Co-treatment of acid mine drainage with municipal wastewater: performance evaluation
- First Online:
- Cite this article as:
- Hughes, T.A. & Gray, N.F. Environ Sci Pollut Res (2013) 20: 7863. doi:10.1007/s11356-012-1303-4
Co-treatment of acid mine drainage (AMD) with municipal wastewater (MWW) using the activated sludge process is a novel treatment technology offering potential savings over alternative systems in materials, proprietary chemicals and energy inputs. The impacts of AMD on laboratory-scale activated sludge units (plug-flow and sequencing batch reactors) treating synthetic MWW were investigated. Synthetic AMD containing Al, Cu, Fe, Mn, Pb, Zn and SO4 at a range of concentrations and pH values was formulated to simulate three possible co-treatment processes, i.e., (1) adding raw AMD to the activated sludge aeration tank, (2) pre-treating AMD prior to adding to the aeration tank by mixing with digested sludge and (3) pre-treating AMD by mixing with screened MWW. Continuous AMD loading to the activated sludge reactors during co-treatment did not cause a significant decrease in chemical oxygen demand (COD), 5-day biochemical oxygen demand, or total organic carbon removal; average COD removal rates ranged from 87–93 %. Enhanced phosphate removal was observed in reactors loaded with Fe- and Al-rich AMD, with final effluent TP concentrations <2 mg/L. Removal rates for dissolved Al, Cu, Fe and Pb were 52–84 %, 47–61 %, 74–86 % and 100 %, respectively, in both systems. Manganese and Zn removal were strongly linked to acidity; removal from net-acidic AMD was <10 % for both metals, whereas removal from circum-neutral AMD averaged 93–95 % for Mn and 58–90 % for Zn. Pre-mixing with screened MWW was the best process option in terms of AMD neutralization and metal removal. However, significant MWW alkalinity was consumed, suggesting an alkali supplement may be necessary.
KeywordsAcid mine drainageActivated sludgeCo-treatmentMetalsNeutralizationSewageRemediationWastewater treatment plant
Acid mine drainage
Five-day biochemical oxygen demand
Chemical oxygen demand
Hydraulic retention time
Mixed liquor suspended solids
Sequencing batch reactor
Solids retention time
Sludge volume index
Total organic carbon
Wastewater treatment plant
Acid mine drainage (AMD) is a major global pollutant caused by the weathering of mineral surfaces exposed during mineral extraction. Oxidation of minerals (particularly pyrite, FeS2) releases dissolved metals, acidity and SO4 (Stumm and Morgan 1981), which contaminate infiltrating waters, leading to severe pollution of surface waters and aquifers in the vicinity of the mine (Nordstrom and Alpers 1999). Acid mine drainage is a significant pollution issue in the US (Cravotta III et al. 2010; US EPA 2000), and the associated fiscal burden is enormous, with estimated cleanup costs at 88 National Priorities List mining sites in excess of $2.8 billion as of April 2002 (US EPA 2004). In Europe, AMD contamination is also a widespread issue, and mine water management is a major concern (Wolkersdorfer and Bowell 2004). A broad range of treatment strategies has been developed and implemented, with remediation typically involving neutralization by an alkali supplement and metal removal via precipitation or adsorption processes (Hedin et al. 1994; Skousen et al. 1998; Watzlaf et al. 2004).
The activated sludge process is a biological wastewater treatment process widely used for domestic, municipal and industrial wastewaters (Gray 1990). It is used to treat a wide variety of recalcitrant and potentially toxic wastewaters, normally in admixture with domestic sewage. In the primary treatment stage, solids are removed from screened wastewater as primary sludge by sedimentation. In the secondary treatment stage, the settled wastewater is mixed in an aeration tank with activated sludge, a low-density (typically 1.5–3.5 g solids/L) sludge that is comprised of a diverse population of bacteria, fungi, protozoa, rotifers and nematodes (Gray 2004). The metabolic activity of these microorganisms naturally degrades organic matter and nutrients, which are removed from wastewater by various mechanisms, e.g. adsorption onto sludge flocs, mineralization, assimilation and oxidation (Comeau 2008). A key element of the biomass structure of activated sludge is the flocculant nature of the solids, resulting from the bacterial formation of extracellular polymeric substances (Brown and Lester 1979). A fraction of the activated sludge is displaced from the main aeration tank into a settlement tank, where the sludge solids (primarily flocculated biomass) settle, and the clarified effluent may then be discharged. The settled sludge (returned activated sludge) is returned to the aeration tank, where it serves as an inoculum. Any excess sludge is disposed of and receives further treatment if required.
Co-treatment of AMD and municipal wastewater (MWW) using the activated sludge process is an innovative approach to AMD remediation that uses MWW alkalinity and the adsorptive properties of MWW particulates and activated sludge biomass to remove acidity and metals from AMD-impacted waters while potentially improving the efficiency of MWW treatment (Hughes and Gray 2013). Metal concentrations in MWW are relatively low compared with AMD, with individual metal concentrations typically <500 μg/L (Chipasa 2003; Oliver and Cosgrove 1974; Santos and Judd 2010); therefore, adsorption sites on MWW particulates and sludges remain potentially available for metal uptake. Also, the high concentrations of suspended solids in MWW may enhance Fe removal by (oxy)hydroxide precipitation by serving as nucleation sites for Fe (Johnson and Younger 2006). In turn, Fe- and Al-(oxy)hydroxides may provide attachment sites for bacteria which have important roles in nutrient removal from MWW, i.e., nitrifying and denitrifying bacteria (Demin and Dudeney 2003). The Fe in AMD can also be used as an effective substitute for commercial coagulants (Rao et al. 1992; Sandström and Mattsson 2001). Another key advantage of co-treatment is that MWW is an alkaline material and therefore has acid-neutralizing capacity. Dilution and neutralization of AMD by mixing with alkaline MWW cause the pH to increase, leading to increased removal of dissolved metals (Jiménez-Rodríguez et al. 2009; Strosnider and Nairn 2010).
Effective, sustainable co-treatment of AMD and MWW depends on the capacity of the activated sludge process to withstand high loadings of acidity, heavy metals and SO4. It is also critical that final effluent quality does not deteriorate, especially in terms of metals, acidity and SO4. Recent findings provide strong indicators that co-treatment is feasible. Hughes and Gray (2012) demonstrated that spiked addition of high concentrations of synthetic high-strength AMD did not cause significant decreases in sludge oxygen uptake rate (OUR) in activated sludge from municipal WWTPs. Acclimatization of activated sludge was also observed over a relatively short period of continuous AMD loading (i.e., approximately 3 weeks), although decreases in protozoa abundance and maximum sludge OUR were observed (Hughes and Gray 2012). Hughes and Gray (in review) observed metal removal rates >90 % for Al, Cu, Fe and Zn in batch tests with MWW and activated sludges from municipal wastewater treatment plants (WWTPs). Neutralization by screened MWW caused the pH of AMD to increase from pH 2.8 to 6.2 at 50 % v/v (Hughes and Gray 2013). Passive co-treatment of MWW and AMD in a multi-stage system (consisting of a primary clarifier, a reducing and alkalinity producing system and an aerobic wetland) consistently removed significant concentrations of dissolved Al, As, Cd, Fe, Mn, Pb and Zn and resulted in a net-alkaline effluent (Strosnider et al. 2011a, c), as well as achieving high organic matter and nutrient removal efficiencies (Strosnider et al. 2011b) and pathogen removal (Winfrey et al. 2010). Johnson and Younger (2006) reported removal of Fe and Mn from net-alkaline coal mine AMD, as well as phosphate, nitrate and suspended solids removal, in a wetland co-treating AMD with secondary sewage effluent. Neto et al. (2010) mixed coal mine AMD and sewage in 1:1 volumetric ratios and reported significant decreases in acidity and Fe and Mn concentrations, as well as decreased concentrations of organic matter and nutrients and complete removal of pathogens. Removal of Pb and Ni (Sirianuntapiboon and Ungkaprasatcha 2007) and Fe, Cu and Mn (Marandi et al. 2007) using sequencing batch reactors (SBRs) has also been observed.
Appropriate sizing of an AMD system, i.e., designing a system that treats the AMD to the desired level efficiently without using excessive space (e.g. treatment wetlands) or materials (e.g. volumes of neutralizing agents) is critical for achieving economical and adequate AMD remediation. The sizing process for designing active or passive treatment systems is commonly based on a set of key AMD parameters, including influent acidity (Zipper and Skousen 2010) and metal concentrations (Hedin and Nairn 1992). The metal removal capacity of activated sludge is of critical importance when designing for co-treatment. The required alkalinity supply must also be considered when designing for treatment of net-acidic AMD. In AMD treatability studies (Hughes and Gray 2012) and metal removal and neutralization studies (Hughes and Gray 2013; Hughes et al. in review), it was demonstrated that alkalinity and buffering processes are very important factors in co-treatment, both in terms of toxicity to activated sludge and the impact of pH on metal removal.
The aim of this study was to examine the impacts of continuous AMD loading to laboratory-scale activated sludge reactors, in terms of removal of chemical oxygen demand (COD), 5-day biochemical oxygen demand (BOD5), total organic carbon (TOC), total phosphorus (TP), total nitrogen (TN), suspended solids (SS), metals, acidity and SO4, as well as in terms of sludge morphological characteristics. This is the first reported study which uses such a broad range of effluent and sludge parameters to compare remediation of AMD at a range of strengths in activated sludge reactors with different hydraulic retention times (HRT), i.e., plug flow reactors and SBRs.
Materials and methods
Three different co-treatment processes and one control were simulated at laboratory-scale (Electronic supplementary material 3, Fig. 1). Process I was the control, with no AMD added. Process II simulated the addition of raw (untreated) AMD to the aeration tanks. Process III simulated pre-treatment of AMD by mixing with digested sewage sludge, sending the mixture to a sedimentation tank and finally adding the supernatant to the aeration tanks. Process IV simulated pre-treatment of AMD by mixing with screened MWW prior to adding to the aeration tanks.
Dissolved metals, sulfate and acidity concentrations in synthetic AMD
Average concentrationa (mg/L)
AMD: process II
AMD: process III
AMD: process IV
Two process configurations, i.e., plug flow and SBRs, were operated at laboratory-scale to simulate co-treatment of AMD and MWW. Operating parameters for reactor volume, sedimentation tank volume, mixed liquor suspended solids (MLSS), food-to-microorganism (f/m) ratio, sludge retention time (SRT), HRT and dissolved oxygen were 6 L, 2 L, 2.5–3.0 g/L, 0.25 kg BOD5/kg MLSS/day, 3–4 days, 24 h and >2 mg/L, respectively, in the plug flow reactors, and 3.4 L, none, 2.5–3.0 g/L, 0.10 kg BOD5/kg MLSS/day, 25 days, 48 h and >2 mg/L, respectively, in the SBRs.
The plug flow reactors (Bio-Simulator, ISCO, Italy) consisted of continuously aerated high-density polyethylene (HDPE) reactors (working volume, 6 L) connected by clear vinyl tubing to HDPE sedimentation tanks (2 L) (Electronic supplementary material 3, Fig. 2). Four plug flow reactors, simulating Processes I, II, III and IV, respectively, were operated simultaneously. Influent and effluent flows and sludge recycle were controlled using the instrument control panels and peristaltic pumps connected to clear vinyl tubing. The pH and temperature in the aeration tanks of the plug flow reactors were measured daily, prior to adding fresh influent, using a portable pH meter (HI 991 300, Hanna Instruments), which was calibrated weekly. Influent (synthetic MWW + AMD) (6 L) was added daily, resulting in a HRT of approximately 24 h.
The SBRs, developed by Dubber and Gray (2011), consisted of glass reactor vessels (working volume, 3.4 L) (Electronic supplementary material 3, Fig. 3). Four SBRs, simulating Processes I, II, III and IV, were operated simultaneously. The SBRs were aerated using porous ceramic airstones attached to aquarium air pumps and were kept covered to minimize ingress of oxygen during anoxic and anaerobic periods. Magnetic stirrers (SB161, Stuart Scientific, UK) were used to ensure homogenous mixing during reaction. Influent was supplied and effluent was drawn using peristaltic pumps (iProcess, USA). All instruments and the operation cycles were computer-controlled using programmable external timer power control units (IP Power 9258, Audon Electronics, UK). The SBRs were operated using one five-stage cycle per day, i.e., fill (30 min), react (22 h), waste (as required), settle (1 h) and draw (30 min). Activated sludge was wasted only as necessary, based on MLSS measurements. During each cycle, 1.7 L effluent was exchanged with 1.7 L influent (synthetic MWW + AMD), giving a HRT of approximately 48 h.
Activated sludge inoculum
To obtain a diverse activated sludge microbial inoculum, mixed liquor to seed the reactors was collected from three different nitrifying municipal WWTPs, located in Leixlip, Co. Kildare, Swords, Co. Dublin, and Athy, Co. Kildare. The samples were mixed in equal proportions (by volume) in the laboratory and then introduced into the reactors, so that every reactor received the same activated sludge inoculum. Swords WWTP incorporates an anoxic stage to perform nitrogen removal via denitrification, so mixed liquor from this WWTP was considered to be a particularly appropriate inoculum for the SBRs, which were designed for denitrification. Reactors were operated for a start-up period of 12 days to acclimatize to the synthetic MWW before AMD loading began. Substrate selectivity can occur after prolonged use of a single synthetic wastewater in laboratory-scale experiments (Meriç et al. 2003). To maintain a diverse microbial population and avoid confounding results due to substrate selectivity, 10 % of the mixed liquor in each reactor was replaced with freshly collected activated sludge from either Leixlip WWTP or Swords WWTP every 7 days (Berg and Nyholm 1996).
Synthetic MWW and AMD preparation
Synthetic MWW was prepared by adding the following components to 1 L distilled water to make a 100-fold concentrate: peptone (16 g), meat extract (11 g), urea (3.0 g), K2HPO4 (2.8 g), NaCl (0.7 g), CaCl2•2H2O (0.4 g) and MgSO4•7H2O (0.2 g) (OECD 1984). The concentrated synthetic MWW was stored at −18 °C and thawed and diluted as required to the desired f/m ratio. Sodium hydrogen bicarbonate (NaHCO3) was added as an additional buffer against pH drop due to nitrification at a final concentration of 0.3 g/L synthetic MWW, based on the fact that oxidation of ammonia to nitrite is an acid-generating process, which consumes 7.14 mg alkalinity as CaCO3 for every milligram of ammonia that is nitrified (Christofi et al. 2003; Ekama and Wentzel 2008).
Synthetic AMD was prepared fresh daily from stock metal solutions and distilled, deionized water (Gray and O'Neill 1995). The pH was adjusted as required with sulfuric acid (H2SO4) or sodium hydroxide (NaOH). All chemicals were of analytical reagent grade. The hypothetical WWTP influent was 6,000 m3/day. The combined median flows from the major adits in the Avoca mining district total 30 L/s, i.e., approximately 2,600 m3/day. Based on these values, the volumetric ratio of synthetic MWW to AMD was held at a constant ratio of 2:1 (v/v) in all processes.
Physico-chemical analysis of process performance
Process performance was quantified in terms of COD, BOD5, TOC, TN, TP and metal removal rates (percent). To compare reactor performance, ANOVA was performed (α = 0.05) using Minitab 15 (Minitab 2007).
Activated sludge biomass assessment
Activated sludge samples (50 mL) were taken weekly from the aeration tanks of the plug flow reactors and from the SBRs during the react phase, for a total of five sampling events. The MLSS was measured gravimetrically according to standard methods (APHA et al. 2005), and MLSS values were used to determine appropriate sewage loads to maintain a constant f/m ratio in all reactors. Microscopic analysis was used to assess the presence of protozoa and rotifers, floc morphology and filamentous growth. For microscopic analysis, 25-μL sub-samples were examined using phase contrast microscopy at ×100 magnification within 8 h of collection (Dubber and Gray 2009) to detect the presence of ciliate protozoa and rotifers. Floc morphology was analyzed for general shape (e.g. spherical or irregular), structure (e.g. compact or diffuse) and size. Three size classifications were used, i.e., small floc (<75 μm diameter), normal range (75–1,000 μm diameter) or large floc (>1,000 μm diameter). Microstructure (small flocs, spherical and compact in structure, few filaments) and the presence of pin flocs (pure microstructure with no filaments) were noted (Gray 2004). Filament abundance was scored from 0 to 6, according to the subjective scale in Jenkins et al. (2004), with a score of 0 indicating that no filaments were observed and scores of 1–6 indicating few, some, common, very common, abundant and excessive filaments, respectively. Settled sludge supernatant was visually assessed for clarity or turbidity. Finally, on the last day of the experiment, the sludge volume index (SVI) was determined on duplicate samples from each reactor (APHA et al. 2005).
The physico-chemical characteristics of synthetic MWW and AMD samples were measured, and actual composition of influent (accounting for dilution) was determined for use in calculating removals. The average COD, BOD5, TOC, TN and TP influent concentrations were 490.0, 422.9, 258.0, 91.0 and 9.5 mg/L, respectively, in the plug flow system and 247.1, 200.8, 182.1, 64.3 and 6.7 mg/L, respectively, in the SBRs. Metals, SO4 and net acidity concentrations of AMD are given in Table 1. The alkalinity of the synthetic MWW with added sodium bicarbonate was relatively high, with an average value of 488.1 mg/L as CaCO3, similar to MWW from a hard water catchment (McKinney 2004). The pH for the synthetic MWW was within the normal range for a municipal WWTP (Henze and Comeau 2008), with average pH 8.1.
MWW treatment performance
Final effluent BOD5 concentrations were determined for effluents sampled on the last two sampling dates (days 30 and 39) (Electronic supplementary material 1). Removal efficiency ranged from 94–99 % in the plug flow reactors and from 89–94 % in the SBRs over the two sampling dates. On both sampling dates, effluent BOD5 was lowest in Process II reactors, ranging from 5.1–10.7 mg/L in the plug flow system and from 12.1–13.8 mg/L in the SBRs.
Final effluent TOC concentrations in the plug flow system were generally <12 mg/L, with average removal efficiency of 96 % (Fig. 1c). Final effluent TOC concentrations in the SBRs were generally <8 mg/L, with average removal efficiency of 96 % (Fig. 1d). Removal was nearly constant in all reactors, with no significant differences (α = 0.05) between removals in different reactors within each system, and no significant differences (α = 0.05) between removals in reactors in different systems.
Final effluent TN concentrations in the plug flow system were very high in all reactors, and increasing TN concentrations indicated that removal of N was not occurring (Fig. 1e). Only in a few cases on the first and second sampling dates were effluent concentrations less than the average influent concentration, and after the third sampling date, significant accumulation of N in the test reactor effluents was observed. In the SBRs, TN removal did occur in all reactors during the second and third sampling dates (Fig. 1f). On these dates, average removal in Processes I, II, III and IV ranged from 15–18 %, 39–48 %, 15–25 % and 13–18 %, respectively. However, TN removal ceased after the third sampling date, and from then on, significant TN accumulation was observed in all reactors.
Average TP removal efficiency in the plug flow system was 95 % in Process II and 82 % in Process III, with final effluent concentrations <2 mg/L in both reactors (Fig 1g). In contrast, TP removal was not occurring in Processes I and IV. There was one anomalous value measured in Process III effluent (day 22), but removal efficiency on the next two sampling dates was in line with the high removal efficiency observed on the first three sampling dates. In the SBRs, average TP removal efficiency was 94 % in Process II and 84 % in Process III, with final effluent concentrations <2 mg/L in both reactors (Fig. 1h). Comparing reactors within the same systems, the TP removal in Processes II and III was significantly greater (p < 0.05) than removal in Processes I (control) and IV for both systems. There were no significant differences (α = 0.05) between removals in reactors in different systems.
Activated sludge biomass assessment
Data from the activated sludge biomass assessments are provided in Electronic supplementary material 2. In the plug flow system, ciliated protozoa and rotifers were initially present in samples from all reactors on the first sampling date. Overall abundance of grazers decreased in all reactors for a short period, but recovered over time, and on the last two sampling dates, ciliated protozoa and rotifers were again observed in samples from all reactors. Floc morphology in the plug flow reactors was small, irregular and compact, with microstructure observed, on the first sampling date. A general change to a diffuse floc structure was observed in Processes III and IV whereas flocs remained more compact in Processes I (control) and II. In Samples from Processes I (control) and III, characteristic floc size increased from small to normal. Pin flocs were observed in all reactors over the last two sampling dates. Filament index values dropped from 5 to 3 in Process I (control) and from 5 to 1–2 in Processes II, III and IV. Effluents from all reactors were turbid on the first sampling date, with gradual improvements noted in samples from Processes I (control), II and IV. Effluent turbidity in samples from Process III improved on earlier sampling dates but subsequently deteriorated. Total SS concentrations were higher than typical discharge limits in all reactors (e.g. 30 mg/L) (Code of Federal Regulations (CFR) 2006) (Electronic supplementary material 3, Fig. 4a).
In the SBRs, ciliated protozoa were observed in samples from all reactors on all sampling dates. In contrast, rotifers were not commonly observed until the last two sampling dates. Floc morphology on the first sampling date was generally small, irregular and diffuse with microstructure observed. There were no real changes in terms of floc shape during the test period. However, floc size notably increased from small to normal over the second, third and fourth sampling dates. Finally, large open flocs with loss of ideal compact floc structure were observed in all reactors on the final sampling date. Pin flocs were observed commonly in samples from Processes II and III. Filament index values dropped from 5 to 1 in Processes I (control), II and III and from 5 to 2 in Process IV. Total SS concentrations were higher than typical discharge limits in Processes II, III and IV (Electronic supplementary material 3, Fig. 4b). Effluent from all reactors became turbid during the test period.
The mean SVI on the final day of the study ranged from 24.1–47.5 mL/g for the plug flow reactors and from 105.0–176.1 mL/g for the SBRs (Electronic supplementary material 3, Fig. 5).
In the plug flow system, Cu removal efficiency during secondary treatment in Process II averaged 60.5 % (Fig. 2c). Final effluent concentrations in Process I (control) and in Processes III and IV reactors ranged from 0.04–0.15 mg/L. In the SBRs, Cu removal efficiency during secondary treatment in Process II averaged 46.8 % (Fig. 2d). Process I (control) and Processes III and IV effluent concentrations ranged from 0.01–0.22 mg/L.
In the plug flow system, Fe removal efficiency during secondary treatment in Processes II and III averaged 86.0 % and 73.9 %, respectively (Fig. 2e). Process IV effluents were generally ≤2 mg Fe/L. In the SBRs, average Fe removal efficiency during secondary treatment in Processes II and III was 81.0 % and 79.3 %, respectively (Fig. 2f). Process IV effluents were generally ≤2 mg Fe/L.
In both systems, Mn removal during secondary treatment in Process II was inconsistent, with effluent concentrations suggesting accumulation of Mn (Fig. 2g–h). However, removal in Process III, which received approximately the same influent concentration, averaged 92.8 % in the plug flow and 94.6 % in the SBRs. A probable reason for the difference in Mn removal is the acidity in Process II influent; as the acidity increases, Mn tends to stay dissolved. Process IV values were generally below detection limits (i.e. <10 μg/L).
Lead removal efficiency during secondary treatment approached 100 % for Processes II, III and IV. All final effluents from both systems had Pb concentrations below detection limits (i.e. <50 μg/L).
In the plug flow system, average Zn removal efficiency during secondary treatment in Processes III and IV was 64.9 % and 58.0 %, respectively (Fig. 2i). However, removal in Process II was very low, averaging <10 %. The same trend was observed in the SBRs, with average removal during secondary treatment in Processes III and IV of 80.1 % and 90.0 %, respectively (Fig. 2j). Average removal in Process II was <10 %.
The same general trend of increasing effluent metal concentrations over time was observed in both systems. There was only one significant difference observed between systems, i.e. for Al removal in Process III, where SBR removal was significantly greater (p < 0.05) than in the plug flow system.
In the plug flow system, SO4 removal did not occur in any reactors. Increasing SO4 accumulation was observed, with the effluent concentrations in Processes II and III exceeding influent concentrations towards the end of the study. In the SBRs, some removal of SO4 did occur on earlier sampling dates in Processes II and III. However, increasing SO4 concentrations over time indicated that removal efficiency in these reactors was decreasing.
In the plug flow system, the final effluents in all reactors were net-alkaline. In general, Process I (control) had the highest concentration of alkalinity in the final effluent, ranging from 90–130 mg/L as CaCO3. The alkalinity of the other three reactors was generally lowest for Process II (i.e. <20 mg/L as CaCO3), moderate for Process III (20–55 mg/L as CaCO3) and greatest for Process IV (25–75 mg/L as CaCO3). In the SBRs, the final effluents in all reactors were net-alkaline. In general, Process I (control) had the highest concentration of alkalinity in the final effluent, ranging from 30–85 mg/L as CaCO3. The alkalinity of the other three reactors was generally lowest for Process II (i.e. <20 mg/L as CaCO3), moderate for Process III (10–40 mg/L as CaCO3/L) and greatest for Process IV (40–60 mg/L as CaCO3).
This study investigated the impact of AMD loading at a 2:1 volumetric ratio with synthetic MWW on treatment performance using laboratory-scale plug flow and SBR systems. In terms of removal of COD, BOD5 or TOC, AMD loading did not cause a significant decrease in performance for any of the simulated co-treatment processes. Whereas both systems received the same proportion (by volume) of AMD in the influent stream, the plug flow reactors were operated at a conventional organic loading rate, SRT and HRT, while the SBRs were operated as a low-rate system with longer SRT and HRT and a lower f/m ratio. Previous workers have investigated the effect of process parameters on heavy metal-loaded reactors and reported that COD removal efficiency was hindered more by metals at short HRTs (Chua et al. 1999) and suggested that increasing the sludge age can improve tolerance to metals (Neufeld and Hermann 1975; Pamukoglu and Kargi 2007). However, the results of the present study did not support those findings. In terms of MWW treatment, the performance of the plug flow and SBR systems was very similar overall, with no significant differences (α = 0.05) between BOD5 removals, TOC removals or TP removals. However, COD and BOD5 removal efficiencies in Process II in the plug flow system were higher than in the SBRs, suggesting that the plug flow system is more robust to highly acidic AMD loads.
Precipitation and adsorption onto wastewater particulates and sludge biomass are the major removal mechanisms for dissolved metals during secondary treatment (Brown and Lester 1979). Metals that are precipitated, adsorbed and associated with fine particulate matter (along with any residual insoluble metals) can be taken up by entrapment in the sludge matrix and removed from final effluents by settling in the secondary clarifiers (Santos and Judd 2010). Removal efficiency by adsorption varies for different metals, depending on factors such as initial metal concentrations, solubility of metal ions, electronegativity, solids concentration and pH (Acheampong et al. 2010).
In the present study, Al and Fe were efficiently removed, with precipitation likely to be the dominant removal mechanism. The pH in the reactors remained at circum-neutral values, creating conditions suitable for (1) removal of Al and Fe as phosphate precipitates (Caravelli et al. 2010; Evangelou 1998); (2) Fe removal via precipitation as schwertmannite, ferrihydrite and goethite and (3) Al removal via co-precipitation with the Fe solids and/or precipitation as hydrobasaluminite or Al(OH)3 (Burgos et al. 2012). The presence of Fe and Al was associated with significantly increased TP removal, indicating that precipitation of these metals with phosphates was occurring. Sorption and/or co-precipitation in the Fe and Al precipitates was also a possible removal mechanism for dissolved Cu, Mn, Pb and Zn (Burgos et al. 2012; Munk et al. 2002; Sánchez España et al. 2006).
Removal of Cu during secondary treatment can occur by hydroxide precipitation, co-precipitation and adsorption and is typically efficient compared with removal of other metals such as Cd, Ni, Mn and Zn (Cheng et al. 1975; Nielsen and Hrudey 1983). Copper has a relatively high affinity for binding sites on organic materials (Gibert et al. 2005; Hammaini et al. 2007; Hammaini et al. 2003; Zhang 2011), and Cu removal rates in batch tests using activated sludge reached approximately 1–2 mg Cu/g MLSS (Hughes and Gray 2013). However, those studies were conducted with comparatively high initial Cu concentrations. In the present study, Cu removal was only moderately efficient, indicating that removal efficiency is decreased at low influent concentrations and that achieving complete Cu removal from final effluents may be a challenge in co-treatment. Elevated Cu concentrations were observed in the control effluents for both systems, along with a slight accumulation in the SBR effluents; the cause of these trends was not identified.
Final effluent Pb concentrations were below detection limits in all three simulated processes. Based on these results, influent Pb at low concentrations is efficiently removed during secondary treatment and is unlikely to present a contamination issue in final effluents. Removal of Pb is typically highly efficient during wastewater treatment, with removal rates higher than those of Cu, Cd, Ni and Mn (Buzier et al. 2006; Cheng et al. 1975; Oliver and Cosgrove 1974). Hydroxide precipitation is a key mechanism for removal of Pb from wastewater (Pagnanelli et al. 2009; Sterritt et al. 1981), along with adsorption onto sludge biomass (Hammaini et al. 2002, 2007; Rozada et al. 2008; Zhang 2011).
In the present study, Mn removal was poor (<10 %) during secondary treatment in the reactors simulating Process II, the process with the highest load of acidity. In the reactors simulating Process III, with similar Mn concentrations to Process II but lower acidity, Mn removal during secondary treatment was >90 %. Zinc removal followed a similar trend, with <10 % removal with net-acidic AMD (Process II), but reaching 58–90 % removal with circum-neutral AMD (Processes III and IV).
Removal efficiency of Mn and Zn in activated sludge is typically poor compared with other metals such as Cu, Cr, Pb and Cd (Chang et al. 2007; Nielsen and Hrudey 1983; Oliver and Cosgrove 1974; Stephenson and Lester 1987). For Mn, this is mainly because uncatalyzed oxidation of soluble Mn (II) to insoluble Mn (IV) does not occur readily below pH 9 (Brezonik 1994; Stumm and Morgan 1981), thus giving Mn a high solubility over a wide pH range. Zinc hydroxide precipitation in activated sludge becomes significant at pH > 8.0 (Katsou et al. 2010) and therefore was not a major removal mechanism in the present study. However, dissolved Mn and Zn can be removed at pH < 7.5 by co-precipitation with and/or adsorption onto Fe- and Al-(oxy)hydroxides, as well as by carbonate precipitation (Azzam et al. 1969; Chang et al. 2007; Gibert et al. 2005; Kempton et al. 1983). Alternatively, Mn and Zn can be removed by adsorption onto activated sludge (Cecen and Gursoy 2001; Hammaini et al. 2007; Katsou et al. 2010; Lei et al. 2008), although this process is limited in the presence of preferentially adsorbed metals (Brown and Lester 1982; Chang et al. 2007; Goldstone et al. 1990b; Hammaini et al. 2003; Sterritt et al. 1981; Zhang 2011). Results of the present study indicate that where acidity is not limiting, Mn and Zn removal was efficient, but acidity limited the removal of Mn and Zn. Tertiary treatment by alkali addition and removal of these metals as carbonates may be required to achieve desired concentrations after co-treatment of highly acidic AMD (Edenborn and Brickett 2002).
Overall, Process IV was the best process option in terms of metal removal, having the lowest final effluent concentrations for most metals. This process simulated a pre-mixing stage with screened MWW to neutralize the AMD and remove the major fraction of metals in the primary sludge, with the remaining dissolved metals entering the secondary treatment stage. Process III was the second most effective process in terms of metal removal. It also incorporated a pre-treatment stage, in this case mixing with digested sewage sludge, to neutralize and remove metals from AMD. Metal removal efficiency in Process II, wherein AMD is added directly to the aeration tank, was similar to Process III for Al, Cu, Fe and Pb but was the least efficient process in terms of removal of Mn and Zn, suggesting that influent acidity significantly decreased removal for those metals. With the exception of Al removal in Process III, where removal in the SBR was significantly greater (p < 0.05) than removal in the plug flow reactor, there were no significant differences (α = 0.05) between metal removals in the different systems. This suggests that, in general, increased HRT in the SBRs did not improve metal removal efficiency. Hughes and Gray (2013) demonstrated that metal removal using MWW and activated sludges is very rapid, with removal complete in <30 min. Experiments using shorter HRTs are recommended to identify the minimum time period required for effective removal of dissolved metals.
Removal of TP was highest in the reactors loaded with AMD which contained Fe and Al. In contrast, TP removal in the control reactors was relatively inefficient, with no net TP removal observed in several cases. The presence of Fe and Al in AMD significantly improved TP removal, most likely by precipitation of phosphates (Caravelli et al. 2010; Clark et al. 1999; Omoike and Vanloon 1999; Yeoman et al. 1992) and/or sorption of phosphates onto Fe-(oxy)hydroxide precipitates (Dobbie et al. 2009; Sibrell et al. 2009; Wei et al. 2008), achieving final effluent TP concentrations in the range 1–2 mg TP/L. Successful phosphate removal by flocculation with Fe and Al and sorption to AMD flocs was reported for a passive AMD treatment system incorporating a primary mixing stage with raw MWW (Strosnider et al. 2011b). This is an important co-treatment synergy, with the potential for AMD to serve as a substitute for proprietary chemicals and coagulants for improving phosphate removal.
For nitrification to occur in conventional systems, the sludge age should be >8 days (Gerardi 2002). Therefore, while nitrification was not expected to occur in the plug flow system, it was expected to occur in the SBRs. Although the SBRs were operated with anaerobic and anoxic periods, and N removal was observed on earlier sampling dates, subsequent increases in NH4 and NO3 effluent concentrations indicated poor nitrification and denitrification. Previous toxicity studies by Hughes and Gray (2012) demonstrated that nitrification could be inhibited by AMD. Metals have previously been reported to inhibit the activity of autotrophic bacteria, e.g. nitrifying bacteria, more than that of heterotrophs (Katsou et al. 2011; Principi et al. 2006), and the nitrification process is generally very susceptible to metals, pH and temperature shifts (Gerardi 2002). However, in the present study, AMD was not considered to be limiting nitrification or denitrification processes, because final effluent TN concentrations were also high in both control reactors, indicating that another factor must be the cause. No clear effect of AMD loading on nitrification and denitrification processes was demonstrated by these results. The alkalinity of final effluents from the plug flow system was consistently higher than effluents from the SBRs, suggesting that, although TN concentrations were increasing because denitrification had ceased, nitrification (an acid-generating process) was still occurring in the SBRs.
Poor floc structure is a major concern in activated sludge, because the floc structure serves several purposes. Well-flocculating sludge settles well during sedimentation, and the floc structure can protect bacteria from toxic substances in the supernatant (Jönsson et al. 2000). In this study, there was a general deterioration in floc morphology in both systems throughout the test period. Changes in the test reactors in both systems were similar overall to changes in the controls, with the exception of higher filament index values and slightly more compact floc structure in the plug flow system control. Therefore, observed changes in floc morphology in test reactors evidently were not direct effects of AMD loading. Microstructure and pin flocs, small and compact flocs with no filamentous structure (Gray 2004), were observed in all reactors in the later stage of the study. Such flocs are undesirable because they do not settle well and can be washed out in final effluents, leading to decreased biomass in the reactors. Microstructure is an indication of deflocculation, a major problem in sludge that causes turbidity and loss of biomass in final effluent. Despite periodic re-inoculations of all reactors with fresh activated sludge, filament abundance also decreased in both systems. While an overabundance of filamentous bacteria has a detrimental effect on settleability, a complete lack of filaments is not desirable either. Operational factors such as ambient temperature or the use of synthetic substrate are possible causes of these undesirable changes in floc morphology. It is significant that throughout the study, diverse populations of ciliates and rotifers were continuously present in all reactors, and the efficient removal of organics indicated normal bacterial activity. Thus, despite the changes in floc structure, reactor conditions supported healthy microbial populations.
The physical characteristics of sludge were often better in the plug flow system. For example, the floc size and structure in SBR sludge was generally larger and more diffuse than sludge in the plug flow reactors. This difference in floc morphology corresponds with the remarkable difference in SVI values between the two systems, and it is concluded that the smaller floc size led to improved settling in the plug flow system (Andreadakis 1993; Jin et al. 2003). In contrast with a general reduction in turbidity in the plug flow reactors, turbidity increased in all SBRs during the study. Often, turbid effluents are caused by deflocculation or the loss of ciliated protozoa (Gray 2004); however, in the SBRs, the flocs were maintaining a large structure (albeit with reduced filamentous structure), indicating that deflocculation was not a major issue, and ciliated protozoa were continuously present in all reactors. The cause for the highly turbid effluents in the SBRs in this study was not identified. Hughes and Gray (2012) also observed increasing effluent turbidity and elevated effluent SS concentrations during extended operation of laboratory-scale SBRs loaded with synthetic MWW and AMD, suggesting that decreased populations of floc-forming bacteria could be a factor. Acid mine drainage was not believed to be the cause of turbid effluents in that study, because the turbidity was also present in control effluents. It was anticipated that Fe-(oxy)hydroxide precipitates would provide attachment sites for SS, or act as a coagulant to improve flocculation and minimize turbidity in final effluents. However, results indicated that higher concentrations of Fe are required for improving settleability.
In the plug flow system, SO4 removal was not occurring at any point throughout the study, and final effluent concentrations indicated SO4 accumulation in the reactors. In the SBRs, final effluent concentrations indicated that SO4 removal was occurring initially, but removal efficiency gradually decreased. Sulfate removal from AMD by its reduction to sulfide is readily achieved in biological treatment systems using SO4-reducing bacteria under anaerobic condition. Metals are also removed under these conditions via precipitation as insoluble sulfides and acidity reduced due to alkalinity generation (Strosnider et al. 2011a). However, SO4 removal from AMD presents a challenge in the conventional activated sludge process. First, the process is primarily aerobic and not conducive to SO4-reducing bacteria activity, although there is recent evidence that SO4-reducing bacteria can survive oxic periods (Kjeldsen et al. 2004). Second, the presence of sulfides can alter sludge floc structure and lead to floc disintegration (Nielsen and Keiding 1998), although acclimatization of activated sludge to sulfide loads has been observed (Burgess and Stuetz 2002). In practice, provided that the activated sludge microbial community suffers no adverse affects from the high SO4 concentrations, there are several ways in which SO4 can be removed from solution in the activated sludge process. At low pH (pH 3.6–5.0), SO4 removal can occur via formation of schwertmannite or Al-hydroxysulfates and sorption to microcrystalline gibbsite Al(OH)3 (Munk et al. 2002). Sulfate ions can form bridging complexes between heavy metals and mine drainage precipitates (Webster et al. 1998). Where calcium and SO4 concentrations reach saturation concentrations, SO4 may be removed via precipitation of gypsum (CaSO4·2H2O). Finally, when the ORP is less than −50 mV (i.e. during the anoxic phase in the SBRs), SO4 reduction can occur (Nielsen et al. 2005), although it is commonly accepted that the optimum for SO4 reduction is anaerobic conditions (ORP −200 to −300 mV) (Boon 1995). However, in this study, reactor conditions did not support SO4 removal.
In this study, AMD acidity was effectively neutralized by the buffered MWW, resulting in net-alkaline final effluents in both the plug flow system and the SBRs. Lower alkalinity in SBR effluents may indicate that limited nitrification (an alkalinity-consuming process) was occurring in the SBRs. However, nitrification is not believed to have been occurring to a significant extent in either system. Highest concentrations of effluent alkalinity were measured in the control, and the values for the remaining reactors increased for Processes II, III and IV, respectively. This trend of decreasing effluent alkalinity with increasing influent (AMD) acidity supports the conclusion that neutralization of net-acidic AMD was consuming the alkalinity present in the reactors. A large fraction of MWW alkalinity may be consumed when mixed with AMD; therefore, an adequate concentration of alkalinity must be maintained during secondary treatment to prevent drops in pH drop below normal operating conditions, i.e. above pH 6.5 (Ekama and Wentzel 2008; Garcia Orozco 2008; Lew et al. 2009). The alkalinity budget of a WWTP is extremely important, and if final effluents are net-acidic or net-alkaline with low concentrations of alkalinity (i.e. <50 mg/L as CaCO3) (Gerardi 2002), the pH stability of the WWTP will be adversely affected, especially in WWTPs which are designed for nitrification. Ideally, during co-treatment, the acidity of AMD should be neutralized without consuming all of the available alkalinity from the influent MWW. Process performance at a lower ratio of AMD to MWW should be explored as a way to avoid the need for alkali additions. Alternatively, highly alkaline materials available from a WWTP (e.g. digester supernatant) can be used in a pre-treatment step to neutralize AMD and remove metals by precipitation.
Co-treatment of AMD with MWW using the activated sludge process is an innovative remediation strategy with several practical advantages. Using a WWTP for AMD treatment can lead to significant reductions in the materials and energy costs which are associated with building and operating AMD treatment systems. Evidence from this study demonstrated that, as well as removing metals and acidity from AMD, co-treatment can improve P removal and sludge settleability during MWW treatment. Thus, co-treatment may eliminate the need to add costly proprietary chemicals to remove P and increase sludge settleability during MWW treatment, leading to further reductions in operating expenses. The results of this study demonstrate the feasibility of co-treatment and contribute new and valuable information for the development of AMD co-treatment systems.
Acid mine drainage, loaded to laboratory-scale activated sludge reactors in a 2:1 volumetric loading ratio with wastewater, did not cause a significant decrease in performance for any of the simulated co-treatment processes in terms of removal of COD, BOD5 or TOC.
In Processes II and III, co-treating AMD which contained approximately 30 mg Fe/L and 20 mg Al/L, TP removal was dramatically enhanced, most likely by precipitation with phosphates, resulting in final effluent TP concentrations <2 mg/L. In contrast, control reactors and reactors not heavily loaded with Fe and Al achieved little or no TP removal.
Significant metal removal occurred during the co-treatment of AMD and wastewater in laboratory-scale plug flow and sequencing batch reactor systems, with similar trends in metal removal efficiency observed in both systems.
Average Al and Fe removals reached 52−84 % and 74−86 %, respectively, with metal removal most likely associated with the removal of P via phosphate precipitation.
Net-acidic AMD was effectively neutralized by the highly buffered synthetic wastewater, and all final effluents were net-alkaline. Addition of alkali material during co-treatment may be required where influent AMD is net-acidic to maintain net-alkaline conditions in activated sludge systems.
T. Hughes gratefully acknowledges the support provided by the Irish Research Council for Science, Engineering, and Technology (IRCSET) Embark Initiative and Science Foundation Ireland (SFI) (Grant Number: 08/RFP/ENM993). In addition, the authors extend sincere appreciation to the personnel at the WWTPs located in Leixlip, Co. Kildare, Swords, Co. Dublin, and Athy, Co. Kildare, for their assistance.