Conference PaperPDF Available

MATHEMATICAL MODELLING OF HEAT TRANSFER AND THERMAL EXPANSION: AIDING THE DEVELOPMENT OF HEATING STRATEGIES FOR SILICA-LINED FURNACES

Authors:

Abstract and Figures

The heating of silica-lined furnaces in the frame of forced convection conditions combines fundamental and complex physical and chemical processes. In this work, a numerical investigation has been performed using CFD calculations, in order to gain more insight in involved processes. This work has been carried out using the capabilities of Glass Service GFM (Glass Furnace Model) software, which provides a framework for 3D-CFD simulations, including the solver COMBUSTOR for solving chemical kinetics. As a general methodology, this study is based on simulating combustion fumes injected into silica-lined furnaces, corresponding to the conditions of real heating operations carried out later by Thermojet. Specifically, this research checked the effect of quantity and positioning of burners on thermal homogeneity during transient heating. In addition, the 1D software ThermoSil 1.0 has been applied on the development of an optimised heat-up curve, designed to minimise thermal stresses. The results of this investigation led to the reduction of the number of allocated burners and supported the definition of positioning, including their inclination. Calculations results have been compared to field data, and good agreement has been observed.
Content may be subject to copyright.
MATHEMATICAL MODELLING OF HEAT TRANSFER AND THERMAL
EXPANSION: AIDING THE DEVELOPMENT OF HEATING STRATEGIES FOR
SILICA-LINED FURNACES
Costa, I. S.; Vidal A., B.; Passos, R. L.: Thermojet do Brasil Ltda., Brazil
Budik, P.: Glass Service Inc., Czech Republic
ABSTRACT
The heating of silica-lined furnaces in the frame of forced convection conditions combines
fundamental and complex physical and chemical processes. In this work, a numerical investigation
has been performed using CFD calculations, in order to gain more insight in involved processes. This
work has been carried out using the capabilities of Glass Service GFM (Glass Furnace Model)
software, which provides a framework for 3D-CFD simulations, including the solver COMBUSTOR for
solving chemical kinetics. As a general methodology, this study is based on simulating combustion
fumes injected into silica-lined furnaces, corresponding to the conditions of real heating operations
carried out later by Thermojet. Specifically, this research checked the effect of quantity and positioning
of burners on thermal homogeneity during transient heating. In addition, the 1D software ThermoSil
1.0 has been applied on the development of an optimised heat-up curve, designed to minimise
thermal stresses. The results of this investigation led to the reduction of the number of allocated
burners and supported the definition of positioning, including their inclination. Calculations results have
been compared to field data, and good agreement has been observed.
INTRODUCTION
Heating of industrial furnaces
Industrial furnaces and ovens which are lined with refractory material and operate at high
temperature usually need, prior to their operational start-up, a preliminary step of pre-
heating. This requirement may arise from limitations of conventional process burners, which
are mainly designed to operate under a narrow range of conditions. As a consequence,
transient heating performed by such burners may cause inefficiency and heterogeneity in the
furnace environment, leading to hot spots and high gradients of temperature inside the lining,
with subsequent warping. Another reason for pre-heating is the need to reduce or virtually
eliminate the water present on or within the lining, in order to avoid sudden increase in
pressure inducing cracks or even explosion (Peret et al, 2009).
Due to the given reasons, the heating of industrial furnaces has significant effects upon the
refractory lifetime, which may be comparatively extended when the heating strategy fulfills a
sort of requirements (Peret et al, 2006). These conditions are particularly stringent when the
equipment is lined in silica, as detailed in the following paragraphs.
Challenges on heating silica-lined furnaces
Silica undergoes a series of
polymorphic phase transformations,
accompanied by abrupt changes on
the volume of the material (Beerkens,
2005), as presented in Figure 1 (Peret
et al, 2006). Accordingly, strict control
of the heating conditions is imperative
on preserving the structure of silica-
lined equipments, such as glass
furnaces or coke ovens.
Figure 1: Expansion curve of silica-based
refractory materials
The thermal expansion is a material property indicative of the extent to which a material
expands upon heating and it is related to stresses induced in a body as a result of changes
are exposed to elevated temperature fluctuations or as a result of the difference in heating
rate and thermal expansion between the surface and interior regions. These thermal stresses
are important in brittle ceramics, since they may weaken the material or, in extreme cases,
lead to fracture, which is termed thermal shock. Normally, attempts are made to avoid
thermal stresses, what may be accomplished by heating the piece at a sufficiently controlled
and slow rate (Callister, 2007).
Description of the model equipments studied
Two different concepts of silica-lined equipments were studied: (i) conventional coke ovens
and (ii) heat recovery coke ovens. The corresponding heating operation comprised 80 and
72 coke ovens, respectively. In both cases, the temperature was continuously monitored by
over 300 thermocouples, generating extensive data for validation.
As the heating operation in both of the model equipments studied followed the convective
method, the flow pattern inside the coke ovens was critical for the final results. It impacted
the uniformity of temperatures throughout the internal environment, having direct
consequences on the thermal profile inside the refractory lining at each moment. Thus, the
flow patterns were investigated, considering the quantity and positioning of the high velocity
burners.
Details on the operation of each coke battery concept are presented by HRC (2011) and
Uhde (2009).
Model A: conventional coke ovens
Figure 2: Conventional coke oven battery
Model A consisted of evaluating the impact of changing
the quantity of burners on the features of the heating. The
original strategy would be to position burners at the ends
of each of the 80 coke ovens: one at the Pusher Side, the
other at the Coke Side (Figure 2). Nonetheless, given the
drive of machinery on the Pusher Side it would be
operationally preferable to have burners only on the Coke
Side.
Simulations were thus
performed, taking into account
the following four cases:
Case 1:
Two burners per oven,
positioned horizontally
Case 2:
One burner per oven, with
boosted air injection
Case 3:
One burner per oven, with
standard air injection
Case 4:
Two burners per oven,
positioned angled
Model B: heat recovery coke ovens
The model for heat recovery coke ovens
was developed with the main goal of
studying the positioning of the burner at
the door of the main chamber (Figure 3)
of each of the 72 coke ovens.
The heating strategy included one
burner per oven, only at the Coke Side.
following cases:
Case 1: burner positioned at the
centerline of the door.
Case 2: burner positioned at the right-down corner
of the door
Figure 3: Heat recovery coke oven
METHODOLOGY AND RESULTS FROM MODELLING AND SIMULATION
Model A: conventional coke ovens
3D Modelling and simulation
The results of heating-up procedures based on convective method and high velocity burners
are strongly influenced by the fluid dynamics taking place within the target equipment.
A CFD evaluation of the air flow was accomplished using Glass Service COMBUSTOR
at evaluating the effects of positioning only one burner per oven, compared to positioning two
burners per oven. Accordingly, the following cases were simulated:
Case 1: Two burners positioned horizontally at each oven
Case 2: One burner positioned per oven, with boosted air injection
Case 3: One burner positioned per oven, with standard air injection
Case 4: Two burners positioned per oven angled relative to the floor
The air flow field and temperature distribution were calculated as steady-state. For these
studies, such limitation is not a problem, since the heating rate is very small.
As a matter of simplification, focussing only on the core issue analysed here, results are
presented in Figure 4 for only case 1 and case 4:
(a) (b)
1100 1110 1120 1130 1140 1150 1160 1170 1180 1190 1200 1210 1220 1230 1240 1250 1260 1270 1280 1290
1300
Figure 4: Flow patterns from a side view: (a) Case 1 (two burners, positioned
horizontally); (b) Case 2 (one burner, with boosted air injection)
From the information illustrated by the pictures, it may be concluded that the heating
operation using only one burners is possible and, with boosted air injection, even higher
homogeneity may be reached, given the formation of an interface between two horizontally
positioned jets.
Model B: heat recovery coke ovens
1D Modelling and simulation
As mentioned in a previous chapter, the adoption of an appropriate heat-up curve is critical
for the preservation of the designed features of refractory-lined equipments. In particular, if
the constructive material is made of silica, a relevant variable is the progress of thermal
expansion as a function of temperature, as indicative of possible thermal stresses being
induced.
In order to minimize thermal stresses during the heating-up of silica-lined furnaces,
Thermojet developed a software which calculates transient temperatures inside the refractory
resulted from heat conduction: ThermoSil 1.0.
The heat-up schedule was defined based on the calculation of the thermal profile within the
silica bricks of the crown arch. This region was chosen as a reference because it is the silica
refractory that will present the greatest thermal gradients, since it is exposed to two
An optimized heat-up schedule is obtained from the following procedure (Peret et al, 2009):
An original heat-up schedule is presented as input.
The corresponding level of expansion is calculated for each point and each time
interval of the heat-up schedule, using data from experiments using silica bricks.
Then, an iterative method is started from ambient temperature to design a heat-up
curve so that all of the following criteria are met:
The average rate of expansion is kept constant or below an average for the
number of days with temperature increase;
The rate of temperature increase is below a maximum, predefined level in
order to prevent thermal shock;
The computation stops when the defined final temperature is reached.
If the total heat-up time resulting from calculation is above or below the user-defined
parameter, the set-point for thermal expansion rate is modified, and the entire heat-up
schedule is recalculated with the new value.
The heat-up schedule resulting from the unsteady-state simulation is presented on Figure 5.
Figure 5: Heat-up schedule for the coke oven batteries
The schedule presents a
continuous increase on the
heating rate, especially above
silica bricks cease to expand.
imposed to the program to
avoid the risk of temperature
heterogeneities.
Figure 6 presents the evolution of
the expected expansion rate during
the Dry-out and Heat-up phases.
According to the program logic, the
increase in temperature should be
stopped whenever the expansion
rate attained a value corresponding
to the expected average
(considering the total expansion
difference divided by the
corresponding time period).
The scattering presented on Figure 6 is due to the
resolution of temperature increase step defined by
the program.
Figure 6: Evolution of the expansion rate as a
function of time
3D Modelling and simulation
A CFD evaluation of the air flow was accomplished using Glass Service COMBUSTOR
adjacent ovens (to analyze the effect of heating a wall by two contiguous ovens). Specifically,
the study aimed at deciding the better position for the burner, out of two possible
configurations. Accordingly, the following cases were simulated:
Case 1: burner is positioned at the right-down corner of the oven door, inclined to the
upper-left corner;
Case 2: burner is positioned at centerline of the oven door, horizontally.
The air flow field and temperature distribution were calculated as steady-state. For these
studies, such limitation is not a problem, since the heating rate is very small.
(a)
.
.
.
.
.
.
. . . . . . . . .
.
.
(b)
.
.
.
.
.
.
. . . . . . . . .
.
.
380 390 4 00 410 420 4 30 440 4 50 460 470 4 80 490 50 0 510 520 53 0 5 40 550 560 570
down corner); (b) Case 2 (centered burner).
On Case 1, the flow at the side opposite to the burner is controlled partly by a secondary
vortex induced by the main flow. This new vortex exchanges particles and heat with the main
stream, and improves the thermal homogeneity throughout the entire oven. If the burner
were aligned with the oven longitudinal axis (Case 2), on the other hand, a symmetrical
vortex formation would occur, and the average gas speed would be lower close to the walls.
The images on Figure 8 represent the expected level expansion based on a direct post-
processing of the results on temperature, using a data table of expansion level as a function
along the silica bricks.
Considering the results for temperature and thermal expansion, the simulations pointed out
that thermal homogeneity during the heating-up should be better for the positioning
corresponding to Case 1.
(a)
0.65
0.63 0. 64
0.63
0.63
.
0.603851
.
0.651947
.
0.606827
.
0.577426
.
0.651318 .
0.65044
.
0.654286
.
0.654855
.
0.609684 .
0.637 81 .
0.604706
(b)
0.6 0. 61 0.6
0.63
.
0.583117
.
0.634509
.
0.586207
.
0.57553
.
0.636588 .
0.635893
.
0.634609
.
0.629322
.
0.591936 .
0.618 466 .
0.586419
Sil_expan s.dir
0.2 0.25 0.3 0 .35 0.4 0.45 0.5 0 .55 0.6 0 .65
0.7
Figure 8: Main chamber floor surface expansion level at average air temperature of
FIELD DATA
Model A: conventional coke ovens
In order to inspect the temperature profile along the length of a single oven, 32
Figure 9: Positioning of thermocouples throughout the battery
High homogeneity
could be verified, with
maximum tempera-
ture difference along
the longitudinal direc-
tion of each furnace
21, for instance
(Figure 10). This
result is in good
agreement with pre-
dictions presented in
Figure 4.
Model B: heat recovery coke ovens
A validation for the heat recovery coke ovens model is presented in Figure 11, which
of its interior mapped by six thermocouples installed through its crown arch. The picture
exhibits the absolute value of deviation between predicted and measured temperatures as a
function of position relative to the Coke Side, based on data presented in Figure 7. A
Figure 11: Deviation between simulation and field data as a function distance to the
CONCLUSIONS AND PERSPECTIVES
This work presents two examples of successful application of mathematical modelling on
predicting operating conditions of silica-lined furnaces. The validation of both cases reveals
this tool as an applicable approach with the following advantages, compared to design based
only on experiments (Veersteeg and Malalasekera, 2007): (i) Reduction of lead times and
costs of new designs; (ii) Ability to study systems where controlled experiments are difficult
or impossible to perform; (iii) Ability to study systems under hazardous conditions and
beyond their normal performance limits; (iv) Practically unlimited level of detail of results, as it
is relatively cheap to perform parametric studies; (v) Possibility to optimize equipment or
process performance.
Immediate perspectives for further development of the model are the following: (a) Inclusion
of effects of the transient changes; (b) Insight on the heat conduction through the silica
bricks.
REFERENCES
BEERKENS, R.G.C., VERHEIJEN, O.S.: Reactions of alkali vapours with silica based
refractory in glass furnaces, thermodynamics and mass transfer. Glass Technology -
European Journal of Glass Science and Technology Part A, Volume 46, Number 6,
December 2005 , pp. 371-382.
CALLISTER JR., W. D.: Materials Science and Engineering - An Introduction. John Wiley
& Sons, Inc., Seventh Edition, 2007.
HRC: Coke production technologies in use worldwide. Available at
http://www.hrc.lv/uhdeteh.pdf. Accessed on June 9, 2011.
PERET, C. M., DONKE, S. L. and PASSOS, R. L.: Heat-up schedules for silica
refractories for coke ovens.
PERET, C.M., PASSOS, R. L. e BUDIK, P.: Design of heat-up procedures for furnaces
using numerical simulations
process simulation, Glass Service Inc. and the Czech Glass Society, 2009.
Competence and know-how by tradition: a new dimension
in cokemaking technology. Printed in Germany, 2009. Available at http://www.uhde.eu/cgi-
bin/byteserver.pl/archive/upload/uhde_brochures_pdf_de_18.00.pdf. Accessed on June 9,
2011.
VEERSTEEG, H. K. and MALALASEKERA, W.: An introduction to computational fluid
dynamics: the finite volume method. Pearson Education. Second edition, 2007.
ACKNOWLEDGEMENTS
The authors are thankful to the Glass Service team, in special to Glenn Neff, Petr Chmelar
and Josef Chmelar for breaking the paradigms and allowing the use of GFM software outside
the boundaries of the Glass Industry.
ResearchGate has not been able to resolve any citations for this publication.
Heat-up schedules for silica refractories for coke ovens
  • C M Peret
  • S L Donke
  • R L Passos
PERET, C. M., DONKE, S. L. and PASSOS, R. L.: Heat-up schedules for silica refractories for coke ovens.
Design of heat-up procedures for furnaces using numerical simulations process simulation
  • C M Peret
  • R L Passos
  • P Budik
PERET, C.M., PASSOS, R. L. e BUDIK, P.: Design of heat-up procedures for furnaces using numerical simulations process simulation, Glass Service Inc. and the Czech Glass Society, 2009. Competence and know-how by tradition: a new dimension in cokemaking technology. Printed in Germany, 2009. Available at http://www.uhde.eu/cgibin/byteserver.pl/archive/upload/uhde_brochures_pdf_de_18.00.pdf. Accessed on June 9, 2011.
An introduction to computational fluid dynamics: the finite volume method. Pearson Education
  • H K Veersteeg
  • W Malalasekera
VEERSTEEG, H. K. and MALALASEKERA, W.: An introduction to computational fluid dynamics: the finite volume method. Pearson Education. Second edition, 2007.