WSRC-MS-2001-00128
Comparison of Constructed Wetland Mesocosms Designed
for
Treatment of Copper-Contaminated Wastewater
J. B. Gladden, W. L. Specht, and E. A. Nelson
Westinghouse Savannah River Company
Aiken, SC 29808
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
This report has been reproduced directly from the best available copy.
Available for sale to the public, in paper, from: U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161, phone: (800) 553-6847, fax: (703) 605-6900, email: orders@ntis.fedworld.gov online ordering: http://www.ntis.gov/support/ordering.htm
Available electronically at http://www.osti.gov/bridge/
Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062, phone: (865 ) 576-8401, fax: (865) 576-5728, email: reports@adonis.osti.gov
Abstract
This study compared the performance of two constructed wetland mesocosms used to model a full-scale wetland system designed for treatment of copper-contaminated wastewater. One mesocosm (designated "site-specific") was built near the construction site of the full-scale wetland using on-site soil, commercially available vegetation [Scirpus californicus (C.A. Meyer) Steud.], and water from the targeted wastestream. A second mesocosm (designated "generic") was constructed at Clemson University using local soil, cultured S. californicus, and local municipal water amended with CuSO4. Performance objectives were to achieve 22 m g/L total copper and no toxicity (Ceriodaphnia dubia Richard, 7-d/static/ renewal) in wetland outflows. Total inflow copper to the site-specific and generic mesocosms ranged from non-detect to 87 m g/L and from 27 to 68 m g/L, respectively. Overall total copper removal was 40% (± 33) for the site-specific mesocosm and 73% (± 14) for the generic mesocosm. In seven of nine monthly toxicity tests, C. dubia reproduction was significantly decreased (a =0.05) in outflow of the site-specific mesocosm. No outflow toxicity was observed for the generic mesocosm. Although performance of the two mesocosms differed, both studies contributed to full-scale design by highlighting critical aspects of wetland function and augmenting operation and maintenance plans, enhancing overall constructed wetland design.
Keywords: Constructed wetland design; Mesocosms; Copper; Wastewater treatment
1. Introduction
To efficiently and effectively design constructed wetlands to reduce risks or adverse effects of pollutants in wastewater, three common sources are typically investigated for information regarding expected transfers and transformations of the targeted wastewater constituent(s). The preliminary information required for constructed wetland design usually consists of rate coefficients for wetland processes and physical-chemical equilibrium events that may be operative in the system to transfer or transform the targeted constituent(s) (U.S. EPA, 1988; Rodgers and Dunn, 1992; Cueto, 1993; Johnston, 1993; Tchobanoglas, 1993). During the conceptual design stage for a constructed wetland, an initial source of information is theoretical models based upon fundamental thermodynamics that regulate the fate of materials in aquatic systems (Allison et al., 1991). Also useful at this stage may be models of nutrient cycling, energy flow, trophic dynamics, or other processes in wetlands. A second source of information is empirical or published data from well designed systems that may provide accurate rate coefficients and anticipated removal of the targeted constituent(s) from the wastewater (Hammer, 1989; Moshiri, 1993; Kent, 1994). In particularly critical situations or in cases where there is limited experience or published information, physical models of constructed wetlands, or mesocosms, can provide the requisite data regarding removal rate coefficients and anticipated treatment performance, providing the confidence required to proceed with the full-scale constructed wetland (Hawkins et al., 1997; Gillespie et al., 1999; 2000).
Physical wetland models (mesocosm or pilot-scale systems) have effectively established design parameters and predicted performance for constructed wetland systems receiving a variety of point and non-point source wastewaters, including petroleum refinery effluents (Hawkins et al., 1997; Gillespie et al., 1999; 2000; Huddleston et al., 2000), agricultural runoff (Moore et al., 2000), and pulp and paper wastestreams (Thut, 1993). In order to decrease uncertainties regarding the performance of a full-scale constructed wetland system at a specific site for a given wastestream and targeted constituents, a mesocosm-scale system may be built at the site. This strategy minimizes the need for site-to-site scaling which may not always be linear (Rodgers, 1994). Options for conducting wetland mesocosm studies for the purpose of wastewater treatment range from on-site construction using "site-specific" materials (e.g., the actual wastewater, site-specific soils and vegetation) to "generic" systems constructed at research facilities using readily available materials that attempt to mimic site-specific conditions (Hawkins et al., 1997; Gillespie et al., 1999; 2000). Some studies may combine aspects of the two extremes, such as shipping wastewater, soils, or plants to a research facility for evaluation in wetland mesocosms (Huddleston et al., 2000). Both site-specific and generic mesocosm studies have advantages and limitations. Decisions regarding where and how these studies should be conducted are unique to the study objectives and the site, and are influenced by a number of logistic factors, such as transportation or shipping costs, available personnel for sampling and data collection, and analytical requirements and capabilities.
This study compared two mesocosm-scale constructed wetland systems for establishing design parameters and predicting treatment performance of a full-scale wetland treatment system receiving copper-contaminated wastewater. One system was constructed on site and treated a portion of the actual wastestream. This site-specific constructed wetland mesocosm system incorporated site soils and utilized vegetation (Scirpus californicus [C.A. Meyer] Steud., giant bulrush) obtained from a commercial vendor. The second constructed wetland mesocosm system was built at Clemson University, approximately 200 km from the proposed construction site. This generic system consisted of locally obtained soils, S. californicus from cultures maintained at the research facility, and municipal water amended with copper to mimic the on-site wastestream.
The wastestream proposed for treatment by the full-scale constructed wetland system consisted of process flow from a technical facility and stormwater runoff from the surrounding watershed. Average daily process flow consisted of approximately 0.04 m3 s-1 (0.97 million gpd), with low flow of 0.01 m3 s-1 (250,000 gpd) and peak flow of 0.11 m3 s-1 (2.6 million gpd). Land uses and covers of the drainage area of approximately 77 ha included an industrial setting (47%), pavement (1%), grass (9%), gravel roads (1%), and forest (42%). Copper was identified as problematic in the wastestream through a toxicity identification evaluation (TIE), with total copper (acid soluble) occasionally exceeding the regulatory discharge limit of 22 m g/L. Copper concentrations in the wastestream were variable, ranging from non-detect to 87 m g/L.
The specific objectives of this study were: 1) to compare performances of the two mesocosm-scale models for developing design parameters for a full-scale constructed wetland system to treat copper in this wastestream, 2) to evaluate the ability of the wetland mesocosm systems to achieve the site-specific copper discharge limit of 22 m g/L and no toxicity in the wetland effluent (Ceriodaphnia dubia Richard, 7-d, static, renewal), and 3) to discern scaling considerations for the full-scale system.
2. Materials and Methods
2.1 Design Theory
The goal of this constructed wetland system is to decrease risks to receiving-water biota by reducing exposure to bioavailable copper in the wastestream. In natural systems, copper bioavailability and associated toxicity are largely dependent on the chemical forms or species of this metal (Deaver and Rodgers, 1996). Copper speciation in aquatic systems is a function of a variety of intrinsic system characteristics, such as pH, hardness, ionic strength, particle size distribution, dissolved and particulate organic matter, and redox potential (Baudo et al., 1990; Breault et al., 1996). Constructed wetland systems can be designed using characteristics that exploit the thermodynamic processes responsible for copper speciation. The fundamental basis of this constructed wetland design is to provide a system in which the thermodynamic reactions that transfer and transform copper to non-bioavailable species were both possible and likely to occur.
Design of the constructed wetland system began with theoretical modeling to predict copper speciation. In most natural aquatic systems, prevalent copper species include cupric oxide, hydroxides, carbonates, and sulfides, as well as copper adsorbed to mineral particles, complexed with dissolved and particulate organic matter, and adsorbed to and absorbed by wetland vegetation (Fig. 1) (Leckie and Davis, 1979; Elder, 1988; Jenne and Crecelius, 1988; Cabaniss, 1992; Breault et al., 1996; Suedel et al., 1996). The geochemical equilibrium speciation model, MINTEQA2, Version 3.0 (Allison et al., 1991) was utilized to predict initial speciation of copper in the constructed wetland system. Input values for the model were based on present wastestream characteristics (Table 1) and anticipated characteristics of the constructed wetland system. These values included Cu+2 (0.05 mg/L), alkalinity (20 mg/L), sulfate (10 mg/L), Ca+2 (8 mg/L), and CO3-2 (12 mg/L). Also specified were the redox couple HS-1/SO4-2 and redox potential of 150 mV.
2.2 Published Literature
A strategic review of published literature was performed to assist in selection of wetland hydroperiod, hydrosoil, and vegetation (macrofeatures) for this constructed wetland system. Each macrofeature and related literature are described below.
2.2.1 Hydroperiod
Determination of hydroperiod used in this constructed wetland design was based on a hydraulic retention time (HRT) adequate for transfers and transformations of copper to non-bioavailable forms, and a water depth providing a sufficient oxygen barrier for maintaining reduced hydrosoil conditions (for sulfate reduction). Published literature indicated that, given sufficient ligand concentrations, precipitation of copper with sulfur and binding of copper to dissolved and particulate organic matter is relatively rapid (instantaneous to <24 h) (Morel and Hering, 1993; Ma et al., 1999; Kim et al., 1999). Other studies reported that 24 to 48-h HRT and 30-cm water depth were sufficient for removal of divalent metals from wastewater using constructed wetlands (Sinicrope et al., 1992; Hawkins et al., 1997; Gillespie et al., 1999; 2000).
2.2.2 Hydrosoil
Previous studies were useful in establishing the criteria for selecting wetland hydrosoil (Sinicrope et al., 1992; Suedel et al., 1996; Hawkins et al., 1997; Gillespie et al., 1999; 2000). These criteria included: 1) availability; 2) soil compatibility with wetland vegetation; 3) nutrient status, including sufficient sulfur for promoting copper sulfide precipitation; and 4) organic matter content for maintaining reduced hydrosoil conditions (-100 to -250 mV). Specific amendments for achieving these criteria are provided for each wetland mesocosm below (Section 2.3).
2.2.3 Vegetation
Vegetation chosen for this constructed wetland design was based on published and experimental data. Giant bulrush, Scirpus californicus (C.A. Meyer) Steud., was proposed for this constructed wetland based on: 1) the plants ability to maintain reduced hydrosoil (Josselyn et al., 1990; Sinicrope et al., 1992; Hawkins et al., 1997; Gillespie et al., 1999; 2000); 2) its ability to provide an organic carbon source for sulfate reducing bacteria (Sinicrope et al., 1992; Araujo de Oliveira et al., 1994; Richardson et al., 1995); and 3) sufficiently slow degradation of detritus for increasing organic binding sites over time (Chendorain et al., 1998).
2.3 Construction of Wetland Mesocosms
To provide scaling information for full-scale design, a site-specific wetland mesocosm and a generic wetland mesocosm were constructed and evaluated for a period of eight months. The site-specific mesocosm was constructed adjacent to the wastestream to be treated by the full-scale wetland system. This system consisted of four 2.4 x 1.2 x 0.9 m (length x width x depth) butyl-lined plywood boxes, each divided along the center forming two identical 2.4 x 0.6 x 0.9 m cells per box, providing four sets of two cells in series (Fig. 2). The system was plumbed using PVC and vinyl tubing. Soil collected from the proposed construction area provided wetland hydrosoil, which was placed in the mesocosm cells at a depth of 46 cm. Initial hydrosoil amendments included organic material (wood mulch) at 3% by volume, agricultural lime at 71 g m-2, and OsmocoteÒ time-release fertilizer at 60 g m-2. After three months of operation, pelletized gypsum (CaSO4· 2H2O) and grass clippings were added to achieve 0.5% sulfur per top 10 cm of hydrosoil and 5% organic matter (by hydrosoil volume), respectively. These amendments were intended to promote dissimilatory sulfate reduction within the wetland system. Scirpus californicus plants were purchased as rootstock from a commercial vendor and planted on 15-cm centers. Because of high initial mortality, this system was replanted to the specified density in June 1999. Water was pumped directly from the wastestream to the mesocosm system using piston pumps (Fluid Metering, Inc., Oyster Bay, NY) calibrated to provide a 48-h HRT and overlying water depth of 30 cm.
The generic mesocosm study was conducted at Clemson University, approximately 200 km from the proposed wetland construction site. Ten wetland cells (0.69 x 0.64 x 0.61 m, length x width x depth) were constructed using 378.5-L RubbermaidÒ utility tanks, providing five sets of two cells in series (Fig. 3). One set was an untreated control. Plumbing consisted of PVC and vinyl tubing. A locally obtained hydrosoil (creek sediment) with similar characteristics to the site-specific soil (85% sand, 15% silt and clay, and 2% organic matter) was used. Initial amendments to this hydrosoil included agricultural lime at a rate of 49 g m-2 and OsmocoteÒ fertilizer at 144 g m-2. After two months of operation, pelletized gypsum (CaSO4· 2H2O) and a mixture of pond detritus and grass clippings were added to the wetland mesocosm system to achieve 0.5% sulfur per top 10 cm of hydrosoil and 5% organic matter (by hydrosoil volume), respectively. Each wetland cell contained 30 cm of hydrosoil and S. californicus shoots planted on 15-cm centers. Plants were obtained from stock cultures maintained at the research facility. Wastewater was simulated by amending municipal water with reagent grade copper sulfate (CuSO4· 5H2O, SigmaÒ Chemicals) to a nominal concentration of 50 m g Cu l-1. Water amended with copper was delivered to the system using piston pumps (Fluid Metering, Inc., Oyster Bay, NY) calibrated to provide a 48-h HRT. To simulate an upstream retention basin for the full-scale constructed wetland system, wastewater was retained in a 6867-L reservoir, which was renewed weekly and amended with copper.
2.4 Sampling and Analyses
Inflow of the site-specific mesocosm consisted of the wastestream targeted for treatment by the proposed full-scale constructed wetland system. Inflow samples were collected by an automated composite sampler (24-h) used to monitor the wastestream. Initially, outflow samples were collected by grab technique. In September, all sampling of the specific mesocosm was conducted using automated 24-h composites with a 48-h lag between inflow and outflow collections.
Inflow of the generic mesocosm system consisted of a continuously mixed reservoir containing municipal water amended to a nominal concentration of 50 m g/L total copper. The reservoir and outlets of each series of wetland cells were sampled by grab technique. Outflow samples were collected 48 h following inflow sampling. Samples from individual wetland outlets were combined for analysis.
Methods for analysis of water samples and hydrosoil are referenced in Table 2. Toxicity of inflow and outflow water was evaluated in 7-d, static, renewal exposures of Ceriodaphnia dubia Richard (U.S. EPA, 1994). In the site-specific study, toxicity experiments were performed on outflow water from each series of wetland cells. Outflow water samples were combined for toxicity evaluation in the generic mesocosm study.
3. Results
3.1 Design Theory
MINTEQA2 predicted precipitation of copper as the mineral covellite (CuS), with the equilibrated mass distribution of copper within the system as >99% precipitated. The remaining fraction (<1%) was dissolved or adsorbed , which included Cu+2 (40% of the unprecipitated fraction), Cu(OH)2 AQ (21%), CuSO4 AQ (18%), CuCO3 AQ (15%), CuHCO3+ (5%), and CuOH+ (2%). Other sources of thermodynamic information provided additional predictions of copper speciation. Brookins (1988) developed Eh-pH relationships based on thermodynamic data describing interactions between metals and potential inorganic ligands (oxides, hydroxides, carbonates, and sulfides) in natural aquatic systems. According to the Cu-C-S-O-H model (Fig. 4), CuS and Cu2S would be the dominant copper species in the wetland hydrosoil given sufficiently negative redox (< -100) and circumneutral pH.
3.2 Published Literature
Wetland macrofeatures (hydroperiod, hydrosoil, and vegetation) for this constructed wetland system were selected based on information provided by published literature and experimental data. Each macrofeature was carefully considered for its integral contribution to decreasing copper bioavailability within the system.
3.2.1 Hydroperiod
The hydroperiod selected for this constructed wetland design included a conservative 48-h HRT and a 30-cm water depth. To achieve this hydroperiod given an average daily discharge of 0.04 m3 s-1, the calculated total area of the constructed wetland system was approximately 3.2 ha. Eight wetland cells (0.4 ha each) were proposed for this constructed wetland system (four pairs of cells, each pair providing a 48-h HRT for one quarter of the wastestream flow). A 30-cm water depth was selected for assisting in maintaining anaerobic hydrosoil conditions sufficient for sequestering copper (Hawkins et al., 1997; Gillespie et al., 1999; 2000).
3.2.2 Hydrosoil
Hydosoils used in the constructed wetland mesocosms were amended with lime for maintaining circumneutral pH, organic matter for supporting anaerobic microbial respiration, sulfur for promoting dissimilatory sulfate reduction and copper sulfide precipitation, and a time-release fertilizer for plant growth (Table 3). Greenhouse experiments confirmed the compatibility of selected hydrosoils with wetland vegetation.
3.2.3 Vegetation
Published information, experimental data, and results from the mesocosm studies (Section 3.3) confirmed the effectiveness of S. californicus for use in this constructed wetland design. S. californicus allowed the targeted hydrosoil redox range of 100 to 250 mV to be achieved, and other characteristics described in the scientific literature (e.g., organic carbon contribution, slow degradation rate) supported its selection for use in the full-scale constructed wetland system.
3.3 Constructed Wetland Mesocosms
The site-specific mesocosm system received water directly from the wastestream, resulting in considerable variation of influent copper concentrations (see performance data below). Inflow copper concentrations in the generic mesocosm system were less variable due to the simulated upstream retention basin amended to nominal 50 m g/L total copper. Both mesocosm systems were operated with a 48-h HRT for the duration of the study.
3.4 Performance of Constructed Wetland Mesocosms
Total copper in the inflows of the constructed wetland mesocosms averaged 28 (± 19) m g/L (range: non-detect to 87 m g/L) in the site-specific study (Fig. 5) and 46 (± 9) m g/L (range: 27 to 68 m g/L) in the generic study (Fig. 6). Outflow concentrations averaged 16 (± 11) m g/L (range: non-detect to 39 m g/L) and 12 (± 7) m g/L (range: 3 to 29 m g/L), respectively. Both systems removed total copper from aqueous phase, although to varying degrees. On average, total copper removal by the site-specific mesocosm was 40% (± 33), while the generic mesocosm removed 73% (± 14) of influent total copper.
The rate coefficient for copper transfers and transformations was calculated by the first-order or pseudo-first-order model:
Ct = Ci e-Kt
where Ct and Ci are total copper concentration at time t and initial concentration, respectively. For this model, K is the overall copper transfer and transformation (i.e., removal) rate coefficient with units of t-1. In the case of this constructed wetland system, a useful expression of K is the transfer and transformation half-life (T1/2), or the time required to decrease the inflow copper concentration by 50%, given by the equation:
T1/2 = ln2/K = 0.693/K
Using the average inflow and outflow copper concentrations and 48-h HRT for the generic mesocosm, the resulting K for this system was 0.028 h-1, with a transfer and transformation half-life of 24.8 h.
Survival and reproduction of C. dubia in 7-d exposures to inflow and outflow waters were also used as a measure of treatment effectiveness. Mortality in inflow water ranged from 30 to 100% in the site-specific mesocosm study and from 40 to 100% in the generic mesocosm study (Table 4). Outflow mortality ranged from 0 to 25% in the site-specific study and 0 to 20% in the generic study. Of the nine monthly toxicity tests conducted for the site-specific system, only two (September and November 1999) indicated no toxicity in each of the four outflows based on C. dubia reproduction. In the remaining tests, reproduction was significantly decreased (a =0.05) compared to control organisms in at least one outflow. In the generic mesocosm study, outflows were composited for toxicity evaluation. With the exception of the initial test (June 1999), no toxicity was observed in outflow water from the generic mesocosm system.
Redox potential and acid volatile sulfide (AVS) concentrations of wetland hydrosoils indicated that both systems supported dissimilatory sulfate reduction, which was necessary in these systems for continued precipitation of copper sulfide minerals. The site-specific mesocosm achieved redox in the targeted range (-100 to -250 mV) within one week of initiating flow (Fig. 7). The generic mesocosm was slower to respond, reaching the targeted redox in about six weeks (Fig. 8). In the site-specific mesocosm system, hydrosoil redox increased above the range for sulfate reduction during the winter months, averaging approximately -50mV during this time. However, redox decreased to less than -100 mV by April (Fig. 7). In the generic mesocosm study, hydrosoil redox remained within the range of sulfate reduction from August through the remainder of the study [8 months](Fig. 8).
Additional hydrosoil conditions indicative of wetland function for removing copper from aqueous phase and limiting copper bioavailability included AVS, pH, and organic matter content. In the site-specific system, average AVS to simultaneously extractable copper (SECu) ratio was 4.13 (± 3.61) m mole g-1 : 0.35 (± 0.92) m mole g-1. Average AVS:SECu in the generic mesocosm study was 3.10 (± 0.11) m mole g-1 : 0.04 (± 0.1) m mole g-1. Hydrosoil pH and organic matter were comparable between the two systems following amendments (Table 3). Cation exchange capacity (CEC) for both site-specific and generic hydrosoils was 1.4 meq/100g, indicating limited availability of mineral surfaces to bind copper.
From the time of planting (May-June) through the growing season (November), S. californicus shoots in the site-specific mesocosm increased in height from approximately 0.5 m to 2 m. Shoot length in the generic mesocosm increased from approximately 1.5 m to 2.7 m. Colonization of S. californicus as measured by increase in shoot density was slowed in the site-specific wetland mesocosm due to insect herbivory (Delphacidae, plant hopper) and shading from surrounding trees. Shoot density increased from approximately 30 to 60 shoots m-2 in the site-specific mesocosm and from 30 to 300 shoots/m2 in the generic mesocosm during the growing season.
Anaerobic hydrosoil conditions were verified by calculating the net oxygen supply rate (NOSR) within the generic mesocosm system (Kadlec and Knight, 1996):
NOSR = K(DOsat DOavg ) SOD
where K = mass oxygen transfer rate, 0.1 m/d
DOsat = equilibrium dissolved oxygen concentration at 1 atm, g/m3 (mg/L)
DOavg = average dissolved oxygen concentration in overlying water, g/m3 (mg/L)
SOD = sediment oxygen demand, g O2/m2/d
The experimentally determined SOD for generic mesocosm system was 0.446 g O2/m2/d. Using the equation above, the NOSR for this system was 0.372 g O2/m2/d, or a net oxygen consumption as indicated by the negative sign.
4. Discussion
Design of this constructed wetland system for treatment of copper-contaminated wastewater included fundamental principles of thermodynamics and kinetics that govern the transfers and transformations of materials in aquatic systems and consequently exposures leading to risks for biota. Specifically, this design was based on integration of the carbon and sulfur cycles in this constructed wetland system. Organic matter in wetlands serves as a carbon and energy source for microbial metabolism. As diatomic oxygen is consumed for respiration, hydrosoil becomes reduced (anaerobic). Within the redox range of -100 to -250 mV and circumneutral pH of the hydrosoil, sulfate is utilized as a terminal electron acceptor by a heterogeneous assemblage of sulfate reducing bacteria (Brock and Madigan, 1988; Mitsch and Gosselink, 1993). Dissimilatory sulfate reduction produces H2S, which readily reacts with available divalent metals which adsorb or co-precipitate with commonly formed iron sulfide minerals (Wetzel, 1983; Morse, 1995). Thus, essential characteristics of this constructed wetland system included: 1) a source of organic carbon for microbial respiration; 2) anaerobic and circumneutral hydrosoil; and 3) available sulfate as a terminal electron acceptor. In natural wetlands, these conditions can establish over time (Mitsch and Gosselink, 1993). However, in the case of this constructed wetland system, achieving performance in a short time was critical, so the hydrosoil was amended to hasten establishment of the necessary hydrosoil conditions. Also essential for long-term function of this system was a recurring source of organic carbon. By utilizing S. californicus, which has a detrital decomposition half-life of greater than six months (Huddleston, unpublished data), organic matter will accrete in this wetland over time. The organic matter will serve not only to bind copper directly, but will also provide a carbon and energy source for sulfate reducing bacteria (Wetzel, 1983; Mitsch and Gosselink, 1993). Additionally, accretion of organic matter will increase the life of this constructed wetland system by adding hydrosoil mass over time, providing new binding sites for copper.
Biogeochemical interactions of carbon, sulfur, and copper in aquatic systems were used to model this constructed wetland treatment system both theoretically and physically, with the information gained from each exercise used to enhance design of the full-scale system. Theoretical modeling using the equilibrium speciation model MINTEQA2 (Allison et al., 1991) provided a first-order estimation of potential products of copper transfers and transformations within the constructed wetland. Models such as MINTEQA2 are useful for predicting equilibrium processes in aquatic systems when the operative water characteristics are known or accurately estimated. However, wetlands may require additional input components, such as a specified redox potential or redox couples for predicting speciation. In this study, MINTEQA2 predicted >99% precipitation of copper from the aqueous phase, given the actual hydrosoil Eh-pH conditions and availability of sulfur. Covellite (CuS), the primary precipitate expected to form in this system, is stable (log K = 36.1; Morel and Hering, 1993), suggesting that copper bioavailability in this system would be limited. Numerous studies have demonstrated decreased copper bioavailability resulting from speciation to more stable, less soluble forms due to sediment-water interactions (Deaver, 1995; Deaver and Rodgers, 1996; Suedel et al., 1996; Huggett et al., 1999). Formation of stable copper sulfide precipitates and co-precipitation with other metal sulfides in anoxic sediments can immobilize copper and decrease its bioavailability (Morse, 1995; Carbonell et al., 1999a; 1999b). Thus, MINTEQA2 supported the constructed wetland design theory, indicating that thermodynamic processes responsible for precipitation of copper sulfide were possible and likely to occur in this system, if the input or required parameters or conditions are maintained.
However, theoretical models alone do not provide information about specific wetland characteristics or macrofeatures (hydroperiod, hydrosoil, or vegetation) that must be integrated to achieve conditions necessary for transferring and transforming the targeted contaminant(s). Published scientific literature provided information for selecting wetland macrofeatures to be included in a conceptual design and for evaluation in physical models. Review of previous studies was useful for estimating removal of divalent metals under various hydroperiods (Sinicrope et al., 1992; Hawkins et al., 1997; Gillespie et al., 1999; 2000), hydrosoil characteristics (Suedel et al., 1996; Hawkins et al., 1997; Gillespie et al., 1999; 2000), as well as the production dynamics of S. californicus in constructed wetlands. Given our present understanding of constructed wetlands, the conceptual design alone is insufficient to predict treatment effectiveness of the full-scale system. Therefore, physical scale models, or wetland mesocosms, can provide data and increase confidence that the system will perform as designed.
Successful performance of constructed wetlands for treating wastewater is crucial for compliance with regulatory requirements, which include specific discharge limits. Mesocosm studies based on the conceptual design for the full-scale system can provide important information regarding potential performance as well as critical functional aspects of the system. For example, evaluation of the two mesocosm systems in this study illustrated the importance of an upstream reservoir to thoroughly homogenize the wastewater prior to its entrance into the wetland cells. Water treated by the site-specific mesocosm was pumped directly from the wastestream, with no previous mixing. As a result, the inflow copper concentrations fluctuated from non-detect to 87 m g/L, with concomitant variability in treatment effectiveness, both in terms of copper removal and toxicity. Variability in both inflow and outflow of the site-specific mesocosm system decreased our ability to predict performance and to calculate meaningful rate coefficients for transfers and transformations of copper. Incorporating an upstream retention basin stabilizes influent copper concentrations, as represented by the reservoir utilized in the generic mesocosm design. One advantage of a stabilized inflow is that the necessity of time-dependent (actually flow-dependent based on HRT or careful dye studies) outflow sampling is precluded. If inflow concentrations vary widely, subsequent outflow sampling must take place after a period of time equal to the hydraulic retention time of the system in order to observe treatment performance based on a specified inflow concentration. If inflow concentrations are stabilized by an upstream retention basin, outflow concentrations should also stabilize, which would allow detection of upsets in the system or situations that may compromise treatment efficiency. The calculated half-life for copper in water (based on mean concentrations) was 57.8 h for the site-specific mesocosm and 24.8 h for the generic mesocosm. The longer half-life for copper observed in the site-specific mesocosm was likely due to greater variability in inflow copper concentrations and treatment efficiency.
Another difference in the two mesocosms which could have affected performance was growth of wetland vegetation. The plants utilized in the site-specific mesocosm were obtained as rootstock from a commercial vendor and were likely stressed during transportation. Insect herbivory was observed in this system within a few weeks of planting, and shading from surrounding trees likely slowed plant growth and colonization. In contrast, plants in the generic mesocosm study had been maintained in culture at the research facility for more than a year. These plants were not subjected to shipment over a long distance, and stresses associated with harvesting and transplanting were minimized. Because full-scale constructed wetland systems often require a quantity of plants that are only available through commercial vendors, care must be taken to minimize plant stress during shipment, storage, and transplanting.
Other information gleaned from the two mesocosm systems related to poising the wetlands to promote precipitation of copper as sulfide minerals. In this study, time was critical, and initial results from both systems indicated that hydrosoils were not adequately poised for promoting precipitation of copper. Therefore, the hydrosoil was amended (organic matter and gypsum) to encourage reduced conditions with sufficient sulfur for concomitant sulfate reduction. Both systems responded favorably to the amendments with a decrease in redox potential and an increase in copper removal. However, these conditions were not sustained in the site-specific system. During the winter months (December-February), redox increased periodically out of the range of sulfate reduction, which was also reflected by decreased copper removal and increased toxicity in wetland outflows. Based on these observations, hydrosoil characteristics were modified for the full-scale constructed wetland system. To ensure the full-scale system will be poised for dissimilatory sulfate reduction, amendments of organic matter, sulfur, and lime (CaCO3) to hydrosoils will be included, based on experience with the mesocosm studies.
If wetland macrofeatures had been sufficiently similar in the two mesocosm systems, comparable performances would have been expected. However, differences in wastewater character and loading, hydrosoil, and vegetation affected performance, as would be expected. These results and observations provide insight regarding whether mesocosm studies should be conducted at the site of full-scale construction or at a research facility. An expected advantage of a site-specific mesocosm study using on-site materials is that realistic conditions will be represented. By contrast, a generic mesocosm study conducted at a remote or removed research facility may offer experimental flexibility, including observation with greater resolution. A number of criteria could be considered in arriving at a decision to conduct site-specific or generic mesocosm studies. For example, wastewater containing hazardous substances would likely preclude shipment off site for evaluation in mesocosm studies. Also, simulation of some wastewaters in generic mesocosm systems may not be possible. Site-specific studies may not be possible if an untreated control mesocosm is required. In addition, it may be desirable to conduct some studies in greenhouses, independent of climate, which may require they be conducted at a research facility. Regardless, the physical model of a wetland treatment system should mimic as closely as possible the conceptual design of the full-scale system, which permits more linear extrapolations when scaling from the mesocosm to the full-scale system.
Performance of the two constructed wetland mesocosm systems illustrated that the discharge limit of 22m g/L total copper could be achieved at this site using this constructed wetland system design. Several other alternatives for treatment of this wastestream were considered, including corrosion inhibitors and pH adjustment, rerouting the wastestream to a larger receiving system, ion exchange, blending with other wastestreams at the site, and movement of the compliance point. Based on performance of the mesocosm systems and feasibility evaluations of the treatment alternatives (i.e., technical considerations, economic factors, regulatory issues, and implementation time), the constructed wetland system was determined the most viable alternative for mitigation of copper and associated risks from this wastestream. Scaling this system to full-scale and incorporating the modifications to wetland macrofeatures identified in this study will provide an effective, long-term solution for treatment of this wastestream.
Acknowledgments
This study was partially funded by the United States Department of Energy under Contract No. DE-AC09-96SR18500. Mr. T. Harris, Mr. M. Boerste, and Mr. W. Payne of Westinghouse Savannah River Company were instrumental in facilitating portions of the site-specific study.
References
Table 1. Chemical characteristics (ranges) of waters treated
by a site-specific
and a generic constructed wetland mesocosm system.
Parameter |
Site-specific Mesocosm |
Generic Mesocosm |
pH (s.u.) |
5.3 - 7.8 |
6.0 7.8 |
Hardness (mg/L as CaCO3) |
4 - 28 |
12 16 |
Alkalinity (mg/L as CaCO3) |
21 - 59 |
18 24 |
Conductivity (m S/cm) |
80 - 164 |
83 94 |
Dissolved Oxygen (mg/L) |
4.0 - 11.8 |
7.8 9.4 |
Table 2. Chemical, physical, and toxicological
methods used to evaluate performance of a
site-specific and a generic constructed wetland mesocosm system.
Analysis |
Site-specific |
Generic |
Total copper in |
U.S. EPA Method 220.1 (1983) |
APHA (1995) |
Hydrosoil redox potential |
In situ platinum electrodes
(three per wetland cell, 2-6 cm depth) |
In situ platinum electrodes(four per wetland cell, 2-6 cm depth) (Faulkner et al., 1989) |
Hydrosoil particle |
U.S. EPA Method 9081 (1986) |
Hydrometer method |
Acid volatile sulfides |
Purge and trap method |
Purge and trap method |
Hydrosoil organic |
Loss on ignition |
Loss on ignition |
Hydrosoil cation exchange capacity |
ASTM (1998) |
Plumb (1981) |
pH |
Portable HACH OneÒ pH meter |
OrionÒ Model 250A pH meter with TriodeÒ electrode |
Hardness |
APHA (1995) |
APHA (1995) |
Alkalinity |
APHA (1995) |
APHA (1995) |
Conductivity |
VWRÒ Model 604 Conductivity Meter |
OrionÒ Model 142 Conductivity Meter |
Dissolved oxygen |
YSIÒ Model 58 Dissolved Oxygen Meter |
YSIÒ Model 52 Dissolved Oxygen Meter |
Toxicity |
Ceriodaphnia dubia, 7-d, static, renewal (U.S. EPA, 1994) |
Ceriodaphnia dubia, 7-d, static, renewal (U.S. EPA, 1994) |
Table 3. Initial and amended hydrosoil characteristics in
a site-specific
and a generic constructed wetland mesocosm system.
Parameter |
Site-specific |
Generic |
||
Initial |
Amended |
Initial |
Amended |
|
pH (range) |
5.8
|
5.8 6.5 |
5.7
|
6.1 6.4
|
Hydrosoil organic |
2 % |
|
|
|
Particle size distribution: Sand |
|
- |
|
- |
Hydrosoil cation |
1.4 meq 100g-1 |
- |
1.4 meq 100g-1 |
- |
Table 4. Ceriodaphnia dubia toxicity (7-d, static, renewal
exposures) in site-specific and
generic mesocosm studies. Reproduction refers to the average number of
young produced per surviving adult. Toxicity refers to a significant
difference (p=0.05) in reproduction relative to control animals.
Site-specific Mesocosm |
Generic Mesocosm |
||||
Date |
Sample |
Toxicity (P/F)* |
Sample |
Toxicity (P/F) |
|
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
P |
Composite Outflow |
F |
||
Jun-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
No Test Conducted |
|||
Outflow 1 |
F |
||||
Jul-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
No Test Conducted |
|||
Outflow 1 |
F |
||||
Aug-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
P |
Composite Outflow |
P |
||
Sep-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
F |
Composite Outflow |
P |
||
Oct-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
P |
Composite Outflow |
P |
||
Nov-99 |
Outflow 2 |
P |
|||
Outflow 3 |
P |
||||
Outflow 4 |
P |
||||
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
F |
Composite Outflow |
P |
||
Dec-99 |
Outflow 2 |
F |
|||
Outflow 3 |
F |
||||
Outflow 4 |
|||||
Inflow |
F |
Inflow |
F |
||
Outflow 1 |
F |
Composite Outflow |
P |
||
Jan-00 |
Outflow 2 |
F |
|||
Outflow 3 |
|||||
Outflow 4 |
|||||
Inflow |
F |
Inflow |
F |
||
Feb-00 |
Outflow 1 |
F |
Composite Outflow |
P |
|
Outflow 2 |
F |
||||
Outflow 3 |
F |
||||
Outflow 4 |
P |
* P = Pass; F = Fail.
Fig. 1. Schematic of potential transfers and transformations of free divalent copper ion in constructed wetlands.
Fig. 2. Schematic of the site-specific constructed wetland mesocosm.
Fig. 3. Schematic of the generic constructed wetland mesocosm.
Fig. 4. Eh-pH diagram for part of the system Cu-C-S-O-H, modified from Brookins (1988).
Fig. 5. Total copper in site-specific wetland mesocosm inflow and outflow.
Fig. 6. Total recoverable copper in generic
wetland mesocosm inflow and outflow.
Fig. 7. Hydrosoil redox potential in the site-specific wetland mesocosm (n = 4 probes). Error bars represent standard deviation.
Fig. 8. Hydrosoil redox potential in the generic wetland mesocosm (n = 4 probes). Error bars represent standard deviation.