Removal of heavy metals from mine water by ... - Springer Link

13 downloads 0 Views 460KB Size Report
Removal of heavy metals from mine water by carbonate precipitation in the. Grootfontein-Omatako canal,. Namibia. Michael O. Schwartz ´ Dieter Ploethner.
Cases and solutions

Removal of heavy metals from mine water by carbonate precipitation in the Grootfontein-Omatako canal, Namibia Michael O. Schwartz ´ Dieter Ploethner

Abstract Water from the Kombat mine was delivered to the Omatako dam via the 263-km-long Grootfontein-Omatako canal during test runs in 1997. It is intended to supply water from Kombat and other underground mines in the Otavi Mountain Land to the capital Windhoek. The Cu-Pb-Zn orebodies are hosted by carbonate rocks and the mine waters are supersaturated with respect to calcite and CO2. Along the length of the canal, the CO2 partial pressure drops from 10 ± 2.1 atm at the inlet of the Kombat mine to 10 ± 3.5 atm at the end of the canal. This is accompanied by a drop in Ca concentration from about 60 to about 20 mg/l. The heavy metal concentrations (Cd, Cu, Mn, Pb and Zn) drop along the course of the canal to values far below the national drinking-water standard. Scavenging by calcium carbonate precipitation is the major depletion mechanism.

Introduction The Grootfontein-Omatako canal is part of the Eastern National Water Carrier, linking the Otavi Mountain Land in northern Namibia and the capital Windhoek (Fig. 1). In 1997 the canal offered a good opportunity to study large-scale heavy metal scavenging by precipitating carbonate. Water from a single source, the copper-lead-zinc underground mine at Kombat, was delivered to the canal in 1997, when the chemical changes along its course were investigated.

Received: 21 June 1999 ´ Accepted: 29 August 1999 M.O. Schwartz ()) ´ D. Ploethner Bundesanstalt für Geowissenschaften und Rohstoffe, P.O. Box 510153, D-30631 Hannover, Germany e-mail: [email protected]

The Kombat mine can yield 5 million m3 water per year. Water from other underground mines in the Otavi Mountain Land also constitutes a large potential source of drinking water. The Tsumeb mine, active at the time of sampling but closed since 1998, and the dormant Berg Aukas, Abenab and Abenab West mines have a total yield of about 15 million m3 water per year (Fig. 2, Table 1). The waters from the dormant mines are earmarked to be supplied to the canal. These waters were also sampled in order to make predictions about their chemical behaviour. This research was conducted under a technical co-operation program between Namibia and Germany, commissioned by the Federal Ministry of Cooperation and Development (BMZ). The partners in the groundwater exploration project were the Department of Water Affairs (DWA, Windhoek), Namibian Water Corporation and the Bundesanstalt für Geowissenschaften und Rohstoffe (BGR, Hannover).

Mines in the Otavi Mountain Land The Otavi Mountain Land in northern Namibia is a dolomitic massif rising up to 500 m above the surrounding plain. Its four synclines composed of fractured dolomite comprise one of the most important groundwater sources of Namibia despite the fact that their recharge depends on rare, excessive rainfall events. The mean annual rainfall of 550 mm is high compared to the country-wide average of 270 mm (United Nations 1986). It is difficult to construct wells with a high yield in the fractured dolomite. However, the cavities produced by underground mining represent easily accessible groundwater. The polymetallic orebodies were emplaced along hydraulically favourable structures such as paleokarst and fault conduits (Ploethner and others 1998). The degree of supergene oxidation of the sulphide ore prior to mining varies considerably and may reach unusual depths of 1500 m, as shown by the Tsumeb deposit (Table 1). Despite the relatively high proportion of oxidized ore, mobilisation of heavy metals and hence polluEnvironmental Geology 39 (10) September ´  Springer-Verlag

1117

Cases and solutions

Fig. 1 Location of the Grootfontein-Omatako canal (GOC), Namibia

Fig. 2 Location of mines in the Otavi Mountain Land, Namibia

tion of mine waters, is strongly inhibited by the ubiquitous presence of carbonate.

of which 203 km are open and concrete-lined and the remainder consists of 23 inverted siphons or underground pipeline. The canal has a maximum width of 3.7 m and a maximum depth of 1.65 m. It is a gravity flow canal with a gentle slope of about 1 : 3000. When The Grootfontein-Omatako canal full, the flow velocity will be 0.8 m/s. At the time of sampling in March 1997, the estimated flow velocity was The Grootfontein-Omatako canal was built between 1981 0.4 m/s at km 80.21 (Ploethner and others 1998). and 1987. The Kombat mine is connected to the canal via Water from the Kombat mine has been delivered to the a 20-km-long pipeline. The length of the canal is 263 km, Omatako dam during test runs in 1997. It is intended to

1118

Environmental Geology 39 (10) September ´  Springer-Verlag

Cases and solutions

Table 1 Geological characteristics of Pb-(Cu-Zn-V) deposits in the Otavi Mountain Land, Namibia Kombat mine

Tsumeb mine

Berg Aukas mine

Abenab West mine/ Abenab mine

Status at the time of water sampling Concentration of ore metals (production or ore reserves) Depth of mining

Operating

Dormant

Dormant

Dormant

3.1% Cu, 1.1% Pb, (Zn)

4.0% Pb, 16.8% Zn, 0.5% V 800 m

16% Pb, 37% Zn, (V)

800 m

3.1% Cu, 3.8% Pb, 1.2% Zn, 0.04% Cd 1800 m

Type of ore

Oxidized (0±60 m), sulphidic/oxidized (>60 m)

Oxidized/sulphidic (0±1500 m), Sulphidic/oxidized sulphidic (>1500 m) (0±800 m)

References for ore deposit data

Geological Survey (1992), commun. by mine staff

Expected ground water yield (million cubic meters/year)

5

Schneider (1991), Geological Survey (1992), Gebhard and Schlüter (1995) 6±9

supply water from Kombat and other underground mines in the Otavi Mountain Land to the capital Windhoek during long droughts, when surface water becomes scarce. Up to now, Kombat mine water has been treated at Okakarara in order to meet rural water demand in that area. There have been plans to link the canal via a 240-kmlong pipeline to the Okavango River. Disastrous consequences for the flora and fauna in this unique game reserve would result if the Okavango water were diverted to Windhoek. All alternatives, including those described in this paper, deserve to be seriously evaluated.

Geological Survey (1992)

380 m (Abenab West) 215 m (Abenab) Oxidized/sulphidic (Abenab West), mainly oxidized (Abenab) Verwoerd (1957), Geological Survey (1992)

Up to 4.5

3±4

1997 were filtered through a 0.45-mm membrane prior to acidification. The minor elements were determined by inductively coupled plasma mass-spectrometry (ICP-MS). The water samples for analysis of major elements were not filtered or acidified. The analyses of unfiltered samples tend to overestimate the concentration of heavy metals, because mine waters contain suspended particles, which may dissolve during acidification prior to laboratory analysis. In contrast, the analyses of filtered samples tend to underestimate the concentrations of heavy metals because scavenging by precipitating phases (such as carbonates) during filtration takes place.

Sampling and analytical methods The water samples of the Kombat mine were taken from the raw water reservoir of the main shaft (sample 1008) and from the pipeline inlet into the Grootfontein-Omatako canal (sample 1007 at km 234.29). The pumping rate was 1400 m3/h in March 1997. The water sample of the Tsumeb mine was collected from Shaft No. 1 in July 1997 when the pumping rate was 220 m3/h. The water sample from the Berg Aukas mine was taken with a 5-l PVC bailer at 743 m below water level. The same bailer was used for sampling the Abenab West and Abenab mines at 80±100 m below water level. Redox potential, pH, dissolved oxygen and water temperature were determined in situ using WTW electrodes plugged into a 1-l-flow cell at an approximate flux of 10 ml/s. Alkalinity (HCO3± and CO3± 2) and dissolved CO2 were determined by potentiometric titration, applying 0.1-N HCl and 0.05-N NaOH solutions. The water samples taken in March 1997 from the Kombat and Berg Aukas mines and from the Grootfontein-Omatako canal, which were analysed for minor elements in the BGR laboratory, were acidified with 1 vol. % HNO3 (65%) to pH < 2; these samples were not filtered. In contrast, the water samples for minor element analysis taken at the Tsumeb, Abenab West and Abenab mines in July

Chemical composition of mine water The water samples from the underground mines have high partial pressure of CO2 (10 ± 1.3 ± 10 ± 2.3 atm), i.e. considerably higher than that of atmospheric CO2 (10 ± 3.5 atm). The waters are also supersaturated with respect to calcite and dolomite (Table 2). The water from the different mines varies considerably with respect to heavy metal concentration (Cd, Cu, Mn, Pb and Zn), partly reflecting the composition of the respective ore. This paper focuses on geochemical modelling; it is based on samples taken in March and July 1997 and analysed in the BGR laboratories. Water samples taken between 1980 and 1995 and analysed in other laboratories are reviewed by Ploethner and others (1998). They are not included in the modelling for the sake of analytical consistency. However, the predictions would not be significantly different if they were based on analyses from other laboratories.

Environmental Geology 39 (10) September ´  Springer-Verlag

1119

Cases and solutions

Table 2 Chemical analyses of water from the Grootfontein-Omatako canal, the Kombat and Berg Aukas mines (sampling period 1±4 March 1997) and the Tsumeb, Abenab West and Abenab mines Grootfontein-Omatako canal (km ~0.3), downstream of Omatako Dam by-pass

GrootfonteinOmatako canal (km 80.21)

Sample 1004 1005 Sampling date 1.03.97 02.03.97 Field measurements Temperature [ C] 28.7 28.6 pH 8.98 8.84 pe 5.34 5.51 ±3.55 ±3.43 log(pCO2)a ± 299.4 280.3 HCO3 (mg/l, total aqueous C) Filtration No No Major element concentrations (mg/l) [laboratory analyses] K 13.5 14.77 Na 15.51 14.77 Cl 18.34 19.1 Mg 53.6 49.5 Ca 19 25.9 49.4 62.2 SO±2 4 (total S) 17.9 17.7 SiO2 (total Si) Minor element concentrations (mg/l) [laboratory analyses] Al 0.12 0.083 Cd 0.00013 0.000090 Cu 0.0037 0.0018 Fe 0.10 0.081 Mn 0.010 0.012 P 0.021 0.012 Pb 0.0011 0.00063 Zn 0.026 0.0041 a Saturation index Aragonite 11.03 11.02 Azurite 1±7.01 1±7.73 Calcite 1 1.17 1 1.17 Cerusite 1±2.68 1±2.84 Chalcedony 1±0.09 1±0.07 Pyromorphite ±10.19 ±11.14 Cuprous ferrite 10.06 1 9.68 Dolomite 1 3.21 1 3.02 Hydrous ferric oxide 11.39 11.35 Goethite 17.28 1 7.24 Hydroxyapatite 10.86 10.55 Magnesite 11.45 11.27 Malachite 1±3.53 1±4.05 Otavite 1±0.91 1±1.18 Rhodochrosite 1±0.54 1±0.48 Siderite 1±5.89 1±5.83 Strengite 1±4.51 1±4.44 Tenorite 1±1.07 1±1.39 Smithsonite 1±2.52 1±3.20 Willemite 1±0.35 1±1.96 a

GrootfonteinOmatako Canal (km 234.29), inlet of Kombat Mine pipeline

Kombat mine (Main shaft)

Berg Aukas mine (Shaft 2)

Tsumeb mine (Shaft 1)

Abenab West mine

Abenab mine

1007 03.03.97

1008 03.03.97

1010 04.03.97

1020 22.07.97

1022 22.07.97

1024 22.07.97

25.5 7.81 7.26 ±2.08 590.7

26.3 7.95 7.07 ±2.26 526.6

27 7.15 6.21 ±1.35 746.3

28.2 6.99 3.86 ±1.55 366.2

24.7 7.06 6.40 ±1.25 813.3

20.9 7.06 1.886 ±1.3 772.1

No

No

No

Yes

Yes

Yes

13.5 14.17 16.41 47.5 61.1 50.1 17.7

14.96 13.81 16.49 44.8 55.6 40.9 18.2

111.35 114.94 115.1 156.8 100.7 121.5 120.7

1121.1 1160 1178.6 1155 1454 1807 1120.5

10.7 15.8 12.9 94.5 71.8 17.11 28.8

11.7 19.7 15.1 97.3 54.6 0.55 24

1.62 0.0016 0.59 1.9 1.00 0.16 1.21 0.15

1.59 0.0011 0.45 1.72 0.59 0.45 0.27 0.109

0.0089 0.0015 0.0020 0.042 0.0032 0.0076 0.019 0.70

1 decrease and distribution coefficients < 1 increase, as precipitation rates increase; in both cases, the distribto the left and produces calcium carbonate precipitation. ution coefficient approaches 1 (Lorens 1981; Tesoriero The Mg concentrations remain nearly unchanged (apand Pankow 1996). Table 3 lists the experimental DMe(approximately 50 mg/l), and the degree of supersaturation parent) values for Cd and Mn for low precipitation rates with respect to dolomite increases along the course of the [about 0.03 nmolCa/(min*mgcalcite)] and high precipitation canal. Precipitation of dolomite is inhibited for kinetic rates [about 300 nmolCa/(min*mgcalcite)]. The precipitation reasons (Wedepohl 1974). rate increases with the saturation index of calcite, defined Heavy metal concentrations drop along the length of the as the logarithm of the ratio between the product of the canal. Scavenging by precipitating carbonate is the major Ca+2 and CO3± 2 activities in the solution to the solubility depletion mechanism for heavy metals. Note that the product of calcite (Parkhurst 1995). The exact relationauthors do not approve of the canal being used as an ships have not been established experimentally, but the exclusive dump site for heavy metals. It is feasible to pre- data of Lorens (1981) suggests that high precipitation cipitate sufficient carbonate in a settling pond prior to rates [about 300 nmolCa/(min*mgcalcite)] are compatible delivering the water to the canal. This paper elucidates with a calcite saturation index of about 1, i.e. the satthe heavy metal removal by carbonate precipitation. The uration index for calcite along the whole length of the results are intended to be used in the layout of settling Grootfontein-Omatako canal (Table 2). ponds. For Cu and Zn, the exact relationships between the distribution coefficient and precipitation rate are not known. The experiments of Kitano and others (1980) yield a high distribution coefficient for Zn [DZn(apparent)=25], whereas Distribution coefficients Crocket and Winchester (1966) determined a low partition coefficient [DZn(apparent)=4]. Different set-ups were The mineralogical composition of the precipitate in the used for the experiments. It can be assumed that DZn(apcanal is not known. Thermodynamically stable calcite or parent)=25, which is close to the thermodynamic distrib-

Fig. 3 Ca concentration, pH and log partial CO2 pressure (atm) of the water in the Grootfontein-Omatako canal. Only water pumped from the Kombat mine entered the canal (at km 234.29)

Environmental Geology 39 (10) September ´  Springer-Verlag

1121

Cases and solutions

Table 3 Distribution coefficients DMe‡2 and DMe(apparent) for the partitioning of divalent trace metals between liquid and calcite at 25 C. The underlined apparent distribution coefficients are

Ba Cd Co Cu Fe Mg Mn Pb Ra Sr Zn

Low precipitation rate (or equilibrium)

High precipitation rate (nonequilbrium)

DMe‡2

References

DMe‡2

Tesoriero and Pankow (1996) Lorens (1981), Tesoriero and Pankow (1996) Lorens (1981) Kitano and others (1980) Dromgoole and Walter (1990) Hartley and Mucci (1996) Lorens (1981) [See text for explanation] Gnanapragasam and Lewis (1995) Tesoriero and Pankow (1996) Kitano and others (1980)

0.053

0.012 1240 3.7 0.018 180±910 0.021

DMe(apparent) 70 18.4 25 51 10.013 25

ution coefficient (see following section), corresponds to a low precipitation rate and DZn(apparent)=4 to a high precipitation rate. In an analogous manner, it is assumed that DCu(apparent)=25 (Kitano and others 1980), which is close to the thermodynamic distribution coefficient, corresponds to a low precipitation rate. The extrapolation of this distribution coefficient to a high precipitation rate corresponds approximately to DCu(apparent)=4, if a relationship analogous to those of Cd, Mn or Zn is assumed. The theoretical Pb distribution coefficient Since no experimental data is available for Pb partitioning, DPb‡2 has to be estimated. The distribution of Ca+2 and a divalent cation Me+2 between a liquid and a solid phase with end-member compositions CaCO3 and MeCO3 (e.g. PbCO3) is described by the equilibrium CaCO3(solid)+Me+2(liquid)=MeCO3(solid)+Ca+2(liquid)

(4)

The equilibrium constant (Kd) for this reaction is termed thermodynamic distribution coefficient. The standard states of CaCO3 and MeCO3 are the pure end-members CaCO3 (calcite) and MeCO3 (e.g. PbCO3) at the temperature of interest. The solubility product constants KCaCO3 and KMeCO3 for the reactions CaCO3(solid)=Ca+2(liquid)+CO3± 2(liquid)

(5)

and MeCO3(solid)=Me+2(liquid)+CO3± 2(liquid)

Kd ˆ KCaCO3 =KMeCO3

…7†

Assuming that the Me+2/Ca+2 concentration ratios equal the Me+2/Ca+2 activity ratios, the activity coefficient of the solid phase trace component (gMeCO3 ) can be calculated from the thermodynamic distribution coefficient and the true distribution coefficient, assuming the activity of the CaCO3 component (gCaCO3 ) is unity: Environmental Geology 39 (10) September ´  Springer-Verlag

1.9

0.14

DMe(apparent)

References

10 12 14

Tesoriero and Pankow (1996) Lorens (1981) Lorens (1981) [See text for explanation] Dromgoole and Walter (1990)

16 10

Lorens (1981) [See text for explanation]

14

Tesoriero and Pankow (1996) Crocket and Winchester (1966)

gMeCO3 ˆ Kd =DMe‡2

…8†

Experimental true distribution coefficients (DMe‡2 ) are available for Ba, Cd, Fe, Mg and Sr. Experimental apparent distribution coefficients [DMe(apparent)] are available for another five metals (Co, Cu, Mn, Ra and Zn) (Table 3). The activity coefficient (gMeCO3 ) increases with the absolute value of the difference in ionic radius |rCa ± rMe| between Ca+2 and Me+2. Small cations (Cd+2, Co+2, Cu+2, Fe+2, Mg+2, Mn+2 and Zn+2) correspond to a smaller increase of the activity coefficient than large cations (Ba+2, Ra+2 and Sr+2), probably because the calcite lattice accommodates small cations more readily than large cations (Rimstidt and others 1988). Plotting gMeCO3 versus [rCa ± rMe], the value for gPbCO3 can be interpolated by cubic regression (Fig. 4). The value gPbCO3 ˆ 102:25 is obtained using the curve Table 4 Effective ionic radius (Š) of divalent metals in sixfold co-ordination (Shannon 1976), solubility product constant (log KMeCO3 and log KCaCO3 ) of divalent metal carbonates and activity coefficient (log gMeCO3 ) of the solid phase trace component in calcite at 25 C. The activity coefficient is calculated from KMeCO3 and KCaCO3 as well as DMe‡2 or DMe(apparent) (in parentheses) in Table 3 Cation

Ionic radius

logK

log gMeCO3

References for logK

Ba+2 Ca+2 Cd+2 Co+2 Cu+2 Fe+2 Mg+2 Mn+2 Pb+2 Ra+2 Sr+2 Zn+2

1.35 1.0 0.95 0.65 0.73 0.61 0.72 0.67 1.19 1.37 1.18 0.74

1±8.57 1±8.29 ±11.41 1±9.54 1±9.92 ±10.50 1±7.46 ±10.61 ±12.8 1±9.58 1±9.18 1±9.92

2.19

HSC Chemistry HSC Chemistry HSC Chemistry HSC Chemistry HSC Chemistry HSC Chemistry Lindsay (1979) HSC Chemistry HSC Chemistry HSC Chemistry HSC Chemistry HSC Chemistry

(6)

are conveniently used to calculate the thermodynamic distribution coefficient:

1122

used for predicting the composition of the water in the Grootfontein-Omatako canal

0.027 (0.33) (0.23) 1.64 0.91 (0.61) (3.18) 2.57 (0.23)

(1997) (1997) (1997) (1997) (1997) (1997) (1997) (1997) (1997) (1997) (1997)

Cases and solutions

Predicting the chemical changes in water from the Kombat mine The heavy-metal ± calcium ratio of the water changes along the length of the canal and, accordingly, the heavymetal ± calcium ratio of the precipitating calcite. This case is described by the heterogeneous distribution law of Doerner and Hoskins (1925; quoted by McIntire 1963): ln[Meinitial/Mefinal)]=D ln[Cainitial/Cafinal],

Fig. 4 Log of the activity coefficient (log gMeCO3 ) of the solid phase trace component (MeCO3) in calcite versus difference in ionic radii between Ca+2 and Me+2 (rCa ± rMe). The solid circles refer to activity coefficients derived from DMe‡2 and the open circles refer to activity coefficients derived from DMe(apparent) (Table 3 and 4). The dotted line is the cubic regression of data derived from DMe‡2 and the solid line is the cubic regression of all data

(9)

where [Cainitial/Cafinal] is the ratio between Ca concentration in the water before crystallization of calcite (mine water reservoir) and Ca concentration at the end of crystallization of calcite (at a given position along the canal) and [Meinitial/Mefinal] is the corresponding heavy-metal concentration ratio. The Kombat mine water has a Ca concentration of 55.6 mg/l. After flowing 154 km through the canal (from km 234.29 to km 80.21), the Ca concentration decreases to 25.9 mg/l. The concentration ratio 55.6/25.9=2.15

has been used to predict the heavy metal concentrations in the canal at km 80.21 (Fig. 5). The modelling produces satisfactory results for Mn and for gMeCO3 derived from DMe‡2 (dotted line). The same Zn (Fig. 5, prediction A). The modelling overestimates value for gPbCO3 corresponds to the regression of all slightly the depletion of Pb (by 6 times), whereas the gMeCO3 (solid line), which are derived from both DMe‡2 depletion of Cd is considerably overestimated (by 180 and DMe(apparent). times). Predicted and actual concentration are in agreeThe distribution coefficient DPb‡2 is calculated from ment for DPb(apparent)=8 (instead of 10) and DCd(appargPbCO3 ˆ 102:25 , using the solubility product constant for =3.3 (instead of 10). The mine water is supersaturated ent) PbCO3, which is in the range of 10 ± 13.5 ± 10 ± 12.8 according with respect to the phosphates hydroxyapatite, pyromorto various authors (reviewed by Van©t Dack and others phite and strengite (Table 2). It is possible that precipi1983), and the solubility product constant for CaCO3 tating pyromorphite also removes some Pb. (10 ± 8.29): The predicted Cu concentration is higher than the actual value. Variations in Cu concentrations cannot be attribDPb+2=10 ± 8.29+13.5 ± 2.25=910 uted to calcite precipitation alone. The mine water is or supersaturated with respect to azurite, cuprous ferrite, +2 ± 8.29+12.8 ± 2.25 malachite and tenorite (Table 2). Precipitation of these DPb =10 =180. phases may remove Cu dissolved in the water. These very high values correspond to low precipitation The effect of adsorption on hydrous ferric oxide is only rates or equilibrium. Using a different interpolation pro- minor. The Fe concentration changes from 1.72 mg/l at cedure, Rimstidt and others (1998) estimated a Pb disthe Kombat mine to 0.081 mg/l along the course of the tribution coefficient of 2630 at equilibrium. Note that a canal, or from 0.031 mmol/l to 0.0015 mmol/l. Assuming very high distribution coefficient has also been deterthat all Fe is removed by precipitation of hydrous ferric mined for Cd near equilibrium conditions (DCd‡2 ˆ 1240; oxide there are Tesoriero and Pankow 1996), whereas DCd(apparent) is as 0.0308 ± 0.0015=0.029 mmol/l Fe(OH)3 low as 10 at high precipitation rates (Lorens 1981). In analogy to Cd, DPb(apparent)=10 is considered to be a suit- available for complexing at the surface. Using the diffuse able value for the modelling carried out below (see double-layer model of Dzombak and Morel (1990), the underlined distribution coefficients in Table 3). adsorption effect has been calculated with the program y=a+bx+cx2+dx3

PHREEQC (Parkhurst 1995). The calculation is based on the assumptions: l area per mass of surface material=600 g/m2 l moles of weak binding sites per mol Fe=0.2 l moles of strong binding sites per mol Fe=0.005. If the adsorption of hydrous ferric oxide is taken into consideration in addition to calcite scavenging, the pre-

Environmental Geology 39 (10) September ´  Springer-Verlag

1123

Cases and solutions

Fig. 5 Histogram showing the heavy metal concentration of water in the Kombat mine as well as the predicted and actual composition of the water in the Grootfontein-Omatako canal at km 80.21. Prediction A takes only heavy-metal removal by calcite precipitation into consideration. Prediction B takes both adsorption on hydrous ferric oxide and calcite precipitation into consideration. See text for explanation

dicted heavy-metal concentrations (prediction B) are little different from those based on calcite precipitation alone (prediction A). Prediction B produces slightly lower values for Cu than prediction A, whereas the values for the other heavy metals are nearly identical (Fig. 5).

Predicting chemical changes in water from other mines The Tsumeb, Berg Aukas, Abenab West and Abenab mines have not supplied water to the Grootfontein-Omatako canal yet. The expected chemical changes undergone by the water from these mines, as it flows through the canal, have been modelled using two different procedures (Fig. 6): l Prediction A (theoretical model) is based on calcite precipitation alone, using the apparent distribution coefficients in Table 3 (underlined values) and the calcium depletion ratio [Cainitial/Cafinal]=2.15 (same procedure as used for the case of the Kombat mine). l Prediction C (empirical model) uses the heavy-metal depletion ratio observed for water from the Kombat mine, e.g.

concentrations are only critical in the case of the Tsumeb mine. However, even the empirical model forecasts a Cd concentration close to the national drinking-water standard (0.01 mg/l Cd) after the water from the mine has passed through the Grootfontein-Omatako canal. For Cu, Mn, Pb and Zn, drinking-water quality is predicted by both models (