451

Finite element analysis of temperature distribution

using ABAQUS for a laser-based tile grout sealing
process

J Lawrence* and L Li
Manufacturing Division, Department of Mechanical Engineering, University of Manchester Institute of Science and
Technology, Manchester, UK

Abstract: A ®nite element (FE) model has been developed utilizing the commercial ABAQUS package
to determine the temperature distribution generated during the high-power diode laser (HPDL)
processing of a novel ceramic tile grout seal, known as an amalgamated oxide compound grout
(AOCG). The model was developed speci®cally to ®nd the temperatures reached at the base of the
ceramic tile grout seal, since temperatures in excess of 100 8C must be attained so as to ensure
certain chemical changes (the complete irreversibility of the water-glass reaction required by the
process). Fourier's law and Laplace's equation were employed to build the temperature ®eld
equations under appropriate boundary conditions in two-dimensional coordinates, while the
ABAQUS FE model was used to calculate the temperature ®eld and model the thermal pro®le in
the ceramic tile grout seal. Both the FE-generated thermal pro®les and the resultant operating maps
predicted temperatures and vitri®cation depths which were consistently higher than those obtained
experimentally. Notwithstanding this, a comparison of the results revealed that the predicted depths
of vitri®cation were in very close agreement with those observed on AOCG samples treated with the
HPDL.

Keywords: high-power diode laser (HPDL), ®nite element, FE, ABAQUS, thermal pro®le, ceramic,
grout, enamel

NOTATION 1 INTRODUCTION

k thermal conductivity Ceramic tiles are considered to be one of the most

kx thermal conductivity in the x direction cleanable surfaces available [1], and are applied to the
ky thermal conductivity in the y direction walls and ¯oors in a whole host of places. Currently, cera-
n node mic tiles are sealed with tile grout (typically epoxy based).
q generated heat ¯ux A major diculty with tiled surfaces is that contaminants
Q generated heat can enter into and exit a space via a tiled surface, through
t time the tile grouts used to ®ll the void between adjoining tiles
T surface temperature [1]. Moreover, the predominant problem with commer-
Ti surface temperature after laser beam cially available tile grouts is that because they are organic
interaction based, porous and very dicult to clean, over time they
To original surface temperature become contaminated, and have to be removed physically
or mechanicallyÐan arduous and costly undertaking. A
 thermal di€usivity novel technique has been developed [2, 3] whereby a new
ti thermal boundaries (i ˆ 1, 2, . . . , 8† tile grout was developed, namely the amalgamated oxide
r three-dimensional operand compound grout (AOCG), which, after high-power diode
laser (HPDL) surface treatment, was sealed by means of
The MS was received on 12 April 1999 and was accepted after revision the application of a vitreous enamal glaze which was ®red
for publication on 15 October 1999. using the HPDL.
*Corresponding author: Manufacturing Division, Department of
Mechanical Engineering, UMIST, PO Box 88, Manchester M60 1QD, The development of an accurate analysis procedure for
UK. many laser applications, including the ceramic tile grout
B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B
452 J LAWRENCE AND L LI

sealing process, is extremely complicated due to the ceramic tile grout materials after laser processing, and
multitude of process parameters involved. Moreover, not necessarily during processing, from which it will be
laser process parameters must be carefully selected in possible to obtain such information. However, in order
order to achieve satisfactory results. Thus the develop- to fully verify the model and ultimately obtain such
ment of a suitable model to predict such changes information, the transient condition across the entire
can not only eliminate the need to perform experiments range of the heating cycle leading to the steady state con-
with every possible range of operating parameters, but dition is examined. Furthermore, the creation of a model
can also provide information outside the range of that generates thermal contours rather than an isochro-
operating parameters currently possible. Furthermore, matic model allows time to be considered in an analysis,
mathematical modelling is required in order not only thus allowing the prediction of the e€ects of more laser
to predict the behaviour of the two-stage ceramic tile parameters, such as traverse speed, into the ®nal
grout seal during processing with various laser param- model. Also, thermocouple readings along with the
eters but also to investigate the basic process mechanisms results of the FE model revealed that during the second
in detail. In particular, it is needed to relate the physics of stage of the sealing process the maximum temperatures
the thermal transfer process and the subsequent chemical reached were consistently lower than those reached
changes within the interaction zone and its vicinity to the during the ®rst stage [14]. Since no further thermally
process parameters. induced chemical reactions will occur as a result of the
For many laser processes a large number of analytic second stage of the process, its inclusion in this current
models currently exist for the prediction of the thermal study was not deemed necessary.
behaviour of materials during laser surface processing.
These are mainly based on classic analytical solutions
to temperature distribution derived by Jeager [4],
Rosenthal [5], Carslaw and Jeager [6] and Patankar [7]. 2 CERAMIC TILE GROUT SEALING PROCESS
However, any model developed to predict the tempera-
ture pro®le in the two-stage ceramic tile grout seal Because of the attendant inadequacies of current
should give the temperature pro®le throughout ceramic commercially available tile grouts, a new grout was
tiles and the substrate material at any stage during the developed using a range of vitri®able compounds in
process. In order to achieve this, the model developed suitable ratios with other substances, and indeed with
can predict the temperature pro®les in the materials each other, in order to optimize the bene®cial properties
during both the ®rst and second stages of the ceramic and reduce the detrimental properties of the constituent
tile grout sealing process, wherein the HPDL beam was materials to form the AOCG. The composition of the
incident on the untreated AOCG and the enamel frit. AOCG is given elsewhere [2].
Because of the complex boundary conditions present in In order to form a manageable paste, the AOCG was
the two-stage ceramic tile grout seal, use of such analyti- mixed with approximately 50 wt % water-diluted sodium
cal techniques is likely to be so involved and the compu- silicate solution. Sodium silicate solution, or water-glass
tation so laborious, if possible at all, that a numerical as it is more commonly known, is a viscous colourless
method is the only feasible option. solution of colloidal sodium silicate. It is a silica-contain-
The numerical method used was the software package ing aqueous solution that, when combined with other
ABAQUS, a commercially available general-purpose solutions such as the AOCG, forms a gel-like mass of
®nite element (FE) package developed by HKS Inc. silicate hydrate. Such a mass remains soft and malleable
This selection was made on the grounds of the wide- until it is exposed to CO2 gas, either by means of a direct
spread availability of the package, its relative ease of ¯ow or through contact with the atmosphere, whereupon
use and, most importantly, because it produced the it hardens as a result of the complete hydrolysis of the
required thermal data in the desired format. Because of dissolved sodium silicate by the reaction [15]
these attributes ABAQUS has been employed by a
number of workers to model various laser materials pro- Na2 SiO3 ‡ CO2 ! Na2 CO3 ‡ SiO2 …colloidal† …1†
cessing techniques such as laser welding [8], laser brazing
[9] and laser forming [10±12]. Notwithstanding this, exposure of the hardened mass to
The principal mode of heat transfer examined in the water results in a reversal of the process and the mass
model was conduction. This was mainly for reasons of returns to a gel-like state. However, heating the hardened
simpli®cation, but also because in the HPDL ceramic mass in e€ect ®res the water-glass (similar to that of a
tile grout sealing process conduction appears to be the ceramic material) [15], increasing its strength and
predominant heat transfer mechanism [13]. The primary enabling it to withstand exposure to water.
aim of the analysis was to determine the temperature- The variations in thermal characteristics of the AOCG
induced chemical changes that occur in the ceramic tile with increased temperature were investigated in terms of
grout seal, in particular the AOCG, as a result of the compressive strength [2]. It was found that the
HPDL interaction. Thus, it is the condition of the strength of the AOCG in terms of increasing temperature
Proc Instn Mech Engrs Vol 214 Part B B02799 # IMechE 2000
 FE ANALYSIS OF TEMPERATURE DISTRIBUTION USING ABAQUS FOR A LASER-BASED PROCESS 453

is determined primarily by the water-glass, since the treatment arrangement of the AOCG during the ®rst
strength of the individual oxide compounds varies little stage of the two-stage ceramic tile grout sealing process.
or not at all at relatively low temperatures (<500 8C) Here, pairs of ceramic tiles were applied on to an ordin-
[16]. Moreover, at temperatures above approximately ary Portland cement (OPC) substrate in the standard
100 8C the complete irreversibility of the water-glass manner with a conventional epoxy tile grout. The multi-
reaction, i.e. the rehydration of the water-glass when mode defocused laser beam (Fig. 1b) was ®red once
exposed to water, was observed. Thus, when the HPDL across the surface of the AOCG placed in the void
surface treatment is used on the AOCG it is essential (standard 2 mm wide) between adjoining vitri®ed cera-
that the temperature at the base of the AOCG is in mic tiles. This was achieved by traversing the samples
excess of 100 8C. beneath the laser beam using the x and y axes of the
The laser used in the study was a surgical HPDL CNC gantry table at speeds ranging from 1 to 20 mm/s,
(Diomed Limited), emitting at 810 nm  20 nm and while 3 l/min of coaxially blown O2 assist gas was used
operating in the continuous-wave (CW) mode with to shield the laser optics as the fumes produced were
rated optical powers ranging from 0 to 60 W. The laser removed with an extraction system.
beam was delivered to the work area by means of a 4 m
long, 600 mm core diameter optical ®bre, the end of
3 MATHEMATICAL MODEL
which was connected to a 2 :1 focusing lens assembly
mounted on the z axis of a three-axis computer numeri-
cally controlled (CNC) gantry table. The AOCG was In general, a steady state heat conduction problem
irradiated using the defocused high-order mode HPDL consists of ®nding the temperature at any point within
beam with a beam spot diameter of 1.75 mm and laser a speci®ed solid whose surface has been subjected to a
powers (measured at the workpiece after ®bre and known set of boundary conditions. Fourier's law and
optics losses using a Power Wizard powermeter) of 10± Laplace's equation are utilized to assemble the tempera-
45 W. Figure 1a illustrates the typical HPDL surface ture ®eld equations under appropriate boundary condi-
tions in the two-dimension Cartesian coordinates, while
the FE is employed to calculate the temperature ®eld
and model the heat transfer. According to Fourier's
law, general heat conduction can be described by [6]
@t
q ˆ ÿk ˆ ÿkrT …2†
@n
where q is the generated heat ¯ux and r is the three-
dimensional operand. However, in a two-dimensional
situation where kx and ky are the thermal conductivities
in the x and y directions respectively, as is the case in
question with the ceramic tile grout seal, the governing
equation of the conductive heat transfer can be expressed
by the rearranged Poisson equation [6, 17]:
   
@ @T @ @T 1 @T
kx ‡ ky ‡Qˆ …3†
@x @x @y @y  @n

However, the elements comprising the ceramic tile grout

seal (the AOCG, the vitreous enamel, the ceramic tiles
and the OPC substrate) can be considered individually
to be isotropic, within which no heat, Q, is generated.
As such, the conductive heat transfer is given by the
Laplace equation [6, 17]:
   
@ @T @ @T @T
kx ‡ ky ˆ …4†
@x @x @y @y @t

Consequently, the two-dimensional stationary tempera-

ture distribution can be derived by obtaining the solution
Fig. 1 (a) Schematic diagram of the HPDL surface treatment to Laplace's equation (r2 T ˆ 0, when kx and ky are
of the AOCG during the ®rst stage of the two-stage constant) with appropriate boundary conditions.
ceramic tile grout sealing process and (b) laser beam Consider Fig. 2, which shows a schematic cross-
pro®le of the HPDL section view of the two-stage ceramic tile grout seal prior
B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B
454 J LAWRENCE AND L LI

increasing temperature. In particular, the thermal

properties of the enamel frit are assumed to be
constant through the molten to the solid state.
4. For ease of analysis, the ceramic tiles in the lateral
direction are considered to be semi-in®nite (as veri®ed
by thermocouple readings).
5. For ease of analysis, the OPC substrate in the lateral
and vertical directions is considered to be semi-in®-
nite (as veri®ed by thermocouple readings).
Fig. 2 Schematic pro®le view of the ®rst stage of the two-stage
ceramic tile grout seal showing the thermal boundaries,
i (where i ˆ 1; 2; . . . ; 8), within the seal 4 FE MODEL CALCULATED RESULTS

to the ®rst stage of the sealing process, and indicates the 4.1 FE model calculated temperatures
thermal boundaries, ti (where i ˆ 1; 2; . . . ; 8), within the
By ascertaining the temperatures for the regions around
seal. Thus, the boundary conditions for the ®rst stage
the thermocouples from the nodal values calculated by
FE model are given by
ABAQUS at predetermined times, it was possible to pre-
dict the temperature through the central components
x; y 2 t1 ; t2 ; T ˆ To
(the AOCG and OPC substrate) of the two-stage ceramic
x; y 2 t3 ; T ˆ Ti tile grout seal. This procedure was carried out for both
the ®rst and second stages of the two-stage ceramic tile
@T
x; y 2 t4 ; t5 ; ÿkAOCG ˆ q1 grout sealing process. Figure 3 shows how the tempera-
@n
ture through the centre of the two-stage ceramic tile
@T grout seal (through the AOCG and OPC substrate) in
x; y 2 t6 ; t7 ; ÿktile ˆ q2
@n the ®rst stage of the process alters with time for the
@T varying depths of the thermocouples at a ®xed power
x; y 2 t8 ; ÿ kAOCG ˆ q3 density of 1.5 kW/cm2 . From Fig. 3 it can be seen that
@n
the temperature typically increases with time and
in which T is the transient surface temperature, To is the decreases with depth.
initial (original) surface temperature of the AOCG, Ti is
the temperature pro®le of the HPDL beam incident on
the AOCG and q1 , q2 , q3 are the various resultant heat 4.2 FE model calculated thermal pro®les
¯uxes. Additionally, so as to accurately model the heat
Figures 4, 5 and 6 show the variation in temperature
transfer within the ®rst stage of the sealing process result-
predicted within the two-stage ceramic tile grout seal
ing from HPDL beam interaction, it was essential to
during HPDL processing by the FE model in terms of
determine the exact nature of the heat source, i.e. the
depth and time, depth and speed and depth and power
laser beam characteristics, in particular the beam pro®le
density respectively, during the ®rst stage of the ceramic
shown in Fig. 1b, thus making it possible to introduce
tile grout sealing process. The thermal pro®les predicted
into the model the exact shape of the heat source. Such
are through the centre of the seal. The model shows the
beam characteristics were obtained using a Promotech
predicted thermal pro®le on the surface in one direction
UFF-100 laser beam analyser (LBA) [14].
only, since the two-stage ceramic tile grout seal is symme-
In order to develop the FE models of the ®rst and
trical about the centre, and therefore the pro®le will be
second stages of the sealing processes in the simplest
mirrored about the centre. In order for the model to
and most accurate way the following important assump-
represent realistically the two-stage ceramic tile grout
tions were made:
sealing process the power density used in the simulations
1. The latent heat e€ects resulting from chemical shown in Figs 4 and 5 was 1.25 kW/cm2 . This is within
reactions in the AOCG and the surface melting the optimum process parameters for the ®rst stage of
caused by HPDL interaction are ignored. the process, being towards the lower end of the power
2. Because of the comparatively small meltpool in density scale. However, in the simulation shown in
relation to the entire seal, the e€ects of convection Fig. 6 the interaction time used was 2.5 s, which converts
heating within the meltpool are assumed to be to an equivalent traverse speed of 240 mm/min for a
negligible. beam spot diameter of 1.6 mm and lies well within the
3. The thermal properties of the elements comprising the optimum process parameters for the ®rst stage of the
two-stage ceramic tile grout seal (the AOCG, the process. Only the depths of the thermal pro®les in the
vitreous enamel, the ceramic tiles and the OPC AOCG are considered, because the AOCG section
substrate) are assumed to remain constant with in situ within the two-stage ceramic tile grout seal is
Proc Instn Mech Engrs Vol 214 Part B B02799 # IMechE 2000
 FE ANALYSIS OF TEMPERATURE DISTRIBUTION USING ABAQUS FOR A LASER-BASED PROCESS 455

1000

800 1.25 mm depth (AOCG)

2.50 mm depth (AOCG)
3.75 mm depth (AOCG)

Temperature (°C)
600 5.00 mm depth (AOCG)
6.50 mm depth (OPC)

400

200

0
0 1 2 3 4 5 6
Time (s)

Fig. 3 FE model calculated relationship between temperature and time at various depths within the two-stage
ceramic tile grout seal in the ®rst stage of the process

relatively narrow; thus lateral changes in temperature towards the bottom end of the power density scale, the
may be deemed to be negligible. whole section of the AOCG experiences a temperature
As discussed previously, the FE model does not in excess of 100 8C in an interaction time of less than
account for any of the reactions occurring in the 0.2 s, therefore indicating that the complete irreversibility
AOCG during HPDL interaction, or any optical of the water-glass reaction throughout the whole section
phenomena intrinsic within the sealing process. During of the AOCG, i.e. the rehydration of the water-glass, is
the ®rst stage of the process the predicted temperatures achieved when it is exposed to water. It was found that
are slightly higher than the measured temperatures. the highest temperature experienced by the AOCG,
Notwithstanding this, Figs 4, 5 and 6 show how the ther- which occurs during the ®rst stage of the process, is
mal pro®le varies with time through the centre of the only around 200 8C, even at the maximum interaction
AOCG and the OPC substrate. Indeed, from Figs 4, 5 time.
and 6 the change in the thermal pro®le between the
AOCG and the OPC substrate at around 5 mm can be 4.3 FE model calculated process operating maps
seen.
More importantly perhaps, Figs 4, 5 and 6 show that From a manipulation of the thermal pro®le data ®les
when using a power density of 1.25 kW/cm2 , which lies resulting from the FE model and simultaneous cross-

Fig. 4 FE model predicted thermal pro®le in relation to depth and time during the ®rst stage of the ceramic
tile grout sealing process (1.25 kW/cm2 power density)

B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B
456 J LAWRENCE AND L LI

Fig. 5 FE model predicted thermal pro®le in relation to depth and traverse speed during the ®rst stage of the
ceramic tile grout sealing process (1.25 kW/cm2 power density)

referencing with the thermogravimetric-di€erential earlier, during the second stage of the sealing process
thermal analysis (TG-DTA) results shown in Fig. 7, it the maximum temperatures reached are consistently
was possible to generate detailed operating maps indi- lower than those reached during the ®rst stage. Thus
cating the depth to which HPDL-induced chemical no further thermally induced chemical reactions will
reactions occur within the AOCG in terms of power occur; therefore operating maps for the second stage of
density and traverse speed for the ®rst stage of the two- the sealing process are unnecessary. As before, only the
stage ceramic tile grout sealing process. The generated depth of the thermally induced changes in the AOCG
operating maps predict the depth of dehydration, the are considered because the AOCG section in situ within
depth of Mg3 (OH)2 Si4 O10 decomposition, the depth of the two-stage ceramic tile grout is relatively narrow;
vitri®cation, the Zn melt depth and the depth of the thus lateral changes in temperature are deemed to be
Fe2 O3 a±b solid state phase transformation/crystalliza- negligible. Figure 8 shows the operating map for the
tion of Na2 O
2SiO2 . However, as has been discussed predicted depth of vitri®cation in the AOCG.

Fig. 6 FE model predicted thermal pro®le in relation to depth and power density during the ®rst stage of the
ceramic tile grout sealing process (240 mm/min traverse speed)

Proc Instn Mech Engrs Vol 214 Part B B02799 # IMechE 2000
 FE ANALYSIS OF TEMPERATURE DISTRIBUTION USING ABAQUS FOR A LASER-BASED PROCESS 457

–200.0 102.0
I II III IV V

100.0
–300.0 267 °C

B 349 °C
417 °C
98.0
A 366 °C 662 °C
–400.0 C
E
D
Heat flow (mW)

Weight change (%)

96.0
Heat flow (mW)

–500.0

94.0

–600.0
92.0

Weight change (%)

–700.0
90.0

–800.0 88.0
0.0 200.0 400.0 600.0 800.0 1000.0 1200.0
Temperature (Celsius)

Fig. 7 TG-DTA traces for the AOCG

Fig. 8 Operating map for the ®rst stage of the two-stage ceramic tile grout sealing process depicting the
predicted depth of vitri®cation in the AOCG

B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B
458 J LAWRENCE AND L LI

Fig. 9 Schematic of the set-up for the temperature monitoring during laser processing of the AOCG

5 COMPARISON BETWEEN FE MODEL the AOCG seal using a 0.45 mm diameter diamond-
RESULTS AND EXPERIMENTAL RESULTS tipped drill so as to achieve as ®ne a hole as possible.
Water cooling of the drill was employed in order that
Standard vitri®ed ceramic tiles were cut into smaller the nature of the OPC substrate, the ceramic tiles or
pieces, 20 mm  20 mm, for experimental purposes, and the AOCG was not a€ected by the heat generated
applied in pairs to an OPC substrate using standard during the drilling process. The holes were then ®lled
epoxy tile grout. The spacing between the vitri®ed with thermal conducting paste prior to the insertion of
edges of each tile pair was 1.5 mm. The ®xed ceramic the thermocouples so as to provide good thermal contact
tile pieces were then allowed to set for the standard between the materials and the thermocouples. To obtain
setting time of 12 h. The tiles were then sealed by apply- accurate temperature readings, the thermocouple pairs in
ing the AOCG into the void between adjoining ceramic each sample were attached to digital thermometers, with
tiles, ¯ush to the surface of the tiles. The AOCG was the thermometer readings being recorded on video along
then allowed to cure for 8 h. with the interaction times. The samples were irradiated
To gauge the thermal pro®le generated in the ®rst for 6 s using a defocused HPDL beam with a power
stage of the ceramic tile grout sealing process, thermo- density of 1.5 kW/cm2 (for a beam spot diameter of
couples were then inserted to either one of four predeter- 1.6 mm). The beam was aligned such that it was incident
mined depths and two predetermined distances from the axially with the centre-line through the ceramic tile grout
centre of the ceramic tile grout seal in the samples in the seal samples.
manner shown in Fig. 9. The thermocouples used were Figure 10 shows the results of the experimentally
type-K class 2, with an accuracy of 2.5 8C and a determined temperature distribution in the ®rst stage of
response time of 0.015 s, while the thermometer used the ceramic tile grout sealing process, in comparison
had a measurement rate of 4 readings/second, with a with the FE model predicted temperature distribution
resolution of 0.1 8C below 1200 8C with an accuracy of for one set of parameters. The results depicted in
2 8C at ÿ50±1200 8C. The holes for the thermocouples Fig. 10 are only for a single point in the centre of the
were drilled into the OPC substrate, the ceramic tiles and AOCG at a depth of 1.25 mm subjected to a ®xed

1000

800
Temperature (°C)

600

400
Stage 1 FEM Predicted
Stage 1 Experimental
200 Stage 2 FEM Predicted
Stage 2 Experimental

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Time (s)

Fig. 10 Comparison between the FE model predicted and experimental temperatures for a point in the centre
of the AOCG 1.25 mm from the surface of the seal

Proc Instn Mech Engrs Vol 214 Part B B02799 # IMechE 2000
 FE ANALYSIS OF TEMPERATURE DISTRIBUTION USING ABAQUS FOR A LASER-BASED PROCESS 459

Table 1 Comparison between measured and predicted depths within the components of the two-stage ceramic tile
of vitri®cation on the AOCG for certain laser power grout seal, in particular the AOCG, during the heating
densities and traverse speeds cycle to the steady state condition. However, since the
Depth of vitri®cation (mm)
increase in temperature resulting from HPDL interaction
results in the occurrence of certain chemical reactions
Laser parameters Measured Predicted within the AOCG and the surface melting, latent heat
2
1 kW/cm at 480 mm/min 370 420
e€ects due to the energies of these reactions (endothermic
1.25 kW/cm2 at 360 mm/min 1025 1100 reactions) will be apparent.
1.5 kW/cm2 at 240 mm/min 520 585 If Fig. 5 is considered, then it can be seen that for the
®rst stage of the sealing process agreement between the
temperatures predicted by the FE model and the experi-
mentally obtained temperatures hold consistently for
power density of 1.5 kW/cm2 . Figure 10 shows that for
approximately the ®rst 1±1.5 s of the heating cycle to
the ®rst stage of the two-stage ceramic tile grout sealing
the steady state condition. Beyond this point in time,
process, the temperature distribution predicted by the
when the error between the predicted and experimental
FE model was consistently higher than those obtained
sets of temperature data ceases to be consistent, the
experimentally. Nevertheless, the temperatures predicted
error was seen to increase. From the experimental
by the FE model and those obtained experimentally were,
curves it can be seen that the temperature at which this
in general, in very close agreement with one another.
consistent agreement ceases is around 250 8C. As the
Of the ®ve operating maps generated for the AOCG
TG-DTA showed, within this temperature range consid-
[depth of dehydration, depth of Mg3 (OH)2 Si4 O10
erable dehydration of the AOCG occurred. Moreover,
decomposition, depth of vitri®cation, Zn melt depth
these dehydration reactions were identi®ed as being
and the depth of Fe2 O3 a±b solid state phase transfor-
endothermic [14]. The dehydration reactions remove
mation/crystallization of Na2 O
2SiO2 ], only the depth
heat which is not accounted for in the FE model. Conse-
of vitri®cation has been considered in this study since it
quently, the temperatures predicted by the FE model are
is the most readily veri®able for the vitri®cation depths
higher than those observed during the experiments since
observed on actual HPDL-treated AOCG samples.
the model does not account for the e€ective heat sink
Table 1 shows, as was the case with the temperature pro-
arising from the endothermic reactions.
®les, that the predicted depths of vitri®cation observed in
However, the TG-DTA revealed that the endothermic
Fig. 8 were consistently higher than those measured [14].
reactions in the AOCG occurred from around 80 8C
Notwithstanding this, such a comparison revealed that
onwards. Accordingly, the e€ects of the heat sink
the predicted depths of vitri®cation were in very close
would therefore be expected to be seen from 80 8C
agreement with those observed on AOCG samples
onwards, whereas in Fig. 10 these e€ects become appar-
treated with the HPDL.
ent from 250 8C onwards. It is suggested that since the
di€erence between the actual onset of the endothermic
6 DISCUSSION reactions (80 8C) and those experimentally observed
(250 8C) is relatively small, the rapid heating rate
achieved through HPDL interaction results in the
The fact that there are discrepancies between the
`telescoping' of the reactions, causing the reaction tem-
temperatures predicted by the FE model and the tem-
peratures to increase [18]. Furthermore, the TG-DTA
peratures recorded during experimentation indicates
indicated that the other reactions that occurred in the
that some degree of inaccuracy is inherent within the
AOCG as the temperature increased were also endother-
FE model. This is due perhaps to the theoretical idealiza-
mic, and will therefore compound the error between the
tions and the assumptions made, as well as the many
temperatures predicted by the FE model and those
unknown aspects involved, such as the thermocouple
obtained experimentally as the temperature increases.
readings, etc. Also, since none of the laser parameters
Indeed, this is the case as depicted in Fig. 10, where the
alter during the experiments it is highly probable that
complete range of temperatures and corresponding
certain external factors in¯uence the degree of agreement
endothermic reactions in the AOCG are experienced in
between the results. These external factors can be identi-
the ®rst stage of the sealing process.
®ed as the endothermic reactions, changes in the thermal
properties of the two-stage ceramic tile grout seal com-
ponents and experimental errors during the temperature
monitoring. 6.2 Changes in the thermal properties of the
amalgamated oxide compound grout

6.1 Endothermic reactions Prominent within the assumptions of the FE model is the
fact that the thermal properties of the components
The FE model assumes there to be no latent heat e€ects comprising the two-stage ceramic tile grout seal remain
B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B
460 J LAWRENCE AND L LI

constant, in particular the thermal conductivity, k. How- not be drilled perpendicular to the surface, then an
ever, the vitri®cation of the AOCG surface during the error of some millimetres in the depth and width mea-
®rst stage of the sealing process at around 366 8C, surements could result, since the thermocouple is only
which is an endothermic reaction, causes a change in 0.3 mm in diameter and therefore signi®cant at depths
the thermal conductivity of the a€ected area. However, of only a few millimetres.
the change is small, altering from 1.49 W/mK before The precise characterization of the HPDL beam spot
laser treatment to 1.53 W/mK after laser treatment. incident on the AOCG surface is essential in order to
Consequently, with such a small change in the thermal give an accurate model of the heating process. Inexact
conductivity of the AOCG, it is reasonable to assume measurements of the beam pro®le, and thus the power
that the e€ect of this change on the FE model will be intensity distribution of the beam, would lead to the
negligible. Nevertheless, when the e€ects of this FE model consistently predicting an inaccurate thermal
unaccounted-for change in thermal conductivity are pro®le. However, since the exact beam pro®le of the
combined with those resulting from the same endo- laser beam, as measured by the Promotech UFF-100
thermic reaction, which again is unaccounted for, the LBA and shown in Fig. 1b [14], is used in the FE
resulting error may not be insigni®cant. model, then it can be assumed that little or no experi-
Additionally, it is assumed that the components mental error can be apportioned to the measurement of
comprising the two-stage ceramic tile grout seal are iso- the HPDL beam characterization.
tropic, and no direction is more favourable for the con-
duction of heat than any other. This is indeed the case
for the AOCG. The same situation, however, does not 7 CONCLUSION
exist with regard to the OPC substrate. This is because
OPC is a non-uniform composite material and its Due to the complexities of modelling the thermal
thermal properties will therefore vary since the thermal behaviour of the two-stage ceramic tile grout sealing
properties of the cement are di€erent from those of the process, a numerical technique was employed which
aggregate pieces [19]. Also, the aggregate pieces will utilizes the ABAQUS ®nite element (FE) package. The
undergo di€erent reactions from those of the cement model created generated thermal contours, as opposed
matrix. The value for the OPC thermal conductivity to isochromes, and so allowed laser parameters such as
used in the FE model represent an average for the bulk traverse speed to be considered in the modelling, thereby
concrete. In contrast, however, the measurements of facilitating a more realistic analysis. Furthermore,
the thermocouples depend entirely on their speci®c manipulation of the thermal pro®le data ®les predicted
positioning. Consequently, a thermocouple positioned by the FE model resulted in the generation of operating
within an aggregate piece will measure the temperature maps which showed the extent to which the temperature-
of the aggregate rather than the bulk OPC, thus induced chemical changes in the AOCG, which were
a€ecting the average of the readings and generating a identi®ed in the TG-DTA, occur. Both the FE
discrepancy. predicted temperature pro®les and the operating maps
were seen to be in close agreement with experimentally
obtained measurements, arming the accuracy of the
6.3 Experimental errors during temperature FE model.
monitoring Consequently, generation of the model can facilitate
the establishment of the ideal laser operating param-
It is highly probable that experimental errors account for eters necessary to ensure that both adequate surface
a signi®cant proportion of the discrepancies observed melting of the AOCG [20] and a suciently high tem-
between the temperatures predicted by the FE model perature (>100 8C) at the base of the AOCG is achieved,
and the results obtained experimentally. In particular, ensuring the complete irreversibility of the water-glass
two areas of the experimental preparation and pro- reaction.
cedures can be identi®ed as likely sources of experimental
error. These are the accuracy of the measurement of the
depth of the thermocouples and the determination of the ACKNOWLEDGEMENTS
HPDL beam characteristics.
The inaccurate measurement of the depth of the The authors would like to express their gratitude to the
thermocouple can introduce signi®cant errors into the EPSRC: Process Engineering Group (Grant GR/
experimental results. The positions of the holes drilled K99770) and the EPSRC: CDP Group (CASE Award
into the two-stage ceramic tile grout seal samples were 95562556) for their ®nancial support. Special thanks
measured relative to the surface of the samples, with also go to Dr Osman Beg of Ove Arup and Partners
the assumption being that the actual position of the and Dr Shuguang Li of UMIST for their expert advice
thermocouple below the sample surface is the same. on the subject of thermal modelling and the use of
However, should the sample not be square or the hole ABAQUS.
Proc Instn Mech Engrs Vol 214 Part B B02799 # IMechE 2000
 FE ANALYSIS OF TEMPERATURE DISTRIBUTION USING ABAQUS FOR A LASER-BASED PROCESS 461

REFERENCES 11 Odumodu, K. U. and Das, S. Forceless forming with a laser.

Proc. ASME: Mater. Div., 1996, 74, 169±170.
12 Hsiao, S., Shimizu, H., Firth, L., Maher, W. and Masubuchi,
1 Taylor, G. D. Construction Materials, 1991 (Longman
K. Finite element modelling of laser forming. LIA Proc.,
Scienti®c and Technical, London).
1997, 77, 31±40.
2 Lawrence, J., Li, L. and Spencer, J. T. A two-stage ceramic
13 Mazumder, J. Overview of melt dynamics in laser proces-
tile grout sealing process using a high power diode laser.
sing. Opt. Engng, 1991, 30, 1208±1219.
Part I: grout development and materials characteristics.
14 Lawrence, J. Characteristics of a two-stage ceramic tile
Optics and Laser Technol., 1998, 30, 205±214.
grout sealing process using a high power diode laser. PhD
3 Lawrence, J., Li, L. and Spencer, J. T. A two-stage ceramic
thesis, UMIST, 1999.
tile grout sealing process using a high power diode laser.
15 DeGarmo, E. P., Black, J. T. and Kohser, R. A. Materials
Part II: mechanical, chemical and physical properties.
and Processes in Manufacturing, 1997 (Prentice-Hall,
Optics and Laser Technol., 1998, 30, 215±223.
Upper Saddle River).
4 Jeager, J. C. Moving sources of heat and temperature at
16 Petzold, A. and Rohrs, M. Concrete for High Temperatures,
sliding contacts. Proc. R. Soc. NSW, 1942, 76, 203±224.
1970 (MacLaren and Sons, London).
5 Rosenthal, D. The theory of moving sources of heat and its
17 OÈzisik, M. N. Heat Transfer: A Basic Approach, 1985
application to metal treatments. Trans. ASME, 1946, 68,
(McGraw-Hill, New York).
849±866.
18 Warne, S. Introduction to thermal analysis. In Thermal
6 Carslaw, H. S. and Jeager, J. C. Conduction of Heat in
Analysis: Techniques and Applications (Eds E. L. Charsley
Solids, 1959 (Oxford University Press, London).
and S. B. Warrington), 1992, pp. 1±16 (Royal Society of
7 Patankar, S. V. Numerical Heat Transfer and Fluid Flow,
Chemistry, London).
1980 (Hemisphere Publishing, New York).
19 Gani, M. J. Cement and Concrete, 1997 (Chapman and
8 Yagla, J. J., Haag, R. S. and Scott, M. E. Laser welding
Hall, London).
analysis and experiments. J. Ship Prod., 1995, 11, 102±110.
20 Lawrence, J., Li, L. and Spencer, J. T. High power diode
9 Park, J. S. and Na, S. J. Heat transfer in a stud-to-plate
laser modi®cation of the wettability characteristics of an
laser braze considering ®ller metal movement. Weld. J.,
Al2 O3 /SiO2 based oxide compound for improved enamel-
1988, 77, 155±163.
ling. J. Mater. Sci. Engng A, 1999, 266, 1067±1074.
10 Onut, S. Finite element analysis of laser forming. MSc
thesis, Strathclyde University, 1993.

B02799 # IMechE 2000 Proc Instn Mech Engrs Vol 214 Part B