Universidade Porto: Decision Support System in The Design, Production and Quality Control of Glass Containers

UNIVERSIDADE DO PORTO
Decision Support System in the Design, Production

and Quality Control of Glass Containers
Programa Doutoral em Engenharia Mecânica (PRODEM) Bruno Miguel Rodrigues Martins

Departamento de Engenharia Mecânica (DEMec)
Orientador: Prof. Dr. José Cesár de Sá
Faculdade de Engenharia (FEUP)
Universidade do Porto (UP) Co-orientador: Prof. Dr. Ana Reis
PORTUGAL July 2017
ii
“They are wild, they are free, they are not different from us…”
Jeremy Wade
À minha família e amigos.
iii
Acknowledgements
Acknowledgements
O trabalho apresentado apenas foi possível fruto da colaboração e apoio de algumas pessoas, pelas quais
não posso deixar de prestar o meu reconhecimento e agradecimento.
Ao Professor Doutor José Manuel de Almeida César de Sá um profundo agradecimento por ter-me
acolhido como seu orientando, pelo apoio prestado durante o decorrer deste trabalho. O conteúdo das
discussões cientificas foram sempre muito produtivas e esclarecedoras. Agradeço ainda pela paciência,
disponibilidade e amizade demonstrada.
À Professora Doutora Ana Reis um agradecimento pelo apoio, empenho e aconselhamento dado durante
o desenvolvimento do trabalho. Ainda um agradecimento pelo acolhimento e confiança em acreditar na
finalização do projecto com sucesso.
Ao Doutor Pedro Teixeira pelo suporte dado durante o projecto. Sem a sua colaboração grande parte deste
trabalho não teria sido possível. Adicionalmente, um enorme agradecimento pela incessante partilha de
conhecimentos, paciência e amizade demonstrados.
A Doutora Margarida Machado pelo apoio dado, as discussões científicas foram sem dúvida
enriquecedoras e ainda pela ajuda em determinar as linhas de rumo para levar este trabalho a bom porto.
Ao Doutor Jaime Rodrigues pelo precioso auxílio dado em áreas estratégicas que sem a sua ajuda seriam
muito mais complicadas em realizar. Um enorme gosto em trabalhar com uma pessoa com um enorme
conhecimento científico.
À Empackglass por todo o apoio técnico para a execução do trabalho e pelo fornecimento de resultados
experimentais. Quero agradecer também pelas orientações e sugestões práticas.
Ao Professor Doutor Abel Dias dos Santos um especial agradecimento pelo acolhimento inicial, por ter
concedido a oportunidade de trabalhar num ambiente científico e por ter despertado em mim o gosto pela
investigação científica. Por toda a paciência demostrada e também por todo o conhecimento científico
ensinado. Não tenho dúvidas que o seu apoio prévio foi sem dúvida essencial para concretizar este
trabalho até ao fim.
Ao INEGI - instituto de ciência e inovação em engenharia mecânica e engenharia industrial desejo

agradecer à unidade CETECOP/UTAF - Unidade das Tecnologias de Conformação Plástica/ Unidade de
Tecnologias Avançadas de Fabrico. Pelo todo apoio técnico e humano dado durante e execução do
trabalho.
Ainda a toda a minha família e amigos.
Finalmente, a todos os que diretamente ou indiretamente contribuíram para que este trabalho chegasse a
bom porto.
iv
Abstrat
Abstrat
In this work a numerical model, based on the finite element method, for all processes of glass forming of
glass containers is presented. Glass forming processes involve coupled thermomechanical phenomena
which heat transfer and viscous flow are dependent, as glass viscosity is highly dependent on temperature
and the significant shape changes and contact conditions affect the temperature distribution. During the
overall process glass changes from a molten state to a solid state. Therefore, adequate cooling conditions
must be set appropriately. The ultimate aim is to set the better process parameters so that the final products
will have the required geometrical shape and thickness distribution.
From the numerical point of view geometrical modelling must be robust so as to adjust to the different
sequenced stages. Remeshing techniques requiring adequate transfer the data correctly. Different thermal
and mechanical contact conditions between glass and moulds must be taken into account adequately. Also
effective treatment of the incompressible conditions associated with glass flow must be dealt
appropriately. The numerical model developed, addressing all these issues is validated with real industrial
products. Numerical aspects associated with the model implementation and its validation with real formed
products are discussed.
Keywords: Glass container forming, Finite element method, Thermomechanical model
v
Resumo
Resumo
Neste trabalho é apresentado um modelo numérico baseado no método dos elementos finitos para todas
as etapas da produção de garrafas de vidro. O modelo trata o processo de prensado/soprado e
soprado/soprado. O processo de produção de garrafas de vidro envolve um modelo termomecânico tendo
a viscosidade do material grande influência no fenómeno térmico e mecânico. Uma vez que a viscosidade
é altamente dependente da temperatura influenciando significamente a forma e as condições de contacto
afetando desse modo a distribuição de temperatura. Durante o processo tecnológico o vidro muda o seu
estado de líquido para sólido num período de tempo curto. Assim, as condições de arrefecimento entre o
vidro e os moldes têm de ser conhecidas com rigor. O objetivo deste trabalho é simular o processo de
produção de garrafas de vidro de modo que os parâmetros do processo possam ser alterados de forma a
obter a forma da garrafa e a distribuição de espessura.
O modelo deve ser robusto do ponto de vista numérico para que durante todas as etapas desde a queda da
gota, a formação da primeira forma até à forma final durante o prensado ou o primeiro soprado. A
transferência da informação entre malhas deve ser realizada adequadamente. As condições de
incompressibilidade associadas ao vidro devem ser também alvo de atenção. O modelo desenvolvido
tendo em conta todos estes aspetos foi validado com resultados reais resultantes da produção de garrafas
de vidro. Os aspetos numéricos associados ao modelo implementado e a sua validação são discutidos
neste trabalho.
Keywords: Produção de garrafas de vidro, Método dos elementos finitos, Modelo termomecânico
vi
Index
Index
ACKNOWLEDGEMENTS ............................................................................................................. IV
ABSTRAT ......................................................................................................................................... V
RESUMO ......................................................................................................................................... VI
INDEX ............................................................................................................................................. VII
LIST OF FIGURES .......................................................................................................................... X
LIST OF TABLES ......................................................................................................................... XIX
NOMENCLATURE ........................................................................................................................ XX
C H A P T E R 1 .................................................................................................................................1
1 INTRODUCTION ..................................................................................................................... 1
1.1 HISTORIC REVIEW OF GLASS MAKING........................................................................................1

1.2 HISTORY OF GLASS MOULDING ..................................................................................................2
1.3 GLASS CONTAINERS INDUSTRY ..................................................................................................3
1.4 GLASS CONTAINER QUALITY......................................................................................................4
1.5 GLASS CONTAINERS TERMINOLOGY .............................................ERRO! MARCADOR NÃO DEFINIDO.
1.6 GLASS CONTAINER SIMULATION ................................................................................................5
1.7 OBJECTIVES ................................................................................................................................7
1.8 THESIS SYNOPSIS ........................................................................................................................7
2 GLASS MATERIAL PROPERTIES ........................................................................................ 9
2.1 BRITTLENESS ..............................................................................................................................9

2.2 DENSITY .....................................................................................................................................9
2.3 THERMAL EXPANSION ..............................................................................................................10
2.3.1 Glass Cord .......................................................................................................................10
2.4 TEMPERATURE DEPENDENCE OF VISCOSITY ..............................................................................11
2.5 CHEMICAL DURABILITY ...........................................................................................................14
2.5.1 Dealkalization ..................................................................................................................15
2.6 LIQUIDUS TEMPERATURE .........................................................................................................15
2.6.1 Devitrification ..................................................................................................................15
2.7 LIGHT TRANSMISSION – COLOUR .............................................................................................15
2.8 CULLET.....................................................................................................................................16
2.9 STRENGTH OF THE GLASS .........................................................................................................17
2.10 THERMAL SHOCKS ....................................................................................................................17
2.11 GLASS COMPOSITION ................................................................................................................18
3 GLASS CONTAINER PRODUCTION .................................................................................. 21
vii
Index
3.1 BATCH ......................................................................................................................................22

3.1.1 Batch – Storing ................................................................................................................22
3.1.2 Batch – Weighting ............................................................................................................23
3.1.3 Batch – Mixing .................................................................................................................23
3.2 MELTING ..................................................................................................................................24
3.2.1 Regenerative Furnace ......................................................................................................25
3.2.2 Recuperative Furnace ......................................................................................................28
3.2.3 Boosting ...........................................................................................................................30
3.2.4 Bubbler.............................................................................................................................32
3.2.5 Barrier Wall .....................................................................................................................33
3.2.6 Refining Bank ...................................................................................................................33
3.2.7 Drain ................................................................................................................................34
3.2.8 Cullet Preheating .............................................................................................................34
3.2.9 Melting Process and glass formation ...............................................................................35
3.3 GLASS CONDITIONING ..............................................................................................................36
3.3.1 Forehearth .......................................................................................................................37
3.4 FORMING ..................................................................................................................................42
3.4.1 Feeder ..............................................................................................................................42
3.4.2 Blow and Blow .................................................................................................................52
3.4.3 Press and Blow ................................................................................................................54
3.5 HOT END COATING ...................................................................................................................56
3.6 ANNEALING ..............................................................................................................................58
3.6.1 Internal Stress ..................................................................................................................59
3.6.2 Cycle ................................................................................................................................60
3.6.3 Quality control .................................................................................................................61
3.7 COLD END COATING .................................................................................................................62
3.8 INSPECTION AND QUALITY CONTROL ........................................................................................63
3.8.1 Defects .............................................................................................................................68
3.8.2 Quality Control Laboratory .............................................................................................78
3.9 PALLETIZING ............................................................................................................................79
3.10 SHRINK-WRAPPING ...................................................................................................................81
4 MATHEMATICAL MODELLING OF GLASS FORMING PROCESSES ......................... 83
4.1 FINITE ELEMENT FORMULATION FOR THERMOMECHANICAL BEHAVIOUR................................83

4.1.1 Constitutive Model ...........................................................................................................83
4.1.2 Finite Element Method Formulation ................................................................................85
4.2 FINITE ELEMENT FORMULATION FOR HEAT TRANSFER ............................................................88
4.2.1 Transient Analysis ............................................................................................................88
4.2.2 Time Integration...............................................................................................................91
5 SIMULATING MODEL FOR GLASS CONTAINERS ......................................................... 95
5.1 ELEMENT TECHNOLOGY ...........................................................................................................96

5.1.1 Library of Elements..........................................................................................................96
5.1.2 Spurious Oscillations Due to Smaller Time Increments ..................................................96
5.1.3 Hourglass and Locking ....................................................................................................96
5.2 CONTACT ..................................................................................................................................97
5.2.1 Contact Algorithm ............................................................................................................98
5.2.2 Sticking Condition ..........................................................................................................100
5.2.3 Sliding Conditions ..........................................................................................................102
5.2.4 Updating Boundary Conditions .....................................................................................102
5.3 MESH UPDATING AND REMESHING .........................................................................................106
5.3.1 Mesh Updating ...............................................................................................................107
viii
Index
5.3.2 Remeshing ......................................................................................................................108
6 ELEMENTS VALIDATION AND TESTING .......................................................................115
6.1 HEAT TRANSFER VALIDATION ................................................................................................115

6.2 MECHANICAL VALIDATION ....................................................................................................129
7 COMPUTER PROGRAM .....................................................................................................156

7.1 SOFTWARE ENVIRONMENT ......................................................................................................156
7.1.1 Graphic Engine ..............................................................................................................156
7.1.2 Preprocessor ..................................................................................................................156
7.1.3 Postprocessor .................................................................................................................158
7.2 SOFTWARE CORE ....................................................................................................................160
7.3 COMPUTER PROGRAM FILES ...................................................................................................161
7.3.1 Input File........................................................................................................................162
7.3.2 Result File ......................................................................................................................163
7.3.3 Warning File ..................................................................................................................164
7.3.4 Error File .......................................................................................................................164
7.3.5 Restart File ....................................................................................................................164
7.3.6 Remeshing Files .............................................................................................................164
7.3.7 Update File ....................................................................................................................164
7.3.8 List File ..........................................................................................................................164
8 GLASS CONTAINERS SIMULATION CASES ...................................................................166
8.1 BLOW AND BLOW PROCESS ....................................................................................................173

8.1.1 Gob Loading ..................................................................................................................173
8.1.2 Settle Blow .....................................................................................................................182
8.1.3 Corkage Reheat ..............................................................................................................189
8.1.4 Counter Blow .................................................................................................................190
8.1.5 Reheat ............................................................................................................................205
8.1.6 Invert ..............................................................................................................................206
8.1.7 Rundown ........................................................................................................................209
8.1.8 Final Blow......................................................................................................................216
8.1.9 Mould Open ...................................................................................................................223
8.1.10 Bottle Thickness .............................................................................................................225
8.2 PRESS AND BLOW PROCESS ....................................................................................................230
8.2.1 Gob Loading ..................................................................................................................230
8.2.2 Baffle On ........................................................................................................................240
8.2.3 Plunger Up .....................................................................................................................241
8.2.4 Reheat ............................................................................................................................262
8.2.5 Invert ..............................................................................................................................263
8.2.6 Rundown ........................................................................................................................267
8.2.7 Final Blow......................................................................................................................278
8.2.8 Mould Open ...................................................................................................................288
8.2.9 Bottle Thickness .............................................................................................................293
8.3 EVALUATION OF SOME NUMERICAL ASPECTS ..........................................................................299
8.3.1 Volume Evolution ...........................................................................................................299
8.3.2 Evolution of the Number of Nodes .................................................................................303
8.3.3 Plunger Displacement ....................................................................................................306
8.3.4 Time Evolution ...............................................................................................................307
9 CONCLUSIONS AND FUTURE WORK..............................................................................312
10 REFERENCES ...................................................................................................................315
ix
List of Figures
List of Figures
Figure 2.1: a) Examples of a cord defects after cooling. b) Glass cord observed by viewing in polarized
light . ........................................................................................................................................................11
Figure 2.2: Viscosity evolution in working zone temperature [12]. ........................................................12
Figure 2.3: Chemical durability of a container. ......................................................................................14
Figure 2.4: Glass colours presented in some container. .........................................................................16
Figure 2.5: Effects of some oxides in glass properties. ...........................................................................19
Figure 3.1: Production line of glass containers. .....................................................................................22
Figure 3.2: a) Glass plant batch house....................................................................................................23
Figure 3.3: Weighting the raw material. .................................................................................................23
Figure 3.4: Mixing the raw material operation. ......................................................................................24
Figure 3.5: Melting plant scheme with the temperature gradient. ..........................................................25
Figure 3.6: Regenerative furnace layout. ................................................................................................26
Figure 3.7: Regenerative end fired furnace [18]. ....................................................................................27
Figure 3.8: Regenerative furnace heat exchange [18]. ...........................................................................28
Figure 3.9: Recuperative furnace [18]. ...................................................................................................29
Figure 3.10: Side fired recuperative furnaces [18]. ................................................................................30
Figure 3.11: Comparison of energy consumption of a regenerative and a recuperative furnace at part
load [18]. .................................................................................................................................................30
Figure 3.12: Boosting system in an end fired furnace. ............................................................................31
Figure 3.13: a) Melting booster b) Barrier booster c) Local booster [18]. ............................................31
Figure 3.14: Bubbling principle [18]. .....................................................................................................33
Figure 3.15: Barrier wall. .......................................................................................................................33
Figure 3.16: Drain cross section [18]. ....................................................................................................34
Figure 3.17: Furnace temperature evolution and the respective convection. .........................................35
Figure 3.18: Flux of glass inside a furnace. ............................................................................................36
Figure 3.19: Woking end furnace. ...........................................................................................................36
Figure 3.20: Glass conditioning general view. ........................................................................................37
Figure 3.21: Glass Conditioning heat transfer. .......................................................................................38
Figure 3.22: Forehearth Type K. .............................................................................................................38
Figure 3.23: Forehearth Type K. .............................................................................................................39
Figure 3.24: Forehearth longitudinally cooled. ......................................................................................39
Figure 3.25: Radiation cooled forehearths. .............................................................................................40
Figure 3.26: Muffle cooled forehearths. ..................................................................................................40
Figure 3.27: System 500 forehearths. ......................................................................................................41
Figure 3.28: System 500 forehearths a) cooling cycle b) Heating cycle c) Normal operation. ...............41
Figure 3.29: System 500 forehearths cooling zones. ...............................................................................42
Figure 3.30: System 500 forehearths cooling zones. ...............................................................................42
Figure 3.31: Forming feed machine. .......................................................................................................43
Figure 3.32: a) Tube height raised. .........................................................................................................44
Figure 3.33: Examples of a) Spout bowl b) Feeder tubes c) Orifice ring. ...............................................44
x
Figure 3.34: a) Small Orifice Ring. .........................................................................................................45
Figure 3.35: Glass feeder plungers. ........................................................................................................45
Figure 3.36: Representation of plunger motion and gob shear cutting [1]. ............................................46
Figure 3.37: Effect on gob with different plungers [20]. .........................................................................46
Figure 3.38: a) Plunger height lowered. .................................................................................................47
Figure 3.39: Gob cutting shears. .............................................................................................................47
Figure 3.40: Shear tension effect on gob shearing [20]. .........................................................................48
Figure 3.41: Shear overlap effect on gob shearing [20]. ........................................................................48
Figure 3.42: Plunger/shear differential adjustments a) Differential 0º setting b) Differential 10º setting
c) Differential 30º setting. [20]. ...............................................................................................................49
Figure 3.43: Different types of tuning a feeder a) Glass temperature increased. b) Glass temperature
decreased. c) Machine speed increased. d) Machine speed decreased. [20] ..........................................49
Figure 3.44: Gob shape parts affected by feeder controls [20]. ..............................................................50
Figure 3.45: Gob distributor scheme. ......................................................................................................51
Figure 3.46: Gob distributor. ..................................................................................................................51
Figure 3.47: a) Gob loading b) Settle blow c) Counter blow d) Invert e) Final blow f) Take out ...........53
Figure 3.48: a) Gob loading b) Baffle on c) Plunger up d) Invert e) Rundown f) Final Blow g) Take out
.................................................................................................................................................................56
Figure 3.49: Hot end coating...................................................................................................................57
Figure 3.50: Application of hot end coating. ...........................................................................................57
Figure 3.51: Application of hot end coating. ...........................................................................................58
Figure 3.52: Rapid cooling from forming to annealing lehr. ..................................................................59
Figure 3.53: Stress after coating process. ...............................................................................................59
Figure 3.54: Tension inside the container thickness. ..............................................................................60
Figure 3.55: Tension inside the container thickness. ..............................................................................61
Figure 3.56: Measurement of glass stress with a polariscope. ................................................................62
Figure 3.57: Different surface treatments though a convey belt. ............................................................62
Figure 3.58: Application of cold end coating. .........................................................................................63
Figure 3.59: The packaging materials include pallets, cartons, trays, and shrink wrap and stretch wrap
materials. .................................................................................................................................................64
Figure 3.60: Furnace control. .................................................................................................................65
Figure 3.61: Hot end coating quality control. .........................................................................................66
Figure 3.62: Cond end coating quality control. ......................................................................................67
Figure 3.63: Defects can be presented in a glass container [Emhart ]. ..................................................70
Figure 3.64: Vertical palletizing..............................................................................................................81
Figure 3.65: Shrink-wrapping a pallet. ...................................................................................................81
Figure 5.1: Model domain contact problem. ...........................................................................................97
Figure 5.2: Target for glass node contact the mould. ..............................................................................98
Figure 5.3: Contact algorithm flowchart for each node. .........................................................................99
Figure 5.4: Contact representation for sticking condition. ...................................................................100
Figure 5.5: Contact flowchart for candidate segment selection. ...........................................................101
Figure 5.6: Contact velocity components a) Nodes in contact to the mould with the velocity components.
...............................................................................................................................................................102
Figure 5.7: Local coordinates at each contact node between the plunger and the glass. .....................103
Figure 5.8: Contact between two bodies at different temperatures considering the different boundaries
for each physical field. ...........................................................................................................................105
Figure 5.9: Contact between two bodies at different temperatures considering the different boundaries
for each physical field. ...........................................................................................................................106
Figure 5.10: a) Example of a triangular linear mesh. ...........................................................................106
Figure 5.11: Example of an element with negative Jacobian. ...............................................................107
Figure 5.12: Meshing a curve with different mesh size. ........................................................................108
Figure 5.13: Flowchart of remeshing technique....................................................................................109
xi
Figure 5.14: Correction of the new boundary nodes. ............................................................................110
Figure 5.15: Example of a boundary segments. ....................................................................................110
Figure 5.16: Cross product between a node that belong to a new mesh and the older element. ...........112
Figure 6.1: Mesh used to simulation for heat transfer validation a) linear triangle b) quadratic triangle
c) linear quadrilateral d) quadratic quadrilateral.................................................................................116
Figure 6.2: Scheme to validate the prescribed temperature at the boundary of a cylinder. ..................116
Figure 6.3: Results for the analytical equation for the prescribed boundary temperature a)
Temperature/Radius b) Temperature/Time evolution. ...........................................................................117
Figure 6.4: Results for Abaqus simulation for the prescribed boundary temperature a) Linear
triangular elements b) Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic
quadrilateral elements. ..........................................................................................................................118
Figure 6.5: Results for Empaktor software for the prescribed boundary a) Linear triangular elements b)
Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic quadrilateral elements.
...............................................................................................................................................................119
Figure 6.6: Results for Empaktor software for the prescribed boundary temperature using only the
diagonal values for the capacity matrix a) Linear triangular elements b) Linear quadrilateral elements
c) Quadratic triangular elements d) Quadratic quadrilateral elements. ...............................................120
Figure 6.7: Results for Empaktor software for the prescribed boundary temperature using a pondered
capacity matrix a) Linear triangular elements b) Linear quadrilateral elements c) Quadratic triangular
elements d) Quadratic quadrilateral elements. ......................................................................................121
Figure 6.8: Scheme to validate the prescribed flux at the boundary cylinder. ......................................121
Figure 6.9: Results for the analytical equation for the prescribed boundary flux a) Temperature/Time b)
Temperature/Radius evolution. ..............................................................................................................122
Figure 6.10: Results for Abaqus simulation for the prescribed flux a) Linear triangular elements b)
...............................................................................................................................................................123
Figure 6.11: Results for Empaktor software for the prescribed Flux a) Linear triangular elements b)
...............................................................................................................................................................124
Figure 6.12: Results for Empaktor software for the prescribed flux using only the diagonal values for
the capacity matrix a) Linear triangular elements b) Linear quadrilateral elements c) Quadratic
triangular elements d) Quadratic quadrilateral elements. ....................................................................125
Figure 6.13: Results for Empaktor software for the prescribed flux using a pondered capacity matrix a)
Linear triangular elements b) Linear quadrilateral elements c) Quadratic triangular elements d)
Quadratic triangular elements. ..............................................................................................................126
Figure 6.14: Mesh used for heat transfer validation for quadratic triangular and quadratic
quadrilateral. .........................................................................................................................................127
Figure 6.15: Results for prescribed temperatures a) Quadratic triangular elements in Abaqus b)
Quadratic triangular elements Empaktor software c) Quadratic quadrilateral elements in Abaqus d)
Quadratic quadrilateral elements Empaktor software. .........................................................................128
Figure 6.16: Results for prescribed flux temperatures a) Quadratic triangular elements in Abaqus b)
Quadratic triangular elements Empaktor software c) Quadratic quadrilateral elements in Abaqus d)
Quadratic quadrilateral elements Empaktor software. .........................................................................129
Figure 6.17: Axisymmetric deformation modes for the linear triangular element using 1 Gauss point.
...............................................................................................................................................................131
Figure 6.18: Orthogonalizing deformation modes for the linear triangular element using 1 Gauss point.
...............................................................................................................................................................131
...............................................................................................................................................................132
...............................................................................................................................................................132
...............................................................................................................................................................132
xii
Figure 6.22: Axisymmetric deformation modes for the MINI linear triangular element using one Gauss
point. ......................................................................................................................................................133
Figure 6.23: Orthogonalizing deformation modes for the MINI linear triangular element using 1 Gauss
point. ......................................................................................................................................................133
Figure 6.24: Axisymmetric deformation modes for the MINI linear triangular element using one Gauss
point. ......................................................................................................................................................134
Figure 6.25: Axisymmetric deformation modes for the MINI linear triangular element using three
Gauss point. ...........................................................................................................................................134
Figure 6.26: Axisymmetric deformation modes for the MINI linear triangular element using three
Gauss point. ...........................................................................................................................................135
Figure 6.27: Axisymmetric deformation modes for the MINI linear triangular element using seven
Gauss point. ...........................................................................................................................................135
Figure 6.28: Axisymmetric deformation modes for the quadratic triangular element using one Gauss
point. ......................................................................................................................................................136
Figure 6.29: Orthogonalizing deformation modes for the quadratic triangular element using one Gauss
point. ......................................................................................................................................................136
Figure 6.30: Axisymmetric deformation modes for the quadratic triangular element using one Gauss
point. ......................................................................................................................................................137
Figure 6.31: Axisymmetric deformation modes for the quadratic triangular element using three Gauss
point. ......................................................................................................................................................138
Figure 6.32: Axisymmetric deformation modes for the quadratic triangular element using three Gauss
point. ......................................................................................................................................................139
Figure 6.33 Axisymmetric deformation modes for the quadratic triangular element using seven Gauss
point. ......................................................................................................................................................139
Figure 6.34 Axisymmetric deformation modes for the quadratic triangular element using seven Gauss
point. ......................................................................................................................................................140
Figure 6.35: Axisymmetric displacement modes for the linear quadrilateral element using one Gauss
point. ......................................................................................................................................................141
Figure 6.36: Orthogonalizing deformation modes for the linear quadrilateral element using one Gauss
point. ......................................................................................................................................................142
Figure 6.37: Axisymmetric deformation modes for the linear quadrilateral element using one Gauss
point. ......................................................................................................................................................142
Figure 6.38: Axisymmetric deformation modes for the linear quadrilateral element using 2x2 Gauss
point. ......................................................................................................................................................143
point. ......................................................................................................................................................143
point. ......................................................................................................................................................143
point. ......................................................................................................................................................144
Figure 6.42: Axisymmetric deformation modes for the quadratic quadrilateral element for one Gauss
point .......................................................................................................................................................145
Figure 6.43: Orthogonalizing deformation modes for the quadratic quadrilateral element using one
Gauss point. ...........................................................................................................................................145
Figure 6.44: Axisymmetric deformation modes for the quadratic quadrilateral element using one Gauss
point. ......................................................................................................................................................147
Figure 6.45: Axisymmetric deformation modes for the quadratic quadrilateral element for 2x2 Gauss
point. ......................................................................................................................................................148
point. ......................................................................................................................................................149
point. ......................................................................................................................................................150
xiii
point. ......................................................................................................................................................151
point. ......................................................................................................................................................152
Figure 6.50: Scheme for the compression test. ......................................................................................153
Figure 6.51: Mesh used for compression test a) Linear triangular elements b) Quadratic triangular
elements c) linear quadrilateral elements d) Quadratic quadrilateral elements. ..................................153
Figure 6.52: Displacement results for the compression test for the different elements type. ................154
Figure 6.53: Scheme to validate the prescribed temperature at the boundary cylinder. .......................155
Figure 6.54: a) Pressure teste for one linear quadrilateral element using two Gauss points to integrate
the force b) using one Gauss point to integrate the force. .....................................................................155
Figure 7.1: Layout of the preprocessor. ................................................................................................157
Figure 7.2: Option to model the simulation. ..........................................................................................158
Figure 7.3: Software postprocessor overview. ......................................................................................159
Figure 7.4: Software postprocessor. ......................................................................................................159
Figure 7.5: Software postprocessor thickness distribution. ...................................................................160
Figure 7.6: Software core diagram. ......................................................................................................161
Figure 7.7: Input file example. ..............................................................................................................163
Figure 7.8: Example of the results file. ..................................................................................................163
Figure 7.9: Example of the update file. .................................................................................................164
Figure 7.10: Example of the list file. .....................................................................................................165
Figure 8.1: Process selection for container height (cm). ......................................................................167
Figure 8.2: Gob shape calculation. .......................................................................................................167
Figure 8.3: Temperature distribution in gob loading for the case 1 for 0.334s, 0.342s, 0.427s and
0.763s. ....................................................................................................................................................175
Figure 8.4: Boundary temperature for gob loading stage for case 1. ...................................................175
Figure 8.5: Temperature distribution in gob loading for case 2 for 0.365s, 0.382s, 0.429s and 0.555s.
...............................................................................................................................................................176
Figure 8.6: Boundary temperature for gob loading stage for case 2. ...................................................176
Figure 8.7: Displacement module in gob loading for case 2 for 0.368s, 0.375s and 0.381s. ................177
Figure 8.8: Displacement module with the neckring detail in gob loading for case 2 for 0.3877s,
0.3973s 0.4165s. ....................................................................................................................................178
Figure 8.9: Displacement vector with neck ring detail in gob loading for case 2 for 0.3877s, 0.3973s
0.4165s ...................................................................................................................................................178
Figure 8.10: Temperature distribution in gob loading for case 3 0.326s, 0.335s, 0.369s and 0.490s...179
Figure 8.11: Boundary temperature for gob loading stage for case 3. .................................................179
...............................................................................................................................................................180
...............................................................................................................................................................181
Figure 8.16: Temperature distribution in settle blow for the case 1 for 0.796s, 0.934s and 1.083s .....183
Figure 8.17: Boundary temperature for settle blow stage for case 1. ...................................................183
Figure 8.18: Temperature distribution in settle blow for the case 2 for 0.555s, 0.666s and 1.333s. ....184
Figure 8.20: Displacement module with the neckring detail in gob loading for case 2 for 0.555s, 0.744s
0.850s. ....................................................................................................................................................185
Figure 8.21: Displacement vector with the neckring detail in gob loading for case 2 for 0.555s, 0.744s
0.850s. ....................................................................................................................................................185
xiv
Figure 8.28: Temperature distribution in corkage reheat for the case 1 in 1.500s, for the case 2 in
2.044s, for the case 3 in 1.863s, for the case 4 in 4.090s and for the case 5 in 2.682s. .........................189
Figure 8.29: Temperature distribution in counter blow for the case 1 for 1.500s, 1.538s, 1.570s, 1.622s,
...............................................................................................................................................................192
Figure 8.30: Simulation temperature of the boundary for counter blow stage for case 1. ....................192
Figure 8.31: Displacement module in counter blow for the case 1 for 1.500s, 1.538s, 1.570s, 1.622s, 194
...............................................................................................................................................................196
Figure 8.33: Boundary temperature for counter blow stage for case 2. ................................................196
Figure 8.34: Displacement module in counter blow for the case 2 for 2.044s, 2.161s, 2.376s, 2.571s, 198
Figure 8.35: Displacement vectors with the baffle detail in counter blow for the case 2 for 2.939s,
2.962s, 2.974s and 2.991s. .....................................................................................................................198
...............................................................................................................................................................200
Figure 8.37: Boundary temperature for counter blow stage for case 3.................................................200
...............................................................................................................................................................202
...............................................................................................................................................................204
Figure 8.42: Temperature distribution in reheat for the case 1 in 3.958s, for the case 2 in 5.622s, for the
case 3 in 4.705s, for the case 4 in 10.227s and for the case 5 in 6.117s. ...............................................205
Figure 8.43: Temperature distribution in invert for the case 1 in 4.514s , for the case 2 in 6.666s, for the
case 3 in 5.490s, for the case 4 in 11.590s and for the case 5 in 6.823s................................................207
Figure 8.44: Boundary temperature for invert stage for case 1, case 2, case 3, case 4 and case 5. .....209
Figure 8.45: Temperature distribution in rundown for the case 1 in 4.514s, 5.201s, 5.833s. ...............210
Figure 8.46: Boundary temperature in rundown for case 1. .................................................................210
Figure 8.47: Temperature distribution in rundown for the case 2 in 6.666s, 7.222s and 9.111s. .........211
Figure 8.49: Displacement module in rundown for the case 2 in 6.666s, 7.222s and 9.111s. ..............212
Figure 8.52: Temperature distribution in rundown for the case 4 in 11.590s, 12.898s, 14.545s. .........214
Figure 8.56: Temperature distribution in final blow for the case 1 in 5.555s, 5.556s, 5.558s, .............217
Figure 8.58: Displacement module in final blow for the case 2 in 9.111s, 9.125s, 9133s, ....................219
Figure 8.60: Temperature distribution in final blow for the case 4 in 14.545s, 14.552s, 14.563s, .......221
Figure 8.61: Temperature distribution in rundown for the case 5 in 9.882s, 9.884s, 9.892s, ...............222
Figure 8.62: Temperature distribution in mould open for the case 1 in 7.292s, for the case 2 in 12.555s,
for the case 3 in 10.294s, for the case 4 in 22.500s and for the case 5 in 14.282s. ...............................223
Figure 8.63: Final boundary temperature for case 1, case 2, case 3, case 4 and case 5. .....................225
Figure 8.64: Bottle thickness (mm) distribution in case 1. ....................................................................226
xv
Figure 8.66 Bottle thickness (mm) distribution in case 3. .....................................................................228
Figure 8.67: Final bottle shape for the case 4. ......................................................................................228
Figure 8.70: Temperature distribution in gob loading for the case 1 in 0.175s, 0.199s, 0.252s and
0.466s. ....................................................................................................................................................231
Figure 8.72: Temperature distribution in gob loading for the case 2 in 0.096s, 0.140s, 0.238s, 0.383s.
...............................................................................................................................................................232
...............................................................................................................................................................233
...............................................................................................................................................................234
Figure 8.78: Displacement module in gob loading for the case 4 in 0.2513s, 0.2667s, 0.3213s ...........235
Figure 8.79: Displacement vector in gob loading for the case 4 in 0.2513s, 0.2667s, 0.3213s ............235
...............................................................................................................................................................236
Figure 8.82: Temperature distribution in gob loading for the case 6 in 0.076s, 0.128s, 0.303s and
0.460s. ....................................................................................................................................................237
Figure 8.83 Boundary temperature for gob loading stage for case 6. ...................................................237
...............................................................................................................................................................238
...............................................................................................................................................................239
Figure 8.88: Temperature distribution in baffle on for the case 1 in 0.800s, for the case 2 in 0.859s, for
the case 3 in 1.147s, for the case 4 in 0.889s, for the case 5 in 0.972s, for the case 6 in 0.801s, for the
case 7 in 0.957s and for the case 8 in 1.126s.........................................................................................241
Figure 8.89: Temperature distribution in plunger up for the case 1 in 0.8000s, 0.8002s, 0.8003s .......243
Figure 8.90: Boundary temperature in plunger up for case 1. ..............................................................243
Figure 8.92 Boundary temperature for plunger up stage for case 2. ....................................................245
Figure 8.94 Boundary temperature for plunger up stage for case 3. ....................................................247
Figure 8.96 Boundary temperature for gob loading stage for case 4. ...................................................249
Figure 8.97: Displacement module in plunger up for the case 4 in 0.8888s, 0.8890s, 0.8891s ............251
Figure 8.98: Displacement vectors in plunger up for the case 4 in 0.8888s, 0.8890s, 0.8891s .............252
Figure 8.100 Boundary temperature for plunger up stage for case 5. ..................................................254
Figure 8.101: Temperature distribution in plunger up for the case 6 in 0.8013s, 0.8014s ...................256
Figure 8.102: Boundary temperature in plunger up for case 6 and the thickness (mm) in plunger up
stage for case 6. .....................................................................................................................................257
Figure 8.103: Temperature distribution in plunger up for the case 7 in 0.9572s, 0.9575s, 0.9580s .....259
Figure 8.105: Temperature distribution in plunger up for the case 8 in 1.1267s, 1.1272s, 1.1277s .....261
xvi
Figure 8.107: Temperature distribution in reheat for the case 1 in 2.733s, for the case 2 in 2.837s, for
the case 3 in 3.160s, for the case 4 in 2.777s, for the case 5 in 2.847s, for the case 6 in 3.178s, for the
case 7 in 2.871s and for the case 8 in 3.943s. ........................................................................................263
Figure 8.108: Temperature distribution in invert for the case 1 in 3.760s, for the case 2 in 3.910s, for
the case 3 in 3.733s, for the case 4 in 4.194s, for the case 5 in 3.819s and for the case 6 in 3.561s, for
the case 7 in 3.487s and for the case 8 in 4.788s. ..................................................................................265
Figure 8.109: Boundary temperature for invert stage for case 1, case 2, case 3, case 4, case 5, case 6,
case 7 and case 8. ..................................................................................................................................267
Figure 8.110: Temperature distribution in rundown for the case 1 in 3.760s, 5.026s and 5.800s. .......268
Figure 8.111 Boundary temperature for run down stage for case 1 ......................................................268
Figure 8.113: Boundary temperature in rundown for case 2. ...............................................................269
Figure 8.118: Displacement module in rundown for the case 4 in 4.194s, 4.974s and 5.972s..............272
Figure 8.121: Temperature distribution in rundown for the case 6 in 3.561s, 4.298s ..........................274
Figure 8.122: Boundary temperature in rundown for case 6 and the thickness (mm) in run down stage
for case 6. ..............................................................................................................................................275
Figure 8.127: Temperature distribution in final blow for the case 1 in 5.800s, 5.803s, 5.804s ............279
Figure 8.131: Displacement module in final blow for the case 4 in 5.972s, 5.418s, 5.419s ..................283
Figure 8.133: Temperature distribution in final blow for the case 6 in 5.584s 5.583s..........................286
Figure 8.136: Temperature distribution in mould open for the case 1 in 7.399s, for the case 2 in 8.205s,
for the case 3 in 7.600s, for the case 4 in 7.750s, for the case 5 in 7.847s, for the case 6 in for the case 7
in 7.653s and for the case 8 in 10.516s. .................................................................................................290
Figure 8.137: Final boundary temperature for case 1, case 2, case 3, case 4, case 5, case 6, case 7 and
case 8. ....................................................................................................................................................292
Figure 8.138: Bottle thickness (mm) distribution in case 1. ..................................................................293
Figure 8.142: Real thickness (mm) distribution in case 4. ....................................................................296
Figure 8.143: Thickness (mm) distribution in case 5. ...........................................................................296
Figure 8.147: Volume evolution for blow and blow process for the case 1, case 2, case 3, case 4 and
case 5. ....................................................................................................................................................301
xvii
Figure 8.148: Volume evolution for press and blow process for the case 1, case 2, case 3, case 4, case
5, case 6, case 7 and case 8. ..................................................................................................................303
Figure 8.149: Number of nodes for the case 1, case 2, case 3 case 4 and case 5. ................................304
Figure 8.150: Number of nodes for the case 1, case 2, case 3 case 4, case 5, case 6, case 7 and case 8.
...............................................................................................................................................................305
Figure 8.151: Plunger displacement for case 1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8
for a slip penalty value of 1.0e04. ..........................................................................................................307
Figure 8.152: Time increment for blow and blow process for case 1, case 2, case 3, case 4 and case 5.
...............................................................................................................................................................309
Figure 8.153: Time increment for blow and blow process for case 1, case 2, case 3, case 4, case 5, case
6, case 7 and case 8. ..............................................................................................................................311
xviii
List of Tables
List of Tables
Table 2.1: Viscosity reference temperatures. ...........................................................................................11
Table 2.2: Composition of typical glass containers .................................................................................18
Table 2.3: Oxides functions in glass containers. .....................................................................................19
Table 7.1: Files created during the simulation. .....................................................................................161
Table 8.1: Simulation cases for the blow and blow and press and blow process. .................................166
Table 8.2: Simulation variables for blow and blow (left) and press and blow (right) processes. .........168
Table 8.3: Machining timing for the blow and blow cases being respectively the case 1, case 2, case 3,
case 4 and case 5. ..................................................................................................................................169
Table 8.4: Machining timing for the press and blow cases being respectively the case 1, case 2, case 3,
case 4, case 5, case 6, case 7 and case 8. ..............................................................................................171
Table 8.5: Air temperature considered in the different stages. ..............................................................171
Table 8.6: Material chemical composition for different simulations cases in blow and blow and press
and blow processes. ...............................................................................................................................172
Table 8.7: Material behaviour for different simulations cases in blow and blow process. ...................172
Table 8.8: Material behaviour for different simulations cases in press and blow process. ...................173
xix
Nomenclature
Nomenclature
Physical Quantities
Symbol Description Unity [SI]

Scalar
A Area [m2]
A, B, C, D, T0 Empirical constants
Bi Biot number
c Specific heat [JKg-1K-1]
*
c Critical crack length [m]
E Young’s modulus [Pa]
Fo Fourier number
G Shear modulus
h Heat transfer coefficient [Wm-2K-1]
k , kx , k y , kz Thermal conductivity [Wm-1K-1]
L Length [m]
m Mass [Kg]
p Pressure [Pa]
Q Heat Energy [J]
R Molar gas constant [m2Kgs-2K-1mol-1]
r Radius [m]
t Time unity [s]
t0 Initial time [s]
T Temperature [K]
T1,m , T2,m Temperature mould node [K]
Ti Initial temperature [K]

𝑇𝑎 , 𝑇∞ Ambient temperature [K]
u Displacement [m]
V Volume [m3]
Vc Characteristic volume [m3]
 Heat transfer coefficient [Wm-2K-1]
 Penalty value
c Heat transfer coefficient convection [Wm-2K-1]
cr heat transfer coefficient convection/radiation [Wm-2K-1]

 Stefan–Boltzmann constant [Wm−2K−4]
 Emissivity
xx
 Strain Rate [s-1]
e Elastic strain rate [s-1]
p Plastic strain rate [s-1]

 Viscosity [Pa.s]
0 pre-exponential factor
k shear viscosity [Pa.s]

 Penalty number
 Density [Kgm-3]
f failure stress [Pa]
e Stress rate [Pa.s-1]

G Activation energy [J]
L Element size [m]
T Temperature difference [K]
V Volume difference [m3]
Symbol Description
Vector
b Volume forces
fi , f applied forces
Ni , N Shape functions
R Rotation matrix
Qi , Q Thermal loads
Ti Nodal temperature
vi , v Velocity
Symbol Description
Matrix
xxi
B Derivative matrix
Cij , C Capacity matrix
eii Dilatational strain rate
eij Deviatoric strain rate
K ij , K Stiffness matrix
Sij Deviatoric stress tensor

𝛼𝑖𝑗 Thermal expansion tensor
𝜀𝑖𝑗 Strain tensor
 ii Dilatational strain tensor
 ij Strain rate tensor
 ij Stress tensor
m Dilatational stress
Symbol Description
Number
i , m ,n Matrix index
Symbol Description
Abbreviations
VFT Vogel–Fulcher–Tammann equation
AG Adam–Gibbs Equation
WLF Williams-Landel-Ferry Law
CPU Central Processing Unit
FEM Finite Element Method
RAM Random Access Memory
PLC Programmable Logic Controller
HEC Hot End Coating
CVD Chemical Vapour Deposition
CEC Cold End Coating
HTC Heat Transfer Coefficient
RI Reduced Integration
IS machine Individual Section Machine
SPC Statistical Process Control
xxii
Chapter 1 Introduction
CHAPTER 1
1 Introduction
A brief introduction about the glass manufacturing is done in this chapter and, in addition, a thesis synopsis
and the objectives are presented.
1.1 Historic Review of Glass Making

Created millions of years ago, glass is an abundant and easily accessible material that has been present
since the formation of our planet. It is formed by high temperature phenomena such as volcanic eruptions,
lightning strikes or by the impact of meteorites that melt the rock then, rapidly enough, solidifies it, making
the liquid-like structure stay in a glassy state [1].
Lava spewed by volcanos gets cooled to form glass. Meteor impacts are also known to melt the earth’s
crust and with the occurrence of subsequent impacts form smaller bits of glass [2]. Impactites are a natural
occurring type of glass formed by amorphous crystalline material after the impact of a meteorite shockwave
[2]. When glass was invented it was quickly noticed how resistant it was, how it didn’t suffer from
devitrification as well as water corrosion. Our stone-age ancestors noticed the strength and sharpness of
this material when they used obsidian to produce tools and arrowheads[1].
The melting of raw materials to create glass is a process that was discovered thousands of years ago. The
Egyptians, no earlier than 7000 B.C., manufactured jewellery and glass beads [3]. In Mesopotamia, objects
with coloured glazes containing copper compounds have been found dating to 4000 B.C. [2]. Feathery or
zigzag patterns of coloured threads on the surface of some objects have also been found [4]. Glass was
produced by the Egyptians until 1500 B.C. [3]. The recovered glass vessels from Egypt were manufactured
between 14th and 16th century B.C. [2]. The adopted method is believed to have been to melt the glass,
draw the threads out, turn them around a clay or sand object and then remelt the glass threads, using
chemicals to give colour to the glass. Patterns were made by remelting pieces of coloured glass. The
manufacture of Egyptian glass declined around the 11th century B.C. and in the meantime, it spread to
other eastern Mediterranean regions such as Syria, Cyprus and Palestine. The major glass making centres
became Syria and Palestine using the same techniques and colours that the Egyptians used, after 1000 B.C.
they made core-formed vessels with different shapes [2]. Around 200 B.C. the Syrian craftsmen
revolutionized the making of glass by discovering the blowing pipe [3]. With the development of the glass
1
blowing technique the possibilities of shaping glass became endless [4]. Generating an amazing range of
glass applications as well as improving the quality of glass jars and bottles making the glass drinking vessels
popular [3]. The next step was to use clay moulds for contours while blowing the glass to create flasks and
other shapes, they also realized that glass could be shaped freehand to any form they desired. This created
decorative elements, handles, feet that could be added as one desired [4]. Around 400 B.C. Macedonia and
Greece emerged also as centres of glass manufacture [2]. Techniques for making glass tableware such as
bowls, the use of brilliant colours, also the use of lathes and grinding wheels to manufacture decorations
had already been developed. Greeks developed a technique to sandwich gold layers between clear glass
parts and the mosaic forming, this technique was developed in order to impart special colour effects. The
Roman Empire, between 3rd and 2nd century B.C., spread the making of glass through Rome and Italy
with the methods of imparting different colours and glass blowing becoming widely popular. The making
of cameo glass became one of the most sought techniques of this era [2]. Romans became the first
civilization to experiment blowing glass inside moulds leading to improve the glass jars and bottles and the
production of drinking vessels thereafter [5]. After the Roman Empire, the Byzantine regions (Greek
Empire), under Islam, came with new developments. In the 7th century A.C. glass manufacture thrived in
Syria and Palestine. The hedwig glass belongs to this era and it was created using diamond cutting. Baghdad
emerged as the glass making centre of the 9th A.D. century after the Arab invasion that spread through
most part of Asia and the closest Mediterranean parts of Europe. In the 13th century A.C. and through the
next two centuries Persia became the major manufacturer of fine glassware and developed the technique of
grinding glass surfaces. In the 14th century the independent republic of Venice became the centre for glass
manufacture of Europe. Two significant developments occurred at this time: the discovery that the addition
of calcia gave bright, shiny glass and the glass could also be worked to be incredibly thin, crystal-like clear
with feather-light weight; the second was the creation of ice glass by plunging molten hot glass into ice
cold water with an iron pipe for just a few moments and then blown, creating what appeared to be a tiny
cracks within its surface. Glass making only flourished to the other parts of Europe by the end of the 16th
century, it was at this time that English workers develop the waldglass (green or amber tinted glass).
Development in Europe came in the form of techniques and popular designs that made glass articles
affordable by the middle classes. With the 19th century came the birth of the mechanized glasswork not
only in Europe but the United States as well. Opaline, a translucent milky-white glass, was created in France
around 1810 by adding metallic oxides to the glass mixture. The modern method of glass making grew
roots in most countries in Europe, America, Japan, India and China in the end of the 19th century. It was
in the 20th century that glass became part of the human life with an evolution of manufacturing techniques
and a deeper scientific understanding of it. This scientific understanding of glass was proportionally
connected to the growth of chemistry and physics understanding, mainly in the field of inorganic materials.
This growth was born from the observation of the effects of the addition of oxides and other chemicals
produced in silica glass [2]. The invention of glass vacuum tubes paved the path to create most modern
electronics today. The development of glass fibre optics is revolutionizing the telecommunications industry
by replacing copper wires enhancing our ability to transmit data around the world [3]. Since its discovery
glass became a stalwart material for our society, from its commonly day uses to architecture, to science and
many other fields.
1.2 History of Glass Moulding

Blow moulding is believed to have been created around 1821 in Bristol, England. In 1859 Mein of Glasgow
designed the first vertically spilt iron mould which had a movable base plate to make bottles by blowing
compressed air. The biggest change to the manual process was to make the mouth or finish first so that the
glass could be held in place while the rest of the manufacture operations were performed [6]. In 1866, H.M.
Ashley and Josiah Arnall changed the way to fabricate containers. To answer the demands for sterile and
refrigeration containers they created a new way to produce bottles, with semi-automatic machines, using
an inverted mould with a plug at the bottom to create the mouth. [6]. While in hand blowing, the bottle neck
was finished last (after blowing operations), a major change happened when this operation was made first
by the machine [1]. In 1887, Ashley made some improvements, a separate parison 1 mould gave a better
1
Mass of glass that has been partially shaped or molded.
2
distribution of wall thickness and a separate neckring mould to hold glass while being manipulated. This
was done with a simple machine that needed to be fed gobs of glass by hand, reducing the labour cost or so
it was claimed. This achieved an almost consistent weight of glass and internal capacity [6].
In the 1900 the progressive use of soda-lime-silica glass was started in the glass containers production.
Mechanization in the production of glass containers was also in development.
An American engineer called Michael Owens (1859–1923) invented an automatic bottle blowing machine.
This innovation was a successful process, which arrived in Europe after the turn of the century and
considerably transformed the containers industry. A further improvement of the process is given by
Hartford Company which invented the feeder in 1922 to provide the forming machines with a thermally
conditioned glass gob2. The feeder mechanism worked constantly as an extrusion machine which create a
uniformly gob with a specified rate.
The feeders are still employed todays and with the invention of individual section (IS) machines by 1925,
they deliver several forming machines. A great advantage of the machine was that one section could be
shut down to change moulds without stopping the production in the other sections. These inventions
answered the growing demand for glass containers.
Along time glass as become a common material used in diverse applications. Firstly, empirical knowledge
had a principal importance in developing glass industry. Nowadays trial and error is still a manner to
improve quality in glass industry, however this type of production is highly time consuming and expensive.
We live in a world in which glasses play a very important role in all aspects of our daily life and this is
enough reason for us to study its forming processes.
1.3 Glass Containers Industry

Glass forming processes take place at high production rates. At the same time, glass manufacturers desire
to optimize glass products to their customer’s satisfaction. Glass containers process is a very complex
procedure which must be thoroughly controlled in order to be able to satisfy the even more exigent
costumers. Unfortunately, practical experiments are in general quite expensive, whereas the majority has
to be performed under complicated circumstances. In addition, each glass manufacturing company keeps
secret about its own experimental data. Therefore, it is no surprise that in the recent past glass process
techniques were based on experience, rather than on scientific research. Nowadays computer simulation
models are necessary to gain improvement and a better understanding of glass forming processes.
The glass containers manufacturers have to deal with challenges to improve the productivity of the plant or
reduce the productions costs without the decrement of the quality levels imposed by the customer. Glass
industry needs to work 24h per day and 7 days a week in order to keep the furnace working. Combine high
productivities and low production costs are the main goals that the glass industry tries to achieve. Glass
industries require sustainable profitability, quality and increase market share in the packaging industry. To
maintain the glass containers quality the forming process needs stability. Glass industry tends to approve
the new technologies slowly preferring to use the empirical knowledge of their workers. The
competitiveness in glass industry brings the necessity to improve the design and the quality of their
products. Different shapes, glass composition, colours may needs different changes in the processing
parameters which can have very expensive costs.
Nowadays, glass industry has achieved considerable goals in efficiency and production rate without loss of
product quality. They have been doing progresses in developing raw materials and processes using the
knowledge created along the years. Utilizing the knowledge existent in materials properties and tools and
R&D progress and opportunities, the glass industry is improving the production. Glass researchers are
2
A small masss of glass with shape of a cylinder
3
making progress in manufacturing new functional glasses, upgrading existent technology continuously and
also the simulation of the process can play an important role in this field.
1.4 Glass Container Quality

Glass has a wide range of uses. The main characteristics of glass are including shock-resistance,
soundproofing, transparency and reflecting properties, which makes it particularly suitable for a wide range
of applications, such as windows, television screens, bottles, drinking glasses, lenses, fibre optic cables,
sound barriers and many other applications.
Glass products are engineered to fit the needs of customers, the major qualities their claim are resistance,
durability, and ability to support rigorous condition. Special customers may need especial glass with special
chemical composition and the glass industry should be prepared for the melting point and especial
production processes needed to meet these requirements.
A key issue in glass containers is the wall thickness distribution. The thicker the container, the stronger it
is and the less easily it breaks. The thinner the container, the lighter it is and less costs are spent on material.
The optimal wall thickness distribution have to be optimize between these aspects. The ideal thickness
distribution is not necessarily uniform, as some parts of the container are more vital than others, e.g. corners,
and should be thicker for optimal strength. The difficulty in blow moulding a container with a prescribed
wall thickness distribution is that the corresponding initial operating settings are not known beforehand.
There are essentially two ways to deal with this. The first approach is the use of numerical simulation. If
the initial operating conditions can be estimated, for example by empirical expertise of the process, a
mathematical model can be developed that can directly compute the corresponding container production.
Then, the initial operating settings of the blow moulding process can be adjusted to improve the wall
thickness of the computed container. The first approach to attempt the desired wall thickness distribution
is commonly a trial and error procedure, which is often based on the manufacturer’s knowledge of the
process. However, empirical expertise is often insufficient to find the initial operating settings, particularly
when innovative processes are involved. A second approach is to solve an inverse mathematical problem
to find the initial operating settings corresponding to the container with the desired wall thickness
distribution. This class of inverse problems is quite challenging, as in general couple, highly nonlinear
physical systems and complicated geometries are involved [7]. Usually, numerical optimization methods
are employed to find a solution to the inverse problem [8]. Some advantages of lighter weight containers
are, less cost to manufacture, higher forming machine production rates possible, shipping load and space
requirements reduced, more competitive with plastic, cans, and paper containers.
The design of a container must be suitable for weight reduction without a reduction in strength. This is
extremely important because a bottle can usually be effectively made lighter in weight only if can be
designed with a stronger structural shape taking into consideration the particular forming process used to
produce the containers.
Light weighting is important to improve productivity, when the glass containers are thinner and lighter
some contribution for productivity are made and are the following,
o Lighter weight containers means that more containers per ton of glass can be produced.
o Lighter weight means less glass and therefore, less heat to be removed and the result is an increase
in production speed.
o Saving weight also reduces glass cost per container which is a contribution of financial
productivity.
Glass is basically made from sand and sand is largely spread throughout the earth’s crust with very high
purity of silica and uniform grain size. Uniform grain size is a very desirable quality because it can
correspond to a consistent chemical composition. Sand before being used as a raw material needs some
treatments to remove impurities like iron and ferric oxide, chromic oxide and heavy minerals. In the glass
4
the control of chemical composition of raw materials used in the processes is necessary in order to guarantee
a high and constant level of quality of the product.
Glass is chemically inert, and environmentally friendly. Glass material is a preferred material of choice for
the packaging industry worldwide for many years. The knowledge of the temperature is important in almost
all stages of the glass production and glass processing. During the melting process in the glass tank the
temperature influences the homogeneity of the glass melt. Undesired cracks could be the result of a wrong
cooling of the glass [9]. Thus, the knowledge of the right temperature is important to guarantee the quality
of the final products.
Heat conduction, heat convection and radiation plays an important role in the production of glass containers.
Other properties, such as viscosity, thermal expansion, and electrical conductivity, are strongly dependent
on glass composition. Glass is a very versatile material, it is quick to form in shape, cheap to produce and,
depending on the recipe used it can be extremely durable. It can be transparent, opaque and take a greater
range of colour and decorative effects than any other man made material.
Viscosity is the most important mechanical property in glass material and varies enormously with chemical
composition. Since viscosity varies with chemical composition and is highly dependent of temperature,
different glass types have very different behaviours while are worked at high temperatures.
Customers usually take no particular interest in the chemical composition of glass. Most catalogues by glass
producers, therefore, do not indicate the chemical composition of the products. For glass producers,
however, the chemical composition of the products and its control are highly important. To know the
temperature dependence on viscosity is of fundamental importance for glass science and technology [10].
1.5 Glass Container Simulation

Modelling has become an important tool to accelerate the innovation while reducing the trial and error
methods commonly used in production environments. Numerical simulation provides detailed information
about forming process and thereby helps in determining the effects of different parameters on the final form
of the item. Simulation can be done in shorter times and their costs are much lower than that of making trial
moulds. Therefore, numerical simulation technique has the potential capacity to largely replace the other
design techniques.
Glass forming techniques are still mostly based on empirical knowledge and hands on experience.
Experimental trial using glass forming machines are in general very expensive and time consuming.
Simulations give an alternative to an empirical knowledge avoiding the trial error procedures used in
factories. Also help to reduce waste as well as the wall thickness reducing weight and maximizing strength.
In the last years, simulations software based on the finite element method has become an important tool to
improve understanding, controlling and optimized the process. It is expected that this can give the final
shape, thickness distribution, temperature distribution, etc.
The production of glass container takes place at high production rates. Time quality factors of the products,
such as smoothness, strength, weight and cooling conditions, are optimized in order to pass the quality
control process. To optimize and control the process is required knowledge. Unfortunately, measurements
are often complicated, considering that blow moulding processes take place at high rates in closed
constructions under complicated circumstances at high temperatures. Simulation models may offer a good
alternative for complicated experimental measurements.
Glass-packaged products face a significant competitive pressure from products made of alternative
materials. Nowadays, plastic, paper and related material and metallic (Aluminium and steel) containers
share a large part of the container market. This has forced the glass industry to produce lighter containers
(this has pushed the tools toward and the glass containers have more uniform glass distribution all over the
article) for short-term conservation purposes while offering new attractive designs and promoting glass
recycling. The growth of competitiveness on the industries that use glass containers puts demands on the
5
glass industry to develop new designs. In this context, simulations models are essential as virtual
prototyping tools, reducing time and costs. To be effective, the model must give some information for the
optimization of the process, namely thickness distribution, residual stresses and mould temperatures
distributions.
Glass container production can be simulated by means of a mathematical model, which is solved
numerically to visualize the process at discrete times. A representative numerical simulation should give as
output the containers final shape and wall thickness. In addition, numerical tools to simulate the glass
forming process can be used to optimize an existing process. Process simulations can help minimize
undesired variations in the wall thickness and reduce the weight while maintaining the strength. They can
also help optimize cooling conditions and increase the production speed. The goal of all modelling software
is to reduce the risk of new designs, reduce development time and find more optimal solutions in a virtual
environment before physical creation.
The forming of hollow ware glass articles is a complex coupled thermal/mechanical problem with
interaction between the heat transfer analysis and the viscous flow of molten glass. The glass viscosity,
which governs the glass flow during the forming process, is strongly dependent on the temperature [11]
and the contact with the moulds along with the constant geometry changes alters significantly the heat
transfer process. The final product thickness distribution is one of the aspects which is more affected by
changing process conditions. Simulating glass forming with FEM is a particularly difficult task since the
material behaviour lies somewhere between a fluid and a solid. One of the first models was developed by
Cesar de Sa [11]. In his approach, the thermal problem is coupled to the mechanical one at every time step,
in order to handle the strong coupled phenomena. The model has provided good predictions of glass flow,
resulting from gravity and blow pressure. The model was the basis for the development of a commercial
2D code [9]. But, nowadays, manufacturers are facing an evolving and increasing competition in the
development of new complex 3D products. Tools are available which are capable of dealing with the most
important aspects of these forming processes [10-11]. Nevertheless, some particular aspects requires a
dedicated software or some user intervention in order to adapt commercial codes to this particular field
(which may not be an easy task). Besides the referred commercial code, a new approach has been recently
released, [12], based on meshless methods, with 3D modelling. Nevertheless, in those software some
important features in terms of process knowledge still lack or are simplified like, for example, the plunger
movement defined by a prescribed velocity, which does not represent the real case, where a pressure must
be set. A software suite for glass forming should include the calculation of the gob initial geometry and
temperature distribution, as well as moulds heat transfer with the glass and with external ambient
temperature, in order to optimize the final geometry and process efficiency. This last feature is not available
in the mentioned software. Besides the paramount importance in the final container thickness distribution,
it can be also extremely useful for machine developers since it allows to design dedicated cooling systems
that could push the process further to its limits. Therefore, the main objective is to build a robust and reliable
suite of numerical modelling programs designed to provide glassmakers with decision support system tools
to aid in the design, production and quality control of all types of glass containers. The programs should
combine the latest numerical technology to provide practical, representative simulation tools that
glassmakers can use and benefit from. They should be rooted in sound theoretical formulations to deal with
the complex coupled thermal and mechanical phenomena, as glass viscosity is dramatically influenced by
temperature distribution that itself is highly influenced by shape modification and contact with the tools
and moulds. The suite of programs should also comprise pre and post processors and analysis modules for
the complete definition of the problem, analysis and viewing of the results. It should be used to simulate
the forming process of round and non-round glass containers on an IS (individual section) forming machine
from gob delivery through prison formation, invert and final blow until mould opens. The types of forming
process modelled should include BB (Blow and Blow), PB (Wide Mouth Press/Blow) and NNPB (Narrow
Neck Press/Blow) processes. As final result it should be possible to: predict the glass thickness distribution
in the final product under a variety of conditions; analyses the thermal behaviour of the blank and blow
moulds; investigate the influence of process conditions like glass composition, IS Machine timing, gob
temperature, mould temperatures upon final container wall thickness results; evaluate the effects of mould
cooling on the forming process and container wall thickness results. The final program should be not only
6
a design tool for improvement and optimization of existing products but also a means of understanding the
main process parameters role to provide the assessment of new ideas for new products and innovation.
1.6 Objectives
The main objective is to perform a numerical tool to simulate the glass container forming. The program
results are the wall thickness distribution, final shape and temperature distribution. Also the program has
to be a robust and reliable suite of numerical modelling program. In general the program should provide
glassmakers with decision support system tools to aid in the design, production and quality control of all
types of glass containers.
1.7 Thesis Synopsis

Chapter 1 is dedicated to present the current of state glass industry. Then, chapter 2 focusses on the glass
material properties. On chapter 3 an exhaustive description of the glass container production is done,
since the raw material until the packaging. Chapter 4 explains the mathematical modelling of the glass
forming process. The constitutive model for the thermomechanical behaviour and a subsequently the
finite element solution method implementation is discussed. Chapter 5 goes into further detail on the
numerical model of the glass container simulation. In this chapter, numerical aspects are addressed, such
as the elements definition, the contact algorithm, mesh updating and the solution of the resulting system
of equations. Chapter 6 discusses the validation of the thermal and the thermomechanical formulation.
Some particular attention is paid in comparing the analytical heat transfer equation to the formulation
proposed. Also, the thermomechanical formulation is verified. Subsequently, in chapter 7 the software
layout is explained, in particular its environment and the program files. In chapter 8 a description of a set
of the simulation cases is performed. It includes a reference to the material properties, machining timings,
as well as the process parameters. All the process stages results for the blow-blow process and press-blow
process respectively are presented. Finally, in chapter 9 states some conclusions are drawn and some
future tasks are proposed.
7
This page intentionally left blank
8
Chapter 2 Glass Material Properties
CHAPTER 2
2 Glass Material Properties
A description of the glass material properties is presented in this chapter.
2.1 Brittleness
Glass is a brittle material at room temperature, this means that when is subjected to loads it can break
without significant deformations. Brittle materials cannot absorb plastic deformation, failing in a sudden
and catastrophic manner. Glass fracture once a critical stress is reached from a defect present in the glass.
Brittleness is affected by glass composition through hardness and toughness. These type of materials fail
when there is little or no evidence of plastic deformation before failure. However they have usually high
strength.
2.2 Density
Density of a material is defined as the mass divided per unit of volume:
m
 2.1
V
Density calculated is true density of the material considering that the sample is free of bubbles, voids, or
other defects. However, if the samples contain defects, as referred before, the density will be less than the
true density. Glass containing unknown material, such as, impurities from raw material, unmelted batch or
crystals formed during cooling, will cause an imprecise glass density. Viscosity can change the density.
Also, glass structure will vary the density depending on the cooling rate. Varying the cooling rate will
influence the atomic rearrangement of atoms which will change the final volume. This uncertain density
can make difficult to predict the stress.
9
2.3 Thermal Expansion

When the material is subjected to a temperature change, i.e. is heated from room temperature to a
temperature that promotes softening and dilatation. The thermal expansion coefficient has a relation
between the volume of a glass and its temperature being strongly influenced by the chemical composition
of the glass.
The deformation caused by the thermal expansion is given by:
 ij   ij T 2.2
which αij is the thermal expansion coefficient.
Glass material is characterized to have low thermal expansion. This mean that material will not expand or
contract very significant. However, the thermal expansion of glass can lead to residual stresses when it
cools to fast or is subjected to thermal shock. Theses stresses can be high enough to damage or fracture the
container. When the glass is being formed some concerns should be taken due to temperature gradients
inside the container. For example, if the surface of a material is cooling faster than the interior, the surface
will attempt to contract but it will be constrained by the internal material resulting in tension and the interior
in compression. For glass materials this can be sufficient to extend cracks, causing damage or fracture.
Residual stresses can be inside a container during the coating processes or in the thermal tempering.
Thermal expansion can lead to compression or tension stress. If the stresses are tensile, this can lead to
crack formation but if compressive, they can be used for strengthening.
2.3.1 Glass Cord

Different glass compositions can lead to different thermal expansion coefficients. When glass is produced
with the same chemistry composition, the properties will be consistent. When a nonstandard glass is made
from one batch or a series of batches, that glass will be a separate streak in the glass flow. If there is a
difference in the coefficient of expansion between the two glasses, a permanent stress will result as the
glass cools bellow the strain point (≈500℃). The stress occurs because the entire piece begins to contract
as it cools. The nonstandard glass might want to contract at different amount than the base glass. Since it
must contract to the actual amount determined by the base glass, that glass will have a stress induced
proportional to the difference in expansion coefficients. All these factors will influence the capacity of
container to withstand different types of stresses (pressure, thermal shock, impact, etc) before it breaks.
Glass cord is defined as a chemical inhomogeneity of the glass resulting in permanent stress due to different
coefficients of expansion between the layers of inhomogeneity. Cord usually is the result of the normal
glass having a streak of nonstandard glass due to batching problems, Figure 2.1a). Cord can also occur
when all batch ingredients are properly weighted but not well mixed. Each batch should be of relatively
uniform consistency to avoid cord. Segregation of materials in mixed batch storage can result in chronic
cord in the average glass density. Glass cord is observed by viewing in polarized light, Figure 2.1b). The
amount of stress will shift the light as passes through it and a corresponding colour shift (birefringence)
occurs.
10
Figure 2.1: a) Examples of a cord defects after cooling. b) Glass cord observed by viewing in polarized light [12].
2.4 Temperature dependence of viscosity

Viscosity is considered to be the most important criteria in glass making and the way the viscosity of a melt
varies with temperature forms the only determining step for shaping of glass articles since it depends of
capacity to formability at certain temperature. Viscosity also controls the stresses and strains to be applied
to form the glass. It determines the melting conditions, the temperature regimes for forming, annealing and
tempering, and the maximum temperature to avoid devitrification, Table 2.1. Consequently, numerous ways
to relate the glass viscosity as a function of temperature and composition have been made. The relationship
between viscosity and other physical properties can also lead to a better understanding of the glass structure.
For the forming machines used in the container glass making industry, it is essential that a constant viscosity
regime is maintained throughout in order to avoid or minimize any dimensional variation and also to avoid
defect formation.
Log Viscosity (Poise) Approx. Temperature (ºC)

Melt Viscosity 2 100 1449
Gob Viscosity 3 1 000 1185
Softening Point 7.65 44 668 000 724
Conveyor Viscosity 10 10 000 000 000 632
Annealing Point 13.37 23 442 000 000 000 546
Strain Point 14.6 398 110 000 000 000 499
Table 2.1: Viscosity reference temperatures.
Viscosity is highly dependent on temperature, as glass is heated the material resistance to deformation
gradually gets less important. Viscosity measures the resistance to shear deformation, therefore, measures
the resistance to flow.
At certain temperatures, known as the glass transition temperature Tg, some materials are no longer
considered to be liquids but rather amorphous solids or glasses.
11
Viscosity values should be considered with great care. The curve suggests that the rate of increase of
viscosity with decreasing temperature becomes greatly reduced, Figure 2.2.
Figure 2.2: Viscosity evolution in working zone temperature [13].
Arrhenius equation was one of the first equations that could describe the viscosity-temperature evolution.
The relationship between the viscosity-temperature of a glass is based on commonly employed conceptions
in order to describe the transport phenomena in solids and liquids where ionic species have, under the
influence of an electrical field, to overcome a potential barrier to affect conduction or diffusion. In this case
however, using the thermal energy available at a certain temperature, the viscous flow of atoms and/or
molecules needs to overcome the energy barrier [14]. The Arrhenius behaviour equation is given by,
 G 
 T   0 exp   2.3
 RT 
where ΔG and η0 are temperature-independent coefficients called the activation energy (for viscous flow)
and the pre-exponential factor, respectively. An Arrhenius temperature dependence of viscosity is quite
easily used and allows for rapid prediction of viscosity at a given temperature. This exponential expression,
however, doesn’t adequately describe viscosity variation over a wide temperature range since, with glass
forming melts, the activation energy isn’t a constant over the entire viscosity range [14].
Due to the limitations described above, various empirical equations have been put forward to deal with
practical production situations. Notably, Vogel, Fulcher and Tammann and Hesse found independently that
the following general equation provides a very good representation of the viscosity-temperature relationship
which is satisfactory, for most practical purpose, for a wide temperature interval [14]. For silicate glasses,
the VFT (Vogel–Fulcher–Tammann) equation often fits the temperature dependence over ten orders of
magnitude in viscosity [13]. The behaviour of the super cooled liquid (T>Tg) can be fitted empirically by
the Vogel–Fulcher–Tammann. The VFT equation involving three empirical constants (T0, A and B) and is
given by:
12
B
log10   A  2.4
T  T0
This equation can capture relatively well the non-Arrhenius temperature dependence over the melt-forming
region. This law can be considered as a correction of the Arrhenius law adding a T0 term. The VFT
approximation is still easy to use and allows for close prediction of the actual viscosity.
Such models however, can be used only within limited temperature ranges. However they don’t permit a
clear understanding of the changes that occur with temperature or composition [15]. Doremus provided a
further theoretical explanation of the Fulcher equation based on either free volume or temperature
dependence of the size of the structural units that actively participate in the viscous flow [14];
 B   D 
 T   A exp   1  C exp   2.5
 RT    RT  
where T is temperature, R is molar gas constant, A, B, C, and D are constants. This formula can be fitted to
the experimental data so it can give us the correct Arrhenius-type asymptotes of viscosity at high and low
temperatures [15].
An empirical formula proposed by Richert and Bassler in 1990, of some merit, fits the viscosity data,
particularly in the high viscosity regime (77>106 poises) [2].
2
T' 
   exp  0 
0
2.6
T 
where 𝑇0′ is a constant. This formulation is attractive, it avoids having to invoke any other temperature of
divergence as in the VTF equation [2].
Dyre et al. (1996) had a different approach to account for the experimental viscosity variations with
temperature as an alternative to VTF and AG models. They considered the flow in viscous liquids to arise
from unpredictable events involving motion and molecular reorganization. With the mechanism in mind,
the energy needed for such flow is decreased if the surrounding liquid is shoved aside to create the necessary
volume for rearrangement. This volume is completely different from the volume of the free volume theory
and is, in general, an activation volume. The free energy involved has five important thermal parameters,
namely the melting point, crystallization temperatures, glass transition temperature, softening temperature
and thermal expansion coefficient [2];
 F T  
2
  0 exp   2.7
 kT 
where ∆F(T)
F T   G T  .VC 2.8
where G∞ is the shear modulus, Vc is some characteristic volume, which is related to the activation volume,
and ∆V the volume of the liquid before shoving. It is given by [2].
2  V 
2
VC  2.9
3V
13
The non-linearity of the log viscosity as a function of inverse temperature arises due to the fact that G∞(T)
itself is temperature-dependent [2].
 D 
  0 exp  
 f g    L  C  T  Tg  
2.10
 
The previous relation is known as the Williams-Landel-Ferry (WLF) law.
The approach in the Adams-Gibbs model is to consider the configurational entropy Sc of the liquid. Viscous
flow is viewed as a cooperative rearrangement of the molecules [16]. As temperature decreases, less
configurations will be available and rearrangement will need an increasing number of molecules. This leads
to a decrease in configurational entropy and a decrease in mobility. The theory can be conveyed as [13];
 E 
  0 exp   2.11
 TSc 
where E and η0 are constants. The configurational entropy can be approximated by,
 1 1
Sc      2.12
 Tk T 
where α is a constant and Tk is the temperature at which Sc goes to zero.
2.5 Chemical Durability

Chemical durability refers to glass resistance from exposure to warm, humid conditions, etc. Non-silicate
glasses are very susceptible to dissolution in water, which limits their utility for applications involving
contact with aqueous liquids or water vapour, Figure 2.3. Routine tests for surface dealkalization in the
glass container industry generally evaluate the amount of alkali extracted from the glass when it is rinsed
with or exposed to purified water.
Figure 2.3: Chemical durability of a container [12].
14
2.5.1 Dealkalization
Dealkalization is a process of surface modification applicable to glasses containing alkali ions, wherein a
thin surface layer is created that has a lower concentration of alkali ions than is present in the underlying,
bulk glass. Commercial glass containers are made of soda-lime glass, therefore, have a substantial
percentage of sodium ions in their internal structure. Since sodium is an alkali element, its selective removal
from the surface results in a dealkalized surface. Dealkalization treatment hinders this process by removing
alkali from the inside surface. Not only does this mean less extractable alkali in the glass surface directly
contacting the product, but it also creates a barrier for the diffusion of alkali from the underlying bulk glass
into the product. The process of dealkalization should take in account the final product which will be
introduced inside the container.
2.6 Liquidus Temperature

The liquidus temperature is designated as the maximum temperature at which crystals can co-exist with the
melt in thermodynamic equilibrium. Above the liquidus temperature the material is homogeneous and
liquid at equilibrium. In the other hand, below the liquidus temperature more crystals may form in the melt.
The liquidus temperature for glass container is approximately 1024ºC. This means that if we hold the
temperature slightly below this temperature for a long period of time (hours to days) there will be crystals
which form in glass. This only occurs where abnormal cooling of the glass occurs in the refiner, foreheart
or spout areas.
2.6.1 Devitrification
Detrivication is the process of crystallization in a formerly crystal-free (amorphous) glass. The causes of
devitrification, can include to holding a temperature slightly below liquidus temperature for too long, which
causes the nucleation of crystals. In addition, the chemical compositions of some types of glass can make
them more vulnerable to devitrification than others.
2.7 Light Transmission – Colour

Glass colouring can be obtained by adding of colouring ions. Metals and metal oxides can be added to glass
during its manufacture to change its colour which can vary its aesthetic appeal. These oxides can be added
to give or hide colour from glass, Figure 2.4. Brightness is the average amount of visible light transmitted
by the glass.
15
Figure 2.4: Glass colours presented in some container [12].
2.8 Cullet
Cullet from various melting units usually can be added to the respective glass batch. Cullet (recycled glass)
can be used in the glass manufacturing process in variable percentages. The proportion depends on:
availability, type of glass and quality (colour). The use of cullet, or scrap glass, not only reduces waste, but
also aids in reducing batch-free time by both reducing the amount of refractory material in the batch, and
by providing additional liquid throughout the melting process.
Glass cullet is a high-value material since its usage results in:
o Reduction of consumption of raw materials (silica sand, soda, dolomite, etc.);

o Reduction of CO2 emission that is generated while melting the raw materials;
o Prolongation of the service-life of the glass furnace even up to 30%, because of the lower melting
temperature of glass cullet than of the raw materials;
o reduction in the consumption of energy sources for melting the raw materials, which means also
fewer emissions of NOx, SO2 and particles into the environment;
16
o cullet is 100% recyclable, out of 100 Kg of glass cullet 100 Kg of glass containers can be produced;
Impurities have to be removed from cullet, such as: ceramics, stones, iron and non-ferrous metals
(Aluminium, lead), quartz, glass ceramics, opal, and organic waste (paper, plastic, food rests).
The impact of these impurities in the recycled glass is the following:
o Ceramic materials like stones and chinaware, being hardly fusible or which hardly dissolve in the
molten glass, and non-ferrous metals;
o Alumina particles (Al2O3) not completely dissolve in the glass melt in the glass melting furnace;
o Metallic Aluminium in the glass melt reduces SiO2 into Si – globules (Si – spherically shaped due
to high interface tension) which have a different expansion coefficient;
o Other glass types like vitreous silica and glass ceramics, generally occur less often as impurity, but
if present will also create stress concentrations within the glass product;
o The organic components in the recycling glass can only be separated to a certain extent. These
impurities have a strong reducing effect on the glass melt (addition of C). Consequently the redox
state of the melt is decreased which may cause fining problems and colour variations in the glass.
2.9 Strength of the Glass

Strength of the glass is defined as the stress at failure and is measured of the resistance of glass fracture.
Inherently, glass is a strong material, however, it is sensitive do surface flaws. The reduction in strength is
due to surface flaws that severely weaken the glass. These flaws act as stress concentration points,
increasing the local stress to levels that exceed the theoretical strength and causes the fracture of glass.
Griffith handled this problem in detail and derived the expression [13]:
2 E
f  2.13
 c*
where σf is the failure stress and c* is the critical crack length.
The critical crack length attainment is only a necessary condition for crack to grow. It is also crucial for the
stress at the cracks tip to exceed the theoretical strength of the glass before the crack starts to grow
spontaneously [13].
There are two methods to increase glass strength. Removal of flaws is only effective in short periods of
time since new flaws are easily formed, while preventing their formation. We must concentrate on the
prevention of crack growth and accept the fact that there will be flaws. Since crack growth needs the
presence of a tensile stress at the flaw tip, creating compression on a near-surface region may prevent the
cracks to growth. Crack growth won’t occur until the stress applied is large enough to overcome the
compressive stress and produce a tensile stress at the crack tip. Compressive surfaces can be manufactured
by ion exchange, thermal tempering, by application or formation of a compressive coating [13].
2.10 Thermal Shocks

Some manufacturers have tried to create tempered containers. Thermal shock is a genuine problem
wherever glass is rapidly cooled over lengthy temperature ranges. A cooling rate gradient can lead to
thermal tempering of glass by producing distinct fictive temperatures in the surface and bulk of the glass.
Unfortunately, cooling with a temperature gradient in a glass creates temporary stresses that counteract the
permanent stresses because of the differences in fictive temperature [13].
Even though the stresses formed during cooling are not permanent, failure can happen due to the high stress
that occurs when the surface temperature and bulk temperature vary. The maximum amount of stress will
17
be generated if the surface is instantaneously cooled, while the bulk remains at the original temperature. In
these conditions, the stress is given by this equation [13]:
ET
 2.14
1   
where the difference between the surface and bulk temperatures is represented by ∆T and α is the materials
thermal expansion coefficient.
The best thermal shock resistance is for low expansion and low modulus glass. The maximum temperature
differential that can be used, without sample failure, will be very high for a very low thermal expansion
glass. [13].
2.11 Glass Composition

Throughout the years the knowledge developed on how any change in glass chemistry has a correspondent
change in each physical property. This has allowed to calculate an expected (or theoretical) property from
glass composition. Maintaining consistent glass chemistry will lead to maintaining consistent glass
properties. As each component oxide percentage changes in a glass composition, every physical properties
will be affected. Some components have large influence on some properties more than others. Container
glass is usually made with eight major oxide groups. Below is listed a container glass oxide composition.
Composition Oxide Term Raw Material Source(s)

72,4% SiO2 Silica Sand (Nepheline Syenite, Calumite)
1,8% Al2O3 Alumina Nepheline Syenite (Calumite, Feldspar)
0,05% Fe2O3 Iron (Calumite, Sand, Limestone)
14,3% Na2O Soda Soda Ash (Nepheline Syenite)
0,3% K2O Potassia Nepheline Syenite (Feldspar)
10,7% CaO Calcia Limestone (Calumite)
0,3% MgO Magnesia Limestone (Calumite, Dolomite)
0,18% SO3 SulphurTrioxide Salt Cake (sodium sulphate)
Table 2.2: Composition of typical glass containers
18
Oxide Type Function Example (container glass)

Build the glass structure. Form glasses when SiO2
Network-Forming
melted and cooled
Not capable of forming a glass, but it can take Al2O3 , ... ,
part in the glass network. In conjunction with
Intermediate
the glass former oxide builds the glass
structure
Not able to build up the glass network and Na2O
Network Modifier instead weakens the glass structure. Lower
the melting temperature.
Fining Agent Remove bubbles from glass melts. Na2SO4
Colouring FeO (green-blue), Fe2O3
Accessories (green-yellow), Co2O3
(blue), Cr2O3 (green)
Table 2.3: Oxides functions in glass containers.
Figure 2.5: Effects of some oxides in glass properties.
19
20
Chapter 3 Glass Container Production
CHAPTER
3
3 Glass Container Production
On a larger scale glass containers are formed using different processes, we shall focus on blow/blow and
the press/blow processes. All processes start with batching and melting, after conditioning the materials to
the desired temperature, subsequently the hot gob is transferred from the furnace to the forming operations
through the gob feeders. The gob feeder controls the weight, temperature and shape of the gob, which are
critical to the container quality. Many forming sections are distributed by only one gob feeder. The gob
feeder comprises a plunger that drives the glass melt to the aperture and shears that cuts the gob for the
desired volume. The cutting of the gob is helped by the cooling of the glass melt resulting from the shear
contact, increasing the viscosity, while the backward movement of the plunger induces the upward
rearrangement of the glass melt. The formed gob then enters the forming operations. After the forming
operation, the containers go to a convey, annealing, coating, inspection and palletizing operations [1],
Figure 3.1.
21
Figure 3.1: Production line of glass containers [12].
3.1 Batch
The batch is a key element in the glass melting process. The quality of glass is very much affected by non-
homogeneous mixtures and inaccurate administration systems. A glass batch plant needs the capacity to
fulfil the three main functions, as raw material storage, batch, preparation and batch transport [17]. Some
concerns can influence the quality of the batch such as:
o Unclean trucks with foreign residues can introduce contamination;

o Variations in the chemical composition of raw materials: high iron content in sand (and other
raw materials) imparts an undesired green tint to glass;
o Contamination such as metal and ceramic in cullet will cause defects such as “stones” in finished
glass bottle;
o Weighting errors;
o Overmixing of the batch can lead to segregation of the batch materials;
3.1.1 Batch – Storing

Raw materials reach the glass plant batch house by rail or truck, they are visually inspected and sampled to
ensure they fulfil the specifications, Figure 3.2a). Depending on the needed quantity, raw materials can be
supplied in bulk, sacks or drums. The silos are filled pneumatically or mechanically, Figure 3.2b) [17].
22
Figure 3.2: a) Glass plant batch house [12]. b) Bach house of a glass container line [12].
3.1.2 Batch – Weighting

Before going into the furnace, materials are proportioned into batches and weighed on scales located under
the storage silos, Figure 3.3. There is usually one scale for each raw material and each batch must be
accurate. Scales can be electronic or hybrid. In both systems, loading cells measure the weight of the raw
material. To ensure its accuracy the scales should be checked periodically for sensitivity and calibration.
Figure 3.3: Weighting the raw material [12].
3.1.3 Batch – Mixing

After being weighed, materials are sent to the mixer, Figure 3.4. When all raw materials are inside the mixer
the dry mixing begins. After dry mixing, water can be added and the wet mixing starts [17]. This is done to
decrease the possibility of dusting and segregation as well as to supress demixing during transport. A wet
batch enhances the batch control pattern in the furnace, which is important for an efficient melting.
Normally, after mixing, cullet is added on the top of the batch [17]. This is done to minimize wear and tear
23
of the mixer. The cullet may be added continuously or discontinuously to the mixture [17]. The mixed batch
is then carried to the furnace by a conveyor.
Figure 3.4: Mixing the raw material operation [12].
3.2 Melting
In an industry of glass containers, the products are produced from melting raw materials and recycled glass
in tank furnaces at high temperatures. Furnaces are made of three main parts, the melter, refiner and
regenerators or checkers. Most furnaces are designed to use natural gas, they are capable of using alternate
fuels like oil, propane and electricity, if necessary [18].
The melter is a rectangular basin in which the actual melting and fining (seed removal) takes place, Figure
3.5. In a side fired furnace, the batch is fed into the furnace through the doghouse, an extension of the
melter, protruding from the back wall. Along the sides of the melter, above glass level, are three to seven
ports, containing natural gas burners that direct the combustion air and exhaust gases [18]. A combination
of electrical heating and fuel is used to improve heating uniformity and melt efficiency, it also reduces gas
consumption and emission. The melter basin is connected with the refiner by the bridge wall (throat end
wall). Glass passes from the melter to the refiner through the throat, a water or air-cooled tunnel that extends
through the bridge wall [18].
24
Figure 3.5: Melting plant scheme with the temperature gradient [12].
3.2.1 Regenerative Furnace

There are some different types of modern fossil fired furnaces, the heat contained in the released waste
gases leaving the furnace is used to preheat the combustion air to produce higher flame temperatures and
improve efficiency [19].
Regenerators form an intermediate storage medium, consisting of two chambers, each one is filled with a
network of refractories, named the packing [19]. The waste gases from the furnace go through one of the
chambers, heating up the refractories in the chamber. The combustion air goes through the other chamber
into the furnace, Figure 3.6 [19]. Sometime after, both flows of air and waste gases are reversed. The
combustion air now flows through the hot chamber and heats up by heat transfer from the refractories,
whilst the waste gases pass through the other chamber and heat the refractories in this chamber again.
Regenerator chambers are normally constructed vertically making the waste gases pass downwards, while
the combustion air travels upwards [19].
25
Figure 3.6: Regenerative furnace layout [12].
Regenerator chambers are normally constructed vertically making the waste gases pass downwards, while
the combustion air travels upwards. Regenerative furnaces can be divided into two basic types based on the
location of the burner and the flame path which is regenerative furnaces side port (cross fired) and end port
(end fired) [19].
• End fired furnaces;
End fired regenerative furnaces has two burner ports, side by side in the rear wall furnace, and the
regenerators are located behind the furnace. The flame travels forward from the burner port, does a 180°
turn and exits through the second burner port. Making the flame and waste gas path take the shape of a
horizontal “U”. Results is that the combustion gases, in the furnace, have a relatively long residence time,
26
achieving a good energy utilization. The raw materials go through one or two doghouses by the sides of the
furnace, located next to the rear wall [19], Figure 3.7. This type of furnaces have the following advantages:
o Very flexible furnace type;

o Lower construction costs than with cross fired furnaces;
o Lower energy consumption than a cross fired furnace,
Figure 3.7: Regenerative end fired furnace [19].
• Cross fired furnaces;
Cross fired regenerative furnaces have burner ports located along the furnace side walls, normally covering
most of its length. The amount of ports depends on the size of the furnace and usually within the range of
three to five. Each port has from two to four burners, depending on the size of the furnace. The flame travels
in one side to the other of the furnace and the waste gases are exhausted in the opposing entry burner port.
The maximum flame size available is determined by the furnace width. With this type of furnace the
regenerator chamber supplies many burner ports [19], Figure 3.8.
27
Figure 3.8: Regenerative furnace heat exchange [19].
The single doghouse is situated on the furnaces rear wall and the batch usually covers almost the complete
tank width. As a result of the bigger number of ports and larger regenerator chambers, the heat loss area is
greater than with comparable end fired furnaces [19].
3.2.2 Recuperative Furnace

In recuperative furnace the hot waste gases as well as the cold combustion air pass through parallel but
separate channels and heat transfer takes place through the intermediate wall, Figure 3.9. Recuperatively
preheated combustion air provides continuous heating without the flame/waste gas path reversal that is
necessary with the regenerative systems. Lower investment than regenerators. Load energy usage increases
more slowly than with regenerative furnaces. However, the air preheat temperatures are lower than with
regenerators [19].
28
Figure 3.9: Recuperative furnace [19].
Most recuperative furnaces use steel recuperators to melt glass. They are always installed vertically, where
the waste gas flows either upwards or downwards. The basic types of steel recuperator used are [19]:
o Double shell recuperator;

o Tube cage recuperator;
Double shell recuperator is made of two concentric high temperature resistant steel tubes of similar
diameter, so that a narrow annular slit is formed in the centre of the two tubes. The hot waste gases pass
through the inner tube, while the combustion air goes through the annular slit. The air may pass in the same
basic direction as the waste gases (parallel flow) or go in the opposite direction (counter flow). Double shell
recuperators can give a typical air preheat temperature with the range of 450 – 650 °C. The bulk of these
units are used for small furnaces up to a melting capacity of approximately 50 t/24 h [19].
In a tube cage recuperator the combustion air is driven through a large number of individual small diameter
steel tubes arranged in a ring around the inner circumference of a large diameter outer tube, through where
the waste gases flow. The outer tube is made of steel, coated with refractory material [19].
Conventional recuperative furnaces can be branched into two types based on the location of the burner and
the flame path [19]:
o End fired furnaces;

o Side fired furnaces;
There are many installation arrangements for burners and the recuperator in end fired recuperative furnaces.
Similar to regenerative furnace, the burners are on the furnace rear wall, with the waste gas port immediately
above, in the same wall. The flame leaves the burner and rides along the furnace, turns upwards and back
to exhaust right above the burners. The flame path creates the shape of a vertical “U”. The single doghouse
is on a side wall, next to the rear wall. This furnace concept is mainly used for small installations, with
melting capacities of up to approximately 35 t/24 h [19].
29
The side fired burners of the recuperative furnaces are equipped along both side walls. The waste gases are
exhausted towards the rear. Either one or two exhaust ports are outfitted in the furnace rear wall or a side
wall. The furnace can be designed with a single or a double doghouse, either on the rear wall of the tank or
immediately adjacent to the rear wall. Controlling the temperature of larger furnaces requires the use of
control zones along the length of the furnace, Figure 3.10 [19].
Figure 3.10: Side fired recuperative furnaces [19].
The needed energy consumption of side fired recuperative furnaces at full load is higher than that of
comparable regenerative furnaces, yet is lower with partial load, Figure 3.11 [19].
Figure 3.11: Comparison of energy consumption of a regenerative and a recuperative furnace at part load [19].
3.2.3 Boosting
The use of electrical energy as an additional energy source in conventional fossil-fuel fired furnaces is
named as electric boosting [19]. Molybdenum electrodes can be equipped horizontally through the tank
side walls, or vertically through the furnace bottom, Figure 3.12. A mixture of these methods is also
possible. Under some circumstances an electric booster can be equipped in an operating furnace: in some
furnaces, they make arrangements for a later installation during construction. Boosting systems can be
branched into three groups according to their usage [19].
30
Figure 3.12: Boosting system in an end fired furnace [19].
Melting booster - a boosting system in the melting area gives additional energy directly to the glass bath
and this leads to a higher melting capacity, Figure 3.13. a) [19].
Barrier booster - The installation of an electric booster around the hot spot boosts the convection currents
and this increases the bottom temperature of the glass bath. This has a positive effect on the glass quality
[19]. Furthermore, additional energy is given to the glass bath and the temperature of the glass flowing
from the hot spot to the melting area is boosted. This may lead to a small increase in the melting capacity,
Figure 3.13. b) [19].
Local booster - Here the electrodes are equipped in order to heat a precise area of the glass bath. The
majority of these boosters are equipped in the throat and riser, where the glass can cool down more when
the melting rate is low, Figure 3.13. c) [19].
a) b)
c)
Figure 3.13: a) Melting booster b) Barrier booster c) Local booster [19].
31
3.2.4 Bubbler
Air or other gases are blown through special bubbler nozzles equipped in the furnace bottom. This makes
bubbles in the glass, bubbles then rise to the surface and the gas is exhausted into the furnace atmosphere.
The upward movement of the bubbles makes strong localized convection currents around their path, and
these currents move the bottom glass upwards with an increase glass temperature at the bottom of the tank.
The bursting bubbles on the glass surface also creates an effective barrier that prevents the unmelting batch
from moving forwards on the glass bath surface. Generally the bubbler tubes are equipped in a row over
the complete furnace at the location of the hot spot to reinforce the convection currents at this place [19].
The creation of this amount of gas, however, also leads to the formation of an extremely large number of
bubbles, which must be removed from the melt before processing is finished because bubbles in commercial
products are almost always considered flaws and result in rejection of the product [3].
Bubbles can be eliminated from melts either by physically rising to the surface, or by chemical dissolution
of the gas into the surrounding melt. Since the density of a bubble is less than that of the surrounding liquid,
a bubble will automatically surface and burst unless prevented from doing so by some external agent. If
bubbles do not rise fast enough in a quiescent melt, the fluid itself can sometimes be moved by convection
or stirred in such a manner that the bubbles are carried to the surface. Upward fluid motion can be done by
mechanical stirring, by design of a glass tank floor to produce upward currents, by localized heating to
make a locally hotter and thus less dense region in the melt, or by bubbling with a gas introduced near the
bottom of the melt. In practice, all of these techniques were used to varying degrees in either tank or crucible
melting [3].
Bubbles can be built by physically trapping atmospheric gases during the initial phase of batch melting, or
by the decomposition of batch components. The gases in the apertures between batch particles may be
cornered as the particles begin to soften and form a viscous liquid. As the viscosity decreases with
increasing temperature, these apertures become fully surrounded by liquid. Surface forces then cause these
apertures to assume the spherical shape of bubbles. These bubbles contain gases characteristic of the
melting atmosphere, which can be air, combustion gases or some gas deliberately introduced to control
chemical reactions with the batch. Prevention the formation of these bubbles involves eradication of gas
from the apertures of the batch, i.e., melting under vacuum [3].
Bubbles can be created by the following causes;
- Trapping of atmospheric gases is enhanced by the use of very fine sand in the batch or by the use of batch
components of broadly differing particle sizes. Both factors create a large concentration of very fine
apertures within the unmelted batch. Agglomeration of particles is particularly effective in trapping
atmospheric gases inside the agglomerate, becoming effectively surrounded by the viscous melt. Stirring
mechanically a melt can introduce bubbles by trapping air and forcing it into the melt [3].
-Decomposition of batch materials can produce awfully large quantities of gases like CO, SO3, NOx, H20,
etc. Reactions with metals making contact with the melt can generate oxygen, carbon dioxide, or hydrogen
by electrolytic reactions. Erosion of refractories can open previously closed pores to the melt, releasing the
gas contained in those pores into the melt. Residual carbon in refractories, or carbide refractories such as
SiC, can react with oxide melts to create CO, or CO2. The products of these reactions can agglomerate to
form bubbles [3].
-Bubbles can also be created by precipitation from the melt whenever super saturation occurs for a specific
gas. Since many gases have a large enthalpy of solution in glass forming melts, their solubility is a strong
function of temperature. Species which alter their chemical form with temperature or changes in melt
composition are particularly affected by precipitation from melts where they were previously soluble.
Carbon dioxide, for example, is present in silica-rich melts as C02 molecules, where it chemically reacts
with alkali-rich melts to form carbonate species, which are far more soluble than the molecular species.
The solubility of carbon dioxide in the alkali-rich regions neighbouring silica grains during the early stages
of batch melting is quite high. As the silica grains dissolve, the melt becomes locally enhanced in silica,
which converts the dissolved species to CO, molecules, which have a much lower solubility. The melt
32
becomes supersaturated locally and CO, bubbles are created. As a result, CO, bubble generation will keep
happening long after the initial decomposition of the carbonate batch components [3].
Figure 3.14: Bubbling principle [19].
3.2.5 Barrier Wall

In conventional glass furnaces the convection currents in the furnace have a compelling influence on the
melting capacity and glass quality. An upward current is created at the hot spot as a result of the temperature
distribution in the furnace. This current stops the forward flow of the colder bottom current and diverts it
upwards to areas of higher temperatures. This avoids lower quality bottom glass from flowing directly into
the throat. Nowadays this effect is aggravated by a barrier that is anchored in the bottom and projects
upwards. This barrier wall is put up from one side of the furnace to the other at the location of the hot spot.
The height of the barrier banks depends on the glass bath depth, Figure 3.15 [19].
Figure 3.15: Barrier wall [19].
3.2.6 Refining Bank

The Refiner performs as a holding basin where the glass is allowed to cool to a uniform temperature before
entering the forehearths. The melter and refiner are enclosed by crowns to contain the heat [18].
33
Successfully refining glass is largely dependent on two factors: time and temperature, where temperature
has the most influence. The refining bank is an additional device in the tank, devised to raise the glass
temperature in the refining zone, without increasing the superstructure temperature [19].
The bottom of the tank is raised over a precise distance, to give a shallower glass depth. The glass going to
the throat is forced upwards into the hotter area near the glass bath surface. This fulfils one of the
prerequisites for enhanced refining at higher temperatures. Higher glass temperatures in the refining zone
gives lower glass viscosity. Thus, the refining gases reach the glass bath surface more easily. The shallower
glass bath means that the current distance to the surface is shorter [19].
Glass fining produces a melt that is homogeneous in terms of composition, temperature and bubble free. In
fact, the temperature of the glass delivered from the melting furnace, and its composition, has to be
regulated very carefully. This requirement affects the next operation, that is, glass forming, the latter
operation is reliant on viscosity, that is, on the way the glass has been prepared [1].
3.2.7 Drain
Zircon cords in finished articles are the result of deterioration of the refractory material that is used in large
quantities in modern glass furnaces. Zircon has a somewhat higher density than normal soda-lime glass.
Consequently, zircon-rich material created because of refractory corrosion tends to gather on the furnace
bottom. The drain is designed to allow the continuous draining of very small quantities of material from
the furnace, working end or forehearth bottom. The low draining velocity prevents glass from higher
regions to be pulled down as a result of the funnel effect and helps to ensure that that contaminated material
is taken out effectively from the furnace bottom level, Figure 3.16 [19].
Figure 3.16: Drain cross section [19].
3.2.8 Cullet Preheating

When the cullet composition in the batch amounts to more than 60%, the use of cullet preheating to recover
additional heat from the furnace waste gases is an economically feasible proposition. Cullet enters the tower
structure at the top and slowly makes its way downwards to the exit. The waste gases arrive at the bottom
on one side and are exhausted at the top on the other side, and so the cullet and the gases are in cross counter
flow. The waste gases enter the preheater at a temperature of approximately 500°C and leave with
approximately 200 °C. The cullet preheat temperature is normally approximately 400°C [19].
34
3.2.9 Melting Process and glass formation

Most commercial glasses are based upon chemical modifications made to melt and modify the structure of
silica. Crystalline silica melts above 1700ºC. By adding fluxing ingredients such as alkali (Na2O, K2O),
the melting point drops significantly. To improve chemical durability and to obtain other physical properties
other Alkaline Earth’s (CaO, MgO) and other modifiers (Al2O3) are added. The resultant material is
converted from a crystalline structure to a glass.
Glass formation involves a number of key steps – starting by properly formulating specific raw materials
to contribute required oxides. The properly prepared batch ingredients must be kept in close proximity as
they are heated. A serial of chemical reactions initiate some components melting and performing
conversions of others into new, intermediate compounds. Large volumes of gases are evolved during the
thermal decomposition of carbonates, hydrates and sulphates. Eventually, the final glass chemistry is
reached when all raw materials have reacted and intermediate compounds combined into final product.
Chemical finning agents are materials intentionally added to the batch which promote the evolution of gases
generated during the fusion process at elevated temperatures. Some processes make the bubbles larger in
size, which can rise more quickly out of the melt. Others promote more physical agitation of the less soluble
types of gases.
Thermal heat transfer for melting begins with a combusting process generating a flame. Radiation from the
flame is directed to the melting batch and molten glass bath. It can also go to the refractory structure for
reradiation into the melting system.
The heat transfer to the melting batch and freshly formed glass requiring thermal refining is controlled by
flame place location and combustion control. It is difficult to measure temperatures within the melting batch
for controlling that aspect of the process. Most operations establish temperature profiles within the furnace
by a variety of sensors in the crown, bottom, molten glass or melter superstructure surface readings by
portable optical pyrometers. By controlling these reference readings, the resultant glass quality helps
establish control parameters for consistent melting and refining results.
Most furnace operations rely upon establishing and maintaining a hot spot. In this system of furnace the
temperature gradient (or profile) relies upon the principle that hotter glass expands and flows on the surface
towards colder areas. The surface flow can help contain the batch piles behind the hot spot and prevent
partially melted glass of leaving the melting zone, Figure 3.17.
Figure 3.17: Furnace temperature evolution and the respective convection.
35
A convection pattern is established when glass lower in the melt is drawn upward to replace the glass
flowing on the surface away from the hot spot, Figure 3.18. The batch charging establishes the pile shape
and initial direction of flow into the melter. From that point on the operating temperatures and glass surface
convection flow will determine the path the batch follows until is converted to bottles, Figure 3.19.
Figure 3.18: Flux of glass inside a furnace.
Figure 3.19: Woking end furnace.
3.3 Glass Conditioning

Glass conditioning is a thermal process that can be defined as the achievement of a stable, desired glass
temperature, distributed throughout the vertical and horizontal planes of the glass at the entrance of the
spout, Figure 3.20.
36
Figure 3.20: Glass conditioning general view.
3.3.1 Forehearth
The forehearth plays no role in chemical conditioning or refining of the glass nor does it leads to the removal
of bubble or seeds from the glass. These functions are carried out only in the furnace [20]. Forehearth
simply don’t have the subsystems to fix furnace faults. Consequently, furnace derived problems will be
given by the forehearth to all subsequent stages in the manufacturing process [20].
However it is expected that it can mitigate the effects of:
o Upsets in the melter or distributor;

o Fluctuations in air temperature between day & night;
o Fluctuations in humidity;
o Fluctuations in pull rate;
o Fluctuations in glass level either from the melter or from head loss on adjacent forehearths.
Forehearth requirements:
o Temperature stability;
o Temperature homogeneity;
o Glass flow (along the forehearth and trough bends – the forehearth is slightly tilted downward);
o Fast response time (at job change and in response to external disturbances);
o Low energy consumption.
If the forehearth fails to produce its conditioning objectives, the ensuing problems will be fed to the
subsequent forming process. Approximately 100 types of errors have been identified, around 50% may be
related to the temperature and thermal homogeneity of the glass, and by implication, to the effectiveness of
the thermal conditioning properties of the forehearth design [20].
The ability of the forehearth to get thermal stability rapidly after a job change is another key element in the
efficiency of the overall forming process. The time needed to stabilize the glass after a job change is a part
of the response time of the forehearth and is determined by the efficiency of both combusting and cooling
systems, as well as the effectiveness of the superstructure geometry [20].
The homogeneity of the melted glass in the forehearth needs to be monitored by measuring the temperature
in three different points at three different levels in each forehearth zone (typically: rear, middle, front) – 9
points grid, Figure 3.21.
37
Figure 3.21: Glass Conditioning heat transfer [12].
There are some types of different forehearth which include forehearth type K, forehearth longitudinally
cooled, radiation cooled forehearths, muffle cooled forehearths and system 500 forehearths. Each
forehearths will be described as follows:
o Forehearth type K
Cooling is applied through passages in the side of the superstructure, passing under the forehearth roof and
exiting through the central flue with the combustion gases, Figure 3.22.
Figure 3.22: Forehearth Type K [12].
Three problem can be founded in forehearth type K which are;
1st problem: Heat is taken out of the sides of the glass flow by radiation to the colder surface of the roof
block. Effect: This causes the sides of the glass flow to be cooler than the centre, Figure 3.23 a)
2nd problem: The cooling air and combustion gases exhaust through the same single flue. Effect: It is not
possible to separate the cooling functions from the heating functions within the forehearth, Figure 3.23b)
3rd problem: There is insufficient insulation applied to the substructure. Effect: There are substantial
uncontrolled heat losses through the forehearth sides and base, Figure 3.23 c).
38
a) b) c)
Figure 3.23: Forehearth Type K [12].
o Forehearth longitudinally cooled
Cooling is applied longitudinally under the forehearth roof, cooling the underside of the roof block. Heat is
taken out of the glass by radiation to the colder refractory surface, Figure 3.24. Three problem can be
founded in Forehearth longitudinally cooled which are:
1st problem: With only a single flue for both the combustion and cooling, the cooling air and combustion
gases mix together. This effect prevents separate control of the heating and cooling functions within the
forehearth.
2nd problem: The cooling air is quickly spread across the entire forehearth width causing the cooling effect
is insufficiently concentrated on the central flow of glass.
3rd problem: The cooling zones are too long and the cooling effect is spent soon after entering the
forehearth. The effect for a significant part of each zone the cooling air is ineffective, limiting forehearth
response time and flexibility.
Figure 3.24: Forehearth longitudinally cooled [12].
39
o Radiation Cooled forehearths
The method of operation contain large flues are left in the forehearth superstructure enabling heat from the
glass to radiate to the colder atmosphere as it flows beneath a flue, Figure 3.25. Three problem can be
founded in radiation cooled forehearths which are;
1st problem: Allowing heat from the glass to radiate to atmosphere causes massive heat loss beneath the
flues. Effect: As the glass flows beneath a closed section of superstructure the glass reheats, causing cycling
in the glass temperature.
2nd problem: Radiation cooling is extremely severe. Effect: Close control of the glass temperature is
difficult to achieve and often results in overcooling, chilling of the glass surface, and prevention of further
heat removal.
3rd problem: The presence of large open flues prevents control of the internal forehearth pressure. Effect:
Fuel efficiency is very poor.
Figure 3.25: Radiation cooled forehearths [12].
o Muffle cooled forehearths
Cooling wind is passed along a muffle in the forehearth superstructure, cooling conductive tiles set in the
forehearth roof, enabling the glass to cool by radiation to the tile’s cooler surface, Figure 3.26.
The problem associated to this type of forehearth is the heat from the glass has to be conducted through
the muffle tile. Effect: Cooling capacity is restricted and response time is slow. Forehearths need to be
longer to compensate for the lack of cooling.
Figure 3.26: Muffle cooled forehearths [12].
o System 500 forehearths
The main features of this type of forehearth include separate combustion and cooling dampers, fully
automatically controlled, cooling subzones for greater cooling capacity, hot face longitudinal cooling, a
single piece roof block profiled to separate the combustion from the cooling, Figure 3.27.
40
Figure 3.27: System 500 forehearths [12].
Combustion takes place at the sides of the glass flow and exhausts through dedicated side exhaust dampers.
Cooling air passes longitudinally down the forehearth under the roof blocks exhausting though dedicated
central exhaust dampers, Figure 3.28a).
The cooling air is automatically terminated, the side dampers closed, and all the products of combustion
exhaust through the central flue, heating up the entire forehearth width, Figure 3.28b).
In normal operation the forehearth operates automatically between the heating and cooling modes, adjusting
the firing rate and the flow of cooling air to obtain the required glass temperature and optimum thermal
homogeneity, Figure 3.28 c).
a) b) c)
Figure 3.28: System 500 forehearths a) cooling cycle b) Heating cycle c) Normal operation [12].
In conventional longitudinally cooled forehearths the cooling zones are the same length as the combustion
zones approximately 2500-3000mm long. However, with such long cooling zones the cooling effect from
the cooling air is dissipated after travelling 50% of the zone length, Figure 3.29 and Figure 3.30.
41
Figure 3.29: System 500 forehearths cooling zones [12].
Control zones are subzoned into multiple cooling ‘subzones’, significantly increasing the cooling capacity
and flexibility of the forehearth and quicken the response time.
Figure 3.30: System 500 forehearths cooling zones [12].
3.4 Forming
In this subsection the process of the forming of glass containers will be described.
3.4.1 Feeder
The molten glass goes from the melting furnace into the thermal conditioning forehearth. Once the right
thermal homogeneity and viscosity are achieved, the feeding mechanism cuts the string of liquid glass into
gobs of required size, shape and weight, Figure 3.31. The red hot gobs of glass (1.200° C approximately)
vertically drop into the delivery system and then distributes them to the various sections of the I.S. Machine.
42
Figure 3.31: Forming feed machine [12].
The spout bowl is a refractory part in the end of the forehearth channel. It has a circular hole at the bottom
where fits the orifice ring. A cylindrical tube overlies the spout bowl and a selected clearance can be
established between the bottom of the tube and the surface of the spout bowl around the opening to control
glass flow from the spout bowl, Figure 3.33.
Feeder tubes are used in the spout section to homogenize the glass by rotation, and to allow heating or
cooling through the vent holes if present. The tube is mainly used to control the flow of glass through to
the orifice ring below. The feeder tube is also used to shut off the flow of glass if necessary by lowering
onto the surface of the spout. These tubes have a collar on the top to fit the mechanism used for rotation.
There is a scale which gives us the height of the tube – 0 position being at the bottom, with the tube shutting
off the glass flow. The tube is often viewed mostly as a tool for controlling gob weight, but the height of
the tube, in conjunction with the plunger, is also an important factor in gob shaping. There is an optimal
tube height for every orifice ring size. For instance, with a small gob of 14 to 28 grams in weight, the area
of the orifice is very small making the tube height very low and critical. If the tube is too high, that will
make the gob to lose action and gain very little weight. Glass runs from the orifice unencumbered. If the
tube is too low it will cause the orifice to run empty, resulting in loss of action and weight, plus a long and
stringy gob shape [21].
As the tube height is raised, more glass is permitted to flow under it. Also the height of glass below the tube
is increased. This in turn increases the effectiveness of the plunger action within the glass due to more glass
adhering to the plunger, which consequently is pushed out of the orifice. Resulting the gob becomes heavier,
longer and begins to have a ‘waist’ in the middle, Figure 3.32a). As the tube is lowered, the amount of glass
beneath the tube is reduced as is the height of glass below the tube. This also decreases the effectiveness of
the plunger action and less glass is pushed through the orifice. Resulting the gob be lighter, shorter and
more compact in shape, Figure 3.32b).
43
a) b)
Figure 3.32: a) Tube height raised. b) Tube height lowered [21].
The function of the orifice ring is to control the diameter of the molten glass gob, and the number of gobs
required for the particular glass operations. The orifice ring is typically round and contains the appropriate
number of gob holes, normally from one to four holes centrally located in the orifice ring. The orifice rings
are installed in the bottom of the spout. The molten glass is forced through the holes in the orifice ring by
means of mechanical force from the plunger(s) located above the orifice ring, Figure 3.33c).
a)
b) c)
Figure 3.33: Examples of a) Spout bowl b) Feeder tubes c) Orifice ring [12].
The orifice diameter is usually determined by feeder manual charts and previous experience and operation
of similar situations (same weight and gob shape). When making a gob, if unable to get enough weight and
the gob is short and stubby, this usually indicates that the orifice is too small, Figure 3.34 a). If the weight
cannot be reduced to what is required with the tube and plunger on their lowest point, this indicates that the
orifice is too large, Figure 3.34b) [21]. If unable to get a low weighted gob and the tube and plunger are
low, the change to a smaller orifice ring is the most indicated action. In each case the requirement is to
change the orifice ring (assuming that the glass temperature and machine speed are correct and appropriate
for the situation).
44
a) b)
Figure 3.34: a) Small Orifice Ring. b) Large Orifice Ring [21].
Plungers are available in different shapes, diameter and lengths, depending on the type of feeders and
process being used, i.e. size of refractory tube, single, double or triple gob. Plunger tip profile and diameter
are very important to the gob weight and shape control. Taper point (standard, long and offset) and ball
point plungers are the ones often used. The tapered plunger will form a tapered gob while the ball point
will make a straight cylinder type gob, Figure 3.35 [21].
Glass feeder plungers operate in coordination with shears and other components in order to control the
manner in which molten glass is delivered to glassware forming equipment downstream of the feeder. A
glass feeder may have one or more plungers, each axially lined up above an associated orifice in the bottom
of the feeder spout bowl through which emanates a stream of molten glass. Each plunger (and orifice) has
implemented with it a pair of shears, and all plungers and shears are synchronized with the glassware
forming machine and operate in cycles in order to cut the respective stream into predetermined gobs of
molten glass [22].
Figure 3.35: Glass feeder plungers [12].
Each plunger controls the speed with which the glass emanates from its associated orifice. Therefore, the
cyclical motion of each plunger has to be controlled in a predetermined manner in order to achieve desired
gob shapes and weights [22]
Plunger motion profiles are dependent of a lot of variables including glass temperature and type, machine
speed, the type and size of glassware to be produced, etc. As a result, even with job changes, minimal
changes of plunger movement are desirable to keep an optimum machine operation and speed. Furthermore
45
the motion profile of each plunger, adjustments of the plunger differential (i.e. phase relationship of plunger
cam to shear cut), stroke length and height (i.e. defined as the lowest point of the plunger tip above orifice)
also affect gob formation [22], Figure 3.36.
Figure 3.36: Representation of plunger motion and gob shear cutting [1].
The plunger extent is important. It not only controls, in alliance with the tube, the gob weight, but also is
key in determining the ultimate gob shape. When the plunger is lowered, there is a reduction in gob weight,
but it also increases the action inside the gob. When raised the gob weight increases, but reduces the action
in the gob, less control over the glass flow. There is also an optimal plunger height in relation to tube height.
There is a scale that shows the position of plunger height above the orifice ring, Figure 3.37 [21].
Figure 3.37: Effect on gob with different plungers [21].
As the plunger height is lowered, the low point of the plunger is adjacent to the orifice. The plunger is also
farther into the glass inside the tube giving more grip to the glass and restricting the flow, Figure 3.38a).
As the plunger is raised, the low point of the plunger rises farther above the orifice ring and out of the height
of glass inside the tube, Figure 3.38b). This reduces the effect of the plunger action within the glass and
46
consequently control over the glass flow through the orifice ring is moderately lost allowing the glass to
flow more freely [21].
a) b)
Figure 3.38: a) Plunger height lowered. b) Plunger height raised [21].
Gob cutting shears are employed to systematically sever the lower most portion of the vertical glass flow
that comes out of each orifice ring, into a discrete glass gob of fixed volume, Figure 3.39. After the gob has
been severed it then falls or slides down a suitable feed chute into a glass forming mould. The cutting of
gobs of glass is a crucial operation since minor defects or irregularities caused as the gob is cut may result
in a defectively formed glass article [22].
Figure 3.39: Gob cutting shears [12].
The shear height is changeable. The integral shear mechanism is either moved up or down, close to the
orifice or further away from the orifice. Shear height is very useful to get the final frame of the gob. Weight,
temperature differential may be correct but shears height too high or too low can blemish the gobs shape.
Raising the shears closer to the orifice will shorten the gob, lowering the shears farther form the orifice will
lengthen the gob. Both adjustments control shape of the gob at its lower end. Shear tension is very
important. Tension is adjusted by the raising or lowering of one shear arm where the blades are tensioned
one against the other. Care should be taken when this operation is carried out as too much tension creates
excessive abrasion on the blades and the shearing edge is lost very quickly, resulting in heavy shear marks.
Too little tension on the other hand can create feathering of the cut, again creating a bad shear mark or
occasional stringing of the gob, Figure 3.40 [21].
47
Figure 3.40: Shear tension effect on gob shearing [21].
Shear overlapping is controlled by adjusting one shear arm stop. This, through gear sections on each shear
arm, can be used to alter the amount of crossover of each blade with the other. At this point the gob has
been sheared. Excessive overlap can interfere with gob loading and unnecessary adjustments have to be
made to overcome this. It is important that after shearing, the gobs leave the blades clearly vertical without
inclination. Drop guides are generally positioned and fixed on the right hand shear arm and are used to
adjust the gob angle at the shear cut. The adjustment is made side to side and front to rear directions, Figure
3.41.
Figure 3.41: Shear overlap effect on gob shearing [21].
The difference in timing action of Plunger Action/Shear Cut is applied to gob shaping. This is done via the
PLC timing system, accelerating or decelerating the plunger action. If the plunger is at its lowest point at
the same time the shears cut off, glass will lay on the shears, withdraw a large heavy shear mark and flatten
the gob. If the plunger action is advanced starting the upstroke lifting the glass before the shears cut off this
48
gives a small shear scar and leaves a point on the gob that will elongate into a good shaped gob, Figure 3.42
[21].
a) b) c)
Figure 3.42: Plunger/shear differential adjustments a) Differential 0º setting b) Differential 10º setting c)
Differential 30º setting. [21].
As the glass temperature rises, the glass becomes more fluid. More glass will flow under the tube and the
plunger action will be less effective on holding the gob together. The glass will move more freely through
the orifice, Figure 3.43a). As the glass temperature declines, the glass becomes less fluid. Less glass will
move under the tube and consequently the gob will lose weight. Since it is colder, the plunger will have
more grip on the glass during its upstroke, restraining the flow from the orifice, Figure 3.43b) [21].
As the machine speed is increased, the action of the plunger and shears also increases. Due to the increased
motion of the feeder assuming that the glass temperature and all other things are the same, less glass will
be out of the orifices for each gob, Figure 3.43c). As the machine speed is decreased, the action of the
plunger and shears is lesser due to the decrease in feeder motion. All other things being equal, more glass
will come out for each gob, Figure 3.43d) [21].
a) b)
c) d)
Figure 3.43: Different types of tuning a feeder a) Glass temperature increased. b) Glass temperature decreased. c) Machine
speed increased. d) Machine speed decreased. [20]
49
The objective when shaping a glass gob is to deliver a gob of ideal shape and. This shape must be consistent
and reproducible. This is one aspect which, in the past, empirical experience and practice were the best
teachers. This still applies, in part, today. But with increasing machine speeds, a much broader
understanding relating to controlling more precisely, gob shaping and loading techniques has come about.
It is true to say that the best gob to use for the majority of the jobs, is the shortest most compact gob it is
possible to load consistently into the blank mould, Figure 3.44
Figure 3.44: Gob shape parts affected by feeder controls [21].
The gob distributor and delivery system on an IS machine gets the gobs after the shear cut, for delivery to
the individual sections of the machine. The gob distributor moves the scoop with a high speed and accuracy
to the entryway of the troughs in accordance with the firing order (the order in which each section receives
gobs from the gob distributor). The firing order is calculated to reduce the scoop travelling distance keeping
the timings and movements synchronized.
In the delivery of various gobs of molten glass, simultaneously, to one of a plurality of stationary glass
moulding machine sections positioned beneath the gob former, it has been a constant goal that the various
gobs arrive at the mould of a section at the same precise time so that the gobs will all undergo the same
degree of cooling while being transformed into a parison. This goal is an obviously difficult one to attain
because the gobs may be travelling slightly different distances and due to manufacturing differences
between delivery system geometry, Figure 3.45.
50
Figure 3.45: Gob distributor scheme [12].
When considering the problem of achieving consistent and predictable loading characteristics between as
many as 10 or 12 sections which will each get the various gobs at timed intervals from a feeder that is
cutting gobs at a preset rate, it can be seen that there are huge timing problems involved. In order to avoid
inconsistent loading of gobs, various systems have evolved, but most have culminated in what would best
be termed an accommodation between the practical and the expensive systems [23], Figure 3.46.
Studies of gob delivery systems in operation with high speed cameras expose many unforeseen situations
that can affect the actual time of travel of an individual gob from the point of detachment from the stream
at the feeder to the actual channel of the gob into the upper end of the parison mould cavity [23].
The fact remains that gob delivery to each cavity on a machine is occasionally very inconsistent and that
the inconsistency is random in nature. A given cavity may be satisfactory, and then for no foreseen reason,
it becomes bad. One factor that will contribute to an inconsistent delivery is mediocre maintenance of the
equipment [23].
Figure 3.46: Gob distributor [12].
51
3.4.2 Blow/Blow
In this section the stages for the blow/blow process will be explained in detail.
 Gob Loading
The gob should be guided through the funnel into the blank mould with the minimum rubbing and with
approximately 0.8mm clearance on the diameter. This will give accurate loading with a minimum marking
of the gob and give sufficient vent to the blank mould to keep the undesirable formation of an air pocket to
a minimum, which would prevent solid loading. With good loading the gob settles uniformly and almost
entirely into the shape of the finish before settle blow air is applied. To ensure good loading, the clearance
between the plunger and thimble or plunger and finish guide plate should be as specified in the I.S Machine
mould, Figure 3.47a). Equipment must be kept clean to prevent dirt from building up in the vents.
 Settle Blow
First, in the settle blow the glass gob is blown from above to form the bottle neck. Settle blow should be
applied as soon as possible after loading to obtain uniform glass/metal contact at the earliest possible
moment. The settle blow time should be kept to a minimum, using only sufficient time to set up the bore of
the finish to withstand the counterblow. Large or heavy finishes require more time to set up the finish than
the smaller finishes. The minimum settle blow time will also give the minimum settle wave in the finished
bottle, Figure 3.47b).
 Corkage Reheat
As soon as the settle blow action is completed, the plunger should be withdrawn to allow corkage reheat,
or softening of the glass immediately above the plunger, to permit counterblow to form a symmetrical
bubble in the parison without distorting the neck of the bottle.
 Counter Blow
The earlier counterblow is applied, the less settle wave will be apparent in the final bottle. The longer the
counterblow is left on, the better the balance of heat removal between the blank and blow mould, permitting
maximum machine speed and hot mould practice, i.e. operating the blank mould at a maximum temperature
consistent with the production of a firm parison. The pressure used must be suited to the particular size of
the bottle, that is, the larger quantity of glass the greater pressure. In general, more stable operation, better
glass distribution and higher speed can be obtained by using maximum blank contact time, Figure 3.47c).
 Parison Reheat
After the counterblow the blank mould is opened and the parison is allowed to reheat before inverting into
the blow mould. This reduces the cold skin produced by the glass contact with the blank mould.
 Invert
During the time the parison is transferred from the blank to the blow mould, the parison is reheating. The
speed of invert affects the distribution of the finished bottle, if it is too slow, the parison will sag backward
due to gravity; if too fast, the parison is thrown forward by centrifugal force. The speed must be varied to
suit the weight, viscosity and shape of parison, Figure 3.47d).
 Rundown
The only force acting in the rundown stage is gravity. A cooling effect will appear when the glass contacts
the baffle. During this time, the parison continues its reheat or temperature equalization. Too much
reheating on the blank side allows the parison to sag and on the blow mould side to run, and the two effects
have to be counterbalanced.
52
 Final Blow and Vacuum
Vacuum is applied to the blow mould cavity to form the finished bottle, and the final blow air is also used
in helping to form the shape of the bottle. By circulating the air through the internal cooling blowhead arm
when the blowhead is on top of the mould, the internal temperature of the bottle is reduced. In general
terms, the vacuum and final blow time should be the maximum to form the bottle before it is removed from
the blow mould, Figure 3.47e). Blowing is an important stage in production of glass containers. In this stage
the shape of a bottle is reached.
 Mould open and takeout
After the blow mould opens, the takeout tongs pick up the finished bottle and place it on the conveyor dead
plate. Takeout tongs should be aligned correctly and have sufficient clearance so that the bottle hangs freely,
and the takeout mechanism should move smoothly to avoid distorting or chipping the glass, Figure 3.47f).
An important concern in blowing is the distribution of wall thickness at the end of final blow stage.
a) b) c)
d) e) f)
Figure 3.47: a) Gob loading b) Settle blow c) Counter blow d) Invert e) Final blow f) Take out [12]
53
3.4.3 Press/Blow
The narrow neck press/blow (NNPB) process has been introduced to gain improved control over glass
distribution in the bottle. The better control over glass distribution has enabled a significant reduction in
glass weight of up to 33% without negatively affecting the mechanical performance of the container. A key
component in the above process is the plunger, used to form the cavity in the parison in the forming stage.
The function of the plunger is to uniformly distribute the glass within the blank mould cavity and to aid the
elimination of thermal energy from the internal surface of the parison [24].
 Gob loading
With the plunger in loading position, the gob goes through the funnel of the blank and loads on top of the
plunger. The gobs shape and the loading depth are the most important factors [6].
 Baffle on
After gob loading and with the plunger in the loading position, baffle is then positioned on top of the blank
mould.
 Plunger Up
The plunger starts the upward movement as quickly as possible, in order to press the glass while it still
retains the maximum amount of heat and before it sets too hard. During the plunger up stage the plunger
travels in an upward direction forcing the glass going up against the baffle and then, when the baffle is
completely fill of glass, the glass travels to the neckring to complexly fill it. This action forms the finish in
the neck ring and the parison is completely formed. The pressure applied to the plunger up movement must
be regulated to a minimum to avoid splits and marks in the finish and parison, Figure 3.48c). Then, after
the plunger up stage the plunger has to be removed. So, the plunger retracts to its full down position.
 Parison Reheat
After the pressing operation the plunger is withdrawn, the baffle is removed and the blank mould opened
to allow the parison to reheat and equalize the temperatures to eliminate skin conditions before inverting to
the blow mould. Parison reheat will then begin on the inside of the parison. At this point the colder inner
skin of the parison will be heated up by the hotter glass in the middle of the parison [6]. Complete parison
reheat is started when the baffle is up, the blank open, and the plunger is in the down position. This process
will continue until the final blow is applied [6]. Reheat is important in order to equalize the temperature in
the parison and to melt the solidified outside layer to make the blowing possible.
 Invert
During the time the parison is transferred from the blank to the blow mould, the parison is still reheating.
The speed of the invert must be regulated to suit the size of the parison, as the centrifugal force will allow
swing of the parison heel, which will affect distribution in the base of the finished bottle, Figure 3.48d).
 Rundown
As soon as the parison is transported to the mould side, it starts running towards the bottom of the blow
mould. The amount of run time is set by the job being made. The longer the parison is allowed to run, the
thicker the glass will become in the wall area, Figure 3.48e) [6].
 Final Blow and Vacuum
Vacuum is applied to the blow mould cavity to form the finished container. The final blow air is also used
to help form the shape of the container, but also by circulating the air through the internal cooling blowhead
arm and also the internal temperature of the jar is reduced. In general terms, the vacuum and final blow
time should be the maximum to form the jar before it is removed from the blow mould, Figure 3.48f).
54
 Mould open and take out
The mould opens after the blow head is absolutely up and the takeout closes around the finish before the
mould opens. The takeout action is as smooth as possible to decrease the change of defects, Figure 3.48g)
[6].
a) b) c)
d) e)
55
f) g)
Figure 3.48: a) Gob loading b) Baffle on c) Plunger up d) Invert e) Rundown f) Final Blow g) Take out [12].
3.5 Hot End Coating

The container surface is regularly protected from abrasion damage with double coating. The hot end coating
is enforced between forming and annealing. Glass containers are coated with two substances: one coating
occurs at the hot end of the lehr3 immediately after creating (called a hot end coating), and the other occurs
as the bottles go out of the annealing lehr (called a cold-end coating) [25].
The role of hot end coating is to increase strength and abrasion resistance, and also provides sufficient
lubrication to ensure a smooth flow in the high speed procedures of inspection, automatic filling and
packing. The hot end coating is applied by chemical vapour deposition (CVD), directly after forming the
bottle and before they enter the annealing lehr. A thin layer of tin oxide is formed on the glass surface in
seconds [26]. These materials are applied immediately after forming when the container surface
temperature is between 450–600 °C. A range of precursors can be used, including monobutyltin trichloride,
tin tetrachloride, dimethyltin dichloride, and titanium tertrachloride, Figure 3.49.
3
A furnace used for the annealing of glass
56
Figure 3.49: Hot end coating [12].
Other coating processes have been attempted but failed in the marketplace. None of these is used by
manufacturers currently, primarily because of their higher cost and because they only increased the burst
pressure without also improving the impact resistance. One step coatings were applied at the cold end and
found to be inferior to the dual HEC/CEC coatings. On the positive side, however, there appear to be no
technical barriers to achieving significant improvements to the properties of glass containers. Much of the
technology already exists, Figure 3.50.
Figure 3.50: Application of hot end coating [12].
57
The tin compound is evaporated in air and comes into contact with the hot glass surface (600°C) during the
vapour phase. In bottle coating, the mole concentration of product in air is approximately 0.02 per cent,
leading to a very high excess of oxygen. As a result, the alkyl group is absolutely oxidized. The deposition
rate of the film is mainly dependent on the tin concentration in the coating hood. Other causes are gas
velocity and glass temperature. The gas temperature may range from 70 to 200°C and has nearly no effect
on the deposition rate, as long as it is high enough to contain the tin compound in the vapour phase and low
enough to bypass premature formation of tin oxide in the gas phase [26].
The coating air must contain water vapour; the water is supplied to the hood by exchange of ambient air.
This usually leads to a large excess of water vapour. Under practical conditions, a residence time of 2 to 3
seconds is sufficient to deposit the 10nm thick film of tin oxide [26]. It is known that the structure of hot
end film can be strongly affected by temperature; at relative low temperature the grain size is relative small
while at relative high temperature a more column-like structure is obtained.
The inner loop receives and circulates a highly concentrated stream of vaporized hot end coating material.
The outer loop captures any coating material vapour escaping from the inner loop and recirculates it in a
low concentration. Coating vapour that escapes the outer loop is captured through exhaust slits located at
each corner of the hood. The flow of fresh air through the centre section of the hood assures finish
protection, Figure 3.51.
Figure 3.51: Application of hot end coating [12].
When a "hot end" coating is applied to glass containers, it is normally necessary to avoid depositing metal
oxide coating on the finish of the container. The metal oxide coating on the finish can interact with the
dissimilar metal of the cap, lid, crown, and can produce an unsightly corrosion issue [27]. Such corrosion
issue, with its visual blemishes, is particularly acute since the coated glass containers are normally
employed for beers, wines, soft drinks and foodstuffs, and the ultimate consumers won’t buy products even
appearing to be tainted [27].
3.6 Annealing
Development of stresses during the creation of glass may lead to fatigue of the product, or even to
dimensional changes due to relaxation or optical refraction. The process of removing and reducing the
stresses due to relaxation is called annealing [3]. In an annealing process the glass objects are arranged in
a so called annealing lehr, where they are reheated and then again gradually cooled down. The rate of
cooling is determined by the allowable stresses and property variations throughout the glass [3]. The
annealing has to be overseed carefully in order to avoid excessive temperature differences that may affect
the quality of the product or even lead to cracks in the material [28]
58
3.6.1 Internal Stress

The development of internal stresses in a glass container during the forming process is explained because,
there is a very rapid cooling of the glass in a short period of time. Approximately 850ºC in just 20 seconds,
from cutting the gob until the finished glass container enters the annealing lehr, Figure 3.52.
Figure 3.52: Rapid cooling from forming to annealing lehr [12].
Glass has a low thermal conductivity, therefore, the thermal dissipation does not process at the same
velocity in all of its mass. It is established a thermal gradient from the centre of the glass mass, up to the
external surface, Figure 3.53.
Figure 3.53: Stress after coating process [12].
Temperature gradients across glass wall thickness produces different cooling rates. Different cooling rates
produce different times at which the glass undergoes the glass transition. This in turn produces thermal
stress across sample. Surfaces cool more quickly, bulk cools more slowly. Surface gets put into
compression, bulk in tension. The combination of tensile stresses and surface discontinuities at sufficient
magnitude or severity occurring at same place and time result in a fracture, Figure 3.54.
59
Figure 3.54: Tension inside the container thickness [12].
Containers must be free of internal stresses so that they don’t spontaneous fail under pressure. Annealing
has the objective of reducing the internal residual tensions to minimum values, enhancing its mechanical
properties. This is achieved by the rearrangement of the molecular structure. Annealing is used to heat treat
the glass near the softening point to relieve stresses. Glass has to cool more slowly through the transition
region so as to not reintroduce thermal stresses. Glass which has not been annealed is liable to crack or
shatter when subjected to a relatively small temperature change or mechanical shock. Annealing glass is
critical to its durability. If glass is not annealed, it will retain many of the thermal stresses caused by
quenching and significantly decrease the overall strength of the glass.
3.6.2 Cycle
Each different type of glass containers should have a determined annealing cycle. The duration and thermal
condition of that cycle depends of the: glass characteristics (chemical composition), container geometry
(non-cylindrical containers have intrinsically, because of its shape, more stress concentrators in the bottom
and shoulder areas, therefore, need to spend more time to stabilize and to slow cool down) and container
glass thickness.
Some issues should be taken into account in annealing:
o Thicker glass containers anneals more slowly, has larger stresses;

o Thinner glass containers anneals more quickly, has smaller stresses;
o Glass containers with large thermal expansion requires more annealing;
o Glass containers with smaller thermal expansion requires less annealing;
In the annealing process, the following 4 phases should be considered (that is 4 thermal gradients to reduce
the internal stress in a glass container):
1. Preheating the container until the annealing point temperature. The preheating time depends of the
temperature of the container when it enters the lehr, glass thickness and the value of the annealing point
temperature. At this temperature the glass is still too hard to deform, but is soft enough for the stresses to
relax, Figure 3.55;
2. Stabilization or homogenization at the annealing point temperature. At this temperature the
internal stresses are reduced dramatically in a short period of time (a few minutes) through the
homogenization of the molecular structure by internal viscous flow. The piece is allowed to heat-soak until
its temperature is even throughout. At a glass container corresponds to a temperature of 540 to 565ºC and
a viscosity of log(13,4) (1013, 4 poises). At this temperature glass has the characteristics of a viscouselastic
material. The stabilization time depends of the glass thickness and glass geometry, Figure 3.55.
3. Slow cooling from the annealing point temperature to the strain point temperature at a relatively
slow and uniform velocity in order to assure that no new internal stress appears or is recovered. At this
temperature the internal stresses are reduced in a period of hours and this is the highest temperature from
which strain free glass can be cooled quickly without introducing permanent strains. At this temperature
glass is predominantly in a solid state. At a glass container corresponds to a temperature of 495 to 520ºC
60
and a viscosity of log(14,6) (1014,6 poises). The slow cooling time increases with the glass thickness and
container geometry (increases for non-cylindrical type of containers), Figure 3.55.
In order to assured that all parts of the glass container are at a higher temperature than the strain point
temperature this phase should be extended to a temperature of approximately 475ºC. For typical soda-lime
container type of glass the difference between the annealing point and the strain point should be around 35
to 45ºC (viscous transition zone or annealing zone).
4. Rapid cooling from the strain point temperature until room temperature. The cooling speed should
be limited in order to avoid breakages due to thermal shock, Figure 3.55.
The thermal gradients (cooling velocities) and permanency times at a certain temperature depend essentially
of the glass thickness of a container and the glass expansion coefficient, which is to say the glass chemical
composition.
Figure 3.55: Tension inside the container thickness [12].
The annealing process is carried out in a temperature controlled kiln known as a lehr. A temperature
controlled curve is inputted on the lehr, depending on the container characteristics. The tunnel is comprised
of individual modules or zones of equal length. Width varies widely based on capacity but remains uniform
from zone to zone. These zones are designed to produce a gradually decreasing temperature curve, which
typically starts slightly above the calculated annealing point of the glass and decreases zone by zone until
the desired glass exit temperature is achieved.
For the most efficient heat transfer method most modern lehrs use convection fans and can be heated with
gas burners or electric elements. The annealing lehr continuously receives glassware directly from the
forming process. Ware is conveyed through the tunnel via a metal mesh lehr belt that provides a stable
surface for conveyance as well as allowing recirculated air to pass through and around the glass products.
3.6.3 Quality control

Transparent materials having internal mechanical stress are found to be birefringent, due to their anisotropic
nature difference, when measured along different axes, in a material's physical or mechanical properties.
That is, they demonstrate double refraction (having two indices of refraction). Transparent materials having
internal mechanical stress can be analysed by a Polariscope 4. The Polariscope incorporates a polarizing
filter to convert white coloured into a polarized light source that is then directed through a glass sample. If
the glass is free from stress, the light simply passes through unaffected. Stresses in the glass, however,
retard the transmission of light, and produce various spectral colours when viewed through the
4
An instrument for measuring or exhibiting the polarization of light or for examining substances in
polarized light, often to determine stress and strain in glass.
61
Polariscope, Figure 3.56. The level of stress present on the glass affect the amount that the light waves are
retarded. Actual levels of stress (temper grade) can easily be determined by comparing colours exhibited
during the test with those produced by calibrated standards (Standard Strain Disks), taking into account the
thickness of the glass.
Figure 3.56: Measurement of glass stress with a polariscope [12].
3.7 Cold End Coating

In the past, when glass containers were heavy and line speeds slow, there was little need for surface
treatment. In result the ware was incredibly heavy, the speed was awfully slow, and the lehr was covered
with broken glass. However, in almost all modern manufacturing, bottles have become increasingly lighter
in weight and speeds have accelerated extraordinarily, both in the glass plant and on the customers filling
line. Because of these changes, the need for surface treatment is crucial, Figure 3.57 [25].
Figure 3.57: Different surface treatments though a convey belt [12].
62
Although glass is in theory a very strong material, conditions exist throughout the entire manufacturing
process which can damage the ware to the extent that, unless the surface is protected, damage can occur.
Resistance to scratching is paramount to keep the inherent bottle strength high [25].
Water based polyethylene emulsions are normally applied by some variation of the traverse spray system.
This machine delivers the diluted material, that is mixed with approximately 100 to 200 parts of water, over
at the cold end of the annealing lehr, from the shoulder of the bottle on down [25]. Care should be taken
that the suggested amount of concentrate coating be mixed with water. Overuse of the concentrate won’t
enhance its performance values. If the mixture is too heavily concentrated, appearance of the ware can be
debilitated. The glass containers will come out there streaked and can also be too slippery, resulting in
filling-line pop-out. Too heavy concentration will also affect the labelling ability and can lead to failures in
labelling. Conversely, too little concentrate will result in bottles without sufficient coverage to avoid surface
abrasion, Figure 3.58 [25].
Temperature of application is also an aspect involved in successful cold end coating. The “cold end” isn’t
cold at all, with the bottle temperature ranging from 66° C to 177°C. It is important that the polyethylene
coatings be applied in the range of 82°C to 113°C for the best labelling ability. If the coating is applied at
a too low temperature, it will inadequately cure; if the temperature is too high, the danger of thermal shock
is present. Another problem of high temperature application is that labelling failures may happen [25].
Figure 3.58: Application of cold end coating [12].
3.8 Inspection and quality control

It is essential to have good raw material specifications. A technical specification sometimes defining ideal
or optimum limits in addition to the normal reject limits is necessary and is often part of the commercial
contract against which raw material deliveries are made. Physical and chemical checking of incoming raw
materials on receipt has now largely been replaced by either the provision of supplier certificates of
conformance, or by the availability of a system whereby the glass manufacturer is given access to the raw
material supplier’s process control information when necessary. The latter facility may also be
supplemented by audits of the raw material supply locations. The purpose of such audits is to ensure
familiarity with the process control methods and standards used by the supplier and to conform their ability
to adequately control the raw material within the specification.
When raw materials are received in the glass factory, the incoming check can simply confirm the nature of
the material and the quantity being delivered if a certificate of conformance system is in operation. For
certain materials or where there is a specified need, small samples can be taken and retained for laboratory
examination. This may be done either on a random basis or regularly if required. An important part of the
63
raw material control system is ensuring the right material goes into the right raw material storage silo and
this is usually done by means of an electrical interlock system or a manual padlock system on the silo input
pipes. Certain checks on the raw materials are done on a regular basis as specified by the glass manufacturer.
Cullet is a very important raw material. Cullet is inspected at various stages in the cullet processing
operation as it is converted from raw cullet e.g. as received from bottle banks, into a form which can be
incorporated into the batch mix. Any large contaminating objects such as refractory bricks, are eliminated
by hand packers as the cullet enters the processing plant on the input conveyor. Magnetic separators remove
magnetic materials and light sources of contamination such as aluminium retaining rings are blown away
from the crushed cullet. Control tests are regularly carried out on the processed cullet to ensure it is within
the glass manufacturer’s cullet specification. When processed cullet is received from an external cullet
producer the manufacturer either carries out control checks similar to the above when deliveries are received
or operates a formal compliance certificate system.
As with the conventional raw materials there is a close connection between the glass manufacturer and the
mould suppliers, usually through the design and mould ordering departments of the glass manufacturer. In
addition to any formal compliance certificate in the case of moulds, the supplier often forwards a summary
of the final inspection data from checks carried out on the mould equipment at the time of manufacture. If
the sampling raises a concern, this is either investigated in conjunction with the supplier, or if necessary,
further samples of mould equipment are checked.
The packaging materials include pallets, cartons, trays, and shrink wrap and stretch wrap materials. Again,
as with the raw materials, the main control is exercised by the packaging supplier who is monitored through
certificates of conformance systems and supplier audits as appropriate, Figure 3.59.
Figure 3.59: The packaging materials include pallets, cartons, trays, and shrink wrap and stretch wrap materials [12].
There are four main aspects of furnace control:
o To ensure the effective melting of the raw materials and to provide good quality glass to the
forming machines. Samples of the glass being produced are retained for specified periods so that
any change in appearance can be quickly detected.
o To ensure the most economic operation of the furnace and to minimize emissions, which might
cause atmospheric pollution.
64
o To ensure the furnace fabric is maintained, e.g. to check that no hot spots or thin spots are
developing and that the cooling systems are working effectively.
o To provide early warning of any aspect of furnace control or equipment which may require
attention.
The controls carried out at furnace process control could typically include measurement of:
o Furnace temperatures.
o Regenerator temperatures.
o Glass melt temperatures.
o Furnace pressures – measurements of differential pressure relative to the atmosphere. The ideal
melter pressure is the lowest amount which prevents any inspiration of the cold outside
atmosphere;
o Glass level depth in the furnace – typical measuring location is in the refiner/distributor or the rear
zone of forehearth. The glass position is referenced relative to a defined “zero”. Measuring
precision must be to the nearest one hundredth of an inch. The reading must be accurate and
reliable.
o Bubbler rates (when bubblers are used to assist melting);
Such items are regularly monitored and the results of checks carried out are either recorded in furnace log
books or are available as a computer record, especially if the furnace is fully computer controlled as is often
the case with modern glass melting units, Figure 3.60.
Figure 3.60: Furnace control [12].
The main test to asses melting / glass quality are:
o Glass density (dif < 0,0040 g/cm3).

o Homogeneity (this test indicates the presence of any inhomogeneous glass which shows up as
coloured streaks under polarized light. The degree of inhomogeneity can be quantified if necessary
by examination under a petrological microscope).
o Bubble (seed) count.
o Stone count (and determine origin).
o Colour (purity, brightness, dominant wave length).
65
The density test is carried out using a density bath which consists of several boiling tubes partially immersed
in water within a water bath. The boiling tubes contain a special mixture, e.g. tetrabromoethane and
bromonapthaline or bromobenzene, the density of which approximates to the density of the glass. Small
fragments of glass are then introduced into the boiling tubes together with control samples of the known
density. As the temperature of the water bath is then increased at a controlled rate, the density of the liquid
mixture changes slightly and the samples floating on the top gradually sink to the bottom. The temperature
at which the samples (known density and unknown density) sink past a mark on the boiling tube is noted,
and from this the density of the unknown glass can be determined.
Samples from each individual mould cavity on the forming machine are taken off at regular predetermined
intervals and allowed to cool for inspection in a light box. Glass weight is another parameter which is
regularly monitored. All checks are carried out having regard to the container specification and any other
important control information made available by the quality assurance department. Such additional
information could typically relate to previous customer complaints or concerns, or to special customer
requirements for the job being manufactured. All bottles checked at the forming machine are rejected for
cullet as they cannot be subsequently annealed. Another part of the manufacturing control system involves
the regular monitoring of the mould numbers which are working on which forming sections of the machine
and precisely when mould equipment is changed and for what reason.
One key aspect in the quality control system in a glass plant is the communication between the operators
of hot end coating and cold end coating in the same production line. This communication should be
bidirectional:
o Hot end operator should inform the cold end whenever a defect is detected in his normal sampling.
This way cold end can react and eliminate the defect.
o Cold end operator should inform the hot end whenever a defect is detected in his normal sampling.
This way hot end can react and correct (and eliminate) the defect.
Two test machines both supplied by A.G.R. are in common use. One is used for checking the surface coating
on the body of the container as it is rotated in a vertical position against a sensing head. The other checks
the surface coating on the finish of the container. In both cases as the coating is very thin (only angstrom
units thick) the test machines assess the reflection from the depth of the coating rather than measuring the
actual amount of material present, Figure 3.61.
Figure 3.61: Hot end coating quality control [12].
Associated with the annealing lehr it will have a process control variable the temperatures of each of the
zones in the annealing lehr, against the temperature specified as necessary to give the correct annealing
profile in the lehr usually in the form of preset controls. Unlike the other tests carried out on the finished
container which are all performed on “rounds” or one container from each forming cavity taken at regular
66
intervals from the cold end, the annealing test is carried out on samples taken from designated positions
across the lehr, e.g. either from the centre or from each side of the lehr, or from all three positions. The test
samples are examined in a polariscope (or strain viewer) equipped with a tin plate. When looking through
the eye piece of the polariscope, the amount of strain in the glass is proportional to the degree of colour
observed in the field of view of the polariscope. A quantitative assessment of the strain can be made by
comparing the observed colour against standard strain discs or against reference samples with the same
design characteristics which themselves have been assessed against standard strain discs. This test is usually
performed in the quality control laboratory of the plant, although for quick assessments it is possible to see
polariscopes near production lines.
For process control cold end coating should be monitored for the correct operation of the application
systems and also the temperatures of application. Simply rubbing the containers together as they leave the
lehr normally provides sufficient confirmation that the surface treatment has been satisfactory. This test is
usually performed by the cold end operators. A further confirmation is obtained by means of quantitative
surface treatment checks done in the quality control laboratory. In this test two containers are placed side
by side in a horizontal position on the stage of the machine and a third container is placed on the top. The
stage is then raised at one end until the top container starts to slide. The angle at which this occurs is known
as the slip angle and provides an assessment of the level of the cold end treatment, Figure 3.62.
Figure 3.62: Cond end coating quality control [12].
The cold end process controls are carried out after the containers emerge from the annealing lehr. Some
controls are based on the examination of representative samples from each mould cavity at regular intervals
(attributes inspection for stratified sampling). Such controls could include the gauging or measurement of
specification dimensional features (qualitative assessment with go-no go gauges) and general assessments
of the visual quality of the containers being produced from particular moulds.
If defects are found the cold end operator must reject the concerned cavity numbers and inform is hot end
counterpart of all nonconformities detected. The effectiveness of the inspection machines, if present, in
rejection that defect, must be assessed. If the production line is equipped with automatic inspection
machines the cavity rejection is performed automatically. For that, the container has the mould number
code engraved either as a dot code (digital body or bottom of the container) or with characters
(alphanumeric bottom of the container).
One group of machines sidewall visual inspection check for features which may result in internal
contamination of the container. The contamination may be caused during manufacture, e.g. a birdcage, or
may be introduced subsequent to manufacture, e.g. a piece of broken glass. Basically these machines
compare the image of the container against the expected image and deviations such as dark areas due to
extraneous features cause the container to be rejected. Similarly, these machines can reject excessive visual
67
deviations which might be objectionable cosmetically. This machine is able to detect all stress defects
generating tensions in the sidewall.
Another group of multi inspection machines uses light reflection principles to detect and automatically
reject features such as splits. If a split is present, this reflects light from a lamp source as the container is
rotated in front of it. The reflected light is picked up by a sensor and the container is rejected. This machines
have several inspection stations, containers indexed by a star wheel check selected dimensions such as the
bore of a bottle or the undulation of the sealing surface of a jar and glass thickness. If the container has the
dot mould number code engraved on the body, the systematic mould rejection is made on this machine.
Using cameras to perform the automatic inspection, there is another group of machines that performs
inspection to the base and finish of the containers and finish inspection machines. These machines are able
to inspect a number of different visual defects on the base and finish and stress inspection on the base of
the container. If the mould number code is engraved on the base of the container, it is on this machine that
it is possible to reject systematically a specific mould number. The automatic inspection machines are
regularly checked with control samples to ensure they are performing satisfactorily. The results of such
control checks are recorded as are also the checks on the representative samples referred to in above. These
results are useful should there be a need for subsequent investigation and they provide a valuable input for
the preventative control measures employed at the hot end during forming.
These samples – must follow certain rules or good practices:
o The sample must present only one defect for a specific detector or camera.
o The defect on the sample must be of low magnitude constituting a real challenge to the inspection
machine.
o The sample must be clean in order to avoid false rejections because of the dirtiness rather than the
defect itself.
This procedure is key to assess the effectiveness of the inspection machines. Also at regular intervals the
rejections of the inspection machines should be assessed by the cold end operator. This is to provide to the
hot end valuable information for correction and at the same time assessing the machine functioning. After
the inspection machines in the cold end usually have a manual visual inspection as the containers pass in
front of an on line light box. Here the cold end operator inspects at regular intervals during a short period
of time the containers that are passing in the line. The objective of this inspection is only to assess the visual
quality of the containers, therefore, check the effectiveness of the inspection machines regarding this
particular aspect.
Some containers, e.g. carbonated beverages, require a mechanical stress control to be applied to each
individual bottle and this is achieved by passing the bottles through an on line impact simulator, or squeeze
tester. In this device, the bottle is squeezed between two vertical rollers or pads which place a stress on the
bottle so that weak containers are broken and eliminated.
After the inspection machines and just before the containers are packed on pallets, in the cold end of a glass
plant, there can be another equipment: bottle (jar) inverter. This equipment turns the containers upside down
and a clean air flow is blown inside the container. The aim is to remove any physical contamination from
inside the container. Together with the inspection machines this is a typical Critical Control Point (CCP)
for food safety in a glass plant. This is a critical equipment for the production of wide mouth jars. In some
production lines that only produce bottles narrow mouth. A good practice to follow is to investigate any
occurrences of loose glass found in the inverter trays. Additional statistical checks, or audits, are done as
the containers are being, or when they have been, assembled on pallets for shrink-wrapping.
3.8.1 Defects
For the purpose of these checks, imperfections are often classified into three main groups. The first group
concern critical defects, defects that could or are likely to result in hazardous or unsafe conditions for
individuals using the glass container or be injurious to their health, under normal conditions of use as
defined between the glass manufacturer and the filler. The second group concerns major defects, i.e. those
68
which may affect the functionality of the container in some way. We can subdivide this group in two
subgroups: Major A defects: any non-critical defect which the consumer could perceive as being harmful
and/or which damages the bottler's or trademark owner's legal image and/or which can result in legal action
against the Company; and Major B Defects: any defect which is likely to reduce the usability of the product
for its intended purpose. The third group concerns minor defects any departure from established standards
having little bearing on the effective use of the product (usually a defect that materially affects the
appearance but not the performance of the product aesthetic or dimensional slight defects). Some defects
are presented in the Figure 3.63.
69
Figure 3.63: Defects can be presented in a glass container [12].
Critical Type Description
Any container which is so badly

deformed that it will not pass through
Freaks
a normal filling line and it can be
seen at almost any distance
Small projection of glass in the

Spike internal surface of the container
(bottom) (bottom). For Press/Blow and NNPB
processes
Air bubbles near the inside surface.

Internal These bubbles can easily break
surface loosing glass particles inside the
blisters container. Usually associated with
thin glass thickness.
70
Thin strand of glass across the inside

of a container either between the
Birdswing walls or between the wall and the
bottom. Contact of the parison
internal walls before final blow
A finish which has excessive glass

Overpress projecting upward from the inside
edge of the finish.
Excess of glass in the neck which

Stuck or loose
partially or completely obstructs the
glass inside
bore (thermal or mechanical origin).
Rough or sharp glass on the outside

Stuck ware surface of the container due to hot
contact between two containers
Offset of the finish (out of the correct

position). If the finish rim is
Offset finish
protruded there is a danger of
breakage (mechanical origin).
71
Major A Type Description
Surface crack in the base of the

neck. Usually occurs near the
Check on the
seam but it can occur anywhere
neck
along the neck diameter
(thermal or mechanical origin).
Thin but deep crack – V shaped

– on the body of the container.
Check on the
Two types of check: pressure –
body
mechanical - check and hot
panel – thermal - check.
Thin and vertical cracks that

Check on the
occur near or on the blow
seam
mould seam
Crack on the bottom part of the

Check on the
container (near or on the
bottom
bottom parting line).
Checks on the
Small cracks on the engravings
engravings,
or other moulded container
orientation
features.
marks
72
Crack on the finish of the

container. Probably one of the
Check on the most common type of defects
finish (thermal or mechanical origin).
Temporally corrected with
swabbing.
A finish from which a small

Chipped
section is broken on the top or
finish
side.
A finish which is incompletely

Unfilled
filled, in the top, the bead or the
finish
thread
A finish which is bulged out of

shape, either by blowing
(thermal) or mechanical action.
Bulged finish It may prevent the good
capping of the container, which
will spoil the filled product or
lead to breakage
The sides of a container

Sunken /
(usually panel shape) is sunken
bulged sides
or pushed out.
A neck which is tilted to one

side or that has another type of
Bent neck
deformation (mechanical or
thermal origin).
73
Typically the containers

Broken ware crack/break inside the
annealing lehr
Non-uniform glass
distribution. Thin spots on the
Thin ware container, which are below the
minimum acceptable level
(thermal origin).
Major B Type Description
The body of the container has a non-

Out of round cylindrical / oval shape (thermal
origin).
A bottom which has sagged, so that

Rocker
the container is unstable when
bottom
placed on a flat surface.

or major
A
Broken or very fragile bubbles on the

Blister on the
top of the finish (risk of glass
top of finish
particles inside the container).
74

or Major
B
Unfilled body shape or

Out of shape
deformed after the opening of
container
the blow mould.
The body of the container is

tilted from the vertical axis
Leaner
(not perpendicular to the base
– thermal origin).
Seams which are relatively

large, extending from the
Blank and
shoulder to the bottom of the
blow mould
container. Blank seams tend
seam
to veer off from the blow
mould seam.
A fin or seam of glass lying

Neck ring
across the top or the side of
seam
the finish (mechanical origin).
75
Surface crack under or in the

Check under lower part of the finish, near
finish or on the neck ring parting
line.

or Major
B or
minor
A seam occurring between the baffle
and the blank mould (mechanical or
Baffle mark thermal origin). It can show a crack
or a tear which extends until the
body of the container.

or Major
or minor
Line over Small crack in the outer top surface

finish of the finish. It does not reflect light.
Crizzled A finish with many fine surface

finish fractures, mainly across the top.
Major or Type Description

minor
Wedge A bottom thicker than specified in

bottom, heel one of the sides. Mainly a cosmetic
tap, slug defect but it can cause thin spots in
bottom other parts of the container.
76
Major B Type Description

or minor
A small surface section torn from

Tear
under it.
Superficial and external mark with

Open mark
two separated, irregular rims.
The baffle mark has swung to one

side and is not central in the bottom
Swung baffle of the container. The container can
have low glass thickness in the
opposite side of the swung baffle.
Minor Type Description
Orange peel / Rough, grainy and dirty aspect

toad skin similar to the texture of an orange
marks peel or toad skin.
77
A series of horizontal waves or

Wash boards
folds on the side of the container
Wavy appearance in the body of the

Cold moulds
container (thermal origin).
3.8.2 Quality Control Laboratory

A failure during a routine process control test initiates an immediate retest of a larger sample to establish
whether or not there is a downward quality trend. When substandard ware is detected, all the containers
from the suspect mould cavities are rejected until there is a successful test. In addition, all the ware packed
as good from the suspect mould cavities since the last successful test is regarded as suspect and
reinvestigated so that all substandard ware can be rejected.
Any containers which have to withstand thermal shock in use, e.g. refillable bottles, or bottles subject to
pasteurization, are subjected to a specified hot to cold thermal shock. The test samples are placed in a basket
and then immersed in a hot water tank so that they are completely filled with water. They are then allowed
to soak, after some time they are transferred to a cold water bath where they are immersed before removal
and examination for signs of breakage.
In the internal pressure test the bottles are filled with water and then placed. In the tester where the pressure
is applied hydro dynamically once the test head is sealed on top of the bottle. There are two types of testers
in use. These are known respectively as “increment,” or “continuous” (or Ramp) testers.
The impact test is normally carried out as a design evaluation test rather than as a routine process control
test. The test container is placed on the stage of the tester against either a V-shaped back stop, or in some
cases a horizontal cylindrical back stop. The stage is then adjusted vertically so that a swinging impact bob
will impact at the desired position when the arm supporting the bob is raised to a predetermined level and
released. The test level is progressively increased until the container breaks or alternatively, the test can be
terminated once the container has survived a particular test level.
In vertical load test, the test is carried out during design evaluation rather than as a routine process control
test. The container is placed upright on the stage of the testing machine and the load is transmitted along
the vertical axis of the container via a ball swivel head piece as the stage is gradually raised. The vertical
load can be applied through an attachment to a standard internal pressure machine, or by means of any other
machine which simulates this operation.
In the capacity (volume) test containers are weighed empty and then weighed again when filled with water
either brim-full in the case of jars or to the designated fill point in the case of bottles. The difference between
78
the weight of the container when empty and when filled with water represents the capacity of the container.
Capacity, dimensions and glass mass are, of course, interrelated.
The moulds in which glass containers are formed are usually made of cast iron and, when producing glass
containers, operate at a temperature of approximately 750ºC. The moulds have to be cleaned frequently of
scale deposits caused by oxidation and carbon residues produced by mould lubrication. Thus, allowances
have to be made for mould wear. As the external dimensions increase, the mass of the glass may have to
be increased to keep the capacity within tolerance.
There are several ways in which the glass thickness is checked. Some require the container to be cut into
segments or halves while others can be carried out on the whole container. The most common thickness
measuring instruments in some cases are mounted on stands or incorporated into other support devices.
Another instrument uses a magnetic ball which is introduced inside the container to be held directly opposite
to the sensing head of the instrument. The instrument is moved over the external glass surfaces
simultaneously moving the ball on the inner surface and the distance between the ball and the instrument
head is the thickness of the glass.
Some measuring instruments are purpose built, e.g. to measure container height. Others are incorporated
into special support devices, e.g. to measure verticality and ovality. Results are either logged on separate
control sheets or automatically submitted for S.P.C. analysis, when this method of control is appropriate.
The data collated from this activity serves to form part of the continuous improvement programme of
operational control to achieve greater consistency of the glass manufacturing process. When dimensions
are extremely difficult to gauge, or cannot be accurately gauged, profile projectors are normally used for
control. These instruments project a profile of the glass onto a screen. The glass dimensions are then
checked against graduations marked on the screen, or against profile transparencies placed in front of the
screen, or by means of micrometres attached to the projector’s stage. Finish dimensions are usually checked
on some form of profile projector where the finish outline can be precisely measured or compared against
a profile of the finish specification.
The type and number of gauges which are necessary to maintain good dimensional control will vary from
one type of container to another. Most container specification drawings give tolerances for the ring finish,
overall height, and the leading body dimensions, and so these dimensions will be subject to a gauging check,
usually in the production line, both at hot end and cold end. Untoleranced dimensions are not subjected to
a routine gauging check. Other gauges are used as required to control features such as sunken or bulged
body, out of vertical, undulating finish on wide mouth containers, etc., where these features have an adverse
effect on the performance of the container. Certain types of gauge, e.g. ring gauges, are constructed from
metal blanks to predetermined sizes. Other gauges, e.g. height gauges, are often constructed so that they
can be adjusted and preset for different types of containers. The sizes used when manufacturing fixed
gauges or setting adjustable gauges are taken from the bottle or ring finish specification.
In the case of bottles, the internal bore and in some cases the profile is usually specified, particularly where
the bottles are to be filled automatically or where they have closures which seal against the inside of the
ring, e.g. cork closures. In such cases bore gauges would be used. On the other hand, bore gauges are rarely
necessary for jars. There is equipment's who is able to measure automatically the inner profile of a bottle,
giving a quick and visual indication of the profile.
3.9 Palletizing
When glass containers leave the production line they have to be packaged in some way. This packaging has
to satisfy the requirements of both manufacturer and the customer. It has to provide for easy transportation
and to protect against the ingress of foreign bodies and general atmospheric contamination. This is typically,
nowadays, done using some form of automatic palletizer. Dependent on the container application and/or
destination, the glass will be packed in either bulk or modular form. Bulk palletization consists of pallet
sized layers of containers stacked to the required height and covered with suitable protection [29].
79
Modular packages are smaller groups of containers packed into small packages such as cartons, trays, brick-
packs or crates. Each of these styles of pack are then loaded onto a pallet and again stacked to the required
height. The palletized containers are then either shrink-wrapped or stretch wrapped and strapped (bottles)
to aid stability and offer a barrier to the environment before being transferred to the warehouse usually by
fork lift truck or a conveyor system. Each one of the pallets is labelled either manually or automatically to
assure product traceability [29].
Before any glass containers are placed onto a pallet it is necessary to consider the interlayer between the
pallet and the containers. When completed pallets are to be shrink-wrapped it is usual to have a polythene
base sheet installed prior to any packing in order to create a barrier to upward contamination and to provide
a base to seal the shrink-wrap bag. The base sheet will normally extend beyond each pallet side prior to
shrinking. This fuses to the shrink-wrap to act as a damp proof membrane, provides an aid to pack integrity
and aids stability. A pad or tray can be placed over the base sheet dependent on customer requirements.
In order to prevent damage to the glass containers and to help the stability of the bulk pallet, each layer of
containers is separated from its neighbours by a sheet of corrugated or solid material, which is made from
either plastic or cardboard. Alternatively, a cardboard tray may be used. This can be upright or inverted to
provide additional stability. These trays can be either manual or automatically erected. The choice of layer
dividers is often governed by the method of depalletization to be used at the filler premises. A capping tray
or pad is used to complete the pallet.
To assemble the glass containers into modularized pallets nowadays are used palletizer systems from semi-
automatic ones to fully automatic systems. Usually these equipment's only require human intervention for
loading the bulk packaging materials into the palletizer. These equipment's transfer the layers of containers
to the pallet either by dragging them or by vacuum operated or special designed heads. Even the most exotic
products with demanding shapes can be arranged into one layer and can be taken over by the swivel arm
stacker exactly as it is. These type of equipment's can also be used for cardboard box palletizing (modular
packaging).
The shuttle car makes the transfer of completely formed pallets from the palletizers of each production line
to the centralized units of shrink (stretch) wrapping (and sometimes strapping) units. These units make the
traffic of forklift truck traffic between palletizer, shrink wrapping and empty pallet lines unnecessary. In
this way shuttle cars produce a significant reduction in hazards for man and machine and provides
transparent materials-flows.
In some circumstances it may be desirable to use strapping to hold the unit together and to provide
additional support to the pack (e.g. bottles). Strapping can be applied horizontally around a layer of
containers and/or vertically over the finished pack. Horizontal strapping, sometimes known as “banding”,
can be applied to the layer of containers at the bulk palletizer. It is used to minimise the movement of the
layer during transportation but may increase the handling difficulties at the depalletizing stage.
Vertical or compression strapping can be used in conjunction with plastic layer pads. This requires the
application of a returnable (reusable) top frame to distribute the load applied by strapping. The strapping
units can be either integrated in the palletizers of each production line or be centralized units just before the
wrapping operation, Figure 3.64.
80
Figure 3.64: Vertical palletizing [12].
3.10 Shrink-wrapping
The assembled containers on the pallet are either covered with shrink wrap applied at a shrink oven station,
or with stretch wrap applied by an automatic stretch wrap machine. Stretch wrap saves cost and space,
while shrink wrap provides additional product retention, full weatherproofing and higher speed. In either
case the processes are regularly checked against specified standards, and in the case of the shrink wrap
machine, items such as the shrinking time/oven temperature, are additionally monitored and recorded.
Figure 3.65: Shrink-wrapping a pallet [12].
81
Pallets are usually conveyed to the glass manufacturer’s warehouse by forklift truck. They are then stacked
and stored in designated bays, each of which has a separate code to enable easy stock location and to permit
the greatest degree of regular stock rotation. Other important warehouse controls concern protection against
infestation and contamination of the shrink or stretch wrap covers. Additionally, if there is any deterioration
of the pack, or loss of pack integrity, then the pallets concerned are rejected for reprocessing or repack
before they are loaded for dispatch. The main loading controls are checks to confirm the vehicle bed is
clean, to see that the pallets are loaded correctly, and to see that the load is securely held and that all outer
covers are in place before dispatch.
Perhaps the major quality concern at this stage of the process is the physical contamination of the product
and/or pallet load. This is of special importance just before the pallet load is shrink or stretch wrapped.
During the palletization process and the transfer of the unit load to the shrinking unit, the unit load is
unprotected and subjected to possible sources of contamination. A possible and very probable source of
contamination is the contamination by pieces of broken glass originated during the container manipulation
either on the palletizer or during transport. At this stage of the process there are no additional equipment’s
that can detect and eliminate this contamination (either inspection machine or container inverter). This fact
calls for a human control/supervision at this stage and the enforcement of “broken glass” procedures by the
glass manufacturer. In order to avoid contamination the integrity of the shrink wrap must be assured during
this operation and downstream the process until the pallet load reaches the filler. The shrinking time/oven
temperature is a process parameter that should be monitored in order to achieve optimum performance. The
pallets when stored in the warehouse should be inspected for integrity. If found with any hole in the plastic
the pallet loaded should be rejected and returned to the production area for reinspection and reprocessing.
The same inspection should take place just before the pallet is load into the truck or container.
Loading the pallet verticality is another key aspect to take into account. The deviation from vertical position
of the pallet load can hamper the storing, transportation, depalletization and container integrity downstream
the process. The causes of this phenomena are variable and can be related either with the operation:
palletization, transfer of the pallet load from palletizer to shrink wrapping, incorrect shrink wrapping or the
packaging material (not enough stiffness). Some of the shrink wrapping equipment’s are able to re-centre
the pallet load and correct any deviations to the vertical position just before this operations takes place
through a centring device. However, this should be one of the controls to perform when storing the pallet
loads in the warehouse. If the pallet load is strapped another aspect that should be object of control is the
strapping tension. This parameter is of key importance to monitor the integrity of the strapped layers. This
parameter should be monitored in all of the strapped layers and just before the shrink wrap takes place. A
loose layer will definitely cause handling difficulties at the depalletizing stage.
82
Chapter 4 Mathematical Modelling of Glass Forming Processes
CHAPTER 4
4 Mathematical Modelling of Glass
Forming Processes
In this chapter the modelling techniques of glass container forming processes are presented. Modelling
glass forming consists in the reproduction of the physics processes by means of mathematics and numerical
models. The finite element method was here the choice to the development of a methodology for simulating
this technology process. The physical process is a coupled thermomechanical problem evolving on the glass
and its interaction with the air and equipment. In this chapter, the partial differential equations that describe
the temperature field and mechanical behaviour are presented.
4.1 Finite Element Formulation for Thermomechanical Behaviour

4.1.1 Constitutive Model
Glass material can be mechanically characterized as an isotropic Maxwell viscoelastic material. When
deforming the total strain rate  may be considered to be obtained as the sum of the elastic strain rate e
and the viscous strain rate p:
  e   p 4.1
where:

e 
E
4.2

p 

83
Chapter 4 Mathematical Modelling of Glass Containers
where E is the Young’s modulus and η is the viscosity. Therefore;
 
  4.3
E 
Nevertheless, experiments on stress relaxation show that for lower viscosities, i.e., for high temperatures,
the relation between viscosity and the rate of stress release becomes linear indicating that the viscous flow
is the main mechanism and hence permitting the neglect of elastic effects, which means that it can be
assumed that  e  0 [11].
By separating the stress-strain relation into its deviatoric ( sij , eij ) and dilatational (  m ,  ii ) components
we may write:
 ij   m ij  sij 4.4
where the deviatoric stress tensor is given by,
sij  2 eij 4.5
and the hidrostatic stress is given by,
 m  k ii 4.6
being  and  k the shear and bulk viscosities, respectively.
At high temperatures glass deformation can be considered incompressible, i.e.
 ii  0 4.7
and consequently the solution is driven by the deviatoric response and the constitutive equations take the
more general form:
sij  2  ij
4.8
 ij  sij  p ij
being  ij the strain rate tensor and p is the hydrostatic pressure:
84
 ii
p 4.9
3
4.1.2 Finite Element Method Formulation

The finite element model is established from the equilibrium equations for the thermoviscoelastic problem.
Following the standard finite element method implementation and notation taking nodal velocities v as
the primal variables within any element the velocity field v is defined, by means of the standard shape
functions N as:
v = Nv 4.10
The strain rate tensor,
 ij 
1
2
 vi, j  v j ,i  4.11
can then be related with the nodal velocities by the B Matrix5.
ε = Bv 4.12
The deviatoric stress (Equation 4.8) can be expressed in the matrix form:
s  Dε  DBv 4.13
being D the shear stress viscosity matrix.
The penalty method is used to impose the incompressibility constraint by defining:
p
 ii   4.14

where α is the penalty value, that can be related with the bulk viscosity. This penalty value is weighted
locally with the bulk viscosity as
   4.15
being λ a ‘large’ number.
Then, the hydrostatic stress can be related with the dilatational stress as:
5
B matrix is composed of the space derivatives of the shapes function
85
 ii
m    p   ii 4.16
3
or, in the matrix form,
 m   mTε   mTBv 4.17
T
for the axisymmetric case m is a vector given by,
m T  1 1 1 0 4.18
Neglecting inertia forces the mechanical equilibrium equations are:
 ij
+ bi = 0 4.19
x j
where bi are body forces components per unit volume.
Following the Galerkin method the final system of equations, obtained by performing integration over the
full domain  and its boundary  , is then obtained in the standard form as:
Kv = f 4.20
The global stiffness matrix K includes the deviatoric, Kd, and dilatational, Kh, as:
K = Kd + Kh 4.21
where
 B
T
Kd = DBd 4.22
Ω
and
K h    B Tmm T Bd  4.23

The load vector is,
86
f  f p +f G 4.24
where fp and f G represent the vectors of nodal applied ( papplied ) and body ( b ) forces, respectively
f   NT papplied d    NT b d  4.25
 
It is interesting to notice that if a mixed formulation is adopted, utilizing velocities and pressure as primal
variables the final system of equation would take the form:
K v v + Qp = f
 T
4.26
 Q v=0
or in a matrix form:
K v Q   v  f 
Q T 
0  p  0 
4.27

where the second set of equations relate to the incompressibility constraint.
Relaxing that constraint by means of a penalty parameter  such as
1
QT v  Ip  0 4.28

then
K v Q 
  v  f 
Q T 1       4.29
 I p   0 
  
Solving for pressure in the second set of equations,
1
QT v  Ip  0  p   QT v 4.30

and substituting in the first set of equations we obtain a system only dependent on the nodal velocities
K v v   QQT v  f 4.31
87
where K v may be identified with K d , and  QQ T may be identified with K h .
It is well known that, to avoid “locking” effects, the number of variables associated with the constrain
condition (incompressibility here) in the second set of equations must be smaller than the number of
variables associated with the other primal field (velocities here) and that is related with the need of using
reduced integration techniques when evaluating the K h .
In general, a nonlinear static finite element is most effectively performed using an incremental formulation,
in which the static variables are updated incrementally corresponding to successive time steps in order to
trace out the complete solution path. In this solution it is important that the governing finite element
equations are satisfied at each time step to obtain sufficient accuracy, because otherwise, solution errors
can be significantly accumulated that can lead even to solution instabilities [30].
The glass viscosity is dependent on the glass composition and also is highly dependent on the temperature,
influencing greatly the flow of the material as discussed in the section 2.4. Therefore, it is essential to
constantly model heat transfer within the glass and to the air and moulds and to couple it with the
mechanical formulation that describes glass deformation.
4.2 Finite Element Formulation for Heat Transfer

In this section, it was consider the finite element formulation for the thermal problem. The main goal is to
introduce the finite element formulation necessary for the solution of nonlinear transient heat transfer
behaviour of glass containers production. Heat transfer and mechanical flow will be treated in an uncoupled
but staggered manner, because glass viscosity and other material parameters are assumed to be temperature
dependent.
4.2.1 Transient Analysis

The temperature is the most important parameter in almost all stages of glass making and glass processing.
During the glass melting, the temperature influences the homogeneity of the glass melt, the drop in
temperature influences the following forming process and finally the cooling of the hot glass influences
possible thermal residual stresses inside the glass product [31].
The classical partial differential equation governing the transient heat transfer in a solid is:
   T    T    T   T
  kx    ky    kz   Q  c 4.32
 x  x  y  y  z  z   t
where T is the temperature, kx, ky, kz are the thermal conductivity coefficients in x, y, z directions (if the
material is isotropic then the thermal conductivity can be simplified as kx = ky = kz = k),  is the density,
c is the specific heat and Q is a heat source per unit of volume.
To solve it an initial temperature condition for the entire body, at an initial time t0 must be known
T  Ti on  at t  t0 4.33
Thermal boundary conditions may be of different kind and can be classified into two types:
88
• Essential boundary conditions (Dirichlet boundary conditions):
where temperatures are imposed in part of the domain boundary, 1 , are known
T  T1 in 1 4.34
• Natural boundary conditions (Neumann boundary conditions):
where heat flux can be imposed in part of the domain boundary, 2 .
T
qc  k   c T  Ta  4.35
n
where αc is the heat transfer coefficient, Ta the temperature of surrounding fluid (air) and n is the outward
direction normal to the boundary surface.
Also, a radiation flux can be imposed at the boundary surface;

qr    T 4  T4  4.36
where β is the Stefan–Boltzmann constant, ε the emissivity and the T∞ the surround temperature. Along the
free surface in contact with air, a condensed formulation can take into account both convective and radiative
heat flux, the sum of which being written as qcr.
qcr  qc  qr 4.37
The heat flux considering the convection and radiation, may be given by,

qcr   c T  Ta    T 4  T4  4.38
or,
 
qcr   c T  Tamb    T 2  T2 T 2  T2  4.39
and,
 
qcr   c T  Tamb    T 2  T2 T  T T  T  4.40
Therefore, a temperature dependent “equivalent heat transfer parameter” can be defined for radiation as,
89
 r    T 2  T2  T  T  4.41
Assuming that Ta and T are the same in a condensed approach may be written as
qcr   cr T  Ta  4.42
where αcr is an effective temperature dependent parameter or heat transfer “coefficient” (HTC). The heat
transfer coefficient (HTC) depends also on the velocity and on the material. Therefore, in hollow glass
production the heat transfer coefficient can be different in different regions of the mould, the ring, the baffle
or the plunger.
To solve equation 4.32 a finite element solution based on the Galerkin weighted residual method was used.
The temperature T is interpolated from the nodal values Ti using the approximating functions Ni
n
T   NiTi 4.43
i 1
where Γ represent the whole boundary and n is the total number of nodes.
Applying the Galerkin weight residual method to the transient analysis gives:
    T    T    T   T 
 i   x  x x  y  y y  z  z z 
N   k  k  k   Q   c d  0
t 
4.44
 
for i=1,n.
Integrating by parts we obtain for each equation,
 Ni  T  Ni  T  Ni  T   T  

  x  x  y  y  z  z 
   k x    k y    k z   N i Q  N i  c

 d 
t  
4.45
 T   T   T 
  Ni  k x d    Ni  k d    N i  k d   0
  x    y    z 
with the boundary conditions are:
 T   T   T 
 N  k d    Ni  k d    Ni  k d     Ni qd    h T  Ta d  4.46
x  y  z
i x
       
where h is the heat transfer coefficient
90
The final finite element formulation for the thermal behaviour in matrix form is generally given by,
KT+CT = Q 4.47
being,
T
T 4.48
t
the time derivative of the nodal temperatures T ,
K is the of the “conductivity” matrix, with Kij terms for i,j=1,n
 N N j Ni N j Ni N j 
Kij    i k xT j  k yT j  k zT j d     hN i N j d   0 4.49

x x y y z z  
and C is the heat source term with Qi terms defined as,
Cij    cN i N j d  4.50

and Qi the terms of the “thermal load” vector,
Qi   Ni Qd    Ni qd    N i hTa d  4.51
  
4.2.2 Time Integration

For the integration in time we introduce a set of approximations to the time derivative by linear interpolation
of two consecutive time values
tn1  tn  tn 4.52
by
Tn 1  Tn  t 1    Tn   Tn 1  4.53
where 0    1.
Writing the previous equation at time t n
91
KTn + CTn = Qn 4.54
using that linear interpolation, assuming that K and C are kept constant in the time interval tn and
obtaining Tn1 from equation:
KTn1 + CTn1 = Qn1 4.55
we may write
 1   1 
 K  t C Tn+1  1    Qn   Qn+1  1    K  t C  Tn 4.56
where
ˆ   K  1 C 
C  t 
4.57
1 
Kˆ   C  1    K  4.58
 t 
Different values for  may be used that lead to different type of solutions:
  0 - Forward difference or Euler scheme (conditionally stable)
1
 - Crank-Nicolson or trapezoidal rule scheme
2
2
 - Galerkin scheme
3
  1 - Backward difference scheme (unconditionally stable)
When   0 the scheme is termed explicit. The others schemes are termed implicit.
Implicit schemes are unconditionally stable in linear problems. This means that the numerical process will
not diverge even if Δt is large, but the solution may be coarse because important changes may take place
on a small time scale. Implicit schemes are second order accurate, which means that numerical errors in T
produced by the algorithm is approximately quartered when Δt is halved.
Among the implicit methods, the choice of   1 is popular, but sharp transient conditions may provoke
2
annoying oscillations in the solution. Oscillations can be reduced using smaller values of Δt or numerically
damp the solution by using a value of γ somewhat greater than ½.
92
If Q represents a thermal shock the solution may display some spurious oscillation, which can be reduced
by using a value of γ greater that ½ so as to introduce algorithmic damping.
The explicit scheme may be unstable unless we adopt a time step smaller than
2
tcr 
1  2  max
4.59
ˆ T.
ˆ T C
where max is maximum eigenvalue of the eigenvalue problem K
93
94
Chapter 5 Simulating Model for Glass Containers
CHAPTER 5
5 Simulating Model for Glass Containers
Some of the main features of the general computational model for glass forming derived, namely for
press/blow and blow/blow processes are described here. Various modelling aspects are discussed for each
process, while several key issues, such as the movement of the plunger, contact, and remeshing technique
are examined thoroughly.
Certain steps in formulating a finite element analysis of a physical problem should be embodied in the finite
element software. Before running a simulation a model has to be created. So as a first step a preprocessor
tool is needed to create a model. The preprocessing step is, basically, used to define the model and it
includes
 Definition of the geometric domain of the problem.

 Definition of the element type(s) to be used.
 Definition of the material properties associated to the elements.
 Definition of the physical constraints (boundary conditions).
 Definition of the loads.
 Generation of the mesh.
The preprocessing is an important step to obtain a “good” result. Then, to run a simulation some few steps
are needed. The main procedures in the finite element analysis are:
 Input reading.
The input needed for the finite element analysis comprise:
o Geometry definition,
o Finite element mesh information (number of nodes, number of elements, nodal
connectivity,
o Finite element parameters,
o Material properties,
o Node temperatures,
o Load information;
95
o Process parameters.
 Obtain the element associated matrices and vectors.
 Assembling element matrices.
 Apply constrains to the matrix and vector according to the imposed boundary conditions.
 Solve the equation system of resulting non-linear equations, with an iterative process to compute
the unknown values of the primary field variables.
 Compute secondary variables, such as, from the known nodal temperatures/velocities, if required,
strains and stresses can be computed.
 Post Processing, plot results.
The visualization of the results is performed through a postprocessor. The customised post
processor is used to allow the user to rotate, translate and zoom into and out from the objects. The
postprocessor software contains sophisticated routines used for sorting, printing, and plotting
selected results from the finite element solution. Animations and movies can also be produced to
simulate the dynamic aspects of a problem. Outputs in form of text files and x–y plots are also
available. The output data are processes in a desired format for plotting. While solution data can
be manipulated in many ways in the postprocessing, the most important objective is to apply sound
engineering judgment in determining whether the solution results are physically reasonable [32].
5.1 Element Technology

The software is able to perform a simulation with several elements type such as triangular and quadrangular
elements with linear and quadratic integration. Next these different elements and the problems associated
to them will be described.
5.1.1 Library of Elements

The most used 2D shapes the triangle with 3 nodes, 3 nodes with bubble function and the quadrilateral with
4 nodes or 8 nodes are included.
5.1.2 Spurious Oscillations Due to Smaller Time Increments

Some oscillation in the solution of the temperatures that can occur in the time domain. Such behaviour is
directly associated with the numerical characteristics of the time integration scheme involved and the time
increment being employed in the solution.
In particular, the solution of time dependent field problems using quadrilateral or triangular elements is
often accompanied by increase or decrease in temperature that violate the physical aspects of the problem.
In transient heat transfer analysis with second-order elements there is a relationship between the minimum
usable time increment and the element size that should be taken into account [33]. A simple guideline is:
c
t  L2 5.1
6K
where ρ is the density, c the specific heat , K the conductivity and ΔL the element size.
5.1.3 Hourglass and Locking

One of the important issues that have been discussed in the literature is the locking problem. Some finite
elements exhibit in some applications overly stiff behaviour characterised by the fact that important modes
of deformation cannot be effectively reproduced. This is the case when using elements with low order
interpolating polynomials in problems involving the incompressible constraint in the deformation of a
medium, as is the case in glass forming. Typical solutions may involve the use of high order elements,
mixed formulations or selective reduced integration techniques. However these solutions may, in certain
cases, introduce some spurious energy deformation modes, commonly known as hourglass effects, and
96
some stabilization techniques may be needed. In the next chapter an analysis of the underlying deformation
modes associated with different types of elements used here will be analysed.
5.2 Contact
Most of the problems in the simulation of manufacturing processes involve contact between two or more
bodies, sometimes involving complex geometries and undergoing large deformations. From a mechanical
point of view, at macroscopic scale, contact is the notion of the interaction between bodies coming in to
touch and exchanging load and energy [34]. The key condition on these processes is the condition of
impenetrability, namely, the condition that two bodies cannot interpenetrate. Nonetheless, all physical
processes involved in contact such as friction, heat transfer, must be considered.
To model the contact in glass container forming we may consider two states. The first is the so called
'sticking’ state in which a point of the glass mesh is in contact and it is impossible to move in the tangential
direction, i.e, against the mould. The second case is ‘sliding’, which means that a node can move freely in
tangential direction through the contact interface. In between these two states a sliding with friction state
can occur. The contact between two bodies is described by a choice of both behaviours, such as, normal
and tangential conditions. Contact with sticking friction is a good macroscopic description of perfect
adherence whilst sliding contact can be considered as a perfectly lubricated contact.
In the case of thermomechanical contact problems two fields - deformation and temperature - interact, and
thus have to be considered within the formulation. In the general settings, these fields are coupled, since
the constitutive parameters depend upon the temperature. Furthermore, the evolution of the thermal field is
related to the deformation.
In the contact of two bodies, two different areas can be distinguished. The area where there is contact and
the area where there is no contact, Figure 5.1.
Figure 5.1: Model domain contact problem.
The velocity vector at each point in contact can be expressed in the rotated axis in relation to the contact
surface.
v p  vn  vt 5.2
97
where vp are the velocity of the node and vn and vt are the normal and tangential components of velocity.
The position of a point at time t +Δt is obtained as:
xt t  xt  ut 5.3
with ut being the displacement measured at time t and
ut  t v t 5.4
In order to prevent node penetration of the glass through the mould a correction of current increment time
can have to be performed, Figure 5.2. This will be clarified in the next sections
Figure 5.2: Target for glass node contact the mould.
5.2.1 Contact Algorithm

The initial assumption of the algorithm is that only one node of the FEM mesh representing the glass can
touch the mould at each time interval. This may redefine the initial assumed time increment as it will be
explained for either sliding or sticking conditions.
The contact detection is a search of closest glass node that can come into contact with the mould during the
current solution step. Large contact problems may imply a large number of contacting nodes. This simple
detection technique, based on a comparison of distances from each ‘slave’ node to all components of the
‘master’ surface, may lead to an excessive time consumption, especially, if contact elements must be
frequently updated. The flowchart of the general contact algorithm is presented in Figure 5.3.
98
Inicialization of contact
Node =1 ITURN =2 Node
Calculate node
Calculate node
→ Displacement
→ Displacement
Yes →Coordinates
→Coordinates
For new increment
For initial increment time
time
No Node Update node

Node fixed? Yes
fixed? coordinates
No
No
No
Boundary
Boundary node?
node?
Update
Confirm the
→Boundary
Determine the minimum delta Yes
conditios
minimum delta time in time in Plunger
→Coordinates
Plunger
No
ITURN = 2
Confirm the Update
minimum delta time →Boundary
Determine the Yes
in blank or blow conditios
minimum delta time in
mold →Coordinates
blank or blow mold
No
Update
Confirm the
→Boundary
minimum delta time Yes
conditios
in neckring
→Coordinates
Determine the
minimum delta time in
neckring
No
Update
Confirm the
→Boundary
minimum delta time Yes
conditios
in baffle
→Coordinates
Determine the
minimum delta
time in baffle
End Program
Figure 5.3: Contact algorithm flowchart for each node.
99
5.2.2 Sticking Condition

A point which sticks to the mould does not move in a tangential direction. This is a valid assumption for
certain stages of the process.
Firstly all boundary nodes are candidate to contact the mould. The algorithm tracks all the surface nodes,
excluding the nodes that have already touched the mould. The next step is to calculate which mould element
can be reached, according to the calculated velocity, and the time needed to reach the mould.
A mathematical condition can be built to satisfy the node positioning over the mould and the time needed
to reach it, as represented in Figure 5.4.
Figure 5.4: Contact representation for sticking condition.
The two conditions must be satisfied:
x   A  t m   u  B  A 
5.5
x  P  t v
where A and B are two points of the mould segment, P and v the position of the candidate node to contact
and its velocity, u is the position of the mould element where the node will stick, Δt is the time needed to
reach the mould element and m is the mould velocity.
The intersection is given by;
 xA  tm1  u  xB  xA   xP  tv1

 y A  tm2  u  yB  y A   yP  tv2
5.6
or, in matrix form,
  m1  v1   xB  xA   t    xP  xA  
 
 m2  v2   yB  y A   u   yP  y A 
5.7
100
Taking:
 m1  v1   xB  xA 

 m2  v2   yB  y A 
5.8
and using the Cramer’s rule we obtain,
 x P  x A   xB  x A 
 yP  y A   yB  y A 
t  5.9

and,
 m1  v1   xP  xA 
 m2  v2   yP  y A 
u 5.10

For each node then is checked in which mould segments u lies between [0,1] and the smaller time increment
(Δt) the touching segment is selected and the time step redefined. The coordinates of all nodes and boundary
conditions are then updated.
Initialization of tool
contact
Node candidate data
Tool element
No Calculate Δt and u
Δt>0 End routine

0≤u≤1
No If iturn = 2
Update node
Δt ≤ old Δt Yes
coordinates and
boundary conditions
Figure 5.5: Contact flowchart for candidate segment selection.
101
5.2.3 Sliding Conditions

In the sliding case, some differences have to be implemented in the previous algorithm. The first obvious
difference is when the node is in contact with the mould. This means that the node can slide along the mould
segment. Slipping may result in two parallel vectors, Figure 5.6b). The tangential velocity and the tool
element vector are parallel, therefore, Δ is equal zero in equation 5.8. So, for this case the algorithm has to
be changed. In that scenario Δt is calculated so that the distance between the contact node and the mould
node in the velocity direction is calculated using:
 x A  xP    y A  yP 
2 2
t  5.11
v12  v22
For the moulds and the plunger, we may allow both slip and no slip boundary conditions.
Figure 5.6: Contact velocity components a) Nodes in contact to the b) Sliding component in a single mould
mould with the velocity components. component.
5.2.4 Updating Boundary Conditions

In this section the update of boundary conditions will be briefly discussed.
 Mechanical contact boundary conditions
During the glass container production in gob loading stage it is allowed that the glass nodes slips along the
mould. This can be a good approximation due to the high gob initial velocity due to the falling and low
viscosity. This stage time is also very small and the cooling effect almost does not affect viscosity. In others
stages the boundary conditions are redefined as sticking conditions as the node of the finite element mesh
touches the mould. Mechanically the node is then assumed to be fixed to the mould, which seems to be a
good approximation due to the sudden drop of temperature and consequent large increase of viscosity [35].
The particular case of plunger up stage had to be addressed differently, sliding friction is present. In
practice, the plunger is driven by a piston, which means that the movement is the result of an external force
applied to the plunger.
The plunger contact with the glass can be added to the formulation by the minimization of a functional, or
more precisely an energy term, associated to this contact that should include the work of the external force
of the plunger, the internal resisting force associated to glass deformation and the constrain of
impenetrability.
102
From the finite element solution point of view a new variable is added, V , which is the plunger vertical
displacement, which is assumed to be a rigid body with only vertical displacement under an applied vertical
force, Fv .
Then the external work is:
we  Fv .V 5.12
The internal work involved will be associated with the displacement (or velocity) variables at the contact
nodes of the finite element mesh used in the glass. As the plunger will be geometrically modelled by lines
that may not be parallel to the coordinate axes, so it is preferable to assign those restrain conditions
(impenetrability and slide friction) in local coordinates. Therefore, for every contact node a local coordinate
system, as in Figure 5.7, is assigned.
Figure 5.7: Local coordinates at each contact node between the plunger and the glass.
being,
u   u '
ui   i  , ui '   i  5.13
 vi   vi ' 
where,
ui  Ri ui ' 5.14
with,
cos i seni 
Ri  
cos i 
5.15
 seni
103
Here, i  1, m , where m is the number of points of the mesh contacting the plunger geometrical
description. Please note that the number of contacting points may change with every time step.
An energy term related with the friction, may be added as:
1
  ui ´
2
wif  5.16
2
where  is a friction factor, so that fi   ui is the friction force at any contact point. Note that for  0
the sliding condition is recovered, for a large value of  the sticking condition is present.
The impenetrability constrain can be introduced in a penalized form including an energy term as:
1
wip    vi ´V ´
2
5.17
2
where  is a penalty value imposing the impenetrability constrain.
Then, the global energy term may then be written as:
1 m  cos i  sin i    cos i  sin i   ui  

 V ,...ui ...,...vi ...    ui vi  V   
cos i    sin i  cos i   vi  V  
2 i 1   sin i
 Fv .V 

1 m

2 i 1
  
 cos 2 i   sin 2 i ui 2   sin 2 i   cos 2 i  v V 
i
2

 2 cos  i sin  i     ui  vi  V  
 Fv .V
5.18
By performing the minimization of this function the extra equation associated with plunger movement is
obtained as
   
m
   cos
i 1
i sin i     ui   sin 2 i   cos 2 i vi   sin 2 i   cos 2 i V Fv
5.19
and the extra terms to be added to the global stiffness in equations associated with node i are
...    cos 2 i   sin 2 i  ui   cos i sin i     vi  ...    cos  i sin  i     V  0
...   cos i sin i      ui    sin 2  i   cos 2  i  vi  ...   sin 2  i   cos 2  i  V 0
5.20
104
 Thermal contact boundary conditions
In the case of thermomechanical contact problems the two fields - deformation and temperature – interact.
These fields are coupled, since the constitutive parameters depend upon the temperature, i.e., the evolution
of the thermal field is related to the deformation by viscosity.
The majority of the heat flow during contact between glass and the metal in the moulds is by conduction
because the metal is opaque to radiation. The rate of cooling depends on the heat transfer coefficient and
on the temperature difference between the glass and metal [36]. In general, two different types of boundary
condition have to be considered Figure 5.8. The nodes that are in contact with the mould surface a
conduction boundary conditions is assigned, in the other hand, the remaining surface nodes are assigned
with convection/radiation boundary conditions.
Figure 5.8: Contact between two bodies at different temperatures considering the different boundaries for each
physical field.
In modelling thermal contact problems by the finite element methods some difficulties arise, Figure 5.8.
First, it is impossible to model the contact at every point along the boundary. For example, in the two-
dimensional case in which boundaries of individual elements are represented by straight line segments, as
shown in Figure 5.9, some nodes may be in contact with the body but the segment between the nodes may
not be in contact.
Also, the element type can complicate the contact between glass and mould. For example, a single element
along one sides may not be touching completely the mould. Typically, in a quadratic element, along one
mould element, one or two nodes may be contact the mould but another may be in contact with the air
inside the mould and in these cases it have to be particularized. In Figure 5.9b) the first case has the node
T3 only in contact with the mould, the nodes T2 and T3 have convection/radiation boundary conditions. In
the second case only two nodes contact the mould and similar analysis can be done. Then, in the third case
all nodes contact the mould and the boundary conditions are the same.
105
Figure 5.9: Contact between two bodies at different temperatures considering the different boundaries for each
physical field.
5.3 Mesh Updating and Remeshing

Our finite element method starts with a finite element mesh generation to discretize the geometry as
illustrated, for example, in Figure 5.10.
Mesh generation must be carefully done as the mesh is responsible for the accuracy of the solution
computed. A special attention should be paid to the way the boundaries of the domain are approximated in
order to capture the physical behaviour of the problem as accurately as possible. It means that both the
number of nodes and elements must be adequate (in terms of shape, size, density, variation from region to
region, etc [37]. However, mesh construction can be very expensive in terms of time.
Figure 5.10: a) Example of a triangular linear mesh. b) Example of a quadrilateral linear mesh.
The size of elements should be chosen carefully as it can influence the convergence the solution. The
smaller the size of the elements the better is the accuracy of the solution but bigger the computational time
[38].
106
5.3.1 Mesh Updating

The mesh is updated every time step, by adding to the current node coordinates the corresponding current
displacements as:
xt t  xt  ut  xt  t vt 5.21
where xt t are the coordinates at time tt t , xt the coordinates at time t and vt the velocity at t .
At a certain stage of the simulation the mesh may be distorted and have to be redefined. Therefore, the
finite element mesh should be constantly checked to detect excessive element distortion. This can be done
by calculating the determinant of the Jacobian, which tends to zero. If the Jacobian becomes negative in
part of the element, this implies a negative density. It may then correspond to a violation of mass
conservation and continuity of the displacement field, Figure 5.11.
Figure 5.11: Example of an element with negative Jacobian.
Therefore, a parameter for mesh distortion is introduced. For example, for a linear quadrilateral element
where full numerical integration require 4 sampling Gauss points this control parameter is:
 min J 
DP  4    5.22
 area 
where min J is the minimum value of the determinant of the Jacobian at the Gauss points and area is
the area of the element. Similar formulas are adopted for other types of elements.
If DP is less than a tolerance, then the element is considered unacceptably distorted and the analysis must
terminate using the current mesh and a redefinition should be undertaken.
A new mesh is then created and the field variables transferred from the old to the new mesh.
A meshing tool available as open source code “GEOMPACK” was utilised. The mesh generator needs a
boundary well defined. So, a Jacobian ratio was created to control the mesh distortion. The ratio is given
by,
J n  J n 1
MDratio  5.23
Jn
107
Where Jn is the Jacobian at tn and Jn+1 at tn+1. This mesh distorted ratio gives a number between [0,1] which
can be effectively done on the percentage of element distortion. When those parameters are exceeded the
simulation is stopped and a new mesh is created, as will be explained in the next section. This procedure is
needed to maintain the accuracy of the results. What is acceptable in one situation may be unacceptable in
another [9.8].
5.3.2 Remeshing
Forming glass containers requires large material deformations, in most cases, to perform the simulation
completely with the same mesh it is quite difficult, therefore, a remeshing tool is required.
The definition of the mesh should be chosen according to the element type and element size in order to find
the best relation between the accuracy and the computational effort, Figure 5.12, note the example shows
triangular elements but the same analogy can be done for quadrilateral elements. A refined mesh will
improve the results, but it will need high investments in computer time. The accuracy of the solution
depends on the element type and mesh used to represent the domain and the representation of the actual
geometry. Using linear elements to represent a curved boundary is a coarse approach and mesh refinement
may be needed to improve the representation of the specific geometry. A triangle can easily be decomposed
into congruent subtriangles. Thus, we can easily perform a global grid refinement to halve the mesh size.
Figure 5.12: Meshing a curve with different mesh size.
If sticking boundary conditions are adopted as deformation continues elements near the boundary may
become increasingly distorted diminishing the reliability of the solution. Eventually a singular matrix may
become inevitable. This situation can be prevented by introducing a remeshing procedure. Remeshing is
also used to prevent grid distortion, inside the domain, as the interpolation functions used are only
appropriate for convex elements. The remeshing process redraws the grid and interpolates field variables
from the old to the new mesh [36]. The flowchart of the remeshing technique is illustrated in Figure 5.13.
Some aspects that were taken into account will be referred next.
108
Remeshing
Pass all the nodes and elements

information to temporary arrays
Calculate the volume of the

older mesh
Create a boundary to perform a

new mesh
Calculate the volume of the

boundary No
Compare the volume from

the older mesh and the boundary.
If is less than the tolerance then
refine boundary
Call mesh tool
Read new mesh
Compare the volume from

the older mesh and new mesh.
If is less than the tolerance then
refine new mesh
Renumber the new mesh
Pass the information to new mesh.

Store the nodes and elements
information in glogal arrays
End Subroutine
Figure 5.13: Flowchart of remeshing technique.
 Creation of the boundary through the older mesh
The data required for the remeshing stage consists of the boundary data, or more precisely, a polygonal
discretization of it, input as a set of segments. The original boundary, which may be a curve, is the collection
of the exterior element boundaries. The boundary uses the nodes of the older mesh to create the segments
that will create the final shape of the new mesh. If the mesh required is a quadratic elements mesh, the
remeshing tool will include in the middle of the segments an additional node. However, these additional
nodes can lead to wrong boundary conditions if not properly dealt with. These additional nodes can also be
109
located such that they are interior or outside to the tool. Likewise, the refined mesh may end up with nodes
that do not coincide with the boundary, Figure 5.14, note the example shows triangular elements but the
same analogy can be done for quadrilateral elements. This means that these possible source of errors must
be dealt with.
Figure 5.14: Correction of the new boundary nodes.
For a concave tool surface, we must prevent any node to penetrate into the tool domain. In order to satisfy
the non-penetration condition a projection of the nodes to the tool is performed. However, errors such as
material losses can become important and must be controlled.
 Volume control
The volume of the glass may increase or decrease if coarse or refined meshes are required. Then, a volume
control must be implemented to maintain the glass volume constant during simulation. Two types of volume
control were implemented. One control is on the boundary and other control on the new mesh.
In order to calculate the volume of material from the node boundary positions, Figure 5.15, the following
algorithm is used. The boundary was parameterize by each segment as:
Figure 5.15: Example of a boundary segments.
110
1 1
x 1  u  x1  1  u  x2
2 2
5.24
1 1
y  1  u  y1  1  u  y2
2 2
Consequently,
dx 1
  x2  x1 
du 2
5.25
dy 1
  y2  y1 
du 2
Using the Green Theorem we may evaluate the total volume by a line integral along the full boundary that
at each segment, takes the form
2
 1
1 
  1  u  x1  1  u  x2  .  y2  y1  du 5.26
2 1  2 
Simplifying,
2

 y2  y1    1  u  1  u  x22  du
1
1 2
 
x12  1  u 2 x1 x2
1 2
5.27
1  
2 4 4
Resolving the integral we obtain,

1
 y2  y1   x12u  x12u 3  x1 x2u  x1 x2u 3  x22u  x22u 3  5.28

1 1 1 1 1 1
2 4 12 2 6 4 12  1
Simplifying,

 y2  y1   x12  x1 x2  x22  5.29
3
The volume is the sum of the integrals on each of the n segments,

 y2  y1   x12  x1 x2  x22 
n
Vb   5.30
i 1 3
To control the new mesh volume, the volume is calculated by adding the individual volumes associated
with each element of the new mesh by performing a numerical integration as.
111
 J   r w
nelem ngaus
Vn   2
i 1 j 1
j j j j 5.31
where nelem is the total number of elements, ngaus the number of Gauss points per element and at each
Gauss point Jj is the determinant of the Jacobian, j the local coordinates, rj the radius and wj the
weighting factors for the Gauss numerical quadrature.
These two volumes are compared with the a previous volume calculated, Vl , given a certain tolerance as
Vb  Vl   tolerance
Vn  Vl   tolerance
5.32
If the conditions are not satisfied a new refined mesh is created. After the mesh overcome the two conditions
the program runs normally.
 Transfer of information to the new mesh
When the mesh is too distorted a new mesh is generated and field variables must be interpolated to the new
mesh. This issue is crucially important to be able to pursue the simulation, after a remeshing stage and to
preserve accuracy, robustness and convergence properties of the finite element solution [39].
Velocities and temperatures fields which are stored on nodes of the old mesh must be transferred to the
nodes of the new mesh. This may be done by determining the spatial position of a node on the new mesh,
inside a n element of the old mesh that contains it. Then, easily interpolation can be performed by using the
element shape functions.
To find that element a simple algorithm is implemented as described in Figure 5.16. A node A of the new
mesh is detected inside an element of the old mesh if all the third components (normal to the plane of the
figure) of the three product vectors indicated are positive.
 v1  u12 k  0   v1  u23 k  0   v1  u21 k  0 5.33
Figure 5.16: Cross product between a node that belong to a new mesh and the older element.
112
For other type of elements it is easy to divide in triangle and generalize the procedure.
Boundary conditions have also to be updated. In the new geometry, for practical reasons, each node of the
glass is considered as being in contact with the mould when the distance between the node and the mould
element is less than a very small tolerance.
113
114
Chapter 6 Elements Validation and Testing
CHAPTER 6
6 Elements Validation and Testing
The glass forming simulation involves complex heat transfer processes with the material properties
changing constantly with temperature and different type of non-linear boundary conditions changing
throughout the forming process. The model developed should be able to deal with all these aspects and a
validation of its robustness must be performed comparing the solutions obtained in real forming processes.
This will be done in Chapter 8 but, prior to that, it is always helpful to run simple example tests on the code
developed before embarking in more complex ones. The successful application of finite element analysis
should always include an initial validation of the element to be used and its implementation in a specific
computer program. Usually, the elements utilized in most problem are very well understood and tested.
However, some applications can be difficult to model, and the elements used for the analysis may be more
prone to numerical difficulties [40].
In our case the performance of the elements should be tested in simple thermal and thermomechanical
problems, to assess their robustness. This is particularly important in glass forming problems, both from
the thermal and the mechanical points of view, due to the fact that the material flow is incompressible and
depends on viscosity, which is highly influenced by the temperature distribution. Thus, this requires some
investigation on the behaviour of the different available elements on some specific situations that may cause
some “pathological” responses.
6.1 Heat Transfer Validation

In transient heat transfer problems, for certain type of elements used in the simulation, some specific
boundary conditions and geometric discretization, spurious temperature oscillations may occur particularly
at initial time stages of glass forming simulations due to the small time increment presented. Our aim in
this section is to understand those oscillations in temperature induced by small time increments. We want
to know when those oscillations may appear and how to prevent such behaviour.
In order to validate the numerical approach to the heat transfer phenomenon the analysis was made for each
one of the most common element types, i.e., Lagrangian linear triangular and quadrilateral elements and
quadratic triangular and quadrilateral elements. In the first analysis not all meshes have the same element
size, in order to comply the equation 5.1, but have the same number of nodes. Later the same analysis will
be performed with the same element size.
115
The validation of the in house software was performed using analytical solutions and Abaqus software. In
Figure 6.1 the meshes used with Abaqus and the inhouse software are shown. Also, the time increment was
kept equal for all cases.
Figure 6.1: Mesh used to simulation for heat transfer validation a) linear triangle b) quadratic triangle c) linear
quadrilateral d) quadratic quadrilateral.
The validation was performed creating two different boundary conditions, such as prescribing temperatures
and heat fluxes. These two cases are analysed as follows.
 Prescribed temperature
The first analysis was to prescribe a boundary temperature of 500℃. A cylinder with a unitary radius and
the double dimension value on the height. A material with diffusivity equal to one was used in order to
simplify the analytical calculation. The initial temperature was 1000℃. The scheme can be seen in Figure
6.2.
Figure 6.2: Scheme to validate the prescribed temperature at the boundary of a cylinder.
116
o Analytical Calculation
The analytical solution for the infinite cylinder with convection boundary condition is given by:
 J   
T  Tf  J R
 2
1

exp n2 Fo 0 n
6.1
Ti  T f n 1 n 1 n
where Ti is the initial temperature, Tf the temperature of the outside, Jo is the Bessel function and λn are the
values of the Taylor series which can be determined using,
J 0  n   0 6.2
Fo the Fourier modulus is given by:
t
Fo  6.3
R2
where α is diffusivity coefficient and t is the time. The analytical solutions were obtained in [41].
The analytical solution was obtained using the equation 6.1 for a Tf=500℃, Ti=1000℃, R=1m and α=1.
The results shown in Figure 6.3 are for initial time of 0 to 0.1 seconds with a time increment of 0.01 seconds.
The low time increment was used intentionally in order to induce the spurious oscillations. Also, during the
glass simulation the time increment may be very small because the contact conditions induce small time
increments. The results are shown combining temperature/radius or temperature/time evolution. In the
Figure 6.3, it can be seen the analytical results for the example previous explained. The results appear be
consistent with the temperatures showing a quadratic evolution.
Figure 6.3: Results for the analytical equation for the prescribed boundary temperature a) Temperature/Radius b) Temperature/Time
evolution.
117
o Abaqus results
A model in Abaqus software was created to verify the temperature evolution with time. As discussed before
the temperature evolution can have oscillations. In Figure 6.4 the results for Abaqus simulation with the
different elements types are shown. The results were obtained at the following nodes of the meshes: nodes
1, 4, 7, 10, 13 for x=0.0, nodes 2, 5, 8, 11, 14 for x=0.5 and the nodes 3,6,9,12,15 for x=1.0. For the quadratic
quadrilateral the nodes considered were 1, 4, 6, 9, 11 for x=0.0, 2, 7, 12 for x=0.5 and 3, 5, 8, 10, 13 for
x=1.0. It can be seen that the linear elements doesn’t show any kind of temperature oscillation, presenting
a small deviation from the analytical solution. In other hand, the quadratic elements show a big oscillation,
including nodes with higher temperatures than the initial, which is an unphysical behaviour. However, the
quadratic quadrilateral elements tend to converge to the analytical solution faster than the quadratic
triangular elements.
a) b)
c) d)
Figure 6.4: Results for Abaqus simulation for the prescribed boundary temperature a) Linear triangular elements b) Linear quadrilateral
elements c) Quadratic triangular elements d) Quadratic quadrilateral elements.
118
o Empaktor Software
The same analysis was performed using the code developed (Empaktor software). In this case a full capacity
matrix was used. Figure 6.5 shows the results for the linear triangular and quadrilateral elements as well as
for the quadratic triangular and quadrilateral elements. It can be seen that the oscillations also appear in the
linear elements. Nevertheless, all the situation appear to converge to the analytical solution.
a) b)
c) d)
Figure 6.5: Results for Empaktor software for the prescribed boundary a) Linear triangular elements b) Linear quadrilateral elements c)
Quadratic triangular elements d) Quadratic quadrilateral elements.
Inasmuch as we want to avoid the temperature oscillations, to use a full capacity matrix seams not to be
advisable. Therefore, two different ways of building a lumped capacity matrix were tested. The first one is
to force the non-diagonal values to be zero. The results of this case are presented in Figure 6.6. As it can
be seen the temperature oscillations in the linear elements disappear but the temperatures are smaller than
that analytical solution. The quadratic element present higher temperatures than the initial temperature, a
situation that should be avoided. The second case is to weight the diagonal capacity matrix with the non-
diagonal values, such as the diagonal values are given by:
119
n n

i 1, j 1
Cij   Cii
i 1
Cii 
i j
n
, Cij  0, for i j 6.4
C
i 1
ii
being n the number of nodes of the element. The results for this case are presented in Figure 6.7. A lowest
deviation from the analytical solution is obtained. However, the quadratic elements still present unphysical
heating.
a) b)
c) d)
Figure 6.6: Results for Empaktor software for the prescribed boundary temperature using only the diagonal values for the capacity matrix
a) Linear triangular elements b) Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic quadrilateral elements.
120
a) b)
c) d)
Figure 6.7: Results for Empaktor software for the prescribed boundary temperature using a pondered capacity matrix a) Linear triangular elements
b) Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic quadrilateral elements.
 Prescribed Flux
Here a prescribed flux was considered as the boundary conditions. The scheme of the analysis is presented
in Figure 6.8.
Figure 6.8: Scheme to validate the prescribed flux at the boundary cylinder.
121
o Analytical Calculation
The analytical solution for the infinite cylinder with convection boundary conditions is given by:
T  Tf
 2


BiJ 0  n R  exp n2 Fo 
 
6.5
Ti  T f n 1
2
n  Bi J 0  n 
2
Ti is the initial temperature, Tf the temperature of the outside fluid, h the convection coefficient and λn the
values of the Taylor series which can be determined using,
n J1  n   BiJ 0  n   0 6.6
being Bi the Biot modulus that is given by:
hR
Bi  6.7
k
R is radius of the cylinder, Fo is the Fourier modulus given by:
t
Fo  6.8
R2
where α is diffusivity coefficient and t the time.
To solve the analytical equation the following data was considered Tf=500℃, Ti=1000℃, R=1cm, α=1 and
h=1 Wm-2K-1. The results shown in Figure 6.9 are from the initial time up to 0.1 seconds with an increment
of 0.01 seconds. The results are shown in Figure 6.9 for temperature versus time and another for temperature
versus radius.
Figure 6.9: Results for the analytical equation for the prescribed boundary flux a) Temperature/Time b) Temperature/Radius
evolution.
122
o Abaqus results
As in the previous case an Abaqus simulation with prescribed flux at the cylinder boundary was performed.
In Figure 6.10 the results for Abaqus simulation with the different elements types are shown. Again the
linear elements have a good correlation in relation to the analytical solution and the quadratic elements
present the biggest deviation including heating.
a) b)
c) d)
Figure 6.10: Results for Abaqus simulation for the prescribed flux a) Linear triangular elements b) Linear quadrilateral elements c) Quadratic
triangular elements d) Quadratic quadrilateral elements.
o Empaktor Software
The same analysis was performed using the Empaktor software. Again three different cases the capacity
matrix were tested, a full capacity matrix, a lumped capacity matrix with zero non-diagonal values, and
lumped capacity matrix with non-diagonal pondered values. In Figure 6.11, Figure 6.12 and Figure 6.13
the three different studies are presented, respectively.
In the first case, the results show temperature oscillations in all the elements, however, they seams to
converge to the analytical solutions. The second case, the linear elements present a higher cooling than
expected and the quadratic elements present heating. In the third case, the linear elements present a good
agreement with the analytical solution but the quadratic elements presents unexpected heating.
123
a) b)
c) d)
Figure 6.11: Results for Empaktor software for the prescribed Flux a) Linear triangular elements b) Linear quadrilateral elements c) Quadratic
triangular elements d) Quadratic quadrilateral elements.
124
a) b)
c) d)
Figure 6.12: Results for Empaktor software for the prescribed flux using only the diagonal values for the capacity matrix a) Linear triangular
elements b) Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic quadrilateral elements.
125
a) b)
c) d)
Figure 6.13: Results for Empaktor software for the prescribed flux using a pondered capacity matrix a) Linear triangular elements b)
Linear quadrilateral elements c) Quadratic triangular elements d) Quadratic triangular elements.
It may be concluded that the linear elements using a lumped capacity matrix with non-diagonal values equal
to zero present the better agreement to the analytical solution. Therefore, this procedure will be adopted for
heat transfer with this type of elements.
In the previous test, as shown in Figure 6.1, quadratic and linear element meshes had the same number of
nodes but different element sizes. Next we will use elements with the same sizes as in Figure 6.14.
126
Figure 6.14: Mesh used for heat transfer validation for quadratic triangular and quadratic quadrilateral.
In Figure 6.15 the results for boundary prescribed temperature are shown. A full capacity matrix was used
for Empaktor software. Although the results present temperature oscillations they are not so significant in
relation to the results using the mesh of the Figure 6.14. Moreover, the results converge faster to the
analytical solutions. This results agree with the equation 5.2.
127
a) b)
c) d)
Figure 6.15: Results for prescribed temperatures a) Quadratic triangular elements in Abaqus b) Quadratic triangular elements Empaktor
software c) Quadratic quadrilateral elements in Abaqus d) Quadratic quadrilateral elements Empaktor software.
Figure 6.16 present the results for the mesh of the Figure 6.14 performing the example with prescribed flux
of the analogous conclusions can be drawn, i.e. the temperature oscillations are reduced when a finer mesh
is used.
128
a) b)
c) d)
Figure 6.16: Results for prescribed flux temperatures a) Quadratic triangular elements in Abaqus b) Quadratic triangular elements Empaktor software c)
Quadratic quadrilateral elements in Abaqus d) Quadratic quadrilateral elements Empaktor software.
6.2 Mechanical Validation

Here we intend to validate certain aspects related to the modelling of the mechanical behaviour of the
developed program. It is not an easy task to use commercial softwares for that purpose, as the constitutive
law used, with a high dependence on temperature and viscosity cannot usually be found implemented. One
important aspect has to do with incompressibility of the material flow, which is a source of numerical
problems in the finite element solutions. So, this restricted validation will concentrate on the ability of the
elements to deal with that problem, by avoid locking, and a simple reproduction of the process performed.
The ability of the elements to represent glass flow with no locking was done determining the quantity and
the quality of the displacement modes associated with the incompressibility constrain. Then, a simple
pressing test was performed in order to verify the displacements of each element type.
129
 Displacement Modes
Referring back to the previous chapter and the underlying approximation of the space of incompressible
modes we have seen that it is associated to the solution obtained by a mixed formulation having
displacements (velocities) and pressure as main variables:
K Q   v  f 
Q T 
0   p  0 
6.9

In which the incompressible constraint is represented by the second set of equations:
QT v = 0 6.10
The solution v should then lie in the subspace of incompressible deformations, which represents the null
T
space of Q . However, the finite dimensional approximation space may not reproduce all the possible
solutions. As discussed in the section 5.1.3, locking occurs when, for a given set of boundary conditions,
the expected solution does not belong to the subspace solutions. In this case the solution is the trivial one
to equation 4.22, i.e., v = 0 . Different elements will yield different approximations to that subspace,
therefore, different abilities to reproduce these type of situations.
Establishing the incompressible constraint at the integration points (Gauss points, ngauss).
 N i N i 
 kk    u j  0, j  1, ngauss (2D)
   
i 1, nodes 
 N Ni N i 
 kk   i
  u j  0, j  1, ngauss (axisymmetric)
  rj  
 i 1, nodes 
6.11
the basis modes, for each type of elements can be obtained. This is illustrated next.
o Linear Axisymmetric Triangle Element
In Figure 6.17 the deformation modes for the axisymmetric linear triangle are presented, for one point
Gauss integration. Five modes are obtained but it is clear that the modes obtained shows spurious
deformation patterns. Some of the deformation modes have movement through the axisymmetric axis,
which is unphysical.
130
In order to remove these spurious deformation modes, the solution was chosen so that the modes would be
orthogonal to them and also to the rigid body mode, shown in Figure 6.18. In practical terms, the
orthogonalization procedure intends to prevent these modes, as it is done in practice by imposing boundary
conditions and by restricting the horizontal movement of the nodes which are in the axisymmetric axis.
Figure 6.18: Orthogonalizing deformation modes for the linear triangular element using 1 Gauss point.
The Figure 6.19 shows the resulting deformation modes. Thus, two deformation modes were obtained.
131
The deformation modes for three and seven Gauss points are presented in Figure 6.20 and Figure 6.21
respectively. They coincide and only two modes, apart from the rigid body motion one, are obtained. No
spurious modes are present. In this case, only one rigid mode was included.
o MINI Axisymmetric Triangle Element
The MINI linear triangle axisymmetric element, using only one Gauss point depicts a larger basis of
deformation modes, as presented in Figure 6.22. It means that is better “equipped” to deal with
incompressibility. Nevertheless one of the modes, mode 5, is clearly not satisfying the incompressibity
condition and is, therefore, a spurious mode.
132
Figure 6.22: Axisymmetric deformation modes for the MINI linear triangular element using one Gauss point.
As in the case of the linear axisymmetric triangle element additional deformation modes are used, Figure
6.23, to prevent spurious and rigid body modes, by the same orthogonalization procedure.
Figure 6.23: Orthogonalizing deformation modes for the MINI linear triangular element using 1 Gauss point.
The Figure 6.24 shows the resulting deformation modes. Four deformation modes were obtained.
133
Figure 6.24: Axisymmetric deformation modes for the MINI linear triangular element using one Gauss point.
The MINI linear triangular element with three Gauss points, Figure 6.25, presents also a spurious mode,
which is removed if seven Gauss points are used, Figure 6.27, but the results are then the same as for the
linear triangular element.
Figure 6.25: Axisymmetric deformation modes for the MINI linear triangular element using three Gauss point.
If the previous procedure is orthogonalised, the solution against the spurious modes and rigid body modes
is performed, the result obtained is illustrated in Figure 6.26. These two deformation modes are the same
when calculation is made with seven Gauss points.
134
Figure 6.26: Axisymmetric deformation modes for the MINI linear triangular element using three Gauss point.
Figure 6.27: Axisymmetric deformation modes for the MINI linear triangular element using seven Gauss point.
o Quadratic Axisymmetric Triangle Element
In Figure 6.28 the quadratic axisymmetric triangle deformation modes, using one Gauss point are presented.
As expected some spurious modes are observed.
135
Figure 6.28: Axisymmetric deformation modes for the quadratic triangular element using one Gauss point.
Othogonalizing the solution against the modes of Figure 6.29 the seven resulting deformation modes are
shown in Figure 6.30.
Figure 6.29: Orthogonalizing deformation modes for the quadratic triangular element using one Gauss point.
136
Figure 6.30: Axisymmetric deformation modes for the quadratic triangular element using one Gauss point.
The quadratic axisymmetric triangle with three Gauss points still presents some spurious modes, Figure
6.31. They are eliminated when seven Gauss points are used but the basis dimension is reduced, Figure
6.33.
137
Figure 6.31: Axisymmetric deformation modes for the quadratic triangular element using three Gauss point.
The Figure 6.32 shows the resulting deformation modes. These modes were calculated taking in account
the rigid motion calculated in Figure 6.29. Thus, it were obtained five deformation modes without
movement throughout the axisymmetric axis. The obtained modes are the same modes calculated for the
seven Gauss points.
138
Figure 6.32: Axisymmetric deformation modes for the quadratic triangular element using three Gauss point.
Figure 6.33 Axisymmetric deformation modes for the quadratic triangular element using seven Gauss point.
The Figure 6.34 shows the resulting deformation modes after ortogonalization to the rigid body mode.
139
Figure 6.34 Axisymmetric deformation modes for the quadratic triangular element using seven Gauss point.
o Linear Axisymmetric Quadrilateral Element
In Figure 6.35 the deformation modes for this element using one Gauss point are shown.
The number of deformation modes is reduced if the number of Gauss points are 2x2, Figure 6.38 or 3x3,
Figure 6.40 , increasing the likelihood of locking to happen.
140
Figure 6.35: Axisymmetric displacement modes for the linear quadrilateral element using one Gauss point.
Orthogonalizing the solution against the modes represented in Figure 6.36 the four modes represented in
Figure 6.37 are obtained.
141
Figure 6.36: Orthogonalizing deformation modes for the linear quadrilateral element using one Gauss point.
Figure 6.37: Axisymmetric deformation modes for the linear quadrilateral element using one Gauss point.
142
Figure 6.38: Axisymmetric deformation modes for the linear quadrilateral element using 2x2 Gauss point.
If the same type ortogonalization is performed only one mode is obtained, Figure 6.39.
Figure 6.40 shows the deformation modes for the 3x3 Gauss points. In this case only two deformation
modes, apart from the rigid bode mode, were obtained. With the orthogonalization procedure only one
mode is obtained, Figure 6.41.
143
o Quadratic Axisymmetric Quadrilateral Element
In Figure 6.42 the deformation modes for quadratic axisymmetric quadrilateral element using one Gauss
point are shown. As it can be seen many spurious modes are present. This number is reduced with a 2x2
integration rule, but still some spurious modes are present, Figure 6.45.
144
Figure 6.42: Axisymmetric deformation modes for the quadratic quadrilateral element for one Gauss point
The deformation modes were orthogonalized against the modes in Figure 6.43.
Figure 6.43: Orthogonalizing deformation modes for the quadratic quadrilateral element using one Gauss point.
The Figure 6.44 shows the resulting deformation modes.
145
146
Figure 6.44: Axisymmetric deformation modes for the quadratic quadrilateral element using one Gauss point.
147
Figure 6.45: Axisymmetric deformation modes for the quadratic quadrilateral element for 2x2 Gauss point.
The Figure 6.46 shows the resulting deformation modes after the orthogonalization procedure as in the
previous cases.
148
Figure 6.47 show the deformation modes for the 3x3 Gauss points. In this case seven deformation modes
were obtained. If rigid deformation modes are not include only six deformation modes should be
considered, Figure 6.48.
149
150
The Figure 6.49 show the deformation modes for the 4x4 Gauss points. In this this case six deformation
modes were obtained, including the rigid body mode.
151
 Plunger Compression Test
In this section a simple compression model was tested. A rectangular domain was considered. The material
within the rectangle can only flow out through the right side. Furthermore, on top and on the bottom sides,
stick conditions are prescribed, whereas the left side represents the symmetry-axis. The scheme of the
testing is presented in Figure 6.50. Structured meshes of triangular and quadrilateral elements were
constructed by partitioning the domain into different elements, Figure 6.51. The height of the domain is
equal to 10cm.
152
Figure 6.50: Scheme for the compression test.
Figure 6.51: Mesh used for compression test a) Linear triangular elements b) Quadratic triangular elements c)
linear quadrilateral elements d) Quadratic quadrilateral elements.
The results of the compression test with displacement/time are presented in Figure 6.52. The selective
reduce integrations cases are referred as RI. It can be seen that the linear elements show a locking behaviour
if full integration is used and that it can be prevented resorting to selective reduce integration. Quadratic
elements do not show locking evidence.
153
Figure 6.52: Displacement results for the compression test for the different elements type.
 Pressure Compression Test
An important case, not reported anywhere, as far as we know, must be taken when dealing with these
axisymmetric selective reduced integrated elements when pressure loads are applied, especially near the
symmetry axis.
The problem is illustrated in the following example, representing a cylinder of an incompressible material
(glass at high temperature) with an applied pressure. An axisymmetric linear quadrilateral element is used,
as in Figure 6.53. Any movement should be prevented due to the imposed boundary conditions and to the
fact that the material is incompressible. However, if the correct full integration is used along the boundary
where the pressure is applied, two different nodal force values will be obtained: a bigger value for the outer
node and a smaller value for the inner node. This may activate the deformation mode, as in Figure 6.54a),
that exists in the underlying subspace of incompressible modes associated to this element, which is not
physically acceptable. To alleviate this problem a simple solution is to evaluate the nodal forces using also
a reduced integration technique for integrate the loads and recover the physical solution, as shown in Figure
6.54b).
154
Figure 6.53: Scheme to validate the prescribed temperature at the boundary cylinder.
a) b)
Figure 6.54: a) Pressure teste for one linear quadrilateral element using two Gauss points to integrate the force b)
using one Gauss point to integrate the force.
After the thermal and mechanical analyses performed, the linear quadrilateral element with selective
reduced integration (one Gauss point) for the incompressibility term was mostly adopted, as it was
considered the most balanced element, both in controlling artificial inner heating effects as well as the
possible locking effects associated with the incompressible law of glass at high temperatures.
155
Chapter 7 Computer Program
CHAPTER 7
7 Computer Program
The finite element method has become an essential analysis tool in many industries as a result of the
increase in computer power and the breakthrough in numerical computation technologies. The method has
increased in both reputation and performance with the development of user-friendly pre and postprocessing
software. Effective use of the method was enhanced with the introduction of low cost computer graphics
systems that brought both the possibility of visually displaying both the model and the computed results
[42].
The requirement for a software environment to provide an integrated computational tool to support
engineering design and manufacture is ever growing at the industry level. A key component of an integrated
environment is the use of computational engineering to assist and support solutions for complex design
[42]. In a software package the analysis must be accompanied by preprocessor and postprocessor tools [43].
In this direction a preprocessing and postprocessing which was developed to support the present work is
described here.
7.1 Software environment

In order to design the software environment several technical requirements have been considered. These
requirements are as follows: graphic engine, data structure, input and output phases.
7.1.1 Graphic Engine

The graphical software packages include slices (sectioning), hidden lines, present or removed, views from
various directions, windowing, zoom, and perspectives. The graphic engine is developed by a set of
graphical features. Geometry and results in this work have been drawn and visualized using customized
GID software.
7.1.2 Preprocessor
In finite element preprocessing, the geometric model are the basis for defining a finite element mesh. The
main objective of the geometric model is to display geometry objects. This allows an interaction control
156
over details, such as point of view, appearance of surfaces and lines, etc. The software used should handle
any number of objects and control over them [42].
The finite element preprocessing defines the types of meshing algorithms that can be used, the type of
elements that can be selected the resulting mesh density, the quality of the mesh and the feasibility of
creating a finite element mesh. The geometric model can be used to associate the mesh with the physical
attributes such as material properties, element properties, boundary conditions and loads. When the mesh
is modified these attributes are assigned to the geometry and remain constant. This implies that a direct
correspondence between the geometry and the mesh must be maintained. As soon as the geometry model
is correctly defined it can be discretized to create finite elements that contains the appropriate type of
elements required for the specific analysis. This process is known as mesh generation [42].
In Figure 7.1 a view of the preprocessor menu and appearance is depicted. The title bar indicates the name
of the model database. The menu bar contains all the available menus; the menus give access to all the
functionality in the product. Different menus appear in the menu bar depending on which module you
selected from the context bar. The toolbar provides quick access to items that are also available in the
menus. The Model Tree provides you with a graphical overview of your model and the objects that it
contains, such as parts, materials, steps, loads, and output requests. The toolbox area displays tools in the
toolbox that are appropriate for that module. The toolbox allows quick access of the module functions that
are also available from the menu bar. Viewports are windows to displays your model. The command line
interface is used to make an action when a task is applied.
Figure 7.1: Layout of the preprocessor.
The preprocessor is used to define the element types, material properties, initial conditions, gob shape, the
mould shape, machining parameters, process parameters. The boundary conditions and loads are calculated
automatically by the software, Figure 7.2. The geometries of the moulds are essential as an input for the
simulation.
157
Figure 7.2: Option to model the simulation.
7.1.3 Postprocessor
Postprocessing is designed into two distinct categories; graphical postprocessing and postprocessing text
files, both forms are extremely useful in interpreting the results of finite element analysis. The combined
option makes the finite element analysis more productive. The graphical postprocessing allows to quickly
assess the behaviour of the model. The postprocessing text files provides detailed numerical data necessary
to fully understand the results of finite element analysis [42].
The software developed produces lists and tables of many types of results that are normally displayed on
the screen. However, it is possible to save the results to a file to be later analysed or included in a report.
The general postprocessor is used to review results over the entire model. Using the postprocessor contour
displays the results to review and interpret the results. It is able to display temperatures and displacements
contours. In Figure 7.3 and Figure 7.4 it can be seen the overview of the postprocessor software. Also, the
software is able to create a movie of the simulation. At the end of the simulation a thickness distribution is
generated, Figure 7.5.
158
Figure 7.3: Software postprocessor overview.
Figure 7.4: Software postprocessor.
159
Figure 7.5: Software postprocessor thickness distribution.
7.2 Software Core

The software is composed by two main executable files. These two executable files perform different tasks.
Meshing and remeshing, that must be performed periodically due to high deformation present in the glass
forming is deal automatically by an executable file, the preflow file. The second executable file, called flow,
performs the finite element calculation.
Figure 7.6 show the layout of the software. When the user creates a model and runs the simulation, the
preprocessor interface creates a file (*.epk) that contains the user model. Then the model is loaded by the
preflow which reads the data creates the first mesh. When preflow is ready, it passes the first calculation
input for flow. At that point the calculation is done. When the mesh is too distorted flow writes the results
and an update file and transfers the data to preflow to deal it. At this point a cycle is performed at the point
the simulations ends.
The programing language used to write the software was FORTRAN
160
Figure 7.6: Software core diagram.
7.3 Computer Program Files

Software allows a "check run" that stops of solving global equations or even generating them. The check
run applies automatic data checks and may also estimate storage requirements and solution time that will
be required by the actual analysis.
When a simulation starts texts (ASCII) files are created. This files are needed to share information between
the software and the remeshing tool and also gives information to the user (result, warning, error, debugging
files, etc). The software have a restart function that restart the simulation from a desirable process stage and
a file is also created. The Table 4 presents the files generated by the software.
Type of file File extension Description

Input model data *.dat Finite element model data
Input simulation file *.mef Input file for several solutions during the simulation
Result File *.res Result file for several solutions during the simulation
Warning File *.war Save warnings during simulation
Error File *.err Save the error description
Restart File *.res Restart the simulation from a previous stage
Remeshing File *.msh File containing a mesh from a remeshing tool
Update File *.upd File containing data at the end of the simulation
Mesh Information *.post.msh File contain mesh coordinates and connectivity
Information file *.info Information data during the simulation
Logging file *.log Information registry during the simulation
Table 7.1: Files created during the simulation.
161
7.3.1 Input File

The input file developed contains all information about the problem to be analysed such as control
information (type of analysis, etc.), material properties, mesh (element types, nodal coordinates,
connectivities), applied force and heat flux data, boundary conditions (e.g., prescribed flux, displacement
restrictions), initial conditions and the parameters of the solver as well as various control parameters
required in the finite element analysis. The result is an input file which is used then by the computational
program, Figure 7.7.
162
Figure 7.7: Input file example. (Cont.)
7.3.2 Result File

After running the program successfully, a second file (output file) which contains the simulation results is
generated. Furthermore, a file (postscript file) is also generated so that the user can view the contour plots
graphically. The program also produces an output file that is intended to be an input file for a postprocessor.
The contents and structure of this file are shown in Figure 7.8.
Figure 7.8: Example of the results file.
163
7.3.3 Warning File

Listing of all warning messages
7.3.4 Error File

The error file list all error messages that can occur during the simulation. This file is generated during the
simulation. The error encountered may stop the simulation. The file contains the complete database at the
time of its creation.
7.3.5 Restart File

During a solution a restart file is saved containing the current solution data in a file. A restart file is always
created when a process stage is successful completed. When the restart file is called the finite element
analysis is restarted from the position where the restart file was created.
7.3.6 Remeshing Files

The remeshing files are a set of files necessary to perform the remeshing. These set of files are the input
files from the remeshing tool and the output. The input data are the boundary consisting of the coordinates,
connectivity and parameter from the remeshing tool. The output is a file containing the new mesh.
7.3.7 Update File

The update file includes the information that connects the solver and a data management file. This
information is mostly on the actual displacements and temperatures at the end of each calculation, Figure
7.9.
Figure 7.9: Example of the update file.
7.3.8 List File

The list file is created in order to write the information during a simulation. The file is helpful for debugging
while performing the calculation or even when an error was encountered. Figure 7.10 shows an example.
It can be seen that besides the simulation data the file contains the convergence ranges for each step and
other numerical information, errors, etc.
164
Figure 7.10: Example of the list file.
165
Chapter 8 Glass Containers Simulation Cases
CHAPTER 8
8 Glass Containers Simulation Cases
In this chapter real forming operations were modelled, using, real values for the various physical
parameters.
Two different forming processes of glass containers were used, namely blow/blow and press/blow
processes. Five different cases of blow/blow process and eight in press/blow processes were performed.
Each case and the respective description is shown in the Table 8.1.
Process Case Description

Case 1 Kingfisher650ml
Case 2 Bacardi
Case 3 Ardagh750ml
Case 4 Senzo750ml
Case 5 Wine bottle750ml
Case 1 NNPB
Case 2 Ambev343ml
Case 3 Malta7oz
Case 4 Parkam
Case 5 Sgbeer
Case 6 Gallipot
Case 7 Fruitz
Case 8 President650ml
Table 8.1: Simulation cases for the blow/blow and press/blow process.
Usually to choose the process to produce a glass bottle or other container several indicators have to be
accounted for. One of the indicators is the size of the container. For a small container in size the narrow
neck press/blow process (NNPB) is the advisable. On other hand, blow/blow processes are a preferable
choice for big containers, Figure 8.1.
166
Indifferent
Narrow Neck P&B Blow & Blow
20 40 60 80 100 120 140 160 180 200 220 240 260
10 30 50 70 90 110 130 150 170 190 210 230 250
Figure 8.1: Process selection for container height (cm).
To start the simulations a gob has to be created. As discussed in the section 3.4.1 the gob can have different
shapes. These shapes can be drawn at the graphical interface. However, the gob is usually very similar to a
cylindrical shape, thus, the software can create the gob using the mass of glass and some parameters to
define the shape of the cylinder, Figure 8.2.
Figure 8.2: Gob shape calculation.
The diameter of the cylinder is calculated by,
1
 Weight ( g )  Gobshapefactor  3 5
 (mm)  10    25.4   8.1
    16 
and the length,
  Weight (mm)    . 3  
 
3   
  Density ( g / mm )   6  
Lenght (mm)    8.2
  2
4
The several simulations reported were performed using the parameters of the Table 8.2. . These parameters
may be given or altered by the user. Those parameters comprise the gob definition, the process air pressures,
the equipment temperatures, the machine timings and the material parameters.
167
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8
Gob Weight [Kg] 0.415 0.450 0.400 0.580 Gob Weight [Kg] 0.170 0.180 0.136 0.152 0.189 0.285 0.200 0.200
Gob Definition
Gob Definition
Gob Shape Factor 0.4 0.4 0.4 0.4 0.4 Gob Shape Factor 0.6 0.6 0.6 0.6 0.55 0.6 0.7 0.6
Gob Initial Temp [℃] 1135 1070 1150 1132 1140 Gob Initial Temp [℃] 1170 1170 1162 1185 1190 1153 1151 1139
Gob Inital Speed [m/s] 6.5 6.5 6.5 6.5 6.5 Gob Inital Speed [m/s] 6.5 6.5 7.0 6.5 6.5 6.5 6.5 6.5
Gob Initial Position [mm] 6.5 6.5 6.5 6.5 6.5 Gob Initial Position -10 -10 -10 -10 -10 -10 -10 -10
Pressures
Settle Blow 2.5 2.8 2.8 2.5 2.4 Final Blow [bar] 2.5 2.0 1.5 2.4 1 2.5 2.5 2.5
Process
Pressures
Process
[bar]
Counter Blow 2.5 2.5 2.5 2.5 2.2 Vacuum [bar] 0.5 0.5 0.5 0.5 0.0 0.5 0.0 0.5
Final Blow 2.5 2.0 2.0 2.00 1.2 Plunger Force [N] 400 400 400 400 350 900 900 900
Temperatures
Vacuum 0.5 0.6 0.6 0.00 0.5 NeckRing 550 420 450 410 420 550 400 400
Equipment
Temperatures
NeckRing 550 550 420 450 420 Blank Mould 450 450 470 450 470 450 460 450
Equipment
[℃ ]
Blank Mould 450 450 470 450 465 Baffle 450 450 450 410 450 450 460 450
[℃ ]
Baffle 450 450 450 450 450 Plunger 500 550 500 410 550 550 450 480
Plunger 550 450 550 550 500 Blow Mould 450 480 480 475 480 480 480 480
Blow Mould 480 450 480 480 480 Machine Speed [BPM] 450 326 375 576 288 203 195 142
Machining
Timming
Machine Speed [BPM] 360 120 136 22 170 Number of Sections 12 10 10 16 12 8 8 8
Machining
Timming
Number of Sections 10 8 8 6 12 Gobs per Section 3 3 3 3 2 2 2 2

Gobs per Section 3 2 2 1 2 Cavity Rate 12.5 10.9 12.5 12.0 12.0 12.7 12.19 8.87
Cavity Rate 12.0 7.5 8.5 3.7 7.1 Cycle Time per Bottle [s] 4.80 5.52 4.80 5.00 5.00 4.73 4.92 6.76
Cycle Time per Bottle [s] 5.00 8.00 7.06 16.36 8.47 Oxidized
Material
Glass Color Flint Flint Flint Flint Flint Flint Green Amber
Material
Glass Color Flint Flint Flint Flint Flint

Theorical Thickness [mm] 1.0 2.5 2.5 5.0 3.5 Theorical Thickness [mm] 1.0 1.0 2.5 1.0 2.0 1.0 1 2.5
Hot Glass Density [Kg/m3] 2730 2450 2430 2430 2430 Hot Glass Density [Kg/m3] 2430 2430 2430 2500 2430 2430 2387 2430
Table 8.2: Simulation variables for blow/blow (left) and press/blow (right) processes.
In Table 8.3 and Table 8.4 the machine timing is shown for the cases of the blow/blow process and
press/blow process. The machine timing is measured in degrees as it is usual in the shop floor. Such measure
can be difficult to understand but, the machine timing may be converted into seconds so that it can be used
in the simulation. The cycle time is calculated as:
Gob per Section×60

Cycle Time [s]= Number sections× 8.3
Machine Speed[BPM]
thus, the machine timing is converted to seconds as
MachiningTiming[deg]×Cycle Time[s]
MachiningTiming[s] = 8.4
360
168
Machine Timings Machine Timings

# Name Deg Initial Time [s] End Time [s] # Name Deg Initial Time [s] End Time [s]
1 Blank Close 0 0.000 0.306 1 Blank Close 0 0.000 0.111

2 Gob In 22 0.306 0.764 2 Gob In 5 0.111 1.667
3 Settle Blow on 55 0.764 1.083 3 Settle Blow on 75 1.667 2.667
4 Settle Blow off 78 1.083 1.153 4 Settle Blow off 120 2.667 2.778
5 Plunger Down On 83 1.153 1.500 5 Plunger Down On 125 2.778 2.889
6 Counter Blow On 108 1.500 2.986 6 Counter Blow On 130 2.889 4.444
7 Counter Blow Off 215 2.986 3.264 7 Counter Blow Off 200 4.444 5.000
8 Blank Open 235 3.264 3.958 8 Blank Open 225 5.000 5.889
9 Start Invert 285 3.958 4.514 9 Start Invert 265 5.889 7.667
10 Mould Close 325 4.514 4.792 10 Mould Close 345 7.667 7.889
11 Neck Ring Open 345 4.792 5.833 11 Neck Ring Open 355 7.889 8.022
12 Final Blow On 60 5.833 7.222 12 Final Blow On 1 8.022 8.067
13 Final Blow Off 160 7.222 7.292 13 Final Blow Off 3 8.067 11.667
14 Mould Open 165 7.292 14 Mould Open 165 11.667 16.000
Case 1 Case 2


2 Gob In 15 0.294 0.490 2 Gob In 5 0.144 1.007
3 Settle Blow on 25 0.490 0.784 3 Settle Blow on 35 1.007 1.439
4 Settle Blow off 40 0.784 1.627 4 Settle Blow off 50 1.439 1.583
5 Plunger Down On 83 1.627 1.863 5 Plunger Down On 55 1.583 2.590
6 Counter Blow On 95 1.863 3.627 6 Counter Blow On 90 2.590 4.748
7 Counter Blow Off 185 3.627 3.824 7 Counter Blow Off 165 4.748 5.468
12 Final Blow On 40 7.843 10.196 12 Final Blow On -40 9.209 14.101
14 Mould Open 165 10.294 14 Mould Open 135 14.245 20.720
Case 3 Case 4
Machine Timings
# Name Deg Initial Time [s] End Time [s]
1 Blank Close 0 0.000 0.863

2 Gob In 30 0.588 1.381
3 Settle Blow on 48 0.941 2.245
4 Settle Blow off 78 1.529 2.360
5 Plunger Down On 82 1.608 3.281
6 Counter Blow On 114 2.235 5.899
7 Counter Blow Off 205 4.020 6.158
8 Blank Open 214 4.196 7.482
9 Start Invert 260 5.098 8.346
10 Mould Close 290 5.686 9.439
11 Neck Ring Open 328 6.431 1.727
12 Final Blow On 60 8.235 6.417
13 Final Blow Off 223 11.431 7.108
14 Mould Open 247 11.902 10.360
Case 5
Table 8.3: Machining timing for the blow/blow cases being respectively the case 1, case 2, case 3, case 4 and case 5.
169


2 Gob In 35 0.467 0.533 2 Gob In 5 0.077 0.333
3 Baffle On 40 0.533 0.000 3 Baffle On 25 0.383 0.000
4 Remove Plug 0 0.000 0.800 4 Remove Plug 0 0.000 0.747
5 Plunger Up 60 0.800 2.000 5 Plunger Up 56 0.859 1.667
6 Plunger Down 150 2.000 2.067 6 Plunger Down 125 1.917 1.800
11 Final Blow On 75 5.800 7.267 11 Final Blow On -40 4.907 6.627
13 Mould Open 195 7.400 9.600 13 Mould Open 175 8.203 9.600
Case 1 Case 2


2 Gob In 5 0.067 0.800 2 Gob In 17 0.236 0.556
Case 3 Case 4


2 Gob In 5 0.069 0.625 2 Gob In 5 0.066 0.460
Case 5 Case 6
170

2 Gob In 35 0.478 0.820 2 Gob In 5 0.094 0.488
Case 7 Case 8
Table 8.4: Machining timing for the press/blow cases being respectively the case 1, case 2, case 3, case 4, case 5,
case 6, case 7 and case 8.
The air temperature was considered fixed in every stage. However, the air temperature can vary for different
stages in the process according the conditions, as shown in, Table 8.5.
Process Stages Air Temperature

Blank Close 45
Gob In 45
Settle Blow on 50
Settle Blow off 350
Plunger Down On 450
Blow Blow
Counter Blow On 500

Counter Blow Off 800
Blank Open 60
Start Invert 40
Mould Close 60
Neck Ring Open 200
Final Blow On 250
Final Blow Off 300
Mould Open 150
Blank Close 45
Gob In 45
Baffle On 300
Remove Plug 350
Plunger Up 500
Press Blow
Plunger Down 800

Blank Open 60
Start Invert 40
Mould Close 60
Neck Ring Open 200
Final Blow On 250
Final Blow Off 300
Mould Open 150
Table 8.5: Air temperature considered in the different stages.
The chemical compositions of the glass for the different cases are presented in the Table 8.6. The material
properties, needed for the simulation, are calculated using the chemical properties. The material properties
vary with the temperature, as illustrated in the Table 8.7 and Table 8.8 for blow/blow process and
press/blow process.
171
Material Percentage [%] Case 1 Case 2 Case 3 Case 4 Case 5

Material Percentage [%] Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8
Si02 71.742 71.953 72.490 72.490 71.060
Si02 71.742 72.000 71.200 70.110 71.06 72.120 71.635 71.634
Al2O3 1.771 1.343 1.310 1.310 1.900
Al2O3 1.771 1.450 2.000 1.900 1.90 1.400 1.675 1.675
Fe2O3 0.051 0.067 0.051 0.051 0.470
Fe2O3 0.051 0.280 0.500 0.225 0.47 0.075 0.357 0.357
CaO 10.198 11.250 7.470 7.470 11.000 CaO 10.198 10.900 11.000 10.400 11.00 11.000 10.890 10.890
MgO 2.402 2.123 4.480 4.000 2.000 MgO 2.402 0.800 2.000 0.000 2.00 1.950 0.258 0.258
BaO 0.015 0.050 0.017 0.017 0.000 BaO 0.015 0.020 0.000 13.750 0.00 0.000 0.100 0.100
Na2O 12.841 12.430 13.240 13.240 12.210 Na2O 12.841 13.900 12.000 0.050 12.21 12.090 14.400 14.400
K2O 0.578 0.527 0.710 0.710 0.690 K2O 0.578 0.340 1.000 0.000 0.69 0.710 0.670 0.670
LiO2 0.001 0.000 0.000 0.000 0.000 LiO2 0.001 0.055 0.000 0.180 0.00 0.000 0.000 0.000
Cr2O3 0.001 0.003 0.002 0.002 0.270 Cr2O3 0.001 0.013 0.250 0.040 0.27 0.004 0.006 0.006
SO3 0.316 0.131 0.182 0.182 0.000 SO3 0.316 0.050 0.050 0.000 0.00 0.000 0.070 0.070
TiO2 0.023 0.000 0.028 0.028 0.000 TiO2 0.023 0.035 0.000 0.000 0.00 0.000 0.000 0.000
B2O3 0.000 0.000 0.000 0.000 0.000 B2O3 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
NiO 0.001 0.000 0.000 0.000 0.000 NiO 0.001 0.000 0.000 0.000 0.00 0.000 0.000 0.000
F 0.008 0.000 0.000 0.000 0.000 F 0.008 0.000 0.000 0.000 0.00 0.000 0.000 0.000
Cl 0.027 0.000 0.000 0.000 0.000 Cl 0.027 0.000 0.000 0.000 0.00 0.000 0.000 0.000
PbO 0.009 0.012 0.002 0.002 0.000 PbO 0.009 0.000 0.000 0.000 0.00 0.000 0.000 0.000
ZnO 0.003 0.000 0.000 0.000 0.000 ZnO 0.003 0.000 0.000 0.000 0.00 0.000 0.000 0.000
CdO 0.000 0.000 0.000 0.000 0.000 CdO 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
CuO 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
CuO 0.000 0.000 0.000 0.000 0.000
CoO 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
CoO 0.000 0.000 0.000 0.000 0.000
MnO 0.012 0.000 0.000 0.000 0.00 0.000 0.000 0.000
MnO 0.012 0.000 0.000 0.000 0.000
ZrO 0.000 0.000 0.000 0.000 0.00 0.000 0.010 0.010
ZrO 0.000 0.000 0.000 0.000 0.000
V2O5 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
V2O5 0.000 0.000 0.000 0.000 0.000 Se 0.002 0.000 0.000 0.000 0.00 0.000 0.000 0.000
Se 0.002 0.000 0.000 0.000 0.000 S 0.000 0.000 0.000 0.000 0.00 0.000 0.000 0.000
S 0.000 0.000 0.000 0.000 0.000
Table 8.6: Material chemical composition for different simulations cases in blow/blow and press/blow processes.
Table 8.7: Material behaviour for different simulations cases in blow/blow process.
172
Table 8.8: Material behaviour for different simulations cases in press/blow process.
8.1 Blow/Blow Process

In this chapter complete simulations of blow/blow process are presented, for the cases referred in the
previous chapter. The results are shown at all stages.
8.1.1 Gob Loading

The gob loading is the first stage of the blow/blow process. At the beginning the gob falls from the feeder
to the interior of the moulds. At this stage it was admitted that the glass can slip along the moulds edges in
order to reproduce the possible slipping effect resulting from the initial velocity.
For case 1 the shape and temperature contour of the glass are plotted in Figure 8.3. Since the blank mould
has a narrowing zone and the gob diameter is higher than the diameter of the blank near the neckring, the
first contact between the glass and the mould is in the blank mould wall. Note that the initial time does not
correspond to the initial machine time in the Table 8.3, this is due to the fact that the time that the gob
spends due to the gob falling from the feeder until the mould contact. Then, the glass slips, by gravity, along
the blank mould until it reaches the plunger. Once the glass contacts the plunger, the neckring begins to
form. In addition, it can be seen that at the end of this stage the glass container neck is almost filled. Also,
in Figure 8.4 it can be seen at the boundary temperature of the glass. It can be seen that the glass that is in
contact with the mould is cooling faster than the glass that is cooling to the air.
For case 2 the shape and temperature contours of the glass are plotted in Figure 8.5. Also, the first contact
between the glass and the mould was in the blank mould wall. Again the glass slips along the blank mould
until reaching the plunger. When the glass contacts the plunger, the neckring begins to form. As the
neckring of this container is very narrow the bottle neck is not completely filled at this stage. In Figure 8.6
the boundary temperature for the case 2 is presented. Similarly, the glass cools faster near to the mould then
to the air. Additionally, in Figure 8.7 it can be seen the displacement values, while the glass slips around
the edge of the blank mould. The results shows clearly that when the blank mould becomes narrower the
173
glass tends to slow down. Also, the Figure 8.8 and Figure 8.9 shows in detail the beginning of the neckring
filling, where the glass flows into the neckring due gravity force until finishing the stage.
For case 3 the shape and temperature contours of the glass are plotted in Figure 8.10. Similar results in the
previous cases are observed. In Figure 8.11 the boundary temperatures for the case 3 are presented. For
Case 4 the shape and temperature contours of the glass are plotted in Figure 8.12. This bottle is very
geometric different from the last examples, however, similar results were obtained. In the boundary
temperatures for the case 4 are presented, Figure 8.13. It can be observed that due to this stage is longer
than in the previous cases the glass presents a more evident cooling. For case 5 the shape and temperature
contour of the glass are plotted in Figure 8.14. The result are quite similar to the last ones. Figure 8.15 the
boundary temperatures for the case 5 are presented.
174
Figure 8.3: Temperature distribution in gob loading for the case 1 for 0.334s, 0.342s, 0.427s and 0.763s.
Figure 8.4: Boundary temperature for gob loading stage for case 1.
175
176
Figure 8.7: Displacement module in gob loading for case 2 for 0.368s, 0.375s and 0.381s.
177
Figure 8.8: Displacement module with the neckring detail in gob loading for case 2 for 0.3877s, 0.3973s 0.4165s.
0.4558s 0.4842s 0.5414s
Figure 8.9: Displacement vector with neck ring detail in gob loading for case 2 for 0.3877s, 0.3973s 0.4165s
0.4558s 0.4842s 0.5414s
178
Figure 8.10: Temperature distribution in gob loading for case 3 0.326s, 0.335s, 0.369s and 0.490s.
179
180
181
8.1.2 Settle Blow

The settle blow is the stage that follows the gob loading. In order to fill completely the neckring a top
pressure is applied.
For case 1 the shape and temperature contours of the glass are plotted in Figure 8.16. In this stage a pressure
is applied to force the glass to fill completely the neckring zone, it can be seen that in the end of the stage
the neckring is completely filled. In addition, it can be noted that the top of the gob, due to the air pressure,
becomes slighted flattened. In Figure 8.17 it can be seen that the glass in contact to the moulds cools faster.
Also inside the neckring and near to the corner it can be seen that the temperature drops faster relatively to
the zone where there is more glass.
For case 2 the shape and temperature contours of the glass are plotted in Figure 8.18. The filling of the
neckring is more evident, as well as the flattening of the top of the gob. Also, in Figure 8.19 it can be seen
that the temperature along the glass in contact with the mould is constant. However, the temperature near
the neckring varies significantly in the zones near to the neckring corner. In addition, in Figure 8.20 and
Figure 8.21 the displacement value (meaning its norm) and the displacement vector are shown in order to,
in detail, show the filling movement of the neckring.
For case 3 the shape and temperature contours of the glass are plotted in Figure 8.22. The results for this
case are very similar with the case 2. In Figure 8.23 the boundary temperatures are represented and the
pattern is analogous to what is verified in case 2.
For case 4 the shape and temperature contours of the glass are plotted in Figure 8.24. The longer cooling
effect in the glass in contact with the mould is more visible here. In Figure 8.25 representing the boundary
temperature.
For case 5 the shape and temperature contours of the glass are plotted in Figure 8.26 and the boundary
temperature in Figure 8.27. Again, similar comments to the previous one apply here.
182
Figure 8.16: Temperature distribution in settle blow for the case 1 for 0.796s, 0.934s and 1.083s
Figure 8.17: Boundary temperature for settle blow stage for case 1.
183
Figure 8.18: Temperature distribution in settle blow for the case 2 for 0.555s, 0.666s and 1.333s.
184
Figure 8.20: Displacement module with the neckring detail in gob loading for case 2 for 0.555s, 0.744s 0.850s.
1.005s 1.154s 1.333s
Figure 8.21: Displacement vector with the neckring detail in gob loading for case 2 for 0.555s, 0.744s 0.850s.
1.005s 1.154s 1.333s
185
186
187
188
8.1.3 Corkage Reheat

After the settle blow stage the plunger is removed to allow the corkage to reheat in the plunger area, i.e. in
the glass nearby the plunger, therefore, to “soften” this zone just before the counter blow pressure is applied.
The baffle tool is also set in the top of the blank mould. The results from this stage are shown in Figure
8.28 for all the cases. That effect is more evident in case 4.
Case 1 Case 2 Case 3
Case 4 Case 5
Figure 8.28: Temperature distribution in corkage reheat for the case 1 in 1.500s, for the case 2 in 2.044s, for the
case 3 in 1.863s, for the case 4 in 4.090s and for the case 5 in 2.682s.
189
8.1.4 Counter Blow

Air pressure is applied in the bottom part of the glass, pushing the glass to the baffle zone and forming the
parison. The pressure used must be suited to the particular size of the bottle, that is, the larger the quantity
of glass the greater the pressure.
For case 1, the shape and temperature contours of the glass are plotted in Figure 8.29. The gob starts to rip
a hole in the zone where the glass has a higher temperature creating a wall against the blank mould. The
thickness of the wall depends on the cooling rate between the blank moulds and the glass. During the
counter blow the wall of the parison is cooling in contact with the mould. After the glass reaches the bottom
of the blank mould, filling the baffle zone, the parison is completely formed. For a short time air pressure
continues to push the glass against the mould with the only purpose to cool the parison. At the end of the
stage the parison is formed and cooled enough to pass to other stages.
In Figure 8.30 the temperatures at the glass boundary are represented. It is clear that at the boundary that
was in contact with the mould the glass is cooler than at the boundary where the air pressure was applied.
The boundary zone that had contact with the mould, since the gob loading stage is cooler than the glass
zone that only contacted the mould during the counter blow stage. The same happens with the boundary
where the air pressure was applied. The boundary contacting the mould is around 730℃ and the boundary
contacting the air is around 950℃.
At the end of the stage, the glass has an almost constant wall thickness, Figure 8.30. The bottom part of the
container presents a higher thickness. The cooling rate between the glass and moulds during this stage can
influence the final thickness. If the glass cools more than it should the parison will have a larger wall
thickness and a smaller thickness at the bottom zone and if the cooling is enough the opposite happens. So,
it can be concluded that the cooling rate influences dramatically the final parison shape. Off course, also
other parameters play here an important role as conductivity, capacity, mould temperatures, etc.
Additionally, in Figure 8.31 displacement contours are represented during the counter blow stage. The
displacement is higher just above the counter blow pressure and while the glass is spread around the edges
of the blank mould the displacement increases. When the glass hits the baffle the displacement becomes
almost null when the parison is formed.
For case 2, the shape and temperature contours of the glass are plotted in Figure 8.32 Again, the air pressure
pushes the material to the baffle in order to form the parison. Here the baffle has a complex shape, so the
remaining material must accommodate in the baffle gaps. After the parison is formed, the exterior boundary
cools in contact with the moulds and the interior boundary with the air. In Figure 8.33 it can be seen that at
the parison the exterior boundary temperature is around 870℃, while the interior is around 1050℃.
In addition, the thickness at the end of the stage is presented. Also, in Figure 8.34 and Figure 8.35 it can be
seen the evolution of the displacement while the counter blow stage forms the parison, as well as the detail
of the displacement vector when the baffle is filled. The results are very similar to the case 1, but because
the baffle has here a more complicated shape, a detail of the displacement vector is shown to illustrate the
filling of the edges of the baffle.
For case 3, the shape and temperature contours of the glass are plotted in Figure 8.36. In this case the
behaviour is very similar to the case 3. In Figure 8.37 the boundary temperatures are represented. The
exterior parison wall ends this stage with around 770℃, while the interior with around 1100℃. It also can
be noted that the neckring zone has the lower temperatures, as it is a thin zone and has more area in contact
with the mould.
For case 4, the shape and temperature contours of the glass are plotted in Figure 8.38. The cooling between
glass and moulds are more evident due to the high stage time. Air pressure pushes the hot glass to the baffle.
When the parison is formed the glass cools around the exterior boundary at a higher rate than in the interior
Figure 8.39 the boundary temperatures are represented.
190
For case 5, the shape and temperature contours of the glass are plotted in Figure 8.40. Quite similar results
were obtained when comparing to the previous examples. In Figure 8.41 the boundary temperatures are
represented. The exterior boundary ends the stage with around 650℃ and in the bottom zone around 720℃.
In the interior of the parison temperature is around 1100℃.
191
1.664s 1.705s 1.722s 1.727s
1.729s 1.731s 1.733s 2.986s
Figure 8.30: Simulation temperature of the boundary for counter Simulation thickness (mm) in counter blow stage
blow stage for case 1. for case 1.
192
193
Figure 8.31: Displacement module in counter blow for the case 1 for 1.500s, 1.538s, 1.570s, 1.622s,
1.664s 1.705s 1.722s 1.727s
1.729s 1.731s 1.733s 2.986s
194
195
2.711s 2.769s 2.865s 2.939s
2.962s 2.974s 2.991s 4.777s
Figure 8.33: Boundary temperature for counter blow stage for case 2. Thickness in counter blow stage for case 2.
196
197
Figure 8.34: Displacement module in counter blow for the case 2 for 2.044s, 2.161s, 2.376s, 2.571s,
2.711s 2.769s 2.865s 2.939s
2.962s 2.974s 2.991s 4.777s
Figure 8.35: Displacement vectors with the baffle detail in counter blow for the case 2 for 2.939s, 2.962s, 2.974s
and 2.991s.
198
199
2.036s 2.081s 2.116s 2.124s
2.130s 2.134s 2.139s 3.627s
Figure 8.37: Boundary temperature for counter blow stage for case 3. Thickness (mm) in counter blow stage for case 3.
200
201
4.237s 4.253s 4.268s 4.276s
4.298s 4.303s 4.580s 7.499s
202
203
2.847s 2.879s 2.905s 2.918s
2.925s 2.931s 2.939s 4.823s
204
8.1.5 Reheat
A reheat stage follows, which is essential for obtaining the right temperature distribution to initiate the next
stage. Excessive or insufficient heating may cause too thin or too thick walls at the run down stage that
follows. The results from the stage can be seen in Figure 8.42.
Case 4 Case 5
Figure 8.42: Temperature distribution in reheat for the case 1 in 3.958s, for the case 2 in 5.622s, for the case 3 in
4.705s, for the case 4 in 10.227s and for the case 5 in 6.117s.
205
8.1.6 Invert
Before inverting the parison the blank mould is opened and reheating is allowed. While the parison is moved
to the blow mould reheating of the parison is allowed, and heat is transferred from the hotter zones (inside)
to the colder ones (outside). Stretching and cooling of the parison can be helped by the use of overhead
cooling over the blow mould. The results from this stage can be seen in Figure 8.43. It is clear that the
temperature at the exterior boundary of the parison increases due to the temperature equalization. Heat from
the inside part travels to the outside to equalize the parison temperature. In Figure 8.44 the boundary
temperatures at the end of the invert stage are presented.
206
Case 4 Case 5
Figure 8.43: Temperature distribution in invert for the case 1 in 4.514s , for the case 2 in 6.666s, for the case 3 in
5.490s, for the case 4 in 11.590s and for the case 5 in 6.823s.
207
Case 1 Case 2
Case 3 Case 4
208
Case 5
Figure 8.44: Boundary temperature for invert stage for case 1, case 2, case 3, case 4 and case 5.
8.1.7 Rundown
For case 1 the shape and temperature contour of the glass are plotted in Figure 8.45. The glass falls due to
gravity until it reaches the bottom of the blow mould cooling until the end of the stage. In Figure 8.46 the
boundary temperature can be seen, where is clear that the neckring continues to lose heat. The exterior
boundary of the parison is around 920℃ while the interior 1090℃. In addition, the bottom zone has a
higher temperature due to the fact that it has a greater mass of glass.
For case 2 the shape and temperature contours of the glass are plotted in Figure 8.47. Again, the glass falls
until the bottom of parison completely accommodates the bottom of the blow mould. The exterior boundary
is around 900℃ while the interior is around 970℃ and in the bottom zone of the parison lower temperatures
are visible, as shown in Figure 8.48. Additionally, in Figure 8.49 it can be seen the displacement field.
While the glass does not hit the bottom of the blow mould, the highest displacement is located at the bottom
zone of the parison. This effect is due to the fact that parison has more mass at the bottom zone. When the
glass parison reaches the blow moulds then the higher displacements are located at the middle wall of the
parison.
Very similar results are presented for case 3, case 4 and case 5. The respective shape and temperature
contours of the glass are plotted in Figure 8.50, Figure 8.52 and Figure 8.54. The boundary temperatures
are presented in Figure 8.51, Figure 8.53 and Figure 8.55.
209
Figure 8.45: Temperature distribution in rundown for the case 1 in 4.514s, 5.201s, 5.833s.
Figure 8.46: Boundary temperature in rundown for case 1. Thickness (mm) in rundown stage for case 1.
210
Figure 8.47: Temperature distribution in rundown for the case 2 in 6.666s, 7.222s and 9.111s.
211
Figure 8.49: Displacement module in rundown for the case 2 in 6.666s, 7.222s and 9.111s.
212
213
Figure 8.52: Temperature distribution in rundown for the case 4 in 11.590s, 12.898s, 14.545s.
214
215
8.1.8 Final Blow

Vacuum is applied to the blow mould cavity to form the finished bottle, and the final blow air is also used
in helping to form the shape of the bottle. Circulating air through the internal cooling blow head arm, when
the blow head is on top of the mould, causes the internal temperature of the bottle to be reduced. Vacuum
and final blow time should be the maximum to form the bottle before being removed from the blow mould.
For case 1 the shape and temperature contours of the glass are plotted in Figure 8.56. As it can be seen the
air pressure forces the glass to contact the mould. Hotter and thinner zones have higher velocities moving
to the boundaries, therefore, the “correct” temperature distribution is required for an even thickness
distribution to be obtained. When the final shape of the glass container is completely formed the glass cools
against the mould.
For the case 2, case 3, case 4, case 5 the temperatures contour of the glass are plotted in Figure 8.57, Figure
8.59, Figure 8.60, Figure 8.61. Despite the fact that the shape of the containers are different the behaviour
with the final blow is quite similar.
The displacement field for the case 2 is shown in Figure 8.58. At the beginning the higher displacements
are located in the zone that has higher temperature, forcing the glass to hit the blow mould wall. Then the
glass wall gets progressively into contact with the mould, ending with the contact at the container shoulder.
216
Figure 8.56: Temperature distribution in final blow for the case 1 in 5.555s, 5.556s, 5.558s,
5.561s 5.570s 7.222s
217
9.139s 9.188s 12.333s
218
Figure 8.58: Displacement module in final blow for the case 2 in 9.111s, 9.125s, 9133s,
9.139s 9.188s 12.333s
219
7.853s 7.857s 10.196s
220
14.569s 14.576s 22.272s
221
Figure 8.61: Temperature distribution in rundown for the case 5 in 9.882s, 9.884s, 9.892s,
9.902s 9.916s 13.717s
222
8.1.9 Mould Open

Mould open is the last stage the process, in which basically, the blank opens allowing the glass cooling to
the ambient air, however, the simulation at this stage is considered thermomechanical to be possible some
defects. The results are presented in Figure 8.62. It can be noted that the exterior temperature are around
600℃ as it can be seen in Figure 8.63. It can be seen that in the container there are zones where temperatures
are lower. This effect is caused by the thickness difference, thin zones cool faster and thick zones cool
slower. This is clearly seen in the bottom on the container which is always thicker than the wall.
Case 4 Case 5
Figure 8.62: Temperature distribution in mould open for the case 1 in 7.292s, for the case 2 in 12.555s, for the
case 3 in 10.294s, for the case 4 in 22.500s and for the case 5 in 14.282s.
223
Case 1 Case 2
Case 3 Case 4
224
Case 5
Figure 8.63: Final boundary temperature for case 1, case 2, case 3, case 4 and case 5.
8.1.10 Bottle Thickness

Next, the thickness distribution for the processes of blow/blow examples are presented.
 Case 1
In Figure 8.64 the thickness distribution for the case 1 is shown. The wall of the container has some
variance, showing a decrease of the thickness in the shoulder. Also, at the end of the wall the glass container
has a slender profile due to the fact that is the zone where the final blow pressure deform the glass firstly.
Additionally, the bottom zone of the container has a higher thickness.
225
Figure 8.64: Bottle thickness (mm) distribution in case 1.
 Case 2
Figure 8.65 shows the thickness distribution for the case 2. The neck wall has a decreasing thickness, ending
at the container shoulder. Then the thickness distribution slowly gets an increasing thickness until reaching
the bottom zone of the wall, where it is thinner. The bottom part of the container has a higher thickness,
which is decreasing to the centre of the container.
226
 Case 3
Figure 8.66 shows the thickness distribution for the case 3. As in the case 2 the container has a decreasing
thickness profile in the neck wall until it reaches the container shoulder. Then, the bottle wall has an
increasing profile thickness until the bottom part. Then, as in the previous results, the glass wall has a
smaller thickness, finishing at the bottom zone with a larger thickness.
227
Figure 8.66 Bottle thickness (mm) distribution in case 3.
 Case 4
Figure 8.67 presents the case 4 final shape, as well as, the experimental glass bottle. As it can be seen the
profile obtained with the simulation is quite similar to the experimental. The bottom zone has a large
thickness while in the shoulder a thinner profile is observed. The neck wall has an almost constant thickness.
In Figure 8.68 the simulation thickness profile is presented. Considering the comparison to the experimental
bottle, it can be seen that the results are similar.
Figure 8.67: Final bottle shape for the case 4. Experimental bottle thickness for the case 4.
228
 Case 5
Figure 8.69 shows the results for the case 5. This container presents a very similar behaviour comparing
with the previous cases, thinner zones located at the shoulder and at the bottom of wall and a thicker zone
located at the bottom zone.
229
8.2 Press/Blow Process

In the press/blow process a glass gob is dropped down into a mould. Once the gob is inside the blank mould,
the baffle closes and the plunger moves up. When the plunger is at its highest position, the ring closes itself
around the plunger, so that the mould plunger tool is closed below. Finally, when the plunger is lowered,
the ring is decoupled from the blank mould and the glass preform is carried by means of a robotic arm to
the blow mould. In the blow stage the parison is usually first left to sag due to gravity for a short period.
Then pressurized air blows inside to force the glass in a mould shape. It is important to know the right shape
of the preform beforehand for an appropriate distribution of glass over the mould wall.
During the parison forming a large amount of heat is removed from the glass and the temperature
distribution in the glass changes. Due to the temperature dependence of the viscosity the glass flow depends
on the initial temperature of the gob and the subsequent steps of heat transfer in the glass, blank mould and
plunger. Any disturbance in heat transfer during parison forming can result in an uneven glass distribution,
circumferentially as well as axially.
The input parameters for simulations consist of the parison geometries, i.e. the mould and the plunger
descriptions, the initial position of the plunger, the initial position of the computational (glass) domain, and
a number of parameters characterizing the properties of glass as were used previously.
8.2.1 Gob Loading

In the gob loading stage, glass is dropped directly against the plunger and starts to cool. This stage is similar
to the blow/blow process.
For case 1 the shape and temperature contours of the glass are plotted in Figure 8.70. After falling from the
feeder the gob contacts the plunger and only a little accommodation in the gap between the plunger and the
neckring. In this stage, the gob must be under baffle so that the baffle doesn’t hit the gob while is closing.
In Figure 8.71 the boundary temperatures are represented.
For case 2, case 3, case 4, case 5, case 6, case 7 and case 8 the shape and temperature contour of the glass
are plotted in Figure 8.72, Figure 8.74, Figure 8.76, Figure 8.80, Figure 8.82, Figure 8.84 and Figure 8.86,
in which a similar behaviour for all cases is depicted.
Also, Figure 8.73, Figure 8.75, Figure 8.77, Figure 8.81, Figure 8.83, Figure 8.85 and Figure 8.87 present
the boundary temperatures for the cases 2 to 8. A very similar pattern of the temperature at the boundary is
evident. Except for the case 6, since then the there is no contact with the mould, the gob only contact the
plunger.
The displacement contours for the case 4 are presented in Figure 8.78 as well the displacement vectors in
Figure 8.80. It quite evident the gravity force the glass to flow inside the neckring.
230
Figure 8.70: Temperature distribution in gob loading for the case 1 in 0.175s, 0.199s, 0.252s and 0.466s.
231
232
233
234
Figure 8.78: Displacement module in gob loading for the case 4 in 0.2513s, 0.2667s, 0.3213s
0.4797s 0.5349s 0.5555s
Figure 8.79: Displacement vector in gob loading for the case 4 in 0.2513s, 0.2667s, 0.3213s
0.4797s 0.5349s 0.5555s
235
236
Figure 8.82: Temperature distribution in gob loading for the case 6 in 0.076s, 0.128s, 0.303s and 0.460s.
Figure 8.83 Boundary temperature for gob loading stage for case 6.
237
238
239
8.2.2 Baffle On
The baffle is positioned on top of the blank mould allowing the plunger to move towards the baffle with a
prescribed force. While the tool is being positioned in the top of the blank mould the glass continues to
cool. In Figure 8.88 all the press/blow cases in baffle on stage are presented.
240
Case 7 Case 8
Figure 8.88: Temperature distribution in baffle on for the case 1 in 0.800s, for the case 2 in 0.859s, for the case 3 in 1.147s, for
the case 4 in 0.889s, for the case 5 in 0.972s, for the case 6 in 0.801s, for the case 7 in 0.957s and for the case 8 in 1.126s.
8.2.3 Plunger Up
In this section the results of plunger up stages are plotted to observe how they behave. In the beginning of
this process the gob is already above the plunger and ready for the plunger movement. In practice, the
plunger is pushed by a piston. This means that the flow velocity is the result of an external force applied to
the plunger. The plunger moves upwards, forcing the glass to fill in the space between the mould and the
plunger. At a certain point the glass hits the bottom of the mould (the baffle part) and gets closer to the
neckring part. At this point the pressure in glass increases. In the last instants the glass is obliged to move
and fill the neckring. Here, a bad misalignment may create finish defects.
For case 1 the shape and temperature contours of the glass are plotted in Figure 8.89. While the plunger
moves up the nodes that are in contact with the plunger are allowed to slip on the mould with a friction
coefficient, but in the other hand the nodes that are in contact with the rest of the moulds are maintained
fixed. This assumption is justified by practical observations. When the glass hits the baffle zone the glass
starts to spread and fills all the remaining space. Then glass flows down from the top of the parison to
completely fill the neckring area. Next, heat transfer from the moulds to the glass takes place.
In Figure 8.90 the boundary temperatures are represented. The inner and outer boundary have similar
temperatures as in this stage the two tools “remove” more “evenly” the heat from the glass from the two
sides of the parison. The results also show that the neckring has a higher temperature since it was the last
zone be formed. Additionally, the thickness distribution at the end of the plunger up stage is presented.
For cases 2 to 8 the shape and temperature contours of the glass are plotted in Figure 8.91, Figure 8.93,
Figure 8.95, Figure 8.99, Figure 8.101, Figure 8.103 and Figure 8.105 A very similar behaviour for all cases
was observed. The boundary temperatures from the remaining cases are presented in Figure 8.92, Figure
8.94, Figure 8.96, Figure 8.100, Figure 8.102, Figure 8.104 and Figure 8.106.
The displacement contours during the plunger up are shown in Figure 8.97. It can be seen that at the
beginning of the stage the maximum displacement is near the plunger and when the neckring is completely
241
filled the displacement is almost null. Also in Figure 8.98 the displacement vector is presented when the
baffle and the neckring is being filled with the glass.
242
Figure 8.89: Temperature distribution in plunger up for the case 1 in 0.8000s, 0.8002s, 0.8003s
0.8005s 0.8011s 0.8018s
0.8035s 0.8050s 2.0000s
Figure 8.90: Boundary temperature in plunger up for case 1. Thickness (mm) in plunger up stage for case 1.
243
244
0.8594s 0.8599s 0.8606s
0.8616s 0.8630s 1.9171s
Figure 8.92 Boundary temperature for plunger up stage for case 2. Thickness (mm) in plunger up stage for case 2.
245
246
1.1470s 1.1472s 1.1475s
1.1482s 1.1500s 2.0533s
247
248
0.8896s 0.8907s 0.8915s
0.8925s 0.8931s 1.8750s
Figure 8.96 Boundary temperature for gob loading stage for case 4. Thickness (mm) in plunger up stage for case 4.
249
250
Figure 8.97: Displacement module in plunger up for the case 4 in 0.8888s, 0.8890s, 0.8891s
0.8896s 0.8907s 0.8915s
0.8925s 0.8931s 1.8750s
251
Figure 8.98: Displacement vectors in plunger up for the case 4 in 0.8888s, 0.8890s, 0.8891s
0.8896s 0.8907s 0.8915s
0.8925s 0.8931s 1.8750s
252
253
0.9733s 0.9737s 0.9744s
0.9759s 0.9777s 2.2222s
254
255
Figure 8.101: Temperature distribution in plunger up for the case 6 in 0.8013s, 0.8014s
0.8015s 0.8020s
0.8030s 0.8050s
0.8067s 1.9967s
256
Figure 8.102: Boundary temperature in plunger up for case 6 and the thickness (mm) in plunger up stage for case 6.
257
258
0.9587s 0.9597s 0.9623s
0.9651s 0.9666s 2.0512s
259
260
1.1298s 1.1332s 1.1351s
1.1370s 1.1400s 3.098s
261
8.2.4 Reheat
After the plunger down stage the baffle is removed and the blank mould opens to allow the parison to
reheat, i.e. heat is transferred from hotter zones to the cooler zones. This effect will eliminate skin conditions
before the invert stage. The reheat stage is equivalent to the one in the blow/blow process. The results from
the reheat stage can be seen in Figure 8.107. The lowest temperature is verified at the wall parison. This
temperature effect is due to the heat transfer from the glass to moulds during the gob loading and the baffle
on.
262
Case 7 Case 8
Figure 8.107: Temperature distribution in reheat for the case 1 in 2.733s, for the case 2 in 2.837s, for the case 3 in 3.160s, for the case
4 in 2.777s, for the case 5 in 2.847s, for the case 6 in 3.178s, for the case 7 in 2.871s and for the case 8 in 3.943s.
8.2.5 Invert
During the time to change tools, i.e. while blank mould and blow mould are swapped, the parison is always
reheating. When invert stage starts, speed must be regulated to the size and temperature of the parison, as
centrifugal forces will oscillate the parison, pushing the glass mass to the bottom of the parison and affecting
the mass distribution. This is the period of time between the end of the parison transfer and the start of the
final blow. During this time the parison continues to reheat and gravity stretches the parison while holding
it in the blow mould. The inertia forces involved in this stage was not considered.
Time is also an important factor influencing the invert stage, as an excessive time can create a sag in the
parison and this effect must be counterbalanced. Also, overhead cooling can happen causing a deficient
rundown and even preventing the parison to fall. A short time, in other hand, may cause the opposite effect,
an excessive stretching in the necking area that may produce a very thin wall glass during the rundown.
Thus, the time of the invert stage should be in according the desired final shape.
The results from the invert stage are presented in Figure 8.108. It is clearly noted that the bottom zone of
the parison has more mass, therefore, a higher temperature than the parison wall and neckring. Additionally,
the boundary temperature are shown in Figure 8.109. The coolest zones are located at the neckring while
the hottest zones are at the bottom zone of the parison.
263
264
Case 7 Case 8
Figure 8.108: Temperature distribution in invert for the case 1 in 3.760s, for the case 2 in 3.910s, for the case 3 in 3.733s, for the
case 4 in 4.194s, for the case 5 in 3.819s and for the case 6 in 3.561s, for the case 7 in 3.487s and for the case 8 in 4.788s.
Case 1 Case 2
265
Case 3 Case 4
Case 5 Case 6
266
Case 7 Case 8
Figure 8.109: Boundary temperature for invert stage for case 1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8.
8.2.6 Rundown
At the rundown stage glass stretches due to gravity until it reaches the shape to start the final blow. Figure
8.110 shows the result for the case 1. The glass parison starts to fall due to gravity until it reaches the blow
mould. The boundary temperatures as well the thickness are presented in Figure 8.112. The results show
that the parison wall has an increase of temperature, while the bottom zone of the parison has a large drop
on the temperature due to the contact to the blow mould.
Figure 8.112, Figure 8.114, Figure 8.116, Figure 8.119, Figure 8.121, Figure 8.123 and Figure 8.125 present
the results for the cases 2 to 8. A quite similar behaviour is depicted for all those cases. The boundary
temperatures and the parison thickness is presented in Figure 8.113, Figure 8.115, Figure 8.117, Figure
8.120, Figure 8.122, Figure 8.124 and Figure 8.126 The results also show a very similar behaviour.
Additionally, the displacement contours for the case 4 can be seen in Figure 8.118. In the beginning the
higher displacements are located at the bottom zone of the parison. When the parison hits the blow moulds
the higher displacements are located in the middle of the parison.
267
Figure 8.111 Boundary temperature for run down stage for case 1 Thickness (mm) in run down stage for case 1.
268
269
270
271
Figure 8.118: Displacement module in rundown for the case 4 in 4.194s, 4.974s and 5.972s.
272
Figure 8.120: Boundary temperature in rundown for case 5. Thickness (mm) in run down stage for case 5.
273
Figure 8.121: Temperature distribution in rundown for the case 6 in 3.561s, 4.298s
4.898s 5.584s
274
Figure 8.122: Boundary temperature in rundown for case 6 and the thickness (mm) in run down stage for case 6.
275
276
277
8.2.7 Final Blow

Insufflated pressurize air forces the parison to finalize the container shape. Additionally, vacuum is also
applied from the blow mould, not only to give stability when the parison walls is forced to stick to the blow
mould, but also to reduce the internal pressure. After the vacuum and the final blow pressure the final
product should have no geometric imperfections.
The temperature distribution is presented in Figure 8.127 for case 1. The air pressure forces the glass to
touch the blow mould. When all the glass is in contact with the mould heat transfer between the glass and
moulds occurs along all its length.
In Figure 8.128, Figure 8.129, Figure 8.130, Figure 8.131, Figure 8.132, Figure 8.133, Figure 8.134 and
Figure 8.135 present the temperature distribution for the cases 2 to 8. Note that the neckring and the
shoulder are cooler than the other zones, as expected.
Additionally, Figure 8.131 shows the displacement contours during the final blow stage. The results show
that the highest displacement is located at the end of the parison wall and then “travels” to the container
shoulder. After all the glass boundary is in contact with the mould no further displacement is observed.
278
Figure 8.127: Temperature distribution in final blow for the case 1 in 5.800s, 5.803s, 5.804s
5.805s 5.833s 7.266s
279
4.911s 4.912s 7.621s
280
5.470s 5.479s 7.467s
281
5.419s 5.420s 7.375s
282
Figure 8.131: Displacement module in final blow for the case 4 in 5.972s, 5.418s, 5.419s
5.419s 5.420s 7.375s
283
5.706s 5.909s 7.361s
284
285
Figure 8.133: Temperature distribution in final blow for the case 6 in 5.584s 5.583s
5.584s 5.585s
5.586s 7.174s
286
5070s 5.080s 7.179s
287
6.862s 6.871s 9.859s
8.2.8 Mould Open

Mould open is the last stage of the press/blow process. The results can be found in Figure 8.136 for all
cases. The boundary temperature at the end of the process is presented in Figure 8.137. The results show
that the side that was in contact with the mould show a range of temperatures between 600℃ and 700℃
while in the inner side of the container temperatures are around 700℃. In addition, the inner part of the
bottom zone has the highest temperatures due to the fact that largest glass mass is located in this area.
288
289
Case 7 Case 8
Figure 8.136: Temperature distribution in mould open for the case 1 in 7.399s, for the case 2 in 8.205s, for the case 3 in 7.600s, for the case
4 in 7.750s, for the case 5 in 7.847s, for the case 6 in for the case 7 in 7.653s and for the case 8 in 10.516s.
290
Case 1 Case 2
Case 3 Case 4
291
Case 5 Case 6
Case 7 Case 8
Figure 8.137: Final boundary temperature for case 1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8.
292
8.2.9 Bottle Thickness

Next, the thickness distribution for the processes of press/blow examples are presented.
 Case 1
Figure 8.138 presents the thickness distribution for the case 1. The thinner zones are located at the end of
the neckring and under the container wall.
 Case 2
Figure 8.139 presents the thickness distribution for the case 2. The profile shows a thickness around 2.5
mm.
293
 Case 3
Figure 8.140 presents the thickness distribution for the case 3. The profile presents thicker zones at the end
of the neckring and also at the wall. The experimental cut of the container is presented in Figure 8.140. The
real container has an almost constant thickness along the wall. The simulation result present 1.5 mm of
thickness near the shoulder while in the experimental data it was near 2.5 mm. Above the shoulder the
simulation result is about 2mm. Finally, at the bottom zone the thickness obtained in the simulation is
around 4.75 mm. The distribution is quite similar with the experimental results.
294
Figure 8.140: Bottle thickness (mm) distribution in case 3. Real bottle thickness distribution in case 3.
 Case 4
Figure 8.141 presents the thickness distribution for the case 4. The container wall present a constant
thickness around 1.5 mm while in the bottom is observed a thickness of 6 mm. The experimental thickness
distribution is presented in Figure 8.142.
295
2,43 2,30
2,00 1,99
1,90 1,51
2,28 2,02
1,86 1,69
1,89 1,75
1,92 1,85
2,03 2,0
2,35 2,17
2,50 2,32
2,88 2,34
2,86 2,26
2,61 2,09
3,33 2,35
4,38 5,04 3,40
Figure 8.142: Real thickness (mm) distribution in case 4.
 Case 5
Figure 8.143 presents the thickness distribution for the case 5. Comparing the simulation with the real case
a different thickness distributed was obtained, requiring a better parameter calibration.
Figure 8.143: Thickness (mm) distribution in case 5. Real thickness (mm) distribution in case 5.
296
 Case 6
The Figure 8.139 presents the thickness distribution fort the case 6. The profile presents a thickness around
2.0 mm on the walls and 6 mm at the bottom zone.
Figure 8.144: Thickness (mm) distribution in case 6.
 Case 7
The Figure 8.139 presents the thickness distribution for the case 7. The profile presents at the walls a
thickness around 2.0 mm while in the bottom zone the thickness is around 6 mm. The experimental results
present for the neck an average thickness of 2.14cm, in the sidewall an average of 2.32cm, in the heel a
average of 1.99cm and in the bottom an average of 4.63. It can conclude that the simulation results are very
close to the experimental result.
297
Figure 8.145: Thickness (mm) distribution in case 7. Experimental thickness distribution in case 7.
 Case 8
The Figure 8.139 presents the thickness distribution for the case 8. Again, the profile presents at the walls
a thickness around 2.0 mm while in the bottom zone the thickness is around 6 mm. The thickness
distribution present some fluctuation in the side wall. Thus, it is difficult to compare to the experimental
data but in general the bottle shape appears to be quite similar.
298
Figure 8.146: Thickness (mm) distribution in case 8. Experimental thickness (mm) distribution in case 8.
8.3 Evaluation of some numerical aspects

In this section some additional information is provided on some numerical aspects which illustrate some
problems that need to be addressed in the near future in order to improve the performance of the software
developed.
8.3.1 Volume Evolution

Since incompressibility is assumed, the glass volume should be conserved. Nevertheless, because
incompressibility is imposed via a penalty method and many mesh refinements and redefinitions take part
some violations are expected.
Figure 8.147 presents the volume evolution for all de cases of blow/blow process. The vertical line refer to
the end of the most important stages such as gob loading, settle blow, counter flow and final blow
respectively. The results show that in gob loading and settle blow the volume slightly decreases and is
accentuated in the counter blow and final blow.
299
Case 1
Case 2
Case 3
300
Case 4
Case 5
Figure 8.147: Volume evolution for blow/blow process for the case 1, case 2, case 3, case 4 and case 5.
Figure 8.148 presents the volume evolution for all the cases of press and blow process. The volume
evolution starts to decrease almost uniformly during the plunger upstage having a huge drop at the final
blow. However, the total volume decrease seams not be significant enough to compromise the final results.
This type of evaluation can be an important tool to calibrate the penalty terms and improve the reliability
of the mesh refinement tools.
Case 1
Case 2
301
Case 3
Case 4
Case 5
Case 6
302
Case 7
Case 8
Figure 8.148: Volume evolution for press/blow process for the case 1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8.
8.3.2 Evolution of the Number of Nodes

The evolution of the number of nodes, therefore, the number of degrees of freedom, was evaluated in each
simulation, as this is an important indicator how it may affect the simulation time required.
Figure 8.149 presents the evolution of the number of nodes for all de cases of blow/blow process. The
vertical line means the end stages of gob loading, settle blow, counter blow and final blow respectively.
The results show that in gob loading and settle blow almost a constant number of nodes is kept. As the
simulation proceeds to create the parison and the final container the number of nodes increases with
intermediate remeshing procedures, sometimes with some significant peaks corresponding to the remeshing
in some complex shape boundary zones.
303
Case 1 Case 2
Case 3 Case 4
Case 5
Figure 8.149: Number of nodes for the case 1, case 2, case 3 case 4 and case 5.
The Figure 8.150 presents the evolution the number of nodes for all de cases of press/blow process. The
number of nodes seams to maintain almost constant until the final blow stage. In the final blow stage an
increase of number of nodes is noted corresponding to some complex shapes during the final forming of
the container.
This type of evaluation can be an important to trigger the need to use better remeshing tools and solvers for
the nonlinear system of equations involved.
304
Case 1 Case 2
Case 3 Case 4
Case 5 Case 6
Case 7 Case 8
Figure 8.150: Number of nodes for the case 1, case 2, case 3 case 4, case 5, case 6, case 7 and case 8.
305
8.3.3 Plunger Displacement

The plunger displacement versus time can be an important data to validate the plunger up stage. Therefore,
the plunger displacement during the simulation was registered. Figure 8.151 shows the results for the six
cases. It can be noted that while the glass doesn’t hit the baffle the displacement has an almost linear
behaviour. Then, when the glass moves to the neckring the plunger displacement rate is decreasing until
zero when the neckring is complexly filled.
These type of curves can be used to calibrate the friction between the glass and the plunger to reproduce
experimental data that may be obtained at the industrial site and later be used in the simulation. It is
noteworthy that the existing software’s used on glass forming simulation do not include the possibility of
applying a force to a plunger, as it was developed in this work,, i.e., they rely only on an imposed
displacement condition, which is neither real nor acceptable from the industry point of view.
Case 1 Case 2
Case 3 Case 4
306
Case 5 Case 6
Case 7 Case 8
Figure 8.151: Plunger displacement for case 1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8 for a slip penalty value of 1.0e04.
8.3.4 Time Evolution

The increment of time used throughout the simulation is highly dependent on the contact procedure.
The Figure 8.152 presents the time increment and the total time per step increment for all de cases of
blow/blow process. In the beginning of the blow/blow process, i.e. gob loading, a lot of contact involving
slipping conditions is very much present, which greatly limits the size of the time increment. Then, when
settle blow is applied higher time increments are possible as sliding is no longer present. In the corkage
reheat stage time increments are be bigger, as the thermal problem is dominant. At the counter blow stage,
despite the fact that low contact situations are present, small time increments are needed due to quick mesh
distortions resulting from the applied pressure. In the reheat, invert and run down stages, a higher time
increment is used. Finally, the final flow stage require very small time increments due to the heavy
deformation that takes place. On the right side of the Figure, the time evolution shows clearly the stages
that require more increments. An almost constant time evolution is seen corresponding to gob loading,
settle blow, counter blow and final blow stages. The other stages are performed faster as higher time
increments may be used.
307
Case 1
Case 2
Case 3
308
Case 4
Case 5
Figure 8.152: Time increment for blow/blow process for case 1, Time evolution for blow/blow process for case 1, case 2, case 3,
case 2, case 3, case 4 and case 5. case 4 and case 5.
Figure 8.153 presents a similar information for all de cases of press/blow process. The results are consistent
in all the cases. In the beginning gob loading stage uses small time increments that are increasing until the
baffle on stage. Then in the plunger up stage the time increments are very small. This effect is due to the
slipping contact condition in this stage. In the following stages time increments are higher until the final
blow. With the contact to the blow mould time increments decrease again.
This type of evaluation of the time increment size used in each stage can be an important indicator in future
developments to optimise the contact algorithms, therefore, diminish the computing time.
Case 1
309
Case 2
Case 3
Case 4
Case 5
310
Case 6
Case 7
Case 8
Figure 8.153: Time increment for blow/blow process for case Time evolution for blow/blow process for case 1, case 2, case
1, case 2, case 3, case 4, case 5, case 6, case 7 and case 8. 3, case 4, case 5, case 6, case 7 and case 8.
Together with the previous improvement proposals and with parameter calibration, which is now possible
at industrial ambient, the developed software is a strong basis product in the direction of obtaining a
practical decision support system in the design, production and quality control of glass containers.
311
Chapter 9 Conclusions and Future Work
CHAPTER 9
9 Conclusions and Future Work
A development of numerical software for forming of glass containers has been performed. Several different
glass containers have been simulated for the two typical blow/blow and press/blow processes.
The main goal was to perform automatically all the process stages without user handling. The software is
able to efficiently choose the timestep, boundary conditions, loads, remeshing parameters and to transfer
all information to a new mesh, as well as all to perform the simulation successfully. The software is capable
to compute and visualise the container shape, the thickness of the final product as well as the temperatures
and displacement that the glass and mould undergo during the process. Contact algorithms accounting for
sticking and slipping conditions were implemented.
A mesh updating and a remeshing scheme for linear and quadratic triangles and quadrilaterals finite
elements was included.
Various types of equation solvers were implemented, such as the MKL Library and a Lapack solver. For
this type of problem the skyline solver presented more efficiency. In addition, the software was developed
having in mind computation resources optimization. The examples presented ran with less that 20Mb of
RAM memory.
Quadratic elements were in some cases unsatisfactory to model temperature in transient analysis, giving
some non reliable physical results. Solution relied on finer meshes with consequent increase in
computational time. As glass viscosity is dramatically dependent on temperature the element choice was
an important decision. Linear quadrilateral elements was chosen as they depicted the most balanced
performance, taking into account thermal response, incompressibility control and contact handling.
The software development requirement had various initial milestones to be reached, that were attained
successfully such as:
i. Different element types were implemented, the software is able to run with linear and quadratic
triangles and quadrilateral elements.
ii. Thermal and mechanical contact were implement with all the inherent features.
iii. Remeshing techniques were included for all element types.
iv. Automatic mesh distortion detection and remeshing triggering.
312
v. Automatic boundary conditions definition after remeshing, with no user intervention required
vi. Automatic data transfer from older to new meshes after the remeshing stage.
vii. Implementation of triangle and quadrilateral elements meshing tools
viii. Implementation of renumbering techniques
ix. Implementation of plunger movement driven by force or velocity
x. Integration of a complete friction formulation.
xi. Several equation solver implemented.
xii. Optimization of the computer resources, such as the RAM usage.
xiii. Automated gob creation.
xiv. Implementation thermal and mechanical processes in the several stages and its automatic
execution.
xv. Easy process stages configuration (active tools, gravity direction, etc.)
xvi. Material properties code integration
xvii. Integration of a preprocessor with the solver
xviii. Inclusion of an intelligent file management.
xix. Definition of strategy for output files and postprocessing
xx. Error detection checks
xxi. Restart option (continue analysis) after each calculated stage
xxii. Implementation of the possible of different system units (Imperial, SI, cgs, etc)
xxiii. Intermediate validation of numerical results performed.
Several cases tests were performed. The results showed the reliability and robustness of the developed
computational tool.
When starts a Ph.D. thesis and even more when the work is related to software development, improvement
and researching, is a continuous task that will never end. It is expectable that new challenges and
developments in the glass container simulations domain appears. Considering some perspectives
concerning the continuity of the present work and further developments for future research some proposals
may be outlined.
i. Further improvement of the computational efficiency, robustness, rigor and reliability.

ii. Testing of different industrial cases and for different materials.
iii. Enhancing the developed models, aiming at the description of more features, such as taking into
account more complex models for heat radiation or the inclusion of inertia effects.
iv. Introduction of a selfcontact algorithm.
v. Investigation on the influence of the moulds in the heat transfer problem
vi. Generalizing the software for full 3D forming processes.
The actual state of the software has a good perspective to simulate accurately the glass container forming
process from the initial stage to final shape container. It would be decisive to test some more real cases to
enhance its performance. Since glass forming processes have stages that are clearly separated, experimental
data from the individual stages can be very useful for a multistage validation of the software.
As the mechanical behaviour of glass depends highly on viscosity, validation of heat transfer coefficients,
lubrication and cooling moulds parameters are critical. Also, friction between glass and moulds is very
important to understand the nature of an appropriated friction law. A parallel experimental work to provide
an accurate indication of the physical parameters involved is needed.
Although some work has done in the mesh size effect, a mesh sensitivity study in order to obtain converged
results is still necessary. Presently, a new full mesh is always created when a critical distortion value is
exceeded. This distortion can be in a certain zone or just in one element. The mesh creation could be
improved in a way that the remeshing would be only on the distorted zone in order to reduce the computing
time.
Incremental time is very much influenced by contact. Thus, improving the contact algorithm is mandatory.
313
The evolution of the design can lead to develop non axisymmetric glass containers opening the software to
others potentialities such as the extension from axisymmetric to three dimensional. Thus, the possibility the
software to perform 3D simulation are quite interesting. This can involve a huge expansion in the number
of variables, increasing the computer time as well as the storage space. Remeshing and visualization of the
results would also be challenging tasks.
Additionally, other modules could be added to the software of the glass container process related to the
mould design, melting optimization, quality control, etc. Mould design module could include the cooling
pipes design, material, shape, etc. In relation to quality control a structural module to inspect admissible
pressures and loads could be included, i.e., to possibility to evaluate the integrity of the final containers
subjected to vertical, lateral, combined loads and internal pressure.
Actually, the software is already almost capable to perform a full simulation with the discretised moulds.
However, the results were not included here as is still an ongoing development. This will be a decisive tool
to test and optimize different mould cooling systems such as the axial or radial cooling or the use of different
materials in some zones in the mould.
314
Chapter 10 References
10 References
1. Bourhis, E.L., Glass: Mechanics and Technology. 2014: Wiley.
2. Rao, K.J., Structural Chemistry of Glasses. 2002: Elsevier.
3. Shelby, J.E., Introduction to Glass Science and Technology. 2005: Royal Society of Chemistry.
4. Ropp, R.C., Handbook of Glass Fractography. 2008: AuthorHouse.
5. Groot, J.A.W.M., C.G. Giannopapa, and R.M.M. Mattheij, Numerical Shape Optimization in
Industrial Glass Blowing. Journal of Pressure Vessel Technology, 2014. 136(6): p. 061301-
061301.
6. Warude, A., Analysis of glass mold to enhance rate of heat transfer. 2004.
7. Groot, J., C. Giannopapa, and R. Mattheij, Development of a numerical optimization method for
blowing glass parison shapes. Journal of Manufacturing Science and Engineering, 2011. 133(1):
p. 011010.
8. Kaipio, J. and E. Somersalo, Statistical and Computational Inverse Problems. 2006: Springer New
York.
9. Siedow, N., Radiative Heat Transfer and Applications for Glass Production Processes II, in
Mathematical Models in the Manufacturing of Glass: C.I.M.E. Summer School, Montecatini
Terme, Italy 2008, A. Fasano, Editor. 2011, Springer Berlin Heidelberg: Berlin, Heidelberg. p.
135-171.
10. Louzguine-Luzgin, D.V., L.V. Louzguina-Luzgina, and H. Fecht, On limitations of the viscosity
versus temperature plot for glass-forming substances. Materials Letters, 2016. 182: p. 355-358.
11. César de Sá, J., Numerical modelling of glass forming processes. Engineering Computations,
1986. 3(4): p. 266-275.
12. Empakglass, GLASS CONTAINER TECHNOLOGY TRAINING. 2013. p. 609.
13. Green, D.J., An Introduction to the Mechanical Properties of Ceramics. 1998: Cambridge
University Press.
14. Mallick, K.K. and U.o.W.D.o. Physics, Strengthening of Container Glasses and Related
Compositions. 2013.
15. Ojovan, M.I. and W.E. Lee, Viscosity of network liquids within Doremus approach. Journal of
applied physics, 2004. 95(7): p. 3803-3810.
16. Rouxel, T., Thermodynamics of viscous flow and elasticity of glass forming liquids in the glass
transition range. The Journal of chemical physics, 2011. 135(18): p. 184501.
17. Modernisation of batch plant for efficient operations, in Glass International. 2010.
18. Institute, G.P. Glass Furnace Operations.
19. Sorg, Glass Melting Technology.
20. McMinn, J., Forehearth of the future: The forehearth plays no role in chemical conditioning or
refining of the glass nor does it contributes to the removal of bubble or seeds from the glass, in
Glass Forming.
21. BH-F, Gob Shaping and Adjustments: Operation Manual. 1995.
22. Duga, R.J., et al., Electronic glass feeder plunger operating mechanism. 1985, Google Patents.
23. Davey, R.G. and D.E. Schupbach, Molten glass gob delivery system. 1988, Google Patents.
24. Sarwar, M. and A. Armitage, Tooling requirements for glass container production for the narrow
neck press and blow process. Journal of Materials Processing Technology, 2003. 139(1): p. 160-
163.
25. Snyder, H.M., Cold-end coatings in glass container manufacture. American Ceramics Society
Bulletin, 1990. 69(11): p. 1831-1833.
26. van Wijk, E., et al., GLASS COATING: THE KEY TO GLASS PRODUCTION. Sterling
Publications Ltd, 1995., 1995: p. 125-128.
27. Lindner, G.H. and R.W. Barkalow, Coating hood with air flow guide for minimizing deposition of
coating compound on finish of containers. 1987, Google Patents.
28. Mikelic, A., et al., Mathematical Models in the Manufacturing of Glass. 2011, Springer.
29. Glass, B., General Information on the Packaging of Glass Containers for their Safe Transport and
Delivery to the Customer. British Glass Publication, 1998. TEC 11.
30. Bathe, K.J., Finite Element Procedures. 2006: Prentice Hall.
315
Chapter 10 References
31. Farina, A., et al., Mathematical Models in the Manufacturing of Glass: C.I.M.E. Summer School,
Montecatini Terme, Italy 2008. 2010: Springer Berlin Heidelberg.
32. Hutton, Fundamentals Of Finite Element Analysis. 2005: McGraw-Hill Education (India) Pvt
Limited.
33. Hibbitt, K. and Sorensen, ABAQUS: Theory Manual. 1997: Hibbitt, Karlsson & Sorensen.
34. Yastrebov, V.A., Numerical Methods in Contact Mechanics. 2013: Wiley.
35. César de Sá, J., et al., A computational model for glass container forming processes. 1999.
36. Humphreys, C.E., Mathematical modelling of glass flow during a pressing operation. 1991.
37. George, P., Automatic mesh generation and finite element computation. Handbook of numerical
analysis, 1996. 4: p. 69-190.
38. Rao, S.S., The Finite Element Method in Engineering. 2011: Elsevier Science.
39. Perić, D., M. Dutko, and D. Owen, Aspects of adaptive strategies for large deformation problems
at finite inelastic strains. Studies in Applied Mechanics, 1998. 47: p. 349-363.
40. Akin, J.E., Finite Element Analysis with Error Estimators: An Introduction to the FEM and
Adaptive Error Analysis for Engineering Students. 2005: Elsevier Science.
41. VanSant, J.H. and L.L.N. Laboratory, Conduction Heat Transfer Solutions. 1983: Lawrence
Livermore National Laboratory.
42. Khoei, A., Computational Plasticity in Powder Forming Processes. 2010: Elsevier Science.
43. Cook, R.D., Finite element modeling for stress analysis. 1995: Wiley.
316
Faculdade de Engenharia da Universidade do Porto
Rua Dr. Roberto Frias, s/n 4200-465 Porto PORTUGAL
ww.fe.up.pt
317

Universidade Porto: Decision Support System in The Design, Production and Quality Control of Glass Containers

Enviado por

Direitos autorais:

Formatos disponíveis

Universidade Porto: Decision Support System in The Design, Production and Quality Control of Glass Containers

Enviado por

Dados do documento

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Universidade Porto: Decision Support System in The Design, Production and Quality Control of Glass Containers

Enviado por

Direitos autorais:

Formatos disponíveis

UNIVERSIDADE DO PORTO

Decision Support System in the Design, Production

Programa Doutoral em Engenharia Mecânica (PRODEM) Bruno Miguel Rodrigues Martins

À minha família e amigos.

Ao INEGI - instituto de ciência e inovação em engenharia mecânica e engenharia industrial desejo

Ainda a toda a minha família e amigos.

Keywords: Glass container forming, Finite element method, Thermomechanical model

INDEX ............................................................................................................................................. VII

LIST OF FIGURES .......................................................................................................................... X

LIST OF TABLES ......................................................................................................................... XIX

1.1 HISTORIC REVIEW OF GLASS MAKING........................................................................................1

2 GLASS MATERIAL PROPERTIES ........................................................................................ 9

2.1 BRITTLENESS ..............................................................................................................................9

3 GLASS CONTAINER PRODUCTION .................................................................................. 21

3.1 BATCH ......................................................................................................................................22

4 MATHEMATICAL MODELLING OF GLASS FORMING PROCESSES ......................... 83

4.1 FINITE ELEMENT FORMULATION FOR THERMOMECHANICAL BEHAVIOUR................................83

5 SIMULATING MODEL FOR GLASS CONTAINERS ......................................................... 95

5.1 ELEMENT TECHNOLOGY ...........................................................................................................96

5.3.2 Remeshing ......................................................................................................................108

6 ELEMENTS VALIDATION AND TESTING .......................................................................115

6.1 HEAT TRANSFER VALIDATION ................................................................................................115

7 COMPUTER PROGRAM .....................................................................................................156

8 GLASS CONTAINERS SIMULATION CASES ...................................................................166

8.1 BLOW AND BLOW PROCESS ....................................................................................................173

9 CONCLUSIONS AND FUTURE WORK..............................................................................312

Symbol Description Unity [SI]

Ti Initial temperature [K]

cr heat transfer coefficient convection/radiation [Wm-2K-1]

p Plastic strain rate [s-1]

k shear viscosity [Pa.s]

e Stress rate [Pa.s-1]

eii Dilatational strain rate

eij Deviatoric strain rate

Sij Deviatoric stress tensor

 ij Strain rate tensor

1.1 Historic Review of Glass Making

1.2 History of Glass Moulding

1.3 Glass Containers Industry

1.4 Glass Container Quality

1.5 Glass Container Simulation

1.7 Thesis Synopsis

This page intentionally left blank

A description of the glass material properties is presented in this chapter.

2.3 Thermal Expansion

The deformation caused by the thermal expansion is given by:

which αij is the thermal expansion coefficient.

2.3.1 Glass Cord

2.4 Temperature dependence of viscosity

Log Viscosity (Poise) Approx. Temperature (ºC)

Gob Viscosity 3 1 000 1185

Softening Point 7.65 44 668 000 724

Conveyor Viscosity 10 10 000 000 000 632

Annealing Point 13.37 23 442 000 000 000 546

Strain Point 14.6 398 110 000 000 000 499

Table 2.1: Viscosity reference temperatures.

Figure 2.2: Viscosity evolution in working zone temperature [13].

The previous relation is known as the Williams-Landel-Ferry (WLF) law.

where α is a constant and Tk is the temperature at which Sc goes to zero.