1120 Ind. Eng. Chem. Res.

2009, 48, 1120–1125

Performance Analysis of a Smelting Reactor for Copper Production Process

Pimporn Chamveha, Kattiyapon Chaichana, Anon Chuachuensuk, Suthida Authayanun, and
Amornchai Arpornwichanop*
Department of Chemical Engineering, Faculty of Engineering, Chulalongkorn UniVersity, Bangkok 10330, Thailand

A primary smelting reactor (PSR) is an important unit in the pyrometallurgical process for the production of
copper. This article presents a performance analysis of a PSR with respect to the effects of various operating
parameters. The model of the PSR was developed using a metallurgical process simulator, METSIM. Reactor
parameters, i.e., heat loss and phase distributions of matte and slag, were estimated from actual plant data.
The model predictions were validated with plant data, and good agreement was observed. The developed
PSR model was used to analyze the influence of the feed rates of copper concentrate, oxygen, silica flux, and
revert on the PSR performance in terms of the percentage of copper in white metal and PSR slag, the percentage
of magnetite and silica in PSR slag, and reactor temperature.

1. Introduction silica flux on the performance of the PSR in terms of the

percentages of copper in the white metal and the slag.
Copper is an important commodity that is widely used in
electrical/electronic products, building construction, industrial 2. Description and Modeling of a Primary Copper
machinery, and general consumer goods. In the production of Smelting Reactor
copper, most producers generate full plate copper cathodes as
a final product from the complex manufacturing process A schematic diagram of a primary smelting reactor is shown
consisting of smelter and refinery units. However, some in Figure 1. Dry copper concentrate is fed into the reactor with
producers ship their output as wire bars to hot rolling mills, revert and silica flux. Oxygen is injected into the reactor in order
and a few specialized producers provide copper powders and to promote oxidation reactions.8 Minerals in the concentrate are
bronze and brass alloys.1 decomposed and oxidized (eqs 1-7)1 at temperatures above
In general, a pyrometallurgical process is applied to extract 1250 °C, resulting in segregated liquid layers: matte and slag.
copper from copper-iron sulfide (Cu-Fe-S) ores consisting The matte, a mixture of copper and iron sulfide (Cu2S and FeS)
of 0.5-2% Cu.2,3 In this process, a primary smelting reactor containing 45-75% Cu, is further processed to obtain the final
(PSR) is the major unit used to smelt “copper concentrate”, copper product, whereas the slag, composed of iron oxide (FeO)
which is obtained from the isolation of Cu-Fe-S minerals in and silica, is treated for copper recovery. The off-gas comprising
ores via physical means, i.e., a froth floatation process. The aim mainly sulfur dioxide (SO2) is sent to a sulfuric acid production
of the smelting reactor is to oxidize S and Fe in the copper plant.
concentrate to produce a high-Cu molten sulfide (“white metal”)
Decomposition of Cu-Fe-S mineral
and a molten oxide slag.1 Rosales et al.4 showed that the
metallurgical efficiency of smelting reactors depends strongly 4CuFeS2 w 2Cu2S(matte) + 4FeS(matte) + S2 (1)
on the utilization of oxygen within the reactor during the 4Cu5FeS4 w 10Cu2S(matte) + 4FeS(matte) + S2 (2)
smelting and converting reactions of Cu-Fe-S minerals in the
copper concentrate that leads to the production of white metal. Pyritic sulfur oxidation
Luraschi and Canas5 performed a thermodynamic analysis of a S2+2O2 w 2SO2 (3)
conversion process in a smelting reactor and compared the
model predictions with operating data. Considering the copper FeS/FeS2 oxidation
production process, it has been accepted that major copper loss 2FeS + 2O2 w 2FeO(slag) + 2SO2 (4)
is found in a discarded slag.6 Therefore, the avoidance of copper
loss in the discarded slag is an important issue for enhancing 2FeS2+5O2 w 2FeO + 4SO2 (5)
the efficiency of the copper smelting process, resulting in higher Cu2S/CuS oxidation
yields. To achieve this objective, an understanding of the effects
of operating parameters on the performance of smelting reactors 3
Cu2S + O2 w Cu2O(slag) + SO2 (6)
is needed in order to operate the reactor efficiently and to 2
minimize copper losses in the copper smelting process.7 2CuS + O2 w Cu2S + SO2 (7)
Therefore, the aim of this study was a performance analysis
of a PSR with respect to various operating parameters. Modeling As FeO is highly reactive liquid, particularly in the presence
of the PSR was performed using a metallurgical process of oxygen, it can further react with oxygen (eq 8). This leads
simulator, METSIM. Reactor model parameters used for to the formation of magnetite (Fe3O4), which is maintained in
simulations of the PSR were estimated from actual smelting the solid phase at the operating temperature of the smelting
plant data. The model developed was used to investigate the reactor (about 1250 °C), thus complicating reactor operation.
influence of the feed rates of copper concentrate, oxygen, and Because of agitation in the reactor, the FeS in the matte phase
reacts with Fe3O4 to generate SO2 (eq 9). Upon the generation
* To whom correspondence should be addressed. Tel.: +66-2- of SO2 gas, bubbles are produced that allow the white metal
2186878. Fax: +66-2-2186877. E-mail: Amornchai.A@chula.ac.th. particles (Cu2S) to be dragged to the slag by floatation, resulting
10.1021/ie800618a CCC: $40.75  2009 American Chemical Society
Published on Web 10/09/2008
 Ind. Eng. Chem. Res., Vol. 48, No. 3, 2009 1121

Figure 1. Primary smelting reactor.

Table 1. Comparison of Simulation Results and Plant Data

key variables plant data simulation
Cu percentage in white metal 72.43 72.40
white metal flow rate (tph) 530 530.92
Cu percentage in slag 6.02 6.29
slag flow rate (tph) 880 884.24
T (°C) 1199.5 1197.15
Fe/SiO2 ratio 1.473 1.60
Fe3O4 percentage in PSR slag 18.5 18.18

Table 2. Plant Data for Model Validation

dry blowing garr injection
data concentrate flux revert air oxygen gun air air
set (tpd) (tpd) (tpd) (Nm /h) (Nm /h) (Nm /h) (Nm3/h)
3 3 3

1 1565 120 239 36 584 7500 4006 1816

2 1567 119 240 36 745 7865 3996 1832
Figure 2. Flow diagram of a primary smelting reactor. 3 1555 120 240 37 854 7867 4007 1947
4 1540 120 240 37 912 7886 3997 1909
in a higher loss of copper in the final slag. To reduce the 5 1537 121 240 37 948 7889 3999 1903
chemical activity of FeO, silica (SiO2) as the flux is added to 6 1525 120 240 37 881 7860 4003 1878
7 1565 120 228 37 870 7883 3997 1917
the reactor (eq 10) in order to form ferrous orthosilicate, 8 1493 119 120 37 931 7169 4000 1790
2FeO · SiO2 (fayalite), as slag. 9 1482 121 192 36 469 7099 4004 1840
10 1481 121 192 36 295 7035 4013 1869
Magnetite formation
oxygen with 95% purity, blowing air from air compressors, and
6FeO + O2 w 2Fe3O4 (8) annular air that protects tuyeres from aggressive oxidation
3Fe3O4 + FeS w 10FeO + SO2 (9) reactions.
To perform a simulation of the FRL furnace module, the
FeO oxidation reactor model parameters of the heat loss and phase distribution
within the reactor are required. A least-squares parameter
2FeO + SiO2 w 2FeO·SiO2(slag) (10)
estimation problem was formulated and solved with the objective
It is noted that the heat generated by exothermic oxidations of minimizing the difference between the model predictions and
inside the reactor is used for the endothermic decomposition of real plant operating data. The optimization results showed that
the copper concentrate. This process is known as an autogenous the reactor heat loss is 5.1 Mcal/h, the white metal product
smelting process because no external heat is supplied. The heat derives from 87% matte, and the PSR slag consists of 13% matte
generation from the smelting process depends on the standard and 100% slag. Table 1 reports a comparison of plant data and
enthalpies of reaction of the minerals contained in the copper the model prediction based on the estimate reactor parameters;
concentrate. In general, the total oxygen flow is a key parameter good agreement of the results is observed.
for improving energy balance and controlling matte quality.4 To apply the FRL furnace model of METSIM to the
In this study, the modeling of a copper smelting reactor was simulation of a primary smelting reactor unit with confidence,
performed using METSIM, simulation software for metallurgical the reliability of the model was tested by comparing simulation
reaction processing. The FRL furnace module in METSIM was results with various available plant data sets. With the operating
used to represent the reactor (Figure 2). Feed streams of the conditions given in Table 2, the model results and actual plant
PSR comprise dry copper concentrate, revert, silica flux, and data in terms of the percentage of copper in the white metal,
enriched air. Using data from actual plant operation, mineral the percentage of copper in the PSR slag, the percentage of
compositions of the copper concentrate obtained from a mixture Fe3O4 in the PSR slag, and the reactor temperature were
of three different copper concentrates (Antamina, Escondida, compared as shown in Figures 3-6, respectively. It can be seen
and Collahuasi) were determined to be mainly chalcopyrite that the copper content in the white metal, the magnetite content
(CuFeS2), pyrrotite (FeS), and pyrite (FeS2). Revert is a in the PSR slag, and the reactor temperature as calculated by
byproduct of the copper smelting process and is recycled to METSIM are in good agreement with the plant data. However,
the PSR because of its high copper content (40% Cu), whereas it is observed that the copper content in the PSR slag is slightly
the silica flux used for the formation of slag has a major mineral different from the actual plant data. This can be explained by
content of silica (95% SiO2) and alumina (5% Al2O3). The the way in which slag sampling was performed. Different
enriched air derives from three different sources: industrial operators take samples at different times and positions in the
1122 Ind. Eng. Chem. Res., Vol. 48, No. 3, 2009

Figure 3. Comparison of the Cu percentage in white metal from model Figure 6. Comparison of the reactor temperature from model predictions
predictions and plant data. and plant data.

Table 3. Major Mineral Composition (%) of Copper Concentrates

copper concentrate
component 25% Cu 30% Cu 35% Cu
chalcopyrite CuFeS2 60 43 25
chalcocite Cu2S 0.5 6 13
covellite CuS 1 1 0
bornite Cu5FeS4 5 15 25
pyrrotite FeS 3 8.5 14
pyrite FeS2 12 13 14

Table 4. Operating Ranges of Parameters for Simulation Studies

parameter nominal value range
dry concentrate feed rate 70 tph 60-80 tph
oxygen flow rate 9836 Nm3/h 6000-10 000 Nm3/h
Figure 4. Comparison of the Cu percentage in slag from model predictions flux feed rate 6 tph 2-10 tph
and plant data. revert feed rate 0 tph 0-20 tph

and slag, the percentage of magnetite (%Fe3O4) in the slag, the

percentage of silica (%SiO2) in the slag, and the reactor
temperature as functions of key operating parameters was
performed. Because the composition of the copper concentrates
from the many sources around the world varies, the effect of
the copper concentrate composition on the reactor performance
was also investigated. Table 3 lists the major components of
the copper concentrates used in this study. It is noted that, to
illustrate the effect of chemical compositions, the main mineral
components in the copper concentrate such as chalcopyrite,
chalcocite, and covellite were calculated to obtain copper
concentrates with 25%, 30%, and 35% Cu. Table 4 lists the
ranges of the adjusted parameters that can normally be changed
to continue stable operation without production interruption.
Figure 5. Comparison of the Fe3O4 percentage in slag from model
predictions and plant data.
3.1. Effect of Concentrate Feed Rate. Figure 7 shows the
effects of the variation of the Cu content in the copper
PSR. Because of the maldistribution of the white metal in the concentrate from 25% to 35% and the variation of the feed rate
slag within the reactor, the PSR slag product removed from a of the copper concentrate from 60 to 80 ton/h (tph) on the
slag tap hole consists mainly of the slag phase. Therefore, the performance of the smelting reactor. It was found that an
Cu percentage in the PSR slag obtained from plant data is lower increase in the concentrate feed rate increases the Cu percentage
than the predicted result. To reduce the error from different in the white metal only slightly compared to an increase in the
operators sampling the slag, a work instruction should be Cu percentage in the copper concentrate. It can be seen from
prepared. The schedule of tapping can affect these data values Figure 7a that an increase of 5 tph in the concentrate feed rate
because of the different phase levels in the smelting reactor. It can increase the Cu percentage in the white metal by 0.1%
is noted that, by considering the content of Fe3O4 in the PSR whereas a 5% increase in the copper content in the concentrate
slag, the method of slag sampling has a slight effect on the can increase the Cu content in the white metal by 1%. This is
deviation between the model and plant data as Fe3O4 is dissolved mainly due to the higher degree of mineral decomposition. A
only in the slag phase. greater quantity of copper in the concentrate can increase the
copper content in the white metal and, consequently, has a
greater effect on the composition of the white metal than the
3. Simulation Results and Discussion
feed rate.
A sensitivity analysis of the smelting reactor performance in Increasing the concentrate feed rate also increases the Cu
terms of the percentages of copper (%Cu) in the white metal content in the slag slightly because the amount of white metal
 Ind. Eng. Chem. Res., Vol. 48, No. 3, 2009 1123

Figure 7. Effect of copper concentrate feed rate on the performance of a

PSR at different copper percentages in the concentrate feed. Figure 8. Effect of oxygen flow rate on the performance of a PSR at
different copper percentages in the concentrate feed.
distributed to the slag phase is slightly increased. In addition, could be due to the differences in mineralogical composition
although the slag-forming reaction is still the same, there is an as the copper content in the concentrate increases. As a
increase in the quantity of slag formed. The combined effects consequence of the higher reactor temperature resulting from
result in a slight increase in the percentage of copper in the the higher concentrate feed rate, an increasing trend of the yield
slag phase. In comparison, there is higher copper loss in the in the PSR is also observed (Figure 7f). It is noted that the yield
slag with a higher copper content in the concentrate than with of the PSR is defined as the total content of copper in the white
an increased concentrate feed rate. As can be seen from Figure metal divided by the total input amount of copper, which is
7b, a 5% increase in the copper content of the concentrate results derived from the copper concentrate and the revert.
in a 1.5% increase in the Cu content in the slag because of the 3.2. Effect of Oxygen Flow Rate. The oxygen supply to
higher amount of the white metal and the higher copper content the reactor is one of the key parameters in a PSR because most
of the white metal. important reactions in the PSR involve oxygen. It is noted that
It is shown from the trends in Figure 7c that the amount of the heat generated from these combustion reactions is required
Fe3O4 increases slightly when the concentrate feed rate and for an autogenous operation. For this reason, the optimum
copper concentration in the dry concentrate are increased. The oxygen supply to the furnace should be calculated. Oxygen
copper concentrate with 25% Cu has the lowest magnetite comes from two main sources: industrial oxygen and blowing
content. It is possible that, with the lower Cu percentage in the air. This study considered the variation of industrial oxygen as
white metal, the FeO for the slag-forming reaction is also at its the adjusted parameter.
lowest and, consequently, the magnetite content is reduced. Figure 8a demonstrates that there is a slight change in the
On the other hand, the amount of SiO2 in the slag shows a percentage of copper in the white metal with variations in the
decreasing trend when the flow rate of the concentrate is oxygen flow rate because the quantity of copper that reacts with
increased, as shown in Figure 7d. This is because of the oxygen is still the same and the use of oxygen is in the same
increased amount of FeO in the slag and the addition of slag- proportion. The unreacted oxygen goes to the off-gas stream.
forming reactions. It should also be noted that the concentrate The utilization of oxygen in the reactor is generally called the
feed with 25% Cu has the highest percentage of SiO2 in the oxygen efficiency. In normal operation, for every copper
slag because of the assumption that the gangue that is used to concentrate blend, the oxygen coefficient, which indicates the
balance the mineral compositions of the copper concentrate at amount of oxygen feed for combustion and slag-forming
different percentages of copper is represented by silica so that reactions per ton of concentrate feed, is determined. The
the concentrate with less copper has higher silica content. coefficient of oxygen in this simulation case increased as a
The effects of increasing the concentrate feed rate and the consequence of the increase in the industrial oxygen flow rate.
Cu percentage in the concentrate are to increase the reactor Normally, operation of a PSR with a higher coefficient of
temperature, as shown in Figure 7e. This is a consequence of oxygen than required for the concentrate blend results in a higher
the higher net heat generated from the decomposition and Cu content in white metal and a higher temperature.
combustion of the copper concentrate. It is noted that the The dependence of the Cu percentage in the slag on the
proportion of increasing the Cu content in the concentrate from oxygen flow rate shows a flat trend, as can be seen in Figure
25% to 30% has an effect on the PSR temperature in a different 8b. It is observed that, when a concentrate with higher copper
range than the case of increasing from 30% to 35% Cu. This is used, the copper loss in the slag is higher because the quantity
1124 Ind. Eng. Chem. Res., Vol. 48, No. 3, 2009

Figure 10. Effect of revert flow rate on the performance of a PSR at different
copper percentages in the concentrate feed.
Figure 9. Effect of silica flux feed rate on the performance of a PSR at
different copper percentages in the concentrate feed.
of white metal that distributes to the slag phase remains the
of white metal is increased and, consequently, the amount of same. This results in a decrease in the proportion of copper in
white metal distributed to the slag phase increases. the slag.
It should be considered from the simulation results shown in Figure 9c,e shows that flux feed rates below 4 tph result in
Figure 8a-c that increasing oxygen flow rate has only a minimal high magnetite levels in the slag and high temperatures,
effect on the quality of white metal and slag (percentages of respectively. The high magnetite-forming reaction generates
Cu and Fe3O4) because of the fixed extent of the reactions. It is more heat, thus increasing the PSR temperature. This reaction
assumed that, when all of the copper and iron react with the occurs when the amount of silica added is not enough to form
oxygen at the specified extent of reaction, all of the excess fayalite slag and the remaining FeO is overoxidized to form
oxygen added will not react any further but instead will go to Fe3O4. When the flux feed rate is increased to more than 4 tph,
the off-gas system. However, the trend in the silica content the quantity of silica is sufficient to react with FeO to produce
shown in Figure 8d is different because of the effect of the fayalite slag, and the magnetite content is gradually reduced.
copper content in the concentrate. It can be seen that the con- In Figure 9d, the trend shows the opposite way. As more silica
centrate with 35% Cu has the lowest silica content in the slag, is added in excess, the silica goes to the slag phase. Considering
as the gangue used to balance the concentrate composition is the yield of PSR, it can be seen from the trends in Figure 9f
assumed to be silica. It is noted that the range of oxygen flow that the flux feed rate does not has an influence on the PSR
rates used, 7000-10 000 Nm3/h, is high enough to make the yield, as it has no significant effect on the Cu percentage in the
FeO react completely with the oxygen, resulting in a constant white metal, as shown in Figure 9a. In general, silica flux has
percentage of Fe3O4 in slag. The effect of the oxygen flow rate a greater effect on the slag quality and quantity.
on the reactor temperature is shown in Figure 8e: the temperature 3.4. Effect of Revert Feed Rate. The variation of the flow
decreases when the flow of oxygen is increased. This indicates rate of the revert, which consists mainly of Cu and Fe, was
that an oxygen flow rate of 7000 Nm3/h is sufficient to react determined by the design capacity to be 0-20 tph. The purpose
with the copper concentrate. Increasing the oxygen further does of revert addition to the PSR is to recover the copper in the
not result in further reactions but rather causes heat loss from smelting process and to control the reactor temperature. The
the PSR. Because the Cu percentage in the white metal does composition and quantity of the revert added have many effects
not change with oxygen flow rate, the PSR yield is constant on the process, as shown in Figure 10a-e.
(Figure 8f). It can be seen from the trends in Figure 10a that the Cu
3.3. Effect of Silica Flux Feed Rate. It can be seen from percentage in the white metal increases with revert addition, as
the trends in Figure 9a that flux addition has no effect on the the copper sulfide in revert, when melted, goes directly to the
copper in the white metal. With an increase in the feed rate of matte phase. The effect of revert addition on the Cu percentage
flux, the silica in the flux reacts with iron oxide to form slag in the slag is shown in Figure 10b, which indicates that the
and has no consequence on the white metal, which is mainly low-copper-content concentrate (25% Cu) is more sensitive to
composed of copper sulfide and iron sulfide. It can be seen from the variation of the revert feed rate. Figure 10c shows that the
the trends in Figure 9b that the Cu percentage in the slag Fe3O4 in the slag increases with increasing revert feed rate.
decreases with increasing flux feed rate because the amount of The magnetite present in the revert is melted and dissolved in
slag increases with the addition of silica whereas the quantity the slag phase. In Figure 10d, the reduction of silica in the slag
 Ind. Eng. Chem. Res., Vol. 48, No. 3, 2009 1125
from the addition of revert is the effect of the slag-forming Acknowledgment
reaction where the amount of FeS increases from revert melting
Support from the Thailand Research Fund (TRF), the Office
and consumes silica to form slag. Figure 10e illustrates that the
of Small and Medium Enterprises Promotion (OSMEP), and
PSR temperature decreases when the feed rate of revert is
the Graduate School of Chulalongkorn University is gratefully
increased. This is consistent with more heat being required to
acknowledged. The authors thank Dr. Romeo U. Pagador for
melt the revert. It is noted that the adjustment of the revert feed
useful discussions and suggestions and Thai Copper Industries
rate is used to control the reactor temperature when it exceeds
Public Co. Ltd. for providing the technical process data used
1250 °C.
in this study.
The effect of revert addition on the PSR yield is shown in
Figure 10f. When the revert feed rate is greater than 10 tph, the
increase in yield is lower than for revert flow rates in the range Literature Cited
of 0-10 tph. The addition of more revert to the reactor can
improve the yield because the copper in the revert reacts and (1) Davenport, W. G.; King, M.; Schlesinger, W. G.; Biswas, A. K.
increases the amount of copper sulfide in the white metal. ExtractiVe Metallurgy of Copper, 4th ed.; Elsevier: New York, 2000.
(2) Mackey, P. J.; Campos, R. Modern Continuous Smelting and
Copper in the slag also increases because of the amount of
Converting by Bath Smelting Technology. Can. Metall. Q. 2001, 40, 355.
copper oxide. However, the amount of copper oxide in the revert
(3) Moskalyk, R. R.; Alfantazi, A. M. Review of copper pyrometallur-
is less than the amount of copper sulfide, so there is still an gical practice: Today and tomorrow. Miner. Eng. 2003, 16, 893.
improvement in the yield. It is noted that the revert composition (4) Rosales, M.; Ruz, P.; Fuentes, R.; Moyano, A. Oxygen Efficiency
has a significant impact on the yield, so the analysis of the revert Calculation in Teniente Converters. In Proceedings of Copper 2003-Cobre
composition should have a certain level of accuracy. In this 2003: The Hermann Schwarze Symposium on Copper Pyrometallurgy; Diaz,
study, the revert composition was obtained through reconcili- C., Landolt, C., Utigard, T., Eds.; Canadian Institute of Mining, Metallurgy
and Petroleum: Montreal, Quebec, Canada, 2003; Vol. VI-Pyrometallurgy
ation with the laboratory data for the analysis of Cu, Fe, S, of Copper, pp 241-249.
Fe3O4, and SiO2. (5) Luraschi, A.; Cañas, J. D. Thermodynamic Fundamentals of Teniente
ConVerter Smelting; CADE IDEPE: Santiago, Chile, 1997.
4. Conclusion (6) Sridhar, R.; Toguri, J. M.; Simeonov, S. Copper Losses and
Thermodynamic Considerations in Copper Smelting. Metall. Trans. B. 1997,
In this study, a performance analysis of a primary smelting 28B, 191.
reactor (PSR) was carried out to investigate the effects of reactor (7) Harjunkoski, I.; Borchers, H. W.; Fahl, M. Simultaneous Scheduling
operating parameters. An FRL furnace module of the METSIM9 and Optimization of a Copper Plant. Comput.-Aided Chem. Eng. 2006, 21,
simulator was used to model the reactor. The validation of the 1197.
model coupled with estimation of reactor parameters was (8) Alvarado, R. ; Lertora, B.; Hernandez, F.; Moya, C. Recent
development in the Teniente converter. In Proceedings of Copper 95-
performed with actual industrial process data. From simulations, COBRE 95 International Conference; Chen, W. J., Diaz, C., Luraschi, A.,
it was found that the feed rate of copper concentrate has a Mackey, P. J. Eds.; Canadian Institute of Mining, Metallurgy and Petroleum:
significant influence on the reactor performance. The effect of Montreal, Quebec, Canada, 1995; Vol. IV-Pyrometallurgy of Copper, pp
varying the oxygen flow rate, if the amount of oxygen is not 83-101.
sufficient to oxidize the minerals, is to decrease the quality of (9) METSIM Simulations Modules (http://www.metsim.com).
the white metal. The silica flux feed rate is a key variable for
controlling slag quality because it is used to generate fayalite ReceiVed for reView April 17, 2008
slag. Finally, the revert feed rate affects PSR performance ReVised manuscript receiVed August 27, 2008
Accepted August 28, 2008
because revert components consist of both white metal and slag
materials. IE800618A