See

discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/266403475

Furnace lining analysis and design by

mathematical and physicochemical
modeling
Conference Paper August 2006

READS

27
4 authors, including:
Roberto Parra
University of Concepcin
52 PUBLICATIONS 169 CITATIONS
SEE PROFILE

Miguel Angel Barbs

Independent Researcher
32 PUBLICATIONS 26 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

Available from: Christian Goni

Retrieved on: 15 August 2016

Sohn International Symposium ADVANCED PROCESSING OF METALS AND MATERIALS

VOLUME 8 - INTERNATIONAL SYMPOSIUM ON SULFIDE SMELTING 2006
Edited by F. Kongoli and R.G. Reddy TMS (The Minerals, Metals & Materials Society), 2006

FURNACE LINING ANALYSIS AND DESIGN BY MATHEMATICAL

AND PHYSICOCHEMICAL MODELING
Roberto Parra1, Luis Felipe Verdeja2, Mara Florentina Barbs2 & Christian Goi3
1
Department of Metallurgical Engineering, University of Concepcin.
Emundo Larenas 285, Concepcin Chile
2
Ctedra de Siderurgia, Escuela Tcnica Superior de Ingeniera de Minas,University of Oviedo.
C/ Independencia 13, 33004 Oviedo Spain.
2
Department of Metallurgical Engineering, Universidad Catlica del Norte
Av. Angamos 0610, Antofagasta Chile.
Keywords: Corrosion models, mathematical model,
Abstract
The wear and corrosion phenomena in the furnace lining of the ferrous and non ferrous
pyrometallurgical processes are complex problems where chemical, interfacial and mechanical
phenomena are presents. The characteristics of the processes determine the conditions to which
these materials are submitted, where the thermal conditions reached during the operation is one
of the most important parameter. Assuming that all the degradation phenomena are thermal
activated, the rate of the different steps of the wear mechanism will depend on the thermal field
in the lining.
In view to apply the independent physicochemical knowledge of degradation mechanisms to the
analysis and design of furnace lining the methodology of the Nodal Wear Model (NWM) was
done. It combines the mathematical modelling of the heat transfer in the lining with the
physicochemical control equation for the wear-corrosion phenomena. The NWM has been
applied to different examples: theoretical analysis, laboratory scale test and industrial
applications.
We present in this paper the theoretical background of the model and two examples of
applications: the sole of an electric furnace and the earth of a blast furnace.
Introduction
The pyrometallurgical primary production of metals and alloys is facing different challenges in
view of the optimization for their processes. The optimization looks for high quality products as
well as high quality processes in order to propose in a competitive way a specific product,
process or technology.
The technological aspects in which this optimization is focused are related with the
intensification of the process, with higher operating time and a higher production rate in every
unit (expressed in Tonuhr-1um-3 reactors volume or Tonuhr-1um-2 of installation area). For the
evaluation of any technological optimization it is necessary to analyze the economic results, the
environmental constraints and nowadays the conditions to achieve a sustainable process. From
the different variables to consider, the quality of the materials used in the lining of the different
furnaces and reactors in the ferrous and non-ferrous pyrometallurgical processes are relevant. For
the economic evaluation the quality of a specific material is, in this case, more important in
determining the operating time of a furnace than the reflected in direct cost. From the
561

environmental and sustainability point of view the materials used in the lining must be ideally
recycled at the end of a campaign, so the replacement of the lining would no produce residues. If
it is necessary to dispose the refractory material in a landfill their chemical stability and non risk
must be guaranteed.
The refractory materials used in furnaces, oven, and transport pots or storages equipments must
maintain their chemical, physics and mechanical properties at high temperatures to resist the
attack for the longest possible time. The global optimization of the process searches to extend
their working life stopping as much as possible the attack. In the general optimization
perspective another complementary point is to achieve the physical chemistry transformations
faster, working at a higher temperature and transform discontinuous operations in continuous.
Those two objectives are relevant in the optimization challenges previously defined. To
accomplish this optimization in most of the cases a technical limitation exists relating the quality
of the refractory materials and design of the lining.
Up to day the most classical option for the lining design of different equipments has left
overcoming through the repetition of trial and error tests to define the best material with the most
attractive cost/performance relation. In parallel laboratory test could be used to compare the
corrosion resistance in some specific conditions that suppose simulate industrial environment.
The Nodal Wear model presented in this paper is a tool to analyze the attack of the lining by the
molten phases and its application can help in the choice of the type of refractory material and the
design of the lining. It combines the knowledge of the basic mechanism of the wear-corrosion
(physical chemistry and thermomechanics) with the thermal modeling of the heat transfer using
the Finite Elements Method (FEM) to establish the evolution of the wear profile.
The experimental analysis of corrosion/wear phenomena in ceramic materials and the
Nodal Wear Model.
The most common materials used in the construction of the lining of pyrometallurgial reactors
and furnaces are metallic oxides, carbides and nitrides. Different studies have been done in view
of the characterization of the wear-corrosion mechanism that take place when those materials
contact with corrosives fluids. The first systematic studies were done by Noyes & Nernst, who
suggested, by the end of the 19th century and early 20th century, that the dissolution rate was
controlled by a layer of saturated solution that forms instantly around a solid particle. In 1904
Nernst and Brunner modified the Noyes-Whitney equation by applying Fick's law of diffusion
and a relationship between the dissolution rate and the diffusion coefficient was established. The
formalism and mechanism of those studies are the basis of the work done by Cooper &
Kingery[1,2,3] (1965), Hrma[4] (1970) and R. Cook[5] (1990) who have tried to expressed
quantitatively the wear-corrosion by mean of phenomenological and/or semi empiric equations.
The proposed equations can not been applied directly to other experimental configurations than
those used for their derivations. The lining of a furnace is in this sense a very complex case
where the direct application of those equations does not represent the real wear and corrosion
phenomena that take place in it.
In the industrial practice an important support for the design and evaluation of refractory
materials are the post-morten studies. The characterizations techniques (EDAX, Microprobes and
Cathodluminiscence) have been relevant for the development of those types of studies giving the
chemical characterization on the phases presented in the attacked zone that allows proposing
chemical reactions of corrosion. A useful complement that is available from the last decades is
thermochemical softwares for equilibrium calculation[6-8]).

562

Nevertheless all those possibilities, the post morten analysis could provide wrong conclusion
such the final corrosion mechanism could have no relation with the one that exists at the
beginning of the campaign. At the same time these analysis doesnt provide any information
about the rate of corrosion.
By the side, the dynamic and static laboratory tests are standards[9]. The simplest static test is the
analysis of the reaction when a sample of a molten phase is versed in the surface of a refractory.
The so called finger tests are two others types of test that could be done with different initial
reactive conditions for the liquid/solid system. One of them considers a molten phase filling a
hole made in a refractory and in the second, a piece of refractory is immerged in the liquid. In the
first case the liquid phase could be saturated by the dissolution of a component of the refractory,
in the second this possibility is avoid and a specific step of diffusion through a boundary layer
could be identified. For all these tests the analysis is done on the penetration of the refractory
before a specific time under experimental conditions that try to reproduce as close as possible
those where the material will be placed (temperature, chemical potentials and composition of the
melts). In the dynamic finger test the probe of the refractory immerged in the molten phase is
rotating to simulate the convection movement of the liquid that can accelerate the corrosion and
avoids any saturation of the possible dissolved species on the liquid. Another configuration for a
dynamic test is the case where the probe is static and a metallic liquid is stirred by induction
(ASTM C-622). In a third configuration (ASTM C-768) a known mass of molten phase is drop
by drop dropped in a refractory. Finally, a rotary furnace where different type of materials
conformed the lining and a liquid slag is heat by a burner provides a good option to compare the
qualities of the different materials tested and to obtain information on the corrosion process
(chemical dissolutions).
The conclusions of those types of test, based on the analysis of the altered microstructure of the
refractory, help to understand the corrosion/wear mechanism but the extrapolation to real
industrial conditions must be done carefully such the laboratory conditions are not the real
conditions of the process. Moreover, it is expected that in few hours the results could be
experimentally quantified by the depth of the zone attacked by the melts. In order to achieve this
objective the corrosion/wear mechanism has to be accelerated in the perspective to have a
measurable result in this time (2 to 4 hours) with a penetration about 10-3 m. To accelerate the
wear process some chemical activation is considered by means of the addition of clorines,
borines and specially fluorines into the melts. This modification finally changes completely the
characteristics of the melts from the real system and the rate of corrosion/wear is up to 50 to 100
times the real rate[10]. At the same time the thermal dynamic conditions in the real process are far
from the used in those test and also, even using an important amount of the melt regarding the
area of the refractory exposed the physical chemistry variables cant remain constant during the
test. Those tests are valid only for the comparison of the quality for different materials based on
their resistance under the experimental conditions.
The points that have been briefly described show that a comprehensive knowledge on the
physical chemistry of the corrosion/wear phenomena exists at laboratory scale. Nevertheless
there is no systematic approach for the analysis of the wear-corrosion of refractory in industrial
equipments.
The Nodal Wear Model take the basic phenomenological quantification of the individual
mechanism, each of ones is thermally activated, to establish a corrosion equation that express the
advance of the corrosion interface in units of (L/T) (Table 1). From the description of the
geometry using a Finite Element Method (FEM) grid the thermal field is obtained choosing the
adequate border conditions. The surface defined by the interface melts/refractory is modified
using the appropriate corrosion equation in the nodes of a second grid that represents this
563

interface. The types of equations to use in this step represent the control mechanism of the wearcorrosion.
Formalism of the model.
The resolution of the temperature field in the refractory lining of a furnace allows knowing the
temperature in the melt/refractory interface. To solve this mathematical problem in the steady
state, two border conditions have to be chosen. The temperature on the shell and the heat transfer
conditions between the melt and the refractory, represented by the convective heat transfer
coefficient, are the simplest in this case. The first can be measured and the second estimated.
With this data, the thermal conductivity of materials and using a FEM approach, the heat transfer
from the melt to the refractory lining can be determined and the temperatures in each node of the
FEM grid are known. The specific temperature in the interface melts/refractory is called nodal
temperature.
However, the corrosion phenomenon modifies the geometry of the system, therefore, the thermal
field before a 't is not longer valid since a new distribution of temperature has left consolidating
along the refractory lining of the furnace. The calculation of the new geometry is obtained
applying to each of the nodes in the interface melt/refractory a corrosion equation. This equation,
characteristic according to the control mechanism of the wear process is formally an empiric
equation based on the phenomenological analysis of the wear phenomena that allow defining the
control mechanism of this very complex process. The temperature dependence of the parameters
used for the specific corrosion equation considers the nodal temperature and the difference of
temperature between the node and their adjacent nodes.. With this formulation, the corrosion rate
in the node i expressed in units of length by unit of time is function of Ti and the difference of
temperature between this node and the adjacent ones 'Ti:
v( corrosion )i

L
f ( Ti ; 'Ti ) ;
t

(1)

The wear in each node is determined supposing that the temperature remains constant during the
interval of time 't:
(Wear )i

v( corrosion )i 't

; > L@

(2)

As the geometry of the system has been modified, the new temperature field should be solved to
obtain a new wear profile in the lining by means of an iterative process. If each iterative cycle
takes place after a certain 't, the calculation process for the i node in the nth iteration is:
Temperature in the i node by FEM: Ti ,n 1
Corrosion rate in the node i: v( corrosion )i

f ( Ti ,n 1 ; 'Ti ,n 1 )

Wear during the 't of the iteration n in the i node: (Wear )i v( corrosion )i 't
Definition of the new geometry and calculation of the new temperature field, Ti , n in the
whole refractory lining and so the temperature of the interface melt/refractory is defined.
It is important to note that the interface melts/refractory, where the attack occurs, defines the
surface S build by the nodes where the wear-corrosion equations are applied. This 2D grid is
independent of the FEM grid and the numeric resolution used for the calculation of temperature
field in the lining. The Figure 1 shows the geometrical difference between the surface where the
NWM is applied and the FEM grid for the tuyere zone of a Pierce Smith Converter.
564

New NWM Surface before a 't

Initial NWM Surface

FEM grid for the thermal profile evaluation

Figure 1: Detail of the tuyere zone of a Pierce Smith converter where the FEM grid and the
surface of the NWM nodes are shown. The red dots represent the nodes of the NWM surface[11].
The mathematical problem to define the new FEM grid is one of a moving surface where the
initial surface S(x,y) has been move due to the attack of the fluid. The differential movement is:

dS

wS
wS
dx 
dy
wx
wy

(3)

The rigorous formalism for equation ( 2 ) represents the evolution of S where the wear-corrosion
equation, v(corrosion)i, is applied:
'S wear thickness

&

&
S x v  dt

(4)

&

Where v(corrosion)i is represented by the vector function v . The discretisation of this equation
allows modifying the surface S by the calculation of the unidirectional 'S:
'S
'S
vx 
v y * 't
'S wear thickness #
x
'
'y

(5)

To apply this algorithm it is necessary to develop corrosion equations [v(corrosion)i] that allow
the geometry correction of the refractory lining for a defined 't in the calculation process. The
corrosion rate presented in equation ( 1 ) depends on the specific mechanism considered as
control step of the corrosion/wear of the lining. As it was said, the thermal activation of the
mechanism defines the dependence of this corrosion rate on the nodal temperature and the
temperature of the adjacent nodes: Ti and 'Ti.
Two basic mechanisms are presented in the following case studies: infiltration in the open
porosity and chemical dissolution. This chose was made due to the good understanding of the
physical chemistry process and the possibility to develop the respective corrosion equations
v(corrosion )i
565

Infiltration
Porous refractory materials have a very complex microstructure, which can be hardly described
as being straight, cylindrically shaped with constant-section channels. The channels are not
straight but in zig-zags, their section varies widely from channel to channel and along any given
channel, and their internal surface is not smooth but rough. However, the main features of both
the thermodynamics and the kinetics of pore infiltration can be easily understood, using the very
simple configuration of a unique, cylindrically-shaped, straight-line pore of radius r where the
equilibrium value of liquid rise in the capillary can be calculated by minimizing the variation in
total free energy 'F as a function of the rise, z. Setting d( 'F) / dz 0, the following equation is
obtained, giving the equilibrium rise ze[12]:

ze U g 

2V lv cos T
r

Po  Pv

(6)

Where Uis the liquid density, g the gravity, Vlv the surface energy between the liquid and the gas
phase, T the wetting angle between the liquid and the solid, Po is the pressure applied on the
liquid at the capillary entrance z = 0 and Pv the pressure of the vapor phase ahead of the
infiltration front.
If we assume that the flow of the infiltrated melt is incompressible and laminar with viscosity K
the velocity u can be expressed by equation ( 7 ):
u

dz
dt

1 r 2 dP
K 8 dz

1 r 2 'Pt
K 8 z

(7)

The total pressure drop driving infiltration is equal to the applied pressure Po-Pv minus the
capillary pressure 'Pc

2V lv cos T
, and minus the hydrostatic pressure Ugz. In most practical
r

situations, z << ze, so 'Pt is nearly constant during the infiltration process ('Pt | Po - Pv - 'Pc).
After integrating equation ( 7 ) once with respect to z and then with respect to time taking z = 0
at t = 0, it gives
1/ 2

r2
4K

'Pt1/ 2t1/ 2

(8)

Applying this equation, also known as Washburn's equation[13], to an industrial furnace where the
height of the liquid metal and slag represents the metallostatic pressure, it can be transformed in
a unidirectional equation (L/T) to apply in each of the nodes of the FEM surface S:
v ( corrosion ) i

2 ( U S g z S  U M g z M  Patm ) 2 2
r ;
J m / g cos T K

L
t

(9)

Assuming that the melts in the furnace are a metal and a slag the parameters of equation ( 9 ) are:

US the slag density, zS slag eight, UMmetal density, zm metal eight and Patm the atmospheric
pressure over the bath.
Chemical dissolution:
The chemical dissolution takes place when one of the constituents of the refractory can dissolved
in the melts. To obtain the nodal equation, the dissolution is considered as a process where a
566

constituent of the refractory diffuses into the melt, if any chemical reaction occurs in the
interface it is not taken into account due to the high rate of chemical reactions at high
temperature, and so no chemical control is expected. The basic equations that represent these
phenomena are Ficks laws of mass transfer. For a steady state condition the diffusional mass
flow of the attacked species J i , could be calculated by the well known equation:
Ji

ki 'Ci

( 10 )

Where k i is the mass transfer coefficient, 'C i the difference of concentrations of the specie that
diffuses between the interface and the bulk of the melt. This equation can be transformed to
represent this control mechanism in units of length by unit of time:
v(corrosion)i

U
ki 'Ci i
U
g

100 L

;
%cm t

( 11 )

In equation ( 11 ) the difference of concentration 'C i should be expressed in fraction; U i , is the

nodal density of the melt in contact with the refractory; U g , is the bulk density of the refractory
and %cm is the percentage of the constituent that dissolves. The value for the nodal mass transfer
coefficient from the refractory surface to the melt is obtained by the following equation:
ki

D
0.332 Re1/ 2 Sc1/ 3 i
L

( 12 )

Where Di , is the diffusion coefficient or the attacked specie; L, is the lineal characteristic
dimension; Re, the Reynolds number and Sc the Schmidt number, both of them chosen as the
most representative of the conditions in the interface refractory melts in the furnace.
The corrosion equation that results from substituting the expression ( 12 ) in ( 11 ) is:
v(corrosin)i

D
0.332 Re1/ 2 Sc1/ 3 i
L

Ui
U
g

100 L

;
%cm t

( 13 )

To be able to use this corrosion equation it is necessary to know the movement of the melt in the
interface with the refractory. The nodal velocity vi used to calculate the Reynolds number is
estimate from the specific conditions where the equation will be applied.
Table 1 resumes the equations used in the case studies.
Table 1: Corrosion equation applied in the Nodal Wear Model.
Mechanism
Equation
2 ( U S g z S  U M g z M  Patm ) 2 2 L
v ( corrosion ) i
r ; (9)
Infiltration
J m / g cos T K
t
Chemical Dissolution

v(corrosin)i

D
0.332 Re1/ 2 Sc1/ 3 i
L

Ui
U
g

100 L

; (
%cm t

Case Study I - The Sole of an Electric Arc Furnace

567

13 )

The electric arc furnace shown in Figure 2 is used for ferromanganese production. It has a
magnesite castable sole with high MgO content (% MgO>95%). For the initial conditions the
corrosion profile shown in the Figure 2 was obtained by a continuous campaign of measurements
of the sole. The average duration of the sole was only 19 tappings that represent a corrosion rate
of 0.52 cmhr-1. The control mechanism identified for the wear of the sole in this initial
condition was the penetration of the alloy through the open porosity. A FEM calculation of the
temperature field was done using as border conditions the outside wall temperature. The value of
18 Wm-2K-1 for the convective heat transfer coefficient between the ferroalloy and the
refractory lining adjust the results of the NWM to the experimental measurements of the wear in
the sole[14]. The parameters for equation ( 9 ) are listed below. The temperature dependence of
viscosity is presented in Table 2.
1. Time of one cycle for refining: 3 hrs ('t)
2. Surface tension of ferromanganese: J l / s 0.20 N u m 1
3.
4.
5.
6.

Density of ferromanganese: U ( Fe  Mn80%) 7125  0.893T ( K ) kg u m 3

Contact angle between ferroalloy and refractory: T 30
Thermal Conductivity: O 10 W u m 1 u K 1
The overall size of the open pores of the refractory is between 12 and 20 microns.

Table 2: Temperature dependence for viscosity of a ferromanganese with 80 % Mn and 1.40% C.

Temperature
Viscocity (Paus)
975
1012.5
1100
108
1125
107
1180
6u10-3

Wear profiles for the sole made by a castable

magnesite refractory (monolithic).
Experimental measurements.

Cr2O3-FeO-MgO conformed
Refractory. D

Silicious - Aluminous
Conformed Refractory. E
Magnesite castable
refractory A
8th tapping

19th tapping

3th tapping

5 W. m -1. K -1.

Security refractory: MgO cold

conformed and sintered. B

Aluminous Castable Refractory C

Figure 2: Lining characteristic of an electric arc furnace in ferromanganese production. Initial

corrosion profiles for a monolithic magnesite sole.

568

To extend the number of tapping many simulations were done changing the design of the sole by
means of other materials with a higher thermal conductivity O and changing the conditions of
external refrigeration.
i)
Region B changes from sintered magnesite (O = 5.0 Wm-1K-1), to magnesia-carbon
graphite refractory (O = 25 Wm-1K-1). The number of tapping can be increased up
to 25.
ii)
Region B, C y D, to magnesia-carbon graphite refractory (O = 25 Wm-1K-1). The
number of tapping can be increased up to 33.
iii)
Combining ii) with air forced convection: hc = 10 100 Wm-2K-1. The furnace can
achieve a campaign up to 90 tapping.
Conditions ii and iii were tested in the industrial furnace allowing to increase the number of
tapping at the level predicted.
Before the analysis the reason from which the best design is the one that has a high level of
thermal loses seems clear. The high conductivity proposed for the refractory in zones B,C and D
displace the isotherm corresponding to the solidification of the ferro-manganse alloy to a higher
position in the lining.

D
A
B
C

a: initial design
b: proposed design.
Figure 3: Initial (a) and final (b) configuration of the lining of the electric arc furnace. B,C and D
refractories were replaced by a magnesia-carbon with a higher thermal conductivity.
Case Study II - The hearth of a Blast Furnace.

In view of a deep understanding on the corrosion phenomena that occurs in the crucible of the
blast furnace for pig iron production the application of the NWM was done to quantify the
factors that define the shape of the wear profile in the crucible: central well, mushroom and
elephant foot. The simulation considers the analysis of different heat transfer conditions defined
by the refrigeration used as well as the dimensions and contact conditions between the bottom of
the crucible and the inactive coke (deadman). To represent theses different conditions on the
heat transfer conditions, different conditions of the convective heat transport hc were defined the
long of the crucible, in contact with the pig iron and with the external refrigeration conditions.
The details of these conditions (C-I, C-II and C-III) are shown in Table 3. The C-II and C-III
conditions simulate the case when the deadman is leading on the bottom of the crucible. In
both cases this situation produces a higher thermal resistance at the surface of the refractory in
contact with the deadman (lower hc values). C-II represent a conditions when the deadman is
only in the center of the crucible and in this case a high circulation of the pig iron exists near the
corners. The C-I case could be identified with an operation with a little dead man floating in
the pig iron and the profile is of the type of central well.

569

Table 3: Refrigeration characteristics used for the simulation of the corrosion process in the
hearth of a blast furnace for pig iron production.
Values for the distribution of hc Wum-2uK-1. The position is referred to the nodes 1 to 22 in Figure 4.
Condition
Bottom exterior
Wright exterior
Interior
Left exterior
nodes
1-2
2-4
4-5
5-6
6-8 8-9 9-11 11-12
12-14
14-16
16-18 18-19 19-22
22-1
C-I
600
90
600
170
320 600
85
650
320
170
*
*
85
650 2400 140/2400
C-II
2400 3000 2400 140/2400 2400 600 85
95
(12:140) /40 (14:140) /95
C-III
2400 3000 2400
2400
2400 600 85
95
40
95
85
650 2400
2400

*: The value of hc in the corners is 140 in a distance of 1 meter around the nodes 12 and 14. This
high value is due to the preferential circulation field that exists for the pig iron when the dead
man contact only in the center of the crucible.
The characteristics of the crucible used for the simulation are the following (Figure 4).
- Dimensions: External diameter: 13.8 m; initial diameter: 9.5 m;Crucible Height: 4.5 m. The
design production is about 4000 tonuday-1
- Refractory characteristics:
Refractory in contact with the pig iron: silicious aluminous with thermal conductivity
O = 3.5 Wum-1uK-1. This refractory is in contact with a graphite refractory that in the bottom has
a O = 22 Wum-1uK-1 and in the wall of the crucible 12 Wum-1uK-1.
- Refrigeration:
Air or water forced convection depending on the zones. The air convection represents an average
convective heat transfer coefficient hc of 100 Wum-2uK-1 and the water of 3000 Wum-2uK-1.

Figure 4: Characteristics of Blast Furnace crucibles analyzed by the NWM. 22 points are showed
as reference points from the FEM grid.

570

The corrosion that takes place in the crucible of the Blast Furnace was studied according to the
chemical dissolution equation, equivalent to the example of the finger test. Both equation Error!
Reference source not found. and Error! Reference source not found. are used to evaluate the
corrosion profile. According to the thermal conditions for the refrigeration in the crucible,
different profiles were obtained (Figure 5, Figure 6 and Figure 7).

Figure 5: Central well corrosion profile for a carbon based crucible (C-I in Error! Reference
source not found.).

Figure 6: Elephant foot corrosion profile for a carbon based crucible (C-II in Error!
Reference source not found.).

571

Figure 7: Mushroom shape corrosion profiles, with pronounced differential wear at the wall
(C-III in Error! Reference source not found.).
In Figure 6 the elephant foot appears clearly and differs from the profile of Figure 7 where the
attack on the bottom of the crucible seems stopped. The difference is only the high heat transfer
in the corners that exist due to the circulation of the pig iron. This result confirms the
experimental evidence for the relation between the dimensions, position and contact of the
inactive coke zone (deadman) with the bottom of the crucible[15,16]. To avoid the preferential
corrosion that produce the elephant foot shape the use of refrigerated corners build by
graphite bricks of high thermal conductivity (O= 44um-1uK-1) is a simple solution. It reduces the
wear in the corners up to 40% and delay the apparition of the elephant foot.
Changing the materials properties and the value of hc the simulation of new configurations and
the evolution during a campaign is possible. The results of this type of analysis were done in
collaboration with Saint Gobain CREE[17,18]. It was shown that the refrigeration became the most
important parameter for the control of the wear as the thickness of the refractory diminishes. A
control strategy could be defined to stop as much as possible the advance of the corrosion
boundary.
With low refrigeration conditions in bottom of the crucible, hc < 100, the perforation of this zone
or lateral walls is provoked when the furnace has not yet reached six years of operation. On the
other hand, due to the geometric characteristics of the crucible, the use of low thermal
conductivity materials introduce the risk of perforation after the second year of campaign, even
using extreme cooling conditions in the walls and in the bottom of the crucible.
In any case the evolution of the corrosion increases the volume of the crucible that represents an
increase on the capacity of the furnace. If the corrosion profiles are well developed the volume
increase facilitates the thermal stability of the whole system. A crucible with high thermal
conductivity materials will have less corrosion and a uniform profile. For the case of low thermal
conductivity materials, although they present a greater corrosion rate during the first years, the
final stabilization shape is similar to the case of high thermal conductivity. The corrosion
phenomena in the crucible of a Blast Furnace has to be analyzed as a dynamic process in view of
a control strategy to achieve the increase of the volume and no only to stop the corrosion.
572

The criteria to define a best design for the crucible of a blast furnace is certainly complex, but a
combination of the choice of the materials and the control on the refrigeration in function of the
advance of the campaign can extend the live of the crucible and prevent specific catastrophic
events.
Discussion and Conclusions

The combination of the corrosion equations with the NWM provides a tool for the analysis of the
corrosion/wear phenomena in many different configurations. Even of the fact that the application
of the model needs powerful computation facilities the bases for the calculations are simple.
The two step of the model require a comprehensive understanding of the mechanism that can
control the corrosion/wear phenomena and a precise evaluation of the temperature field in the
lining. Those requirements are without doubt the limitation to apply this model to other furnaces
and process. For the corrosion/wear phenomena all the possibilities of experimental test and post
mortem analysis can help to define the control step. It is relevant for this step to have reliable
data for the physical chemistry properties, both for the solid materials and the melts.
Unfortunately the data available doesnt cover all the needs, and nowadays very few researches
are doing in this field. The value of interfacial properties such contact angle and interfacial
energy are scare and specially in the non ferrous industry the gap is impress and very few data is
available, for example, in the matte/oxide systems.
The border conditions for the calculation of the thermal profile that define the nodal temperature
must be evaluated directly on the furnace. Some on line instrumentation are a very important
support for this calculation. If it is possible to have on line measurements the application of
NWM can give on line estimation of the modification of the geometry of the lining and in this
sense the on line thermal measurement could be transform on a on line monitoring of the lining
integrity.
The equations derived in this model have the same driving force that the previous obtained from
literature. The dependence on the nodal temperature makes the difference due to the explicit
relation on the temperature field of the domain where the equation will be applied and no only on
a nominal temperature. It is useful to underline that a Nodal Property depends on the nodal
temperature and the temperature of the adjacent nodes.
Some paradigms in the analysis of the phenomena that occur in the interface of a refractory
lining exist up to date. Many analyses consider the temperature at the interface melts/refractory
as the bulk temperature and supposed constant the long of the lining. If we can imagine a
constant field in the bath, situation far from a real condition, the differential corrosion/wear that
appears in the lining of any type of furnace must produce an inhomogeneous dissipative heat
transfer in the lining, and consequently the temperature in the interface cant be constant in the
whole interface. Certainly the direct measure of this temperature field is almost impossible to
achieve. Another consideration which is usually accepted is the fact that an insulation between
the work refractory and the shell of the furnace is of advantage, due to a lower heat loses. It was
proved that this configuration affects negatively the live of the refractory. When this insulation is
replaced by a material with a higher thermal conductivity, or just eliminated, the performance of
the refractory that is in contact with the melts increases. In this last case the work refractory
contacts with the shell by means of low rigidity polymeric joining.
Different new analyses are under development. In the case of the Pierce Smith[19] converter the
mechanism of thermal fatigue was introduce with a special equation that relates the advance of a
defect up to the apparition of a crack. In this case the nodal equation derived is combined with
573

the infiltration of the melts in the open porosity that produces a change in the mechanical
properties. The thermal cycles during the blowing steps, charging and tapping introduce a
mechanic stress due to the different thermal expansion coefficient.
The NWM also helps to explain the discontinuous corrosion profile (preferential corrosion) that
appears for example in the triple point of liquid/solid/gas contact where two type of driving
forces could participate: the thermal discontinuity that can induce a thermal Marangoni effect or
the solute Marangoni effect due to a gradient of composition.
Finally, it is important to underline that this model will no replace all the other analysis done on
the corrosion/wear phenomena in refractory materials. More than this, the NWM needs all the
information that is provided by experimental test and post mortem analysis to have a well
understanding on the basic physical chemistry phenomena. In this sense and with the new
materials proposed as alternative to the classic ones used in the pyrometallurgical reactors, more
experimental research must be done and a combination of experimental work with the NWM is a
option for the design of new materials and lining in furnaces.
Acknowledgments

For the help and collaboration received to develop this work and the support in our research we
want to thank the Education and Science Ministry of Spain (grant MAT2003-00502);
ECOS CONICYT Grant C02.E04; Foreign Affairs and Cooperation Ministry of Spain and
International Cooperation Agency of Spain (grant /1629/04); Chagres Smelter, Angloamerican
Chile; Saint Gobain CREE in France; Andes Foundation (grant C-13755-1) and the Argentinean
Institute of Iron and Steelmaking.

References

1. R. Cooper and W. D. Kingery, Dissolution in Ceramic Systems I : Molecular diffusion,

natural convection and forced convection studies of sapphire dissolution in calcium
aluminum silicate, J. Am. Ceram. Soc., 47 [1] 37-43 (1964).
2. B. N. Samaddar, W. D. Kingery, and A. R. Cooper. Dissolution in Ceramic Systems II:
Disolution of alumina, mullite, anorthite and silica in a calcium aluminum silicate salg,
J. Am. Ceram. Soc., 47 [5] 249-254 (1964).
3. Y. Oishi, A. R. Cooper, and W. D. Kingery. Dissolution in Ceramic Systems III: Boundary
layer concentration gradients, J. Am. Ceram. Soc., 48 [2] 88-95 (1965).
4. P. Hrma. Contribution to the study of the function between the rate of isothermal corrosion
and glass composition, Verres Refract., 24 [4-5] 166-168 (1970).
5. Cook, L. P.; Bonnell, D. W. & Rathnamma, D. Model for molten salt corrosion of
ceramics, Ceramic Transactions 10 [Corros. Corros. Degrad. Ceram.] 251-75 (1990).
6. Factsage, Ecole Polytechnique, Canada.
7. Mtdata - National Physical Laboratory, England
8. Thermocalc, KTH, Stockholm, Sweden
9. Annual Book of STM Standards, Vol. 15.01, 1990, American Society for Testing &
Materials.
10. McCauley, R.A.(1995): Corrosion of Ceramics. Marcel Dekker, New York, USA.
574

11. Christian Goi, Desarrollo y aplicacin de modelos de corrosin refractaria en un

Convertidor Peirce Smith, Departamento de Ingeniera Metalrgica, Universidad de
Concepcin, 2004.
12. N. Eustathopoulos, R. Parra & M. Snchez, Role of capillarity in corrosion of refractories
by melts, Copper 2003 International Conference.
13. Washburn, E. W, Dynamics of capillary flow Physical Review 17, 374-517 (1921).
14. L. F. Verdeja, R. Parra, J. P. Sancho & J. Bulln, Corrosion mechanism and wear prediction
on the sole of an electric arc furnace, ISIJ-Int.,43 [2] 192-200 (2003).
15. D. Jamenson, H Lungen & D. Lao Technical study into the means of prolonging furnace
campaign life. European Commission Technical Steel Research. EUR 17247, Luxembourg.
December 1997, pp. 133.
16. M. J. Venturini, J. P. Bolsingner & J. Iezzi Computations and measurements of liquids flow
in the hearth of the blast furnace. 57th Ironmaking Conference Proceedings. Vol. 57,
Toronto (Canada), Iron and Steel Society Ed,, Warrendale, Pennsylvania (USA), 1998, pp.
615-662.
17. Mara Florentina Barbs, Tsis Doctoral El modelo de desgaste nodal en la simulacin y
diseo de los crisoles de alto horno, Departamento de Ciencia de los Materiales e Ingeniera
Metalrgica, Universidad de Oviedo (2003).
18. Verdeja, L. F.; Gonzlez, R.; Alfonso, A. & Barbes, Ma. F.. Nodal wear model: Corrosion
in carbon blast furnace hearths. Revista de Metalurgia, 39 [3], 183-192 (2003).
19. Vanesa Bazn, Ph. D. Thesis, Departamento de Ingeniera Metalrgica, Universidad de
Concepcin, unpublished data.

575