SULFATE REMOVAL from COAL MINE WATER in WESTERN PENNSYLVANIA: REGULATORY REQUIREMENTS, DESIGN, and PERFORMANCE1 William J Walker,2 Jorge Montoy, and Tyler Chatriand

JASMR, 2015 Volume 4, Issue 1

SULFATE REMOVAL FROM COAL MINE WATER IN WESTERN PENNSYLVANIA: REGULATORY REQUIREMENTS, DESIGN, AND PERFORMANCE1 William J Walker,2 Jorge Montoy, and Tyler Chatriand

Abstract: The listing of the Monongahela River as an impaired waterway prompted the Pennsylvania Department of Environmental Protection (PADEP) to adjust aqueous discharge limits to the river to no more than 250 mg/L of sulfate. In response to this, an analysis of water treatment options for a coal mining company was conducted at several non-operating mines in western Pennsylvania that discharge directly or indirectly to the Monongahela River. Given the extremely high capital and operations costs for typical sulfate reduction methods such as reverse osmosis and ion exchange, novel passive and semi-passive treatment options were explored. An ethanol-fed bioreactor system was selected, designed, and constructed in 2014 to test whether sulfate reducing bacteria could be utilized to remove sulfate in alkaline mine water to meet discharge limits. The unique design elements consist of metals removal circuit, ethanol feed circuit, and twin bioreactors bedded with large cobbles and seeded with sulfate reducing bacteria, but containing no additional carbon source. Biochemical performance has shown -3/ that sulfate reduction approaches 1500 mmol SO4 m day during warmer weather, one of the highest rates recorded in the literature. Effluent sulfate ranged from 58 to 400 mg/L at 16ºC and about 1400 mg/L at 2ºC compared to influent sulfate concentrations that averaged 2800 mg/L. In addition, the bioreactor produced 500- 1500 mg/L of total alkalinity due to microbial metabolism supported by the ethanol, typically corresponding with sulfate decreases. Effluent metal concentrations were decreased to 1 mg/L Fe and 0.2 mg/L Mn. The recirculation loop was found to remove 90% of iron in the original settling pond prior to entering the reactors to minimize sludge accumulation. Additional Keywords: bioreactors, AMD, sulfate reduction, sulfate reducing bacteria, mine water ______1 Paper to be presented at the 2015 National Meeting of the American Society of Mining and Reclamation, Lexington, KY Reclamation Opportunities for a Sustainable Future June 7–11, 2015. R.I. Barnhisel (Ed.) Published by ASMR, 3134 Montavesta Rd., Lexington, KY 40502. 2 Dr. William J Walker, Senior Geochemist, Sovereign Consulting, Seattle, WA 98121, Jorge Montoy, Project Engineer, Sovereign Consulting, San Francisco, CA 94103, and Tyler Chatriand, PE with Engineer, Sovereign Consulting, Seattle, WA 98121 Journal American Society of Mining and Reclamation, 2015 Volume 4, Issue 1 pp 73-93 DOI: http://doi.org/10.21000/JASMR15010057 For some unknown reason, the DOI doesn’t always work so if this is the case use in place of the DOI http://www.asmr.us/Portals/0/Documents/Journal/Volume-4-Issue-1/Walker-WA.pdf

73 JASMR, 2015 Volume 4, Issue 1

Introduction The listing of the Monongahela River as an impaired waterway prompted the Pennsylvania Department of Environmental Protection (PADEP) to adjust effluent discharge limits to the river to contain no more than 250 mg/L of sulfate in accordance with USEPA secondary maximum contaminant levels. In response to this regulatory driver, an analysis of water treatment options was conducted for a coal mining company at several non-operating mines in western Pennsylvania that discharge directly or indirectly to the Monongahela River. The analysis reviewed cost and performance data for active and passive systems deemed capable of treating sulfate to 250 mg/L in mine water containing 3000 mg/L at an average flow rate of 1000 gpm.

A common method for removing high concentrations of sulfate from water is through addition of hydrated lime (Ca(OH)2), which precipitates CaSO4 as shown in reaction (1):

Na2SO4 + Ca(OH)2 => CaSO4 + 2 NaOH (1)

Calcium sulfate, which hydrates to become the common mineral gypsum, has a solubility of approximately 2000 mg/L as SO4. Sulfate reduction below 2000 mg/L has previously been possible only through expensive and resource-intensive technologies such as reverse osmosis (RO) or ion exchange (IX). Large volumes of liquid waste are generated with RO and IX, which typically create additional treatment and disposal costs. Overall the capital and especially operation and maintenance (O and M) costs for RO and IX systems are untenable for inactive mine sites that will require treatment in perpetuity.

As a lower-cost semi-passive alternative to RO and IX for treatment of sulfate, bioreactors were explored. Bioreactors have been used in recent years to treat AMD (McCauley et al., 2009; USEPA, 2006). These anaerobic reactors utilize sulfate-reducing bacteria (SRB) to (1) reduce sulfate to sulfide, (2) precipitate divalent heavy metals to sulfide solids, and (3) produce alkalinity in proportion to carbon utilization by the microbial population. As such, a simple bioreactor can, at least theoretically, treat AMD to water standards normally achieved by lime plants. When high metal concentrations are removed as sulfides, carbonates or oxides, the acidity decreases, and pH increases. Many of the bioreactors are constructed using a flow-through design wherein mine water flows through the treatment media, which also serves as the carbon source. Carbon media typically utilized includes mushroom compost, manure, and wood chips (Choudhary et al., 2012), though other efforts have explored use of liquid carbon sources (Zamzow et al., 2007). Reactors

74 JASMR, 2015 Volume 4, Issue 1 with a solid carbon media source are usually very effective initially but can lose effectiveness over time due to consumption of easily utilized carbon, significant changes in hydraulic conductivity, and sludge accumulation in the reactors. The primary purpose of these reactors has typically been acidity and metals removal; sulfate reduction is used as a mechanism to create enough sulfide to treat metals and generate alkalinity.

In this work, sulfate reduction is the main concern, and design was tailored to optimize sulfate reduction rates while minimizing capital and O and M costs. Instead of flow through media, the carbon source, ethanol, is fed continuously into the reactor as a source of chemical oxygen demand (COD) to achieve reduction. Ethanol theoretically produces 2.09 mg COD/mg ethanol (Haandel et al., 2012). The reactors are only filled with large, unreactive cobbles. A recirculation loop and settling pond serve to remove a large percentage of the solids generated. In this way, the carbon source and hydraulic properties of the reactors can be held constant eliminating much of the variability found in other systems. In addition, the reactors can accommodate much higher flow rates and achieve higher sulfate reduction rates.

The pilot system was designed to begin the process of optimizing the microbial transformation of sulfate to reduced sulfide in order to reach a proposed regulatory goal of 250 mg/L. The microbial process also is expected to remove metals via metal sulfide formation and precipitation, create alkalinity from carbon metabolism, and produce elemental sulfur from the oxidation of excess sulfide. Early experimental results allowed the mining company to enter into a consent order with PADEP to build and operate a full-scale system by December 2016. In the broader picture, the pilot study was seen as an opportunity to assist in bringing these types of treatment processes into more common use.

Methods and Materials

The pilot test was designed to treat a 10 gpm source of mine water containing approximately 2800 mg/L of sulfate, 120 mg/L Fe, and 1.8 mg/L Mn. Typical characteristics of the site’s mine water intended for treatment are shown in Table 1.

75 JASMR, 2015 Volume 4, Issue 1

Table 1. SRB Pilot Plant Influent Average Concentrations pH Sulfate Iron Manganese Aluminum TDS Alkalinity (SU) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) (mg/L) 6.8 2,800 120 1.8 0.06 5,000 600

The design objectives, process flow, and design details for each component of the system are described below. The layout of the system is provided as Fig. 1. A Process Flow Diagram is provided as Fig. 2.

Figure 1. SRB System Layout.

Objectives Overall, the pilot was designed to address the following items: • Test the viability of using SRB to reduce sulfate concentrations in mine water. • Determine sulfate removal rates within the volume of the reactors (mmol sulfate/m3 reactor-day). • Test effects on other parameters (pH, metals, organic carbon). • Determine sludge handling requirements. • Minimize power requirements by utilizing gravity where possible. • Allow for design of full-scale sulfate reduction system to meet compliance period.

76 JASMR, 2015 Volume 4, Issue 1

Figure 2. SRB System Process Flow Diagram.

Design and Construction The design considerations for the system were derived primarily from literature sources and private communications with several other researchers in the area. These are summarized below. • The pilot test relies on SRB for sulfate reduction. • The carbon source for the SRB is a liquid alcohol (ethanol) source fed at the influent to the reactors. • Water is pulled from the mine pool to a settling pond and circulated through a settling pond and the reactors to aid in metal sulfide precipitation. • Water will flow from the second reactor to a polishing pond before being discharged to the existing impoundment.

• Each reactor contains 5 multi-depth (10 total) sampling points consisting of slotted PVC piping bedded within reactor cobbles to allow for monitoring of conditions within the reactors.

77 JASMR, 2015 Volume 4, Issue 1

Overall, the process flow (Fig. 2) consists of: • The pilot test flow rates range from 10-20 gpm. • Raw water is pumped from the mine shaft into a 10,000 gallon holding tank. From the holding tank it is pumped directly to the Settling Pond. • Water is gravity fed from the Settling Pond into a 10,000 gallon holding tank before being pumped up through Reactor 1 and being gravity fed back to the Settling Pond. Water also pumps from this holding tank into Reactor 2. • Outfall from Reactor 2 gravity feeds to the Polishing Pond before discharge to the existing impoundment. Reactor 2 also has the ability to recirculate through the Settling Pond, and typically about 1/3 of this water is recirculated to aid in mixing and metals precipitation. • Overall hydraulic retention time in this system is approximately 3 days. • Tanks holding the ethanol feed into the influent of each reactor using metering pumps. These metering pumps are also available to feed nutrients to the reactors as needed.

Pilot Startup and Sampling/Analysis Construction of the pilot system took place from June 30 to August 15, 2014. Photos showing system construction are displayed in Fig. 3 and 4. The system was installed in accordance with the design information provided above. Additionally, fencing and signage was placed around all system equipment and ponds/reactors.

Following construction, a startup period followed to test system controls and to establish a viable SRB population in the reactors. The reactors were filled with mine water, fed ethanol, and inoculated with bacteria. Inoculation was accomplished by mixing about 10 kg of fresh manure with mine water in a 55 gallon drum. The mixture was allowed to sit for 10 days and then the effluent fed into the mixing pond for distribution through the reactors. No special bacterial inoculant was used. Nitrogen and phosphorus were both added in amounts recommended for good carbon utilization, or 3g/L of H2KPO4, 1 g/L of NH4Cl, and 0.1 g/L of MgCl2 after Isa et al. (1986). Geochemical parameters, primarily dissolved oxygen (DO) and oxidation-reduction potential (ORP), were monitored to determine when anaerobic conditions reached levels conducive for sulfate reduction and SRB growth. Additionally, field test kits were used to determine an estimate of the SRB present in each reactor cell (Fig. 5). The COD concentration

78 JASMR, 2015 Volume 4, Issue 1

Figure 3. Photo of constructed and filled 120’x30’x6’ reactor.

Figure 4. Photo of Pump enclosure and tanks near end of construction. was also monitored to ensure concentrations of ethanol in the system were sufficient for sulfate reduction. Initially the COD to sulfate ratio (COD mg/L/sulfate mg/L) was kept near 1, a ratio that

79 JASMR, 2015 Volume 4, Issue 1 should allow complete reduction of sulfate (Bekmezci et al., 2011; Rodriguez et al., 2012). Ethanol theoretically produces 2.09 mg COD/mg ethanol, but COD was confirmed with laboratory analytical samples. The COD per unit mass varies widely between carbon sources and must be considered when determine carbon source dosing rates (Haandel et al., 2012).

Figure 5. The SRB Test System (Sani Check) for counting sulfate-reducing bacteria. Figure note: The vial in the far left is the influent/mixing pond water. The next four vials are samples from the two reactors and the last two are from the polishing pond and effluent pipe. The dark color (iron sulfide) indicates proliferation of SRB.

By early September 2014 optimal anaerobic conditions were present throughout the reactors with a DO <0.1 mg/L and an ORP < -300 mV. Samples were collected in the multi-depth sample points present throughout the reactors. As ethanol degradation proceeded the water in the reactors rose to pH 7.7 compared to the raw mine water pH of 6.8.

The pilot system was sampled every few days initially and then approximately weekly from September through current operation. Weekly sampling and analysis included field parameters pH, DO, ORP, T (0C). Sulfate (method EPA 300.0), sulfide (method SM 4500), alkalinity (method SM20-2320B), COD (method EPA 410.4), Fe, Mn, and Al (method EPA 200.7) were determined by an off-site analytical laboratory (Fairway Laboratories in Altoona, PA). The sampling locations typically included the raw mine water influent, the mixing pond, the reactor outlets, the polishing

80 JASMR, 2015 Volume 4, Issue 1 pond, and the effluent. All analyses were performed using USEPA/ASTM standard methods in state and federal certified laboratories. The QA/QC reports were scrutinized with each submittal to determine if analyte recovery and dilution were accurate.

Full-time system operations began on September 11, 2014 with an average influent/effluent flow rate of 10 gpm. During system operation, pump flow rates and pressures were monitored to ensure the desired flow rate was maintained, in addition to the recirculation rates and metering of ethanol. Ethanol was fed to the system at a target flow rate of 0.016 gpm (1.0 gph).

Results and Discussion System Chemistry The pilot system was designed to begin the process of optimizing the microbial transformation of sulfate to reduced sulfide. Based on observed sulfate reduction rates in the literature, a sulfate reduction rate of 500-1000 mmol/m3-day was targeted (SME, 2009; Tsukamoto et al., 2014). The microbial process also is expected to remove metals via metal sulfide formation and precipitation, create alkalinity from carbon metabolism and produce elemental sulfur from the oxidation of excess sulfide.

The basic microbial processes are well known and they include the following generalized reactions.

I. For sulfate and carbon, utilization by the sulfate reducing bacteria produces sulfide (dissolved and as a gas) and alkalinity as shown in reaction (2). The reaction shown is an overall representation of several microbial processes and chemical pathways. The sulfide produced will be partially deprotonated and solubilized based on ambient pH.

2- - SO4 + 2CH2O = H2S + 2HCO3 (2)

II. The divalent metals in the mine water (mainly Fe and Zn in this mine water) will react with sulfide and precipitate as sulfide solids (3).

2+ + H2S + M = MS(s) + 2H (3) III. Excess sulfide will off-gas or form either elemental sulfur (4).

- o - 2HS + O2 = 2S + 2OH (4)

IV. and/or convert back to sulfate in high O2 waters (5).

- 2- + 2HS + 4O2 = 2SO4 +2H (5)

81 JASMR, 2015 Volume 4, Issue 1

Performance Results: Water Quality, Metals Removal and Sulfur Transformations

The physiochemical data used in the performance analysis is presented in Table 2. All of the data collected as part of this study will be made available through the corresponding authors. The data tables will be updated periodically during the expected performance period of two years through September 2016.

Table 2. SRB Pilot Plant Physiochemical Performance Data – System Effluent Data Sampling Sulfate Temp ORP Sulfide Iron Mn Alkalinity Date (mg/L) (oC) (mV) (mg/L) (mg/L) (mg/L (mg/L) 9/25/2014 2900 16.1 -116 0 103 1.45 620 10/1/2014 1950 14.4 -299 12 0.49 0.139 na 10/9/2014 1600 13.9 -393 85.6 2.75 0.08 1610 10/16/2014 700 12.6 -379 82 0.3 0.03 na 10/23/2014 58 10.5 -366 61.2 0.3 0.01 1910 10/30/2014 101 8.9 -391 76.8 0.78 0.027 1980 11/6/2014 493 2.8 -390 107.4 2.7 0.25 1720 11/13/2014 808 1.8 -401 94 1.7 0.14 1670 11/20/2014 997 3 -374 69.2 1.5 0.11 na 12/11/2014 1488 2.5 -389 na na na na 12/19/2014 1450 2.6 -381 na na 0.05 na 1/29/2015 1510 2.1 -377 56 1.26 0.89 1200 2/5/2015 1500 2.5 -397 66 1.27 0.94 1010 3/12/2015 1870 4.1 -399 2.8 28.1 1.04 642 4/20/2015 976 5.5 -393 100 26.6 0.59 1490 The tabulated data show several important trends for evaluating performance. For example, sulfate concentrations in the effluent began to decrease after 2 weeks into the pilot study from 2900 mg/L to 1600 mg/L. After 6 weeks sulfate in the pilot effluent reached a minimum concentration of 58 mg/L or a 98% reduction (Fig. 6) corresponding to a sulfate reduction rate of approximately 1500 mmol SO4/ m-3/day. This reduction rate was maintained for 2 weeks. However, as water temperature began to decrease in early November, sulfate concentrations in the effluent began to increase and level out at about 1450 mg/L or only a 50% reduction. Water temperature in the reactors was 16º C at the beginning of the study, decreased to 9º C in November, and to 2º C or less in mid-December through mid- February. The dependence of the reduction process on water temperature is shown in Fig. 7 and is similar to other studies where large temperature swings were noted (Praharaj and Fortin, 2008). Warming temperatures in April 2015 resulted in increased reduction rates following the colder winter period (March 2015 reduction rates were depressed due

82 JASMR, 2015 Volume 4, Issue 1 to inability to obtain a delivery of a carbon source and decreased COD). Figure 8 shows a photo of an operating reactor in icy conditions in January 2015.

Figure 6. Sulfate Influent and Effluent of SRB Pilot Plant

Figure 7. Sulfate in Effluent (mg/L) vs Temperature (0C)

83 JASMR, 2015 Volume 4, Issue 1

Figure 8. Photo of reactor in icy conditions, January 2015.

The data also demonstrates the expected decrease in Fe concentrations due to FeS precipitation, a common observation in anaerobic bioreactors (Bekmezci et al., 2011; McCauley et al., 2009). Total iron concentrations decreased from about 120 mg/L to < 2 mg/L once sulfide production occurred. These concentrations have remained stable throughout the study period. Manganese concentrations in the effluent decreased from about 2 mg/L to about 0.2 mg/L in the pilot plant effluent. Sampling at the inlet to the reactors shows that 90% of metals removal occurs in the Settling Pond due to the recirculation loop and prevents sludge accumulation in the reactors. Calculation of metals (primarily Fe) deposition in the bioreactors given these loading rates indicates that porosity would be decreased by approximately 1% annually within the bioreactors; actual rates would need to be confirmed over time. Removal of Mn is not due to sulfide formation but more likely due to carbonate formation in the reactors where alkalinity is produced from ethanol metabolism (Edwards 2008; Paulina et al., 2011). Some Mn adsorption to FeS is also possible.

Operation of the reactors has shown formation of alkalinity corresponding to the amount of sulfate reduction. The alkalinity produced has ranged from 500-1500 mg/L (as CaCO3) through

84 JASMR, 2015 Volume 4, Issue 1 operation of the system. The alkalinity is highest in the system effluent and this may present opportunities for long-term source treatment by recirculation back into the mine pool.

The sulfide (mainly HS- and minor amounts of S=) concentrations observed in the reactor effluents are shown below in Fig. 9.

Figure 9. Soluble Sulfide (mg/L) in System Effluent from start up through April 2015.

Since the pH in the effluent was typically about 8.5, most of the soluble sulfide would exist as the partially protonated HS- species.

A sulfur mass balance and species distribution for treated sulfur was attempted from the available chemical data, although the difficulty associated with sampling the reactors and obtaining direct species information precludes a quantitative result. The estimates, however, are instructive especially with regards to the eventual handling of sulfur based residuals in the effluent. The sulfur mass balance estimates and species distribution are presented in Fig. 10. Speciation calculations were determined from the basic mass balance equation (6):

= = 0 Stotal = SO4 + MS + S + (S + H2S) (6) Where

Stotal = total S = influent sulfate in mixing pond feed to reactors (direct measurement).

85 JASMR, 2015 Volume 4, Issue 1

SO4 = sulfate in effluent (direct measurement). = MS = Metal Sulfides = FeS (mmoles of Fe equal mmoles of S ) FeS2 not considered. Calculated from measured Fe in influent pond and assumes all Fe precipitated was in the form of FeS.

S= = (S= + HS-) = Soluble sulfide = sulfide directly measured in effluent (most S= at pH 8 is HS-). Much of this “soluble” sulfide would eventually outgas in the receiving pond. Removal of this component will occur at end-of-pipe where the measurements are taken.

The estimate of elemental sulfur and sulfide gas is represented by equation (7):

0 (S + H2S) = elemental sulfur plus sulfide gas = [Stotal -SO4 –MS] and are computed by difference as these two species are not yet amenable to direct determination. (7)

Figure 10. Sulfur Speciation in Pilot Plant Reactors at Optimal Sulfate Reduction Rate

The sulfur distribution plot demonstrates that under the higher temperature (highest reduction rate) conditions observed in the late summer/fall, sulfate reduction yields in decreasing amounts elemental sulfur and gas > metal sulfides > non-reduced sulfate > soluble sulfide. Elemental sulfur has not been quantified but is easily recognized in the reactors where it occurs as a thin film of milky white particles at the reactor surface where it slowly settles on the large cobbles below (see Figs. 11 and 12). Treatment options for the elemental sulfur include accumulation in the reactors or harvesting using biofilms and/or booms to collect the floating portion. Sulfide gas has not yet

86 JASMR, 2015 Volume 4, Issue 1 been quantitated either. However, in-field personal air monitors have yet to detect sulfide gas above the current American Conference of Governmental Industrial Hygienists (ACGIH) 15- minute STEL threshold value of 5.0 ppm(v) in the breathing zone.

Figure 11. Reactor displaying formation of elemental sulfur on surface of reactor, April 2015.

From a performance perspective this distribution provides some advantages. For example the more elemental sulfur the system can produce will result in a decrease in the gaseous forms of sulfide which could result in noxious air conditions. The sulfate reduction system can be limited in effectiveness if the reduction product, soluble sulfide, is too high in the effluent since sulfide can re-oxidize to sulfate in well oxidized receiving waters (Glombitza 2001; Bekmezci et al., 2011). In the present case, the small amount of soluble sulfide in the effluent keeps effluent sulfur at an acceptable level even if the remaining soluble sulfide re-oxidizes to sulfate. Taken as a whole, the distribution pattern allows for an identification of probable residuals management as the project progresses.

One of the primary goals going forward will be to keep physiochemical conditions optimal for the formation of elemental sulfur compared to soluble sulfide. If soluble sulfide requires removal, a separate polishing pond could be required. For a larger scale plant, formation and buildup of

87 JASMR, 2015 Volume 4, Issue 1

H2S gas must also be addressed; under the current system the system creates nuisance odors but are below all worker exposure limits in the breathing zone.

Figure 12. Reactor cobbles showing settling of elemental sulfur, October 2014.

Reactor Solids Sampling and Analysis During the January 20, 2015 sampling event, solids coating the surface of the rocks in the reactors were sampled. While the system design removes solids mainly in the settling pond upstream from the reactors, some solids are not settling quickly and precipitate out in the reactors near the influent point. The settling pond will be sampled during the investigation but was not possible at this time, so these solids serve as a surrogate for the solids in the settling pond. Sampling was accomplished by simply removing the upper layer of rocks in the reactor to about 1 ft bgs and then scraping samples of the coating off of individual cobbles into pre-cleaned, wide mouth 100 ml glass bottles. The samples were sent to the RJLee microscopy lab for optical

88 JASMR, 2015 Volume 4, Issue 1 microscopy examination and XRD inspection. The initial analysis indicated that the main solids formed were amorphous iron sulfides likely containing both FeS and FeS2 (Paulina et al., 2011; Sheoran, 2010). Because iron is the only metal in significant quantities in the mine water (the raw mine water contains about 120 mg/L of iron and only trace amounts of other metals) it was expected that the solids would consist primarily of iron sulfides and oxyhydroxides. The sulfide solids form from the mixing of the Fe(II) in the mine effluent with the reduced reactor water rich in soluble sulfide. Some iron oxyhydroxides would be expected as well, as some of the mine water is oxidizing rapidly before it reaches the mixing pond allowing for some hydroxide solids. An example of the appearance and morphology of the samples is shown below in Fig. 13.

Figure 13. Amorphous iron sulfide formation seen under microscope from precipitates scraped off of reactor rocks. Figure note: The dark blackish material is the amorphous iron sulfide, the lighter orange material the iron oxyhydroxide particles, and the very light material scrapings from the rock support.

89 JASMR, 2015 Volume 4, Issue 1

Manganese in the mine water averages about 1.8 mg/L. The reactors typically decreased Mn concentration to about 0.5 mg/L. Since Mn sulfides don’t readily occur, the removal mechanisms are likely Mn oxidation and formation of MnO2 or formation of Mn carbonates without Mn oxidation. A comparison of the effluent Mn and alkalinity concentrations with known Mn solid phases suggest that Mn carbonates are the most likely solid phase forming in the reactors which is also consistent with stability diagrams for this pH and Eh profile. More work on the reactor residues is planned for the duration of the pilot study in order to more definitively identify the solids forming and to understand the long term stability of the solids.

Summary and Conclusions

An ethanol fed bioreactor system was designed and constructed in 2014 to remove sulfate in near-neutral mine water with the goal of maximizing sulfate reduction prior to discharge. The unique design elements consist of metals removal circuit, ethanol feed circuit, twin bioreactors bedded with large cobbles and seeded with sulfate reducing bacteria but containing no additional carbon source.

The pilot test was designed to treat a 10 gpm source of mine water containing approximately 2800 mg/L of sulfate, 120 mg/L Fe, and 1.8 mg/L Mn. After 4 weeks of operation, the bioreactors were able to achieve sulfate reduction rates not typically seen in the literature (1500 mmol/m3- day) and showed to be a viable alternative for meeting treatment goals. As water temperature decreased to 9º C and then 2º C, sulfate reduction decreased to 500 mmol/m3-day, an observation not uncommon in the literature. Sulfate reduction rates increased as expected with the warming temperatures in April 2015.

Effluent sulfate ranges from 58 to 400 mg/L at 16º C and about 1400 mg/L at 2º C. In addition, the bioreactor produced 500-1500 mg/L of total alkalinity due to ethanol metabolism, typically corresponding with sulfate decreases. Effluent metal concentrations were decreased to 1 mg/L Fe and 0.2 mg/L Mn. The recirculation loop was found to remove 90% of Fe in the settling pond prior to entering the reactors to minimize sludge accumulation.

O and M and operating costs for the pilot system were kept to a minimum and consisted primarily of basic pump cleaning, refilling of ethanol tanks, winterization of pump infrastructure, and adjustment of flow rates for optimal performance.

90 JASMR, 2015 Volume 4, Issue 1

Micro-mineralogical analysis of precipitates on reactor cobbles shows Fe occurring mainly as

FeS and FeS2 with some mixed sulfides and some iron oxyhydroxides. While precipitates in the settling pond were unable to be sampled directly, precipitates formed in the pond are presumably in the same form. Manganese solids are likely primarily carbonates and formed in the reactors where alkalinity is produced and not in the settling ponds. Material balance and species o = distribution for sulfur was found to be: S (32%) > Metal Sulfides (25 to 31%) > SO4 (25%) > HS- and S= (g) (11%).

While the system performance is promising, there is much work to accomplish. Several important system processes require closer inspection:

• Sulfate reduction rates over time and seasonal variations must be further studied. • The residual sulfur speciation may require control. Sulfide in the effluent may need to be removed either by adsorption to an inexpensive media or converted to elemental sulfur by

carefully controlling ORP and pO2. Elemental sulfur may require collection.

• The formation and accumulation of H2S gas in a large scale system must be explored to not create a hazardous environment for workers or the surrounding community. • The Settling Pond will eventually require dredging. Solids buildup in the Settling Pond will be assessed by emptying the pond after a year of operation or if accumulation becomes evident visually or in performance testing. The metal sulfides within the sludge will be kept in an anoxic environment probably by pumping back into the mine pool. Calculation of metals (primarily Fe) deposition in the bioreactors given calculated loading rates indicates that porosity would be decreased by approximately 1% annually; actual rates would need to be confirmed over time. Assessment of pump back water on the mine pool chemistry would also require investigation. • The carbon source will be inspected as well to determine the optimal dose for sulfate reduction. The carbon source cannot become cost prohibitive so other admixtures with ethanol or other liquid carbon sources could be tested.

Scheduled 2015 activities include monitoring performance with temperature, optimization of flow rate, ethanol feed and elemental sulfur production, and assessment of requirements for a full- scale design to treat all the water pumped from the mine pool using this technology.

91 JASMR, 2015 Volume 4, Issue 1

Acknowledgements

The authors wish to express thanks to Drs. G. Miller and T. Tsukamoto for their valuable discussions and to John McCollums and Brian Becker for their tireless efforts in the field.

Literature Cited

Bekmezci, O., D. Ucar, A. Kaksonen, and E. Sahinkaya. 2011. Sulfidogenic biotreatment of synthetic acid mine drainage and sulfide oxidation in anaerobic baffled reactor. Journal of Hazardous Materials, 189, 670-676. https://doi.org/10.1016/j.jhazmat.2011.01.087

Choudhary, R.P. and A. Sheoran. 2012. Performance of single substrate in sulphate reducing bioreactor for the treatment of acid mine drainage. Minerals Engineering 39; pg 29-35. http://dx.doi.org/10.1016/j.mineng.2012.07.005.

Edwards, J. 2008. Removal of Manganese from and alkaline mine drainage using a bioreactor with different carbon sources. University of Kentucky Master’s Thesis Paper 522.

Glombitza, F. 2001. Treatment of acid lignite mine flooding water by means of microbial sulfate reduction. Waste Management 21 (2001) 197±203.

Haandel, A. and J. Lubbe. 2012. Handbook for Biological Wastewater Treatment – Design of Activated Sludge Systems. pp 355 to 390

Hayes, W., J. Davis, V. Silva and A.P. Jarvis. 2011. Treatment of zinc rich acid mine water in low residence time bioreactors incorporating waste shells and methanol dosing. Journal of Hazardous Materials, 193, 279-287. http://dx.doi.org/10.1016/j.jhazmat.2011.07.073 PMid:21864976.

Isa, Z., S. Grusenmeyer, and W. Verstraete. 1986. Sulfate reduction relative to methane production in high rate anaerobic digestion: Technical aspects. Applied and Environmental Microbiology, 51(3) 572-579.

Liamleam, W. and A.P. Annachtre. 2007. Electron donors for biological sulfate reduction. Biotechnology Advances, 25: 452-463. http://dx.doi.org/10.1016/j.biotechadv.2007.05.002.

McCauley, C., A. O'Sullivan, M. Milke, P. Weber, and D.A. Trumm. 2009. Sulfate and metal removal in bioreactors treating acid mine drainage dominated with iron and aluminum. Water Research 43; 961-970. http://dx.doi.org/10.1016/j.watres.2008.11.029

92 JASMR, 2015 Volume 4, Issue 1

Paulina, K., E. Remoundaki, A. Hatzikioseyian, F. Battaglia-Brunet, C. Joulian, V. Kousteni, and M. Tsezos. 2011. Metal precipitation in an ethanol-fed, fixed-bed sulphate-reducing bioreactor. Journal of Hazardous Materials, 189, 677-684. http://dx.doi.org/10.1016/j.jhazmat.2011.01.083

Praharaj, T. and D. Fortin. 2008. Seasonal variations in microbial sulfate and iron reduction in alkaline Pb-Zn mine tailings (Ontario, Canada). Applied Geochemistry 23, 3728-3740. http://dx.doi.org/10.1016/j.apgeochem.2008.09.008.

Rodriguez, R.P., G.H.D Oliveira, I.M. Raimundi, and M. Zaiat. 2012. Assessment of an Upflow Anaerobic Sludge Blanket reactor for the removal of sulfate for acid mine water. International Biodeterioration and Biodegradation 74, 48-53. http://dx.doi.org/10.1016/j.ibiod.2012.07.012.

Sheoran, A., V. Sheoran, and R.P. Choudhray. 2010. Bioremediation of acid-rock drainage by sulfate reducing prokaryotes: A Review. Minerals Engineering 23, 1073-1100. http://dx.doi.org/10.1016/j.mineng.2010.07.001.

Society for Mining, Metallurgy, and Exploration Inc. (SME). 2009. Mitigation of Metal Mining Influenced Water. Volume 2. Edited by J. Gusek and A. Figueroa.

Tsukamoto, T. and G.C. Miller. 2014. Bioreactors for Treatment of Acid Mine Drainage at the Leviathan Mine, (Personal Communication with both authors.)

U.S. Environmental Protection Agency (USEPA). 2006. Bioremediation of Acid Mine Drainage Using Sulfate Reducing Bacteria. Office of Solid Waste and Emergency Response. Office of Superfund Remediation and Technology Innovation.

Zamzow, K., T. Tsukamoto and G.C. Miller. 2007. Biodiesel waste fluid as an inexpensive carbon source for bioreactors treating acid mine drainage (California-Nevada) IMWA Symposium 2007: Water in Mining Environments, R. Cidu and F. Frau (eds.), 27th - 31st May 2007, Cagliari, Italy.