Advanced Thermochemical Fundamental and Applied Research to Improve the Integrity of the Steel Water Jacketed Furnace at Port Pirie

Abstract

Lead blast furnace water-jacket stability was investigated by examining samples from the Port Pirie Lead Smelter. The compositions of the phases formed in the samples were analysed using Scanning Electron Microscopy (SEM) and Electron Probe X-ray Microanalyser (EPMA). Phase equilibria analysis and thermodynamic modelling was undertaken using FactSage with latest thermodynamic database for the slag /matte/metal/speiss/solids multi-component Cu₂O–PbO–ZnO–Al₂O₃–CaO–MgO–FeO–Fe₂O₃–SiO₂–S major and As–Bi–Sb–Sn–Ag–Au minor elements system. Results indicated that the stability of the water-jacket steel wall without freeze-lining is influenced by formation of the complex Fe-rich speiss phase at critical As concentrations. The integrity of the furnace walls with freeze-lining is critically dependent of the stability of the oxide-based freeze-lining which is dependent on various factors including most critical one—the proportion of solids determined by the slag chemistry and fluxing. Thermodynamic modelling of the lead blast furnace operation indicated limiting thermochemical conditions for the water-jacket stability. Factors and controls critical to the water-jackets integrity are discussed.

Introduction

Port Pirie Lead Smelter located in South Australia has been in continuous operation for more than 128 years. The smelter annually produces 225,000 t refined lead and 500 t silver in dore [1], making it one of the largest pyrometallurgical primary lead smelting facilities and one of the top five largest silver producer globally. A redevelopment of the Port Pirie Smelter is underway to transform its operation from that of primary lead smelter into a highly flexible poly-metallic processing and recovery facility [2].

The lead smelting process begins with the oxidation of lead concentrate to lead oxide sinter/slag and SO₂ gas. The lead oxide is then reduced to produce lead bullion and zinc rich slag . At the later stages, the lead bullion is sent to a pyrometallurgyical lead refinery to increase its purity and to recover precious metals concentrate, while the zinc rich slag is processed in a slag fumer to recover Zinc Oxide .

At the Port Pirie Smelter, the reduction of the lead oxide sinter/slag to the lead bullion is carried out in a blast furnace . The blast furnace sidewalls are made of several hollow steel jackets, which are water cooled. The use of water-cooled jackets results in the formation of an accretion (frozen slag ) layer, which protects the surface of the furnace sidewalls from the molten charge. Due to the formation of protective accretion layer, no refractory lining is required in the furnace .

Severe interruption in the continuous operation of the blast furnace can happen due to the failure in the water-cooled jackets. Previous visual examination showed the presence of marks and holes on the hot surface of the jackets, which indicated aggressive chemical attacks of the jackets from the inside of the furnace .

The aim of this work is to investigate the lead blast furnace water-jacket stability using a combination of microanalysis techniques using SEM and EPMA, phase equilibria analysis and thermodynamic modelling using FactSage [3] with an advanced thermodynamic database . The advanced thermodynamic database contains slag /matte/metal/speiss/solids phases and multi-component Cu₂O–PbO–ZnO–Al₂O₃–CaO–MgO–FeO–Fe₂O₃–SiO₂–S major and As–Bi–Sb–Sn–Ag–Au minor elements system, which has been developed within a lead consortium research project at PYROSEARCH funded by several major lead producers.

Sample Collection and Analysis Technique

Steel water jacket samples were collected from the area where chemical attack of the wall were visually observed. Figure 1a shows the schematic of the bottom part of the blast furnace [4] and usual area where water jacket is chemically attacked and Fig. 1b shows example of the corroded water jacket. The damaged water jacket sample was cut into smaller pieces approximately 40 mm × 40 mm × 10 mm at the plant operation and was sent to PYROSEARCH for analysis.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig1_HTML.gif — Fig. 1
a Schematic section of the bottom part of the lead blast furnace [4] and approximate area of water jacket sample collection is indicated by the dotted square; and b Water jacket area with clear evidence of corrosion

Macro appearance of the water jacket samples are provided in Fig. 2. The side of the samples with rough surface was identified as the “hot-face”, i.e. the surface which was facing into the molten bath. Visual observation of the steel jacket sample shows evidence of reacted layer of the jacket with thickness of ~3 mm.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig2_HTML.gif — Fig. 2
Macro-appearance of steel jacket samples: a As-received samples from Port Pirie; and b Cross-sectioned of samples showing ~3 mm of reacted layer

Tapped lead bullion sample was also collected from the blast furnace . Macro appearance of the Pb bullion samples is provided in Fig. 3. Most part of the bullion had metallic appearance except on its top side where black coating was observed.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig3_HTML.gif — Fig. 3
Macro-appearance of as-received tapped lead bullion sample

The steel jacket and lead bullion samples were sectioned to approximately 15 mm × 15 mm × 10 mm and were mounted in epoxy resin. The mounted samples were polished using automatic polisher TegraPol-31 (Struers, Denmark) and carbon coated using QT150TES carbon coater (Quorum Technologies, UK). Optical microscopy and SEM with Energy Dispersive X-ray Spectroscopy (EDS) detector were used for initial microstructural and phase analyses.

Compositions of the phases in the steel jacket and tapped lead bullion samples were accurately determined using an EPMA, JEOL JXA 8200L (Japan Electron Optics Ltd., Tokyo). An acceleration voltage of 15 keV and a probe current of 20 nA were used. The Duncumb-Philibert ZAF correction procedure supplied with the JEOL JXA 8200L probe was applied. Suitable reference materials from Charles M. Taylor, Stanford, CA which include Cu, Fe, InAs, PbS and ZnS were used as standards.

Microanalysis Results

Steel Jacket Sample

Initial microstructure analysis using SEM with back-scattered electron mode indicates that the reacted layer on the damaged steel jacket surface contains mixture of two distinct phases as shown in Fig. 4. Initial phase analysis using EDS showed that the two phases contain significant amounts of iron and arsenic . The compositions of phases observed in the steel jacket sample from EPMA analysis are summarized in Table 1. The two phases appear to be FeAs-rich speiss (87.4wt%Fe–10.0wt%As) and FeAsZn-rich speiss (47.1wt%Fe–27.6wt%As–16.9wt%Zn).

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig4_HTML.gif — Fig. 4
Back-scattered electron image of the steel jacket sample with reacted layer

Table 1

Summary of phases compositions in steel jacket sample from EPMA analysis

No	Phase	Composition not normalized (wt%)						Total
No	Phase	Pb	Fe	As	Zn	Cu	S	Total
1	FeAs speiss	0.05	87.4	10.0	0.05	2.4	0.03	99.9
2	FeAsZn-speiss	0.31	47.1	27.6	16.9	5.9	1.28	99.1
3	Steel jacket	0.04	99.01	0.00	0.05	0.05	0.03	99.2
4	Metallic lead	99.7	0.05	0.00	0.00	0.14	0.00	99.9
5	CuZn(γ)	0.08	0.06	0.04	58.9	38.0	0.01	97.1
6	As₂Zn₃	0.00	0.00	40.5	53.8	0.34	0.04	94.7

The FeAsZn-rich speiss appears to be in a liquid state at the operation temperature . The FeAs-rich speiss was observed in two different morphologies, i.e. (i) a dense layer attached directly on the surface of steel jacket; and (ii) globular-shaped material within the FeAsZn-rich speiss matrix. The presence of dense layer of FeAs-rich speiss indicates that the phase was solid, while the globular-shaped FeAs-phase appears to be particles detached from the FeAs-dense layer. The mixed speiss (FeAs–FeAsZn) layer was estimated to have average composition of 64.2wt%Fe–20.5wt%As–9.9wt%Zn–4.6wt%Cu.

The binary phase diagram of As–Fe taken from Okamoto [5] is provided in Fig. 5. It can be observed that the FeAs-rich speiss is part of αFe solid solution with solidus temperature approximately 900 °C. The composition of FeAsZn-rich speiss, after normalized to Fe–As, is also plotted in Fig. 5. The liquidus temperature of the FeAsZn speiss is ~900 °C, which can decrease further with the addition of Zn and Cu in the speiss solution.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig5_HTML.gif — Fig. 5
The As–Fe binary phase diagram [5]

Apart from the speiss phases, intermetallic compounds of CuZn(γ) (melting point = 834 °C for pure compound [6]) and As₂Zn₃ (melting point = 1015 °C [5]) were also observed in the sample. The presence of the Zn-containing speiss and intermetallic compounds can be related to the high-Zn pressure condition in the area where the sample was collected.

Tapped Lead Bullion Sample

Figure 6 shows the microstructure of the tapped lead bullion sample taken using SEM with back-scattered electron mode. The softness of Pb metal created challenges for the metallography polishing technique, i.e. an uneven Pb metal surface and a presence of inclusions from polishing material as shown in Fig. 6. The black coating on the surface of the metal was found to be mixture of PbCu matte (43.6wt%Pb–40.1wt%Cu–14.9wt%S) and Cu-rich matte (3.3wt%Pb–75.4wt%Cu–20.9wt%S) phases. The Cu-rich matte appears to have originated from the PbCu matte formed during slow cooling of the lead bullion.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig6_HTML.gif — Fig. 6
Back-scattered electron image of the lead bullion sample

The EPMA measurements in Table 2 show that there are no significant concentration of As in the metallic PbCu matte and Cu-rich matte phases. The EPMA measurement also shows that the metallic Pb matrix does not contain significant solute concentrations (Cu, Fe, S, As), this may be due the precipitation of solute elements during cooling forming inclusions within the Pb matrix (see Fig. 6). From the historical plant data, the bulk tapped lead bullion actually contains between 0.11 and 0.92wt% As.

Table 2

Summary of phases compositions in the lead bullion sample from EPMA analysis

No	Phase	Composition not normalized (wt%)						Total
No	Phase	Pb	Fe	As	Zn	Cu	S	Total
1	Metallic lead	98.1	0.09	0.00	0.00	0.00	0.00	98.2
2	PbCu matte	43.6	0.17	0.00	0.00	40.1	14.9	98.8
3	Cu-rich matte	3.3	1.08	0.09	0.20	75.4	20.9	100.9

Thermodynamic Analysis

Thermodynamic modelling to investigate a possible water jacket failure phenomena has been carried out using FactSage [3] with the most recent thermodynamic database developed by PYROSEARCH specifically for lead smelting . The advanced thermodynamic database contains sophisticated mathematical models of Gibbs free energies for all relevant phases in the chemical system. The Modified Quasichemical Formalism in Quadruplet approximation [7]: [Pb²⁺, Cu¹⁺, Fe²⁺, Fe³⁺, Si⁴⁺, Al³⁺, Ca²⁺, Mg²⁺, Zn²⁺, As³⁺, Sn²⁺, Sb³⁺, Bi³⁺, Ag¹⁺, Au¹⁺] [O²⁻, S²⁻] is used for the slag . The matte, metal and speiss phases are modelled in one solution based on the Modified Quasichemical Formalism in Pair approximation [8]: (Pb^II, Cu^I, Cu^II, Fe^II, Fe^III, Zn^II, As^III, Sn^II, Sb^III, Bi^III, Ag^I, Au^I, O^II, S^II). The database also includes a wide range oxide and metal solid solutions, pure compounds and more than 100 gas species.

There are two regions of concern related to the stability of the water jacket. In contact with the lead bullion there is no freeze lining. Above the level of the bullion a slag freeze lining is formed on the inner surface of the water jacket.

Speiss Formation

The stability of the water-jacket steel wall without freeze-lining is related to the formation of the complex Fe-rich speiss phase. Thermodynamic prediction of the formation of speiss has been carried out by considering interaction between solid iron and lead bullion. The prediction was done by reacting 5 grams of solid iron and 95 grams of liquid Pb. This solid iron and liquid Pb ratio was selected since in practice the steel jacket was continuously exposed to a stream of lead bullion for a significant period of time. The base case composition for lead bullion used in the prediction was taken from average historical plant data from 2017, i.e. 95.5wt%Pb–2.4wt%Cu–0.47wt%As–0.70wt%S–0.86wt%Sb.

Figure 7a provides the prediction result showing proportion of phases as a function of As in lead bullion at 900 °C. During the normal operation of 0.47wt% As in lead bullion, solid solution of Fe is stable. Above a critical condition, which is approximately 0.52wt% As in lead bullion, formation of liquid speiss and aggressive corrosion of solid Fe are predicted. Figure 7b shows the influence of temperature on the critical As concentration in lead bullion for the formation of liquid speiss. With increasing temperature , it can be observed that the critical As concentration in the lead bullion decreases and hence the possibility of liquid speiss formation increases. The presence of Zn in the reactor gas may further enhance the formation of liquid speiss as observed in the corroded steel jacket sample.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig7_HTML.gif — Fig. 7
Thermodynamic prediction of speiss formation based on contact of 5 grams of solid iron and 95 grams of liquid Pb: a Phase proportion as a function of As in lead bullion at T = 900 °C; and b Critical As concentration in lead bullion for the formation of liquid speiss at T = 850 °C–950 °C

Freeze Lining Thickness/Stability

The integrity of the furnace walls with freeze-lining is critically dependent of the stability of the oxide-based freeze-lining. The thickness/stability of freeze lining has been thermodynamically evaluated using the improved database. Formation of accretion/solids from a blast furnace slag was predicted as a function of temperature . Figure 8a shows the proportion of all phases as a function of temperature for slag composition at point 1 (see inserted ternaries). At this condition, the slag solidus temperature was found at approximately 1090 °C below which slag solidifies to melilite, olivine and willemite. Above the solidus temperature the proportion of liquid slag increases while the proportion of solid phases gradually decreases.

../images/468727_1_En_18_Chapter/468727_1_En_18_Fig8_HTML.gif — Fig. 8
Thermodynamic prediction of freeze lining thickness/stability: a Proportion of phases as a function of temperature (slag composition at point 1); b Proportion of liquid slag as a function of temperature ; c and d Slag compositions (point 1, 2 and 3)

Figure 8b shows the proportion of liquid slag as a function of temperature and slag composition. The prediction results show that a slight change in CaO/FeO/ZnO/SiO₂ proportion in the slag can drastically change the proportion of liquid. It can be seen that phase assemblage containing 70% liquid slag + 30% solid phases can be found at higher temperature when moving the slag composition from point 2, point 1 to point 3. This phase proportion (70% liquid slag + 30% solid) is selected as an indicator for comparison. The ability to predict the phase proportion can be used to optimize the freeze lining stability in the furnace .

Blast Furnace Simulation

The database has also been used to predict the conditions in the blast furnace based on the plant data inputs and parameters. The prediction assumed inputs of 1180 t/d of PbO sinter/slag (containing 45.7wt%Pb), silica flux , lime flux , reverts and coke. Oxygen enrichment of 26% was introduced. The simulation outputs are summarized in Table 3.

Table 3

Summary of Blast Furnace simulation outputs

Reactor	Temp [°C]	Product [t/d]
Bullion	1210	465
Liquid slag		500
Solids		11
Reactor gas		1359
Condensor	Temp [°C]	Product [t/d]
Recirculated dust	400	398
Off gas	400	1068

The prediction result gives 512 t/d of bulk slag containing 2.9wt% Pb and lead bullion with 95.1wt% purity. The prediction shows high As concentration in the bullion of 0.55wt% As and high quantity of recirculation dust containing 0.86wt% As. All of these may contribute to the increase of As within the furnace , which can lead to the formation As-containing speiss and the failure of water jackets.

Several possible factors that influence the stability of lead blast furnace water-jacket have been identified from the present work, such as: (i) Hot face temperature of the water jacket; (ii) Freeze-lining formation; and (iii) Speiss formation or speiss attack. Possible measures to prevent failure of the water jacket have been proposed which include: (i) Control of cooling water and blast furnace hearth temperatures; (ii) Adjustment of the blast furnace hearth slag composition to optimize slag liquidus ; and (iii) Control of maximum As intake in the furnace feed. It has been proposed to review and introduce controls and limits of important operation parameters, as well as to conduct laboratory and modeling studies to quantitatively confirm the operational limits.

Conclusion

The stability of lead blast furnace water-jacket has been investigated using coupled microanalysis techniques using SEM and EPMA, phase equilibria analysis and thermodynamic modelling using FactSage [3] with an advanced thermodynamic database . The analysis of water-jacket sample showed the presence of As-containing speiss which lead to aggressive corrosion of the steel jacket. The stabilities of the water-jacket steel wall without and with freeze-lining have been evaluated based on various factors. The present study shows that the advanced thermodynamic database developed by PYROSEARCH for lead smelting can be used to predict the outcomes of the operation and to optimise the operation chemistry.

Acknowledgements

The authors would like to thank Nyrstar for the financial support and permission to publish the present data. The authors acknowledge the support of the AMMRF at the Centre for Microscopy and Microanalysis at the University of Queensland. The authors are grateful to Prof. Peter C. Hayes for valuable discussion during the preparation of this manuscript.

References

1.
Watt W (2015) Nyrstar Port Pirie redevelopment, in MetPlant 2015. Perth, Western Australia, The Australasian Institute of Mining and Metallurgy
2.
Nyrstar (2018) http://www.portpirietransformation.com/
3.
FactSage 7.0 (2015) FactSage thermochemical software. http://www.factsage.com/
4.
Fern JH, Jones RW (1975) The development and operation of a continuous tapping lead blast furnace at Port Pirie. In: Paper presented at the AusIMM annual conference, Melbourne, Victoria
5.
Okamoto H (1991) The As-Fe (arsenic-iron) system. J Phase Equilib 12(4):457–461Crossref
6.
Miodownik AP (1990) Binary alloy phase diagrams, 2nd edn. Materials Information Society, Materials Park, Ohio
7.
Pelton AD, Chartrand P, Eriksson G (2001) The modified quasichemical model. iv—two sublattice quadruplet approximation. Metall Mater Trans A 32A(6):1409–1415Crossref
8.
Pelton AD, Chartrand P (2001) The modified quasichemical model. ii—multicomponent solutions. Metall Mater Trans A 32(6):1355–1360Crossref