general overview of sugarcane composition, the

milling process and equipment, conventional milling terminology and the process

modelling packages used to simulate the milling process. It describes the scope of the

research and the objectives of the project. Finally the chapter is concluded with an

overview of the thesis.

1.2 The Australian Sugar Industry

The Australian sugar industry is one of Australia’s largest and most important

rural industries and sugarcane is Queensland largest agricultural crop. The

Queensland sugar industry produces about 35 million tonnes of sugarcane from

400,000 hectares annually. This sugarcane crop produces approximately 5,000,000

tonnes of raw sugar, 1 million tonnes of molasses and 10 million tonnes of bagasse

annually. Approximately 85% of the raw sugar produced is exported, generating up

to $1.5 billion in export earnings for Queensland (Australian Sugar Milling Council).

1.3 Composition of Sugarcane

1.3.1 Botany of sugarcane

Sugarcane is a grass grown in tropical and subtropical countries. It is a

complex hybrid of various species, derived largely from Saccharum officinarum and

other Saccharum species. It is propagated vegetatively by planting pieces of cane

2 Introduction

stalk. New growth originates from buds on the cane nodes, ensuring uniform

progeny. In the process of cane breeding, new varieties are produced and tested in a

constant search for improved characteristics. This process has been a major factor in

improving productivity in the sugarcane industry (Rein, 2007). Table 1.1 shows a

typical chemical composition of sugarcane (Barnes, 1974).

Table 1.1. Chemical composition of sugarcane (Barnes, 1974)

Constituent Value (% by mass) Water 69–75 Soluble solids (Brix) Sucrose 8–16

Reducing sugars 0.5–3 Organic matter 0.5–1 Inorganic compounds 0.2–0.6 Nitrogenous bodies 0.5–1

Insoluble solids Fibre (dry), lignin, cellulose 10–16 Dirt (soil, extraneous matter) 0.78–1.63

1.3.2 Composition of juice

Sugarcane is by definition a combination of juice and fibre. The mixture of brix

and water constitutes the juice of the sugarcane. Brix refers to the water-soluble

solids in the cane and includes the sugar. Technically, brix is the concentration of a

solution of pure sucrose in water having the same density as a sample of juice at the

same temperature (Bureau of Sugar Experiment Stations, 1984). Brix typically

constitutes about 17% of the cane. The density of the juice is a function of the brix

of the juice and is approximately 1080 kg/m3 for a juice with a typical brix fraction

of 0.2 (Bureau of Sugar Experiment Stations, 2001).

The juice expressed from the cane is an opaque liquid covered with froth due

to air bubbles entangled in it. The colour of the juice varies from light grey to dark

green. Cane juice has an acidic reaction. It has a pH of about 5 to 5.5. The cane juice

is viscous owing to the presence of colloids. Besides water and sucrose, other

constituents of the juice include glucose, fructose, minerals, proteins, gum,

polysaccharides and organic acids. Although juice is often treated as a solution of

sucrose in water, a variety of other extracted compounds are present, some of which

may affect clarification and subsequent processing. These compounds may be

conveniently divided into groups described in Table 1.2 (Walford, 1996).

Introduction 3

Table 1.2. Soluble solids in juice (Walford, 1996)

Groups Compounds % Brix Sugars Sucrose 81–87

Reducing sugars 3–6 Oligosaccharides 0.06–0.6 Polysaccharides 0.2–0.8

Salts Inorganic salts 1.5–3.7 Organic non-sugars

Organic acids 0.7–1.3 Amino acids 0.5–2.5 Dextrans 0.1–0.6 Starch 0.11–0.5 Gums 0.02–0.05 Waxes, fats, phospholipids 0.05–0.15 Colourants 0.1

1.3.3 Composition of fibre

Fibre is the dry, water-insoluble matter in the cane (Bureau of Sugar

Experiment Stations, 2001). It typically constitutes about 14% to 19% of the cane

and includes any dirt, soil and other insoluble extraneous matter as well. The density

of the fibre is approximately 1530 kg/m3 (Pidduck, 1955). The percentage of sugar in

the cane varies from 8 to 16% and depends to a large extent on the variety of the

cane, its maturity, soil condition, climate and agricultural practices followed. A

proportion of the water in the sugarcane, known as hygroscopic water, is absorbed

into the fibre. According to Foster (1956) the weight of hygroscopic water (brix free

cane water) is typically 25% of the weight of fibre.

1.4 The Milling Process

1.4.1 Overview

The milling process involves the separation of juice from the sugarcane. It

consists of two main processes: cane preparation and extraction. There are two main

extraction processes used in the sugar industry: diffusion and milling. This study is

about the milling process since over 90% of the Australian crop is processed by

milling.

1.4.2 Cane preparation

Sugarcane accumulates sucrose at concentrations up to 62% dry weight (16 to

25% fresh weight) in the storage parenchyma cells of the mature culm

(Gnanasambandam & Birch, 2006). The object of cane preparation is to pulverise

4 Introduction

cane into small pieces for feeding the mills and also to rupture the cells, without

extracting juice. The preparatory devices commonly employed and installed before

the milling tandem are classified into three types-

1. Knives which cut the cane into pieces,

2. Shredder which shred cut cane into long fine pieces,

3. Fibrizer combining the features of (1) and (2).

The percentage of cells opened, which is indicative of the preparation of the

cane is-

1. 50-60% in the case of two sets of knives.

2. 85-90% with a combination of knives and a shredder or a heavy-duty

shredder alone.

3. 75-80% for a fibrizer.

It has been established that fine cane preparation is important for efficient

milling in as much as it exerts influence on mill extraction as well as throughput. In

Australia great emphasis is placed on cane preparation by employing heavy duty

shredders and the power consumed for cane preparatory devices is very high, around

70–90% of the total power for mills (Kulkarni, 2009).

1.4.3 The milling train

The milling process essentially involves the removal of juice from sugarcane

by squeezing the cane between pairs of large cylindrical rolls in a series of milling

units collectively called a milling train, as shown in Figure 1.1. The first milling unit

in the milling train is generally identified as #1 mill; the second milling unit is

generally identified as #2 mill, and so on. The last milling unit is generally called the

final mill. The milling units between the first and final mills are collectively known

as the intermediate mills.

Introduction 5

Figure 1.1. Schematic diagram of milling train

After passing through a pair of rolls and expressing juice, the remaining

sugarcane material is known as bagasse. Only the first milling unit in the milling

train processes prepared cane. The remaining milling units process bagasse. After

being processed by a mill, bagasse typically consists of 30% to 50% of fibre, 45% to

60% of water and a diminishing quantity of brix as subsequent milling units process

the bagasse. Although prepared cane and bagasse are defined as different materials,

this thesis uses bagasse as a general term to collectively refer to both prepared cane

and bagasse.

To aid in the extraction of juice from the much drier bagasse, water or diluted

juice is added to the bagasse before it enters the milling unit in a process called

imbibition. The water or juice added to the bagasse is called imbibition water or

imbibition juice. The common version of the imbibition process, called compound

imbibition, is shown in Figure 1.1. Imbibition water is added to the bagasse entering

the final milling unit at a rate that is typically 200% to 300% of the rate of fibre

passing through the milling train. The juice expressed from the final milling unit is

used as imbibition juice for the second last milling unit. The juice from the second

last milling unit is then used as imbibition juice for the third last milling unit. This

process continues back to the second milling unit.

After first passing through a juice screen to remove most of the fibre in the

juice, the juice from the first and second milling units, called mixed juice, is taken

away for processing into sugar. The fibre removed in the juice screen, called cush, is

1

2 3 4 5

JUICE SCREEN

CANE

MIXED JUICE IMBIBITION

FINAL BAGASSE

CUSH IMBIBITION JUICE

6 Introduction

returned to the milling train, usually before the second milling unit. The bagasse

from the final milling unit is taken away for further processing, typically for burning

in the boiler furnace.

1.4.4 The six-roller mill

Although there are many mill designs used around the world, in Australia, the

six-roller mill is widely used. The main crushing part of the milling units consists of

three rollers shown in the bottom right hand side of Figure 1.2. These rollers are

named the top roller, feed roller and the delivery roller and are collectively called the

mill rollers. There are two nips in this part of the milling unit known as the feed nip

and the delivery nip. A trash plate scrapes the bagasse away from the feed roller and

helps to feed the bagasse into the delivery nip.

A pressure feeder consists of two rollers known as the top pressure feeder

roller and the bottom pressure feeder roller. The pressure feeder feeds bagasse

through the pressure feeder nip along a pressure feeder chute into the main crushing

part of the milling unit.

Over 95% of Australian milling units also contain a sixth roller known as the

underfeed roller. The underfeed roller forms a nip with the top pressure feeder roller

that is known as the underfeed nip. Bagasse exiting the underfeed nip is fed into the

pressure feeder. A vertical or nearly vertical closed feed chute feeds the bagasse into

the underfeed nip (Kent, 2003).

The six rollers are arranged so that the prepared cane or bagasse passes through

four squeezes or nips between the entry and the exit of the milling unit. The

collection of milling units is known as a milling train.

Introduction 7

Figure 1.2. Typical layout of a milling unit (Neill et al., 1996)

1.4.5 Extraction theory

The fundamental equation for the mass balance of the entire milling train is

given by equation 1.1. Evaporation is ignored in the equation.

푚̇ + 푚̇ = 푚̇ + 푚̇ 1.1

Where: 푚̇ is mass rate of cane (kg/s),

푚̇ is mass rate of added water (kg/s),

푚̇ is mass rate of mixed juice (kg/s),

푚̇ is mass rate of final bagasse (kg/s).

This equation assumes that the fifth mill is the final mill, consistent with Figure

1.1. Not all milling trains have five mills. There are milling trains in Australia with

four and six mills and milling trains overseas with three and seven mills.

Extraction is calculated by the percentage of sucrose extracted from cane. The

percentage of the sucrose in the original cane removed with the mixed juice is termed

“sucrose extraction”, EP (Rein, 2007).

8 Introduction

퐸 = 100 ×

푚̇푚̇ 1.2

Where: 푚̇ is mass flow of sucrose in mixed juice (kg/s),

푚̇ is mass flow of sucrose in cane (kg/s).

The mill engineer does not have control over the sucrose but has control over

the brix and hence it would be reasonable to report brix extraction.

퐸 =

푚̇푚̇ × 100 1.3

Where: 퐸 is brix extraction,

푚̇ is mass flow of brix in cane (kg/s),

푚̇ is mass flow of brix in mixed juice (kg/s).

On the basis that the fibre rate through each milling unit is the same, the brix

extraction of the single milling unit is determined from:

퐸 =

푃 ( )푃 ( )

− 푃푃

푃 ( )푃 ( )

1.4

Where: 퐸 is brix extraction of nth mill,

푃 ( ) is percent brix of (n-1)th mill,

푃 ( ) is percent fibre of (n-1)th mill,

푃 is the percent brix of nth mill,

푃 is the percent fibre of nth mill.

1.4.6 Simulation and modelling of the sugar factory

Process modelling of raw sugar factories has been undertaken using different

techniques over the years for the purpose of understanding and improving the

performance of the unit operations. For a mathematical model to be useful, it must be

sufficiently complete and accurate to represent the system in the range of variables to

Introduction 9

be studied. Today, several commercial process modelling packages are available

with the ability to incorporate various unit operations and integrate them together.

Software packages like HYSIS (Aspentech, 2012) and SUGARS (SUGARS

International, 2011) have been used in the past to simulate sugar factory processes.

The HYSIS software is general process modelling software and does not contain a

specific milling unit model. The SUGARS package does contain a milling unit

model, but does not allow in-house process knowledge to be incorporated into the

models (Peacock, 2002).

The use of process modelling software in the South African sugar industry was

discussed and demonstrated by Peacock (2002). SIMULINK is a commercial

software system overlaid on the MATLAB programming language, which is widely

used in modelling, simulation and analysing steady state and dynamic systems using

block diagrams. Peacock explored the SIMULINK system for the mass and energy

balance model in the cane diffuser system (an alternative extraction process to the

milling process). The model, which was at an early stage, consisted of a simple

material and enthalpy balance.

The use of a spreadsheet with constraint equations and Newton-Raphson

technique to solve heat and mass balance models was demonstrated by Loubser

(2004). Loubser discussed the applicability of such techniques, when commercial

software such as SUGARS and SIMULINK is not available.

SysCAD is a commercial process modelling software package developed by

Kenwalt Australia (KWA) and extensively used in mineral industries (SysCAD). The

software package offers dynamic modelling and simulation of individual unit

operations and the ability to develop “whole of plant” packages. This thesis describes

the application of SysCAD for developing a comprehensive process model of the

milling train of a raw sugar factory.

1.5 Scope of Research

1.5.1 Research problem

Currently, the process modelling of Australian sugar factories utilizes

comprehensive models of individual sugar processing unit operations. These models

are used extensively in research and consulting for sugar factories. Several milling

companies also have their own models for individual stations. However the

10 Introduction

capability to model an entire plant using the proven models for the individual unit

operations does not exist. This is a significant limitation compared with the

modelling capability of other process industries which use commercial modelling

packages with the embedded facility to incorporate industry specific unit operations

models.

Models have been developed to predict the extraction performance of the

milling process. While many desirable features exist in the various models, there is

no single model that incorporates all of the desired features and no model that can

satisfactorily be used as a component on a whole of factory process model. The

unavailability of a proven comprehensive milling process simulation package for the

sugar milling industry has restricted its ability to assess advanced control options for

improved efficiency through dynamic simulation.

1.5.2 Objectives

The prime objective of the project was to develop an enhanced and

comprehensive milling process simulation model to analyse the performance of the

milling train and to assess the impact of changes and advanced control options for

improved operational efficiency. The specific aims of the project are:

To develop an enhanced mill extraction model to address the shortcomings of the

previous models.

To develop a model to track the specific soluble and insoluble cane components

in the milling process and their distribution in juice and bagasse.

To merge the models together to develop a comprehensive simulation model that

represents the milling process.

To incorporate the model in a process modelling and simulation package

“SysCAD” for dynamic simulation and for availability in a “whole of plant”

package.

1.5.3 Individual contribution to the research team

This project was conducted as part of a larger research project to develop a

suitable ‘whole of plant’ simulation to address the shortcomings of previous models.

KWA developed the SysCAD process modelling software which is used extensively

Introduction 11

in the minerals processing industry. The larger research project involves KWA and

QUT working together to develop the sugar SysCAD modelling software.

The research student, along with the KWA team, has developed a milling train

model which is a part of the ‘whole of sugar factory’ process model. The student has

individually contributed to the development of the advanced milling train model and

validated it. The research student has developed the milling unit model in the

SysCAD environment to be applicable in the sugar industry.

1.6 Overview of Thesis

Chapter 1 provides a general overview of sugarcane composition, the milling

process and equipment, the conventional terminology and the process modelling

packages used to simulate sugar factory operations. It describes the scope of the

research and the objectives of the project.

Chapter 2 reviews the previous models that have been used to predict

extraction performance of the milling train in Australia and overseas. It identifies

that the model currently used by the Australian sugarcane factories is the best

extraction model currently available for the purpose. Various other models developed

for mass balance purposes have been explored and their limitations are discussed.

Chapter 3 describes the MILEX model i.e. the enhanced mill extraction

model developed as a part of the research project. The mill performance parameters

have been redefined and an entire milling train with the juice screen is modelled.

The chapter presents the solution method for solving the MILEX model. The model

is applicable in both analytical and predictive mode and both modes have found

application in the sugar industry. The model is tested with a base case and is further

explored with some specific scenarios to study the performance of the milling

process and determine the optimum operating condition.

Chapter 4 describes the cane component model, which tracks the flow of

specific cane components and the factors affecting their distribution in juice and

bagasse. The model explores the soluble and insoluble solids in cane and defines new

mill parameters to track the components. The effect of the soluble and insoluble

solids cane components on extraction performance of the milling train is explored.

12 Introduction

Chapter 5 deals with integrating the model into the SysCAD software in both

analytical and predictive mode. The model is tested with some real time factory data.

Chapter 6 describes the application of such comprehensive models in the sugar

industry through research and consulting and concludes the thesis, providing

recommendations for further work to improve the milling train model.

Literature Review 13

CHAPTER 2: LITERATURE REVIEW

2.1 Introductory Remarks

In the previous chapter, basic extraction concepts were introduced. This

chapter deals with the application of the extraction theory for modelling and

simulation of the milling trains.

The chapter explores the previous extraction models developed to determine

the extraction performance of the milling process, along with the soluble solids and

the insoluble solids models developed. The chapter concludes with a discussion of

the limitations of the described models.

2.2 The MILSIM Model

2.2.1 Introductory remarks

The MILSIM model (Russell, 1968) was developed at the University of

Queensland. The MILSIM model was the first and only milling train simulation

model that has been widely used in the Australian sugar industry. The model has

been extensively used by the Sugar Research Institute for research and consulting

purposes and modified over the years. Russell also developed software to determine

the mill settings of the milling train (MILSET) to calculate filling ratio, an important

parameter in the MILSIM model. However, the MILSET model is outside the scope

of this project and hence is not described in the thesis.

Over the years, researchers at the Sugar Research Institute made modifications

to the original MILSIM model in an effort to make the performance indicators in the

model more relevant to factory engineers. Blaik and Edwards (1994) proposed a

model ‘MILSM93’ which used ‘crushing factor’ for the division of brix in delivery

bagasse and expressed juice. Kent et al (2000) proposed two new models based on

the tests conducted at the Victoria mill in 1997. The first model consisted of new

empirical relationships for reabsorption factor and imbibition coefficient parameters

for the first and bagasse mills. The second model replaced the imbibition coefficient

parameter with crushing factor and ‘mixing efficiency’ to determine the division of

brix in delivery bagasse and expressed juice.

14 Literature Review

Although the original MILSIM model is still being used in the industry, the

new models assisted in understanding the causes for low extraction. The new models

were validated over a small range and hence could not be applied with confidence.

The theory described below is a step by step procedure for determining extraction of

the milling unit of the original MILSIM model developed by Russell (1968). The

modifications of the MILSIM model are described in section 2.3.

2.2.2 Extraction performance parameters

Extraction is accepted to be a volumetric process and extraction performance

of a single milling unit may be calculated from the filling ratio, reabsorption factor

and imbibition coefficient (Russell, 1968). If these three parameters are known, the

output quantities from the milling unit can be calculated from the input bagasse

analysis.

1. Filling ratio

Filling ratio is the non dimensional representation of the delivery setting of the

mill. It is defined as the no void volume of fibre per unit escribed volume (Murry,

1959). The no void volume is defined as the volume without air entrapped in it.

퐶 =

푉̇푉̇

2.1

Where: 푉̇ is volume rate of cane fibre (m3/s),

푉̇ is the volume rate escribed by the top and delivery rollers of nth mill

(m3/s).

The escribed volume is defined as the product of the work opening (W), the top

roller surface speed (S) and the length of roller (L).

푉 = 퐿 × 푊 × 푆 2.2

Where: 퐿 is length of roller (m),

푊 is work opening (m),

푆 is top roller surface speed (m/s).


The filling ratio definition for a mill is based on bagasse occupying the

escribed volume of the delivery roller. In practice, filling ratio is a function of

mechanical loading and control parameters of the milling unit.

2. Reabsorption factor

If the no-void volume of bagasse leaving a mill in unit time is measured, it is

found in most cases to be in excess of the escribed volume of the delivery nip.

Hence, juice which ideally should be extracted must be passing the feed side of the

delivery rollers through the work opening to the delivery side (Crawford, 1957).

Reabsorption factor is a term used to describe this phenomenon. Reabsorption factor

represents the volumetric juice extraction performance of the mill.

퐾 =

푉̇푉̇

2.3

Where: 퐾 is reabsorption factor of nth mill,

푉̇ is no void volume rate of bagasse of nth mill (m3/s).

Egeter (1928) postulated that the juice in the feed blanket moved through the

nip at a speed in excess of roller surface speed, while the fibre moved forward at the

roller surface speed. Egeter called the ratio of hypothetical juice velocity to the roller

surface speed, the ‘squirting factor”.

3. Imbibition coefficient

The imbibition coefficient is defined as the ratio of the actual brix extraction to

the theoretical brix extraction of the mill (assuming perfect mixing of the imbibition

liquid and residual juice in bagasse from the previous mill) (Munro, 1963, 1964). It is

the measure of the performance of the mill in extracting brix.

퐼 =퐸퐸 2.4

Where: 퐼 is imbibition coefficient of the nth mill,

퐸 is brix extraction of the nth mill,

퐸 is theoretical brix extraction of the nth mill.


From the above definition, it would be expected that the imbibition coefficient

is a strong function of the imbibition level and the brix levels of both the imbibition

juice and the bagasse to which it is applied. Also, the value of imbibition coefficient

approaches unity as the brix fractions of imbibition and residual juice approach each

other.

Measurements show that the imbibition coefficient decreases along the

milling train, because the difference between the brix fraction of the juice in the

bagasse, and the brix fraction in the imbibition applied to it increases for later mills

in the train. This gradient occurs because, after initial cell breakage in the preparation

devices, there is only a small increase in the proportion of broken cells along the

train, and extraction of brix is mainly a washing process (Murry & Russell, 1969).

2.2.3 Determining extraction of a single milling unit

Figure 2.1 shows the flow of brix and water through a single milling unit ‘n’.

The input quantities, 푃 ( ) , 푃 ( ) and 푃 ( ) , are the mass percentage of

brix, moisture and fibre respectively in bagasse from the preceding mill. The output

quantities 푃 , 푃 and 푃 , are the mass percentages of brix, moisture and fibre

respectively in bagasse. 푃 is the mass percentage of brix in imbibition and 푃 is

the mass percentage of moisture in imbibition. The extraction quantities, 푋 and

푋 are the mass fractions of brix and moisture in juice respectively.

Figure 2.1. The inlet and outlet quantities of single mill ‘n’

PInB PInW

Output PBnB, PBnW PBnF

Expressed Juice XJnB XJnW

Input PB(n-1)B PB(n-1)W PB(n-1)F

Imbibition


Using the three mill parameters viz. filling ratio, reabsorption factor and

imbibition coefficient defined in section 0, the composition of delivery bagasse and

expressed juice may be calculated from the given feed quantities. It is assumed that

no fibre passes into the expressed juice, so that the weights of fibre in the feed and

delivery bagasse are identical.

푚̇ ( ) = 푚̇ 2.5

Where: 푚̇ ( ) is the mass rate of fibre in delivery bagasse of the preceding mill

(kg/s),

푚̇ is the mass rate of fibre in delivery bagasse of the nth mill (kg/s).

From equation 2.1 and 2.3, the ratio of reabsorption factor to the filling ratio is

given by,

퐾퐶 =

푉̇푉̇

2.6

Where: 퐾 is the reabsorption factor of the nth mill,

퐶 is the filling ratio of the nth mill,

푉̇ is the volume of delivery bagasse of the nth mill,

푉̇ is the volume of fibre in delivery bagasse of the nth mill.

Rearranging equation 2.6, we get,

푉̇ =퐾퐶 × 푉̇ 2.7

The volume of juice in bagasse is given by,

푉̇ = 푉̇ − 푉̇ 2.8

Where: 푉̇ is the volume of juice in delivery bagasse of the nth

mill.

Substituting equation 2.7 into 2.8 and rearranging the terms we get,

푉̇ = 푉̇ ×퐾퐶 − 1 2.9


The theoretical brix fraction in bagasse is calculated from,

푌 =

푋 + 푋 ( )

푋 + 푋 ( ) + 푋 + 푋 ( ) 2.10

Where: 푌 is the theoretical brix fraction in bagasse of the nth mill,

푋 is the mass fraction of brix in imbibition of the nth mill,

푋 ( ) is the mass fraction of brix in bagasse of the preceding mill,

푋 is the mass fraction of moisture in imbibition of the nth mill,

푋 ( ) is the mass fraction of moisture in bagasse of the preceding mill.

The theoretical density of juice in bagasse is determined from the brix-fraction-

density factor (Russell, 1968). The theoretical mass of juice in bagasse is calculated

from,

푋 = 푉̇ × 푑 2.11

Where: 푋 is the theoretical mass of juice in bagasse of the nth mill,

푑 is the theoretical density of juice in bagasse of the nth mill.

The theoretical mass of brix in bagasse is calculated from,

푋 = 푋 × 푌 2.12

Where: 푋 is the theoretical mass of brix in bagasse of the nth mill.

The actual mass of brix in bagasse is calculated from,

푋 = (1 − 퐼 ) × 푋 ( ) + 퐼 × 푋 2.13

Where: 푋 is mass of brix in bagasse of nth mill,

퐼 is the imbibition coefficient of nth mill.

The brix fraction-density factor as demonstrated by Russell (1968) is given by,

푌 푑 =푋푉̇

2.14

Where: 푌 푑is the brix fraction-density factor of nth mill.


The actual brix fraction is bagasse is determined from the brix fraction density

table (Russell, 1968). The mass of juice in bagasse is calculated from,

푋 =푋푌 2.15

Where: 푋 is actual weight of juice in bagasse of nth mill.

The mass of moisture in bagasse is determined from,

푋 = 푋 − 푋 2.16

Where: 푋 is mass of moisture in bagasse of nth mill.

The brix and moisture in expressed juice is calculated from the following

equations:

푋 = 푋 ( ) + 푋 − 푋 2.17

푋 = 푋 ( ) + 푋 − 푋 2.18

Where: 푋 is mass of brix in expressed juice of nth mill,

푋 is mass of moisture in expressed juice of nth mill.

The delivery bagasse composition is determined from:

푃 =푋

푋 + 푋 + 푋 × 100 2.19

푃 =푋

푋 + 푋 + 푋 × 100 2.20

푃 = 100− 푃 − 푃 2.21

Where: 푃 is percent brix in delivery bagasse of nth mill,

푃 is percent moisture in delivery bagasse of nth mill,

푃 is percent fibre in delivery bagasse of nth mill.

2.2.4 Calculating reabsorption factor & imbibition coefficient

The extraction model can be used in two modes and both have found

application in the sugar industry. In the analytical mode, if bagasse analysis results

are available, the extraction model can be used to calculate the performance


parameters for each milling unit. These performance parameters can then be assessed

to determine how well each milling unit is performing. The predictive mode involves

defining performance parameters for each milling unit so that the extraction model

can predict bagasse analysis for each mill and hence predict extraction for the milling

train. The predictive mode of the extraction model was demonstrated in the previous

section where filling ratio, reabsorption factor and imbibition coefficient were used

to determine bagasse analysis.

Determining the reabsorption factor and imbibition coefficient for each milling

unit requires the same equations described in the previous section. Filling ratio is not

necessarily a performance parameter as it remains an input in both modes of

operation. Filling ratio is determined from the loading and control parameters of the

milling unit as described in section 0. Equation 2.6 to 2.21 can be rearranged and

used to determine the reabsorption factor and imbibition coefficient from bagasse

analysis.

The values to be used for filling ratio, reabsorption factor and imbibition

coefficient must be properly selected if accurate results are to be obtained from this

theory. The filling ratio depends on the mill speed and delivery settings and many

methods have been put forward for calculating the required setting. Most of these

depend either on arbitrarily selected desired bagasse moistures or merely on filling

ratios from practical experience. Such arbitrary selection takes no account of the

mechanical ability of the particular milling unit and hence correct values may not be

obtained. It is necessary to obtain correct filling ratios for the milling units, if good

predictions are desired (Murry & Russell, 1969).

2.2.5 Reabsorption factor & imbibition coefficient multipliers

Murry & Russell (1969) defined reabsorption factor and imbibition coefficient

multipliers. These multipliers are the ratio of the quantity (reabsorption factor or

imbibition coefficient) to the value predicted by an empirical equation for that

quantity. The multiplier thus gives an indication of the deviation of the actual values

from an expected value.

In case of reabsorption factor, a function was developed which predicts the

reabsorption factor given the filling ratio, mill surface speed and mill number and the

predictions were quite reliable. The imbibition coefficient is predicted from a


function utilizing imbibition brix, imbibition level and brix fraction (brix to juice

ratio) of delivery bagasse from the previous mill. The imbibition coefficient function

is not well validated and the predictions are not entirely reliable. Such failures occur

when changes in the operating parameters such as imbibition level are large and

outside the range of parameters from which the function was developed.

2.2.6 Concluding remarks

The MILSIM model predicts the extraction performance of the milling train

using the input bagasse analysis and the three mill performance parameters. The

model is based on a convenient but incorrect assumption of constant fibre rate. Kent

(2001) explored the assumption in later years with the fibre flow model and though

Kent mentions that the factor was taken into account in the empirical models through

the use of the multipliers for calibration purposes, a substantial error can be

anticipated in the analytical and predictive mode of the extraction model.

2.3 The modified MILSIM extraction model


The Sugar Research Institute made modifications to the existing milling train

model (MILSIM) by replacing the imbibition coefficient with other two performance

indicators; “crushing factor” and “mixing efficiency”. The crushing factor deals with

the preparation that occurs in a milling unit, while mixing efficiency determines how

well the imbibition liquid mixes with the feed bagasse in a milling unit.

2.3.2 Development of new mill performance parameters

Kent (1997) explained the modifications made in the existing milling train

model (MILSIM) and the disparity between the original MILSIM model and the

modified MILSIM model. Imbibition coefficient describes the mixing of imbibition

liquid with the bagasse. The barrier to perfect imbibition is mostly due to the closed

cells in the bagasse which are not available for mixing.

Edwards (1995) put forward a concept of crushing factor to model the opening

of closed cells in the feed. Crushing factor describes the number of cells opened after

passing through a milling unit. The concept for the mixing process was that the brix

and water in the imbibition could only mix with the ‘free’ brix and water in bagasse,

where that ‘free’ brix and water is measured by a test like the POC (Pol in Open


Cells) test as defined by Method 5 of the BSES laboratory manual (Bureau of Sugar

Experiment Stations, 2001). Experiments showed that the mixing was less than

perfect; hence to overcome this imperfect mixing, mixing efficiency was introduced.

This modified model uses both crushing factor and mixing efficiency to replace

imbibition coefficient.

In the modified MILSIM model, the crushing process is divided into four

stages.

1. Mixing

2. Separation

3. Preparation

4. Remixing

The mixing stage involves the imbibition juice mixing with the bagasse from

the previous mill. The imbibition juice mixes with the portion of juice in open cells.

The separation of juice involves the expression of juice in the bagasse. The

preparation stage involves opening of some closed cells in the bagasse. The remixing

stage is used only so that all the juice in open cells leaving the milling unit is of

uniform brix. The processes separation, preparation and remixing occur more or less

simultaneously within the milling unit.

The extraction performance indicators are:

1. Mixing efficiency (휂 )

휂 = 1−

퐵퐵

2.22

Where: 퐵 is the brix in open cells per unit mass of fibre prior to adding

imbibition in the delivery bagasse from the previous mill,

퐵 is the brix in open cells per unit mass of fibre which does not

participate in mixing.

A high mixing efficiency (approaching unity) indicates that most of the juice in

open cells in the bagasse mixes with the imbibitions.


2. Reabsorption factor (푘 ) (separation stage)

The reabsorption factor definition is given in section 0. Essentially, the

reabsorption factor describes the difference between the expected bagasse rate and

the actual bagasse rate through a milling unit (Kent, 1997).

3. Crushing factor (퐺 )

The crushing factor describes the preparation stage and is defined by:

퐺 =

퐵 − 퐵퐵 2.23

Where: 퐵 is the brix in closed cells per unit mass of fibre from the previous mill,

퐵 is the brix in closed cells per unit mass of fibre from the mill.

A high crushing factor (approaching unity) means a large percentage of the

closed cells are opened in a milling unit.

2.3.3 Analytical mode

The data required to solve the analytical mode of the modified MILSIM model

is almost the same as the data for the original MILSIM model. The only difference is

the need for the brix in open cells measurement for prepared cane and bagasse from

each milling unit. The brix in open cells measurement determines relative sizes for

the mixing efficiency and the crushing factor.

It is usual to assume the brix in open cells is the same as the pol in open cells.

While brix in open cells can be measured in the same way as pol in open cells, the

brix measurement is less precise than the pol measurement (Kent et al., 2000). The

lack of precision of the pol in open cells measurement makes the mixing efficiency

and crushing factor calculations less precise than the imbibition coefficient

calculation. This is due to the fact that there is little data with which to compare the

mixing efficiency and crushing factor (Kent, 1997).

2.3.4 Predictive mode

The predictive mode of the model involves defining the performance

parameters for each milling unit, so that the model can be applied to predict the

bagasse analysis.


To use the extraction model for prediction purposes, it is necessary to know

how the performance parameters vary as the operating parameters in the model are

changed. The reabsorption factor is defined as a function of delivery nip compaction,

roller speed and bagasse fineness (Russell, 1968). The mixing efficiency and the

crushing factor empirical relationships were defined in later years at the Sugar

Research Institute. As the parameters were validated over a small range of values

they could not be applied confidently outside the range. This caused a limitation in

using the model in the predictive mode.


The modified MILSIM model can provide insight into the causes of low

extraction and methods to improve extraction through the use of performance

parameters mixing efficiency and crushing factor.

The model was not well validated, and consequently this caused a limitation in

the application of the model. Further work is required before this model could be

confidently used.

2.4 Wienese’s extraction model


Wienese (1990, 1994) developed a model to determine the extraction

performance of the milling train. The model used similar mill performance

parameters to those used in the MILSIM model. The prime difference between the

MILSIM model and the South African model was that Wienese had considered the

fibre in juice to be a substantial value and proposed a new mill performance

parameter named separation efficiency to account for it. Hence the model overcame

the constant fibre assumption.

2.4.2 Mill performance parameters

The South African model defines the performance parameters separation

efficiency, reabsorption coefficient, and imbibition efficiency and calculates them

from known factory data.

1. Reabsorption coefficient

The reabsorption coefficient is essentially the same as the reabsorption factor,

representing the split of juice in delivery bagasse and expressed juice.


푅푒푎푏푠표푟푝푡푖표푛푐표푒푓푓푖푐푖푒푛푡 =푓푖푏푟푒%푏푎푔푎푠푠푒푖푛푡ℎ푒푚푖푙푙

푓푖푏푟푒%푏푎푔푎푠푠푒푎푓푡푒푟푡ℎ푒푚푖푙푙 2.24

2. Imbibition efficiency

Wienese defined imbibition efficiency based on the brix fraction of expressed

juice and input bagasse.

퐼푚푏푖푏푖푡푖표푛푒푓푓푖푐푖푒푛푐푦 =푏푟푖푥%푙푖푞푢푖푑푖푛푗푢푖푐푒푏푟푖푥%푙푖푞푢푖푑푖푛푐푎푛푒

× 100 2.25

3. Separation efficiency

It is known that some fibre ends up in the expressed juice stream from each

mill in the form of suspended solids. Wienese (1990) defined separation efficiency as

one hundred minus suspended solids% expressed juice and assumed it to be constant

for all mills. The definition was later altered by Wienese (1994) and the new

equation considered the input fibre% cane and output fibre% juice rate in the milling

train.

푆푒푝푎푟푎푡푖표푛푒푓푓푖푐푖푒푛푐푦 =푓푖푏푟푒%푐푎푛푒 − 푓푖푏푟푒%푗푢푖푐푒

푓푖푏푟푒%푐푎푛푒× 100 2.26

The definition of separation efficiency is not entirely correct since fibre% cane

and fibre% juice have different denominators. Hence, one shouldn’t strictly be

subtracted from the other. It does, however, have the desirable characteristic that

separation efficiency is 100% if there is no fibre in juice.

2.4.3 Model application

Wienese (1994) developed the extraction model to analyse the effect of

imbibition on extraction. The data required for a mass balance around the milling

unit are the moisture% bagasse, brix% bagasse and the fibre (suspended solids) %

expressed juice. These data were used to solve the analytical mode to determine the

mill performance parameters and the predictive mode solves the model to determine

the bagasse analysis, similar to the MILSIM model.


Wienese (1994) developed the model to determine the optimum imbibition rate

in the milling process. The South African extraction model proposed the separation

efficiency to account for the fibre in the expressed juice stream. The concept could

not provide the mill engineer with actual fibre flows in the milling train. Wienese’s


milling train model did not include the juice screen, which is used to reduce and to

some extent control the amount of fibre in mixed juice. The return stream from the

juice screen is returned to the milling train usually before the second mill. The return

stream from the juice screen has high fibre content and affects the overall fibre rate

and hence affects extraction.

2.5 Other mass balance models


The mass balance of the milling train is the initial step in modelling the milling

process. Several mass balance models have been developed to model the milling

process. The mass balance models are explored in this section.

2.5.2 The fibre flow model

The MILSIM model (Russell, 1968) assumes that all the fibre ends up in

bagasse. The model assumes that each milling unit processes the same fibre rate. The

assumption is convenient, but incorrect. The assumption that each milling unit

processes the same fibre rate is based on an assumption that the amount of fibre is

expressed with juice and subsequently recycled through the milling train in the

imbibition system is negligible.

Kent (2001) developed a mass balance model to examine fibre flow through a

milling train. The model includes the flow of fibre suspended in juice streams that is

recirculated along the milling train as part of the imbibition system. Taking into

account the recirculated fibre, the model is capable of estimating the fibre rate

recirculation for most milling train calculations.


Figure 2.2. Milling train fibre mass flows

Figure 2.2 shows the fibre flows in a five-mill milling train. In Figure 2.2, ‘C’

represents cane, “B” represents bagasse, ‘J” represents expressed juice, ‘Js’

represents the return stream from the juice screen, ‘JM’ represents mixed juice and

‘I” represents added water. The numbers refer to respective mills. The return stream

from the juice screen is shown to discharge before #2 mill, the most common (but

not universal) position. In total, there are 14 streams included in the model: cane,

added water, bagasse and expressed juice from the five milling units, the return

stream from the juice screen and mixed juice. The following equations apply for the

fibre mass flows. In the following equations, the general form of any parameter

is푞 , where q is a quantity, p is a product, and c is a component.

Mass must be conserved around each milling unit and the juice screen.

푚̇ = 푚̇ + 푚̇ 2.27

푚̇ = 푚̇ + 푚̇ + 푚̇ − 푚̇ 2.28

푚̇ = 푚̇ + 푚̇ − 푚̇ 2.29

푚̇ = 푚̇ + 푚̇ − 푚̇ 2.30

푚̇ = 푚̇ + 푚̇ − 푚̇ 2.31

JM

C B1 B2 B3 B4 B5

J1 J2 J3 J4 J5

Js

I

Legend: C Cane I Imbibition B Bagasse streams J Expressed juice streams Js Return stream from juice screen JM Mixed juice stream

Juice Screen


푚̇ = 푚̇ + 푚̇ − 푚̇ 2.32

푚̇ represents the fibre flow for a stream x, where x is one of the streams listed

in Figure 2.2. These equations ignore spillage of fibre and juice that could affect the

fibre and total flow equations and evaporation of water from the milling train that

could affect the total flow equations.

For product stream p, the fibre fraction (푃 ) is determined from,

푃 =

푚̇푚̇ 2.33

Where: 푚̇ is the total mass flow in product stream p.

For each of the 14 streams, the model accounts for three parameters; the total

flow, the fibre flow and the fibre content of the flow. Consequently, there are total 42

parameters in the model. Generally, the cane rate, the added water and the 14 fibre

content will be inputs to the model, leaving a total of 26 unknown parameters. These

26 unknown parameters are determined by solving 26 equations: equation 2.27 to

2.32 which define the fibre flows, the equivalent six equations that define the total

flow and equation 2.33 for the 14 streams. Once the model equations are solved,

equation 2.34 to 2.38 can be used to calculate the total fibre rates to be processed by

the individual milling units.

푚̇ = 푚̇ 2.34

푚̇ = 푚̇ + 푚̇ + 푚̇ 2.35

푚̇ = 푚̇ + 푚̇ 2.36

푚̇ = 푚̇ + 푚̇ 2.37

푚̇ = 푚̇ + 푚̇ 2.38

Where: 푚̇ is mass flow of fibre in feed to the nth mill.

Routine factory performance data can generally provide most of the required

input parameters for the model; the cane and added water rates and the fibre content

of each of the bagasse streams. The fibre content of mixed juice is also readily

available from many factories. The less well known parameters are the fibre contents

of the mill expressed juice streams and the fibre content of the return stream. Kent


(2001) provided typical values for these parameters. Table 2.1 shows the parameter

input values used by Kent (2001).

Table 2.1. Standard values for milling unit fibre analysis (Kent, 2001)

Product stream Fibre (% stream) Cane 14 #1 mill bagasse 30 #2 mill bagasse 36 #3 mill bagasse 41 #4 mill bagasse 44 #5 mill bagasse 47 Expressed juice 2, 3 Return stream 8, 12 Mixed juice 0.2, 0.3 Imbibition (225% fibre) 0.00

The mass balance model described above was used to determine the fibre rate

processed by each milling unit in a five-mill milling train for the data as shown in

Table 2.2. Kent (2001) reported that the second mill processes the highest fibre rate

as the mill processes the fibre in expressed juice from #1 mill, #2 mill and #3 mill.

Table 2.2. Effect of fibre content in juice on milling unit fibre rate (Kent, 2001)

Fibre content (%) Fibre rate (% of #1 mill fibre rate) Expressed juice

Juice screen return

Mixed juice #2 mill #3 mill #4 mill #5 mill

3 8 0.2 129 116 115 106 0.3 128 116 114 106

12 0.2 124 116 115 106 0.3 123 116 114 106

2 8 0.2 115 110 109 104 0.3 115 109 108 103

12 0.2 114 110 109 104 0.3 113 109 108 103

The fibre flow model explored the assumption of constant fibre rate in the

milling train. The model considered only fibre and juice and so did not split juice

into its constituent moisture and brix flows. The model proved to be beneficial to

ongoing research to better understand the milling process.


2.5.3 Wienese’s model

Wienese (1995) proposed a mass balance model for a milling train. The model

is based on the assumption of three constant ratios:

1. 푓푖푏푟푒%푗푢푖푐푒푓푖푏푟푒%푏푎푔푎푠푠푒

2. 푏푟푖푥%푗푢푖푐푒푏푟푖푥%푏푎푔푎푠푠푒

3. 푚표푖푠푡푢푟푒%푗푢푖푐푒푚표푖푠푡푢푟푒%푏푎푔푎푠푠푒

For a given input three independent output variables are required in order to

complete mass balances on fibre, brix and water around the extraction plant.

Wienese mentioned that any change in the concentration of fibre, brix and

water in the input affects the concentration of these components in the output in the

same direction. This means that at a constant fibre throughput, an increase in brix%

input results in an increase in both brix% bagasse and brix % juice i.e. a constant

brix% input must give a constant brix% bagasse and brix% juice. This feature of the

model also applies to fibre and water. The constant ratios obtained can be used to

calculate the output composition given a certain input.

The model proves to be useful in estimating the output quantities of the milling

train when less information is known. Its application is restricted as the constant ratio

assumptions are not entirely correct and cannot be calibrated to take into account

known results. The model determines the final bagasse and mixed juice composition

but does not determine the individual milling unit mass flows.

2.5.4 Loubser’s model

Loubser (2004) proposed a mass balance model of the milling train, as shown

in Figure 1.1 on page 5 and solved it using mass balance equations and the Newton-

Raphson technique in an Excel spreadsheet. The model is an efficient mass balance

model of the milling train. The model includes the return stream from the juice

screen and monitors the flow of brix, moisture and fibre in the milling train.

The mass balance equations were developed using the material balance for

overall mass, brix and fibre. The model is similar to the fibre flow model developed

by Kent (2001), although Loubser’s model was able to determine the brix, moisture


and fibre mass flows unlike Kent’s model which focussed on the fibre mass flows in

the milling train.

For each of the 14 streams, the model accounts for three parameters: total flow,

brix flow and fibre flow. Hence we have 42 unknown parameters in the model. There

are 21 inputs to the model, similar to those of Kent (2001), leaving 21 unknowns.

Equations 2.27 to 2.32 show the fibre mass flows in the milling train. Similar sets of

equations apply for the total and brix mass flows in the milling train, giving a total of

18 equations. Hence we have 21 unknowns and 18 equations. It is logical to assume

the brix and fibre fraction of the imbibition stream as zero, accounting for two of the

three remaining unknowns. Hence the equation set is under defined by one equation.

To provide the extra equation, Loubser (2004) made an assumption that the brix to

water ratio of the juice arriving at the juice screen was the same as that of the juice

leaving the screen. The model determines the overall mass balance of the milling

train, but the model does not determine the extraction performance of milling train.


The MILSIM model (Russell, 1968) does a good job of tracking the flow of

brix and moisture in the milling train, although the constant fibre assumption is not

entirely correct. The fibre flow model (Kent, 2001) is a complete model describing

the fibre and juice flows. Loubser (2004) extended Kent’s model, splitting juice into

brix and water and made a convenient assumption of similar brix fraction in cush

return and mixed juice streams. The separation efficiency parameter, developed and

added to the milling train model by Wienese (1995), is a suitable basis for an

additional performance parameter to describe the quantity of fibre in juice.

2.6 Soluble solids model


The soluble solids (brix) in cane include around 80-90% sucrose and 10-20%

impurities. These soluble impurities are further classified as reducing sugars, organic

matter, inorganic compounds, nitrogenous bodies etc as shown in Table 1.2 on page

3.

Few models have been developed to track the flow of soluble solids; in

particular sucrose in the milling process. The models that do exist are described

below.


2.6.2 Sucrose

The most important component in the cane from the sugar production point of

view is the sucrose and several models have been developed to determine the sucrose

extraction of a milling train. The previous work done in determining the sucrose

extraction of the milling process and the limitations and drawbacks of such models

has been explored.

Lionett (1981) proposed an empirical equation to determine the sucrose

extraction from brix extraction based on mixed juice purity and cane purity. The

monthly data from the milling tandems of several South African sugar factories

showed that the ratio of cane purity to mixed juice purity is proportional to the

sucrose extraction. The equation was put into a general form by Wienese (1990) and

is described below.

푆푢푐푟표푠푒푒푥푡푟푎푐푡푖표푛 = √1919 + 187.6 × 퐵푟푖푥푒푥푡푟푎푐푡푖표푛 − √1919 2.39

Wienese (1995) attempted to extend his extraction model to determine the

sucrose extraction of milling tandem and diffusers. The equation was of a general

form based on Lionett’s equation, consisting of three constants determined from

three ideal conditions. The equation is described below.

푆푢푐푟표푠푒푒푥푡푟푎푐푡푖표푛

= 푎 × 퐵푟푖푥푒푥푡푟푎푐푡푖표푛 + 푏 × 퐵푟푖푥푒푥푡푟푎푐푡푖표푛 + 푐 2.40

Where: “a”, “b” and “c” are constants.

In order to calculate these constants three brix extraction results and their

corresponding sucrose extraction results must be known. The first is 0% brix

extraction corresponding to 0% sucrose extraction. The second is 100% brix

extraction corresponding to 100% sucrose extraction. The third is the standard brix

extraction corresponding to standard sucrose extraction determined from the data.

With these three constant known, matching sucrose extraction can be calculated from

non-standard data.

A more appropriate model for sucrose was developed by the Sugar Research

Institute in the early 1980’s. This approach was based on the concept of a purity ratio

to determine the pol content of bagasse from a milling unit from the brix content.


Although the purity ratio concept has no theoretical basis, it was convenient as it

gave good results. The purity ratio was defined as:

푍푅 = 푍푍 2.41

Where: 푍푅 is the purity ratio of the nth mill,

푍 is the purity of bagasse of the nth mill,

푍 is the purity of cane.

The purity ratio can be calculated from cane and bagasse analysis data when

running the model in analytical mode. The purity ratio values are then assumed to be

constant when predicting the pol in bagasse from brix in bagasse for a different cane

analysis.

It should be noted that the purity ratio concept has no theoretical basis. It was

adopted in determining the flow of sucrose in the milling train and subsequently the

sucrose extraction as it gave good results. The detailed theory is described in Chapter

4 of the thesis.

2.6.3 Reducing sugars

The most abundant non-sucrose components in cane are the monosaccharides

glucose and fructose, also known as reducing sugars (Rein, 2007).

The major concern with modelling reducing sugars is the potential for

inversion of the sucrose into reducing sugars. The amount of inversion that occurs is

not well known.

Van Der Pol and Alexander (1955) reported that the reason for inversion of

sucrose is the destruction of sucrose by enzymes. Van Der Pol stated that the

potential losses by inversion due to the combined effect of temperature and pH

would be small under normal operating conditions and can be neglected. Rein

(2007) supported this view.

Fernandes (2003) proposed empirical equations to determine the reducing

sugars in mixed juice from mixed juice purity and reducing sugars in cane from

reducing sugars in mixed juice and fibre content in cane. The equations proposed by


Fernandes are documented by Wright et al (2007), although the basis of the

equations are not known.

푃 = 3.641− 0.0343 × 푍 2.42

Where: 푃 is reducing sugars% mixed juice,

푍 is purity of mixed juice.

푃 = 푃 × (1 − 0.01 × 푃 ) × (1.0313− 0.00575 × 푃 ) 2.43

Where: 푃 is reducing sugars% cane,

푃 is fibre% cane.

2.6.4 Ash

The inorganic component of brix in cane is often reported by the method from

which it is analysed (involving combustion of all organic material): ash. Soluble ash

consists of the inorganic salts present in the cane.

Wienese and Reid (1997) proposed an ash mass balance model based on total

ash in cane, total ash in bagasse and soluble ash in mixed juice. The model is based

on the assumption that soluble ash extraction is equal to brix extraction and tracks

the flow of total ash across the entire milling train. The model can be applied to

track the flow of ash in the milling process when less information is available.


The purity ratio concept was found to be a convenient way of modelling the

sucrose component of soluble solids. Once brix in bagasse is known, the purity ratio

can be used to convert brix into sucrose.

The reducing sugars are not routinely measured through the milling process in

sugar factories, although the inversion does affect the sugar production process: in

terms of sugar losses and the removal of these invert sugars in clarification.

Fernandes (2003) has provided some information on the flow of reducing sugars in

the milling process. The model presented by Fernandes is described in greater detail

and critically examined in section 4.3.3.

The assumption that the extraction of soluble ash is equal to brix extraction

(Wienese & Reid, 1997) is not likely to be correct. Since sucrose (the largest


component of brix) extraction is different to brix extraction, it seems unlikely that

soluble ash extraction will be the same as brix extraction. The assumption contradicts

the pol model described by Wienese (1994, 1995) and also does not determine an ash

balance for individual milling units, as required for this study.

Although the above described models have been applied in the sugar industry,

there are no models in existence addressing all the major soluble impurities through a

single milling unit as required in this study.

2.7 Insoluble solids model


The insoluble solids account for 10-20% of the cane, termed as cane fibre. The

cane fibre includes about 80-90% of true fibre and 10-20% of mud solids. The true

fibre, also termed vegetative or cellulosic fibre (Gilfillan et al., 2012), is the actual

plant fibre in cane.

There are insoluble ash components in both true fibre and mud solids. Some

Australian sugar factories have included the measurement of insoluble ash in cane in

the routine factory analysis. However, it is not common to measure either true fibre

or mud solids in cane.

2.7.2 Bagasse production model

Kent (2010b) developed a model to determine bagasse production from an

insoluble solids mass balance. The model is based on the measurement of total and

insoluble ash in cane, total ash in bagasse and mud solids in mud. The model

includes an extra ash term since ash doesn’t balance without it.

Although the model was developed to determine bagasse production, it has the

potential to also determine the split of true fibre and mud solids between final

bagasse and mixed juice. The model could be applied to the overall milling train

mass balance but not to individual milling units.

2.7.3 Wright’s model

Wright (2003) proposed an insoluble solids model to determine the bagasse

production. The model used similar parameters as Kent (2010b) but had an extra

parameter viz. dirt in cane which is not routinely measured in the Australian sugar

factory. Also the model did not have the potential of splitting the true fibre and mud


solids. Wright has assumed that the split of dirt (mud solids) in final bagasse and

mixed juice is 60:40.


The bagasse production model tracked the ash and non-ash components and

true fibre and mud solids components of the insoluble solids but considered only the

overall insoluble solids mass balance of the milling train and not the individual

milling unit insoluble solids mass flows.

The effect of mud solids on the extraction performance of the milling unit is

not well known. This limitation of the previous models was one of the motivations

for this study.

2.8 Dynamic modelling of the milling process


Dynamic modelling involves modelling a process as a function of time. The

merits of dynamic modelling are that, when developed and validated accurately, the

dynamic model can be used to model process variations and process control and

hence can address a much wider range of problems.

The milling process is a continuous process with the cane rate (crushing rate)

being the prime variable and other variables such as mill speed, added water rate etc

dependent on the crushing rate. The cane composition is an independent parameter,

although it is beyond the control of the factory.

The models described earlier in this chapter were static models of the milling

process. The development of a dynamic model of the milling process was

investigated in Australia but the model has not been adopted. The dynamic model of

MILSIM is explored in the next section.

2.8.2 McWhinney’s dynamic model

McWhinney (1973) proposed a dynamic version of the MILSIM model to

simulate the milling process. He used a time delay function to account for the time

lost between successive milling units and a capacity lag function for the mixing of

juice in bagasse and imbibition during transport between successive mills. The

dynamic model was implemented on the University of Queensland GE225 computer

since the static model as reported by Russell (1968) was available on that computer.


The dynamic model could be used to predict the response of a change in

crushing rate provided the static parameters such as filling ratios, mill speeds and

mean imbibition level were known and some estimates of the time delays between

successive mills were available. The response of the mixed juice flow could be

predicted for various changes in cane quality and rate and disturbances in the outside

imbibition flow rate.

McWhinney reported the limitations of the dynamic model due to insufficient

validation of the model. The escribed volume and mill speed of the intermediate

mills were not recorded; hence the fibre rate variations along the milling train could

not be accurately computed.


With advances in process modelling technologies, fast computers and process

modelling packages, dynamic modelling can be more readily undertaken. Although

dynamic modelling does have complex constraints, most process modelling packages

have the capacity to include dynamic modelling of the unit operations.

Although, the dynamic model of MILSIM was not adopted, it has provided

some important information in developing the advanced milling process model.

2.9 Concluding Remarks

The models described above are used extensively in the sugar industry despite

their shortcomings and limitations. The models need to be interpreted using the

combined effort of the mill engineer’s experience to obtain valuable information.

The Loubser (2004) mass balance model has all the right mass balance

features. The main limitation of the model is that it does not include the performance

parameters of the mill to enable it to be used in predictive mode.

The MILSIM (Russell, 1968) model has good and well defined mill

performance parameters for a single milling unit but is based on the constant fibre

rate assumption. The Wienese (1995) concept of separation efficiency can be added

to the MILSIM model to overcome the constant fibre rate assumption. The MILSIM

model coupled with Loubser’s model is the selected path for developing an advanced

model.


The purity ratio concept was found to be valid in determining the sucrose

content of bagasse from individual milling units. To date no efficient model has been

developed to handle the other soluble components of the cane and predict their

effects on extraction performance of a milling unit. The previous models provided a

lot of background knowledge in the field of milling process modelling, of which the

Fernandes’ reducing sugars model and Kent’s bagasse model deserve further

investigation.

In modelling the insoluble solids components true fibre and mud solids, the

Kent (2010b) model handles the insoluble solids mass balance of the overall milling

train.

The dynamic model of the milling process developed by McWhinney (1973)

was studied and provided a starting point in developing a dynamic model of the

milling process.

The MILEX Model 39

CHAPTER 3: THE MILEX MODEL

3.1 Introductory remarks

In Chapter 1 the previous work done in modelling the milling process and the

limitations and drawbacks of these models were explored. In this chapter, an

enhanced and comprehensive mill extraction model is developed to simulate the

fibre, brix and moisture flows through the milling process. This chapter deals with

the development of the MILEX model. The model extends the extraction theory and

extraction performance parameters described in chapter 2.

3.2 The milling unit model


In section 2.2, the MILSIM model was introduced that used filling ratio,

reabsorption factor and imbibition coefficient to split the fibre, brix and moisture

flows in the feed to a mill into delivery bagasse and expressed juice. In this chapter,

an extraction model is described which extends the MILSIM model by accounting

for fibre in juice flows, revises the definitions of the performance parameters to

account for those flows and includes a separation efficiency term similar to that

defined by Wienese (1994).

3.2.2 Corrected filling ratio

Equation 2.1 on page 14 shows that the filling ratio is determined from the

volume of fibre and the escribed volume. In the MILSIM model, the volume rate of

fibre is determined from the cane fibre rate. In this enhanced model, the volume rate

of fibre needs to be determined from the fibre rate through the mill n.

Because some fibre is expressed with the juice, the fibre rate in the feed to the

mill is not the same as the fibre rate in the delivery from the mill. To determine

filling ratio, one fibre rate has to be chosen. The feed to the mill or the delivery from

the mill are the two obvious contenders. To choose between the options,

consideration was given to where in the mill the juice is expressed.

Thaval and Kent (2012) proposed a juice flow model to determine the flow of

juice through a six-roller mill. A typical six-roller mill is shown in Figure 1.2 on

40 The MILEX Model

page 7. The juice flow model calculated the juice expressed at the four nips of the

six-roller mill. Results from the model are shown in Figure 3.1.

Underfeed Pressure Feeder Feed Delivery

Perc

ent j

uice

expr

esse

d

0

10

20

30

40

50

60

70

80

Figure 3.1 Percent juice expressed at the four nips of a six-roller mill

Results of using the model for typical heavy-duty pressure feeder settings show

that most of the juice is expressed in the pressure feeder. Since the pressure feeder is

remote from the mill, a significant portion of the juice is expressed before the

bagasse enters the delivery nip where filling ratio is calculated. Consequently, most

of the fibre to be expressed with the juice will have been expressed before this point.

Consequently, the delivery bagasse fibre rate is considered a better estimate of

delivery nip fibre rate than the feed fibre rate for use in the filling ratio calculations.

Using the revised fibre rate, the volume rate of fibre (m3/s) is calculated from,

푉̇ =

푚̇푑 3.1

Where: 푚̇ is the delivery bagasse fibre rate of the nth mill (kg/s),

푑 is the fibre density (1530 kg/m3) (Pidduck, 1955).

The MILEX Model 41

The escribed volume, determined from the mill length, the delivery nip work

opening and the top roller surface speed, is independent of the fibre rate.

The corrected filling ratio can be calculated by substituting equation 3.1 into

2.1.

퐶 =

푚̇푑 × 푉̇

3.2

Where: 퐶 is corrected filling ratio of nth mill,

푉̇ is escribed volume of nth mill (m3/s).

3.2.3 Corrected reabsorption factor

The MILSIM model defines reabsorption factor as the ratio of volume of

bagasse to the escribed volume, assuming the volume rate of fibre is constant for all

mills. The corrected reabsorption factor is determined from bagasse, fibre density,

corrected filling ratio and bagasse fibre flow.

Rearranging equation 3.2 we get the escribed volume,

푉̇ =푚̇

푑 × 퐶 3.3

Substituting equation 3.3 into 2.3 we get the reabsorption factor,

퐾 =

퐶 × 푉̇ × 푑푚̇ 3.4

Where: 퐾 is corrected reabsorption factor of nth mill,

푉̇ is volume rate of bagasse leaving nth mill (m3/s).

3.2.4 Corrected imbibition coefficient

The imbibition coefficient and the brix extraction are defined by MILSIM as

shown in equation 2.4 and 1.4 respectively. In correcting the imbibition coefficient,

the brix extraction definition has been changed.

With the constant fibre rate assumption eliminated in the MILEX model, the

use of fibre to bring the feed and delivery brix quantities to the same basis is no

longer valid. A more accurate model of brix extraction can be determined from the

mass flows.

42 The MILEX Model

퐸 =

푚̇ ( ) − 푚̇푚̇ ( )

3.5

Where: 푚̇ ( ) is mass flow of brix in bagasse of (n-1)th mill,

푚̇ is mass flow of brix in bagasse of nth mill.

Substituting equation 3.5 into equation 2.4, the imbibition coefficient can be

calculated from:

퐼 =

푚̇ ( ) − 푚̇푚̇ ( ) − 푚̇ 3.6

Where: 퐼 is the corrected imbibition coefficient of nth mill,

푚̇ ( ) is the mass flow of brix in bagasse of (n-1)th mill (kg/s),

푚̇ is the mass flow of brix in bagasse of nth mill (kg/s),

푚̇ is the theoretical mass flow of brix in bagasse of nth mill (kg/s).

3.2.5 Separation efficiency

As discussed in section 2.4.2, Wienese (1990) introduced the concept of a

separation efficiency to describe the fibre flow into expressed juice. The separation

efficiency represents the proportion of total fibre in the feed that is found in the

delivery bagasse so that 100% separation efficiency results in no fibre in expressed

juice. Wienese (1995) presented the most recent definition of separation efficiency

of 100 × % %%

. While the concept of separation efficiency is

considered suitable, the definition is not considered ideal, with fibre% juice being

subtracted from fibre% cane (or fibre% feed).

A revised separation efficiency definition is presented here:

푆 =

푚̇ − 푚̇푚̇ × 100 3.7

Where: 푆 is separation efficiency of nth mill,

푚̇ is mass flow of fibre in the feed to the nth mill,

푚̇ is mass flow of fibre in expressed juice of nth mill.

The MILEX Model 43

This definition, while quite similar to that of Wienese (1995), is based on mass

flows that can be sensibly subtracted from each other.

3.3 Juice screen model

The juice screen separates the fibre in the expressed juice streams from the first

and second mills and typically returns it to the second mill while screened juice is

sent for processing. The major drawback of the Wienese (1994, 1995) extraction

model was that, even though Wienese accounted for the fibre in the expressed juice

stream, the model did not consider the fibre returning from the juice screen to the

milling train. A screen efficiency, similar in concept to the separation efficiency

described in section 3.2.5, has been introduced to model the flow of fibre from the

juice screen.

푆 =

푚̇푚̇ + 푚̇ × 100 3.8

Where: 푆 is screen efficiency,

푚̇ is mass flow of fibre in the return stream from the juice screen,

푚̇ is mass flow of fibre in expressed juice from the first mill,

푚̇ is mass flow of fibre in expressed juice from the second mill.

The fibre flow model (Kent, 2001) has discussed the impact of the fibre content

of the return stream on the overall fibre flows in the milling train. The fibre content

of the return stream is rarely measured and some standard values were selected for

the model.

The fibre content of the return stream is defined as:

푃 =

푚̇푚̇ 3.9

Where: 푃 is fibre content of return stream,

푚̇ is total mass flow in return stream (kg/s).

44 The MILEX Model

Equation 3.9 can be written as:

푃 =

푚̇푚̇ + 푚̇ 3.10

1푃 = 1 +

푚̇푚̇ 3.11

The ̇̇

is defined as the juice/fibre ratio of the return stream. Rearranging

equation 3.11, we get:

푅 =

1푃 − 1 3.12

Where: 푅 is the juice/fibre ratio of the return stream.

The juice/fibre ratio determines the amount of juice that is returning to the

milling train along with the fibre, increasing the load on the milling process. Hence,

the juice/fibre ratio of the return stream better describes the performance of the juice

screen than the fibre content of the return stream.

3.4 The milling train model

In the previous sections, the MILEX model for a milling unit and a juice screen

are described and the corrected mill parameters have been defined. In this section, a

mass balance model is described to calculate the actual flows of cane constituents

through the milling process including the return stream from the juice screen.

The mass flow equations for a five-mill milling train including the juice screen

are shown in section 2.5.2. For each of the 14 streams, the model accounts for three

mass flows: total flow, fibre flow and brix flow. Hence, there are 42 mass flows in

the model. The cane rate and added water rate are inputs to the model leaving 40

unknown flows. These parameters can be determined by solving equation 2.27 to

2.32 for total flow, fibre flow and brix flow, leaving 22 unknown flows. Fibre and

brix content are known for cane and imbibition and either known for the five bagasse

streams or can be calculated from the performance parameters. Using equation 2.33

for fibre fraction and brix fraction for these seven streams reduces the number of

unknowns to eight. Separation efficiency is known for the five mills and so equation

3.7 reduces the number of unknowns to three. The juice screen efficiency in equation

The MILEX Model 45

3.8 introduces one further equation leaving two unknowns relating to the juice

screen. The juice/fibre ratio of the return stream in equation 3.12 should be provided

as an input, providing another equation. Finally following Loubser (2004), the brix

fraction in the return stream and the brix fraction in mixed juice are assumed the

same as shown in equation 3.13.

푌 = 푌 3.13

Where: 푌 is brix fraction in mixed juice,

푌 is brix fraction in return stream from juice screen.

This assumption seems reasonable because the return stream has been saturated

by mixed juice in the juice screen. When calculating the brix fraction in juice, the

total mass flow is subtracted by the fibre flow, since brix fraction is brix on juice and

not brix on total juice material.

3.5 Solving the MILEX model

3.5.1 Analytical mode

The theory described below is a step by step procedure to calculate the mill

performance parameters in the MILEX model. The inlet and outlet quantities of

single milling unit ‘n’ are shown in Figure 3.2. The input quantities, 푚̇ ( ) ,

푚̇ ( ) and 푚̇ ( ) , are the mass rate of brix, moisture and fibre respectively in

bagasse from the preceding mill. The output quantities, 푚̇ , 푚̇ and 푚̇ , are

the mass rate of brix, moisture and fibre respectively in bagasse. 푚̇ , 푚̇ and

푚̇ are the mass rate of brix, moisture and fibre in imbibition respectively. The

extraction quantities, 푚̇ and 푚̇ , are the mass rate of brix and moisture in

expressed juice respectively. The mass rates are calculated from the mass

percentages as described in section 3.4.

46 The MILEX Model

Figure 3.2 The inlet and outlet quantities of single milling unit ‘n’ (MILEX)

The mass rate of brix in the expressed juice stream is determined from:

푚̇ = 푚̇ ( ) + 푚̇ − 푚̇ 3.14

Where: 푚̇ is mass rate of brix in expressed juice of nth mill (kg/s),

푚̇ ( ) is the mass rate of brix in bagasse of (n -1)th mill (kg/s),

푚̇ is mass rate of brix in imbibition of the nth mill (kg/s),

푚̇ is mass rate of brix in bagasse of nth mill (kg/s).

The mass rate of moisture in expressed juice stream is determined from:

푚̇ = 푚̇ ( ) + 푚̇ − 푚̇ 3.15

Where: 푚̇ is mass rate of brix in expressed juice of nth mill (kg/s),

푚̇ ( ) is the mass rate of brix in bagasse of (n -1)th mill (kg/s),

푚̇ is mass rate of brix in imbibition of nth mill (kg/s),

푚̇ is mass rate of brix in bagasse of nth mill(kg/s).

푚̇ 푚̇ 푚̇

푚̇ 푚̇ 푚̇

Output

푚̇ 푚̇

푚̇

Expressed Juice

푚̇ ( )

푚̇ ( )

Input 푚̇ ( )

Imbibition

The MILEX Model 47

The mass rate of juice in bagasse is determined from:

푚̇ = 푚̇ + 푚̇ 3.16

Where: 푚̇ is mass rate of juice in bagasse of nth mill (kg/s).

The brix fraction of juice in bagasse is determined from:

푌 =

푚̇푚̇ 3.17

Where: 푌 is the brix fraction of juice in bagasse.

The volume rate of juice in bagasse is determined from:

푉̇ =

푚̇푑 3.18

Where: 푉̇ is volume rate of juice in bagasse of nth mill (m3/s),

푑 is the density of juice in bagasse of nth mill (kg/m3).

The density of juice in bagasse is determined from the brix-fraction-density

factor (Russell, 1968).

The volume rate of fibre in bagasse is determined from:

푉̇ =

푚̇푑 3.19

Where: 푉̇ is volume rate of fibre in bagasse of nth mill (m3/s),

푚̇ is mass rate of fibre in bagasse of nth mill (kg/s),

푑 is the density of fibre (1530 kg/m3) (Pidduck, 1955).

The escribed volume of the milling unit is determined from:

푉̇ =

푚̇푑 × 퐶 3.20

Where: 푉̇ is escribed volume of nth mill (m3/s),

푚̇ is mass rate of fibre in cane (kg/s),

퐶 is filling ratio as defined by Murry (1959).

The corrected filling ratio (퐶 ) can be calculated from equation 3.2.

48 The MILEX Model

The volume rate of bagasse is determined from:

푉̇ = 푉̇ + 푉̇ 3.21

Where: 푉̇ is volume rate of bagasse of nth mill (m3/s).

The corrected reabsorption factor can be calculated from equation 3.4.

The theoretical brix fraction of juice in bagasse is determined from:

푌 =

푚̇ + 푚̇ ( )

푚̇ + 푚̇ + 푚̇ ( ) + 푚̇ ( ) 3.22

Where: 푌 is the brix fraction of juice in bagasse.

The theoretical mass rate of juice in bagasse is determined from:

푚̇ = 푉̇ × 푑 3.23

Where: 푚̇ is theoretical mass rate of juice in bagasse of nth mill (kg/s),

푑 is the theoretical density of juice in bagasse of nth mill (kg/m3).

The theoretical density of juice in bagasse is determined from the brix-fraction-

density factor (Russell, 1968).

The theoretical mass rate of brix in bagasse is determined from:

푚̇ = 푚̇ × 푌 3.24

Where: 푚̇ is theoretical mass rate of brix in bagasse of nth mill (kg/s),

The corrected imbibition coefficient and the separation efficiency can be

calculated from equation 3.6. and equation 3.8 respectively.

3.5.2 Predictive mode

The predictive mode of the MILEX model is more complex than for the

MILSIM model and requires the calculation of filling ratio to obtain the solution. As

discussed in section 3.4 the MILEX model is run in predictive mode with the

corrected mill performance parameters (filling ratio, reabsorption factor, imbibition

coefficient and separation efficiency) as the inputs to the model. Equations 3.14 to

3.24 can be used to run the model in predictive mode.

The MILEX Model 49

The fibre rate in the expressed juice stream is determined from:

푚̇ = 푚̇ × 1 −푆

100 3.25

Where: 푚̇ is mass rate of fibre in expressed juice of nth mill (kg/s),

푚̇ is mass rate of fibre in feed to the nth mill (kg/s),

푆 is separation efficiency of nth mill (%).

The fibre in delivery bagasse is determined from:

푚̇ = 푚̇ − 푚̇ 3.26

Where: 푚̇ is mass rate of fibre in delivery bagasse of nth mill (kg/s).

The escribed volume of the milling unit is determined from the physical

parameters of mill length, delivery nip work opening and top roller surface speed.

The corrected filling ratio can be calculated from equation 3.2.

The volume rate of fibre in delivery bagasse is determined from:

푉̇ =

푚̇푑 3.27

Where: 푉̇ is volume rate of fibre in delivery bagasse of nth mill (m3/s),

푑 is density of fibre (1530 kg/m3) (Pidduck, 1955).

The volume rate of juice in delivery bagasse is determined from:

푉̇ =퐾퐶 − 1 × 푉̇ 3.28

Where: 퐶 is corrected filling ratio of nth mill,

퐾 is corrected reabsorption factor of nth mill,

푉̇ is volume rate of fibre in bagasse of nth mill (m3/s).

The theoretical brix fraction of juice in bagasse, the theoretical mass rate of

juice in bagasse and the theoretical mass rate of brix in bagasse are determined from

equations 3.22, 3.23 and 3.24 respectively.

50 The MILEX Model

The mass rate of brix in bagasse is determined from:

푚̇ = (1 − 퐼 ) × 푚̇ ( ) + 퐼 × 푚̇ 3.29

Where: 퐼 is corrected imbibition coefficient of nth mill.

The brix fraction-density factor is determined from:

푌 푑 =

푚̇푉̇

3.30

The actual mass rate of juice in bagasse is determined from:

푚̇ =

푚̇푌 3.31

Where: 푌 is brix fraction of juice in bagasse of nth mill.

The (푌 ) brix fraction of juice in bagasse is determined from an empirical equation

using the brix fraction-density factor as presented by Russell (1968).

The mass rate of moisture in bagasse is determined from:

푚̇ = 푚̇ − 푚̇ 3.32

The mass rates of brix and moisture in juice streams are determined from equations

3.14 and 3.15.

3.6 Exploring the model

3.6.1 A base case for testing the model

A five-mill milling train with simple compound imbibition is used for testing

the model. To test the model a standard set of input values were adopted (Table 3.1).

The cane and bagasse analysis values were based on a set of routine factory bagasse

analysis data. The fibre content of expressed juice streams, mixed juice stream and

the return stream from the juice screen values were adopted from Kent (2001). The

values measured were from Victoria mill and Method 13 of the BSES laboratory

manual (Bureau of Sugar Experiment Stations, 2001) was used. Hence the values

were actually the true fibre measurements in juice streams. Although, for this case

study the values are assumed to represent the total insoluble solids in juice streams.

These results imply a juice screen efficiency of 92.3%.

The MILEX Model 51

Table 3.1 Input values for the model

Product stream Brix (%) Fibre (%) Cane (198.47 kg/s) 15.91 14.00 #1 mill bagasse 11.93 30.16 #2 mill bagasse 8.43 35.70 #3 mill bagasse 6.24 40.47 #4 mill bagasse 4.65 44.45 #5 mill bagasse 3.19 47.37 Expressed juice - 2.00 Return stream - 8.00 Mixed juice - 0.20 Imbibition (200% fibre) 0.00 0.00

3.6.2 Effect of the constant fibre rate assumption on mill parameters

The corrected mill parameters calculated from the model are compared to their

MILSIM values and are shown in Table 3.2,Table 3.3 and Table 3.4.

Table 3.2 shows the corrected filling ratio results. The ratio essentially

shows how the delivery bagasse fibre rate compares against the cane fibre rate. The

low bagasse fibre flow in the delivery bagasse of #1 mill is an interesting result from

the model. The value shows that only 92% of fibre in cane ended up in the delivery

bagasse of #1 mill, with the remaining 8% in the expressed juice stream. In other

words, 2% fibre in juice equates to 8% fibre in cane. Kent (2001) did not identify this

issue because he examined the fibre rate entering each milling unit rather than the

fibre rate leaving each milling unit. The filling ratios of #1 and #5 mill were less than

their MILSIM values while the filling ratios for the intermediate mills were greater.

Table 3.2 Filling ratio

Mill Filling ratio 푪푶푪 MILSIM (C) Corrected (CO)

#1 0.392 0.360 0.92 #2 0.444 0.463 1.04 #3 0.492 0.507 1.04 #4 0.542 0.559 1.03 #5 0.575 0.567 0.99

The corrected reabsorption factor values shown in Table 3.3 differ from the

MILSIM values by virtually the same amount as the filling ratio and reflect the

52 The MILEX Model

importance of the ratio in the model, rather than either reabsorption factor or

filling ratio alone.

Table 3.3 Reabsorption factor

Mill Reabsorption factor 푲푶

푲 MILSIM (K) Corrected (KO) #1 1.69 1.55 0.91 #2 1.60 1.67 1.04 #3 1.55 1.60 1.03 #4 1.54 1.59 1.03 #5 1.53 1.51 0.98

The corrected imbibition coefficient value shown in Table 3.4 is slightly lower

than the MILSIM value for #1 mill but higher at the other mills. The largest

differences were found for #2 and #5 mills. The return stream from the juice screen

is added to #2 mill, increasing the input brix mass flow to #2 mill. The brix (and

fibre) flow into the final mill is greater than for the MILSIM model because, for the

mass to balance, it contains the brix and fibre expressed in the juice from the final

mill. The brix (and fibre) flow in the bagasse from the final mill has less brix and

fibre because this flow is missing the brix and fibre in mixed juice. The combination

of increased brix flow into the final mill and reduced brix flow out of the final mill in

equation 3.5 causes the large difference in the imbibition coefficient.

Table 3.4 Imbibition coefficient

Mill Imbibition coefficient 푰푪푶푰푪

MILSIM (IC) Corrected (ICO) #1 1.05 1.04 0.99 #2 0.82 0.87 1.06 #3 0.66 0.67 1.01 #4 0.55 0.57 1.03 #5 0.53 0.57 1.07

3.6.3 Separation efficiency

Table 3.5 shows the calculated separation efficiency for each mill, resulting

from 2% fibre in expressed juice for each mill. The separation efficiencies range

from 91.1% at #2 mill to 95.7% at #5 mill. The separation efficiencies are

substantially lower for #1 and #2 mills than the other mills. In the case of #1 mill,

The MILEX Model 53

there is a larger juice flow as a consequence of the higher juice content of the cane.

Since the fibre content of the juice has been defined the same as for the other mills,

the fibre rate in the juice is also higher, giving a lower separation efficiency. In the

case of #2 mill, adding the cush return increases the fibre going into the mill thus

decreasing the separation efficiency.

Table 3.5 Separation efficiency input values for 2% fibre content of expressed

juice

Mill Separation efficiency (%) #1 91.80 #2 91.11 #3 94.80 #4 95.41 #5 95.65

It follows that, if the separation efficiency was assumed constant for each mill,

the resulting fibre content in expressed juice differs substantially from mill to mill

with relatively much less fibre in the juice from #1 and #2 mills (Table 3.6).

Table 3.6 Fibre% juice values for separation efficiency of 95%

Mill Fibre% Juice #1 1.22 #2 1.33 #3 1.93 #4 2.20 #5 2.32

3.6.4 Effect of including juice screen in the model

The MILSIM model was based on the assumption that the fibre content of the

juice streams is negligible. Hence, it seems logical not to include the juice screen in

the MILSIM model although, in one of the earlier version of MILSIM, the juice

screen model was included to model the juice returning from the juice screen to the

milling train. The model was not generally adopted and hence is not described in the

thesis. The Wienese (1994) extraction model did not consider the juice screen, even

though he accounted for the fibre% juice to be a substantial amount.

A case study was undertaken to see if neglecting the juice screen significantly

affects the calculation of extraction. The model was run in analytical mode to

calculate reabsorption factor, imbibition coefficient and separation efficiency for two

54 The MILEX Model

cases. The first case was the base case described earlier in this section. For the

second case, the juice screen efficiency was set to zero so that there was no return

stream from the juice screen. The results are shown in Table 3.7.

Table 3.7 Effect of juice screen on calculated performance parameters

Juice screen KO ICO Case 1 (With juice screen)

1.55 1.04 1.67 0.87 1.60 0.67 1.59 0.57 1.51 0.57

Case2 (Without juice screen)

1.55 1.04 1.46 0.79 1.39 0.63 1.38 0.54 1.30 0.55

Although Table 3.7 shows some significant changes in parameter values, it is

difficult to conceptualize their impact and so brix extraction was calculated for the

base case model (including the juice screen) using both sets of performance

parameters. Using the parameter values calculated with the juice screen in the model,

the brix extraction was 94.2%. Using the parameter values calculated without the

juice screen in the model, the brix extraction was 95.1%. This analysis shows that the

inclusion of the juice screen in the model changes the model parameters to the extent

of almost one unit of extraction. Consequently, the juice screen is an important part

of the overall model.



For a mathematical model to be useful for practical purposes, it must be


be studied. To test the model, results from the model need to be compared to

operating and performance information obtained on the real system.

3.7.2 Factory test at Isis sugar mill

A series of tests were conducted at the Isis sugar mill in Australia as part of a

different project. The Isis sugar mill was chosen for these tests because it had flow

The MILEX Model 55

meters for #3, #4 and #5 mill expressed juice and for mixed juice. The test details

and the laboratory analysis from Kent et al.(2008) are presented below.

Test details

A series of four tests were conducted. Each test lasted approximately four

hours. During each test, samples of the five mill bagasses, cush return (stream from

the juice screen), the five mill expressed juices and mixed juice were regularly taken.

Approximately eight sub samples were collected during each test period and

composited for analysis.

Weighbridge records were used to determine the cane rate during each test and

the corresponding added water rates were recorded. The prepared cane was analysed

with the on-line analysis system and the brix and fibre of the samples were recorded.

Table 3.8 shows the cane supply and imbibition details.

Table 3.8 Cane supply and imbibition details

Test Tonnes crushed

Cane brix (%)

Cane fibre (%)

Added water rate (%fibre)

1 1346.21 16.65 14.06 285.11 2 1322.30 15.95 13.95 293.33 3 1279.04 17.16 14.43 300.29 4 1202.46 16.41 14.61 297.91

Laboratory analysis

The analysis methods used in the Isis mill laboratory for each sample are listed

in Table 3.9. The method numbers refer to the Laboratory Manual for Australian

Sugar Mills (Bureau of Sugar Experiment Stations, 2001). A digital Refractometer

accurate to a brix of 0.03) was used to measure the brix content of the juice samples

from the bagasse and cush disintegrator extracts and from the juice samples.

56 The MILEX Model

Table 3.9 Analysis method

Sample Analysis Method

Bagasse and cush return Moisture 7 Brix 5

Juice Brix Digital Refractometer True fibre 13

Bagasse analysis results

The moisture and brix content of bagasse from each milling unit and the cush

return were analysed and are shown in Table 3.10 and Table 3.11 respectively.

Table 3.10 Bagasse moisture content results (%)

Test #1 mill #2 mill #3 mill #4 mill #5 mill Cush 1 52.19 57.15 54.91 56.13 50.95 77.23 2 54.84 59.16 56.24 55.72 52.06 81.76 3 52.94 58.81 57.33 57.43 51.12 77.30 4 53.79 59.58 54.67 55.91 50.51 77.30

Table 3.11 Bagasse brix content results (%)

Test #1 mill #2 mill #3 mill #4 mill #5 mill Cush 1 12.67 10.49 8.22 6.21 4.60 14.49 2 12.93 9.76 7.49 5.89 4.60 13.68 3 14.62 11.16 7.71 6.65 5.33 16.38 4 14.10 10.30 7.79 6.10 5.43 14.49

Juice analysis results

The brix and fibre content of juice from each milling unit and the mixed juice

were analysed and are shown in Table 3.12 and Table 3.13 respectively.

Table 3.12 Juice brix content (%)

Test #1 mill #2 mill #3 mill #4 mill #5 mill Mixed 1 19.96 9.14 4.64 2.66 1.44 14.53 2 18.96 8.54 4.30 2.39 1.08 13.88 3 20.00 9.02 4.43 2.50 1.28 14.52 4 19.21 8.50 4.25 2.43 1.25 13.84

The MILEX Model 57

Table 3.13 Juice fibre content (%)

Test #1 mill #2 mill #3 mill #4 mill #5 mill Mixed 1 0.81 1.08 1.29 0.91 1.82 0.47 2 0.54 1.13 1.34 1.22 1.55 0.47 3 0.79 1.31 3.53 2.52 1.76 0.23 4 0.68 1.23 1.55 2.57 1.77 0.23

Average bagasse & juice flow measurements

Table 3.14 presents the measured cane rates, added water rates and mixed juice

rates for the four tests. To calculate the final bagasse rates, a mass balance across the

milling train was conducted.

The mass flow of cane constituents across the entire milling train can be

determined from:

푚̇ + 푚̇ = 푚̇ + 푚̇ 3.33

where: 푚̇ is cane rate (t/h),

푚̇ is added water rate (t/h),

푚̇ is final bagasse rate (t/h),

푚̇ is mixed juice rate (t/h).

The final bagasse rates were determined from equation 3.33 and the measured

cane rates, added water rates and mixed juice rates.

Table 3.14 Average flow measurements

Test Cane rate (t/h)

Added water rate (t/h)

Mixed juice rate (t/h)

Final bagasse rate (t/h)

1 429.6 172 469 132.60 2 417.6 171 466 122.60 3 426.3 185 481 130.30 4 426.9 186 486 126.90

3.7.3 Running the model

The model was explored with the series of test results shown in section 3.7.2.

For each of the four tests, the model was run in analytical mode to determine the mill

performance parameters for individual milling units. The results from the model are

presented in the next section.

58 The MILEX Model

Filling ratio & Reabsorption factor

The project conducted at the Isis sugar mill did not require the mill

configuration data. Hence the filling ratio values were not available to determine the

corrected filling ratio and corrected reabsorption factor. Hence the ratio, rather

than the individual parameters, for individual milling units for the four tests

conducted was determined. The significance of the ratio is discussed in section 3.6.2.

In Figure 3.3 the ratio for each milling unit and the four tests is plotted. For the #1,

#2, #3 and #4 mills ratio vary within a range of about 0.4 units.

Mill

#1 Mill #2 Mill #3 Mill #4 Mill #5 Mill

K/C

ratio

2.8

3.0

3.2

3.4

3.6

3.8

4.0

4.2

4.4

4.6Test 1Test 2Test 3Test 4

Figure 3.3 (K/C) ratio results

Imbibition coefficient

The imbibition coefficient results against the mill position are shown in Figure

3.4. The imbibition coefficients were about 1.0 for #1 mill, generally decreasing to

about 0.5 for #5 mill although lower values were obtained for #2 and #4 mills. It can

be seen that imbibition coefficient for the first mill was quite consistent for all the

tests. The #2 mill shows the largest difference for all the tests followed by #3, #4 and

#5 mills.

The MILEX Model 59

Kent et al (1998) made similar observations with imbibition coefficient

parameter based on the tests conducted at the Victoria mill. The first mill imbibition

coefficient was quite consistent for all the tests. The largest difference was found to

be at the #2 mill followed by the #4 mill and final mill.

The variations of the imbibition coefficient multipliers are similar to the

imbibition coefficient results.

Mill


Imbi

bibt

ion

coef

ficien

t

0.0

0.2

0.4

0.6

0.8

1.0


Figure 3.4 Imbibition coefficient results

60 The MILEX Model

Mill


Imbi

bibt

ion

coef

ficien

t mul

tiplie

r

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4


Figure 3.5 Imbibition coefficient multiplier results

Separation efficiency

The separation efficiency results are shown in Figure 3.6. It can be concluded

from the graph that separation efficiency generally decreases from #1 mill to #5 mill.

The separation efficiencies of the #3 and #4 mill show larger variation, although the

reason is not understood. The reason for the decrease in separation efficiency along

the milling train is likely to be a consequence of the reduced size of the bagasse

particles, allowing more fine particles into the juice stream.

The MILEX Model 61

Mill


Sepa

ratio

n ef

ficien

cy

86

88

90

92

94

96

98

100Test 1Test 2Test 3Test 4

Figure 3.6 Separation efficiency results

3.7.4 Discussion of results

In the tests undertaken at Isis sugar mill, no deliberate changes were made in

mill configurations or mill control during the tests. Consequently, the variation in

parameters can be attributed to natural variability in the cane supply, process control

and measurements (including bagasse and juice sampling and analysis).

Although filling ratio and reabsorption factor were not individually determined,

the ratio results can be examined to provide some information. As discussed in

section 2.2.5, the empirical equation calculates reabsorption factor as a function of

delivery nip compaction and top roller surface speed. The control system of the

milling unit tries to keep both delivery nip compaction and top roller surface speed

constant and so most of the variability in the ratio is likely to be due to

variability in reabsorption factor. The cause of that variability is still not well

understood.

There is a lot more variability in the imbibition coefficient parameter and

imbibition coefficient multiplier results for the intermediate milling units than for the

62 The MILEX Model

first mill. The change in imbibition level is at least partly responsible for the

variation in the imbibition coefficient parameter, but the variation in imbibition

coefficient multipliers is not understood. It is observed in Figure 3.5 that the

imbibition coefficient multipliers show almost as much variability as the imbibition

coefficients themselves, concluding that the imbibition coefficient empirical equation

is far from perfect. The imbibition coefficient parameter and multipliers of the #2

mill show the maximum deviation. The cush return from the juice screen added to

the #2 mill is expected to be the cause of this.

The separation efficiency results as shown in Figure 3.6 are mostly quite

consistent although there is one outlier in the #3 mill results and the #4 mill results

are quite variable. The Isis results are not sufficient to make a conclusion regarding

separation efficiency parameter. It does appear that the separation efficiency

decreases along the milling train, presumably as a function of the fineness of the

bagasse.


In exploring the MILEX model using Isis mill results where conditions were

more or less constant, it was found that many mill performance parameters were still

quite variable.

It can be anticipated that the differences in ratio is caused by changes in

reabsorption factor. The ratio varies within a range of 0.4 units for the #1, #2, #3

and #4 mills with typical values of 4.0. The imbibition coefficient parameter varies in

much the same way as the imbibition coefficient multiplier. The #1 mill imbibition

coefficient parameters are relatively consistent with typical values of 0.99. The

results show a lot more variability for the other mills with typical values of 0.2 to 0.8.

The separation efficiency results show that the separation efficiency decreases down

the milling train with typical values of 92% to 96%.

3.8 Case studies

The MILEX model was used to study the effect of parameters on brix

extraction of individual milling units and the overall milling train. Case studies have

been undertaken to study the effect of fibre in expressed juice streams, juice in the

return stream from the juice screen, fibre in the mixed juice stream, imbibition%

The MILEX Model 63

fibre and crushing rate on extraction performance. The case studies and the results

are described in the sections below.

3.8.1 Effect of separation efficiency and juice in return stream from juice screen on extraction performance

It is conceivable that, if all the clearances around the mill, such as between

scrapers and rollers, were reduced, the amount of fibre in expressed juice could

reduce or, in terms of the extraction model, the separation efficiency could increase.

Similarly, it is conceivable that, if the juice screen could be continuously cleaned or

had a larger screen area, a greater amount of juice would pass through to mixed juice

and not be returned to the milling train. The extraction model can examine these

concepts.

The model was solved for separation efficiencies in each mill from 93% to

96% and juice/fibre ratio in the return stream from the juice screen of 0 to 12 (a fibre

in the return stream of 8% as listed in Table 3.1 corresponds to a juice/fibre ratio of

11.5). The brix extraction results are shown in Figure 3.7.

Juice/Fibre ratio

0 2 4 6 8 10 12 14

Brix

extra

ctio

n (%

)

94.1

94.2

94.3

94.4

94.5

94.6

94.7

94.8

94.9

Sn 93%Sn 94%Sn 95%Sn 96%

Figure 3.7 Effect of separation efficiency and juice in cush return on brix extraction

64 The MILEX Model

As separation efficiency increases, brix extraction increases, and as juice/fibre

ratio of the return stream from the juice screen increases, brix extraction decreases.

Increasing the separation efficiency from 93% to 96% increased extraction by 0.2

units. Reducing the juice/fibre ratio of the return stream from the juice screen from

12 to 10, increased extraction by about 0.1 units.

As discussed in section 2.2.5, an empirical equation for imbibition coefficient

has been developed that shows that the imbibition coefficient is a function of

imbibition level and brix concentration. When changes to the juice and fibre flows

are made, as can be expected by changing the filling ratio, reabsorption factor or

separation efficiency, it is logical to assume that the imbibition coefficients will

change and hence the use of imbibition coefficient multipliers is essential when

comparing extraction figures for different values of these parameters.

Figure 3.8 shows brix extraction results in the same way as Figure 3.7, only in

this case imbibition coefficient multipliers, rather than imbibition coefficient

parameters, have been kept constant. The trend of the graph is similar to the Figure

3.7 although the increase in extraction is found to be 0.32 units when increasing

separation efficiency from 93% to 96%, as opposed to 0.2 units as shown in Figure

3.7.

The MILEX Model 65

Juice/Fibre ratio

0 2 4 6 8 10 12 14

Brix

extra

ctio

n (%

)

94.0

94.1

94.2

94.3

94.4

94.5

94.6

94.7

94.8Sn 93%Sn 94%Sn 95%Sn 96%

Figure 3.8 Effect of separation efficiency on brix extraction

The use of multipliers in the model causes a difference in extraction prediction

of 0.15 units, as opposed to when not using multipliers. The results shown in Figure

3.8 are considered more correct as the empirical equation takes into account the

changes in imbibition level to the milling unit as a result of changes in bagasse and

juice mass flows. Hence, model calibration is necessary to ensure reliable predictions

are made.

3.8.2 Effect of fibre in mixed juice on extraction performance

The fibre in mixed juice was varied and the calculated effect on juice screen

efficiency and brix extraction is shown in Figure 3.9.

66 The MILEX Model

Fibre% mixed juice

0.2 0.4 0.6 0.8

Brix

extra

ctio

n (%

)

94.10

94.15

94.20

94.25

94.30

94.35

94.40

94.45

Fibre% mixed juice

0.2 0.4 0.6 0.8

Scre

en e

ffic

ienc

y (%

)

65

70

75

80

85

90

95

Figure 3.9 Effect of fibre in mixed juice on screen efficiency and extraction

As shown in Figure 3.9, increasing the fibre content of the mixed juice is a

consequence of decreasing the screen efficiency. The juice screen efficiency is

defined in terms of fibre returned to the milling train from the juice screen.

Increasing the fibre content of the mixed juice reduces the fibre flow and total flow

in the return stream from the juice screen. The fibre content of mixed juice has little

effect on the brix extraction although it does not indicate that the brix associated with

it, has been completely extracted. The fibre carried with the mixed juice is retrieved

at the filter station and it is the filter’s job of extraction the brix. The filter station

modelling is not a part of the project and is not described in this thesis.

3.8.3 Effect of added water rate on extraction performance

It is well known that increasing the added water rate (Imbibition% Fibre)

increases extraction (Rein, 2007). In this scenario, the load on the evaporator station

increases as there is extra water to be removed, hence increasing the fuel

consumption in the boilers (Lloyd et al., 2010).

A case study was undertaken to observe the effect of added water rate on

extraction performance of individual milling units and overall milling train. The

MILSIM and MILEX model were solved in analytical mode with a set of milling

train data. Table 3.15 and Table 3.16 show the bagasse analysis and the mill setting

values respectively used to calculate the performance parameters and the mill

performance multipliers as shown in Table 3.17, Table 3.18, Table 3.19 and Table

3.20. The separation efficiency parameter is determined by assuming the fibre

The MILEX Model 67

content of juice streams is 2%, fibre content of cush return is 8% and fibre content of

mixed juice stream is 0.2%. Both the models were solved for 207 imbibition% fibre.

The mill performance multipliers are kept constant in the predictive mode of the

model and the models are tested with different imbibition% fibre and extraction

performance is recorded.

Table 3.15 Bagasse analysis

Product stream Brix (%) Fibre (%) Cane (200 kg/s) 16.77 18.15 #1 mill bagasse 12.30 34.29 #2 mill bagasse 8.01 39.99 #3 mill bagasse 5.69 38.57 #4 mill bagasse 3.84 45.18 #5 mill bagasse 3.03 47.22 Expressed juice - 2.00 Return stream - 8.00 Mixed juice - 0.20 Imbibition (207% fibre) 0.00 0.00

Table 3.16 Compaction and roller surface speed of the mills

Mill Compaction 풌품풎ퟑ Roller surface speed 풎풎

풔

#1 mill 452 228 #2 mill 596 208 #3 mill 618 191 #4 mill 698 202 #5 mill 832 211

Table 3.17 Mill performance parameters (MILSIM)

Performance parameters #1 mill #2 mill #3 mill #4 mill #5 mill Filling ratio 0.296 0.389 0.404 0.456 0.544

Reabsorption factor 1.101 1.239 1.356 1.283 1.455 Imbibition coefficient 1.069 0.833 0.589 0.667 0.371

Table 3.18 Mill performance multipliers (MILSIM)

Performance multipliers #1 mill #2 mill #3 mill #4 mill #5 mill Reabsorption factor 0.842 0.863 1.029 0.952 0.990

Imbibition coefficient 1.02 1.00 0.909 1.277 0.748

68 The MILEX Model

Table 3.19 Corrected mill performance parameters (MILEX)

Performance parameters #1 mill #2 mill #3 mill #4 mill #5 mill Filling ratio 0.279 0.404 0.420 0.471 0.537

Reabsorption factor 1.05 1.299 1.42 1.326 1.446 Imbibition coefficient 1.06 0.918 0.618 0.526 0.494 Separation efficiency 94.49 92.43 95.41 95.20 95.72

Table 3.20 Mill performance multipliers (MILEX)

Performance multipliers #1 mill #2 mill #3 mill #4 mill #5 mill Reabsorption factor 0.847 0.874 1.043 0.962 0.993

Imbibition coefficient 1.01 0.842 0.9346 1.312 0.848

The results for the simulation are shown in Figure 3.10. To compare the results

from both the models, conventional definition was used to determine extraction of

the milling units in both the models.

Because imbibition has no effect on the performance of #1 mill, the extraction

performance of the #1 mill remains unchanged in both the models. The surprising

trends in Figure 3.10 are the extraction curves for the #3, #4 and #5 mills in the

MILEX model and #4 and #5 mills in the MILSIM model which show a decrease in

extraction performance with an increase in added water rate. The biggest difference

between the two models is for #2 mill where for the MILEX model, the increase in

extraction performance with increasing added water rate is around 7 units compared

to about 2 units for the MILSIM model. The overall increase in extraction

performance of the MILEX model is around 0.4 units and for MILSIM model is

around 0.15 units.

To determine if the predicted trends are realistic, bagasse analysis data from a

factory experiment where added water levels were changed was obtained from Kent

et al (2000) and brix extraction was determined for #2, #3, #4 and #5 mills using the

conventional definition as presented in equation 1.4. The brix extraction is plotted

against the added water rate and regression lines are shown to understand the trend of

the data. The results are shown in Figure 3.11. Results of the regression analysis

showing the slope of the regression line is presented in Table 3.21.

The MILEX Model 69

Table 3.21 Regression analysis for brix extraction trends

Mill no Slope Std.

Error

t (statistic) Pr(>t) Lower 95%

confidence

Upper 95%

confidence

#2 0.0581 0.0264 2.200 0.045 0.112 8.738

#3 0.0216 0.0281 0.767 0.455 -3.36 7.106

#4 -0.0146 0.0318 -0.4597 0.653 -5.757 3.725

#5 -0.0248 0.0546 -0.4539 0.657 -3.35 2.18

Although the standard error of the slopes is quite high and the confidence

interval is quite wide (Table 3.21) particularly for #3, #4 and #5 mills, the best

estimate of the slope for brix extraction for each mill (Figure 3.11) matches the brix

extraction trends from Figure 3.10 for the MILSIM model. .

To understand the difference in behaviour of the MILEX model to the MILSIM

model, in particular for the #2 and #3 mill, the imbibition coefficient empirical

equation developed by Russell (1968) was investigated as shown in Figure 3.12 and

Figure 3.13.

70 The MILEX Model

#1 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

60.0

60.5

61.0

61.5

62.0

#2 Mill

Imbibition% fibre

200 210 220 230 240

Brix

extra

ctio

n (%

)

20

25

30

35

40

45

50

55

MILEXMILSIM#3 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

20

25

30

35

40

45

50

55

#4 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

20

25

30

35

40

45

50

55

#5 Mill

Imbibition% fibre

210 220 230 240 250

Brix

extra

ctio

n (%

)

20

25

30

35

40

45

50

55

Overall

Imbibition% fibre

210 220 230 240 250

Brix

extra

ctio

n (%

)

93.0

93.1

93.2

93.3

93.4

93.5

Figure 3.10 Effect of imbibition% fibre on extraction performance

The MILEX Model 71

#2 Mill

Imbibition% fibre

160 180 200 220 240 260 280 300 320

Brix

extra

ctio

n (%

)

26

28

30

32

34

36

38

40

42

44

46

48

#3 Mill

Imbibition% fibre

160 180 200 220 240 260 280 300 320

Brix

extra

ctio

n (%

)

30

32

34

36

38

40

42

44

46

48

Brix extractionRegr (extraction trend)#4 Mill

Imbibition% fibre

160 180 200 220 240 260 280 300 320

Brix

extra

ctio

n (%

)

16

18

20

22

24

26

28

30

32

34

36

#5 Mill

Imbibition% fibre

160 180 200 220 240 260 280 300 320

Brix

extra

ctio

n (%

)

5

10

15

20

25

30

35

40

45

Figure 3.11 Brix extraction trends for #2, #3, #4 and #5 mills

The empirical equation developed by Russell (1968) shows that imbibition

coefficient is a function of brix fraction of bagasse entering the mill, brix

fraction of imbibition to the mill and the juice in imbibition to the mill. All the three

parameters and the empirical equation trend are plotted for increasing added water

rate for both the models in Figure 3.12 and Figure 3.13 respectively.

Observing the graphs it can be seen that the only substantial difference is that

the brix fraction of bagasse entering the #2 mill in the MILEX model is twice the

value of brix fraction of bagasse entering the #2 mill in the MILSIM model. In the

MILEX model, the cush return from the juice screen is added before the #2 mill. This

72 The MILEX Model

causes the brix fraction of bagasse entering the mill go higher. The empirical

equation trend is shown to increase in the MILEX model and decrease in the

MILSIM model. Hence in the MILEX model the #2 mill extraction is overestimated

causing the model to predict reduction in extraction of the #3, #4 and #5 milling

units.

It can be concluded from the case study that the empirical equation agrees

satisfactorily for the MILSIM model but not with the MILEX model. There is a need

to develop a new empirical equation in the MILEX model to account for the cush

return before it can be applied with confidence.

The MILEX Model 73

Bagasse brix fraction (MILSIM)

Imbibtion% fibre

205 210 215 220 225 230 235 240

Rang

e

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

#2 Mill#3 Mill#4 Mill#5 Mill

Bagasse brix fraction (MILEX)

Imbibtion% fibre

205 210 215 220 225 230 235 2400.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

Imbibition juice (MILSIM)

205 210 215 220 225 230 235 240

Rang

e

2.0

2.2

2.4

2.6

2.8

3.0

Imbibition juice (MILEX)

205 210 215 220 225 230 235 2402.0

2.2

2.4

2.6

2.8

3.0

Imbibition brix (MILSIM)

205 210 215 220 225 230 235 240

Rang

e

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

Imbibition brix (MILEX)

205 210 215 220 225 230 235 240

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

Figure 3.12 Imbibition coefficient empirical equation functions trend

74 The MILEX Model

Empirical equation (MILEX)

Imbibition% fibre

205 210 215 220 225 230 235 240Ra

nge

0.4

0.6

0.8

1.0

1.2

1.4

#2 Mill#3 Mill#4 Mill#5 Mill

Empirical equation (MILSIM)

Imbibition% fibre

205 210 215 220 225 230 235 2400.4

0.6

0.8

1.0

1.2

1.4

Figure 3.13 Imbibition coefficient empirical equation trend

The MILEX Model 75

#1 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

20

30

40

50

60

70

#2 Mill

Imbibition% fibre

200 210 220 230 240

Brix

extra

ctio

n (%

)

20

30

40

50

60

70

Standard DefinitionRevised definition#3 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

20

30

40

50

60

70

#4 Mill

Imbibition% fibre

210 215 220 225 230 235 240

Brix

extra

ctio

n (%

)

20

30

40

50

60

70

#5 Mill

Imbibition% fibre

210 220 230 240 250

Brix

extra

ctio

n (%

)

20

30

40

50

60

70

Overall

Imbibition% fibre

210 220 230 240 250

Brix

extra

ctio

n (%

)

93.0

93.1

93.2

93.3

93.4

93.5

93.6

Figure 3.14 Extraction figures for MILEX (Standard and Revised definitions)

Figure 3.14 shows the extraction results of Figure 3.10 with the conventional or

standard definition of extraction as presented in equation 1.4 (the same results as in

Figure 3.10) and the revised extraction definition based on mass flows as presented

76 The MILEX Model

in equation 3.5. The trend of the extraction curves is similar for all milling units. The

#2 mill shows the largest difference in extraction from both the definitions as a result

of the juice screen. The #5 and #1 mills show slight difference in extraction. The

overall extraction shows a difference of 0.5 as result of fibre in mixed juice.

3.9 Concluding remarks

A more detailed mill extraction model was developed to predict the extraction

performance of a milling train. The model explains the effect of changing mass flows

of cane constituents through the milling train on mill parameters and eliminates the

assumption of constant fibre rate that is used in the existing MILSIM model. A

modified separation efficiency parameter has been included to quantify the amount

of fibre in expressed juice from a mill. The juice screen is modelled to account for

the juice returning to the milling train along with the return stream.

The solution method of the model is described and the model is explored with

factory data to understand the consistency of the mill performance parameters. It was

observed that without deliberate changes in operating conditions of the milling train,

mill performance parameters still varied, although the exact reason is still not known.

The model was explored by varying the added water rate and examining the

effect of added water rate on extraction. This exercise assisted in understanding the

error of the constant fibre assumption of the previous model and the conventional

definition of brix extraction. More importantly it was found that the empirical

equation developed in MILSIM predicted extraction performance as per the expected

trend. The extraction predictions by the MILEX model did not agree with the

expected trend because of the influence of the cush return and there is a need for a

new empirical equation for imbibition coefficient in the MILEX model.

Case studies were undertaken to study the effect of fibre in juice streams and

juice in cush return on extraction performance of the milling train. It was found that

decreasing the fibre in juice streams and juice content of cush return extraction

performance increased by 0.35 units.

Modelling the Cane Components 77

CHAPTER 4: MODELLING THE CANE COMPONENTS


In the previous chapter, an enhanced mill extraction model (MILEX) is

presented to predict the extraction performance of the milling process. The enhanced

model tracks the flow of brix, moisture and fibre through the milling train and the

distribution of these components in juice and bagasse.

The soluble impurities and mud solids in cane affect the performance of the

milling train and further processing of bagasse and juice. In the MILEX extraction

model, all the soluble solids are termed as brix and all the insoluble solids are termed

as fibre. To date, no comprehensive model has been developed to track the soluble

impurities and mud solids in the milling process. Specific impurities in cane, mixed

juice and final bagasse have been reported by authors worldwide, but none have

reported a holistic approach to track the flow of all major impurities.

In this chapter, a model is described to determine the flow of cane constituents

through the milling process. The focus of this model is to break down the soluble

solids (brix) into sucrose and soluble impurities and insoluble solids (fibre) into true

fibre and mud solids impurities and determine the flow of these components through

the milling process. The model is developed as an extension to the MILEX model to

provide the correct constituents in mixed juice and final bagasse as inputs to the

downstream station models.

4.2 Model framework

As an extension to the MILEX model presented in the previous chapter, the

components model follows the same modelling framework. A milling train model

was constructed from a series of milling unit models and a juice screen model as

shown in Figure 1.1 on page 5. The links between the milling units and juice screen

follow mass balance fundamentals. Equations 2.27 to 2.32 on page 28 show the

conservation of mass across each milling unit and across the juice screen. Similar

equations are applicable to the mass flow of components.

78 Modelling the Cane Components

4.3 Soluble solids model


The soluble solids (brix) in cane include around 80-90% sucrose and 10-20%

impurities. These soluble impurities are further classified as reducing sugars, organic

matter, inorganic compounds, nitrogenous bodies etc (Walford, 1996). In this model,

the soluble solids have been divided into sucrose, reducing sugars, ash, proteins and

the remainder. This section discusses methodologies to calculate the flows of these

soluble solid components.

4.3.2 Sucrose

For the purpose of the sucrose model here, a more appropriate model for

sucrose was developed by the Sugar Research Institute in the early 1980’s. This

approach is based on the concept of a purity ratio to determine the pol content of

bagasse from the brix content. The concept was developed to determine the pol in

bagasse although in this model it is used to determine the sucrose in bagasse.

The purity ratio was defined as:

푍푅 = 푍푍 4.1

where 푍푅 is purity ratio of the nth mill,

푍 is purity of bagasse of the nth mill,

푍 is purity of cane.

The purity ratio can be calculated from cane and bagasse analysis data and

assumed to be constant for the prediction of sucrose in bagasse from brix in bagasse.

Once the fibre, brix and moisture of the bagasse from each milling unit have

been predicted using the MILEX model as presented in chapter 3, the purity ratio

concept can be used to calculate the pol or sucrose component of the brix of each

bagasse.

Firstly, the purity of the bagasse is calculated by rearranging equation 4.1.

푍 = 푍푅 × 푍 4.2


The sucrose of the bagasse is determined from the equation:

푃 = 푍 × 푃 4.3

where 푃 is the sucrose% bagasse of the nth mill,

푃 is the brix% bagasse of the nth mill.

The mass flow of sucrose in bagasse is determined from the equation:

푚̇ = 푃 × 푚̇ 4.4

where 푚̇ is mass flow of sucrose in bagasse of nth mill,

푚̇ is total mass flow in bagasse of nth mill.

Equations 2.27 to 2.32 can be applied to the mass flow of sucrose in the milling

train. While the mass flow of sucrose in the bagasse streams has been determined,

the mass flow of sucrose in the expressed juice streams, return stream from the juice

screen and mixed juice stream are unknown in the model. Hence, there are seven

unknowns and only six equations. To solve the model, the approach reported by

Loubser (2004) was adopted and it was assumed that the purity of the return stream

from the juice screen (cush return) and the purity of mixed juice are the same.

4.3.3 Reducing sugars

This section describes the investigation undertaken to explore the Fernandes

(2003) model and the Vukov (1965) model. The section also describes a new

parameter developed to track the flow of reducing sugars in the milling process.

Investigating Fernandes’ model

Fernandes (2003) proposed empirical equations to determine the reducing

sugars in the milling process and the equations are described in section 2.6.3 of the

literature review.

The validity of Fernandes (2003) model was tested by conducting a 37

numerical factorial experiment to study the effect of individual factors on reducing

sugars levels. In particular, the level of inversion of sucrose to reducing sugars and

the conditions for that inversion were of interest. To conduct this assessment an


invert ratio (퐼푅) was defined as the ratio of reducing sugars to sucrose. The invert

ratio of cane and mixed juice are determined from:

퐼푅 =푃푃 4.5

퐼푅 =

푃푃 4.6

where 퐼푅 is the invert ratio of cane,

푃 is percent reducing sugars in cane,

푃 is percent sucrose in cane,

퐼푅 is the invert ratio of mixed juice,

푃 is percent reducing sugars in mixed juice,

푃 is percent sucrose in mixed juice.

The parameters and their levels examined in the experiment are presented in

Table 4.1 and the theory of the investigation model is described below.

Table 4.1 Parameters and their levels studied to determine the effect on invert

ratio

Parameter Symbol Levels Cane fibre content (%) PCF 10, 15, 20 Ratio of cane sucrose content to cane fibre content XCP 0.6, 0.8, 1.0 Cane purity (%) ZC 80, 85, 90 Total sucrose extraction (%) EP 90, 94, 97 Final bagasse purity ratio ZR 0.60, 0.75, 0.90 Final bagasse moisture content (%) PBnW 45, 50, 55 Added water % fibre XI 200, 250, 300

The sucrose content of cane was determined from:

푃 = 푋 × 푃 4.7

The brix content of cane was determined from:

푃 =푃푍 × 100 4.8


where 푃 is the brix content of cane,

푍 is the purity of cane.

The ratio of moisture to fibre in cane was determined from:

푋 =푃푃 4.9

푃 = 100− 푃 − 푃 4.10

where 푃 is the moisture content of cane,

푋 is the fraction of moisture in cane.

The purity of bagasse was determined from:

푍 = 푍푅 × 푍 4.11

where 푍 is the purity of bagasse,

푍푅 is the purity ratio of the final bagasse.

The ratio of sucrose to fibre in bagasse was determined from:

푋 = 1−퐸

100 × 푋 4.12

where 푋 is the ratio of sucrose to fibre in bagasse,

퐸 is the total sucrose extraction.

The ratio of brix to fibre in bagasse was determined from:

푋 =푋푍 × 100 4.13

where 푋 is the ratio of brix to fibre in bagasse.

The fibre content of the bagasse was determined from,

푃 =100− 푃푋 + 1 4.14

where 푃 is the fibre content of bagasse.


The brix content of bagasse was determined from:

푃 = 100− 푃 − 푃 4.15

where 푃 is the brix content of bagasse.

The sucrose content of bagasse was determined from,

푃 =푃 × 푍

100 4.16

where 푃 is the brix content of bagasse.

The ratio of moisture to fibre in bagasse was determined from:

푋 =푃푃 4.17

where 푋 is the ratio of moisture to fibre in bagasse.

The ratio of brix to fibre in mixed juice was determined from,

푋 = 푋 + 푋 − 푋 4.18

where 푋 is the fraction of brix in mixed juice,

푋 is the fraction of brix in imbibition.

The ratio of sucrose to fibre in mixed juice was determined from:

푋 = 푋 + 푋 − 푋 4.19

where 푋 is the fraction of sucrose in mixed juice,

푋 is the fraction of sucrose in imbibition.

The purity of mixed juice was determined from:

푍 =

푋푋 × 100 4.20

where 푍 is the purity of mixed juice.

The ratio of moisture to fibre in mixed juice was determined from,

푋 = 푋 + 푋 − 푋 4.21


where 푋 is the fraction of moisture in mixed juice,

푋 is the fraction of moisture in imbibition.

The brix content of mixed juice is determined from:

푃 =

푋푋 + 푋 × 100 4.22

where 푃 is brix content of mixed juice.

The sucrose content of mixed juice is determined from,

푃 =

푃 × 푍100 4.23

The empirical equations presented by Fernandes (2003) were used to determine

the reducing sugars in cane and mixed juice. The factorial experiment was setup to

determine the effect of each parameter shown in Table 4.1 on reducing sugars

extraction.

Figure 4.1 shows the results of the experiment. Along the horizontal axis are

the experimental factors. Each factor level is presented on a vertical bar for the

factor. The value for the ratio of the invert ratio for mixed juice to the invert ratio for

cane at each factor level represents the average result for all tests of the experiment

conducted at that factor level. For example, the 20% cane fibre content level shows

an average ratio of about 1.14, representing the average result for the one-third of the

tests conducted at that factor level.


1.05

1.10

1.15

1.20

Inve

rt ra

tio m

ixed

juic

e / i

nver

t rat

io c

ane

Factors

10

15

200.60.8

1 808590

90

94

97

0.60.75

0.9

45

50

55

200

250

300

PCF XCP ZC EP ZR PBnW XI

Figure 4.1 Exploration of invert ratio model

The results show that the ratio of cane pol content to the cane fibre content

(푋 ), purity of cane (푍 ) and purity ratio of the final bagasse (푍푅) were the factors

with the least impact and imbibition level(푋 ), moisture content of bagasse (푃 )

and sucrose extraction of the milling train (퐸 ) were the factors with the highest

impact on the ratio. It seems unlikely that three of the four factors associated with the

sugar content would have less impact than the factors associated with the fibre and

moisture contents, indicating some doubt that this model accurately represents the

flow of reducing sugars from cane into mixed juice. If the model was assumed

correct, the fact that the ratio exceeds 1.0 implies that sucrose inversion in occurring

at a significant level in the milling process.

Investigating Vukov’s model

Vukov (1965) proposed a formula to determine the rate of hydrolysis

(inversion) for solutions with sugar concentration between 0 and 0.9 g/ml,

temperature between 20 and 130 °C and pH between 1 and 6.5. The calculated

reaction rate constant (ka) for each mill in the milling train is shown in Table 4.2.

These values of rate constant were calculated by using the following data. Kent


(2010a) reported typical temperatures of mill juices. The pH of the juices was

assumed to be a constant 5.5. Russell (1968) provided a formula to calculate the

density of juice and the sucrose concentration was obtained from the brix (Bureau of

Sugar Experiment Stations, 2001).

Table 4.2 Rate of hydrolysis in mill products

Temp (0C)

pH Density (g/ml)

Sucrose conc. (g/ml)

Log ka ka (min-1) Stream

98 - - - - - Added Water

30 5.5 1.068 0.196 -7.36 4.37 × 10 #1 Mill 50 5.5 1.053 0.132 -6.23 5.88 × 10 #2 Mill 60 5.5 1.042 0.071 -5.65 2.24 × 10 #3 Mill 69 5.5 1.034 0.043 -5.18 6.61 × 10 #4 Mill 77 5.5 1.024 0.022 -4.79 1.62 × 10 #5 Mill

While it is not easy to convert a reaction rate constant to a likelihood of

inversion, the results show that inversion is much more likely to occur in #4 and #5

mills than earlier in the train, due to an increase in the reaction rate constant of two

orders of magnitude. Since only about 10% of the sucrose is extracted from #4 and

#5 mills and the time that the juice is held at this temperature is at most about two

minutes, it seems likely that inversion is limited to a very small fraction of the

sucrose extracted to mixed juice (probably less than 10%).

The conclusion derived from Vukov (1965), Van Der Pol (1955) and Rein

(2007) contradicts the Fernandes (2003) model. To resolve these contradictory

conclusions, an experimental investigation was undertaken to measure the reducing

sugars in the milling process.

Experimental investigation

The experimental investigation was undertaken at Isis sugar mill in Australia.

Samples of shredded cane, first expressed juice, and mixed juice were collected

while processing large rakes of cane (continuous supply of cane in rail wagons from

a single field) to avoid mixing of cane supplies. The first expressed juice and mixed

juice samples were filtered and frozen to avoid degradation of the sugars. The

shredded cane sample was processed quickly to measure the moisture content and the


sucrose and brix content by disintegration. The disintegrator extract was collected

and stored along with the first expressed juice and mixed juice samples. In total eight

different rakes of cane were sampled in this way. Method 5 of the BSES laboratory

manual (Bureau of Sugar Experiment Stations, 2001) was followed in the sampling

and mixing of shredded cane. All juices were stored at -20 °C and thawed prior to

analysis.

In the laboratory, the brix of samples was measured at ambient room

temperature using a Bellingham and Stanley RFM 342 Refractometer. Edwards

(1996) states that the Refractometer brix measurements are known to be inaccurate

with low purity and the dry substance method is more accurate. The samples

collected in the experiment are shredded cane, first expressed juice and mixed juice,

all of which have relatively high purity (above 80%). The dry substance method

(Method 19) of the BSES laboratory manual (Bureau of Sugar Experiment Stations,

2001) used by Edwards (1996) is tedious and time consuming and, given the time

constraints, was not considered justified for this experiment.

The separation and quantification of sugars, in duplicate, was conducted by

high performance ion chromatography coupled with pulse amperometric detection

(HPIC_PAD) based on the ICUMSA Method GS7/8/7-24. Analyses were carried out

on a Waters HPIC_PAD system coupled to a Waters 2465 Electrochemical Detector

(Thai & Doherty, 2011). Table 4.3 shows the sugar analysis results from the Isis

experiment. Note that reducing sugars (RS) has been defined as the sum of glucose

and fructose.


Table 4.3 Sugar analysis results from Isis experiment

Sample Stream Sugar (% sample) 푹푺푺풖풄풓풐풔풆

Glucose Fructose Sucrose 1 Cane

FEJ JM

0.63 0.76 1.61

0.57 1.60 1.15

7.28 19.17 16.55

0.1706 0.1633 0.1657

2 Cane FEJ JM

0.36 0.44 0.38

0.40 0.47 0.42

7.78 21.51 15.00

0.0986 0.0930 0.0933

3 Cane FEJ JM

0.25 0.59 0.49

0.29 0.64 0.53

9.74 23.56 15.56

0.0558 0.0521 0.0514

4 Cane FEJ JM

0.47 0.97 0.66

0.50 1.48 0.98

9.07 22.36 12.61

0.1003 0.0939 0.0977

5 Cane FEJ JM

0.84 0.33 1.25

0.87 0.29 1.24

12.61 20.07 17.35

0.1335 0.1244 0.1268

6 Cane FEJ JM

0.71 1.12 0.86

0.53 1.25 0.99

10.63 22.20 14.98

0.1159 0.1072 0.1235

7 Cane FEJ JM

0.84 1.67 1.30

0.66 1.43 1.00

8.77 18.81 14.07

0.1710 0.1573 0.1639

8 Cane FEJ JM

0.66 1.27 1.04

0.54 1.23 0.96

8.54 20.07 17.30

0.1255 0.1282 0.1159

Table 4.4 shows the results of the analysis of variance for the reducing sugars

experiment results shown in Table 4.3.

Table 4.4 Analysis of variance for reducing sugars experiment

Degree of freedom Mean square F Value Pr (F)

Block 7 0.00365 38.913 - Sample 2 0.00043 4.6105 0.0289

Residuals 14 0.00009

The analysis of variance found no statistically significant difference between

the invert ratios of cane and mixed juice (the significant result shown in


Table 4.4 was for the invert ratio between first expressed juice and mixed juice. It is

noted, however, that the experiment was conducted with an added water temperature

of 60 °C, a low value by Australian standards.

Considering the available evidence, it is concluded that the inversion of sucrose

in the milling process is negligible and the ratio can be assumed constant

for all mills in the milling train.

The ratio or invert ratio can be calculated from cane analysis:

퐼푅 =푃푃 4.24

where 퐼푅 is the invert ratio,

푃 is the reducing sugars% cane,

푃 is sucrose% cane.

The reducing sugars in the bagasse streams can be determined from:

푃 = 퐼푅 × 푃 4.25

where 푃 is the reducing sugars content in bagasse of the nth mill,

푃 is the sucrose content in bagasse of the nth mill.

As for the sucrose mass balance described above, the mass flow equations 2.27

to 2.32 along with the assumption that the invert ratio of the return stream from the

juice screen is the same as the cane invert ratio, can be applied to model the flow of

reducing sugars through the milling process.

4.3.4 Soluble ash

In Australian factories, soluble ash in mixed juice and final bagasse are not

measured on a routine basis. Some data, however, was collected during the Isis

experiment described above. Table 4.5 shows the ash analysis results and ash/brix

ratio for cane, first expressed juice and mixed juice. The soluble ash was measured

by Method 26, Determination of ash in sugar products by the single sulphation

method (Bureau of Sugar Experiment Stations, 2001).


Table 4.5 Soluble ash in mill products

Sample Stream S.Ash (%)

S.Ash/Brix (× ퟏퟎ ퟐ)

1 Cane FEJ JM

0.58 0.48 0.43

2.92 2.53 3.09

2 Cane FEJ JM

0.60 0.44 0.33

2.94 2.21 2.17

3 Cane FEJ JM

0.64 0.46 0.28

3.17 3.50 3.43

4 Cane FEJ JM

0.94 0.71 0.56

3.07 3.52 4.20

5 Cane FEJ JM

0.72 0.66 0.36

3.54 4.36 2.59

6 Cane FEJ JM

0.59 0.35 0.24

4.46 1.70 1.63

7 Cane FEJ JM

0.57 0.30 0.35

4.36 1.60 5.16

8 Cane FEJ JM

0.52 0.43 0.33

3.06 2.56 2.68

An analysis of variance did not identify any statistically significant differences

between the ratios of the three streams. As a result, the flow of soluble ash is

modelled assuming the ratio is constant for all the bagasse streams in the

milling process, in agreement with the implied assumption of equal soluble ash and

brix extraction made by Wienese and Reid (1997).

푅 =푃푃 4.26

푅 is ash/brix ratio,


푃 is percent soluble ash in cane,

푃 is percent brix in cane.

Applying the mass balance equations 2.27 to 2.32 and assuming the ash on

brix ratio is the same for the return stream from the juice screen; the model can be

used to complete a mass balance of soluble ash.

4.3.5 Proteins

Proteins are present in amino acids in peptide linkages and account for 0.1 to

0.2% of dry solids in cane (Martin, 1958). Wiggins (1958) reported that non-sugars

in cane consist of 9% proteins, 9.5% amino acids and 15.5% amino acid amides.

Protein content was measured in the Isis experiment and the results are

reported in Table 4.6. The colorimetric detection and quantification of protein in the

samples was conducted using the bicinchoninic acid (BCA) Protein Assay Kit

(Pierce, Bonn, Germany); 25μL sample/200 μL BCA working reagent; 37 °C for 30

minutes; 562 nm, based on the manufacturer’s instructions. The quantification (in

triplicate) of protein was carried out on a Beckman AD 200 UV/VIS Plate Reader

(Thai & Doherty, 2011).


Table 4.6 Protein in mill products

Sample Stream Proteins (%) (× ퟏퟎ ퟑ) Proteins/Brix(× ퟏퟎ ퟒ) 1 Cane

FEJ JM

6.39 9.30 7.40

4.80 4.73 8.73

2 Cane FEJ JM

5.99 3.33 8.30

4.19 1.42 5.33

3 Cane FEJ JM

3.80 3.60 0.60

0.21 1.88 0.43

4 Cane FEJ JM

8.87 6.60 8.40

4.69 3.25 6.25

5 Cane FEJ JM

6.27 3.00 3.40

7.97 1.97 9.48

6 Cane FEJ JM

7.53 5.22 2.90

5.65 5.14 1.96

7 Cane FEJ JM

3.75 4.20 8.30

3.00 2.20 1.25

8 Cane FEJ JM

5.33 4.90 1.80

3.11 2.89 1.46

Modelling the flow of proteins raises the issue of where in the sugar production

process proteins are precipitated (Kent, 2010b). When they are precipitated, the

proteins transfer from a component of brix to a component of insoluble solids.

The effect of temperature on precipitation and dissolution of proteins was

explored by Macritchie (1973). Macritchie measured the concentration of BSA (an

animal protein) in the dilute protein phase as a function of total concentration at

temperatures of 45, 50, 55, 60 and 65 °C. It was found that for all temperatures, no

significant precipitation of BSA was recorded.

The results obtained by Macritchie (1973) could not be confidently transferred

to sugarcane proteins. As a result, an experiment was conducted to study the effect

of temperature on protein precipitation in first expressed juice, using samples from


the Isis experiment. The experiment was conducted using the Bio-Rad Protein Assay.

The samples in triplicate were exposed for 1 minute at temperatures of 20, 30, 40, 50,

60, 70 and 80 °C. Figure 4.2 shows the effect of temperature (°C) on proteins

concentration (mg/ml) in solution. By inference, proteins not in solution have been

precipitated.

Tepm (0C), 1 minute exposure

10 20 30 40 50 60 70 80 90

Tota

l pro

tein

(mg/

ml)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35Sample 2-FEJSample 3-FEJSample 4-FEJ

Figure 4.2 Effect of temperature on protein concentration

Sample 2-FEJ had the highest protein content and shows a substantial

reduction in protein concentration at temperatures over 60 °C. The two other

samples with lower protein contents did not show a substantial reduction in protein

concentration in the temperature range explored.

To consider the likelihood of protein precipitation in the milling train, the

temperature data from five-mill milling trains reported by Kent (2010a) were used

(as in the earlier reducing sugars discussion). Even with the highest temperature

added water, the bagasse temperature only exceeds 60 °C from #4 mill to #5 mill.

Since only about 10% of the brix is extracted from #4 and #5 mills and at most only

about half of the proteins in that brix are precipitated, it seems likely that no more


than 5% of the proteins in cane are precipitated in the milling train. Consequently,

the assumption that proteins remain part of brix through the whole milling train

seems adequate.

For modelling purposes, it was assumed that the ratio of proteins to brix

remains constant throughout the milling process:

푅 =

푃푃 4.27

where 푅 is proteins/brix ratio,

푃 is percent proteins in cane,

푃 is percent brix in cane.

Applying the mass balance equations 2.27 to 2.32 assuming the proteins on

brix ratio is the same for the return stream from the juice screen; the model can be

used to complete a mass balance of proteins.


The soluble solids model tracks the flow of sucrose, reducing sugars, soluble

ash and proteins in the milling process. With the sucrose component modelled, the

sucrose extraction of individual milling units and the entire milling train can be

determined. The reducing sugars, soluble ash and proteins are part of the soluble

impurities and affect the further processing of juice and the recovery of the sugar

production process.

4.4 Insoluble solids


The insoluble solids, termed as cane fibre, account for 10-20% of the cane. The

cane fibre includes about 80-90% of true fibre and 10-20% of mud solids. There are

insoluble ash components in both true fibre and mud solids (Bureau of Sugar

Experiment Stations, 2001).

4.4.2 An overall mass balance model

Kent (2010b) developed a model to determine bagasse production from an

insoluble solids mass balance. The model tracked ash and non-ash components and

true fibre and mud solids components of the insoluble solids but considered only the


overall insoluble solids mass balance of the milling train and not the insoluble solids

mass flows for the individual milling units.

The Kent (2010b) model has been used as the basis of a method to determine

the split of true fibre and mud solids into final bagasse and mixed juice. This section

describes the methodology to calculate the insoluble solid component (true fibre and

mud solids) flows.

Kent (2010b) developed a model to determine the mass of insoluble solids in

bagasse from an insoluble ash mass balance as described in section 2.7.2. Kent

(2010b) presented equations to calculate the equivalent mass of insoluble solids in

mud and the mass of insoluble ash in mud.

The model of Kent (2010b) was extended to determine the true fibre and mud

solids in bagasse. This section describes the methodology to calculate the insoluble

solids component (true fibre and mud solids) flows.

The insoluble solids mass flow in cane is determined from:

푚̇ = 푃 × 푚̇ 4.28

where 푚̇ is cane fibre rate,

푚̇ is cane rate,

푃 is percent fibre in cane.

Insoluble solids (fibre) content in cane and cane rate are both routine

measurements at Australian sugar factories.

It is not common to measure either true fibre or mud solids in cane. The

insoluble ash in cane, however is routinely measured in some factories.

푚̇ = 푃 × 푚̇ 4.29

where 푚̇ is the mass flow of insoluble ash in cane,

푃 is the percent insoluble ash in cane.

Assuming the mud solids in cane is 100% ash, the true fibre and mud solids in

cane are determined from:


푚̇ =

푚̇ − 푚̇1− 푋

4.30

푚̇ = 푚̇ − 푚̇

where 푚̇ is mass flow of true fibre in cane,

푚̇ is mass flow of mud solids in cane,

푋 is the fraction of insoluble ash in true fibre in cane.

As indicated by Kent (2010b), Muller et al (1982) reported values of ash in

cane of 0.13% to 0.36%. If the insoluble solids content in cane is assumed to be

around 12% to 15%, the ash content of true fibre can be assumed to be 1.1% to 3.0%.

The 1 − 푋 term refers to the non-ash component of true fibre. Since it was

assumed that mud solids contain 100% ash, the non-ash component of mud solids is

zero.

The insoluble ash in true fibre and mud solids in cane are calculated from:

푚̇ = 푋 × 푚̇ 4.31

푚̇ = 푚̇ − 푚̇

where 푚̇ is the mass flow of insoluble ash in true fibre in cane,

푚̇ is the mass flow of insoluble ash in mud solids in cane.

The total ash in bagasse is measured in some Australian sugar factories. The

mass of ash in bagasse can be calculated from:

푚̇ = 푃 × 푚̇ 4.32

where 푚̇ is mass flow of total ash in bagasse,

푚̇ is total mass flow in bagasse.

The soluble ash in bagasse is calculated from the soluble solids model. The

insoluble ash in bagasse can be calculated from:

푚̇ = 푚̇ − 푚̇ 4.33

where 푚̇ is mass flow of insoluble ash in bagasse,


푚̇ is mass flow of soluble ash in bagasse.

The true fibre in bagasse can be calculated from:

푚̇ =

푚̇ − 푚̇1 − 푋

4.34

where 푚̇ is the mass flow of true fibre in bagasse.

푋 is the fraction of insoluble ash in true fibre in cane (which is assumed to be

the same as the fraction of ash in true fibre in bagasse).

The mud solids in mixed juice can be calculated from:

푚̇ = 푚̇ − 푚̇ 4.35

where 푚̇ is the mass flow of mud solids in mixed juice,

푚̇ is mass flow of mud solids in bagasse.

The ratio of mud solids in mixed juice to the mud solids in cane (푀 ) can be

calculated from:

푀 =

푚̇푚̇ 4.36

4.4.3 The model for a milling unit

In the MILEX model, described in the previous chapter, the separation

efficiency defined the proportion of total insoluble solids that is expressed with the

juice. While validating the MILEX model in section 3.6.1 on page 50, it is mentioned

that the values reported by Kent (2001) were actually true fibre measurements and

not total insoluble solids measurements. Consequently, the separation efficiencies

presented in section 3.6.3 relate more correctly to true fibre separation efficiencies.

In the extended MILEX model, we need to separate the true fibre and mud

solids. Hence, two separation efficiencies for each milling unit need to be defined.

Since the existing measurements of fibre in juice are true fibre measurements, true

fibre separation efficiency (STFn for the nth mill) is defined as one of them. The

separation efficiency for total insoluble solids is related to the true fibre separation

efficiency by:


푆 =푆퐹 4.37

where 푆 is the separation efficiency of total insoluble solids,

푆 is the true fibre separation efficiency,

퐹 is the true fibre factor.

The true fibre factor is a single factor that is applied to the separation efficiency

for all the milling units and the juice screen and provides a capability to ensure that

mud solids are correctly distributed between final bagasse and mixed juice as

discussed in section 4.5.1.

4.4.4 The model for the juice screen

In the juice screen model of the MILEX model, the screen efficiency defined

the proportion of total insoluble solids that is expressed with the cush. While

validating the model, it was mentioned that the measurements of cush return and

mixed juice by Kent (2001) were actually true fibre measurements. Hence the screen

efficiency presented in section 3.3 relates more correctly to true fibre screen

separation efficiency.

Like the milling unit model, the juice screen model has to be able to

distinguish between true fibre and mud solids. An approach similar to that used for

the milling unit model has been used for the juice screen:

푆 =푆퐹 4.38

where 푆 is juice screen separation efficiency of total insoluble solids,

푆 is true fibre juice screen separation efficiency.


Using fibre in juice measurements, the true fibre separation efficiency can be

calculated for a milling unit. Similarly, the true fibre juice screen separation

efficiency can be determined. The unknown parameter is the true fibre factor.

To determine the true fibre factor, the model should be run at different values

for true fibre factor and the ratio of mud solids in mixed juice to mud solids in cane


should be calculated. The required value for true fibre factor will be the value that

gives a ratio of mud solids in mixed juice to mud solids in cane equal to the ratio

calculated using equation 4.36.


The insoluble solids model tracks the flow of true fibre, mud solids and

insoluble ash in the milling process. With the insoluble solids divided as per the

above definition, the mud solids processed by each milling unit and the overall

insoluble solids impurities can be modelled.


4.5.1 The model calibration step

The model has been explored using data from a typical milling train of five

milling units, as used in the previous chapter. The mill performance parameters for

this milling train are presented in Table 4.7. The model has been solved for cane

containing 15.91% brix and 14.00% fibre. The imbibition% fibre was 200% and the

true fibre content of the mixed juice was 0.2%. It was assumed that the ratio

of the return stream was 11.50.

Table 4.7 Mill performance parameters corrected for juice fibre flows

Mill Filling ratio (COn)

Reabsorption factor (KOn)

Imbibition coefficient (ICon)

True fibre separation efficiency (STFn) (%)

#1 0.36 1.55 1.04 91.80 #2 0.46 1.67 0.87 91.12 #3 0.50 1.60 0.67 94.80 #4 0.55 1.59 0.57 95.41 #5 0.56 1.51 0.57 95.66

The effect of true fibre factor on the ratio of mud solids in mixed juice to the

mud solids in cane was determined and is shown in Figure 4.3. As the true fibre

factor increases, the mud solids content of the mixed juice stream increases. From the

data presented by Kent (2010b) for Condong and Broadwater factories in Australia,

the ratio of mud solids in mixed juice to mud solids in cane was calculated to be 0.5.

Figure 4.3 shows that this mud solids ratio can be achieved with a true fibre factor of

1.12.


True fibre factor

0.95 1.00 1.05 1.10 1.15 1.20

Mud

solid

s fra

ctio

n in

mixe

d ju

ice

0.0

0.2

0.4

0.6

0.8

1.0

Figure 4.3 Effect of true fibre factor on the ratio of mud solids in mixed juice to mud solids in cane

4.5.2 Results from the model

The mill extraction model provides the mass flows of brix and fibre in bagasse

and juice of individual milling units along with the return stream from the juice

screen and the mixed juice stream. Figure 4.4 shows the relative flows of brix and

fibre in all the mill streams.


Mill streams

C B1 B2 B3 B4 B5 J1 J2 J3 J4 J5 Js JM

Brix

mas

s flo

ws

(kg/

s)

0

5

10

15

20

25

30

35

Mill streams


Fibr

e m

ass

flow

s (k

g/s)

0

5

10

15

20

25

30

35

Figure 4.4 Mill streams brix and fibre mass flows

Using the purity ratio concept and the brix flow of the mill streams shown in

Figure 4.4, the mass flow of sucrose in the milling train can be modelled. The invert

ratio, ratio and ratio were used to model the flow of reducing

sugars, soluble ash, proteins respectively in the milling process. The invert ratio is

determined from Table 4.3. The ratio and ratio are determined from

Table 4.5 and Table 4.6 respectively.


Mill streams


Pol m

ass f

low

s (kg

/s)

0

5

10

15

20

25

30

Mill streams


Solub

le as

h m

ass f

low

s (kg

/s)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Mill streams


Redu

cing

suga

rs m

ass f

low

s (kg

/s)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Mill streams


Prot

eins m

ass f

low

s (kg

/s)

0.00

0.01

0.02

0.03

0.04

Figure 4.5 Mill streams soluble solids mass flow

The insoluble solids were modelled as discussed previously. Figure 4.6 shows

the flow of true fibre, mud solids and insoluble ash in the mill streams.


Mill streams


True

fibr

e m

ass f

low

s (kg

/s)

0

5

10

15

20

25

30

Mill streams


Inso

luble

ash

mas

s flo

ws (

kg/s)

0

2

4

6

8

Mill streams


Mud

solid

s mas

s flo

ws (

kg/s)

0

2

4

6

8

Figure 4.6 Mill streams insoluble solids mass flow

Figure 4.5 shows that the majority of soluble impurities end up in the mixed

juice stream while Figure 4.6 shows that the mud solids are more evenly divided

between final bagasse and mixed juice. The mud solids mass flow in #1 mill and #5

mill bagasse streams are very low. In the former, about 70% of the juice entering the

mill is expressed and so are the mud solids. In the case of the final mill, input mud

solids flow is reduced since there are no mud solids in the imbibition. The mud solids

flow in the return stream and the #2 mill bagasse streams are quite similar as the

model is calibrated with the true fibre factor to split the mud solids in final bagasse

and mixed juice correctly. The #2, #3, and #4 mill bagasse streams have the highest

of all the mill juice mud solids mass flow. The insoluble ash mass flows follow the

same trend as the mud solids mass flow.


4.5.3 Effect of impurities on extraction

In exploring the MILEX model in section 3.8.1, it was reported that brix

extraction is reduced when more insoluble solids are expressed in juice streams. With

this extended model, impacts of soluble impurities and mud solids on extraction can

be assessed.

The effect of soluble impurities on extraction is reasonably well understood

since both pol and brix are common measurements. The effect of mud solids on

extraction is not well understood.

Figure 4.7 shows the impact of true fibre factor, which varies the amount of

mud solids in the expressed juice and mixed juice streams, on brix extraction and

sucrose extraction. The other mill performance parameters (reabsorption factor and

imbibition coefficient) are kept constant. Although the true fibre contents of the

expressed juice and mixed juice streams remain constant as the true fibre factor

varies, the brix extraction drops about 0.4 units as the true fibre factor increases from

1.0 to 1.12, corresponding to a ratio of mud solids in mixed juice to the mud solids in

cane of 0.5. Although not completely correct, in the presence of modest knowledge

regarding the extraction mechanisms of mud solids and true fibre, it is assumed that

there is no difference in extraction mechanisms of true fibre and mud solids.

A value of 0.5 is typical of the usual split of mud solids between bagasse and

juice. Steindl (1998) assumed that the 40% of the dirt in cane (mud solids) was

washed in mixed juice. The Kent (2010) data showed a 50-50 split of mud solids in

bagasse and mixed juice at Condong and Broadwater sugar mills.


True fibre factor

1.00 1.05 1.10 1.15 1.20

Extra

ctio

n (%

)

93.4

93.6

93.8

94.0

94.2

94.4

94.6

94.8

95.0Brix ExtractionPol Extraction

Figure 4.7 Effect of true fibre factor on extraction

The pol extraction values show a similar trend as that of brix extraction with

increasing true fibre factor. The reduction in pol extraction, corresponding to the true

fibre factor of 1.12 is 0.35 units.


A model was developed using assumptions based on the best information

available to determine the flow of cane constituents through the milling process and

their distribution in juice and bagasse. The model monitors the flow of soluble and

insoluble constituents. It was found that the bagasse from the #1 mill and #5 mill had

less mud solids mass flow then from the other mills, while #2 mill had the highest

mud solids mass flows followed by #3 and #4 mills.

The developed model can assist in understanding the effect of mud solids on

brix and pol extraction. Reducing the true fibre factor increases the separation

efficiencies of individual milling units and reduces the insoluble solids content of

juice streams. Appreciable increases in brix and pol extraction could be achieved

through such actions.


The developed model is an integral part of a ‘whole factory model’ and tracks

the flow of specific cane components in the milling train so that their concentration

and mass flows are known for the downstream models.

The developed model is the first of its kind and provides some additional

insight regarding the flow of soluble and insoluble cane components and the factors

affecting their distribution in juice and bagasse. The model proves to be a good

extension to the extraction model (MILEX) to study the overall performance of the

milling train.

SysCAD Model Development 107

CHAPTER 5: SysCAD MODEL DEVELOPMENT


Process modelling of raw sugar factories has been undertaken using different

techniques over the years for the purpose of understanding and improving the

performance of the unit operations. For a mathematical model to be useful, it must be


be studied. Today, several commercial process modelling packages are available

with the ability to incorporate various unit operations and integrate them together.

An enhanced and comprehensive ‘milling process” simulation model of the

sugar milling industry has been developed as discussed in Chapter 3 and 4 and was

solved using Excel spreadsheets. To enable the developed models to analyse the

performance of the milling train and to assess the impact of changes and advanced

control options for improved operational efficiency, the developed models have been

incorporated in a proprietary software package ‘SysCAD’. The milling process

model in SysCAD is an integral part of a ‘whole of plant’ model.

This chapter describes work with SysCAD to develop a model of the milling

train of a raw sugar factory. The model is developed in static and dynamic modes of

the SysCAD software.

5.2 Modelling the milling process

Software packages like HYSIS (Aspentech, 2012) and SUGARS (SUGARS

International, 2011) have been used in the past to simulate sugar factory processes.

The HYSIS software is general process modelling software and does not contain a

specific milling unit model. The SUGARS package does contain a milling unit

model, but does not allow in-house process knowledge to be incorporated into the

models (Peacock, 2002).

The use of process modelling software in the South African sugar industry was

discussed and demonstrated by Peacock (2002). SIMULINK is a commercial

software system overlaid on the MATLAB programming language, which is widely

108 SysCAD Model Development

used in modelling, simulation and analysing steady state and dynamic systems using

bock diagrams. Peacock explored the SIMULINK system for the mass and energy

balance model in the cane diffuser system (an alternative extraction process to the

milling process). The model, which was at an early stage, consisted of a simple

material and enthalpy balance.

The use of a spreadsheet with constraint equations and a Newton-Raphson

technique to solve heat and mass balance models was demonstrated by Loubser

(2004). Loubser discussed the applicability of such techniques, when commercial

software such as SUGARS and SIMULINK is not available.

5.3 Process Modelling using SysCAD

SysCAD is a commercial process modelling software package developed by Kenwalt

Australia (KWA) and extensively used in mineral industries (SysCAD). SysCAD is a

powerful and versatile plant simulator. SysCAD offers dynamic simulation and the

ability to incorporate in-house models of individual stations. These abilities of

SysCAD were favoured for the development of a milling process model.

5.4 Static milling train model


In Chapter 3 an enhanced milling process model is presented to predict

extraction performance of the milling train. The model was solved using a

spreadsheet with the aid of an iterative calculation procedure and equation solver.

In Chapter 4 a model to track the flow of soluble and insoluble cane

components through the milling train as presented. The developed model was the

first of its kind and provided some novel theories regarding the flow of soluble and

insoluble solid cane components and the factors affecting their distribution in juice

and bagasse. The model was shown to be a good extension to the extraction model to

study the overall performance of the milling train.

The above models have been incorporated in the SysCAD software to develop

a comprehensive milling process model. The developed model was designed as an

integral part of a whole factory model and tracks the flow of specific cane

components so that they are available for the downstream process models.


5.4.2 Developing a milling unit model in SysCAD

The preliminary development work was conducted by modelling a milling unit

as a tank (a standard object in SysCAD) and programming the tank to behave as a

milling unit as defined in Chapter 3. KWA subsequently developed a milling unit

object based on the MILEX model.

Figure 5.1 shows a SysCAD milling unit object. The full model as presented

consists of feeders (for input and output of the process streams), the milling unit and

a vent for heat transfers. The inputs to the milling unit are feed bagasse and

imbibition and the outputs are delivery bagasse and expressed juice. The model when

solved splits the feed components between delivery bagasse and expressed juice.

Some water may be evaporated in the process and the vent accounts for this loss of

water.

Figure 5.1 Single milling unit in SysCAD

Figure 5.2 shows the access window of the milling unit. The white cells

represent input parameters while the grey cells represent the calculated parameters.

The window shows the milling unit performance parameters defined in the MILEX


model: filling ratio (called FillingRatio), reabsorption factor (called

ReAbsorptionFac) and imbibition coefficient (called ImbibitionCoeff). Using the

approach defined in section 4.4.3, the separation efficiency is determined by the true

fibre separation efficiency (called TF_SepEff) and the true fibre factor (called

TrueFibreFactor). The window also shows the bagasse purity ratio (called

BagPurityRatio) that, as described in section 4.3.2, is used to determine the bagasse

sucrose content from the bagasse brix content.

Figure 5.2 Milling unit access window

Mimicking the MILEX model, the SysCAD model can be used in both

analytical mode and predictive mode.

Predictive mode is the default mode of operation of the SysCAD milling unit

model. The performance parameters are mill inputs as shown in Figure 5.2 as the

boxes with the white background.


To operate the SysCAD milling unit model in analytical mode, the

Analytical_Controller object shown in Figure 5.3 is required. This object contains

four PID controllers to determine the performance parameters (..., imbibition

coefficient, ...). Figure 5.3 shows the imbibition coefficient PID controller. The Set

Point_Tag is the bagasse brix to be achieved by the mill, the measured tag

(Meas_Tag) is the bagasse brix calculated using the current imbibition coefficient

value and the Control_Tag is the imbibition coefficient that has been selected by the

controller to achieve the set point bagasse brix.

In section 3.2.2 a corrected filling ratio term is developed which can be

calculated from the conventionally defined filling ratio. The corrected filling ratio is

calculated using the delivery bagasse fibre rate, instead of the cane fibre rate. In

predictive mode, it is considered more likely that the corrected filling ratio will be

used. In analytical mode, however, it is considered more likely that the

conventionally defined filling ratio will be used. In SysCAD, the conventionally

defined filling ratio is called FillingRatioM. The corrected filling ratio is the

FillingRatio shown as an input in Figure 5.2. There is a filling ratio Checkbox called

UseFillingRatio shown in Figure 5.2 that determines which of the two filling ratios is

used as an input.


Figure 5.3 Analytical controller access window

5.4.3 Developing a juice screen model in SysCAD

In the MILEX model a juice screen model is developed to separate the

insoluble solids in the juice streams and return the stream with high fibre content to

the milling train while sending the screened juice for processing. In the extension of

the model, in section 4.4.4 a true fibre factor term is developed to account for the

mud solids content of mixed juice stream. The juice screen model is incorporated in

the SysCAD software in the similar mode, with the true fibre content of the cush

return and the mixed juice being the inputs.

The juice screen model divides the juice stream inputs (from #1 mill and #2

mill in Figure 1.1) between screened juice (mixed juice in Figure 1.1) and the cush

return (return stream from the juice screen in Figure 1.1).


Figure 5.4 shows the access window to a juice screen object. As shown in

Figure 5.4, there are three inputs to the model. The first input is the true fibre

content of the cush return (called ReturnStreamTFConcReq). The second input is the

true fibre content of the screened juice (called MixedJuiceTFConcReq). The third

input is the true fibre factor (called TrueFibreFactorReq). In the juice screen object

(similar to the milling unit object), the true fibre factor works with the true fibre

separation efficiency of the juice screen to determine the overall insoluble solids

content of the screened juice.

It is also possible, instead of defining the true fibre factor, to define the split of

mud solids between screened juice and cush return (called FracMudtoJuice). To

configure this alternative approach, the SetFracMudtoJuice Checkbox should be

selected.

Figure 5.4-Juice screen access window


5.4.4 Developing a milling train model in SysCAD

The milling unit model developed in SysCAD was constructed by linking

together feeders, milling units and a juice screen as shown in Figure 5 Feeders are

only required for inputs and outputs to the entire model (cane, imbibition, final

bagasse and mixed juice). Most object output streams are input streams to another

object and so don’t require a feeder.

Figure 5.5 Milling train model in SysCAD

In Figure 5.5, various controllers are shown around the milling train model. As

shown in Figure 5.5, the PGM controllers are Inputs, Multipliers, Multipliers1,

Analysis, Imbibition_Control and General_Controller. The PGM controller is used to

simulate plant operations and calculate displays.

The Inputs controller is used to enter the bagasse analysis data into the model

for analytical mode. The Filling_Ratio controller calculates the corrected filling ratio

values from the filling ratio values. These filling ratio values are used in the

Multipliers and Multipliers1 controllers.


The Multipliers and Multipliers1 controllers calibrate the model. The

controllers calculate the mill performance parameters from empirical equations,

determine the multipliers and use the multipliers when predicting extraction

performance as defined in the multiplier section earlier.

The Analysis controller shows the bagasse analysis output in the predictive

mode. The Imbibition_Control is used to change the added water rate as fibre% cane

is changed. The General_Controller uses a dynamic tag function which turns the

model from analytical mode to predictive mode and vice versa. It also controls the

flow of cush return to the #2 or the # 3 mill.

TF_SE, Reab_Fac, Imb_Coeff and Pty_Rat in Figure 5.5 are PID controllers,

like Analytical_Controller in Figure 5.3. Because there are so many more

performance parameters to be controlled in a whole milling train, the true fibre

separation efficiency, reabsorption factor, imbibition coefficient and purity ratio

controllers have been split up between separate controller objects.

As mentioned earlier, the default mode of the extraction model is predictive.

The MILL TUNING INPUTS in the General_Controller (Figure 5.6) are used to

switch from predictive mode to analytical mode and vice versa. The Tune01 as

shown in Figure 5.6 turns the TF_SE, Reab_Fac, Imb_Coeff and Pty_Rat PID

controllers on and off with the command of 1 and 0 respectively. This creates

flexibility in operating the model from one controller. When the model is running in

analytical mode, the Multipliers1 controller is switched off. When turning off the

Analytical_Controller, the predictive mode is turned on by default and the controller

turns off all the Analytical_Controllers and turns on the Multipliers1 controller. With

the multipliers determined and fixed, the model calibration process is started and the

model can be run in predictive mode.

The table shown below the milling train model in Figure 5.5 is an annotation

table. Such tables can be created in SysCAD to display outputs of the model. In this

case, the brix extraction of each milling unit and the cumulative brix and pol

extraction of each milling unit are displayed.

A tie is installed in the model to split the cush return from the juice screen to

#2 or #3 mill. A PGM controller can be used to control this split.


Figure 5.6 General controller- Mill tunning

5.4.5 Excel reports

SysCAD includes a function to “communicate” with Microsoft Excel through

automation Inputs can be passed to the model though an Excel interface and values

can be exported to Excel when the model is solved.

When the report is called from the SysCAD software, it looks for the Tag_List

command to generate the values the user has assigned in the excel spreadsheet. When

values from excel spreadsheet are to be used as inputs in SysCAD the SetTag_List

function is used and SysCAD takes the values from excel and feeds into the

controller. An example of the generated report is shown in the next page. As shown

in the report, the blue cells are the inputs and the white cells are the outputs of the

model. The yellow cells show the project details, such as the case study and

date/time of the report generated.


SysCAD-MILEX v1.0 Milling train extraction performance simulator PlantModel.System.Version SysCAD 9.2 Build:134.12203 (Sep 14 2012) PlantModel.System.PrjName Test.scn PlantModel.System.DateAndTime 21/09/2012 8:35 Operation mode Predictive Mode

Calibration On Fields in blue are required Bagasse analysis Cane #1 #2 #3 #4 #5 Cush Return Moisture % bagasse 71.31 52.78 54.45 54.99 57.53 56.13 81.31 Brix % bagasse 15.92 10.53 7.67 5.16 3.78 2.85 11.62 Fibre % bagasse 12.77 38.31 39.46 41.33 39.92 41.59 8.00 Pol % bagasse 13.31 7.97 5.36 3.09 2.04 1.80 9.86 Purity % 83.61 75.72 69.97 59.81 53.99 63.23 84.81 Juice analysis #1 #2 #3 #4 #5 Mixed Juice Moisture% Juice 82.81 89.39 93.80 95.90 96.94 87.34 Brix% Juice 15.50 8.78 4.29 2.14 1.08 12.49 Fibre% Juice 2.00 2.00 2.00 2.00 2.00 0.20 Pol% Juice 13.28 7.30 3.14 1.21 0.38 10.59 Purity % 85.67 83.15 73.17 56.33 35.30 84.81 Extraction Performance Brix - Mill extraction (%) 79.45 17.22 36.16 24.49 33.09 - Total extraction (%) 79.45 82.99 89.14 91.80 94.51 Pol - Mill extraction (%) 81.39 23.51 45.42 31.84 21.63 - Total extraction (%) 81.39 85.77 92.23 94.71 95.85


Operational data

Crushing rate (t/h) 547 Mill length (mm) 2550 2550 2550 2134 2134 Work opening (mm) 80 89 84 99 70 Mill speed (mm/s) 190 168 173 179 178 Compaction (kg/m3) 500.60 508.90 523.61 513.09 734.98 Filling ratio 0.3272 0.3326 0.3422 0.3354 0.4804 Corrected Filling ratio 0.2911 0.3474 0.3566 0.3501 0.4725 Cush Return 0.00 1.00 0.00 0.00 0.00 Imbibition Added water % fibre 0.0 0.0 0.0 7.0 280.0 Performance indicators Separation efficiency (%) 89.21 90.76 94.22 94.47 94.10 Reabsorption factor Actual 1.0306 1.1989 1.1700 1.1924 1.4985 Calculated 1.2788 1.2906 1.1866 1.0910 1.3020 Multiplier 0.8059 0.9290 0.9860 1.0929 1.1509 Imbibition coefficient Actual 1.0290 0.5732 0.6434 0.4394 0.4838 Calculated 1.0500 1.1517 0.5147 0.4406 0.3957 Multiplier 0.9800 0.4977 1.2501 0.9972 1.2228 Purity ratio Actual 0.9055 0.8367 0.7153 0.6456 0.7561 Calculated 0.9150 0.9000 0.8750 0.8600 0.8550 Multiplier 0.9896 0.9296 0.8174 0.7507 0.8843



As described, the SysCAD model has essentially the same functionality as the

spreadsheet models described in Chapter 3 and 4. The main benefits of the SysCAD

model over the spreadsheet models are:

The equations of the model are well defined for the model objects. They

cannot be accidentally changed as is possible with cell formulae in a

spreadsheet. Each object has the sugar properties defined in the object and

when connected with another model object passes on the properties without

possibility of error.

Equations for a milling unit object are only defined once for that type of

object instead of having to be defined for every occurrence of that object as is

usual in a spreadsheet.

The links between objects are visible on the model schematic (such as Figure

5.5) and not hidden in cell formulae. If the objects are connected in a

different way, it is a simple matter of moving the link without having to

redefine a series of cell formulae.

An example of this last point is discussed in a case study below.

In section 3.6.4 the significance of the juice screen in the MILEX model is

reported. One of the things the model did not throw light on was the position of the

juice screen and its significance on extraction. Although the common position is

accepted to be before the #2 mill, several Australian mills add the return stream to

the #3 mill or distribute the cush between #2 and #3 mill. Kent (2001) reported this

scenario and mentioned that redirecting the entire or part of cush return from #2 mill

to #3 mill increases the throughput but at an extraction penalty. Kent (2001) did not

report any values for reduction in extraction.

Figure 5.7 shows the more typical case with the return stream from the juice

screen being returned before #2 mill. Figure 5.8 shows the alternative case where the

return stream from the juice screen is returned before #3 mill. This second scenario

was defined by deleting the link between the juice screen object and the #2 mill

object and creating a new link between the juice screen object and the #3 mill object.

All the analytical controllers are turned off along with the filling ratio controller. The

Multipliers1 controller is turned on to calculate the actual mill parameters from the


multipliers and are used to predict extraction when cush return is shifted to the #3

mill.

Brix extraction and pol extraction results from the models are shown in the

annotation table in each of the figures. A reduction of 0.76 units of brix extraction

and 0.64 units of pol extraction is recorded when the return stream is directed to the

#3 mill.

Figure 5.7- Model application-cush return to #2 mill


Figure 5.8-Model application-cush return to #3 mill


A static milling train model has been developed in SysCAD that has a very

flexible configuration and is easy to use. Compared to a spreadsheet, SysCAD

provides the ability to change the layout of the process such as changing flows and/or

increasing or decreasing the length of the milling train without the need to change a

lot of cell formulae. The SysCAD milling process model can be made available to

other process models in a ‘whole of factory’ model to determine the effect of the

other factory operations on the milling process and vice versa.

5.5 Dynamic milling train model


In the above sections, the static mode of the milling process model developed

in SysCAD is explored. The milling process is a continuous process and the dynamic

simulation of the process is required as variables affecting mill extraction

performance change as cane supply changes.

McWhinney(1973) proposed a dynamic version of the MILSIM model to

simulate the milling process. It used a time delay function to account for the time lost


between successive milling units and capacity lags function for the mixing of juice in

bagasse and imbibition during transport between successive mills. The dynamic

model was implemented on the University of Queensland GE225 computer since the

static model as reported by Russell (1968) was available on that computer.

5.5.2 Developing the dynamic milling train model in SysCAD

Work has commenced to develop a dynamic milling train model in SysCAD as

shown in Figure 5.10. A time delay function is added through a PGM controller to

account for the residence time of the bagasse and juice streams travelling from one

mill to another.

Two new controllers are added in the dynamic mode of SysCAD. The NC_001

is a noise controller used to simulate process noise. It can be used to add disturbances

to feeders, pipe outlet and other variables. PC_001 and PC_002 are profile

controllers used to vary an input parameter such as fibre and pol content of cane in

SysCAD over a period of time using a table of values. The profiler reads a CSV file

set up in excel spreadsheet.

Figure 5.9 Dynamic milling process model

5.5.3 Factors affecting mill extraction performance

Some of the prime variables affecting mill extraction performance are:

Cane composition (brix, fibre, pol)


Cane preparation

Mill torque

Dirt in cane supply

Mill speed/cane rate

Added water

Number of mills

Temperature of process

For a typical Australian sugar factory crushing up to 100 tonnes per hour of

cane, the first seven variables are changing continuously. The number of mills and

temperature of the process are more or less constant through the process but do affect

the performance.

The complete MILEX model developed in SysCAD process modelling

package is explored with real time factory cane analysis data in dynamic mode. The

composition of cane entering the shredder changes continuously and is an

independent variable outside the control of the factory. The changing cane

composition affects the milling process significantly. Changes in fibre% cane

instigate a change in added water rate % fibre, the work opening or mill speed is

changed by the mill torque set point in order to keep torque constant. The MILEX

model is solved with the cane analysis data and extraction of the overall milling train

is plotted against time in Figure 5.10. The extraction of the milling process is seen to

vary with changing cane composition.


Time elasped (min)

0 5 10 15 20

Extra

ctio

n (%

)

96.0

96.2

96.4

96.6

96.8

97.0

97.2

97.4

97.6

97.8

98.0

Brix extractionPol extraction

Figure 5.10 Extraction v/s time


The dynamic milling process model development is under development and

aims to address all significant variables affecting the extraction performance.

Further work is being undertaken to more accurately model extraction processes in a

milling train, to examine extraction issues dynamically and to integrate the model

into a whole factory model.


The main advantages of using the SysCAD software is the flexibility of

configuration, ease of use and the ability to integrate the milling process model with

other factory process models. Using the SysCAD software, the effect of other

process operations can be understood and practical solutions can be provided to

address process problems.

The ability of SysCAD in dynamic modelling presents the possibility of

assessing the response of the milling process to different cane characteristics. It is

expected that this work will enable the effectiveness of different control strategies to

be assessed. It is planned that the milling train model will be integrated with other


process models as part of a whole factory model to provide more accurate prediction

of the effect of changes in any part of the process on the outputs of the whole factory

process.

General Discussions and Conclusions 127

CHAPTER 6: GENERAL DISCUSSIONS AND CONCLUSIONS


This chapter summaries the research and highlights the main conclusions from

the research. The knowledge gained during the research, along with the significance

and benefits to the Australian raw sugar industry is described. Recommendations are

made for further work based on the foundation of this research.

6.2 Aim of the research

The research project aimed to develop an advanced, comprehensive and

dynamic model of the milling process of a raw sugar factory. The research aims

could be summarised into three parts:

To develop a milling train model that accurately evaluates mill performance

parameters and then use the model to predict extraction performance of the

milling train with a view to improving it.

To develop a model that tracks the flow of soluble and insoluble solids in the

milling process and their distribution in juice and bagasse, so that they are

available for the downstream models of the processes of a raw sugar factory. To

understand the significance of the soluble impurities and mud solids on extraction

performance.

To incorporate the developed models into the proprietary software package

“SysCAD” for advanced control options and improved operational efficiency.

6.3 Summary and conclusions of research

6.3.1 The enhanced and comprehensive mill extraction model

The development and validation of an enhanced and comprehensive mill

extraction model called MILEX was the primary focus of this research. The MILEX

model eliminates the constant fibre rate assumption and applies to a wider range of

parameters than previous models.

128 General Discussions and Conclusions

The MILEX model was developed by extending the MILSIM model to account

for fibre in juice streams, adopted the separation efficiency concept from Wienese,

modified its definition and developed a juice screen model as a part of the mill

extraction model. The model development is discussed in section 3.4. The MILEX

model was validated with some real factory measurements and the consistency of the

model was studied. Case studies were undertaken to study the effect of changing mill

configurations on extraction performance.

6.3.2 The cane component model

A cane component model was developed to track the flow of soluble and

insoluble cane components in the milling process. The model can track the flow of

sucrose, reducing sugars, soluble ash, proteins, true fibre and mud solids in the

milling process and can assess the effect of soluble impurities and mud solids on

extraction performance. A term was developed to track the flow of

reducing sugars in the milling process; a term was developed to track

the flow of soluble ash in the milling process; a term was developed to

track the flow of proteins in the milling process. The experimental work and the

development of the ratios are discussed in sections 4.3.2, 4.3.3, 4.3.4, 4.3.5.

A true fibre separation efficiency term was developed to track the flow of true

fibre in the milling process. A true fibre factor (ratio of true fibre separation

efficiency and separation efficiency) was developed to track the flow of mud solids

in the milling train and the juice screen. The true fibre factor is a single term used to

control the separation efficiencies of all the milling units and the juice screen to get

the required amount of mud solids in mixed juice stream.

6.3.3 SysCAD model

The incorporation of the developed models into a proprietary software package

“SysCAD” is the third part of the research. The application of the SysCAD software

as a process modelling tool is demonstrated and discussed in Chapter 5. The model

application example of the cush return discussed in section 5.4.6 shows the flexibility

and ease of use of SysCAD. Returning the cush to the #3 mill, the mill suffers

extensive substantial reduction in extraction performance.


The application of SysCAD in dynamic modelling is introduced. Although the

dynamic model is at a premature stage, further work is planned for the development

of the complete dynamic model.

6.4 Significance of the research


The benefits to be gained from the application of the MILEX model developed

in this project come from being able to evaluate mill performance parameters

accurately and predict extraction performance with changing mill configurations.

The model can provide all the major soluble impurities and mud solids to

downstream process models of a raw sugar factory.

6.4.2 Extraction benefits

Increasing the extraction performance of the existing milling equipment is of

interest to factory staff, as replacing the existing equipment is difficult to justify

because of high capital costs. To increase the extraction performance of the milling

train, understanding the process is vital. Process modelling is the best path of

representing the process and can provide insights into low extraction. The process

model developed should be accurate and cover the range of variables of the process.

The model has been used to identify extraction benefits. As discussed in

section 3.8.1, if the amount of fibre in expressed juice could reduce, such as by

reducing clearances around the mill (for example, between scrapers and rolls),

extraction could be improved. Similarly, it is conceivable that, if the juice screen

could be continuously cleaned or had a larger screen area, a greater amount of juice

would pass through to mixed juice and not be returned to the milling train. Increasing

the true fibre separation efficiency could lead to a decrease in the true fibre factor

and increase in extraction performance. In section 4.6 the resulting increase in

extraction from decreasing the true fibre factor is discussed. Average figures of 0.1 to

0.4 units of extraction could be achieved using these measures.

If a factory producing 150 000 t of sugar achieved a 0.2 unit increase in

extraction (a conservation estimate) through improvements in mill control, added

water rate or through decreasing the true fibre factor, and half of that extraction

improvement resulted in extra sugar production (again, a conservative estimate), an


extra 150 t of sugar would be produced. At $520 a tonne (Queensland Sugar Limited,

2012), a gain of $78000 would be made.

6.4.3 Inputs for downstream models

The processes following the milling train are the clarification process (mixed

juice) and the boiler processes (final bagasse). Many of the juice component values

required for the clarification station are not generally monitored in the milling

process. For example, the clarification station requires the input of mud solids in the

mixed juice stream so the process can be carried out as required.

The final bagasse stream is sent to the boiler station to generate steam for

generating power and process utilization. The mud solids in the final bagasse

decrease the gross calorific value (GCV) of the fuel. A more accurate GCV can be

computed if the mud solids content of the final bagasse is known. With the

developed model, these inputs can be provided to the clarification and boiler station

models. The soluble impurities generally cause problems in the crystallization (sugar

growth) process. Adequate measures can be taken when inputs can be predicted from

models such as this.

6.5 Recommendations for future research

While the research undertaken in this thesis is considered comprehensive, there

remain several tasks required to make the work more applicable.

While the new model (MILEX) was tested against factory measurements to

validate their ability to predict the extraction performance of the milling train, no

work has been undertaken to validate the newly developed performance parameters.

The parameters , are assumed constant through the

milling process in the model. There remains the need to develop empirical

relationships for the above components as a function of temperature, pH and mill

number to accurately model their behaviour. The true fibre separation efficiency term

is well defined in analytical mode but needs to be understood well before it can be

confidently applied in the predictive mode. Similarly for the true fibre factor which is

defined well in analytical mode, no work has been done to generate different true

fibre factors for individual milling units and use it in predictive mode.


As mentioned in section 2.2.5, empirical relationships have been developed for

reabsorption factor and imbibition coefficient, the empirical equations are not well

validated and cause the function to fail when operating conditions are outside the

range of the function. The modified MILSIM model discussed in section 2.3 could be

a better substitute for the imbibition coefficient provided the empirical equations for

crushing factor and mixing efficiency are well validated.

As mentioned earlier, although the model has proven to be superior over the

previous models, there remains the gap to address the above issues. Further work is

being planned to incorporate the modified MILSIM model in the advanced MILEX

model for better prediction of milling performance.

The dynamic mode of the milling train model is introduced in Chapter 5

Further work is being undertaken to understand the variables such as cane

preparation, added water rate and temperature of the process on extraction

performance of the milling process.


In conclusion, this study has contributed to knowledge through the following

discoveries:

The introduction of new definitions for mill performance parameters

(corrected filling ratio, corrected reabsorption factor, corrected imbibition

coefficient and separation efficiency).

The development of an enhanced mill extraction model to predict the

extraction performance of the milling train with a view to improving it.

The development of a soluble solids model to track the flow of sucrose,

reducing sugars, soluble ash and proteins through the milling process.

The development of an insoluble solids model to track the flow of true fibre

and mud solids in the milling process and juice screen.

The development of new performance parameters true fibre separation

efficiency and true fibre factor to monitor the flow of mud solids in the

milling process and the juice screen.


The development of new performance parameters invert ratio, ash/brix ratio

and proteins/brix ratio to track the flow of reducing sugars, soluble ash and

proteins in the milling process.

The incorporation of the models in the proprietary software package

“SysCAD” for advanced operational efficiency.

The development of a dynamic model of the milling process to incorporate

different control strategies to access impacts of changes and increase

extraction.

Bibliography 133

BIBLIOGRAPHY

Aspentech (2012). Aspen HYSYS process modelling system. Retrieved 01/02/2012, from http://www.aspentech.com/hysys

Australian Sugar Milling Council. Retrieved 30/08/2010, from

www.asmc.com.au/content/industry.asp Barnes, A. C. (1974). The Sugarcane: Leonard Hill Books. Blaik, G. W., & Edwards, B. P. (1994). Improving the prediction of milling train

extraction, Technical Report 3/94: Sugar Research Institute, Mackay, Australia.

Bureau of Sugar Experiment Stations (1984). The standard laboratory manual for

Australian sugar mills. (Vol. 1 Principles and practises).

Bureau of Sugar Experiment Stations (2001). The standard laboratory manual for

Australian sugar mills. (Vol. 2 Analytical methods & tables.). Crawford, W. R. (1957). Improving extraction at a No.1 mill. Proceedings of the

Queensland Society of Sugarcane Technologists: 24, pp. (89-103).

Edwards, B. P. (1995). Extraction performance of milling trains- imbibition

processes and calculations. Proceedings of the Australian Society of Sugar Cane Technologists: 17, pp. (346-351).

Edwards, B. P., Bartholomew, H. A., & Miller, K. F. (1996). Analysis of final

bagasse for dry substance, Technical Research Report 4/96: Sugar Research Institute, Mackay, Australia.

Egeter, H. (1928). Studies in bagasse and mill operations. Archief 2, Vol.2, 691.

Fernandes, A. C. (2003). Calculos na Agroindustria da Cana-de-acucar. (2 ed. Vol.

1): Sociedade dos Technicos Acucareiros e Alcooleiros do Brasil (STAB). Foster, D. H. (1956). Analytical methods used in asessing mill tandem perfromance.

Proceedings of IX International Society of Sugar Cane Technologists Congress.

134 Bibliography

Gilfillan, W. N., Sopade, P. A., & Doherty, W. (2012). Moisture uptake and tensile properties of starch-sugar cane fibre films. Proceedings of the Australian Society of Sugar Cane Technologists: 34, pp. (electronic-format).

Gnanasambandam, A., & Birch, G. R. (2006). Isolation of protoplasts and vacuoles

from sugarcane suspension and stem parenchyma cells. Proceedings of the Australian Society of Sugar Cane Technologists: 28, pp. (electronic-format).

Kent, G. A. (1997). Modelling the extraction process of milling trains. Proceedings

of the Australian Society of Sugar Cane Technologists: 19, pp. (315-321). Kent, G. A. (2001). A model to estimate milling unit throughput. Proceedings of the

Australian Society of Sugar Cane Technologist: 23, pp. (457-460).

Kent, G. A. (2003). Increasing the capacity of Australian raw sugar factory milling

units. PhD Thesis, James Cook University Kent, G. A. (2010a). The effect of added water temperature on milling train

operation & performance. Proceedings of the XXVII International Society of Sugar Cane Technologists: 27, pp. (CD-ROM).

Kent, G. A. (2010b). Estimating bagasse production. Proceedings of the Australian

Society of Sugar Cane Technologists: 32, pp. (546-558).

Kent, G. A., Doolan, C. J., Corica, F. A., & McKenzie, N. J. (1998). Milling unit

extraction performance, Technical Report 1/98: Sugar Research Institute, Mackay, Australia.

Kent, G. A., McKenzie, N. J., & Downing, C. M. (2000). Predicting milling train

extraction performance, Technical Report 4/00: Sugar Research Institute, Mackay, Australia.

Kent, G. A., McKenzie, N. J., & Plaza, F. (2008). An alternative method for

measuring milling train performance, Syndicated Report 2/08: Sugar Research Institute, Mackay, Australia.

Kulkarni, D. P. (2009). Cane sugar manufacturing in India (Vol. 1): The Sugar

Technologist Association of India (STAI). Lionett, G. R. E. (1981). The effect of the level of extraction on mixed juice purity.

Proceedings of South African Sugar Technologists' Association: 55, pp. (28-30).

Bibliography 135

Lloyd, T., Eastment, S., & Mitchell, P. (2010). Milling train maceration control utilising NIR technology. Proceedings of the Australian Society of Sugar Cane Technologists.: 32, pp. (electronic-format).

Loubser, R. C. (2004). Heat and mass balance using constraint equations, a

spreadsheet, and the Newton-Raphson technique. Proceedings of South African Sugar Technologists' Association: 78, pp. (457-472).

Macritchie, F. (1973). Effects of temperature on dissolution and precipitation of

proteins and polyamino acids. Journal of Colloid and Interface Science, 45(2), 235-241.

Martin, L. F. (1958). The complex organic nonsugars of high molecular weight.

(Vol. 1). Amsterdam: Elsevier. McWhinney, W. (1973). Simulation of dynamic performnace of sugar mill crushing

trains. Ph.D Thesis, University of Queensland

Muller, R. L., Player, M. R., & Wise, M. B. (1982). An examination of input

disposition and effect of dirt in Queensland sugar mills. Proceedings of the Australian Society of Sugar Cane Technologists: 4, pp. 1-9.

Munro, B. M. (1963). Furhter imbibition experiments. Proceedings of the

Queensland Society of Sugar Cane Technologists: 30, pp. (105-115). Munro, B. M. (1964). An investigation into crushing of bagasse and the influence of

imbibition on extraction. Ph.D Thesis, University of Queensland

Murry, C. R. (1959). Preliminary bagasse tests in the two roll experimental mill.

Proceedings of the Australian Society of Sugar Cane Technologists: 26, pp. (67-72).

Murry, C. R., & Russell, G. E. (1969). Prediction of the extraction performance of

crushing trains, Technical Report 104: Sugar Reearch Institute, Mackay, Australia.

Neill, S. W., McKinnon, S. A., & Garson, C. A. (1996). A library of cane transport

and sugar factory images, Internal Report 1/96: Sugar Research Institute, Mackay Australia.

Peacock, S. D. (2002). The use of SIMULINK for process modelling in the sugar

industry. Proceedings of South African Sugar Technologists' Association.: 76, pp. (444-455).

136 Bibliography

Pidduck, J. (1955). Physical properties of bagasse. Proceedings of Queensland Society of Sugar Cane Technologists, 22, 147-155.

Queensland Sugar Limited (2012). Sugar prices. Retrieved 20/07/2012, from

http://www.qsl.com.au/

Rein, P. (2007). Cane sugar engineering. Berlin: Bartens.

Russell, G. E. (1968). An investigation of the extraction performance of sugarcane

crushing trains. Ph.D Thesis, University of Queensland Steindl, R. J. (1998). Dir-Its implications for the clarifier and filter stations.

Proceedings of the the Australian Society of Sugar Cane Technologists: 20, pp. (484-490).

SUGARS International (2011). Process modelling and simulation software.

Retrieved 30/11/2011, from http://www.sugarsonline.com

SysCAD (2012). Innovative software for plant simulation. Retrieved 06/06/2012,

from http://www.syscad.net/ Thai, C. C. D., & Doherty, W. O. S. (2011). The composition of sugarcane juices

derived from burnt cane and whole green cane crop. Proceedings of the Australian Society of Sugar Cane Technologists: 33, pp. (electronic-format).

Thaval, O. P., & Kent, G. A. (2012). Modelling the flow of juice through a mill.

International Sugar Journal, 1363(114), 36-40.

Van Der Pol, C., & Alexander, J. B. (1955). Decomposition of sucrose in the milling

process. Proceedings of the South African Sugar Technologists' Association: 29, pp. (46-53).

Vukov, K. (1965). Kinetic aspects of sucrose hydrolysis. International Sugar

Journal, 67, 172-175. Walford, S. M. (1996). Composition of cane juice. Proceedings of South African

Sugar Technologists' Association: 70, pp. (265-266).

Wienese, A. (1990). Mill settings and extraction. Proceedings of the South African

Sugar Technologists' Association 64, pp. (154-157). Wienese, A. (1994). Imbibition optimisation at Mount Edgecombe. Proceedings of

South African Sugar Technologists' Association: 68, pp. (137-142).

Bibliography 137

Wienese, A. (1995). The effect of imbibition and cane quality on the front end mass balance. Proceedings of South African Sugar Technologists' Association: 69, pp. (181-185).

Wienese, A., & Reid, M. J. (1997). Soil in cane: Its measurement, its effect on

milling, and method of removal. Proceedings of South African Sugar Technologists' Association: 71, pp. (130-134).

Wiggins, L. F. (1958). The nitrogen-containing non-sugars (the amino acids and

proteins). (Vol. 1). Amsterdam: Elsevier. Wright, P. G. (2003). The effect of dirt on bagasse quantity and heating value.

Proceedings of the Australian Society of Sugar Cane Technologists: 25 pp. (CD-ROM).

Wright, P. G., Fernandes, A. C., & Zarpelon, F. (2007). Control calculations for

factories producing both sugar and alcohol. Proceedings of the Australian Society of Sugar Cane Technologists: 29, pp. (CD-ROM).

Appendices 139

APPENDICES

Appendix A-Glossary of terms

Ash The residue remaining after incineration at the specified temperature or temperature of a sample which was pre-treated with sulphuric acid.

Bagasse The residue after extraction of juice from cane in one or more mills. Hence the terms, first mill bagasse, second mill bagasse etc. and in the case of the last mill final bagasse or simply bagasse, are used.

Brix The brix of a solution is the concentrations (in g solute per 100 g solution) of pure sucrose in water, having the same density as the solution at the same temperature. If refractive index be adopted as an alternative basis of comparison the value derived should be termed refractometer brix.

Cane The raw material delivered to the mill, including clean cane stalk, trash, tops, and any other extraneous matter.

Dry substance The weight of material remaining after drying the product examined under specified conditions, expressed as a percentage of the original weight. The determination of dry substance represents an attempt to measure the total solids, both soluble and insoluble, or, in the absence of insoluble solids, the total soluble solids. The degree of accuracy achieved depends upon the constitution of the sample and the drying technique.

Escribed volume The volume escribed by a pair of mill rollers in a given time. Escribed volume is equal to the product of roller length, work opening and top roller surface speed.

Extraction The percentage of brix or pol extracted from the incoming material by a train of mills either individually of cumulatively.

Fibre Technically, fibre is the dry, water insoluble matter in the cane.

First expressed juice

The juice expressed by the feeding devices and the first two rollers of the first three-roller mill of a milling train.

Hygroscopic water

The brix-free water adsorbed by cane fibre, the amount of which varies with the condition of the solution with which the fibre is in contact. For sugar solutions of low brix and at normal temperature, such as those experienced in bagasse analysis, hygroscopic water is assumed to be 25% on fibre.

Imbibition The process whereby water or juice is added to bagasse to dilute the juice contained therein.

140 Appendices

Impurities (soluble)

A collective term for all substances other than sucrose present in the total soluble solids contained in a sample.

퐼푚푝푢푟푖푡푖푒푠 = 100− 푝푢푟푖푡푦

Invert sugar The equimolecular mixture of glucose and fructose which results from the hydrolysis or inversion of sucrose.

Mixed juice The mixture of juices leaving the milling train or a cane diffuser for further processing.

Mud solids Insoluble matter other than bagacillo in clarifier mud, filter cake and associated materials.

No-void volume The volume of cane (or bagasse) calculated on the basis that it consists of juice and fibre only i.e. that all air and/or other gas has been removed.

Pol The pol of a solution is the concentration (in g solute per 100 g solution) of a solution of pure sucrose in water having the same optical rotation as the sample at the same temperature. For solutions containing only pure sucrose in water, pol is a measure of the concentrations of sucrose present; for solutions containing sucrose and other optically active substances, pol is the algebraic sum of the rotations of the constituents present.

Purity Purity is the percentage of sucrose in the total solids in a sample.

퐴푝푝푎푟푒푛푡푝푢푟푖푡푦 =푃표푙퐵푟푖푥 × 100

Residual juice The juice left in bagasse after milling.

Roller surface speed

The speed on the roller surface in m/s computed from the roller shaft speed and the roller mean diameter.

Suspended solids Insoluble solids in juice or other liquid, removable by mechanical means.

Work opening The mean opening between a pair of mill rollers. This opening takes into account the set opening and allowance for mill grooving. No allowance is made for juice grooves, but where a dirty top roller is employed this must be taken into account.

Modelling the Flow of Cane Constituents through the Milling Process ...

Documents

Transcript of Modelling the Flow of Cane Constituents through the Milling Process ...