Вы находитесь на странице: 1из 11

Proceedings of the Institution of Mechanical

Engineers, Part B: Journal of Engineering


Manufacture
http://pib.sagepub.com/

A review of research in numerical simulation for the glass-pressing process


M Brown
Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 2007 221: 1377
DOI: 10.1243/09544054JEM833

The online version of this article can be found at:


http://pib.sagepub.com/content/221/9/1377

Published by:

http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture can be found at:

Email Alerts: http://pib.sagepub.com/cgi/alerts

Subscriptions: http://pib.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://pib.sagepub.com/content/221/9/1377.refs.html

>> Version of Record - Sep 1, 2007

What is This?

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


REVIEW ARTICLE 1377

A review of research in numerical simulation


for the glass-pressing process
M Brown
Laboratory for Manufacturing and Productivity, Massachusetts Institute of Technology,
77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA. email: mi00brown@gmail.com
The manuscript was received on 3 February 2007 and was accepted after revision for publication on 8 May 2007.

DOI: 10.1243/09544054JEM833

Abstract: Pressing is one of the most versatile and important manufacturing processes
capable of mass-producing complicated glass parts in net shape with excellent dimensional
tolerance. Numerical simulation of glass pressing has been an active research area for many
years, as part quality and yield requirements become more stringent. This paper reviews the
research and development in numerical simulation of glass pressing. It organizes prior
studies into three categories, namely, glass flow, heat transfer, and residual stresses. Based on
the research progress to date, this paper suggests that the coupled and integrated simulation
considering material, mould, and process, the use of empirical models and soft computing
techniques to complement the numerical simulation, and small-scale experiments are the three
most important areas for further advancement.

Keywords: glass pressing, numerical simulation, glass flow, heat transfer, residual stresses

1 INTRODUCTION The advantages of numerical simulation are that,


once it has been verified by comparison to real-life
Pressing is widely used for mass producing discrete situations, it can deal with a large variety of situations
glass parts of complex shape cost-effectively with high and parameters with the minimum of further effort
precision, which gives the glass gob its shape by and there is a problem recovering variables for any
pressing between a mould and a punch. The classical section of the glass at any time during the pressing
usage of glass pressing is as containers. The advanced operation.
applications are also in optics, or electronics, such as Faced with global competition in the glass industry,
picture tube panels. With the rapid development of it is desirable to have analysis tools for improving the
electronic products such as TV sets and computers, mould design and process control in glass produc-
the need for glass-pressed parts increases greatly. tion. Quite a few researchers have attempted various
The major concern in glass pressing is to produce approaches in this area in order to reduce the time to
the formed parts in good quality with an appropriate market and obtain a consistent quality of pressed parts.
manufacturing speed. The quality of formed parts The aim of this article is to review the research in
could be described by their mechanical character- the numerical simulation of glass pressing. Research
istics, optical properties, dimensional conformity, and for different topics, including glass flow, heat trans-
appearance. The quality is affected by many factors fer, and residual stresses, are discussed. The subject
of which the pressing process is a principal. is too large to be covered extensively and summarized
Interrupting the conditions-forming cycle to con- in a short paper; therefore, only an overview will be
duct a series of experiments would be disruptive, given, with an emphasis on showing main aspects
and also factory floor experimentation is not always where numerical simulation has been implemented.
a viable option owing to the very hostile environ-
ment. Large rates of change of temperature are
envisaged, so any instrumentation would need to be 2 GLASS FLOW
both sensitive and robust; also, there is the added
complication of the glass being inaccessible through In a pressing operation the processes involved are
much of the forming cycle. fluid and heat flow. The molten glass is driven by the

JEM833  IMechE 2007 Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


1378 M Brown

action of the plunger. The mould and plunger are not driven by the glass industry. In 1985, six of the world’s
normally at the same temperature as the gob, so there leading glass-container producers established the
will be heat flow between the glass and its immediate research and development partnership, International
surroundings as well as redistribution of heat within Partners in Glass Research (IPGR) [6, 7]. Within this
the glass. This temperature variation within the mol- partnership, a forming program named IFORMTM was
ten glass will affect its flow because the viscosity is developed as early as 1986, which can simulate
strongly temperature dependent. In this section, more parison- and container-forming processes and pro-
attention is paid to the fluid flow problem, and litera- vide final wall thickness distribution [8, 9]. The work
tures that focused on heat transfer will be discussed in done by this group (now Rockfield Software Ltd) is still
the following section. being developed and is probably the first code utili-
In the derivation, the basic assumptions are: zed at the plant level by manufacturers. The software
is now named GD, a glass-container modelling suite
(a) the material is incompressible and purely viscous;
which has four main modules: IFORM, ISTRESS,
(b) inertia is neglected as compared to the viscous
IDYNE, and MOULDS [10].
force.
In universities, Humpherys presented a fundamen-
With these approximations, the flow equations can tal work in this area, and his PhD thesis provided a
be written as simplified mathematical model for axial-symmetrical
@ui pressing [11].
¼ 0 ði ¼ 1, 2, 3Þ ð1Þ Rekhson et al. [12] constructed a model that pre-
@xi
dicted quantitatively the wall thickness distribution
  in a glass bulb, formed by ribbon process. A new
@ @ui @uj @Pi
h þ  ¼ 0 ði ¼ 1, 2, 3, j ¼ 1, 2, 3Þ ð2Þ level of understanding of the effects of various pro-
@xj @xj @xi @xj
cess variables was indicated.
where xi is the dimensional coordinate and ui is the Rekhson et al. [13] analysed the thermal effects on
velocity component, with P being the pressure. In melting, forming, and cooling using GlassProTM, a
addition, h denotes the viscosity. tool for interactive computing. Interactive computing
Unlike polymers, inorganic glass melts are reas- allowed for the immediate use of computer models
onably isotropic (especially silicate-based glass) and with no additional development work. The ‘fountain’
thus show a broad region of strain-independent effect in pressing and the influence of the overall
(Newtonian) viscosity behaviour, albeit with a high temperature distribution were demonstrated.
temperature dependence; that is, at any given temp- Lochegnies et al. [14] introduced numerical simu-
erature there is a linear relationship between stress lations for manufacturing problems during pressing,
and velocity gradient. A number of empirical formu- blowing, centrifuging, and the extrusion of hollow
lae have been suggested to fit the available data. glass items. For pressing, the necessary presence of
One such is the Fulcher equation [1] an added feeder was confirmed by the finite element
B distributions of the strain rate in the item and the
log10 h ¼ A þ ð3Þ
T  T0 evolution of the pressing force.
Sadegh et al. [15] developed a finite element model
where the constants, A, B, T0, vary according to the
of coupled thermomechanical systems in two dim-
actual composition of the glass.
ensions subject to large deformation and general
In fact, glass melts are non-Newtonian at a high
boundary conditions. The model sought a systematic
shear rate. Li and Uhlmann [2] made the first clear
approach in implementing boundary conditions. As
observation of this fact. Their work demonstrated the
such, the constraint forcing term was obtained via
presence of shear thinning in inorganic glass melts,
projections into the range space and null space of
resulting in a decreased viscosity with increased load
the constraint matrix. The technique was illustrated
or shear rate. Simmons performed a long-term study
with a simulation example.
of this phenomenon [3–5], and developed a widely
Hyre et al. [16–18] performed a series of work on
applicable SRD equation specific to describing the
glass-forming simulation and developed the first
strain rate and temperature dependence of the viscos-
comprehensive simulations of the overall container
ity for the soda lime silica family of glasses [4]
forming process – from the forehearth to the final con-
h0 ðT Þ tainer. Their work was also compared to manufactur-
_ TÞ ¼
hð", ð4Þ ing data, adding some level of validity to the results.
_ 0:76
1 þ 3:5 · 106 "h 0 ðT Þ
Some of their work used POLYFLOWTM CFD package
_ and T are the Newtonian viscosity, shear
where h0, ", as the simulation platform.
rate, and temperature respectively. Li et al. [19] described the methodology of a finite
As for the numerical simulation of the glass- element method-based three-dimensional bulk-
pressing process, much of the early work done was forming modelling program by using DEFORMTM as

Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture JEM833  IMechE 2007

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


Numerical simulation for the glass-pressing process 1379

an example, and employed it to model the TV panel- industries (container, flat, and fibre), such as Schott
forming process, especially the glass flow during Glas, Owens Corning, PPG Industries, Stein Heurtey,
pressing. Virginia Military Institute, etc. [17, 18, 26, 27].
Rienstra and Chandra [20] developed a form of
Reynolds’s lubrication flow equations by application
of the slender geometry of mould and plunger and a 3 HEAT TRANSFER
cylindrical symmetry. These equations were solved by
utilizing the incompressibility of the glass, by which Within the glass, the energy equation is derived from
the flux at any axial cross-section was determined for application of conservation of energy to the fluid as
prescribed plunger velocity, leading to analytical res-    
@T @T @ @T
ults in closed form for velocity field and pressure gra- rCV þ ui ¼ Ktrue þ F þ qv ð5Þ
@t @xi @xi @xi
dient. The glass level was implicitly defined by the
integral over the varying volume, which was to remain where T, r, CV, and Ktrue denote the temperature,
constant. density, specific heat, and true thermal conductivity
Olaf Op den Camp et al. [21] at TNO developed a respectively, and F is the heat generated by viscous
software system for the simulation of all process steps dissipation, which can be calculated by
that are required for the production of TV panels that  
h @ui @uj 2
was implemented upon the open software package F¼ þ ð6Þ
2 @xj @xi
SEPRAN. The software package involved the simula-
tion of the forming process from mixing in the spout In addition, qv is called internal heat generation,
to gob forming, pressing, cooling, and annealing. which is associated with the internal radiation in
Feulvarch et al. [22] studied the simulation of glass the glass. This internal radiation can be modelled by
forming by means of a solid approach where the increasing the conductivity to approximate its effects
degrees of freedom were displacements, and pro- on heat transport, i.e. the effective conductivity is a
posed a linear finite element formulation for three- gross approximation made to include internal radia-
dimensional glass forming where the degrees of tion as a diffusive process. Because the proximity
freedom were velocity and hydrostatic pressure. The of solid boundaries lessens internal radiation, it has
mechanical contact was taken into account by means only a minor role in determining the temperature dis-
of an explicit algorithm related to the time integration tribution and heat flux close to a glass/metal interface
scheme. The thermal coupling was not considered. [28]. McGraw [29] asserted that the thinner the glass
Zhou et al. [23] presented a three-dimensional becomes, the less the effect of radiation within. On the
finite element model to deal with the glass flow and other hand, internal radiation is difficult to quantify
heat transfer during the pressing process in TV panel as each small portion of glass absorbs and re-admits
production. In the model, an equal-order velocity– energy. The amount of internal radiation depends
pressure formulation method was employed, as in on many factors, such as the shape of the article, its
their work for injection moulding [24]. opaqueness, or temperature. Therefore, during the
It should also been mentioned that Krause and actual forming operation it is reasonable to assume
Loch [25] edited a book entitled Mathematical simu- that the effect of internal radiation is adequately des-
lation in glass technology, which reported Schott’s cribed by replacing the true thermal conductivity by
work on a large variety of mathematical simulations, an effective thermal conductivity that allows for the
covering all production steps of special glass man- ‘extra’ transport of heat due to the internal radiation
ufacturing: melting, fining, mixing, homogenizing, [30, 31].
hot and cold forming, thermal treatment, and post- As for the boundary conditions, the heat is lost from
processing. the surface of the glass by conduction, convection,
To sum up, the various research efforts for the glass- and radiation. At a glass/air interface, radiation plays
flow simulation mentioned above can be separated an important role, while at a glass/metal interface,
into four primary technologies. Most of the works conduction is the major method of heat transport
have focused on interface tracking techniques, such from the glass to the mould because metal is opaque
as POLYFLOW, DEFORM, IFORM, etc. Another cate- to radiation. This kind of conduction is almost always
gory is the interface capturing techniques, such as the modelled as a contact conductance problem. The
work of University Eindhoven, TNO, etc. Zhou used interface heat-transfer coefficient must be treated as
the BEM method. A newer area is the meshless meth- a variable, as a function of materials, contact time,
ods being introduced that were presented at the first contact pressure, temperature, etc. [29, 32].
MAGICAL meeting (MAGICAL: Mathematics for the Heat transfer in glass-forming operations depends
Glass Industry) by ITWM in 2003. As for the commer- on many factors, e.g. the properties of the glass,
cial software packages, POLYFLOW has the widest use heat-transfer coefficients at the glass/mould and
within industry, including all three of the major glass mould/air interfaces, the characteristics of the mould

JEM833  IMechE 2007 Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


1380 M Brown

material, and the glass/mould contact time [32]. contact conductance between molten glass and
Because many variables influence it, an exact mathe- container-forming moulds. The experimental appar-
matical analysis of the heat transfer associated with atus was capable of independently varying the
the forming process is difficult. However, consider- glass pressure, glass temperature, mould temperat-
able progress has been made in understanding this ure, and glass type. Initial validation of the experi-
complex problem through both experimental and mentally determined contact conductance function
theoretical analyses. in conjunction with the one-dimensional heat trans-
McCraw [29] found that the rate of heat removal fer model utilized within a glass-forming model
from glass is controlled to a great extent by the glass/ indicated good agreement between calculated and
mould heat-transfer coefficient during the period measured results [40].
of contact and decreases rapidly with the time of Viskanta and Lim [41] described a theoretical study
contact. to investigate internal heat transfer in glass under-
Fellows and Shaw [33] provided the first laboratory going cooling between glass and mould, as well as
experiment to investigate the thermal contact bet- plunger, during and after pressing. A thermal model
ween glass and mould during pressing. The variation was formulated to simulate the cooling. The heat-
of the glass–mould heat-transfer coefficient was used transfer analysis accounted for the spectral nature
as a measure of the thermal contact in relation to of radiation in glass, the dependence of the thermo-
other variables such as the initial mould temperature physical properties of glass on temperature, and the
and contact pressure. contact heat transfer between and after pressing, as
Hayashi et al. [34] simulated the pressing process well as subsequent cooling. Heat exchange between
in the manufacturing of glass containers to study glass and mould by contact conduction across a very
the effects of initial temperature distribution of a small gap and that by thermal radiation were consid-
body of molten glass called a gob, heat transfer at ered separately. Numerical solutions were obtained
the glass–mould boundary, and the pressing pressure. for typical conditions simulating symmetric and non-
It was proven that the initial temperature distribution symmetric cooling, and the results obtained were
influences the final temperature distribution and presented and discussed.
deformation speed, and that heat transfer at the wall Hohne et al. [42] set up a laboratory testing unit
boundary also affects the final temperature distribu- to investigate the influence of radiation emitted by
tion and deformation speed due to heat loss. different glass compositions and the influence of
Lochegnies et al. [35] developed an incremental pro- different mould materials used during forming. To
cedure for the thermal analysis in three-dimensional support the experimental results, heat-transfer con-
glass products considering the one-dimensional ana- ditions and temperatures were modelled at given
lytical solutions for a semi-infinite glass wall. When boundary conditions using the CFD code FLUENT
coupled to a code being capable of computing large software program applying discrete models.
mechanical deformations, this allowed the glass flow Druma et al. [43] presented a finite element model
during the pressing, self-deformation, and blowing to determine the temperature history in forming a
processes to be analysed efficiently. The forming of TV glass panel during hot pressing in a metal mould
a reference tumbler was analysed via finite element and the controlled cooling cycle, and DEFORM
simulations. For each forming step, the temperat- 3D software was used to carry out the numeric
ure distribution was calculated from the pressing simulation.
parameters; pertinent results were found for this Choi et al. [44] studied an inverse problem of heat
application. transfer in the glass-forming process to determine a
Siedow et al. [36–39] at Schott evaluated all aspects number of unknown heat-transfer coefficients. An
of glass/mould heat transfer in precision glass press- analysis program for transient heat conduction was
ing. For example, they performed numerical simula- developed to simulate the forming and cooling pro-
tion of three-dimensional radiative heat transfer in cesses. The analysis involved the contact and separa-
glasses and glass melts, by using an improved three- tion of glass and moulds and it was repeated until
dimensional diffusion approximation [36]. Their book the process reached a periodic state. Temperature
[37] modelled the heat transfer in glass melting, hot measurements were taken at several available times
forming, thermal treatment, etc. and spatial positions of glass and moulds during the
Hyre et al. [17, 18, 40] performed a continuous forming operation. Heat removal by the cooling water
study on glass/mould heat transfer. They developed from the plunger was also recorded. An optimization
a simple one-dimensional heat-transfer model in problem was formulated to determine the heat-transfer
conjunction with a complex three-dimensional form- coefficients that minimize the difference between the
ing model to determine the heat flow between the measured data and the analysis results.
glass and forming mould. Furthermore, they described Zhou and Li [45, 46] also pointed out that the
an experimental effort to determine the time-varying continuous operation of glass pressing results in a

Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture JEM833  IMechE 2007

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


Numerical simulation for the glass-pressing process 1381

steady state cyclic heat transfer within the mould. Schwarzl and Staverman [48] classified such materials
This temperature field can be decomposed into two as being ‘thermorheologically simple’ and specified
parts: the cycle-averaged temperature field during the conditions for the material relaxation functions
one cycle and the relatively small fluctuation field. to satisfy the time-temperature equivalence principle.
Therefore, in order to simplify the computation pro- Gurtin and Stemberg [49] extended linear visco-
cedure, a cycle-average approach was developed in elasticity to account for time-dependent tem-
the mould cooling process. perature variations by introducing the concept of
pseudo time, or material time. The model was
applied to calculate transient stresses during the
4 RESIDUAL STRESSES cooling of a plate and of a solid sphere of an iso-
tropic, linear, and thermorheologically simple mate-
In forming, molten glass is pressed into a relatively rial. However, the usual thermal stresses arising
cold mould. As the mould fills, the layers that come from non-homogeneous thermal expansion were
into contact with the mould surface solidify. After neglected.
pressing, heat transfer at the mould surface even- Lee and Rogers [50] showed that Laplace trans-
tually results in a solid part. As layers of the molten forms can be used effectively for viscoelastic stress
material begin to solidify due to the mould cooling analysis only for a restricted class of problems. For
and air cooling, they become capable of support- more general problems, a method in which the her-
ing tensile stress and of supporting shear stresses editary integral form is reduced to a set of Volterra
even in the absence of motion. The sequential solidi- integral equations, was outlined. By applying this
fication of the layers results in a build-up of self- method to a symmetrically cooled glass plate.
equilibrating residual stresses. Because glass exhibits Lee and Rogers [51] were able to account for glass
time-dependent (viscoelastic) thermomechanical relaxation effects. Measured temperature-dependent
behaviour, the local stresses ‘relax’ continually until relaxation characteristics of glass were used in the
the local temperature falls substantially below the form of a thermorheologically simple material model.
glass transition temperature. This relaxation causes Significant differences between the numerical results
the stresses to redistribute continually. for a continuous shift function and those for an elas-
The dependence of the specific volume of glass on tic material below the critical temperature indicated
the temperature and the pressure results in signi- the importance of relaxation effects. The transient
ficant local volumetric changes. These changes, stresses calculated by using a thermorheologically
together with relaxation effects, result in dimensional simple model for a glass plate quenched from well
changes (size and shape) and in residual stresses. The above the glass critical temperature differed signifi-
local strains resulting from volumetric changes dur- cantly from the experimental data.
ing solidification are small. However, these small Narayanaswamy and Gardon [52] showed that
strains and associated residual stresses can be quite these differences result from the numerical solution
important. These residual stresses increase the sus- procedure used. In particular, the trapezoidal rule
ceptibility of the part to stress corrosion cracking used by Lee et al. to approximate the integrals results
caused by following processes such as encapsulation, in large errors when the kernel is strongly non-
and temperature cycling (such as in an annealing linear, which is the case when a glass slab is cooled
cycle) can cause further dimensional changes as a from a state well above the glass critical temperat-
result of stress relaxation. ure. Narayanaswamy proposed a modified integration
Progress in predicting residual stresses and dimen- scheme in which the integral is evaluated with respect
sional changes caused by solidification is closely to the pseudo time instead of the actual time. This
related to the development of thermoviscoelast- scheme takes advantage of the smaller variations of
icity theory and to its application to inorganic glasses. the kernel function with respect to the pseudo time.
The complex, intractable form of the thermovisco- Aggarwala and Saibel [53] considered transient
elastic constitutive equations lead to the develop- stresses in circular plates and cylinders induced by
ment of simpler procedures, many of which are tempering. The response of the glass to shear stress
special cases of the thermoviscoelastic approach. was assumed to be viscoelastic (Maxwell model), the
Linear viscoelasticity is concerned with material viscosity being infinite below a certain critical temp-
behaviour at different but constant temperatures. The erature and zero above it. The volumetric response
linear viscoelastic behaviour at different constant tem- of the material was assumed to be elastic (time-
peratures is related in a particularly simple manner: independent bulk modulus), with the temperature-
the variation of the relaxation modulus versus the induced material expansion during cooling accounted
logarithm of time is the same at different tempera- for through a temperature-dependent thermal expan-
tures, except for a shift in the time scale, which leads sion coefficient. This material model is a highly simpli-
to the concept of time-temperature equivalence [47]. fied version of the general linear thermoviscoelastic

JEM833  IMechE 2007 Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


1382 M Brown

model, in which the transition from the melt to the was used to calculate stresses and strains in a sym-
glassy state is assumed to occur over a wide temper- metrically cooled infinite plate.
ature range, rather than at one fixed temperature Daudeville and Carre [64] used Narayanaswamy’s
as assumed by Aggarwala and Saibel. The simplified model to describe the thermomechanical behaviour
model is a good approximation when the interface of glass, including both stress relaxation (to take into
region of the solidifying material is small compared account the viscous aspect of glass) and structural
with the plate thickness. This condition is well satis- relaxation (to take into account the structural state
fied for high cooling rates. of glass). Computational results were compared with
Rekhson et al. [54] at General Electric Lighting experimental results. Daudeville et al. [65] also pre-
employed the finite-element program MARC to cal- sented numerical simulation of the thermal temper-
culate thermal stresses for a simple sandwich seal ing by the three-dimensional finite element method
and a bead seal under conditions of uniform cooling, to calculate transient and residual stresses near the
reheating, and an isothermal hold and for a tempered edges of a glass plate and near holes.
glass plate. Results of the viscoelastic calculations Zhou et al. [66–69] used the solidification of a
were compared with previously obtained analytical molten layer of glass between cooled parallel plates
solutions and with experimental stress measurements. to model the mechanics of the build-up of residual
Their progressive researches included glass transition, stresses in the pressing process. Flow effects were
non-linear viscoelastic relaxation, structural relaxation neglected, and a thermorheologically simple thermo-
in the evolution of the free strain, and stress relaxation viscoelastic material model was assumed. The equili-
times [55–57]. brium thermomechanical properties of the material
Lalykin and Mazurin [58] refined and checked and the shift function were temperature- and pressure-
experimentally the algorithm previously published dependent.
for the calculation of the heat treatment of a glass, As for shrinkage simulation of glass-pressed pro-
and also presented an algorithm to develop anneal- ducts, few existing studies involved this problem.
ing regimes for flat glass. On the basis of the relaxa- Based on the above analysis of residual stresses,
tion theory of annealing, an interpretation was Zhou et al. [70] developed a finite element simula-
proposed for the physical nature of the decrease in tion for the calculation of shrinkage. The shells model
the effective thermal-expansion coefficient of a glass for injection moulding [71–73] was brought in, as
product with an increase in the cooling rate of the an assembly of flat elements. The approach allowed
glass product [59]. They also presented a universal the prediction of residual deformations layer by
method for analysing the stresses that are caused by layer like a truly three-dimensional calculation, while
differing thermal expansion values of materials and reducing the computational cost significantly. An
depend on the processes of structural and mechanical experimental study was used to verify the developed
relaxation in glass [60]. simulation system.
Dumora et al. [61] developed a computer program During pressing, the dimension of the mould
based on a Maxwell thermoviscoelastic model with machined at room temperature changes due to the
time–temperature-dependent characteristics for resi- thermal and mechanical loading, which determines
dual stresses simulation, which was used in a para- the dimension of the panel at pressing temperat-
metric study of the glass-tempering process and an ure. Therefore, besides the shrinkage of the glass,
example of the determination of the stress field of the mould deformation in pressing also affects the
a glass body for a high voltage insulator. Chabrand surface shape and dimensional precision of the
et al. [62] also dealt with the mechanical and numer- formed part. Zhou et al. [74] also developed a
ical modelling of the thermal tempering of a thermo- three-dimensional thermoelastic boundary element
viscoelastic material by using a Maxwell model. model to predict the deformation occurring on the
Implicit or semi-implicit time-integration methods mould cavity surfaces owing to the thermal and
were used in the numerical solution by finite ele- mechanical loads of the process, which can suggest
ments method. the amounts by which the mould surfaces should
Mauch and Jackle [63] used the framework of the be machined.
theory of thermoviscoelasticity to model the Gardon
and Narayanaswamy theory of annealing and tem-
pering of glass in a unifying way. It was extended by 5 CONCLUSIONS
including the effect of isothermal pressure relaxa-
tion in addition to the relaxation of shear stress. In Glass-pressing simulation has gone through a period
the theory, relaxation of shear stress, isotropic stress of fast development in the past decade, accompanied
(pressure), and thermal stress, which causes thermal by several other key technologies such as numerical
expansion, were treated on the same level. The model methods and information technology. It is useful to

Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture JEM833  IMechE 2007

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


Numerical simulation for the glass-pressing process 1383

review the up-to-date advancements in this area in [77–81]. The software system developed by TNO
order to identify the right path toward the final, fully is called Glass Process Simulator (GPS), and has
optimized production. already been used in production lines.
An organization that cannot be missed is the Inter- Based on the topics discussed in this paper, the fol-
national Commission on Glass (ICG). The ICG is a lowing research issues are suggested for future studies.
non-profit-making international society of scientific
and technical organizations with particular interests 1. The final solid pressed part is obtained at the end of
in glass science and technology [75]. It was founded a series of coupled, fluid, and heat-transfer transi-
in 1933 and has grown to become the recognized ent processes. There are too many interrelated fac-
world-wide organization in the field of glass. The tors involved in the pressing process, all of which
aim of the ICG is to promote and stimulate under- can affect the formed parts quality. For example,
standing and cooperation between glass experts in the temperature distribution and flow are inter-
the fields of science and technology as well as art, related within the molten glass because its viscosity
history, and education. The greater part of the con- is strongly temperature dependent; the mould cool-
tinuing activity of the ICG is carried out by its tech- ing and temperature distribution is of great impor-
nical committees (TCs). One of its TCs is ‘Modelling tance because it significantly affects productivity
of glass forming processes’ (TC 25), whose mission and the thermal residual stresses in the final pro-
is to promote information exchange and activities in duct; and the mould deformation can also affect
the field of numerical simulation of glass-forming the geometric shape and dimensional precision.
processes on both scientific and technological levels. Therefore, glass pressing needs to be treated as an
There are glass-pressing benchmark problems set up integrated system that includes material, mould,
to evaluate the various codes and techniques. and process, like the researches done by Emhart
As mentioned earlier, much of the early work for Glass [16–18]. The well-developed computers make
numerical simulation of glass forming was carried it possible to perform a coupled and integrated
out by the glass industry, which drove much of simulation, which can consider all of the inter-
the follow-on university work. For example, Schott’s related factors involved.
research efforts provided an excellent basis for the 2. Numerical simulation could provide useful infor-
issues that have faced industrial and university res- mation in part design, mould design, and process
earchers [25, 36–39], and Emhart Glass developed design of glass pressing. However, this approach
comprehensive simulations of the overall container- involves creating a finite element model and run-
forming process [16–18, 40]. Other global electronic ning a number of simulations in order to obtain
enterprises have also provided valuable results, such the acceptable forming parameters. As the time
as GE Lighting [12, 13] and Glass Services [76]. for running a simulation could take an hour or
The academic research institutions which have more, it may not be practical to perform a num-
focused on this area include University of Sheffield ber of simulations in a shop-floor production
[11], University of Valenciennes [14], Eindhoven environment. On the other hand, some of the pro-
University of Technology [20], etc. Two important cesses in glass melting, fining, and forming are
institutions that have had active researches in recent still not possible to be mathematically modelled
years will be emphasized as follows. explicitly. Future research would tend to the use
The first is Huazhong University of Science and of empirical models and soft computing tech-
Technology in China. Zhou and colleagues have per- niques to complement the numerical simulation
formed comprehensive studies for the glass-pressing in the optimization of the glass-pressing opera-
process, which covers the whole forming process tion. These approaches include artificial neural
including glass flow, heat transfer, mould cooling, networks (ANN), case-based reasoning (CBR),
residual stresses, shrinkage, mould deformation, etc. design of experiment (DOE), expert systems, fuzzy
[23, 45, 46, 66–70, 74]. Their achievements bring logic, and genetic algorithms (GAs).
a clearer and more systematic insight to this pro- 3. More attention should be paid to experimental
cess, and are of practical value to the modern glass studies. As mentioned earlier, interrupting the
industry. conditions-forming cycle to conduct a series of
The second is TNO Institute of Applied Physics in experiments would be disruptive, costly, and time-
The Netherlands. The glass group at TNO has under- consuming. One option is to conduct a series of
taken long-term research on glass forming, and the small-scale experiments in a laboratory. Through
research interests cover a wider range, which includes this sort of experiment, the necessary basic data
the material properties, design, optimization and con- of the material and process can also be obtained,
trol of melting furnaces and tanks, melting and fin- and the mathematical model and simulation pro-
ing process, annealing and tempering process, etc. gram can be verified [33, 40, 42, 54, 64].

JEM833  IMechE 2007 Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


1384 M Brown

REFERENCES boundary conditions. J. Engng Math., 2001, 39(1–4),


241–259.
1 Tooley, F. V. The handbook of glass manufacture Vol II, 21 Camp, O. O. D., Hegen, D., Haagh, G., and Limpens, M.
1974 (New York: Books for Industry, Inc.). TV panel production: simulation of the forming process.
2 Li, J. H. and Uhlmann, D. R. The flow of glass at high Ceramic Engng Sci. Proc., 2003, 24(1), 1–19.
stress levels. I. Non-Newtonian behavior of homo- 22 Feulvarch, E., Moulin, N., Saillard, P., Lornage, T.,
geneous 0.08 RB2O.0.92 SiO2 glasses. J. Non-Crystalline and Bergheau, J. M. 3D simulation of glass forming
Solids, 1970, 3(1), 127–147. process. J. Mater. Processing Technol., 2005, 164–165,
3 Simmons, J. H., Mohr, R. K., and Montrose, C. J. 1197–1203.
Non-Newtonian viscous flow in glass. J. Appl. Phys., 23 Zhou, H. M., Yan, B., and Li, D. Q. Three-dimensional
1982, 53(6), 4075–4080. numerical simulation of the pressing process in TV panel
4 Simmons, J. H. and Simmons, C. J. Nonlinear vis- production. Simulation-Trans. Soc. Modeling and Simu-
cous flow in glass forming. Am. Ceramic Soc. Bull., lation Int., 2006, 82(3), 193–203.
1989, 68(11), 1949–1955. 24 Zhou, H. M., Geng, T., and Li, D. Q. Numerical filling
5 Simmons, J. H. What is so exciting about non-linear simulation of injection molding based on 3D finite
viscous flow in glass, molecular dynamics simulations of element model. J. Reinforced Plast. Composites, 2005,
brittle fracture and semiconductor-glass quantum com- 24(8), 823–830.
posites. J. Non-Crystalline Solids, 1998, 239(1–3), 1–15. 25 Krause, D. and Loch, H. Mathematical simulation in
6 http://wiegand-glas.de. glass technology, 2002 (Springer-Verlag, Berlin).
7 Tayler, C. High speed changes ahead for the bottling 26 Vanandruel, N. Simulation of glass processes with poly-
business. Process Engng, 1985, 66(6), 83–89. flow, a finite element program. Ceramics, 1996, 40(1),
8 Williams, J. H., Owen, D. R. J., and Cesar de Sa, J. M. A. 11–14.
The numerical modeling of glass forming processes. In 27 http://www.fluent.com/solutions/glass/index.htm.
Collected papers, XIV International Congress on Glass, 28 Rawson, H. Physics of glass manufacturing process.
1986, pp. 138–145. Phys. in Technol., 1974, 5(2), 91–114.
9 Cesar de Sa, J. M. A. Numerical modelling of glass 29 McGraw, D. A. Heat transfer in glass during the forming
forming processes. Engng Computations, 1986, 3(4), process, J. Am. Ceramic Soc., 1961, 44(7), 353–363.
266–275. 30 Manthuruthil, J., Sikri, T. R., and Simmons, G. A.
10 http://www.rockfield.co.uk/gcm.htm. Simplified mathematical model simulating heat trans-
11 Humpherys, C. E. Mathematical modelling of glass fer in glass-forming molds. J. Am. Ceramic Soc., 1974,
flow during a pressing operation. PhD Thesis, University 57(8), 345–350.
of Sheffield, 1991. 31 Clark-Monks, C. Heat transfer in the manufacture of
12 Rekhson, S., Lu, Z. H., Day, C., and Wang, Y. Computer glass containers. Glass Technol., 1974, 15(2), 31–34.
modeling of the forming process. Glass Ind., 1993, 74(5), 32 Manthuruthil, J., Sikri, T. R., and Simmons, G. A.
15–16, 19. Simplified mathematical model simulating heat trans-
13 Rekhson, S. M., Rekhson, M., Ducroux, J. P., and fer in glass-forming molds. J. Am. Ceramic Soc., 57(8),
Tarakanov, S. Heat transfer effects in glass processing. 345–350.
Ceramic Engng Sci. Proc., 1995, 16(2), 19–37. 33 Fellows, C. J. and Shaw, F. Laboratory investigation
14 Lochegnies, D., Martin, C., Carpentier, E., and Oudin, J. of glass to mold heat transfer during pressing. Glass
Finite element contributions to glass manufacturing Technol., 1978, 19(1), 4–9.
control and optimization. Part 2. Blowing, pressing 34 Hayashi, M., Hasegawa, T., and Ohiwa, N. Numer-
and centrifuging of hollow items. Glass Technol., 1996, ical simulation of glass-pressing process. Trans. JSME,
37(5), 169–174. Part B, 1995, 61(591), 4157–4162.
15 Sadegh, N., Vachtsevanos, G. J., Barth, E. J., 35 Lochegnies, D., Noiret, C., Thibaud, C., and Oudin, J.
Pirovolou, D. K., and Smith, M. H. Modelling the glass Computation procedure for the temperature in hot glass
forming process. Glass Technol., 1997, 38(6), 216–218. application to the finite element simulation of hollow
16 Hyre, M. Numerical simulation of glass forming glass forming. Glass Sci. Technol., 1996, 69(8), 253–264.
and conditioning. J. Am. Ceramic Soc., 2002, 85(5), 36 Lentes, F. T. and Siedow, N. Three-dimensional radia-
1047–1056. tive heat transfer in glass cooling processes. Glass Sci.
17 Hyre, M. R. Effect of mould to glass heat transfer on glass Technol.: Glastechnische Berichte, 1999, 72(6), 188–196.
container forming. Ceramic Trans., 2004, 141, 271–279. 37 Brinkmann, M. and Siedow, N. Heat transfer between
18 Hyre, M. R. and Rubin, Y. Modelling of gob and glass and mold during hot forming. In Mathematical
container forming. Int. J. Forming Proc., 2004, 7(4), simulation in glass technology (Ed. D. Krause and
443–457. H. Loch), 2002, pp. 239–262 (Springer-Verlag, Berlin).
19 Li, G., Jinn, J. T., Wu, W. T., and Oh, S. I. Recent devel- 38 Siedow, N., Grosan, T., Lochegnies, D., and Romero, E.
opment and applications of three-dimension finite Application of a new method for radiative heat transfer
element modeling in bulk forming processes. J. Mater. to flat glass tempering. J. Am. Ceramic Soc., 2005, 88(8),
Processing Technol., 2001, 113(1), 40–45. 2181–2187.
20 Rienstra, S. W. and Chandra, T. D. Analytical approxima- 39 Fotheringham, U. and Lentes, F. T. Active thermal con-
tions to the viscous glass-flow problem in the mould- ductivity of hot glass. Glass Sci. Technol. Glastechnische
plunger pressing process, including an investigation of Berichte, 1994, 67(12), 335–342.

Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture JEM833  IMechE 2007

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


Numerical simulation for the glass-pressing process 1385

40 Hyre, M. R. and Underwood, B. L. Experimental meas- 57 Rekhson, S. M. Thermal stresses, relaxation, and hys-
urement of glass to mould heat transfer for computa- teresis in glass. J. Am. Ceramic Soc., 1993, 76(5),
tional modeling of container production. Innovations 1113–1123.
in engineering education 2004, mechanical engineer- 58 Lalykin, N. V. and Mazurin, O. V. Mathematical model
ing education, mechanical engineering technology for the process of annealing flat glass. Glass and
department heads, Anaheim, CA, United States, 2004, Ceramics, 1984, 41(1–2), 9–13.
pp. 319–324. 59 Mazurin, O. V. and Fridkin, R. Z. Effect of the cooling
41 Viskanta, R. and Lim, J. M. Theoretical investigation rate of a glass on the contribution to the residual stresses
of heat transfer in glass forming. J. Am. Ceramic Soc., of the structural component of the thermal expansion
2001, 84(10), 2296–2302. coefficient. Soviet J. Glass Phys. Chem., 1986, 12(6),
42 Hohne, D., Pitschel, B., Merkwitz, M., and Lobig, R. 362–369.
Measurement and mathematical modelling of the heat 60 Borovinskii, S. V. and Mazurin, O. V. Universal method
transfer in the glass forming process, in consideration of stress analysis of viscoelastic-elastic composites.
of the heat transfer coefficients and radiation influ- J. Am. Ceramic Soc., 1986, 69(2), 102–106.
ences. Glass Sci. Technol., 2003, 76(6), 309–317. 61 Dumora, D., Saisse, H., Knosp, B., Goudeau, J., and
43 Druma, C., Alam, M. K., Druma, A. M., and Hoque, A. Licht, C. Thermal tempering study of glass insulators
Finite element analysis of TV panel glass during cool- by means of a finite element modellization. Indian
ing. Mater. Mfg Processes., 2004, 19(6), 1171–1187. Ceramic Soc., 1986, 3, 151–160.
44 Choi, J. H., Ha, D. S., Kim, J. B., and Grandhi, R. V. 62 Chabrand, P., Licht, C., Maisonneuve, O., and Raous, M.
Inverse design of glass forming process simulation Residual thermal tempering stresses. Comput. Struct.,
using an optimization technique and distributed com- 1989, 31(6), 1003–1011.
puting. J. Mater. Processing Technol., 2004, 148(3), 63 Mauch, F. and Jackle, J. Thermoviscoelastic theory
342–352. of freezing of stress and strain in a symmetrically
45 Zhou, H. M. and Li, D. Q. Mold cooling simulation cooled infinite glass plate. J. Non-Crystalline Solids,
of the pressing process in TV panel production. Simu- 1994, 170(1), 73–86.
lation Modelling Practice and Theory, 2005, 13(3), 64 Daudeville, L. and Carre, H. Thermal tempering simu-
273–285. lation of glass plates: Inner and edge residual stresses.
46 Zhou, H. M. and Li, D. Q. Modeling and simulation J. Thermal Stresses, 1998, 21(6), 667–689.
of heat transfer for glass bulb mold. Progress in Nat. 65 Daudeville, L., Bernard, F., and Gy, R. Residual stresses
Sci., 2005, 15(7), 650–655. near holes in tempered glass plates. Mater. Sci. Forum,
47 Leaderman, H. Rheology (Ed. F. R. Etrich), 1958 (Aca- 2002, 404–407, 43–48.
demic Press: New York). 66 Zhou, H. M., Xi, G. D., and Li, D. Q. Residual thermal
48 Schwarzl, F. and Staverman, A. J. Time-temperature stresses simulation of television panel in the forming
dependence of linear viscoelastic behavior. Applied process. Part 1: modeling. Proc. IMechE, Part C:
Phys., 1952, 23(8), 838–843. J. Mechanical Engineering Science, 2006, 220(C5),
49 Gurtin, M. E. and Sternberg, G. E. Further study of ther- 573–582.
mal stresses in viscoelastic materials with temperature- 67 Zhou, H. M., Feng, W., Geng, T., and Li, D. Q. Residual
dependent properties. In Proceedings of International thermal stresses simulation of television panel in the
Symposium on Second-order effects in elasticity, plasti- forming process. Part 2: simulations and validation.
city and fluid dynamics, Haifa, Israel, 1962, pp. 51–76. Proc. IMechE, Part C: J. Mechanical Engineering Science,
50 Lee, E. H. and Rogers, T. G. Solution of stress analy- 2006, 220(C5), 583–591.
sis problems for linear viscoelastic materials based 68 Zhou, H. M., Sun, Q., Xi, G. D., and Li, D. Q. Numerical
on measured deformation functions. ASME Trans. J. prediction of process-induced residual stresses in glass
Appl. Mech., 1963, 30, 127–133. bulb panel. Appl. Math. Mech.-Engl. Edn, 2006, 27(9),
51 Lee, E. H. and Rogers, T. G. On the generation of 1197–1206.
residual stresses in thermoviscoelastic bodies. ASME 69 Zhou, H. M., Xi, G. D., and Li, D. Q. Modeling and
Trans., J. Appl. Mech., 1965, 32, 874–880. simulation of residual stresses during glass bulb press-
52 Narayanaswamy, O. S. and Gardon, R. Calculation ing process. Sci. China Series E: Engng Mater. Sci.,
of residual stresses in glass. J. Am. Ceramic Soc., 1969, 2007, 50(1), 103–117.
52(10), 554–558. 70 Zhou, H. M., Xi, G. D., and Li, D. Q. Modeling
53 Aggarwala, B. D. and Saibel, E. Tempering stress in and simulation of shrinkage during the picture tube
circular plates and cylinders. J. Math. Phys. Sci., 1968, panel forming process. J. Mfg Sci. Engng-Trans. ASME,
2, 321–332. 2007, 129(2), 380–387.
54 Soules, T. F., Busbey, R. F., Rekhson, S. M., 71 Zhou, H. M. and Li, D. Q. Integrated simulation
Markovsky, A., and Burke, M. A. Finite-element cal- of the injection moulding process with stereolitho-
culation of stresses in glass parts undergoing viscous graphy molds. Int. J. Advd Mfg Technol., 2006, 28(1–2),
relaxation. J. Am. Ceramic Soc., 1987, 70(2), 90–95. 53–60.
55 Rekhson, S. M. Computer modeling of glass behavior. 72 Zhou, H. M. and Li, D. Q. Numerical simulation and
J. Non-Crystalline Solids, 1990, 123(1–3), 26–35. experimental study of warpage of injection-moulded
56 Rekhson, S. A model for non-linear viscoelastic relaxa- parts. Polymer-Plast. Technol. Engng, 2005, 44(4),
tion. J. Non-Crystalline Solids, 1991, 131(2), 467–475. 603–617.

JEM833  IMechE 2007 Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014


1386 M Brown

73 Li, D. and Zhou, H. Modelling and simulation of 78 Muysenberg, H. P. H., Bauer, R. A., and Peters, E. G. J.
residual stress and warpage in injection moulding. Advanced control of glass tanks using simulation
Proc. Instn Mech. Engrs, Part C: J. Mechanical Engineer- models and fuzzy control. Ceramic Engng Sci. Proc.,
ing Science, 2004, 218(C5), 521–530. 1998, 19(1), 127–135.
74 Zhou, H. M., Geng, T., and Li, D. Q. Deformation simu- 79 Camp, O. O. D., Verheijen, O., and Kahl, S. R. Appli-
lation of the panel mould in the pressing process. cation of fast dynamic process simulation to support
Engng Analysis with Boundary Elem., 2005, 29(9), glass furnace operation. Ceramic Engng Sci. Proc.,
894–902. 2004, 25(1), 197–207.
75 http://www.icg.group.shef.ac.uk/. 80 Verheijen, O. S. Advanced operation support
76 http://www.gsl.cz/. system for redox control. Ceramic Trans., 2004, 141,
77 Beerkens, R. G. C., Faber, A. J., Muysenberg, E., 421–428.
and Simonis, F. Upgrading glass melting technology 81 Beerkens, R. G. C. and Van Der Schaaf, J. Gas release
by model-based processing. Glass Sci. Technol., 1997, and foam formation during melting and fining of glass.
70(8), 256–257. J. Am. Ceramic Soc., 2006, 89(1), 24–35.

Proc. IMechE Vol. 221 Part B: J. Engineering Manufacture JEM833  IMechE 2007

Downloaded from pib.sagepub.com at The University of Manchester Library on November 1, 2014

Вам также может понравиться