Survey of sulphates in process water of LKAB - Kiruna operation

(1)

Survey of sulphates in process water of

LKAB - Kiruna operation

Ebba Videll

Sustainable Process Engineering, master's level 2019

Luleå University of Technology

(2)

i

Acknowledgement

This thesis was made at LKAB R&D department in Kiruna during spring 2019. The project was carried out as a degree project on master’s level within the programme Sustainable Process Engineering at Luleå University of Technology.

I would like to thank the employees at the LKAB R&D and Kvalitetservice departments for making my thesis work possible and for having the best time. I would like to thank my supervisors from LKAB: Mattias Ylipää and Elsa Peinerud for supervision and for sharing knowledge. Thank you!

Thank you Kari Niiranen and Viktoria Töyrä for taking time for all my questions and thoughts throughout my thesis work! Finally, I would like to thank my supervisor from LTU, Jan Rosenkranz, for helping me forming the thesis to its best version and for valuable advices. Thank you!

(3)

ii

Abstract

Sulphate-rich wastewater is an increasing concern for industries as LKAB. The water chemistry of the LKAB Kiruna water system is characterized by high alkalinity, high pH (pH 7.5-9.0) and high concentrations of chemical species and soluble minerals. The sulphate content in the water system of Kiruna is associated with the dissolution of calcium sulphate as anhydrite (CaSO4)

and gypsum (CaSO4∙2H2O). However, the high concentrations of sulphate in the effluents from

the LKAB Kiruna operation are unique for iron ore mining.

The aim of the thesis was to study and evaluate the behavior of sulphate in the process water system of the Kiruna concentrator plants. This was done by laboratory grinding (leaching tests), equilibrium calculations with the HSC software and mass balancing of the concentrator KA3 in Kiruna. The highest concentrations of sulphate have been detected in the process water of KA3, hence the focus has been on KA3 regarding sampling, evaluation and comparison. Water treatment technologies for sulphate and the effect of process water on ore processing have not been included in this project.

The laboratory grinding was done using process water and ore from the concentrator KA3. During the experimental work with laboratory grinding the parameters pH, temperature and operating times for primary- and secondary grinding, respectively, were varied. The observations from the experimental work were further confirmed by equilibrium calculations and mass balancing. The following conclusions were drawn.

▪ Anhydrite/gypsum is not leached from the ore during ore processing with process water having concentrations of sulphate already close to the saturation point at approximately 1800 mg/L. If the process water is diluted with water with lower sulphate content, e.g. mine water or a diluted return water from the pond system (e.g. during spring flood), anhydrite/gypsum in ore will be dissolved until the sulphate concentration reaches the saturation point.

▪ The ionic strength of the process water controls the saturation point and thus the sulphate concentration.

▪ Leaching of anhydrite/gypsum in saturated process water, with respect to sulphate, is not affected by grinding time or adjusted conditions in the process water, such as temperature or pH value.

For further work, it is recommended to investigate the behavior of sulphate in the tailings pond system to increase knowledge of the overall behavior in the water system. In addition, a similar investigation of the behavior of uranium in the process water is recommended in order to face future environmental standards.

(4)

iii

Sammanfattning

Sulfat betraktas som ett miljöfokusämne tillsammans med uran och nitratkväve i LKAB:s vattensystem. Med miljöfokusämne menas ett ämne som kan ha effekt på recipient, och dessa bör därför utredas med hänsyn till skyddsåtgärder samt miljökonsekvenser. Av LKAB:s verksamhetsorter tillhör Kiruna den orten med högsta sulfatnivåerna i bräddvattnet till recipient. Vattenkemin i vattensystemet kan beskrivas med högt pH (7,5–9,0), hög alkalinitet samt höga koncentrationer av lösta ämnen. Höga sulfathalter i utsläpp från gruvverksamheter är vanligtvis kopplade till surt och metallrikt lakvatten, så kallat Acid Mine Drainage (AMD), från brytning och förädling av sulfidiska malmer. De uppmätta sulfatnivåerna i vattnet från LKAB:s verksamhet i Kiruna är därför unikt för förädling av järnmalm. Sulfatnivåerna i Kirunas vattensystem är främst kopplat till upplösning av kalciumsulfater, såsom anhydrit (CaSO4) och gips (CaSO4∙2H2O).

Syftet med projektet var att utreda sulfats beteende i processvattensystemet i Kiruna, med fokus på anrikningsverk KA3 där de högsta sulfatnivåerna uppmätts. Genom labbmalningar, jämviktsberäkningar samt en massbalans av strömmar i KA3 kunde sulfats beteende i processvatten kartläggas. Vattenrening samt vattenkvalitetens påverkan på processen har inte inkluderats i detta arbete.

Vid labbmalningarna användes processvatten och ingående rågods från den verkliga processen. Under försöken varierades pH, temperatur samt maltider för primär- och sekundärmalning. Resultatet från labbmalningarna visade oförändrade sulfathalter vid malning med processvatten. Observationerna från försöken kunde senare bekräftas med resultat från jämviktsberäkningarna och massbalansen. Arbetet resulterade i följande slutsatser.

▪ Anhydrit/gips lakas inte ut i processvatten med en sulfathalt nära mättnadsnivå (1800 mg/L). Om processvattnet exempelvis späds med smältvatten från vårfloden som innehåller en lägre sulfathalt, kommer anhydrit/gips lakas ut tills sulfathalterna återigen når mättnadsnivå. ▪ Processvattnets jonstyrka kontrollerar mättnadsnivån i systemet och därmed också

sulfatkoncentrationen.

▪ Lakning av anhydrit/gips påverkas inte av maltider, pH eller temperatur vid malmförädling med processvatten.

För fortsatt arbete rekommenderas att utreda sulfats beteende i dammsystemet för att få en överblick av sulfats beteende i hela vattensystemet. För att kunna möta framtida utsläppskrav på miljöfokusämnen rekommenderas att genomföra en liknande kartläggning för uran i vattensystemet.

(5)

iv

Nomenclature

Abbreviations Full notation

AMD Acid Mine Drainage

KA 1, 2 and 3 Kiruna concentrator 1, 2 and 3 KK 2, 3 and 4 Kiruna Pelletizing plant 2, 3 and 4

KP4 Pump station 4, water from clarification pond KP50 Pump station 50, return water from gas cooling

KP57 Pump station 57, mine water

KREC Kjeöy Research & Education Center

PC Principal Components

PCA Principal Component analysis

QEMSCAN Quantitative Evaluation of Materials by Scanning Electron Microscopy

SEM-EDS Scanning Electron Microscopy with Energy Dispersive Spectrometer

TDS Total Dissolved Solids

WLIMS Wet Low Intensity Magnetic separation

XRF X-ray Fluorescence

Minerals Chemical formula (webminerals.com)

Anhydrite CaSO4

Apatite Ca5(PO4)3F

Calcite CaCO3

Gypsum CaSO4∙2H2O

Magnetite FeO∙Fe2O3

Pyrite FeS2

Thorite ThSiO4

(8)

vii

Definitions

D10 10 % have a diameter less than D10 (µm) D50 50 % have a diameter less than D50 (µm) D80 80 % have a diameter less than D80 (µm) D90 90 % have a diameter less than D90 (µm) HSC chemistry software Chemical reaction and equilibrium software Malvern master-sizer Laser diffraction size analyzer

MODDE software Design of Experiments

Process water Water in processing plants; concentrators, sorting- and pelletizing plants

Q2X Goodness of prediction, for which a value of 0.9 is excellent and 0.5 is good

QEMSCAN Mineralogical investigation using automated mineralogy R2X Goodness of fit, for which a value of 1 is perfectly fitting model

and 0 is no fit

Roll-bottle tests Leaching test using rotating bottles for agitation SA45 LKAB standard sieve analysis, from 45 µm to 5.6 mm SEM-EDS Automated particle analysis with energy dispersive

spectrometer (EDS)

Sequential leaching (ALS) Five-stage leaching with increasing strength of leachate SIMCA software Multivariate tool for data analysis

(9)

1

1. Introduction

The project was carried out as a degree project on master’s level within the programme Sustainable Process Engineering, with supervisors from LKAB, Mattias Ylipää, and LTU, Jan Rosenkranz. The thesis includes three main sections, namely experimental work, equilibrium calculations and mass balancing. All sections are presented with separate methods and results to simplify reading. Anyway, a general discussion, conclusions and suggestions for further work are presented in the end of the thesis.

1.1 Background

Luossavaara-Kiirunavaara AB (LKAB) is a mining and mineral company producing high-quality iron ore products in the northern part of Sweden. The iron ore products are mined, sorted, concentrated and pelletized at the three operating mine sites Kiruna, Svappavaara and Malmberget. The Kirunavaara deposit is an apatite-magnetite ore, hence ore processing includes a sequence of two-stage grinding, concentration by WLIMS (Wet Low Intensity Magnetic Separation) and reversed flotation for apatite removal. The main objective with ore processing is to liberate and separate apatite and silica (silicate minerals) as far as possible. [1]

The Kiruna ore, in comparison with Svappavaara and Malmberget ores, shows a higher dissolution of chemical species, particularly higher amounts of sulphate. Sulphate is associated to gypsum (CaSO4∙2H2O) and anhydrite (CaSO4), which appear in the Kiruna ore body in low

quantities. Despite the low concentration of gypsum/anhydrite in the ore, its high solubility results in calcium and sulphate ions being the main dissolved components in the water system of Kiruna. The Swedish government has not yet established threshold values for sulphate in the environment. However, sulphate in effluents is an increasing concern for mine industries, hence it is an interesting compound to study and evaluate. [2][3]

In general, concentrations of chemical species (e.g. sulphate, calcium) are increasing with increasing production and recirculation of water. Large quantities of water are consumed and recycled in LKAB’s concentrators and pelletizing plants for ore processing. By increasing the recirculation of process water, the freshwater demand can be reduced as well as the volume of mine effluents to the recipient waters. Hence, recirculation is crucial for both economic and environmental factors. However, increasing recirculation may result in accumulating dissolved compounds, affecting the water chemistry and mineral processing (e.g. flotation) negatively. [4][5]

(10)

2 depending on season, with more frequent sampling during summer and less frequent during winter. In addition, the mine water has been analyzed during the last 20 years. In Figure 1 historical values of sulphate, chloride and calcium in the process water of the KA2 concentrator are presented. Historically, the levels of sulphate have been approximately 500 mg/L, compared with today’s levels of approximately 1500 mg/L in the process water. Thus, the sulphate concentrations have increased about 1000 mg/L over the last 20 years of mining and ore processing. [2][6]

Figure 1. Historical concentrations of chloride, sulphate and calcium in process water of KA2. Source: [6]

This thesis project has focused on concentrator plant KA3 regarding evaluation, comparison and sampling. The production of KA3 started in 2008, hence process water has only been sampled for 10 years (with start in 2009). In Figure 2 the historical concentrations of sulphate, calcium and chloride are presented. At the start in 2009, the sulphate concentrations were approximately 1000 mg/L. Since then, the sulphate concentrations have increased to 2000 mg/L. [6]

(11)

3

1.2 Objective and scope

The objective was to study the sulphate concentrations in process water and to increase understanding of the behavior during ore processing in the concentrator plants. A project plan including the tasks literature review, experimental work, data evaluation and equilibrium calculations was defined with respect to the thesis requirements and the given objectives. The following questions are summarizing the objective:

▪ What are the amounts of dissolved sulphate in process water? ▪ How does ionic strength affect leaching of gypsum/anhydrite?

▪ Does sulphate settle within the water system and what is the amount of settled sulphate?

(12)

4

2. Description of the LKAB Kiruna operation

To increase the understanding of the water quality of the process water, the mineralogy, seasonal variations and ore beneficiation methods must be described. Accordingly, this chapter covers descriptions of the mineralogy, the processes and the water system of LKAB Kiruna operation.

2.1 Mineralogy of the Kirunavaara deposit

The Kirunavaara deposit is a sheet-like iron ore body approximately 4 to 4.5 kilometers long and 50 to 100 meters thick, with a maximum thickness of 200 meters in the northern part. Today the ore body is well known down to a depth of 1365 meters below the surface. In Kiruna, LKAB is operating an underground mine using sublevel caving as the mining method. The Kirunavaara deposit is a high-grade iron ore deposit consisting of iron oxides (magnetite and hematite), apatite and varying amounts of different gangue minerals. The type of mineralization is known as “Kiruna type” and occurs in several deposits in the northern part of Sweden, of which the Kirunavaara deposit is the largest and the best-known example. The deposit is mainly consisting of magnetite and apatite, with an average grade of 63.8 % iron and 0.4 % phosphorous. Today, magnetite is the mineral of economic value in the deposit. Besides apatite the following gangue minerals are occurring: actinolite, phlogopite, titanite, ilmenite, rutile, quartz, talc and albite. In addition, carbonates (calcite, Fe-dolomite and ankerite), sulphide minerals (pyrite and chalcopyrite) and calcium sulphates (anhydrite and gypsum) are occurring in fewer quantities. [1]

Further, the apatite-magnetite ore is divided into two main types; low phosphorous ore and high phosphorous ore. Before 2009, the different ore types were mined separately, but with increasing production of crude ore the different ore types are mixed to one type of crude ore for further processing. [1]

Leaching of constituents in process water are stated to be dependent of the mineralogy of ore, i.e. there is a relationship between mineralogy and water quality. The impact of the ore on the process water quality depends on several parameters: the composition of ore, the particle sizes and the quality of the initial water. The Kiruna ore body is mainly consisting of oxide and silicate minerals, which have low leaching degree and low impact on the quality of process water. Thus, the main constituents in process water are originating from minerals and salt-inclusions occurring in fewer quantities in the ore body (e.g. carbonates and calcium sulphates). As carbonates and calcium sulphates occur in low quantities, it is difficult to detect the minerals with mineralogical investigations using automated mineralogy (QEMSCAN). [7]

2.2 Process description

(13)

5 magnetic separation. The final product from the sorting plant has already an iron content of approximately 62 %. The sorted product is further processed in the concentrator plant with the aim to reduce particle size and the content of impurities as well as to increase the iron content in concentrate. The concentrators are using large quantities of water and most of the water is recirculated within the plants. Concentration of iron ore includes grinding, magnetic separation, flotation, addition of additives and filtration. Impurities such as silicon, sodium, potassium and phosphorous are separated in order to increase the iron content in concentrate and to reduce the amount of elements detrimental to downstream metallurgical processing [8]. The flowsheet of the concentrator KA3 is presented in Figure 3.

Figure 3. Flowsheet of KA3 (Mikael Olfosson, personal communication 2019)

The last stage of mineral processing includes addition of binders, such as the clay mineral bentonite. The iron ore concentrate forms pellets in rotating drums. The pellets are then dried, pre-heated and sintered for increasing strength. The pelletizing plants in Kiruna are grate-kiln plants, in which pellets are dried and pre-heated in a grate, sintered in a rotating kiln and cooled in a rotating cooler. A schematic image of the ore processing is illustrated in Figure 4. Eventually, the high-quality pellets are transported by train to the ports in Narvik and Luleå, and from there shipped by boat to customers around the world. [8]

(14)

6

2.3 Water system

To provide the processing plants with water, the LKAB water system is divided into inner and outer water circuits (Figure 5). The inner water circuit includes the processing plants and the water is referred to as process water. The outer water circuit also includes mine water and water from the pond system (clarification pond and tailings dam). The water chemistry varies between the different circuits and is differently affected by e.g. seasonal variations. The water chemistry is mainly dependent on dissolution of minerals in the ore and host rock, unexploded explosives, additives and chemical processes occurring in the water system. [2]

Figure 5. The LKAB water system – Kiruna (Rikard Söderström, personal communication 2019)

The main sources for mine water are drainage water from the deformation zones and ground water. However, water from the clarification pond includes water from the tailings dam (mainly process water), seepage water from Lake Luossajärvi and precipitation. Water is leaving the system by evaporation, seepage from the pond system and as absorbed pore water in the tailings dam. The excess water of the water system is discharged to the nearby recipient, the Rakkuri water system. The water chemistry in the outer water circuit is strongly affected by seasonal variations. The water additions are the largest during spring and summer, significantly diluting the water system. During winter, dilution is the lowest due to reduced precipitation. Thus, the highest concentrations of chemical species are attained during winter. Additionally, temperature, pH and alkalinity are parameters affected by seasonal variations in the outer water system. [2]

(15)

7 high temperatures and decrease sludge content. The process water is recirculated using thickeners, where the sludge is dewatered and transported to the tailings dam. The highest concentrations of chemical species and dissolved minerals are found in the inner water circuit, which are mainly due to high recirculation. The highest sulphate levels are obtained in the process water, with levels of approximately 1800 mg/L. [2]

The mine water system of Kiruna operation has increased and been re-built with expanded mining. The mine water system consists of several pump stations and basins for desludging. Figure 6 illustrates the mine water system including the water flows. The water system of the Kiruna mine is necessary in order to decrease the groundwater level and to provide water for drilling. Further, mine water is pumped up through the system with a difference in altitude of approximately 1500 meters, until it reaches ground level and the storage tanks. The last pump station (KP57) in the chain of pump stations, has the highest flow and represents the average mine water chemistry. [2]

Figure 6. Water system in underground mine of LKAB Kiruna. Source: [2]

The mine water is a major source of uranium, sulphate and calcium. Concentrations of calcium and sulphate are increasing with dissolution of gypsum and/or anhydrite associated with the ore. The sulphate concentrations in the mine water are varying with season, even though parameters such as temperature are rather stable during the year in the mine, i.e. at 14-15 °C. [6]

(16)

8 dam. Also pumped water from Lake Luossajärvi and precipitation end up in the tailings dam. The water from the tailings dam is further transported to the clarification pond. Clarification ponds are the common water treatment method utilized in mine industries. The main objective of the clarification ponds is to settle suspended particles to the bottom of the pond. The overflow is recirculated to the processing plants and the excess water is discharged to the recipients. [2][9]

Alkalinity and pH are varying with season in the clarification pond, with the lowest values during winter. Before discharge of the water to the recipient, the water passes a pH-adjusting plant. The purpose with the pH-adjusting plant is to decrease pH during periods with high pH, in order to reduce the ammonia concentration in the effluents. [2]

The chemistry of the water from the pond system is controlled by several processes. In a previous study by Lundkvist and Walder [6], the following processes were mentioned:

▪ Dissolution of salt minerals/saline fluid inclusions ▪ Mineral dissolution/precipitation

▪ Freezing and melting ▪ Dilution with mine water

▪ Dilution with precipitation, snowmelt and run-off water within the catchment area ▪ Long-term mineral weathering

Lundkvist [10] noted an undersaturation of calcite in the pond system during winter. This undersaturation can be explained by temperature and the dynamics of the carbon system. During winter the temperature in pond water decreases, hence solubility of calcite increases. As a result, the pH, alkalinity and calcium concentrations would increase. However, the pH in the pond system of Kiruna decreased during winter, which was explained by build-up of CO2 pressure.

[10]

2.4 Variations of sulphate in the water system

The main sources of sulphate are the concentrators and the mine site. In previous studies [2][6] the concentrations of the main species, sulphate and calcium, are predicted to increase in the water system and in a few years the levels of sulphate and calcium will reach a maximum, which will be controlled by gypsum solubility in the range of 2000-2200 mg/L. The current gypsum solubility and reaction rate in the process water are close to equilibrium, hence concentrations of sulphate and calcium are limited in the process water. However, the solubility of calcite and gypsum increases with increasing total dissolved solids (TDS). Total dissolved solids (TDS) can be estimated from electric conductivity or calculations of the sum of constituents in water. Determination of TDS using electric conductivity can be underestimated with high concentrations of dissolved species. [6]

Sulphate is considered as a complex compound since the concentrations are dependent on the dilution and precipitation of gypsum and calcite (CaCO3). The concentrations of sulphate in the

(17)

9 and the lowest concentrations during spring flood at 800 mg/L, when the water system is diluted by melt water [11].

During the spring flood of 2015, the dilution factor was approximately 3.3, which should result in decreasing sulphate concentrations, from 1500 mg/L to 450 mg/L. However, the sulphate concentrations were reaching a minimum of 800 mg/L, which can be as result of dissolution of anhydrite/gypsum from sediments and/or fine particles from process water. The minimum sulphate concentrations are present for a short period of time and are increasing rapidly, which indicates sources of sulphate from other than process water and mine water. However, the concentrations during summer are lower than during winter season, with concentrations of approximately 1200 mg/L, due to dilution. Thus, the sulphate concentrations in the pond system are correlated to precipitation. [11]

The calcium to sulphate weight ratio for gypsum dissolution is 0.41. The corresponding ratio in mine water was determined to 0.41 and 0.37 in the clarification pond. The results indicated an excess of sulphate in the system, which can be explained by different processes either by oxidation of sulphide minerals or calcium being controlled by calcite precipitation. In the colder parts of the water system (tailings dam and clarification pond), build-up of calcium and sulphate occur. When the water temperature increases, the calcite solubility decreases, and thus calcite may precipitate. The alkalinity is also controlled by calcite, which acts as a buffer for both alkalinity and pH. Increasing gypsum dissolution may result in increased calcite precipitation, hence a reduction in alkalinity. The calcium and sulphate concentrations do not increase stoichiometrically in the clarification pond, which indicates calcite precipitation in the process water and slow reduction of alkalinity. [6]

In process water, where high concentrations of different ions are present, it is difficult to determine the solubility of gypsum. Calcite/aragonite and gypsum/anhydrite are the minerals with the largest potential to precipitate in water. In order to determine whether a system is in equilibrium with chemical species or not, the saturation index, S.I. or Langelier index, L.I, can be used. The saturation index is defined as the difference between measured pH and the hypothetical pH reached at equilibrium (pHs), given in equation 1. If the saturation index is

zero, the system is in equilibrium with the chemical species. However, if the saturation index is negative the system is undersaturated and if the S.I. is positive the system is oversaturated, which will result in precipitation of a solid compound. [11][6][12]

L.I = pH – pHs (1)

(18)

10 Figure 7. Saturation indices for Kiruna water system. Source: [6]

According to the literature [12] the negative indices indicate undersaturation with respect to gypsum/anhydrite, which would result in dissolution of gypsum/anhydrite. In addition, the saturation indices for calcite in the process water were determined. The indices for calcite were rather stable in the process water at 0.5, which indicate precipitation of calcite. For precipitation of calcite, a saturation index of 0.2-0.3 is required. According to the conditions in process water gypsum/anhydrite dissolves in process water and calcite, which is supersaturated, precipitates. As a result of dissolving gypsum and precipitating calcite, the sulphate content in water increases. Increasing temperature results in increasing solubility of gypsum but decreasing solubility of calcite. In the plants, where the water temperature is high (approximately 30 °C) the solubility of calcite and gypsum is affected. According to the study a maximum concentration with respect to calcium, but not sulphate, may be obtained in the process water. [6]

2.5 Other constituents in the water system

The major elements of the LKAB water system include chloride, magnesium, sodium, potassium, calcium, nitrate and sulphate. Examples of trace elements are strontium, barium, copper, uranium and zinc. Chloride, magnesium, sodium and potassium show seasonal variations with high concentrations during winter and early spring, and lower concentrations during late spring and early summer. Historically, the trace element concentrations have decreased rather than increased in the process water. These elements are rather controlled by long-term mineral weathering process, dilution processes and sorption/precipitation processes. [6]

(19)

11 chemical properties as calcium and substitutes for calcium in minerals like gypsum/anhydrite, calcite and Ca-plagioclase. Significant strontium concentrations have been observed in anhydrite. Hence anhydrite is a potential source for strontium. In addition, strontium is relatively stable in the water system, with a low degree of sorption. As for strontium, gypsum is a probable source of barium. Other processes that are potentially controlling the concentrations of trace elements in the water system are carbonate precipitation, sorption on iron oxides as well as precipitation of metal oxides. [6]

Concerns of nitrate and uranium in wastewater have increased; hence, LKAB has environmental focus on those species too. The main sources of nitrate are unexploded explosives (an ammonium-nitrate compound) and residuals from explosives in mine water. Seasonal variations of nitrate are observed in the pond system, which can be explained by dilution and biodegradation during summer. [2]

(20)

12

3. Previous investigations

Sulphate-rich effluents from iron ore mining are unique for the Kiruna operation. The highest sulphate levels are detected in wastewater associated to acid mine drainage (AMD) and mining of sulphide ores as well as coal. This chapter presents previous investigations of sulphate-rich effluents from industrial activities and from the LKAB mine sites.

3.1 External studies

Industrial activities such as mining, metallurgical processes, power plants and papermaking generate sulphate-rich wastewater contaminating the surface and ground water. High sulphate levels in effluents are mainly related to AMD from sulphide ore mining. AMD is generated from oxidation of sulphides, e.g. pyrite, and can occur both naturally by rock weathering processes and artificially from industrial activities. The sulphide ore is chemically stable in saturated environments without contact of oxygen and water. Once the ore is exposed to moisture and oxygen, AMD occurs. Sulphide oxidation lowers the pH, by formation of sulfuric acid, and increases sulphate concentrations as well as the content of heavy metals. The acidic and metal-contaminated effluents are typically treated by lime neutralization in order to remove metals and increase pH. However, the sulphate levels are not affected by lime neutralization. Instead, remediation technologies can be utilized to decrease the sulphate levels. [13][14][15]

Contaminated wastewater is generated from active, closed as well as abandoned mine sites. The water quality of effluents varies between different mine sites. The water quality depends on ore composition, extraction, enrichment methods and climate conditions. The different effluents can be classified according to pH; AMD with pH < 6, neutral mine drainage with pH higher than 6 but lower than 9 and alkaline mine drainage with pH > 9. Sulphur-rich residues, e.g. pyrite, from coal mining contributes to AMD with high concentrations of sulphate contaminating the environment. Thus, both active and historical mine sites for coal requires remediation technologies in order to decrease sulphate and metal levels as well as to increase the pH in the wastewater. [15][16][17]

High levels of sulphate in industrial water systems are concerning the industries from an environmental, technical and economic point of view. The sulphate-rich wastewater increases corrosion of equipment and may induce environmental impacts [14]. Previous study [18] assessed the impacts on organisms such as macroinvertebrates in the freshwater systems downstream a mine site in China. However, the mechanism on how as well as the ecological risk are still unknown.

(21)

13 sustainable alternative of remediation. However, the selection of remediation technology is site-specific and must account for parameters including the water quality of the effluent, the flowrate as well as economic, environmental and regulatory factors. Examples of tested remediation technologies for sulphate removal includes electrocoagulation, ultrafiltration and biological sulphate reduction with bacteria. [19][9][15][13]

Lake Jormasjärvi, located in central Finland, is an example of a recipient affected by sulphate-rich mine water. At the southern end of Lake Jormasjärvi the Terrafame Talvivaara Ni-Zn-Co-Cu mine is located. The Terrafame Talvivaara mine has, thus, altered the water chemistry of the downstream recipients with wastewater mainly containing sulphate, sodium and metals. The wastewater of the Terrafame Talviaara was planned to be purified chemically and further channeled through a settling pond and wetland treatments before discharge to the recipients. Although the wastewater treatment has not operated as planned. The recipients downstream the mine has been affected by unintentional leaks since the start, and during 2012 a dam accident occurred, where sulphate-rich and metal-containing wastewater polluted the environment. The environmental changes of uncontrolled mine effluents have been evaluated in previous studies. The results indicate altered lake ecosystems due to mining activities. The mine effluents were also considered to induce rapid and large-scale changes of recipients, even far downstream the chain of lakes. [20][21]

3.2 Internal studies of LKAB

Previous experiments have been performed at LKAB with the aim to increase the understanding of mineral surfaces and to determine the leaching properties of the ore. These experiments as well as previous studies have been used for designing the experimental work included in this thesis.

Gustafsson [7] evaluated the leaching properties of ores from Kiruna, Malmberget and Svappavaara. The study was using mineralogical studies and process water analysis in order to determine the relationship between mineralogy and process water chemistry. Process water in the grinding circuit of KA1 was sampled in order to evaluate the leaching of sulphate during primary grinding. In addition, the mineralogy of the Leveäniemi and Malmberget ores was investigated with respect to the leaching properties. Mineralogical analysis was performed using QEMSCAN in order to determine and evaluate the potential leaching properties of the ore types. [7]

(22)

14 Figure 8. Sulphate concentrations in process water of KA1. Data taken from [7]

In addition, several leaching tests have been performed with LKAB ores. During 2013, the KREC center [22] performed a bottle roll leach test with the aim to determine the leaching properties of Kiruna and Svappavaara ores, respectively. Each bottle was filled with 300 g solid material and 2000 ml of distilled water. The ores were crushed and ground to a D80 varying between 75 µm and 150 µm before leaching. Further, the bottles were continuously rolled throughout the experiments for four days. The results from the tests indicated higher leaching ratios for most constituents from the Kiruna ore in comparison with Svappavaara ores. Sulphate and calcium were the constituents of the highest concentrations after leaching, and the highest concentrations were obtained from leaching of Kiruna ore. [22]

Further, leaching factors for constituents were determined based on the leaching tests and studies of historical development of process water. The results from leaching tests with de-ionized water indicated higher leaching factors in comparison with calculated leaching factors, especially for sulphate and calcium. The deviant leaching factors for sulphate and calcium can be explained by the high solid ratios used in the leaching tests; hence, the water-to-solid ratio was higher during the tests than in the concentrators. As the leaching of gypsum/anhydrite is limited in process water with higher initial concentrations, the leaching factors for sulphate and calcium were increasing when using de-ionized water [22]. Thus, the drawbacks with the tests was the high water-to-solid ratios and the usage of de-ionized water as leachate.

(23)

(24)

16

4. Experimental work – Materials and methods

In this chapter, the experimental work of the thesis is presented. The methodology was determined based on previous comminution tests [1] and leaching tests [23][22]. Further, the methodology was modified in order to reduce errors and losses of sulphate. The aim of the experimental work was to simulate leaching of sulphate in the grinding circuit of the concentrator; hence, process water was used as solvent.

4.1 Design of experiments

The leaching tests were designed using MODDE 12.1 software [24]. A full factorial design at two levels was set with four factors and three center-points. The levels were determined with the aim to simulate real processes in the grinding circuits at LKAB concentrators. The main differences, compared to previous experiments, were the usage of laboratory mills and process water as well as a lower liquid-to-solid ratio (L/S). In Table 1 the determined factors, their levels and standard conditions (center-points) are summarized. The standard conditions for the experiment were set at 25 °C, pH 8, 9 minutes primary grinding and 30 minutes secondary grinding.

Table 1. Design of experiments - parameters and levels

Abbreviations Factors Levels

(Low/High)

Centre-points

pH pH 7/9 8

Temp. Temperature 10 °C/40 °C 25°C

Time prim. Operating time (Primary grinding) 6 min/12 min 9 min Time sec. Operating time (Secondary grinding) 25 min/35 min 30 min

Relevant responses for the experiment were set to chloride, sulphate and uranium concentrations in process water. The responses were fitted according to equation 2 [25].

Response = constant + pH + temp. + time prim. + time sec. + pH×temp. + pH×time prim. + pH×time sec. + temp.×time prim. + temp.×time sec. + time prim.×time sec. + ε, (2)

where ε is the residual response variation not explained by the model, ε ∊ N (0, σ). Although the responses were not used for further data evaluation.

(25)

17 Table 2. Experimental plan

Sample number Temperature

[°C] pH Operating time [min] (Primary grinding) Operating time [min] (Secondary grinding) 1 (Ore 2018) 25 8 9 30 2 10 7 6 25 3 40 7 6 25 4 10 9 6 25 5 40 9 6 25 6 10 7 12 25 7 40 7 12 25 8 10 9 12 25 9 40 9 12 25 10 10 7 6 35 11 40 7 6 35 12 10 9 6 35 13 40 9 6 35 14 10 7 12 35 15 40 7 12 35 16 10 9 12 35 17 40 9 12 35 18 25 8 9 30 19 25 8 9 30 20 25 8 9 30

21 (50/50, Process water/de-ionized water) 25 8 9 30

22 (100 % de-ionized water) 25 8 9 30

4.2 Sampling of material and sample preparation

(26)

18 Figure 9. Sampling of ore feed

The sampled ore feed had a moisture content of 1.35 % and was dried before splitting at 35 °C for 4 days. The dried ore feed was further homogenized and split in a rotary splitter to samples of 2 kg. The samples were crushed in a closed circuit using a laboratory jaw crusher and a screen with mesh size of 3 mm. Coarse material (+3 mm) was re-crushed to achieve a size fraction finer than 3 mm. However, 3.3 % of the material was coarser than 3 mm (-4.0 to +2.8 mm) after crushing. The D50 of the ore feed to primary mill was 400 µm and the size distribution is presented in Figure 10.

(27)

19 In addition, 100 liters of process water were sampled from the overflow of the thickener of KA3 before the experiments. The sampled water was stored in closed water containers at room temperature.

4.3 Leaching tests with two-stage grinding

The leaching tests were performed with two-stage grinding in laboratory tumbling mills. The grinding test method is defined in a previous study [1] and has been used for the comminution tests. The samples were wet ground in a rod mill and ball mill, respectively. Technical data for the laboratory tumbling mills is presented in Table 3.

Table 3. Technical data for laboratory tumbling mills. Source: [1]

Primary grinding (rod mill) Secondary grinding (ball mill)

Mill inner diameter Mill inner length

200 mm 250 mm 200 mm 250 mm Grinding media Number Weight Steel rods 53 14 200 g ± 50 g Steel balls, Ø 15 mm - 13 100 g ± 50 g Degree of filling Rotation speed 67 wt. % 65 rpm 67 wt. % 65 rpm

The weight of the grinding media was checked before running the mills. If the weight was differing by 50 g, balls or rods were added. Detailed data of the steel rods used for primary grinding is presented in Table 4.

Table 4. Data for steel rods. Source: [1]

Number of rods Diameter [mm] Length [mm]

3 25.4 240 4 22.0 240 4 19.0 240 5 16.0 240 6 12.5 240 8 10.0 240 10 8.0 240 13 6.0 240

(28)

20 material were added to the ball mill with 13 100 g (± 50 g) grinding media. The operating time for the ball mill varied between 25 and 35 minutes. The slurry from the secondary grinding was filtrated in the pressure filter and process water was sampled for water quality analysis. Finally, the filter cake was dried at 105 °C for approximately 24 hours (Figure 11).

Figure 11. Experimental work - ore feed to filter cake

The dried samples were split for chemical- and particle size analysis. Figure 12 illustrates the approach and the analyzed streams.

Figure 12. Experimental setup and analyzed streams

4.4 Analytical methods – Theory

This section describes the analytical methods: QEMSCAN, laser diffraction size analyzer (Malvern master-sizer) and sequential leaching. Other analytical techniques as XRF, SEM, water quality analysis and sieve analysis (SA45) are explained in the Definitions section.

(29)

21 the identification of several mineral types, which are color coded for easy visual inspection. [26]

Malvern master-sizer is a laser diffraction instrument. The main principle of laser diffraction size analysis is that light scattering is utilized to calculate the particle size distribution based on the light distribution pattern, that is the result of finer particles inducing more scatter than coarser particles. The advantages of laser diffraction instruments are the easy and fast usage, giving reproducible results. In mineral processing, laser diffraction size distributions tend to appear coarser than other methods, e.g. sieve analysis. Thus, laser diffraction size analyzers should be used with caution. [26]

The sequential leaching test used by ALS Scandinavia is a five-stage leaching method with increasing strengths of the acids, used as leachates. The leachates from each leaching stage are analyzed. The main objective with the method is to determine the leachability of the solid material, e.g. soils or ores [27]. Table 5 presents the five stages and their leachates.

Table 5. Five-stage sequential leaching performed by ALS Scandinavia. Source: [27] Leaching

stages

Form Solvent

1 Adsorbed and exchangeable metals and carbonates

1.0 M Na-acetate buffer (pH 5) (25 °C)

2 Labile organics 0.1 M Na-pyrophosphate (25 °C)

3 Amorphous

Fe/Mn-oxides

0.25 M NH2OH∙HCl in 0.1 M HCl

(at 60 °C)

4 Crystalline Fe-oxides 1.0 M NH2OH∙HCl in 25 % acetic acid

(at 90 °C)

5 Stabile organics and sulfides K-chlorate in 12 M HCl and 4 M HNO3

(at 90 °C)

4.5 Analytical methods – Experimental work

(30)

22 Table 6. Samples from experimental work and analyses

Sieving analysis (SA45) Malvern analysis (dry) Chemical analysis (XRF) Water quality analysis

Initial ore feed (1 sample) X X

Initial process water (1 sample)

X

Process water primary grinding (22 samples)

X

Process water secondary grinding (22 samples)

X

Filter cake secondary grinding (22 samples)

X X

Flush water (3 samples) X

Parameters such as conductivity, alkalinity and pH were analyzed at the LKAB environmental laboratory. Anions, cations, metals and uranium were analyzed by the ALS laboratory in Luleå. The measurement uncertainty for sulphate was 15 %. In addition, two different methods for particle size analyses were used, the Malvern master-sizer and the SA45 LKAB standard sieve analysis. Laser diffraction size analysis using the Malvern master-sizer was done at the LKAB physical laboratory in Kiruna. The Malvern master-size analyzer was using dry dispersion measurements. SA45 was used for the ore feed, for which particle sizes between 5.6 mm and 45 µm were analyzed.

In order to determine the anhydrite/gypsum content in the solid samples, QEMSCAN analyses as well as sequential leaching were used. Two samples were sent for each analysis, the initial ore feed and sample 12 after secondary grinding. The aim with the analyses was to characterize and determine the amount of leached anhydrite/gypsum during the experimental work. QEMSCAN analysis was performed at LKAB in Malmberget and the sequential leaching method was done at ALS Scandinavia in Luleå.

(31)

23

4.6 Multivariate data analysis

(32)

24

5. Evaluation of results from the experimental work – Water samples

The evaluation of the results from the water quality analysis was divided into different parts: data analysis using SIMCA 15 software and data evaluation of variations of concentrations, alkalinity, pH and conductivity. In this chapter the results from water quality analysis are presented and discussed. In addition, errors related to flush water are evaluated and a comparison with data from the concentrator is presented.

5.1 Multivariate data analysis of results from the water quality analysis

The results from water quality analysis included several variables; hence, data analysis was made with SIMCA 15 software to simplify evaluation of correlations. The water samples are described as x.1 and x.2, for which x.1 presents water samples from primary grinding and x.2 presents water samples from secondary grinding. Samples 1-20 were ground with 100 % process water at varying conditions (i.e. operating time, pH and temperature), which are presented in the experimental plan (Table 2). Samples 21 and 22 were ground with 50 % and 100 % de-ionized water, respectively. The flush water and initial water samples were not included in the data analysis.

Three different models were set: model 1 for the full dataset, model 2 for the samples from the primary grinding and model 3 for the samples from the secondary grinding. The statistic overview of the full dataset (compare Appendix 1) indicated high min/max quotient (>0.1) and low skewness (<2.0). In the project table (Table 7) the goodness of fit (R2X) and goodness of prediction (Q2X) are presented, as well as the different models, their variables and number of significant PC (Principal Components).

Table 7. Project table of data analysis

No. Model Type Number of

significant PC Number of observations R2X (cum) Q2X (Cum) Title 1 M1 PCA-X 4 43 0.698 0.344 Full dataset 2 M2 PCA-X 3 20 0.580 0.0793 Primary grinding 3 M3 PCA-X 3 20 0.567 0.114 Secondary grinding Model M1

(33)

25 samples from primary grinding with a temperature of 10 °C were in the upper right quadrant compared to the samples ground at 40 °C, which were in the lower right quadrant.

To determine whether sample 22.1 was a strong outlier, a Hotelling’s T2 plot was displayed (Figure 14). The Figure 14 presents sample 22.1 as a strong outlier, with high effect on the model. Furthermore, both samples 21 and 22 indicated deviating properties with high effect on the model. Thus, samples 21 and 22 were excluded for further data analysis with SIMCA.

Figure 13. PCA score plot - Full dataset Figure 14. Hotelling's T2 plot - Full dataset Model M2

In Figure 15 and Figure 16 the score and loading plot, respectively, for samples from primary grinding are presented. The score plot indicated four clusters, located in each quadrant. Samples 4, 6, 8, 12, 14 and 16 were located close to each other, hence they were positively correlated. The temperature (10 °C) was the common parameter for the samples. In general, the samples were located according to temperature, with the samples ground at low temperature (10 °C) at the right side and with the samples ground at high temperatures (25 °C and 40 °C) at the left side of the figure. Additionally, the deviating properties for sample 2 can be related to errors during the experimental work.

(34)

26 Figure 15. PCA score plot - Primary grinding Figure 16. PCA loading plot - Primary grinding Model M3

In Figure 17 and Figure 18 the score and loading plot, respectively, for samples from secondary grinding are presented. Generally, differences were observed for samples from primary- and secondary grinding. For evaluation of the samples from the experimental work, the error due to flush water must be considered. The water quality analysis of the flush water indicated leaching of anhydrite/gypsum and this may have affected the concentrations of constituents such as chloride, magnesium or sodium. Thus, the results from the secondary grinding have been excluded from conclusions and comparisons.

The samples from secondary grinding were evenly distributed in the score plot (Figure 17). However, samples were located in different quadrants according to temperature. In comparison with the score plot for samples from primary grinding, the center-points (ground at standard conditions) were located close to the samples ground at lower temperatures. As for the results from the primary grinding, alkalinity was associated to low temperatures and variables as sulphate, nitrate, conductivity and chloride were associated to high temperatures.

(35)

27

5.2 Variation of concentrations, pH, conductivity and alkalinity

For data evaluation and comparison, the measurement uncertainty of the different analyses must be considered. When comparing primary- and secondary grinding the error due to flush water must be considered. In this section, the varying concentrations of sulphate, calcium and uranium are presented. In addition, the variations of pH, conductivity and alkalinity are illustrated. All results are presented in Appendix 2.

The initial sulphate concentration in process water (before grinding) was 1960 mg/L. In Figure 19 and Figure 20, the variations of sulphate and calcium concentrations in the water samples from the experimental work are presented. The sulphate concentration was reaching a maximum of 1990 mg/L during laboratory grinding. The lowest sulphate concentration was observed for the sample with the lowest initial concentration; sample 22, for which the concentration was 895 mg/L. The results show an average sulphate concentration of 1870 mg/L for the samples ground with 100 % process water (sample 1-20). Based on the measured sulphate levels and the measurement uncertainty of sulphate analysis (of ± 15 %), leaching of anhydrite/gypsum during grinding with process water was considered as limited. In addition, the measured sulphate levels indicated saturation and were not affected by leaching of calcium sulphate.

Figure 19. Sulphate concentrations in the samples from the experimental work

(36)

28 Figure 20. Calcium concentrations in the samples from the experimental work

The variations of pH are depicted in Figure 21. The highest pH value for the experiments was 8.1 and the lowest pH value was 7.5. Although the initial process water was pH adjusted (between pH 7 and 9) the water samples after grinding were close to a pH of 8. The results show the buffering capacity of the process water being controlled by the carbonate system [12].

(37)

29 The variations of conductivity in the experimental data are presented in Figure 22. The conductivity was varying between 500 mS/m and 600 mS/m for the samples ground with 100 % process water. The conductivity decreased with increasing dilution with de-ionized water (samples 21 and 22). The conductivity after grinding with 100 % de-ionized water was 258 mS/m after primary grinding and 280 mS/m after secondary grinding. For all samples, conductivity increased after secondary grinding.

Figure 22. Variations of conductivity in samples from the experimental work

(38)

30 Figure 23. Variations of alkalinity in samples from the experimental work

However, uranium concentration in process water indicated the same correlations as alkalinity. The uranium concentrations in the samples from the experimental work are presented in Figure 24. After the primary grinding at low temperatures (10 °C) the uranium concentrations increased up to five times the concentration of samples ground at higher temperatures (25 °C and 40 °C) or in the secondary mill. The high uranium concentrations may be correlated to the increasing alkalinity and the solubility of carbonates.

(39)

31

5.3 Effects of dilution

In order to determine the impact of dilution on leaching, samples with a varying degree of dilution (samples 20, 21 and 22) were compared. All samples were ground at the same conditions. All samples had different initial concentrations as a result of the different degree of dilution with de-ionized water. The initial concentrations of water used for grinding are given in Table 8.

Table 8. Initial concentrations of water used for primary- and secondary grinding

Sample Water composition Initial sulphate concentration [mg/L]

20 100 % process water 1960

21 50 % process water 50 % de-ionized water

892

22 100 % de-ionized water -

In Figure 25 the sulphate concentration after primary grinding in samples 20, 21 and 22 are plotted versus the degree of dilution. The sulphate concentration increases with increasing proportion of process water as well as increasing ionic strength.

Figure 25. Dilution degree of samples from the experimental work

5.4 Comparison of sulphate contents in grinding circuit and in laboratory grinding

(40)

(41)

33

6. Evaluation of results from the experimental work – Solid samples

In this chapter, the results from chemical-, particle size- and mineralogical analysis (QEMSCAN) are presented and discussed. Furthermore, the results from sequential leaching and errors are evaluated.

6.1 Particle size analysis with Malvern master-sizer

Malvern master-sizer was used to determine the particle sizes of the samples from secondary grinding. To determine whether the laboratory grinding is similar to the grinding circuit in KA3, the parameters D10, D50 and D90 were compared. Averages of the D10, D50 and D90 values from the experimental work were compared with the corresponding parameters for samples taken from the overflow of the hydro-cyclone in KA3. The hydro-cyclone is located after the secondary grinding.

The D10, D50 and D90 (given in µm) values from the experimental work are presented in Table 9. The varying grinding times resulted in varying particle sizes (Table 9). The samples ground with the longest grinding times (12 and 35 minutes) resulted in the finest particle sizes, and the samples ground with the shortest grinding times (6 and 25 minutes) resulted in the coarsest particle sizes.

Table 9. Particle sizes of solid samples analyzed with Malvern master-sizer

Samples Grinding time D10 [µm] D50 [µm] D90 [µm]

(42)

34 The particle sizes of KA3 and the laboratory grinding are presented in Table 10. The D10, D50 and D90 values have been determined from previous samplings in KA3 (M. Olofsson, personal communication 2019). The average sizes, of all 22 samples, from the experimental work were similar to the corresponding values in KA3. Thus, the experimental work is considered successful and the observations made in laboratory grinding are comparable to KA3.

Table 10. Comparison of particle sizes - concentrator and samples from experimental work

D10 [µm] D50 [µm] D90 [µm]

Average sizes experimental work

3.96 29.19 75.11

Overflow hydro-cyclone KA3

3.77 29.15 74.50

The D10, D50 and D90 values for different grinding times and for the overflow of the hydro-cyclone in KA3 are illustrated graphically in Figure 27. The D10, D50 and D90 from KA3 indicated similar values as for the samples ground at 9 and 30 minutes (standard conditions). The samples ground with the shortest grinding times (6 and 25 minutes) were differing the most from the corresponding values in KA3.

(43)

35

6.2 Chemical assays

After two-stage grinding, the solid samples were analyzed using XRF. All results are presented in Appendix 3. When evaluating the results, the iron content in samples must be considered. The iron content in the ground samples may be increased and negatively affected by rust from grinding media, if the grinding media is not rinsed before grinding. One of the varying parameters during experimental work was the temperature. The temperature in the laboratory mill was adjusted by heating and cooling of both grinding media and process water. In order to retain the temperatures, the grinding media was not rinsed before grinding. Thus, the iron content in the final samples has been affected by the grinding media.

In the chemical assays, the calcium content was represented as the content of calcium oxide (CaO). In Figure 28, the CaO and total sulphur contents in the ore samples are presented. The figure indicates a decreasing total sulphur content in the samples after grinding in comparison with the initial sulphur content, which indicates leaching. The CaO content, however, varied differently, showing no decrease after grinding in comparison with the ore feed. Additionally, the results indicate larger variations for sulphur than for CaO. The CaO content was between 1.9 % and 2.15 %, to be compared with the sulphur content between 0.1 % and 0.2 %. The variations of CaO may be defined by the measurement uncertainty and difference between ore feed. The measurement uncertainty for analysis of CaO is 5 % for CaO content higher than 1 % (A. Larsson, personal communication 2019).

Figure 28. CaO and sulphur content in ore samples

(44)

36 grinding, which indicates leaching of chlorine. Also, the chloride content in initial water was 608 mg/L, and after primary grinding the chloride content was increasing to an average concentration of 770 mg/L (for samples ground with process water).

Figure 29. Chlorine content in ore and chloride concentrations in process water after primary grinding

6.3 Mineralogical analysis – QEMSCAN

(45)

37 Table 11. Quality/grade of ore feed and sample 12 analyzed with QEMSCAN

Minerals Chemical formula

Source: Webmineral.com Ore feed [wt. %] Sample 12 [wt. %] Magnetite/hematite FeO∙Fe2O3/ Fe2O3 92.60 92.80

FeO (Ti bearing) FeO 0.09 0.10

Titanite CaTiSiO5 0.3 0.28

Ilmenite FeTiO3 0.08 0.06

Rutile TiO2 0.09 0.09

Other Ti-minerals 0.18 0.22

Calcium sulphates CaSO4 0.00 0.01

Calcite CaCO3 0.01 0.07 Dolomite CaMg(CO3)2 0.01 0.03 Ankerite Ca(Fe,Mg,Mn)(CO3)2 0.01 0.02 Pyrite FeS2 0.19 0.14 Quartz SiO2 2.14 1.97 K-feldspar K-AlSi3O8 0.48 0.39 Na-feldspar Na-AlSi3O8 1.69 1.60

White mica minerals 0.01 0.04

Biotite K(Mg,Fe)3[AlSi3O10(OH,F)2 0.21 0.22

Chlorite minerals 0.06 0.04

Zircon ZrSiO4 0.01 0.05

Apatite Ca5(PO4)3F 0.59 0.35

Other minor gangue minerals _1.23 _1.48

The mineral distributions in ore feed and sample 12 are illustrated in Figure 30 and Figure 31, respectively. The minerals from Table 11 were grouped as iron oxides, silicates, sulphides, Ti-minerals, Ca-minerals and other minor gangue minerals. Finally, the amount of leached calcium sulphate could not be determined with QEMSCAN.

Figure 30. Mineral distribution ore feed Figure 31. Mineral distribution sample 12

(46)

38 In Lundkvist [10] pyrite oxidation was considered as the dominating source of sulphur to the dissolved phase in the drainage area of the mine, together with dissolution of gypsum/anhydrite in the process. The results from Lundkvist [10] and the results from QEMSCAN, results in following hypothesis: Pyrite oxidation is regarded as a potential source of sulphate in the process water.

Although the pyrite content was higher than the calcium sulphates, the conditions in the water system of LKAB (high pH) decreases the reaction rate of pyrite oxidation. Instead, soluble calcium sulphates will increase the sulphate concentrations in the process water due to large volumes of ore feed.

6.4 Sequential leaching

The calcium and sulphate concentrations in the leachates after each leaching stage are presented in Table 12. The sulphate concentrations were calculated based on the sulphur concentrations. The measurement uncertainty for sulphur analysis was ± 3.4 %, and for calcium analysis, it was ± 19 %. The results indicated high sulphate concentrations in the first, second and fifth leaching stage. For calcium, the concentrations were high in the first and the third stages of leaching. The increased levels of sulphate in the fifth stage of leaching are correlated to pyrite oxidation and leaching of sulphides. Most of the calcium sulphates are assumed to be leached out in the first stage of leaching, which may be related to the leaching of anhydrite. The high sulphate level in the second stage may be explained by leaching of gypsum, which is poorly soluble in comparison with anhydrite. The high calcium level in the third stage may be correlated to dissolution of calcite or other calcium carbonates.

Table 12. Calcium and sulphate concentrations in leachates (ore feed and sample 12)

Ore feed Sample 12

Leaching stage

Calcium [mg/kg] Sulphate [mg/kg] Calcium [mg/kg] Sulphate [mg/kg]

1 5660 2459.1 5290 1087.3

2 <208 623 <477 1428.8

3 4250 80.9 5250 350.5

4 727 110.8 427 470.3

5 300 2749.7 <196 2887.5

(47)

39

6.5 Particle analysis of filter papers with SEM-EDS

The minerals and the mineral distribution of particles suspended on the 0.45 µm filter papers after vacuum filtration, were determined with SEM-EDS. Two filter papers were analyzed, the filters after filtration of water samples from primary- and secondary grinding, respectively, of sample 12. The mineral distribution is presented in weight percent and is normalized to 100 %. The results from analysis with SEM-EDS (Table 13) indicated similar mineral distribution, mainly silicates and iron oxides, as for the solid samples analyzed with QEMSCAN. Additionally, the detected chromium-rich particles are most likely originating from the grinding media.

Table 13. Detected minerals on filter papers used for sample 12 analyzed with SEM-EDS

Minerals Primary

grinding [wt. %]

Secondary grinding [wt. %]

Ca-rich particles (e.g. calcite, apatite) 10 30

Ca-sulphates (e.g. gypsum/anhydrite) 5 <5

Si-rich particles (silicates) 45 35

Silica (e.g. quartz) 5 <5

Fe-rich particles (e.g. iron oxides) 35 25

Ti-rich particles (e.g. titanite) <5 <5

Barium-Sulphur-rich particles (e.g. Ba-sulphate) <5

Chromium-rich particles with iron <5

6.6 Discussion of measurement errors

During the experimental work, the mills and grinding media were rinsed with tap water. The flushing stage between primary- and secondary grinding was considered as a systematic error since the tap water, with low sulphate concentration of <10 mg/L, promoted dissolution of anhydrite/gypsum in the ore samples. However, only two samples of flush water (used to rinse the primary mill) were analyzed. The results from the water quality analysis are given in Table 14 and indicate high sulphate concentrations of approximately 1400 mg/L. The high sulphate concentrations in the flush water after filtration can partly result from mixing with remaining process water. Although anhydrite/gypsum is assumed to be dissolved during flushing and filtration after primary grinding. To avoid errors, the samples after the secondary grinding are excluded from further evaluation and comparison.

Table 14. Sulphate concentrations in flush water from the experimental work

Samples Sulphate concentrations [mg/L]

Flush water primary grinding sample 10 1430

Flush water primary grinding sample 12 1350

(48)

40 The material losses were determined by measuring the ore feed, the filter cake before and after secondary grinding as well as the filters used for filtration of slurry from primary grinding. The total material losses for sample 12 was 62 g, for which 35 g were lost during filtration. The remaining 27 g were lost during grinding. The material losses for sample 12 are presented in Table 15. However, the material losses were not significant for the final conclusions.

(49)

41

7. Equilibrium calculations

According to Westerstrand [4], leaching tests alone cannot give molecular level understanding of processes. However, they can be combined with geochemical modelling to improve understanding. The geochemical modelling can be used by applying data on Gibbs free energy and equilibrium constants; hence it can be used to determine whether the water system is oversaturated or undersaturated with respect to a specific mineral, e.g. gypsum/anhydrite.

For this thesis, equilibrium calculations were made using the GEM module in HSC Chemistry 9 software [29]. The results from the equilibrium calculations do not represent real scenarios but illustrate the behaviors of components in the system. The aim with the equilibrium calculations was to simulate and confirm the behavior of sulphate during the experimental work and at increasing amounts of e.g. chloride, anhydrite and calcite in the concentrators.

7.1 Methodology

For equilibrium calculations the initial concentrations of the process water used for the experimental work were used. The calculations were made at 40 °C with different additions of anhydrite, calcite and chloride. First, the amounts of constituents in the process water were recalculated from mg/L to kg/L, assuming 1 liter of process water. In addition, silicon and phosphorus were given as separate species in the water analysis; hence, the concentrations were recalculated as PO43- and H4SiO4 to fit the HSC software. The initial concentrations of the

process water entered to the HSC software are presented in Appendix 6. Further, an addition of anhydrite was made with aim to determine the equilibrium amount of sulphate and calcium, respectively, with a solid phase. According to the anhydrite content (0.01 wt. %) determined with QEMSCAN, 0.1 g of anhydrite was added to the system (assuming 1 kg of ore feed). Thereafter, the equilibrium concentrations were calculated and illustrated graphically. For further evaluation of saturation and equilibrium amounts, the amounts of chloride and sodium were adjusted.

7.2 Results and discussion

Survey of sulphates in process water of LKAB - Kiruna operation