Batch Operating Limestone Treatment Systems (BOLTS): Greater Efficiency and Cost Savings

Mine drainage (MD) from abandoned coal mines is a persistent source of water pollution worldwide. Coal MD is usually acidic and contains elevated concentrations of dissolved metals that if left untreated can have serious ecological consequences on receiving water bodies. An estimated 10,500–39,000 miles (17,000–63,000 km) of stream are impacted by MD in the Appalachia range (Appalachian Regional Commission 1969; Hansen et al. 2010; US Environmental Protection Agency 1995). Although substantial efforts have been made to remediate these sites—over $4 billion in grants have been funded to clean up abandoned mine sites since 1977—there are still 5,200 coal-related abandoned mine sites that have yet to be fully reclaimed, resulting in an estimated costs of $3 billion in health and safety problems and $2 billion in general welfare, environmental, and non–coal-related problems (Bureau of Land Management 2021).

Coal MD is often treated in limestone-based treatment systems that rely on natural biological, geochemical, and physical processes to improve water quality (Skousen et al. 2017). Common types of limestone-based systems are anoxic limestone drains, open limestone channels, reducing and alkalinity producing systems (also known as successive alkalinity producing systems), oxic limestone drains (OLDs), vertical flow ponds (VFPs; also known as vertical flow beds), and auto-flushing vertical flow ponds (AFVFPs) (Cravotta et al. 2004; Cravotta & Trahan 1999; Hedin et al. 2010; Kepler & McCleary 1994; Skousen et al. 2017). In OLDs and VFPs, MD discharges are typically collected and transferred to a rectangular bed that is open to the atmosphere and filled with limestone. Flow through the limestone bed can be horizontal or vertical depending on the pipe configuration. As the MD travels through the limestone bed, the acidity in the MD is neutralized by calcite (CaCO₃) and dolomite (CaMg(CO₃)₂) dissolution, thereby increasing the pH, alkalinity, and concentrations of calcium and magnesium in the solution. The size of the treatment system is often calculated based on acidity loadings from the MD discharge and estimated acidity removal rates (Cravotta 2008; Cravotta and Ward 2008; Cravotta et al. 2008; Hedin et al. 2010; Skousen et al. 2017).

Although limestone-based treatment systems are effective for treating MD, challenges can arise when designing and maintaining the systems. Metals in MD can armor and/or clog limestone when they precipitate as iron (III) hydroxide (Fe(OH)₃), aluminum (III) hydroxide (Al(OH)₃), gypsum, and other secondary minerals. Armoring can influence calcite dissolution rates and reduce treatment effectiveness; yet, results of several studies suggest that the impacts of armoring could be relatively small, potentially causing <20% reductions in calcite dissolution rates (Cravotta, 2008; Ziemkiewicz et al. 1997). Clogged limestone also can lead to preferential flow paths, reduced contact time, and consequently reduced treatment efficiency (Cravotta 2008). To alleviate clogging issues, limestone-based treatment systems can be designed to include effluent valves that can be opened manually or automatically at specified time intervals to flush precipitates from the system. Although the effectiveness of flushing systems has been questioned, a field-based experiment revealed that flushing could remove up to 93% of the armoring or clogging in the limestone (Hedin Environmental 2008; Rose et al. 2004).

Preferential flow paths and short circuiting in heterogenous porous media, such as limestone-based treatment systems, result from water following the path of least resistance and can have significant impacts on MD treatment performance. Limestone-based treatment systems are designed such that the MD will be in contact with the limestone long enough to neutralize the MD acidity and produce net alkaline water. However, in horizontal flow systems where MD is added to one end of the limestone bed, the method for introducing the MD to the system and spreading it along the cross-sectional area of the bed (e.g., a pipe, header pipe, rock channel, influent weir box, or baffle wall) can influence short circuiting and the hydraulic retention time (HRT). For example, in a horizontal flow oxic limestone bed near Hegins, PA (Schulkill County), the HRT was estimated as 5.5 hr based on the volume of the basin, limestone porosity, and flow rate; however, a sodium bromide (NaBr) tracer test indicated that the HRT was 2.5 hr (Cravotta et al. 2004). One explanation for the discrepancies between the estimated and measured HRTs is related to preferential flow paths (Cravotta et al. 2004). In another system near Reevesdale, PA (Schuylkill County), MD was added to one end of a buried limestone channel and then moved horizontally through the limestone and exited at the other end of the system. Shortly after construction, treatment effectiveness decreased, which was later attributed to metal accumulation, clogging, and short circuiting near the influent of the system (Cravotta 2008). Short circuiting and partial clogging also were hypothesized to complicate the effectiveness of treating MD in other limestone-based treatment systems (Cravotta 2008; Rakotonimaro et al. 2018; Rötting et al. 2007, 2008).

Some of the MD treatment challenges discussed above could be overcome by treating MD in batch rather than flowthrough mode. Conceptually, MD could be treated in batch by first collecting the MD and directing it into a holding pond equipped with a water level control structure and valve at the outlet. When the holding pond is full of MD, the effluent valve opens, flushing MD from the pond and into a limestone bed also equipped with a water level control structure and valve at the outlet. When the limestone bed is full, the MD can be held for some desired HRT; the valve can then be opened and the MD flushed into a settling pond. In the present study, we compared MD treatment effectiveness between the first batch operating limestone treatment system (BOLTS) ever built (near Puritan, PA) and a limestone-only auto-flushing vertical flow pond (AFVFP). Our hypothesis was that treating MD in batch reduces the clogging, short circuiting, and HRT challenges associated with flowthrough systems, thereby producing higher acidity removal rates in a BOLTS.

In 2020, an innovative BOLTS was built to treat the Puritan Mine discharge located near Portage, PA in Cambria County (40.366827° N, −78.645263° W) (Fig. 1). The discharge originated in an abandoned underground coal mine that was in operation from the late 1800s to early 1900s and had a median flow rate of 209 gal/min (791 L/min), pH = 3.1, 110 mg/L acidity, 6.0 mg/L iron, 1.4 mg/L manganese, 12.4 mg/L aluminum, and 589 mg/L sulfate. The system consists of a holding pond, primary limestone basin (BOLTS), and two settling ponds in series (SP1 and SP2). During high flow, overflow from the holding pond was directed to a smaller AFVFP, which emptied into the second settling pond (SP2).

In the novel design, mine water continually flows into a 1,980-yd³ (1,510-m³) holding pond at the start of the system. At the effluent end (or outlet) of the holding pond, a solar-powered Agri Drain Smart Drainage® (Smart Drain) water level control structure equipped with a float switch is used to determine when the holding pond is full, after which a valve opens and the holding pond drains into the BOLTS (Fig. 2B, Stage 1), which contains approximately 4,600 short tons (5,070 tonnes) of a high-calcium (∼94% CaCO₃ equivalent, 77% as CaCO₃) limestone mixture of size R-3, American Association of State Highway and Transportation Officials (AASHTO) #1, and AASHTO #3 stone. Using methods from Cravotta et al. (2008) and assuming the limestone’s shape most closely resembles an ellipsoid, the average unit surface area was calculated as 556 ft²/short ton (57 m²/tonne) for 30 pieces of limestone randomly selected from the BOLTS. As a comparison, Cravotta (2021) reported a surface area of 439 ft²/short ton (44 m²/tonne) for AASHTO #1 and 703 ft²/short ton (72 m²/tonne) for AASHTO #3. The bulk limestone volume of the BOLTS is approximately 3,400 yd³ (2,600 m³) and contains 1,007 yd³ (770 m³) of void volume. Based on the bulk volume, limestone mass, and void volume, the bulk limestone density in the BOLTS is 1.35 short ton/yd³ (1.95 tonne/m³) and the porosity is 30%. The void volume was calculated by measuring the time needed to fill the limestone bed from empty at a constant flow rate (376 gal/min × 540 min to fill × 1 yd³/202 gal = 1,007 yd³).

The BOLTS utilizes solar-powered Smart Drain controllers that allow the system to treat the MD in a batch configuration (Fig. 2B), which is different from conventional passive limestone treatment systems that are commonly designed to operate in flowthrough mode (Fig. 2A). Once the BOLTS is filled with MD from the holding pond, an upper float switch (high-level indicator) in the BOLTS is triggered, the outlet gate valve (fill valve) for the holding pond closes, and the MD is treated in batches for a user-specified HRT (Fig. 2B, Stage 2). After treating MD for the programmed HRT, the outlet valve for the BOLTS opens (the drain valve), and the treated MD is flushed into a 3,630-yd³ (2,775-m³) settling pond (Fig. 2B, Stage 3). When the BOLTS is empty, a bottom float switch (low-level indicator) is triggered and the outlet valve for the BOLTS closes. When the holding pond float switch is on (i.e., the holding pond is full) and the BOLTS float switches are off (i.e., the BOLTS is empty), another batch treatment cycle is initiated. The times between batch treatment cycles vary depending on the influent flow rate to the holding pond. For example, when flow rates are <282 gal/min (1,067 L/min), the holding pond will not be full by the time a 12-hr batch treatment cycle is complete (282 gal/min × 12 hr hold time × 60 min/hr × 1 yd³/202 gal = 1,007 yd³, the void volume of the limestone basin); a new batch treatment cycle will not start until the holding pond is full and the BOLTS is empty. When flow rates are >282 gal/min, the holding pond will be full before a 12-hr batch treatment cycle is completed in the BOLTS. In this scenario, the overflow is treated in a 1,531-yd³ (1,170-m³) AFVFP operating in flowthrough mode and containing 2,000 short tons of limestone and a flush volume of 394 yd³ (301 m³). Based on the bulk volume, limestone mass, and void volume, the bulk limestone density in the AFVFP is 1.31 short ton/yd³ (1.88 tonne/m³) and the porosity is 26%. The void volume was calculated by measuring the time needed to fill the limestone bed from empty at a constant flow rate (295 gal/min × 270 min to fill × 1 yd³/202 gal = 394 yd³). At the effluent of the AFVFP is a Smart Drain water level control structure programmed to drain the pond once every day for 90 min, but the program can be adjusted for any drain duration and frequency on a 24-hr schedule. All MD that passes through the BOLTS and the AFVFP collects in a final 2,650-yd³ (2,026-m³) settling pond before the treated water is discharged to Trout Run.

In the first set of experiments, the Puritan MD was treated in a flowthrough AFVFP configuration with 4.5- and 9.0-hr HRTs. These theoretical HRTs could be achieved in flowthrough treatment mode based on the size of the limestone beds and influent flow rates. For the 4.5-hr HRT, all flow from the holding pond (295 gal/min or 1,116 L/min) was bypassed around the BOLTS and into the AFVFP basin, which was programmed to flush for 90 min every 24 hr. The HRT was determined by recording the time needed to fill the AFVFP from empty with the continuously flowing MD. A Eureka Manta+20 water quality monitoring sonde (programmed to record water quality every 6 min) equipped with pH, conductivity, pressure, temperature, and dissolved oxygen probes and an ISCO 3700 water sampler (programmed to take samples every 2.75 hr during flowthrough and every 0.1 hr during flush) was installed in the Agri Drain box at the outlet of the AFVFP. Before the start of each 4.5- and 9.0-hr HRT experiment, the Manta+20 was calibrated with reference standards. Samples were collected over a 24-hr period (22.5 hr of flowthrough and 1.5 hr of flushing to empty) and analyzed for total (nonfiltered) and dissolved (<0.45-µm-pore-size nylon filter) metals at G&C Coal Analysis Laboratory (Summerville, PA) or Saint Francis University (Loretto, PA). Nonacidified samples were analyzed for alkalinity with a Hach titration kit and acidity at Geochemical Testing (Somerset, PA). For the 9.0-hr HRT flowthrough experiment, the holding pond and BOLTS flow control structures were altered to allow continuous flow from the holding pond into the limestone basin like a typical AFVFP operating in flowthrough mode (Fig. 2A). The 9.0-hr HRT was determined by recording the time needed to fill the BOLTS basin (operating in flowthrough mode) from empty with the continuous flow of MD from the holding pond at 376 gal/min (1,423 L/min). A sonde and ISCO water sampler were installed in the AFVFP Smart Drain outlet structure. The water sampling and analysis procedures for this experiment were the same as those for the 4.5-hr HRT flowthrough experiment.

For a comparison, the MD was also treated for 4.5- and 9.0-hr HRTs in batches using the BOLTS (Fig. 2B); during these two experiments, the average influent flow rates into the holding pond were 295 gal/min (1,116 L/min) and 272 gal/min (1,029 L/min), respectively. After the MD was treated for the set HRT, water samples were collected from the effluent ca. every 10 min while the BOLTS system flushed to empty. Collected water samples were analyzed for pH, dissolved oxygen, specific conductivity, alkalinity (with a Hach titration kit), acidity (G&C Coal Analysis Laboratory), and total and dissolved metals (G&C Coal Analysis Laboratory or Saint Francis University).

where A_u is the unit surface area of the limestone, ρ_b is the bulk density, and n is the porosity. To calculate A/V for the BOLTS and AFVFP near Puritan, PA, assumptions included A_u = 556 ft²/short ton (57 m²/tonne), n = 0.3 (the pore volume of the BOLTS [1,007 yd³] divided by the bulk volume [3,400 yd³]), and limestone bulk density (ρ_b) of 1.35 short tons/yd³ (the short tons of limestone in the BOLTS [4,600 short tons] divided by the bulk volume [3,400 yd³]). Thus, A/V = 2500 ft²/yd³ (370 m²/m³).

After calculating the mass of limestone, the bulk density (1.35 short tons/yd³) was used to estimate the limestone basin excavation volume, which included installation of a 6-in. clay liner. For the BOLTS, the holding pond excavation volume was estimated as the limestone basin volume times the limestone porosity (assumed to be 30% based on the data collected for the BOLTS) plus an additional 30% of volume to account for solids accumulation. Each basin was assumed to have a depth of 8 ft, side slopes of 2:1, and a length:width of 5:1. Computed dimensions for each limestone basin were used to estimate costs of geotextile liners and pipes. Liner costs accounted for the bottom and all side slopes of the basins. HDPE pipes used for drainage were designed with a central 12-in. pipe spanning the length of the limestone beds and 6-in. perforated pipe branches that extended from the central 12-in. pipe to the outer width of the basin at 20-ft intervals. Installation cost for the pipe was included in their unit cost. The Agri Drain flow control structure for the BOLTS included a 12-ft-tall box and valve for the holding pond and a 14-ft-tall box and valve for the limestone basin outlet. The Agri Drain flow control structure for the AFVFP included a 14-ft-tall box and outlet valve for the limestone basin. Installation costs for the Agri Drain flow control structures were included in their unit cost. Minor components, such as fittings and valves, were not included in the analysis.

In both 4.5- and 9.0-hr theoretical HRTs, the BOLTS design outperformed the AFVFP (Table 1). Under BOLTS conditions, the average pH during flush was 5.50 for the 4.5-hr HRT and 5.71 for the 9.0-hr HRT, whereas the AFVFP during flowthrough was 3.93 and 4.70 at the respective HRTs. The BOLTS also outperformed the AFVFP in acidity removal. The influent MD had approximately 300 mg/L of net acidity; however, effluent from the BOLTS had no remaining acidity under both HRTs and produced an average alkalinity of 29.0 mg/L for the 4.5-hr HRT and 50.2 mg/L for the 9.0-hr HRT. In contrast, during flowthrough, the AFVFP produced <1 mg/L of alkalinity and average acidity concentrations of 182.7 mg/L at the 4.5-hr HRT and 71.46 mg/L at the 9.0-hr HRT. In addition, the bucket tests produced pH, acidity, and alkalinity values that were similar to those of the BOLTS.

Compared with the AFVFP, the BOLTS effluent pH, alkalinity, and acidity was more consistent (Fig. 3). For the 9-hr HRT test in the AFVFP, it took 9 hr to fill the limestone bed (Fig. 3A). After the bed was full and partially treated MD was leaving the system, the effluent pH increased gradually from 3.82 at 9.12 hr to 4.90 at 23.1 hr; alkalinity remained near 0 mg/L during this time (Fig. 3A). When the system was flushed at 24 hr, the flush water chemistry had improved, with pH and alkalinity reaching maximums of 5.41 and 40.17 mg/L, respectively. In contrast, the chemistry of the water flushed from the BOLTS was relatively consistent (average pH = 5.72 ± 0.21; average alkalinity = 50.2 ± 17.3 mg/L).

Alkalinity generation and acidity removal rates were determined assuming first order kinetics for the BOLTS, AFVFP, and bucket test (Fig. 4). The rate of alkalinity generation was highest in the bucket test (0.039 hr⁻¹), relatively similar to that of the BOLTS (0.032 hr⁻¹); no alkalinity was generated for the AFVFP during flowthrough treatment. Analogous to alkalinity generation, the rate of acidity removal in the BOLTS and the bucket test were similar (−0.31 and −0.26 hr⁻¹, respectively) but higher than the AFVFP (−0.08 hr⁻¹). The A/V was calculated (Equation 5) for each system assuming that the BOLTS and AFVFP had limestone surface areas and porosities similar to those of physical analyses of limestone collected from the system. Assumptions were a limestone surface area (A_u) of 556 ft²/short ton (57 m²/tonne), n = 0.3, and limestone bulk density (ρ_b) of 1.35 short tons/yd³. Using these values, A/V = 2,500 ft²/yd³ (370 m²/m³).

In comparison to the BOLTS, the AFVFP had decreased acidity removal likely because of preferential flow paths and short circuiting that resulted in an HRT lower than the theoretical (i.e., design) HRT. During flowthrough, the AFVFP (operating with a 9-hr theoretical HRT) had an average pH of 4.7 ± 0.36 and an average alkalinity of 0.53 ± 0.43 mg/L. At the start of AFVFP flushing, the alkalinity was 1.1 mg/L, which increased to 40 mg/L 5 min after flushing. At 20 min after flushing, the alkalinity decreased to 21 mg/L and decreased to 0 mg/L at 60 min after flushing (Fig. 3A). During flowthrough in the AFVFP, some parcels of MD probably flowed through the AFVFP relatively quickly, whereas other parcels were retained for longer. Because water chemistry is influenced by limestone contact time (i.e., HRT), the variable water chemistry during flushing was likely a result of these parcels of water (with different HRTs) being flushed from the system. In contrast, the chemistry of the MD from the effluent of the BOLTS was relatively consistent (average pH = 5.72 ± 0.21; average alkalinity = 50.2 ± 17.3 mg/L). The differences in treatment results between the BOLTS and AFVFP were likely due to differences in contact time with limestone. In the BOLTS, the MD filled the limestone bed very quickly (ca. 45 min) and was held in the system for 9 hr before being flushed out. The design of a BOLTS ensures that all the MD is in contact with the limestone for the programmed hold time, whereas preferential flow paths, short circuiting, and long fill times in AFVFP can lead to shorter than the theoretical HRTs.

Treatment of the Puritan MD with a BOLTS required less limestone than did treatment with an AFVFP. Equation 7 was used to estimate the mass of limestone needed to treat the Puritan MD with a BOLTS or AFVFP. Assumptions included a design flow rate of 300 gal/min, influent net acidity of 250 mg/L, desired effluent net acidity of −20 mg/L, desired treatment longevity of 20 years (assumed from Pennsylvania Code Title 25, Chapter 87.102(e)(4)(vi)), limestone purity of 0.9, unit surface area of 556 ft²/short ton of limestone, porosity of 30%, k_sa of −1.2 × 10⁻⁴ yd³/ft²hr (−8.3 × 10⁻⁴m³/m²hr) for the BOLTS, and k_sa of −3.1 × 10⁻⁵ yd³/ft²hr (−2.1 × 10⁻⁴m³/m²hr) for the AFVFP. Entering the assumptions into Equation 7, a limestone bed designed as a BOLTS would require 5,945 short tons (6,553 tonnes) of limestone and an AFVFP would require 11,679 short tons (12,873 tonnes). From a design perspective, a reduced limestone mass would result in a cost savings and smaller acreage needed for site development. The higher limestone masses for the AFVFP are reflective of short circuiting that can occur in a horizontal flow system, thereby requiring more limestone to achieve HRTs suitable for the desired effluent quality.

Depending on the flow rate and MD water chemistry, a limestone-based treatment system designed as a BOLTS could be cheaper than an AFVFP. The BOLTS at Puritan requires a holding pond, three float switches, two solar-powered Smart Drain water level control structures, and a limestone bed. In contrast, an AFVFP would require one solar-powered Smart Drain water level control structure and a limestone bed. The BOLTS could have lower limestone costs than an AFVFP; however, the extra technology needed for a BOLTS could make it more expensive than an AFVFP. Therefore, an economic analysis was performed to estimate the total costs for treating 100, 200, 300, or 400 gal/min of Puritan MD and to determine whether design flow rate could be used for which treating the MD in a BOLTS was more cost-effective than using an AFVFP. For these calculations, assumptions included an influent net acidity of 250 mg/L, final effluent net acidity of −20 mg/L, desired treatment longevity of 20 years, limestone purity of 0.9, a limestone mixture of AASHTO #1 and AASHTO #3 (556 ft²/short ton or 57 m²/tonne), porosity of 30%, and k_sa values of −1.2 × 10⁻⁴ yd³/ft²hr (−8.3 × 10⁻⁴m³/m²hr) for the BOLTS and −3.1 × 10⁻⁵ yd³/ft²hr (−2.1 × 10⁻⁴m³/m²hr) to −1.2 × 10⁻⁴ yd³/ft²hr (−8.3 × 10⁻⁴m³/m²hr) for AFVFP. Regardless of the MD flow rate, the BOLTS was always cheaper than the AFVFP when limestone masses for the AFVFP were calculated with the lower k_sa value (−3.1 × 10⁻⁵ yd³/ft²hr) (Fig. 5). At 300 gal/min, the holding pond and limestone basin for the BOLTS would cost $201,790 whereas the AFVFP would cost $301,419. The addition of two solar-powered Smart Drain water level control structures for the BOLTS adds roughly $20,000 in total costs compared with the single solar-powered Smart Drain structure required for the AFVFP. However, compared with the BOLTS, the increased amount of limestone and excavation required for the AFVFP resulted in higher costs. The higher limestone costs for the AFVFP were a result of the low k_sa, which was likely a result of short circuiting. Therefore, the analysis was repeated assuming that the maximum possible k_sa for the AFVFP would be similar to that calculated for the BOLTS. When using a k_sa of −1.2 × 10⁻⁴ yd³/ft² for the AFVFP, the cost of the AFVFP was cheaper than that of the BOLTS at all flow rates. The true cost of the AFVFP is likely somewhere in between the costs calculated based on the low and high k_sa values. Using the low k_sa for the AFVFP, the BOLTS would be 18% cheaper than the AFVFP at 100 gal/min or 35% cheaper at 400 gal/min. Using the high k_sa for the AFVFP, the BOLTS would be approximately 20% more expensive than the AFVFP at 300 gal/min (Fig. 5).

The concept of treating MD in a BOLTS could also lead to new ideas for lowering MD treatment costs. For example, larger limestone sizes (AASHTO #1 to AASHTO #3) are often preferred in AFVFP limestone beds because of their larger void space, which may help prevent clogging issues. However, flushing MD in and out of a BOLTS also could reduce clogging concerns and allow engineers to experiment with smaller stone sizes. Reducing stone size increases the surface area and decreases the mass of limestone needed for MD treatment. Cravotta (2021) reported surface areas of 439 ft²/short ton (45 m²/tonne) for AASHTO #1, 703 ft²/short ton (72 m²/tonne) for AASHTO #3, 1,406 ft²/short ton (144 m²/tonne) for AASHTO #5, and 2,822 ft²/short ton (289 m²/tonne) for AASHTO #7. Based on Equation 7 and assuming that AASHTO #7 would be used to treat 300 gal/min of the Puritan MD in a BOLTS, the limestone mass required for 20 years of treatment would be 4,341 short tons (4,785 tonne) instead of the 5,945 short tons (6,553 tonnes) calculated for the mixture of AASHTO #1 and #3; the reduced limestone mass from using smaller stone size would save $40,000 ($157,000 vs. $201,000) (Fig. 5). This is just one example of modifications that could be explored for lowering MD treatment costs; other innovations might be possible for treating MD in a BOLTS.

A BOLTS is an innovative and effective method for treating MD. In the present study, use of a BOLTS resulted in higher alkalinity concentrations and pH per theoretical HRT than did use of an AFVFP operating in flowthrough mode. Higher alkalinity generation per theoretical HRT translates to less limestone mass and land area required for treatment. Although a BOLTS may have lower limestone costs than an AFVFP designed to achieve the same level of treatment, some extra components (e.g., holding pond, three float switches, and two solar-powered Smart Drain water level control structures) are required for a BOLTS, which add to the cost. Therefore, cost-benefit analyses (such as those in this study) should be performed for determining whether a BOLTS, AFVFP, or other technology is most appropriate for treating MD within project constraints (e.g., desired effluent water quality from the treatment system, land availability, and project budget). Bucket tests should also complement the cost-benefit analyses so limestone mass, land area requirements, and effluent water quality can be accurately predicted for the BOLTS design.

In comparison to an AFVFP, extra float switches, solar-powered water level control structures, electronics, and moving parts in a BOLTS could lead to more complex operation and maintenance (O&M). For example, in January 2021, the actuator arm for the holding pond effluent valve froze, preventing MD from being flushed into and treated in the BOLTS. Cold air was entering the effluent pipe from the holding pond (the influent pipe for the BOLTS), causing ice accumulation in the Smart Drain water level control structure. A hinged flapper was placed on the end of the effluent pipe, resolving the freezing issue. In summer and fall 2023, the BOLTS was routinely getting stuck between stage 3 and 1 (Fig. 2B). The issue was attributed to a faulty float switch and/or rain runoff getting into the system when the holding pond was refilling between treatment cycles. The low-level float switch became stuck in the “on” position, preventing MD from being flushed into the BOLTS. To alleviate this issue, the low-level float switch could be removed entirely, or a timer could be used to indicate the time needed to flush the BOLTS to empty rather than relying on a float switch to signal when the system is empty. Although the BOLTS system near Puritan has experienced some O&M challenges, many of these challenges can be overcome.

Future research could be conducted to explore ways to improve MD treatment in a BOLTS and/or use the idea of treating MD in batches to develop new passive treatment technologies. Reduction in stone size in a BOLTS could decrease the limestone mass, land area, and cost required for MD treatment. Other research on passive MD treatment could focus on new methods for creating a variety of redox environments in batch reactors (operating in a manner similar to a BOLTS), which could improve sulfate removal, dissolved metal removal, and alkalinity generation. The technology in a BOLTS also could be adapted to treat larger MD discharges that commonly require active MD treatment technologies with high O&M costs (Pennsylvania Department of Environmental Protection 2022).

The authors acknowledge the following individuals for their help on the project: Saint Francis University environmental engineering students Mark Koskinen and Jacob Jolly, who assisted with water sampling and data analysis; Dan Guy (BioMost, Inc.) for assisting with programing and troubleshooting challenges related to the auto-flushing valves in the Agri Drain boxes of the BOLTS; and Joel Bandstra (Environmental Engineering, Saint Francis University) for assistance with data analysis and processing ideas for treating MD in a BOLTS.