meerin

andss ss fou5, suH toh reexp

specied residual concentrations of iron or sulfate and the pH. The data indicate that in particular, basic

lth toGold

ccountld are2012).

that forms under natural conditions when geologic strata contain-ing pyrite or other sulde bearing minerals are exposed to theatmosphere or oxidizing environments (Fripp et al., 2000;Gaikwad and Gupta, 2008; Jennings et al., 2008; Taylor et al.,2005). These waters typically pose a risk to the environmentbecause they contain elevated concentrations of metals such asiron, aluminium and manganese, and possibly other heavy metals

l., 2010). ApH valu

vated levels of heavy metals, sulfate and radioactive subssuch as uranium. The impacts of low pH can be immediasevere hence AMD is often detrimental to aquatic life (Joand Hallberg, 2003; Taylor et al., 2005).

AMD has a signicant potential to have an impact on theenvironment and the health of the people that are dependent onthe water around the AMD polluted region (Chapman, 2011).AMD has long-term environmental impacts that include revegeta-tion and rehabilitation difculties (Taylor et al., 2005). This isbecause soils contaminated with AMD have an imbalance of

Corresponding author. Tel.: +27 11 717 7592; fax: +27 86 211 8303.E-mail address: craig.sheridan@wits.ac.za (C. Sheridan).

Minerals Engineering 64 (2014) 1522

Contents lists availab

n

elstechniques to address the situation.AMD is acidic water laden with iron, sulfate and other metals

committee on acid mine drainage (Coetzee et athe Witwatersrand Basin is characterised by lowhttp://dx.doi.org/10.1016/j.mineng.2014.03.0240892-6875/ 2014 Elsevier Ltd. All rights reserved.MD ines, ele-tanceste andhnsonmining in the metropolitan area currently poses challenges such asthe threat of groundwater and surface water contamination arisingfrom Acid Mine Drainage (AMD). Clean water is universally anessential resource and South Africa faces a threat to water securityin the near future if the issue of AMD is not fully addressed(Chapman, 2011; Cobbing, 2008). Considering the environmentaland ecological threats this poses there is a need for innovative

characteristic yellow-boy. Following this, the ferric iron that doesnot precipitate from solution is able to oxidize additional pyrite.

Acid Mine Drainage (AMD) is a growing problem in coal andgold mines areas (Chapman, 2011; Taylor et al., 2005). In SouthAfrica, the main focus of attention is to address the Witwatersrandgold elds around Johannesburg as it has been listed as a problem-atic area with respect to AMD according to the Inter-ministerial1. Introduction

South Africa owes much of its weain areas such as the Witwatersrandlanga Coal elds. Mining currently aof South Africas GDP and coal and gorevenue generators (Statistics South,oxygen furnace slag has signicant potential as a replacement reagent for lime in treating acid minedrainage.

2014 Elsevier Ltd. All rights reserved.

its mineral riches foundelds and the Mpuma-s for approximately 9%two of the three largestHowever, the legacy of

such as uranium (Akcil and Koldas, 2006) and often have a low pHand high sulfate concentrations. AMD is formed through a numberof chemical reaction pathways, namely pyrite oxidation, ferrousoxidation and iron hydrolysis (Akcil and Koldas, 2006; Singer andStumm, 1970; Stumm and Morgan, 1996). The oxidation of ferrousiron to ferric iron follows after which ferric iron precipitates as ironhydroxide (Fe(OH)3) at pH above 3.5 (Tutu et al., 2008) leading toAcid mine drainage maximum treatment capacity of the slag. These experiments indicated that slag replacement strategiesare wholly dependent on the strength of the acid mine drainage, the required residence time and theRemediation of acid mine drainage using

Tendai Name, Craig Sheridan Industrial and Mining Water Research Unit, School of Chemical and Metallurgical Engin2050, South Africa

a r t i c l e i n f o

Article history:Received 20 August 2013Accepted 24 March 2014Available online 19 April 2014

Keywords:Basic Oxygen Furnace (BOF) slagStainless Steel (SS) slag

a b s t r a c t

In this study basic oxygenmine drainage. The stainleable. Basic oxygen slag wamine water with a pH of 2.slag was able to increase pexperiments. In these batcrapid process. Additional

Minerals E

journal homepage: www.etallurgical slags

g, University of the Witwatersrand, Johannesburg, Private Bag 3, Wits, Johannesburg

stainless steel slag were both assessed for potential use in treating acidteel slag was able to effect some pH change but was found to not be suit-nd to have a signicant potential as a remediating agent. For a model acidlfate concentration of 5000 mg/L and iron concentration of 1000 mg/L, the12.1, reduce the soluble iron by 99.7% and reduce sulfate by 75% in batch

actors most reaction was completed within 30 min indicating that this is aeriments were conducted with continuous ow reactors to assess the

le at ScienceDirect

gineering

evier .com/locate /mineng

a major problem as it shields the limestone from reacting with the

ls Enecessary elements vital for plant growth. Detrimental effectsposed by excess iron include interference with the uptake of man-ganese which is important for plant growth; clogging of sh gills;and build-up of iron and acid in animals internal organs can resultin fatal consequences (Fripp et al., 2000). Buildings and infrastruc-ture are also subject to degradation with time due to the corrosiveeffects of AMD (Taylor et al., 2005).

The research conducted around the Witwatersrand Basin areahas also found high levels of radioactive material like uraniumwhich may pose cancer risks (Coetzee and Winde, 2006). TheDWAF (Department of Water and Forestry, 1996) stated that con-centrations of sulfate that are greater 600 ppm causes the waterto taste bitter and may result in diarrhoea. Elevated sulfate concen-trations also results in gypsum formation which degrades concretestructures and causes scaling in pipes and lters (Madzivire, 2009;Swanepoel, 2011).

Appropriate treatment methods need to be implemented toaddress the threats posed by AMD. Historically, focus has mainlybeen on minimisation and control as the best practice (Tayloret al., 2005). However, the generation of AMD is in essenceunavoidable and it is practically difcult to inhibit the formationof AMD at its source. Various methods have thus been proposedto tackle problems posed by AMD with mixed results to ensuretreated efuents meet threshold values set by the government byremoving heavy and toxic metals and maintaining acidity atacceptable levels. Feasible methods for treating AMD were identi-ed and divided into active and passive processes (Akcil andKoldas, 2006; Taylor et al., 2005) in the literature. Active treatmentinvolves addition of alkaline chemicals like limestone, lime, causticsoda and ammonia while passive treatment involves developingnatural chemical and biological systems that are self-operating likeconstructed wetlands (Gaikwad and Gupta, 2008; Ochieng et al.,2010). Advantages with active treatment systems are that theycan cope with higher ow rates, are exible and have smaller foot-prints. Disadvantages include high operating costs associated withalkaline chemicals, constant monitoring and maintenance, skilledmanpower and production of sludge outweigh the advantages.The advantage with passive treatment system is they require littlemaintenance. This is critically important in a country such as SouthAfrica which has a skills shortage.

Passive treatment systems have been used for a number ofyears to treat mine efuent of varying compositions and pH levels(Dvorak et al., 1992). They have been argued to be the long termstrategy to solving AMD problems and with further research theymay become more widely used in future (Jennings et al., 2008).Passive treatment methods used include primarily wetlands(Johnson and Hallberg, 2003; Sheoran and Sheoran, 2006;Sheridan et al., 2013; Taylor et al., 2005; Wallace and Knight,2006) and limestone-based beds or drains (Kleinmann et al.,1998; Taylor et al., 2005). Ochieng et al. (2010) explored the feasi-bility of aerobic and anaerobic wetlands which are also currently inuse in South Africa. Although aerobic wetlands were able toremove various metals from efuents, their main disadvantage liesin the fact that they cannot handle typical AMD efuents and theyrequire vast surface areas for their operation. A further disadvan-tage is their inability to reach pH levels greater than 8 (Tayloret al., 2005). The disadvantage with anaerobic wetlands is therequirement for large area of land for effective treatment. (Battyand Younger, 2004) are of the opinion that vegetation is difcultto establish in AMD treatment applications in both aerobic andanaerobic wetlands. The main disadvantage with limestone bedsis the regular maintenance required to ensure maximum life andeffectiveness. The porosity of the beds and that of organic matter

16 T. Name, C. Sheridan /Minerais reduced as the systems get blocked with treatment precipitates(Cravotta, 2003; Johnson and Hallberg, 2003; Potgieter-Vermaaket al., 2006). Armouring of limestone occurs when ferrous iron isAMD.

1.1. The use of metallurgical slags for treating AMD

Although various processes have been proposed for the treat-ment of AMD using traditional passive treatment schemes, noneof these methods is ideal as a long term solution due to high oper-ating costs and technological failures (Barrie Johnson and Hallberg.,2005). Slags are a highly alkaline by-product of the smelting pro-cess for metals such as steel, copper etc. Slags are highly alkalinebecause they are composed primarily of hydrated amorphous sil-ica, calcium oxide and magnesium oxide (Ziemkiewicz andSkousen, 1998) and often lime is used in the smelting process asa ux. They are widely available in countries such as South Africadue to its large minerals rening industry. The potential use of slagfor the treatment of AMD has been studied and described by vari-ous authors (Gaikwad and Gupta, 2008; Kruse et al., 2012; Shenand Forssberg, 2003; Ziemkiewicz and Skousen, 1998). Researchby Ziemkiewicz and Skousen (1998) suggested direct addition ofsteel slag into streams affected by AMD as an alternative treatmentmethod. Feng et al., 2004 further supported ideas by Ziemkiewiczand Skousen (1998) by citing that slag can increase the pH of acidmine water to almost neutral levels and remove heavy metals.(Bowden et al. 2006) further discovered that rapid iron removalwas possible using steel slag. In South Africa, the use of Slag LeachBeds (SLBs) as a form of passive technology has not been fullyinvestigated, but shows potential in being able to treat AMD(Sheridan et al., 2013). Previous studies on SLBs have mainlyfocussed on stormwater pollution (Taylor et al., 2005) and AMDfrom disused coal mines and direct treatment of water(Ziemkiewicz and Skousen, 1998). The motivation of the presentstudy was to evaluate two common types of slags, which are wasteproducts, to neutralise AMD as an alternative to using lime, whichneeds to be mined. Hence, the use of slags has the potential to pro-vide a cheaper alternative to lime, while at the same time makinguse of a waste stream that is readily available in South Africa.

1.2. The mechanism of slag in acid mine drainage remediation

The CaO found in slag reacts with water to form hydratedcalcium hydroxide (Ca(OH)2). Dissolution of (Ca(OH)2) followsand creates alkalinity. Different forms of iron react with hydroxide(OH) to forms different products depending on the resulting pHvalues obtained by the addition of slag. The removal of iron insolution by slags is due to acid neutralising ability which leads toprecipitation (Feng et al., 2004; Rose, 2010). The removal of sulfatein AMD in slag can be attributed to the formation of gypsum(CaSO42H2O) and other sulfate precipitates that can possibly form.The reactions involved in pH increase, iron and sulfate removal arepresented as Eqs. (1)(4).

CaOH2O! CaOH2 1

CaOH2 ! Ca2 OH 2

Fem mOH ! FeOHm 3

Ca2 SO24 2H2O! CaSO4 2H2O 4

1.3. Research objectivesoxidized and ferric hydroxide precipitates on the limestone is also

ngineering 64 (2014) 1522Within this context, this study sought to investigate the abilityof Basic Oxygen Furnace (BOF) and Stainless Steel (SS) slags to

5 L with Solution A of AMD. The experimental volumes were 5times greater than in Section 2.3 since many samples were takenfrom each experiment and we did not wish to signicantly alterthe mass balance by taking many samples. The masses of slag usedin these experiments ranged from 100 g to 5000 g. The slagsamples were placed in unstirred, open, 5 L glass beakers and5 mL aliquots were sampled after every 30 min and stored in50 mL test tubes for analysis. The sampled solutions were testedfor pH, iron and sulfate levels as described above.

2.5. Testing the maximum capacity of the slag using continuous AMDow

For this experiment, two packed bed reactors were constructedusing 1500 mL beakers. 1200 mL of SS slag and BOF slag wereplaced into the two beakers. Synthetic AMD solution B waspumped (using a peristaltic pump) into the base of each beaker

assaying (Scroobys Laboratory Services). The composition of the

C 1.06 0.76

ls Engineering 64 (2014) 1522 17reduce the acid, sulfate and iron content of a typical Witwaters-rand gold basin AMD. In addition it was sought to investigate thekinetics, determine the rate of pH change and the amount of ironand sulfate removed by the slag from the AMD, with a view tomaximising the lowering of iron and sulfate concentration andacidity of AMD and to determine the long-term capacity of the slagto treat AMD.

2. Experimental methods

In order to complete the research objectives, three experimentswere conducted. These included a test to determine the optimumratio of slag to AMD, a test to understand the rate at which theAMD reacted with the slag and a test using continuous ow todetermine the point of slag saturation.

2.1. Analytical techniques

Iron and sulfate were analysed using a Merck SpectroquantPharo 300 spectrophotometer and Merck reagent kits. Sulfate andiron ion concentrations were analysed using the Merck test kitsNo. 114791 and No. 114761 respectively. The pH of all sampleswas measured using a Metrohm 744 pH meter and calibrated withMetrohm buffer solutions at pH = 4, 7 and 9 at frequencies accord-ing to the manufacturers specications. After use, the electrodewas washed with distilled water and then dried to prevent con-tamination of subsequent samples measured. Data was analysedusing MS Excel.

2.2. Preparation of an articial AMD

A typical Witwatersrand gold basin AMD with low pH andelevated concentration of metals and sulfate was simulated inthe laboratory. The synthetic AMD was created according torecommendations by Potgieter-Vermaak et al. (2006). Twodifferent solutions of AMD were synthesised; Solution A had acomposition of 600 mg/L Fe, 4800 mg/L SO42 and pH of 2.5 repre-senting a low-strength AMDwhilst Solution B had a composition of1000 mg/L Fe, 5000 mg/L SO42 and pH of 2.25 representing a high-strength AMD. The synthetic AMD was prepared by dissolvingindustrial (95% pure) hydrated ferrous sulfate (FeSO47H2O) andsulfuric acid (H2SO4) in distilled water the desired iron and sulfateconcentration. pH was adjusted using analytically pure NaOH fromMerck Chemicals. Solution loss through sampling and evaporationduring the course of experimentation was compensated by theaddition of distilled water and the mass balance was correctedfor these small losses.

2.3. Testing the ratio of AMD to slag

Samples were prepared to give slag to AMD ratios of 20, 40, 60,80, 100, 120 and 140 g/L (grams of slag per litre of AMD) by addingan appropriate amount of slag with AMD from a prepared stocksolution. The masses of slag used in the experiments ranged from20 g to 140 g and the solution volume used was 1 L. The sampleswere placed in unstirred, open, 2 L glass beakers and S left for fourhours to allow for sufcient time for the reaction to take place. Thesamples were ltered from the slag residue and each sample wasthen analysed for acidity, iron and sulfate content. The pH wasmeasured with a digital pH meter.

2.4. Testing the optimum time of contact between the AMD and slag

T. Name, C. Sheridan /MineraThis experiment was designed to test slag to AMD ratios of 20,40, 60, 80, 100, 120 and 140 g/L by adding the mass and topping toS 0.34 0.13MnO 1.42 1.27P2O5 0.46 0.05SiO2 15.2 26.8Cr2O3 0.31 1.91NiO 60.01 0.31CuO 0.19 0.07Al2O3 5.52 5.87V2O5 0.48 0.05TiO2 4.02 0.68CoO 60.01 60.01slags is shown in Table 1. The samples were pulverised to less than75 lm. Parallel samples were prepared by measuring a knownmass, fusing with sodium peroxide and then leaching with HCl,HNO3 and water. Samples were analysed by ICP a Thermo ScienticiCap 6300 radial utilising iTeva software.

The slags are mainly comprised of calcium oxide or lime (CaO),silicon dioxide (SiO2), iron (II) oxide (FeO), magnesium oxide(MgO) and aluminium oxide (Al2O3). The SiO2 was 11.6% more inSS slag than in BOF slag. The BOF slag contained more Fe andCaO than the SS slag.

Table 1Compositions of BOF and SS slag as determined by SLS-ICP (ScienticLaboratory Supplies Inductively Coupled Plasma).

Analysis Composition of BOF slag(mass %)

Composition of SS slag(mass %)using a peristaltic pump and this owed slowly up through themass of slag prior to overowing from a overow nozzle. Flowrates of 4 mL/min, 8 mL/min, 12 mL/min and 16 mL/min were usedwhich translated to 2, 1, 0.5 and 0.25 h of residence time. Resi-dence time is dened as the liquid volume divided by the volumet-ric owrate, and is given by Eq. (6) in this paper. Porosity of theslag was measure at approximately 40%. Each test was completedtwice to check reproducibility. Samples of the outow of the SLBswere taken every 2 h during operating days and samples were ana-lysed for sulfate, iron and pH.

3. Results and discussion

3.1. Slag characterisation

The BOF and the SS slag were sent to an external laboratory forCaO 38.7 36.0MgO 6.80 13.0FeO 16.5 5.54

18 T. Name, C. Sheridan /Minerals E3.2.1. The pH changes of acid mine drainageIn Fig. 1 the increase in pH with increased slag to AMD ratios is

shown. The pH increased from 2.5 to 6.0 for SS slag and to 12.1 forBOF after four hours of contact time. A ratio of 100 g/L of AMD wasobserved to be the maximum ratio as an increase in slag had noappreciable increase in pH, especially for the SS slag. Themechanism of pH increase is hypothesised to be from dissolutionof available CaO present within the slag. Ca is reported as CaObut may also occur in a CaOSi phase. This could also explainwhy the SS slag was less reactive and had less alkalinity in solution.

It can clearly be seen in Fig. 1 that BOF slag had higher alkalinitythan SS slag. This is due to the fact that BOF slag had more calciumoxide content than SS slag as shown by the data in Table 1. It mayalso be that because SS slag had higher silicon dioxide content thanBOF slag, the CaO present was not available for dissolution as itwas locked in an insoluble glassy matrix. This was not, howevertested although the results agree with Shen and Forssberg (2003)who suggested that high silicon dioxide content in compoundstend to make them less alkaline.

3.2.2. Reduction of iron and sulfateThe % removal of iron and sulfate at different slag to AMD ratio3.2. Determining the effect of the ratio of AMD to slag

Fig. 1. The pH changes at different slag to AMD ratios for different slags after fourhours (pHo = 2.5).is presented in Fig. 2. As seen from the gure, iron and sulfateremoval increased with an increase in the slag to AMD ratio for

Fig. 2. Reduction of iron and Sulfate at different slag to AMD ratio for different slagsafter four hours (Feo = 600 mg/L; SO42o = 4800 mg/L).both slags used. The highest percentage iron removal recordedwas 63.6% for SS slag while 99.7% iron removal was recorded forBOF slag. The highest iron removal was recorded at pH values of6.0 and 12.1 for SS and BOF respectively. The maximum sulfateremoval percentage recorded was 40% for SS slag while 75% sulfateremoval was recorded with the use of BOF slag.

It was also noted that maximum sulfate removal was achievedat the optimum slag to AMD ratio (100 g/L), where pH values of11.3 and 5.9 were recorded for BOF and SS slag respectively. Fromthe graph above it is evident that iron was reduced by almost 100%at slag to AMD ratios of 100, 120 and 140 g/L for BOF slag. The pHvalues of 11.3, 11.3 and 12.1 were recorded at those ratios.

Almost all the soluble iron was removed using BOF slag and thiswas attributed to the formation of various precipitates. A Pourbaixdiagram for FeSH2O System at 298 K (Rose, 2010) indicates thatprecipitates such as Fe(OH)3 and Fe(OH)2 are formed, while Rose(1989) claimed that other iron precipitates such as (FeOOH),(Fe2O3) and (Fe5O8H4H2O) could also be formed at various pH val-ues. The type of iron precipitate generated was not assessed in thisstudy.

The low iron removal values obtained for SS slag are likely dueto the fact that lower pH values were found. The EhpH diagramindicates that below pH values of 9, the stable forms of iron areFe2+, Fe3+ and FeSO4+. Thus SS slag caused less iron removal thanBOF slag because of its inability to increase the pH above 9 whichwould result in the formation of precipitates.

Neither of the slags used in this study contained barium or leadin their composition; hence, the mechanism of sulfate removalwould primarily be due to the formation of gypsum which is onlysparingly soluble. As discussed, analysis of the sludge was not con-ducted. The BOF slag caused signicantly greater removal of sulfatethan SS slag.

3.3. Determining the optimum contact time between the AMD and slag

The concentration curves for pH, iron and sulfate for SS and forBOF slag are shown in Fig. 3. The results of this experiment showedthat the pH of AMD increased with an increase in contact timebetween AMD and both slags. The increase in pH was also depen-dent on the slag to AMD ratio. The pH increased rapidly in the rst30 min and thereafter reached steady state for all slag to AMD ratiocombinations. The pH increase was higher for BOF slag comparedto SS slag. The maximum pH reached for BOF was 11.3 while 5.9was reached for SS. The concentration of iron decreased much fas-ter in BOF slag than SS slag and the increase was also shown to be afunction of the ratio of slag to AMD for both slags. The iron concen-tration was lowest after 3 h with BOF slag for a slag to AMD ratio of100 g/L compared to a residual concentration of 220 mg/L of ironfor the SS slag. The rate of removal was most rapid in the rst30 min. For sulfate, similar trends were observed, the rate ofdecrease of concentration was most rapid in the rst 30 min andthe rate was also a function of the slag to AMD ratio with higherratios showing the highest rate.

These experiments indicate that in particular, BOF slag could bedirectly added to AMD as a form of remediation provided heavymetals are not leached from the slag by the AMD (this was nottested in this study). These results agree with those found by oth-ers (Ziemkiewicz and Skousen, 1998; Zurbuch, 1996) where theirtests indicated that AMD could be treated by directly applyingalkaline products into the mine discharge.

3.4. Testing the maximum treatment capacity of SS and BOF slagsusing continuous ow reactors

ngineering 64 (2014) 1522Further experiments were carried out in a continuous process inorder to inform for design or practical purposes in this experiment,

ls Engineering 64 (2014) 1522 19T. Name, C. Sheridan /MineraSolution B was used such that the results would be representworst-case scenario slag-treatment schedules. The experimentswere conducted with ow rates of 4, 8, 12 and 16 mL/min translat-ing to residence times of 2, 1, 0.5, 0.25 h. Given that SS slag wasfound to be less effective than BOF slag, it was not tested anyfurther and thus only results for BOF are presented.

3.4.1. Measuring the effect of ow rate on pHThe pH changes for treated efuent at ow rates of 4, 8, 12 and

16 mL/min in leach bed occupying a volume of 1.2 L for a period of12hrs continuous process using BOF slag are presented in Fig. 4.The feed pH of the simulated AMDwas 2.25. The maximum pH val-ues measured in the outlet were 13.2, 11.0, 9.9 and 7.9 for owrates of 4, 8, 12 and 16 mL/min respectively. Fig. 3 also shows thatpH increased rapidly in the rst two hours of sampling before grad-ually declining. This is primarily a function of contact time andindicates that the alkalinity is being consumed. As expected, thepH increased with an increase in residence time (or decreasingow rate). This is primarily a function of contact time.

3.4.2. Effect of Flow rate on reducing iron concentration in acid minedrainage

The removal of iron at ow rates of 4, 8, 12 and 16 mL/min overa period of 12 h in the continuous process using BOF slag is shownFig. 5. The removal in iron concentration was higher at low ow

Fig. 3. Concentration curves for sulfate, irorates and decreased as the ow rate was increased. The feedconcentration of iron was 1000 mg/L. Iron was decreased to non-detectable concentrations as shown in the Figure for feed ow rateof 4 mL/min and 8 mL/min. The concentrations began to increaseduring the course of the experiment which is related to the pointin the experiment when the pH values began to drop.

n and pH for SS slag and for BOF slag.

Fig. 4. Effect of ow rate on reduction of acid for BOF slag for a period of 12 h.

Precipitates of iron were thought to have been formed at highpH values. These precipitates of iron therefore ensured that nosoluble iron was detected. At low ow rates, these precipitateswould have been ltered by the slag. At higher ow rates, less reac-tion time was afforded to the slag to react and bring about high pHvalues necessary for formation of precipitates.

3.4.3. The effect of ow rate on sulfate concentrationIn Fig. 6 the sulfate concentration at AMD ow rates of 4, 8, 12

and 16 mL/min over a period of 12 h in continuous process usingBOF slag is presented. The removal of sulfate was also higher atlow ow rate and decreased as the ow rate was increased. Thefeed composition sulfate was 5000 mg/L and was decreased to aminimum concentration of approximately 740 mg/L for a ow rateof 4 mL/min. That minimum concentration achieved was stillabove the DWAF general limit for wastewater disposal into a waterresource (Department of Water and Forestry, 1996). Sulfatereduction was lower for high feed ow rates.

3.5. Calculating the effect of residence time on reducing iron and

The removal of sulfate in the AMD is shown in Fig. 10. It can beseen that the feed composition of sulfate was 5000 mg/L and thatthe sulfate concentration was 693 mg/L after 1 h, and outlet ofthe leach bed had a continuously rising sulfate concentration to a

20 T. Name, C. Sheridan /Minerals Engineering 64 (2014) 1522sulfate in AMD

The experiments carried out were able to reduce iron to levelsbelow DWAF general limit for disposal of wastewater into a waterresource, but sulfate levels were still above the limit (Departmentof Water and Forestry, 1996). It was decided to design the contin-uous process in way that would hopefully be to reduce sulfateconcentration to below 400 mg/L. Concentration of sulfate and ironwere thus plotted against residence time and a correlation wasestablished. The residence time capable of reducing sulfate concen-tration below 400 mg/L was tabulated from the equation relatingconcentration to residence time.

In Fig. 7 data is presented which shows iron and sulfate removalin AMD as a result of different residence times for ow rates of 4, 8,12 and 16 mL/min. From the Figure, it can be seen that the removalof iron and sulfate was greatest at a residence time of 2 h. This nd-ing is in agreement with those found by others (Kruse et al., 2012;Skousen and Ziemkiewicz, 2005) whose experiments found thatslag leach beds required one to three hours of residence time fortheir design.

Sulfate removal in synthetic AMD using BOF was correlated totime and the resulting function is given in Eq. (5):Fig. 5. Effect of ow rate on reduction of iron after treatment with BOF slag for aperiod of 12 h.Sulfate concentration 1077 lnresidence time 1457:6 5where y was the sulfate concentration and x, was the residencetime. According to the equation a residence time of 2.67 h wouldbe required to achieve a residual sulfate concentration of 400 mg/L.Thus, we used a ow rate of 3 mL/min of AMD to the slag reactorto achieve that concentration according to the calibration curve.That ow rate was fed to the slag bed for 2 days to ascertain howacid, iron and sulfate concentration changed with time.

The pH changes of AMD for the designed residence time of2.67 h using BOF slag over a period of two days are shown inFig. 8. The pH changed from 2.25 to a maximum of 13.3 in the rsttwo hours and decreased thereafter throughout the course of theexperiment. Data was collected between at 8 am and at 8 pmand hence there is no data between 12 h and 24 h. The data from24 h onwards followed the same trend, however.

For iron, the feed composition iron was 1000 mg/L and this wasreduced to below detection limit as shown in the graph in the rst1012 h. After 12 h the concentration of iron began to increase.The average concentration of iron between 012 h, 1224 h and2436 h is also presented in Fig. 9.

Fig. 6. Effect of ow rate on sulfate reduction after treatment with BOF slag for aperiod of 12 h.Fig. 7. Iron and sulfate reduction at different residence times.

maximum of 3700 after 36 h. The experiment was designed to havean initial sulfate removal to below 400 mg/L; however, this is notpossible due to solubility constraints. We chose this value,however so as to be conservative in our experimental design.

4. Predicting slag requirements for treating different quantitiesof different strength AMD streams

For design purposes, it would be necessary to understand thereplacement strategy of the slag treating different strength AMD.The residence time s, void fraction e, volume V, feed ow rate Qand mass of slag Vslag were all related according to the followingequation.

sVslagQ

s 6

We have chosen to use Vslag for this calculation since the mass ofFig. 9. Reduction of iron for a continuous process using BOF slag for duration of2 days.

Fig. 10. Reduction of sulfate for a continuous process using BOF slag for duration of2 days.

T. Name, C. Sheridan /Minerals Engineering 64 (2014) 1522 21slag heaps is unknown, however there volumes can reasonablyaccurately be estimated. The slag has a bulk density of approxi-mately 1200 kg/m3 which was used for conversion from experi-mental data for this scale-up exercise. In Fig. 11 the data is usedto illustrate the amount of slag required for a set of different feedow rates to achieve a specied set of varying outletconcentrations.

As can be seen, the amount of slag required for a xed AMD owrate will increases with an increase in residence time. For example,if it was required to treat 12 mL/d of AMD and the target residualconcentrations of iron and sulfate were 1.3 and 844 mg/L respec-tively, approximately 3300 m3 of slag would be needed with aresidence time of 2.67 h.

The volume of decant, or efuent, in the Witwatersrand Basin isexpected to range between 12 and 20 mL/d (Coetzee et al., 2010).The target iron and sulfate concentration need to be consideredbefore predicting the amount of slag required to treat the AMD.Under these conditions, i.e. treating these ow rates, the slagwould have to be replaced after the residence time specied. Thegure shows that the amount of slag required would increase withan increase in the feed ow rate. It also shows that more slag isrequired with an increase in residence time for a xed AMD owrate. This implies that if one were to treat 20 mL/d, after 2.67 habout 5500 m3 of slag would need replacing, and the nal concen-tration would be approximately 1.3 ppm Fe and 844 ppm sulfate. Ifone were to treat 20 mL/d after 0.5 h, about 200 m3 of slag wouldneed replacing, and the nal concentration would beFig. 8. Reduction of acid for a continuous process using BOF slag for duration of2 days.

Fig. 11. A prediction of the amount of slag required to treat different feed ow ratesof AMD.

22 T. Name, C. Sheridan /Minerals Eapproximately 188 ppm Fe and 3190 ppm sulfate. Most likely anal design would require several parallel systems such thatreplacement could occur whilst others remain operational.

5. Conclusions and recommendations

Both slags used were able to reduce acid, iron and sulfate con-centration. However, the BOF slag was signicantly better than theSS slag in reducing acid, iron and sulfate concentration. It wasshown that acid, iron and sulfate removal depended on the amountof slag added per 1L of synthetic AMD, the contact time betweenslag and AMD and ow rate of synthetic AMD fed to the SLBs.The ratio tests showed that acid, iron and sulfate removalincreased with an increase in slag to AMD ratio. A ratio of 100 gslag to 1 L of AMD was found to be the optimum at which maxi-mum removal was achieved for both BOF and SS slag in the batchexperiments. At that ratio 63.6% iron removal was found for SS slagcompared to 99.7% iron removal for BOF slag. A 39.8% sulfateremoval was found for SS slag compared to 75% sulfate for BOF slagat the same ratio. SS slag managed to increase the pH of syntheticAMD from 2.5 to 6.0 compared to 12.1 for BOF slag.

Acid, iron and sulfate removal was found to be very rapid in therst hour of contact between slags with AMD in the batch pro-cesses. Remediation of AMD was also successful in a continuousow process at low lower ow rates. For design purposes, theamount of slag required to treat different feed ow rates of AMDfor target iron and sulfate concentration and that any design wouldbe need to understand the strength of the AMD, the ow rate of theAMD and also the desired residual concentration, particularly ofsulfate.

From this study, future research to be conducted includes focus-ing on establishing the toxicity of any trace metals or elements thatmay leach into AMD during the removal of acid, iron and sulfate.Further research could also be carried out to ascertain the compo-sition of the slag residue after pH neutralisation and iron andsulfate removal. This will enable us to know if it can be usedfurther for other purposes such as road construction or as anaggregate in cement. The slag could also potentially be used inconjunction with constructed wetlands as a more efcienttreatment process.

Acknowledgements

We gratefully acknowledge the nancial support from the Glo-bal Change and Sustainability Research Institute of the Universityof the Witwatersrand, Johannesburg. We also extend our thanksto Harsco Metals and Minerals, South Africa for the provision ofthe metallurgical slags.

References

Akcil, A., Koldas, S., 2006. Acid Mine Drainage (AMD): causes, treatment and casestudies. Improving Environmental, Economic and Ethical Performance in theMining Industry. Part 2. Life cycle and process analysis and technical issues. J.Clean. Prod. 14 (1213), 11391145.

Batty, L.C., Younger, P.L., 2004. Growth of Phragmites australis (Cav.) Trin ex.Steudel in mine water treatment wetlands: effects of metal and nutrientuptake. Environ. Pollut. 132 (1), 8593.

Barrie Johnson, D., Hallberg, Kevin B., 2005. Acid mine drainage remediationoptions: a review. Sci. Total Environ. 338 (12), 314.

Bowden, L., Johnson, K., Jarvis, A.P., Robinson, H., Ghazireh, N., Younger, P.L., 2006.The use of basic oxygen steel furnace slag (BOS) as a high surface media for theremoval of iron from circumneutral mine waters. In: 2006 InternationalConference on Acid Rock Drainage, St. Louis, Mo.

Chapman, A., 2011. Acid mine drainage in South Africa: An Emerging EnvironmentalProblem. Institute for Futures Research, Stellensbosch.

Cobbing, J.E., 2008. Institutional Linkages and Acid Mine Drainage: The Case of the

Western Basin in South Africa. Int. J. Water Resour. Dev. 24 (3), 451462.Coetzee, H., Winde, F., 2006. An assessment of sources, pathways, mechanisms andrisks of current and potential future pollution of water and sediments in gold-mining areas of the Wonderfontein Catchment. WRC Report No. 1214/1/06,Water Research Commission, Pretoria.

Coetzee, H., Hobbs, P.J., Burgess, J.E., Thomas, A., Keet, M., Yibas, B., van Tonder, D.,Netili, F., Rust, V., Wade, P., Maree, J., 2010. Mine Water Management in theWitwatersrand Goldelds with Special Emphasis on Acid Mine Drainage:Report to the Interministerial Committee on Acid Mine Drainage. Council forGeoscience, Pretoria.

Cravotta, C.A., 2003. Size and performance of anoxic limestone drains to neutraliseacidic mine drainage. J. Environ. Qual. 32, 12771289.

Department of Water Affairs and Forestry. South African water quality guidelines.2nd ed. Industrial Use vol. 3, 1996. Department of Water Affairs and Forestry,Pretoria.

Dvorak, D.H., Hedin, R.S., Edenborn, H.M., McIntire, P.E., 1992. Treatment of metal-contaminated water using bacterial sulfate reduction: results from pilot-scalereactors. Biotechnol. Bioeng. 40 (5), 609616.

Feng, D., van Deventer, J.S.J., Aldrich, C., 2004. Removal of pollutants from acid minewastewater using metallurgical by-product slags. Sep. Purif. Technol. 40 (1),6167.

Fripp, J., Ziemkiewicz, P.F., Charkavork, H., 2000. Acid mine drainage treament technical notes collection. Report ERDC TN-EMRRPSR-14. Army EngineerResearch and Development Centre, Vicksburg.

Gaikwad, R.W., Gupta, D.V., 2008. Review on removal of heavy metals from acidmine drainage. Appl. Ecol. Environ. Res. 6 (3), 8198.

Jennings, S.R., Neuman, D.R., Blicker, P.S., 2008. Acid Mine Drainage and The Effectson Fish Health And Ecology: A Review. Reclamation Research Publication,Bozeman, pp. 1.

Johnson, D.B., Hallberg, K.B., 2003. The microbiology of acidic mine waters. Res.Microbiol. 154 (7), 466473.

Kleinmann, R.L.P., Hedin, R.S., Nairn, R.W., 1998. Treatment of mine drainage byanoxic limestone drains and constructed wetlands. In: Geller, W., Klapper, H.,Salomans, W. (Eds.), Acidic Mining Lakes: Acid Mine Drainage; Limnology andReclamation. Springer, Berlin, pp. 303319.

Kruse, N.A., Mackey, A.L., Bowman, J.R., Brewster, K., Reier, R.G., 2012. Alkalinityproduction as an indicator of failure in steel slag leach beds treating acid minedrainage. Environ. Earth Sci. 67 (5), 12951389.

Madzivire, G., 2009. Removal of Sulfates from South African Mine Water Using CoalFly Ash. University of the Western Cape, South Africa.

Ochieng, G.M., Seanego, E.S., Nkwata, O.I., 2010. Impacts of mining on waterresources in South Africa: a review. Sci. Res. Essays 5 (22), 33513357.

Potgieter-Vermaak, S.S., Potgieter, J.H., Monama, P., Van Grieken, R., 2006.Comparison of limestone, dolomite and y ash as pre-treatment agents foracid mine drainage. Selected papers from Processing and Disposal of MineralsIndustry Wastes 05, 19(5), 454462.

Rose, A.W., 2010. Advances in passive treatment of coal mine drainage 19892009.In: Barnhisel, R.I. (Ed.), Joint Mining Reclamation Conference. ASMR, Lexington,KY, p. 847.

Shen, H., Forssberg, E., 2003. An overview of recovery of metals from slags. WasteManage. 23 (10), 933949.

Sheoran, A.S., Sheoran, V., 2006. Heavy metal removal mechanism of acid minedrainage in wetlands: a critical review. Miner. Eng. 19 (2), 105116.

Sheridan, C.M., Harding, K., Koller, E., De Pretto, A., 2013. A comparison of charcoal-and slag-based constructed wetlands for acid mine drainage remediation.Water SA 39 (3), 369374.

Singer, P.C., Stumm, W., 1970. Acidic mine drainage: the rate determining step.Science 167, 11211123.

Skousen, J.G., Ziemkiewicz, P., 2005. Performance of 116 passive treatment systemsfor acid mine drainage. In: Proceedings of the 2005 National Meeting of theAmerican Society of Mining and Reclamation. Breckenridge, Colorado, June 1923, pp. 11001133, 2005 (Performance of 116 passive treatment systems foracid mine drainage. In: 2005 National Meeting of the American Society ofMining and Reclamation, Breckenridge, Colorado, pp. 11001133).

Statistics South Africa, 2012. Mineral accounts for South Africa: 19802009.Discussion Document D0405.2. Statistics South Africa, Pretoria.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry. Wiley, New York.Swanepoel, H., 2011. Sulfate removal from industrial efuents through barium

sulfate precipitation, University of the Northwest.Taylor, J., Pape, S., Murphy, N., 2005. A summary of passive and active treatment

technologies for acid and metalliferous drainage (AMD). In: 5th Australianworkshop on Acid Mine Drainage, Fremantle, Australia.

Tutu, H., McCarthy, T.S., Cukrowska, E., 2008. The chemical characteristics of acidmine drainage with particular reference to sources, distribution andremediation: the Witwatersrand Basin, South Africa as a case study. Appl.Geochem. 23 (12), 36663684.

Wallace, S.D., Knight, R.L., 2006. Small-scale constructed wetland treatmentsystems. IWA Publishing, Vancouver.

Ziemkiewicz, P.F., Skousen, J.G., 1998. The use of steel slag in acid mine drainagetreatment and control. In: 19th Annual West Virginia Surface Mine DrainageTask Force Symposium, 28(1), 4656.

Zurbuch, P.E., 1996. Early results from calcium carbonate neutralization of twoWest Virginia rivers acidied by mine drainage, In Seventeenth West VirginiaSurface Mine Drainage Task Force Symposium, Morgantown, WV.

ngineering 64 (2014) 1522

Remediation of acid mine drainage using metallurgical slags1 Introduction1.1 The use of metallurgical slags for treating AMD1.2 The mechanism of slag in acid mine drainage remediation1.3 Research objectives

2 Experimental methods2.1 Analytical techniques2.2 Preparation of an artificial AMD2.3 Testing the ratio of AMD to slag2.4 Testing the optimum time of contact between the AMD and slag2.5 Testing the maximum capacity of the slag using continuous AMD flow

3 Results and discussion3.1 Slag characterisation3.2 Determining the effect of the ratio of AMD to slag3.2.1 The pH changes of acid mine drainage3.2.2 Reduction of iron and sulfate

3.3 Determining the optimum contact time between the AMD and slag3.4 Testing the maximum treatment capacity of SS and BOF slags using continuous flow reactors3.4.1 Measuring the effect of flow rate on pH3.4.2 Effect of Flow rate on reducing iron concentration in acid mine drainage3.4.3 The effect of flow rate on sulfate concentration

3.5 Calculating the effect of residence time on reducing iron and sulfate in AMD

4 Predicting slag requirements for treating different quantities of different strength AMD streams5 Conclusions and recommendationsAcknowledgementsReferences