Applicability of passive compost bioreactors for treatment of extremely acidic and saline waters in semi-arid climates

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Applicability of passive compost bioreactors for treatment of extremely acidic and saline waters in semi-arid climates Vera Biermann a,e, Adam M. Lillicrap a,b,e,*, Claudia Magana c, Barry Price d, Richard W. Bell c, Carolyn E. Oldham e a

Centre of Excellence for Ecohydrology, University of Western Australia, c/o DAFWA, 444 Albany Highway, Albany, WA 6330, Australia b Department of Agriculture and Food, Western Australia, 444 Albany Highway, Albany, WA 6330, Australia c School of Veterinary and Life Sciences, Murdoch University, 90 South Street, Murdoch, WA 6150, Australia d ChemCentre, PO Box 1250, Bentley Delivery Centre, WA 6983, Australia e School of Civil, Environmental and Mining Engineering, University of Western Australia, M015, 35 Stirling Highway, Crawley, WA 6009, Australia

article info

abstract

Article history:

Extremely acidic and saline groundwater occurs naturally in south-western Australia.

Received 25 December 2013

Discharge of this water to surface waters has increased following extensive clearing of

Accepted 6 February 2014

native vegetation for agriculture and is likely to have negative environmental impacts. The

Available online 15 February 2014

use of passive treatment systems to manage the acidic discharge and its impacts is complicated by the region’s semi-arid climate with hot dry summers and resulting periods

Keywords:

of no flow. This study evaluates the performance of a pilot-scale compost bioreactor

Acid drainage

treating extremely acidic and saline drainage under semi-arid climatic conditions over a

Bioremediation

period of 2.5 years. The bioreactor’s substrate consisted of municipal waste organics

Constructed wetland

(MWO) mixed with 10 wt% recycled limestone. After the start-up phase the compost

Semi-arid landscape

bioreactor raised the pH from
3.7 to 7 and produced net alkaline outflow for 126 days.

Sulfate reducing bacteria

The bioreactor removed up to 28 g/m2/d CaCO3 equivalent of acidity and acidity removal

Water treatment

was found to be load dependent during the first and third year. Extended drying over summer combined with high salinity caused the formation of a salt-clay surface layer on top of the substrate, which was both beneficial and detrimental for bioreactor performance. The surface layer prevented the dehydration of the substrate and ensured it remained waterlogged when the water level in the bioreactor fell below the substrate surface in summer. However, when flow resumed the salt-clay layer acted as a barrier between the water and substrate decreasing performance efficiency. Performance increased again when the surface layer was broken up indicating that the negative climatic impacts can be managed. Based on substrate analysis after 1.5 years of operation, limestone dissolution was found to be the dominant acidity removal process contributing up to

* Corresponding author. Centre of Excellence for Ecohydrology, University of Western Australia, c/o DAFWA, 444 Albany Highway, Albany, WA 6330, Australia. Tel.: þ61 8 98928518; fax: þ61 8 98412707. E-mail address: [email protected] (A.M. Lillicrap). 0043-1354/$ e see front matter ª 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2014.02.019

84

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

78e91% of alkalinity generation, while bacterial sulfate reduction produced at least 9e22% of the total alkalinity. The substrate might last up to five years before the limestone is exhausted and would need to be replenished. The MWO substrate was found to release metals (Zn, Cu, Pb, Ni and Cr) and cannot be recommended for use in passive treatment systems unless the risk of metal release is addressed. ª 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

Passive treatment systems for the remediation of acid mine drainage (AMD) have been developed in temperate climates (Younger et al., 2002; PIRAMID Consortium, 2003; Watzlaf et al., 2004). However, acidic surface and groundwaters also occur in semi-arid Mediterranean climates with hot dry summers and cool wet winters, such as in California (Druschel et al., 2004), Spain (Sa´nchez Espan˜a et al., 2005) and Australia (Lillicrap and George, 2010). The application of passive treatment systems to acidic waters in such environments is complicated by high evaporation in summer, which can cause surface water bodies to dry out, restricting periods of surface water flow to winter and spring. The evaporation affects water quality and needs to be considered in passive treatment system design in semi-arid regions (Tyrrell et al., 1997; Degens, 2012). Extremely acidic saline groundwater occurs naturally in the agricultural areas of south-western Australia known as the wheatbelt (Lillicrap and George, 2010), where nearly 47,000 km2 of valley floors and flat terrain have been identified as having a high probability of acidic saline groundwater (Holmes and Lillicrap, 2011). Clearing for agriculture disturbed the hydrological cycle and has led to increased recharge causing groundwater tables to rise (Hatton et al., 2003). Dryland salinisation and increased groundwater baseflow into surface waters are consequences of the rising water tables. Deep drains, greater than one metre in depth, are used to manage dryland salinity. However, the discharge of acidic saline groundwater from arterial drainage networks into surface waters can result in acidification up to 30 km downstream of the drain outlet (Seewraj, 2010) with negative impacts on ecosystems (Jones et al., 2009; Stewart et al., 2009). Acidic saline groundwater found in south-western Australia is similar to acid mine drainage (AMD) as it is characterised by pH values < 4 and high concentrations of metals such as aluminium, iron, lead, copper, nickel and zinc (Degens and Shand, 2010). The acidity, however, is not a result of pyrite oxidation but of ferrolysis (Mann, 1983; Lillicrap and George, 2010), with the sulfate in the system stemming from evaporation of marine aerosols (Bird et al., 1989). In addition, the groundwater is extremely saline with up to 100,000 mg/L of total dissolved solids dominated by sodium chloride (Degens and Shand, 2010). The adoption of AMD abatement technologies in southwestern Australia is hindered by the relatively low value of agricultural land and low profitability of many enterprises compared to mining. Active treatment technologies with high on-going costs are especially unsuitable on economic

grounds. Passive treatment systems generally require no continual input of energy or chemicals and less maintenance than active systems (PIRAMID Consortium, 2003). Most of the passive technologies for AMD were developed for temperate climates and hilly landscapes, where water is often held in open ponds and where treatment is based on the natural hydraulic gradient providing the necessary flows. However, the use of passive treatment systems for acidic saline drainage in south-western Australia is complicated by extreme salinity, very low hydraulic gradients and high evaporation in summer (Tille et al., 2001). Generally, passive treatment systems based on vertical flow are preferable to horizontal flow systems as they are more efficient, present smaller surface areas for evaporation and are less likely to suffer from preferential flow (Younger et al., 2002). Vertical flow systems with dispersed alkaline substrates have been used successfully to treat AMD in semiarid landscapes (Caraballo et al., 2011). However, vertical flow systems require more than two metres hydraulic head to drive flow (PIRAMID Consortium, 2003). This makes them unsuitable for the Western Australian wheatbelt, where hydraulic gradients are in the order of 10e30 cm per 1000 m (Beard, 1999). Given the constraints of cost and low hydraulic head, horizontal flow compost wetlands or bioreactors would be the most suitable passive treatment technology. The efficacy of such treatment systems in a semi-arid climate with periods of no flow is poorly known (Degens, 2012). Initial assessments indicate that it might be possible to adapt passive treatment systems for AMD to the semi-arid, flat landscapes of south-western Australia (Degens, 2009). Laboratory experiments showed that acidic saline drainage found in Western Australia can be treated by bacterial sulfate reduction (Santini et al., 2010). Two pilot-scale horizontal flow sulfate reducing compost bioreactors treating acidic saline drainage with wheat straw and composted sheep manure achieved net alkaline outflow for up to 183 days and partial acidity removal for up to 2.4 years (Degens, 2012). High evaporation in summer was found to impact system performance but neither bioreactor was subjected to extended drying periods where the substrate dried out completely. Substrate availability appeared to be the main limiting factor (Degens, 2012). Neither system used limestone as an additional source of alkalinity. Compost derived from unsorted municipal solid waste (MSW) is a low-cost, relatively uniform and readily available organic substrate. Currently, Western Australia produces about 60,000 t of MSW-derived organics per year, for which markets are still under development. The suitability of this organic material for treating extremely acidic saline water has never been tested. This study evaluates the performance of a

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

pilot-scale compost bioreactor using a mixture of municipal waste organics (MWO) and recycled limestone to treat extremely acidic, saline drainage in the Western Australian wheatbelt under semi-arid climatic conditions that result in design acidity loads being exceeded and the reactor falling dry for part of the year due to high evaporation over summer.

2.

Materials and methods

A pilot-scale surface flow compost bioreactor, from here on called the bioreactor, was built in May 2009 and monitored for 2.5 years until December 2011.

2.1.

Site

The bioreactor is located off Dongolocking Creek near the town of Dumbleyung, 300 km south-west of Perth (Fig. 1). The area has an average annual rainfall of 400 mm and around 1800 mm annual evaporation. Evaporation is highest between November and March, while rainfall is highest from May to October, when it usually exceeds evaporation (Tille et al., 2001; Lillicrap and George, 2010). The mean maximum and minimum temperatures are 30  C and 15  C in summer (December to February) and 16  C and 6  C in winter (June to August). Dongolocking Creek receives acidic water from an arterial drainage system, which was installed in 2007. Since then the water quality and flow rates have been monitored by the West Australian Department of Water at a gauging station 1.8 km upstream of the bioreactor. In 2008, water abnormally flowed all year round, due to the new drainage system (Seewraj, 2010), with mean daily flow rates ranging from 2 to 542 L/s (DoW, 2010). From November 2007 to July 2009 Dongolocking Creek had a median pH of 3.1 and an average electrical conductivity (EC) of 56.6 mS/cm. The water quality varied considerably with time as it was influenced by the flow regime. At the start of the rainy season in late autumn/early winter (May/June) surface flow dominated with circumneutral pH and moderate salinity. Once the system became connected to the regional groundwater, pH decreased and salinity increased due to groundwater inputs. In winter, pH values ranged from 3.3 to 6.8 and EC values from 5.1 to 32.9 mS/cm. In summer, the water turned into an acidic brine with pH values between 2.4 and 3.2 and EC values between 60.6 and 98.3 mS/cm (DoW, 2010).

2.2.

Pilot-scale compost bioreactor

The bioreactor was designed based on summer water quality in Dongolocking Creek (Table 1) and published design guidelines for compost wetlands (Younger et al., 2002; PIRAMID Consortium, 2003). Assuming a worst-case total acidity concentration of 2000 mg/L CaCO3 equivalent the bioreactor was designed to operate at a minimum flow rate of 1 L/min. The required bioreactor surface area for the corresponding acidity load of 2.88 kg/day CaCO3 equivalent was 823 m2 based on the recommended design factor of 3.5 g CaCO3 equivalent acidity removal per square metre compost wetland area per day (Younger et al., 2002; PIRAMID Consortium, 2003).

85

The substrate of the bioreactor consisted of 300 m3 municipal waste organics, which met standards for land application, with 10% recycled limestone (by weight, C-Wise, Mandurah). Substrate selection was based on the results of laboratory experiments as attempts to predict the suitability of organic substrates to treat AMD based on chemical properties have had limited success (e. g. Prasad et al., 1999; Schmidtova and Baldwin, 2011). MWO were tested against five other organic substrates and it was found that MWO could effectively neutralise highly acidic saline groundwater (see Supplementary data). There was some indication that MWO had the potential to release metals such as nickel and chromium but the results were highly variable. MWO has been used as substrate in compost wetlands and no problems with metal release have been reported (Jarvis and Younger, 1999). Overall, MWO represented the most cost effective organic substrate and was selected for use in the bioreactor. The bioreactor was built parallel to Dongolocking Creek (Figs. 1 and 2). It is 125 m long, 9 m wide and 1.15 m deep, except for a 10 m long settling zone near the outlet, which is 2 m deep. The bioreactor was excavated into heavy clay, which eliminated the need for a liner. A 0.2 m high earthen weir was built across Dongolocking Creek to provide sufficient hydraulic head. Water was diverted from Dongolocking Creek into the bioreactor via an off-take channel, a stilling basin and two inlet pipes. The substrate was placed 6 m downstream of the inlet pipes to allow for flow to distribute evenly. The substrate layer was 0.5 m thick and 80 m long. The treated water was collected via a pipe system and returned to Dongolocking Creek via a discharge channel. The bioreactor was filled for the first time at the end of May 2009, but flow through the system only started on 01 July 2009 to allow for the establishment of sulfate reducing bacteria (SRB).

2.3.

Monitoring and maintenance

The water quality at the inlet and outlet of the bioreactor was monitored over 2.5 years from autumn/winter until spring/ summer, depending on water levels in Dongolocking Creek. Water samples and flow measurements were taken twice a month from 01 July 2009 to 07 January 2010 (year 1), every 3 weeks from 07 May 2010 to 20 October 2010 (year 2) and every 4 weeks from 25 July 2011 to 17 October 2011 (year 3). Field measurements included pH, EC, redox potential (Eh), total acidity and total alkalinity. Filtered and acidified water samples were sent for laboratory analysis. Flow into and out of the bioreactor was measured using the bucket and stopwatch method. Flow measurements were repeated at least three times and the average of all measurements was recorded. Residence time (tR) was calculated based on bioreactor dimensions, measured flow rates and the water level in the bioreactor assuming water flowed only in the water layer above the compost substrate (surface flow) with no preferential flow paths. The bioreactor substrate was sampled 20 m, 40 m, 60 m and 80 m downstream of the inlet on 08 December 2010, after the bioreactor had dried out for the second time. The substrate samples were placed into polyethylene bags and immediately frozen until they were sent for laboratory analysis.

86

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Maintenance work was carried out after flow had resumed in winter 2011. On 25 July, the inundated substrate was manually turned over to break up the substrate surface and to bring the deeper substrate layers in contact with the surface water. In November and December 2011, two storm events brought each more than 60 mm of rain within 24 h, flooding the bioreactor and burying the inlet pipes with silt, effectively rendering the system inoperable.

2.4.

Analytical methods

In the field, EC and pH were measured using a calibrated Eutech CyberScan PC 300 multimeter. Eh was determined with a HANNA Instruments HI 98120 ORP pocket meter. Eh values are reported relative to the standard hydrogen potential. Total acidity and total alkalinity were measured with a Hach Acidity Test Kit AC-6 and a HANNA Instruments Alkalinity Test Kit HI3811. Net acidity was calculated as the difference between

Fig. 1 e Schematic (A, plan view, not to scale) and location (B) of the pilot-scale compost bioreactor, location of the sampling points for laboratory experiments (C, see Supplementary data).

87

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Table 1 e Water quality in Dongolocking Creek near the site of the pilot-scale compost bioreactor in summer. pH Dongolocking Creek (n ¼ 4) a b c

2.6b

EC

Aciditya

Na

Mg

Ca

K

Cl

SO2 4

Al

Fe

Mn

mS/cm

mg/L

mg/L

mg/L

mg/L

mg/L

mg/L

mg/L

mg/L

mg/L

mg/L

76.6c

1380

18,900

2830

289

151

34,800

4640

115

213

2.1

Calculated net acidity in mg/L CaCO3 equivalent (Hedin et al., 1994). Median of 13 samples. Average of 13 samples.

total acidity and total alkalinity, where positive net acidity indicates net acidic water and negative net acidity indicates net alkaline water. Water samples were analysed at the ChemCentre in Perth. Total concentrations of Al, B, Ba, Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, SO4_S and Zn were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES) using a Varian Vista-Pro Simultaneous ICP-AES with an axial torch. Chloride was measured colorimetrically with mercuric thiocyanate using an AquaKem Discrete Analyser. The original MWO substrate and the substrate samples taken in December 2010 were also analysed at the ChemCentre. The samples were dried at 80  C for 60 h and all results are reported on a dry weight basis. Subsamples of each sample were subjected to different analyses. Chloride and sulfate content were measured by ion chromatography after extraction with water. The total content of Al, B, Ba, Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Ni, Pb, S and Zn was determined by ICP-AES after ignition and fusion with NaOH. Acid neutralising capacity (ANC) and reduced inorganic sulfur (RIS) were determined using sieved subsamples (dp
2 mm) that had been ground to dp < 150 mm. ANC was determined by back titration according to Australian standard AS 4969.13-2009. RIS was determined as chromium reducible sulfur (CRS) according to Australian standard AS 4969.7-2008 by converting reduced inorganic sulfur to H2S with hot acidic Cr2Cl2 solution. The evolved H2S was trapped as ZnS and determined by titration. This method measures iron disulfides, elemental sulfur and metal sulfides such as FeS and ZnS. Outflow concentrations for the first year were corrected using chloride as a conservative ion to account for dilution and evaporation effects when comparing inflow and outflow concentrations. This correction was only applied where geochemical modelling with the software PHREEQC (Parkhurst and Appelo, 1999) using the Pitzer data base indicated that all chloride minerals were undersaturated (SI < 1)

and the charge imbalance was less than 5%. In addition, a mass balance approach was used. Based on the measured concentrations and flow rates, loads (g/min) into and out of the bioreactor were calculated and used for performance evaluation as loads are independent of evaporation and dilution effects.

3.

Results

The quality of the water being diverted from Dongolocking Creek into the bioreactor was subject to seasonal and annual variability (Table 2, Fig. 3). Inflow pH ranged from 1.9 to 7.0 and net acidity was between 4 and 3600 mg/L CaCO3 equivalent. Generally, the water was less saline and less acidic in winter (June to August) than in summer (December to February). Operation of the bioreactor started in the southern winter 2009 (July), to allow for system establishment under relatively low acidity conditions. Flow rates into the bioreactor varied (Table 2, Fig. 3). Note that inflow ceased when the water level dropped below the inlet pipes in early autumn 2010 (March) and in late spring 2010 (November). Flow out of the bioreactor roughly matched the inflow except for periods of increased evaporation from late spring to early autumn (November 2009 to March 2010 and October 2010 to March 2011, Fig. 3). During these periods the outflow ceased completely when the water level in the bioreactor dropped below the outlet pipe. The calculated residence time (tR) during periods of flow (except for the start-up phase) was one to two weeks (Table 2). In early March 2010 (autumn) and at the end of October 2010 (spring) evaporative losses caused the water level in the bioreactor to drop below the surface of the organic substrate. A
5 mm thick mixture of salt, clay and iron oxide minerals covered the rich black substrate (Fig. 4), which emanated the characteristic smell of hydrogen sulfide (H2S) gas.

Fig. 2 e Simplified cross section of the pilot-scale compost bioreactor (not to scale). Water flow is from left to right driven by hydraulic gradient.

88

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

After the start-up phase the bioreactor produced netalkaline outflow with pH values 6.7 until flow ceased in January 2010 (summer) (Table 2, Fig. 3). After the onset of autumn rains flow into (April 2010) and out of (May 2010) the bioreactor resumed. Some acidity removal was observed from the end of July 2010 (winter) onwards but the bioreactor outflow remained net acidic with pH values
5 until flow stopped again in October 2010 (spring). After maintenance work was carried out in winter (July) 2011 the bioreactor outflow had pH values  5.8 and very little net acidity (Table 2, Fig. 3). A mass balance approach was used to assess the bioreactor’s performance. Loads and area-adjusted loads for each monitoring interval were estimated using measured concentrations and flow rates (Table 3). Average removal rates and efficiencies were calculated for the start-up phase and three subsequent monitoring periods (autumn/winter to spring/ summer, Table 4). After the start-up phase, the acidity removal rate averaged 12.3 g/m2/d CaCO3 equivalent (126% acidity removal efficiency) during the first year of operation (Table 4). Acidity removal declined in the second year to an average removal rate of 0.85 g/m2/d CaCO3 equivalent (41% acidity removal efficiency). After maintenance works were carried out, acidity removal increased in the third year to 2.4 g/m2/d CaCO3 equivalent (99% removal efficiency, Table 4). There was a linear relationship between acidity removal rate and acidity load during the first year and the third year (Fig. 5). The maximum acidity removal rate of 28 g/m2/

d CaCO3 equivalent was observed when the acidity load was highest. Elevated calcium concentrations in the outflow indicated that calcite dissolution was generating alkalinity and contributing to acidity removal (Fig. 3). Although outflow sulfate concentrations did not decrease significantly (Fig. 3) the bioreactor emitted a strong characteristic odour of H2S gas suggesting that bacterial sulfate reduction was taking place. This was confirmed by the analysis of substrate samples taken in December 2010, which showed an increase of reduced inorganic sulfur by up to one order of magnitude compared to the original substrate (Fig. 6). CRS content was twice as high near the outlet as near the inlet. At the same time calcium and ANC content near the inlet had decreased by up to 48% and 58%, respectively, with 81% of the original ANC remaining near the outlet (Fig. 6). The contribution of calcite dissolution and bacterial sulfate reduction to alkalinity generation in the bioreactor during the first two wet cycles was calculated assuming that each of the substrate samples was representative of a 20 m long section of the bioreactor and that each section contained the same amount of substrate. The total ANC and CRS content of the bioreactor after 17 months of operation was compared to the original MWO substrate. The bioreactor lost 39% of its initial ANC, corresponding to the dissolution of 5.4 t CaCO3, and accumulated 488 kg of CRS. Bacterial sulfate reduction produces one mole of H2S and one mole of alkalinity as CaCO3 per mole of sulfate. H2S is

Table 2 e Water quality of the pilot-scale compost bioreactor inflow and outflow during start-up and three subsequent monitoring periods over 2.5 years of operation, average ± standard deviation (range). Parameter Flow

Unit L/min

Residence Days time Temperature  C EC

mS/cm

pH Eh

mV

Net acidity

mg/Lb

Net acidity load

g/m2/db

Aluminium

mg/L

Iron

mg/L

Calcium

mg/L

Sulfate

g/L

a b

Median pH. As CaCO3 equivalent.

Location

01 Jul 2009e 17 Aug 2009

17 Aug 2009e 07 Jan 2010

07 May 2010e 20 Oct 2010

25 Jul 2011e 17 Oct 2011

Inflow Outflow

19  31 17  25 26  30

(1.4e65) (3.1e54) (1.4e59)

5.1  1.1 3.3  2.3 16  4

(3.5e6.4) (0.0e6.3) (12e22)

5.8  3.4 5.4  2.3 14  6

(0.1e13.4) (0.1e8.0) (6.1e30)

12.1  2.6 11.4  2.9 6.9  1.2

(10.0e16.4) (8.4e16) (5.0e8.1)

Inflow Outflow Inflow Outflow Inflow Outflow Inflow Outflow Inflow Outflow Inflow

12.2  0.4 13.1  0.6 53  8 54  4 3.7a 4.9a 340  30 185  130 280  104 49  190 13  20

(11.8e12.5) (12.4e13.6) (44e59) (49e57) (3.2e3.7) (4.6e6.7) (315e360) (95e275) (160e340) (140e239) (0.4e36)

20.2  5.9 20.3  5.5 95  39 83  31 2.7a 7.3a 480  30 70  175 1249  1137 1308  1341 9.6  9.6

(12.5e29.2) (13.3e27.2) (53e165) (48e141) (1.9e3.7) (6.7e7.6) (430e510) (310e115) (240e3600) (3430 to 221) (1.0e32)

15.9  3.9 14.8  3.3 88  45 92  30 3.1a 4.6a 470  35 335  85 450  416 169  192 4.2  6.3

(11.5e22.9) (9.1e19.6) (43e191) (62e139) (2.2e3.6) (2.9e4.9) (405e520) (215e490) (90e1400) (30e660) (0.2e21)

16.6  4.2 16.3  3.3 60  21 63  16 4.6a 6.3a 260  160 105  30 136  165 1.8  15 2.6  3.0

(13.7e22.8) (13.8e21.1) (40e88) (46e80) (3.0e7.0) (5.8e7.3) (120e430) (70e130) (4e340) (16e16) (0.1e6.0)

Outflow Inflow Outflow Inflow Outflow Inflow Outflow Inflow Outflow

6.8  12.0 21  7 8.8  9.5 33  7 13  15 180  24 307  60 2.1  0.2 2.3  0.1

(0.7e21) (13e25) (1.8e20) (27e41) (1.0e30) (154e200) (238e349) (1.89e2.23) (2.16e2.40)

2.2  1.5 136  114 1.1  0.6 129  99 1.4  1.6 453  276 943  609 6.5  4.3 4.0  1.3

(4.7e0.0) (25e370) (0.2e2.1) (38e340) (0.14e4.3) (197e943) (509e2140) (2.63e15.3) (2.92e6.71)

1.5  1.9 58  59 24  28 22  18 11  10 383  191 646  248 5.0  4.0 4.8  1.8

(0.0e6.1) (9.3e200) (5.2e96) (5.9e62) (3.7e37) (189e735) (410e1210) (1.98e14.9) (2.79e8.23)

0.02  0.24 12  16 0.61  0.58 17  19 3.2  3.1 245  83 347  115 2.8  1.2 3.0  0.9

(0.26e0.23) (<0.005e33) (<0.005e1.3) (<0.005e38) (<0.005e7.2) (163e352) (210e461) (1.69e4.33) (2.02e3.95)

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

highly reactive and can undergo further reactions that lead to the formation of reduced inorganic sulfur species, which are accounted for as CRS. It was assumed that the CRS in the bioreactor consisted of only one of elemental sulfur (S0), mackinawite (FeS), greigite (Fe3S4) or pyrite (FeS2). Further it was assumed that bacterial sulfate reduction was the only source of H2S and that all H2S was converted into CRS to determine the amount of alkalinity that had to be generated by bacterial sulfate reduction to account for the accumulated

89

CRS as any of the four species. The required alkalinity ranged from 0.55 t CaCO3 for the formation of FeS to 1.5 t CaCO3 for the formation of S0. Therefore, calcite dissolution contributed 78e91% (5.4 t as CaCO3) of the total alkalinity generated during the first 17 months of operation of the bioreactor while bacterial sulfate reduction contributed 9e22% (0.55e1.5 t as CaCO3). The bioreactor retained aluminium, iron and cobalt (Figs. 6 and 7) but the average removal rates and efficiencies varied

Fig. 3 e Inflow and outflow water quality of the pilot-scale compost bioreactor between July 2009 and October 2011: flow rate (A), water temperature (B), electrical conductivity (C), redox potential (D), aluminium (E), calcium (F), pH (G), iron (H), sulfate (I), net acidity (J, in CaCO3 equivalents) and area-adjusted net acidity load (K, in CaCO3 equivalents). Note the use of two different y-axes in (J) and (K). Open symbols denote no flow, i.e. when the water level in the bioreactor was below the outlet pipe (outflow) or when the inlet pipe was blocked or the water level in the stilling basin was below the inlet pipe (inflow). Vertical dashed lines indicate substrate sampling on 08 December 2010 (e   e) and maintenance works on 25 July 2011 (e e e).

90

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Fig. 4 e Salt, clay and iron oxide minerals on top of the municipal waste organics substrate (black) in March 2010 (A) and in October 2010 (B).

considerably from year to year (Table 4). Aluminium and iron removal followed similar trends as acidity removal with best results in the first year. The analysis of substrate samples taken in December 2010 after two periods of flow showed that the aluminium content had more than doubled while the iron content had increased by more than 46% compared to the original substrate (Fig. 6). Total chromium concentrations in the outflow were higher than those in the inflow for most of the first year of operation, i.e. the compost substrate released chromium (Fig. 8). During the second and third year however, outflow and inflow chromium concentrations were mostly below the detection limit. In contrast, cobalt, nickel and zinc were

released at the start of all three periods of flow until removal commenced after two to four months (Fig. 8). Copper concentrations in the outflow were mostly below or near the detection limit (Fig. 8). For the minor elements, such as chromium and copper, with concentrations below or near the limit of reporting, a mass balance could not be calculated as the uncertainty in the data was deemed too high when combined with the uncertainty of the flow measurements. However, the analysis of the original substrate and substrate samples taken in December 2010 showed that zinc, copper, lead, nickel and to a lesser extent chromium had been leached from the substrate while cobalt had accumulated (Fig. 7).

Table 3 e Mass balance for loads into and out of the pilot-scale compost bioreactor during the monitoring intervals (mass in kg). Interval

01 17 17 14 29 14 30 17 04 21 07 01 18 06 23 16 07 29 25 25 22 a

Jul 09e17 Jul 09 Jul 09e17 Aug 09 Aug 09e14 Sep 09 Sep 09e29 Sep 09 Sep 09e14 Oct 09 Oct 09e30 Oct 09 Oct 09e17 Nov 09 Nov 09e04 Dec 09 Dec 09e21 Dec 09 Dec 09e07 Jan 10 May 10e01 Jun 10 Jun 10e18 Jun 10 Jun 10e06 Jul 10 Jul 10e23 Jul 10 Jul 10e16 Aug 10 Aug 10e07 Sep 10 Sep 10e29 Sep 10 Sep 10e20 Oct 10 Jul 11e25 Aug 11 Aug 11e22 Sep 11 Sep 11e17 Oct 11

Net aciditya

Aluminium

Iron

Calcium

Sulfate

In

Out

In

Out

In

Out

In

Out

In

Out

512 52 47 51 71 87 148 177 217 377 102 25 22 19 33 50 46 19 2.4 53 107

295 9.2 40 31 33 41 32 29 62 35 82 22 20 12 11 23 16 0.30 5.0 1.5 4.6

36 3.8 5.3 4.8 8.5 10 16 21 25 38 13 4.3 4.4 2.7 3.8 6.4 6.7 3.1 0.17 4.1 10

24 1.4 0.25 0.11 0.06 0.03 0.02 0.02 0.04 0.02 13 3.5 3.2 1.8 1.5 2.7 1.8 0.04 0.10 0.35 0.39

45 6.0 7.1 5.8 8.1 9.8 15 19 22 34 6.8 2.8 1.8 1.0 2.7 3.4 2.3 0.84 0.48 9.9 14

37 2.3 0.35 0.09 0.08 0.27 0.26 0.003 0.007 0.005 4.3 0.81 1.0 0.79 1.0 1.7 1.1 0.03 0.54 1.9 2.0

301 30 29 27 33 40 58 71 80 106 96 44 48 37 43 50 49 21 104 105 124

294 66 80 53 65 673 45 17 38 22 136 73 65 55 112 137 70 2 153 162 152

3329 357 410 373 458 537 786 978 1096 1605 976 451 512 394 492 626 609 264 1076 1166 1476

2827 482 470 314 405 435 249 56 120 69 1135 533 515 414 668 925 551 12 1331 1389 1320

Net acidity in kg CaCO3 equivalent.

91

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Table 4 e Average and range of acidity and metal removal by the pilot-scale compost bioreactor during start-up and three subsequent monitoring periods over 2.5 years of operation. Removal efficiency based on mass balance. Analyte

Monitoring period Concentration removal

Load removal

Area adjusted load removal

kg/d

g/m2/d

mg/l Net acidity

b

01 17 07 25 Aluminium 01 17 07 25 Iron 01 17 07 25 a b

4.

Jul 09e17 Aug 09 Aug 09e07 Jan 10 May 10e20 Oct 10 Jul 11e17 Oct 11 Jul 09e17 Aug 09 Aug 09e07 Jan 10 May 10e20 Oct 10 Jul 11e17 Oct 11 Jul 09e17 Aug 09 Aug 09e07 Jan 10 May 10e20 Oct 10 Jul 11e17 Oct 11

179 2536 184 123 9.3 125 26 10 12 117 10 14

(101e257) (593e6365) (8.0e834) (13e257) (4.4e14) (35e308) (2.0e124) (0.1e23) (0.1e24) (45e273) (0.9e29) (0.1e28)

7.5 10.1 0.74 2.1 0.41 0.94 0.10 0.17 0.32 0.88 0.06 0.25

(1.4e14) (3.1e24) (0.11e1.3) (0.24e4.1) (0.08e0.75) (0.18e2.2) (0.00e0.22) (0.00e0.37) (0.12e0.52) (0.24e2.0) (0.01e0.12) (0.00e0.47)

8.4 12.3 0.85 2.4 0.46 1.1 0.11 0.20 0.36 1.0 0.07 0.29

(1.6e15) (3.6e28) (0.13e1.5) (0.27e4.8) (0.09e0.84) (0.20e2.5) (0.00e0.44) (0.00e0.44) (0.14e0.58) (0.28e2.3) (0.02e0.39) (0.00e0.55)

Removal efficiencya % 46 126 41 99 36 100 38 94 23 99 50 81

(42e82) (109e185) (9.4e98) (96e308) (33e62) (95e100) (0.4e99) (40e96) (18e62) (95e100) (22e96) (13e85)

Removal efficiency: (total mass in inflow e total mass in outflow)/total mass in inflow during period. Net acidity as CaCO3 equivalent.

Discussion

The extreme salinity experienced by the pilot-scale compost bioreactor proved to be both beneficial and detrimental. During summer (DecembereFebruary), precipitated salts (mainly gypsum and halite) combined with clay to form a surface cover that reduced evaporation, maintained waterlogged conditions, prevented the substrate from drying out and reduced the risk of metal remobilisation. However, when flow resumed, the salt-clay cover acted as a physical barrier between the acidic water and the substrate, hindering diffusion, the rate-limiting step in surface flow systems. Consequently, the residence time was too short and performance decreased. Substrate analyses after the second period of flow confirmed that substrate exhaustion was not the main reason for the poor performance as the substrate still contained alkalinity (Fig. 6). Breaking up the salt-clay surface layer at the start of the third year of operation increased acidity removal to 99% (4.8 g/m2/d CaCO3 equivalent), indicating that compost bioreactors can perform well even when subjected to extended drying periods over summer. While both limestone dissolution and bacterial sulfate reduction contributed to acidity removal, limestone dissolution was the dominant process. Limestone dissolution was facilitated by the presence of the organic substrate as the decomposition of organic matter increases the partial pressure of CO2, which in turn increases the solubility of limestone. Further, the decomposition of the organic substrate created reducing conditions that prevent armouring of limestone surfaces with iron hydroxyoxides (Rose, 2007). Therefore, a treatment system using limestone combined with organic matter will be more efficient than a system that uses only limestone or other carbonate materials as substrate. Moreover, the organic substrate promotes bacterial sulfate reduction, which contributed at least 9e22% to the overall alkalinity generation of the bioreactor. This figure, based on accumulated CRS, is a conservative estimate as not all H2S produced by SRB is converted to CRS.

The pilot-scale compost bioreactor achieved similar acidity removal results than a semi-active compost bioreactor operating in similar climatic conditions (Degens, 2012), even though the latter suffered less variations in inflow quality and did not dry out over summer due to pumping from a storage basin. Moreover, the water level was kept just below the substrate surface forcing flow through the substrate (Degens, 2012). This means that the surface flow system described here was more efficient as it achieved comparable results with less contact between water and substrate. The compost bioreactor described by Degens (2012) contained a purely organic substrate, consisting of 4.7 vol% composted sheep manure layered with wheat straw, without any limestone. Thus, it relied totally on bacterial sulfate reduction for acidity removal rather than a combination of sulfate reduction and limestone dissolution. Acidity removal in the bioreactor described here is likely to decrease once the limestone in the substrate is exhausted. Based on the observed ANC decrease this would be expected after about five years, when the substrate would need to be replenished. The area-adjusted average acidity removal rates in our bioreactor were comparable to those reported for four to ten year old compost wetlands in the U.S.A., which range from 0 to 47.6 g/m2/d CaCO3 equivalent (Skousen and Ziemkiewicz,

Fig. 5 e Net acidity removal rate versus inflow net acidity load in g/m2/d CaCO3 equivalents.

92

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Fig. 6 e Chromium reducible sulfur concentration (A), acid neutralisation capacity as CaCO3 (B) and aluminium and iron concentration (C) of substrate samples taken from the pilot-scale compost bioreactor 20 m, 40 m, 60 m and 80 m downstream of the inlet in December 2010 after 17 months of operation compared to the original municipal waste organics (MWO) substrate.

2005). The positive correlation between acidity removal rate and acidity load (Fig. 5) is similar to that observed by Matthies et al. (2010) for a reducing and alkalinity producing system (RAPS). Matthies et al. (2010) suggested that load limitation might occur with sufficiently long residence times. Under this hypothesis compost bioreactors should be able to treat higher acidity loads associated with evapoconcentration effects in semi-arid climates. Ideally, flow rates should be regulated so that the water level in the bioreactor remains above the substrate surface but below the outlet pipe, thus maximising the residence time. Aluminium removal rates agreed with those reported for compost wetlands in the U.S.A. (0.06e0.8 g/m2/d), which received similar loads, albeit with lower concentrations at higher flow rates (Faulkner and Skousen, 1994). In contrast, iron removal rates were comparable only during the first and for most of the third year of operation when the performance was not impacted by a salt-clay surface layer. The MWO substrate released zinc, copper, lead, nickel and to a lesser degree chromium (Fig. 7), even though outflow concentrations for lead and copper were mostly below the limit of reporting. Zinc and nickel release occurred after the substrate was disturbed at start-up and during maintenance (Fig. 8). There may also have been some remobilisation of previously retained metals when flow

resumed after an extended drying period (e.g. cobalt and nickel at the start of the second year). However, this was minor compared to metal release from the MWO substrate. Therefore, MWO cannot be recommended as substrate for compost bioreactors unless the risk of metal release is addressed and minimised. The identification of minerals formed in the bioreactor and their stability was beyond the scope of this study. This is a limitation of the study and further work is warranted. Degens (2012) found that high evaporation rates in semiarid climates hinder bioreactor function. But our results show that compost bioreactors with a substrate mix of organic matter and limestone (10 wt%) are suitable to remove acidity, aluminium and iron from extremely saline acidic drainage in semi-arid landscapes despite being subjected to extended drying periods over summer, providing there is sufficient salinity to form a salt-clay surface layer and prevent the substrate from drying out. The need to break up the surface layer annually to improve performance may increase the risk of acidity and metal remobilisation from acidic aluminium and iron salts. This also warrants further investigation. The design of a full-scale treatment system would have to be modified to facilitate easy access for annual maintenance and to ensure the exclusion of surface water to prevent storm

Fig. 7 e Minor ion concentrations of substrate samples taken from the pilot-scale compost bioreactor 20 m, 40 m, 60 m and 80 m downstream of the inlet in December 2010 after 17 months of operation compared to the original municipal waste organics (MWO) substrate.

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

93

Fig. 8 e Inflow and outflow water quality of the pilot-scale compost bioreactor between July 2009 and October 2011: cobalt (A), copper (B), nickel (C), total chromium (D) and zinc (E). Open symbols denote no flow. Crosses denote values below the detection limit plotted at the detection limit. Vertical dashed lines indicate substrate sampling on 08 December 2010 (e   e) and maintenance works on 25 July 2011 (e e e).

water damage. This could be achieved by making the compost bioreactor longer and narrower using the same construction methods used for double-bunded deep drains. This would also maximise residence time and minimise preferential flow paths. In addition, surface wateresubstrate interaction in surface flow systems could be improved by (1) placing the substrate in a series of discrete sections separated by open water to break up the flow, (2) by mixing the substrate with a highly porous spacer material to improve permeability and (3) by placing shallow bunds across the substrate surface perpendicular to the flow path to slow down surface flow. Indrain treatment systems should only be considered where surface water is reliably excluded from the drain and end-ofdrain treatment systems must be protected from storm water by sufficiently high bunds.

5.

Conclusions

 Compost bioreactors using a substrate mix of organic matter and limestone are suitable to remove acidity, aluminium and iron from acidic saline drainage in semiarid landscapes.  A pilot-scale compost bioreactor successfully removed up to 28 g/m2/d CaCO3 equivalent of acidity and produced net alkaline effluent with pH  7 for 126 days after a startup period of 2 months. For sufficiently long residence times acidity removal was found to be load-limited.  Both limestone dissolution and bacterial sulfate reduction contribute to treating acidic drainage. Limestone dissolution produced up to 78e91% of the generated alkalinity while bacterial sulfate reduction contributed at least

9e22%. The biogeochemical processes in the bioreactor warrant further research.  High salinity caused by evaporative losses induced the formation of a salt-clay surface layer on top of the substrate. This surface layer prevented the dehydration of the substrate and ensured it remained waterlogged. When flow resumed the salt-clay surface layer acted as a barrier between the water and substrate decreasing performance efficiency. The surface layer needs to be broken up each year to optimise acidity removal.  To maintain high acidity removal rates, the substrate will have to be replenished when the limestone is exhausted, most likely after about five years.  Municipal waste organics in conjunction with recycled limestone, though successful in removing acidity from acidic saline drainage, was found to release metals and cannot be recommended for use in passive treatment systems unless the risk of metal release is addressed.

Acknowledgments This project was funded by the Australian Government through the Natural Heritage Trust (Regional Competitive Component project 53454). We would like to thank C-Wise, who kindly donated compost substrates for the laboratory tests. The authors thank Todd Gray for providing access to his land to build the pilot-scale compost bioreactor and the staff at the Centre of Excellence for Ecohydrology and the Department of Agriculture and Food WA for helpful discussions and comments.

94

w a t e r r e s e a r c h 5 5 ( 2 0 1 4 ) 8 3 e9 4

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.02.019

references

Beard, J.S., 1999. Evolution of the river systems of the south-west drainage division, Western Australia. J. Royal Soc. West. Aust. 82 (4), 147e164. Bird, M.I., Andrew, A.S., Chivas, A.R., Lock, D.E., 1989. An isotopic study of surficial alunite in Australia: 1. Hydrogen and sulphur isotopes. Geochim. Cosmochim. Acta 53 (12), 3223e3237. Caraballo, M.A., Macı´as, F., Nieto, J.M., Castillo, J., Quisepe, D., Ayora, C., 2011. Hydrochemical performance and mineralogical evolution of a dispersed alkaline substrate (DAS) remediating the highly polluted acid mine drainage in the full-scale passive treatment of Mina Esperanza (SW Spain). Am. Mineral. 96 (6), 1270e1277. Degens, B., 2012. Performance of pilot-scale sulfate-reducing bioreactors treating acidic Saline water under semi-arid conditions. Water Air Soil. Pollut. 223 (2), 801e818. Degens, B., Shand, P. (Eds.), 2010. Assessment of Acidic Saline Groundwater Hazard in the Western Australian Wheatbelt. CSIRO: Water for a Healthy Country National Research Flagship, Yarra Yarra, Blackwood and South Coast. Degens, B.P., 2009. Proposed Guidelines for Treating Acidic Drain Water in the Avon Catchment, Western Australia: Adapting Acid Mine Drainage Treatment Systems for Saline Acidic Drains. In: Salinity and Landuse Impact Series SLUI, vol. 54. Department of Water Western Australia, Perth. DoW, 2010. Hydstra Database e Time Series Data. provided 26 Feb 2010. Department of Water Western Australia, Water Information Provision Section, Perth. Druschel, G.K., Baker, B.J., Gihring, T.M., Banfield, J.F., 2004. Acid mine drainage biogeochemistry at Iron Mountain, California. Geochem. Trans. 5 (2), 13. Faulkner, B.B., Skousen, J.G., 1994. Treatment of acid mine drainage by passive treatment systems. In: Proceedings of the International Land Reclamation and Mine Drainage Conference and 3rd International Conference on the Abatement of Acidic Drainage, Pittsburgh, Pennsylvania, 24e29 April 1994, vol. 2. U.S. Department of the Interior, Bureau of Mines, pp. 250e257 (SP 06B-94). Hatton, T.J., Ruprecht, J., George, R.J., 2003. Preclearing hydrology of the Western Australian wheatbelt: target for the future? Plant & Soil. 257 (2), 341e356. Hedin, R.S., Nairn, R.W., Kleinmann, R.L.P., 1994. Passive Treatment of Coal Mine Drainage. Bureau of Mines Information Circular 9389. U.S. Department of the Interior. Holmes, K.W., Lillicrap, A., 2011. Mapping Acid Groundwater in Western Australia’s Wheatbelt. Resource Management Technical Report 373. Department of Agriculture and Food Western Australia, South Perth. Jarvis, A.P., Younger, P.L., 1999. Design, construction and performance of a full-scale compost wetland for mine-spoil drainage treatment at Quaking Houses. J. Chart. Instit. Water Environ. Manag. 13 (5), 313e318. Jones, S.M., Francis, C.J., Halliday, D.L., Leung, A.E., 2009. The Potential Effects of Groundwater Disposal on the Biota of Wetlands in the Wheatbelt, Western Australia. Prepared for the Avon Catchment Council. Department of Environment and Conservation Western Australia, Perth. Lillicrap, A., George, R., 2010. The Distribution and Origins of Acid Groundwaters in the South West Agricultural Area. Resource

Management Technical Report 362. Department of Agriculture and Food Western Australia, South Perth. Mann, A.W., 1983. Hydrogeochemistry and weathering on the Yilgarn Block, Western Australia e ferrolysis and heavy metals in continental brines. Geochim. Cosmochim. Acta 47 (2), 181e190. Matthies, R., Aplin, A.C., Jarvis, A.P., 2010. Performance of a passive treatment system for net-acidic coal mine drainage over five years of operation. Sci. Total Environ. 408 (20), 4877e4885. Parkhurst, D.L., Appelo, C.A.J., 1999. User Guide to Phreeqc (Version 2), a Computer Program for Speciation, Batchreaction, One-dimensional Transport and Inverse Geochemical Calculations. Water-Resources Investigations Report. U.S. Geological Survey, Denver. PIRAMID Consortium, 2003. Engineering Guidelines for the Passive Remediation of Acidic And/or Metalliferous Mine Drainage and Similar Waste Waters. European Commission 5th Framework RTD Project no. EVK1-CT-1999e000021 “Passive in situ remediation of acidic mine/industrial drainage”. (PIRAMID). University of Newcastle upon Tyne, Newcastle upon Tyne, UK. Prasad, D., Wai, M., Be´rube´, P., Henry, J.G., 1999. Evaluating substrates in the biological treatment of acid mine drainage. Environ. Technol. 20 (5), 449e458. Rose, A.W., 2007. Modeling the controls on acidity removal in vertical flow ponds. In: Barnhisel, R.I. (Ed.), Proceedings of the National Meeting of the American Society of Mining and Reclamation, Gillette, WY, 2e7 June 2007. ASMR, Lexington, KY, pp. 692e704. Sa´nchez Espan˜a, J., Lo´pez Pamo, E., Santofimia, E., Aduvire, O., Reyes, J., Barettino, D., 2005. Acid mine drainage in the Iberian Pyrite Belt (Odiel river watershed, Huelva, SW Spain): geochemistry, mineralogy and environmental implications. Appl. Geochem. 20 (7), 1320e1356. Santini, T., Degens, B., Rate, A., 2010. Organic substrates in bioremediation of acidic saline drainage waters by sulfatereducing bacteria. Water Air Soil. Pollut. 209 (1e4), 251e268. Schmidtova, J., Baldwin, S.A., 2011. Correlation of bacterial communities supported by different organic materials with sulfate reduction in metal-rich landfill leachate. Water Res. 45 (3), 1115e1128. Seewraj, K., 2010. Fence Road Drainage System e Interim Assessment of Monitoring Data. Report. Department of Water WA, Bunbury, Western Australia. Skousen, J., Ziemkiewicz, P., 2005. Performance of 116 passive treatment systems for acid mine drainage. In: Proceedings of the National Meeting of the American Society of Mining and Reclamation, Breckenridge, Colorado. American Society of Mining and Reclamation, pp. 1100e1133. Stewart, B., Strehlow, K., Davis, J., 2009. Impacts of deep open drains on water quality and biodiversity of receiving waterways in the Wheatbelt of Western Australia. Hydrobiologia 619 (1), 103e118. Tille, P.J., Mathwin, T.W., George, R.J., 2001. The South-west Hydrological Information Package, Understanding and Managing Hydrological Issues on Agricultural Land in the South-west of Western Australia. In: Bulletin, vol. 4488. Department of Agriculture Western Australia. Tyrrell, W.R., Mulligan, D.R., Sly, L.I., Bell, L.C., 1997. Trialing wetlands to treat coal mining wastewaters in a low rainfall, high evaporation environment. Water Sci. Technol. 35 (5), 293e299. Watzlaf, G.R., Schroeder, K.T., Kleinmann, R.L.P., Kairies, C.L., Nairn, R.W., 2004. The Passive Treatment of Coal Mine Drainage. Report DOE/NETL-2004/1201. U.S. Department of Energy, National Energy Technology Laboratory, University of Oklahoma, Pittsburgh, Pennsylvania. Younger, P.L., Banwart, S.A., Hedin, R.S., 2002. Mine Water: Hydrology, Pollution, Remediation. Kluwer Academic Publishers, Dordrecht, Netherlands.