Ohio River Total Maximum Daily Load for PCBs
Ohio River Miles 40.0 to 317.1 Public Comment Draft Report Ohio River Valley Water Sanitation Commission 5735 Kellogg Avenue Cincinnati, OH 45228 (513) 231-7719 July 29, 2002 1.0 INTRODUCTION
1.1 TMDL BACKGROUND Section 303(d) of the Clean Water Act requires states to develop Total Maximum Daily Loads (TMDLs) for waters not meeting designated uses after technology-based controls have been implemented. A TMDL establishes the allowable loadings of pollutants for a water body, quantifies the reductions necessary to meet all designated uses, and assigns load allocations. The eight minimum regulatory requirements for TMDLs are as follows: 1) TMDLs must be designed to meet applicable water quality standards. 2) TMDLs must include load allocations (LA) and wasteload allocations (WLA). A load allocation is an allowable pollutant load from non-point sources. A WLA is an allowable pollutant load from point sources. The combined LA and WLA must not result in violations of the applicable water quality standards. 3) TMDLs must consider the impacts of background (natural) pollutant contributions. 4) TMDLs must consider critical environmental conditions. 5) TMDLs must consider seasonal environmental variations. 6) TMDLs must include a margin of safety. 7) TMDLs must include public participation. 8) TMDLs must include reasonable assurance that the reduction goals set forth in the TMDL can be achieved and the applicable water quality standard can be met. In 1997, the United States Environmental Protection Agency (USEPA), Region 3, entered into a Federal Consent Order to complete a TMDL for polychlorinated biphenyls (PCBs) for a section of the Ohio River listed on West Virginia's 303(d) list. The entire length of the Ohio River is listed as impaired due to a long-standing fish consumption advisory resulting from elevated PCB levels in fish. This TMDL establishes the allowable loadings of PCBs for the Ohio River within the study area, and quantifies the reductions necessary to meet the applicable water quality standards. The Ohio River Valley Water Sanitation Commission (ORSANCO) developed this TMDL report on behalf of USEPA, Region 3. ORSANCO is an interstate water pollution control agency for the Ohio River Basin. 1.2 POLYCHLORINATED BIPHENYLS PCBs are manmade compounds that have been used commercially since 1929. These chemicals were manufactured as combinations of chlorinated biphenyls that differed according to the percentage of chlorine in the mixture. PCBs had a wide variety of industrial applications due to their chemical stability and flame resistance, however, these characteristics also enabled them to remain highly persistent in the environment. PCBs were commonly used as plasticizers, heat-transfer fluids, solvent extenders, hydraulic fluids, flame retardants, sealers, ink carriers, organic diluents and dielectric fluids. 1 Approximately 99 percent of commercial PCBs produced for U.S. industry were manufactured by Monsanto Chemical Company in Sauget, Illinois and sold under the trade name Aroclor (USDHHS, 1995). The Aroclors are identified by a four-digit numbering code in which the first two digits denote the number of carbon atoms in the biphenyl group and the last two digits represent the approximate percentage of chlorine in the mixture. The most common PCBs manufactured include Aroclor 1242, Aroclor 1248, Aroclor 1254 and Aroclor 1260 (Cairns et al., 1986). PCBs are not naturally occurring compounds so their presence in the environment is a result of anthropogenic activities. Approximately 1.25 billion pounds of PCBs were purchased by U.S. industry by the time production stopped in 1977 (USEPA, 1993). The USEPA estimates that 60 percent, or 750 million pounds, of PCBs produced are still in use in the United States in some 150,000 PCB transformers and 2.5 million mineral oil transformers (Graham, 1987). Another 36 percent (450 million pounds) of PCBs were either placed in landfills or dumps or were available to biota via air, water, soil and sediments. The remaining four percent (55 million pounds) were destroyed by incineration or degraded to the environment (USEPA, 1993). Although uses of PCBs now are limited to closed-system applications such as sealed capacitors and transformers, most contamination reflects a period when PCBs where used in open systems and losses to the environment were likely. Today, PCBs can be released into the environment from poorly maintained hazardous waste sites that contain PCBs, illegal or improper dumping of PCB wastes, and leaks or releases from electrical transformers containing PCBs. In addition, when PCBs are incinerated small amounts are released into the atmosphere as a result of incomplete combustion (USDHHS, 1995). Main Pathways to Environment · Municipal waste incinerators burning organic wastes. · Industrial incinerators burning organic wastes. · Poorly maintained hazardous waste sites that contain PCBs. · Illegal or improper dumping of PCB wastes such as transformer fluids and leaks or releases from electrical transformers containing PCBs. · Disposal of PCB-containing consumer products into municipal or other landfills not designed to handle hazardous waste. The behavior of PCBs differs depending on the number of chlorine atoms present. In general, these compounds are liquids characterized as stable, relatively insoluble and having the ability to sorb strongly to organic matter. As the chlorine content increases, the solubility of the compound decreases and the mixture becomes more viscous. In addition, PCBs are highly lipophilic and bioaccumulation in fish tissue can result in concentrations that are considered unsafe for human consumption (USEPA, 1980). PCBs exist in the atmosphere as vapors or adsorbed to airborne particulates. The gaseous form predominates, typically comprising over 90 percent of the total PCB concentration in air (Atlas et al., 1986). Once in the atmosphere, PCBs can be carried for long distances until they return to earth by wet or dry deposition (USDHHS, 1995). 2 The ultimate fate of PCBs in the environment is terrestrial or aquatic sediments. Once released into the environment, PCBs bind strongly to sediments where they remain in place or become transported by erosion. The adsorption of PCBs in soil is directly related to the degree of chlorination and the composition of the soil. Generally, adsorption increases as the chlorination of the compound and/or the organic carbon and clay content of the soil increase (USDHHS, 1995). In addition, experiments have shown that PCBs sorbed by soils remain relatively immobile against leaching with water or sanitary landfill leachate (Sawhney, 1986). However, in the presence of organic solvents, PCBs have been shown to leach significantly in soil thereby making it a concern at hazardous waste sites (Chou and Griffin, 1986). In surface waters, small amounts of PCBs remain dissolved but most settle in bottom sediments due to their high specific gravity and affinity for solids. PCBs persist in the environment and may have an estimated half-life in terrestrial soil of several years (USDHHS, 1995). Sediments containing PCBs at the bottom of a large body of water such as a lake or river generally act as a reservoir from which PCBs may be released in small amounts over time. The breakdown of PCBs in water and soil occurs over several years, or even decades (USDHHS, 1995). The ability of PCBs to be degraded or transformed in the environment depends in part on the degree of chlorination of the biphenyl molecule. In general, the persistence of PCB congeners increases as the degree of chlorination increases. PCBs are removed from the environment primarily by photochemical degradation or biodegradation. Photochemical degradation utilizes light energy to replace chlorine atoms with hydroxyls, ultimately dechlorinating PCBs. Generally, chlorobiphenyls with higher chlorine content undergo degradation faster than those with lower chlorine content. However, PCBs in bottom sediments not exposed to light will not degrade in this fashion. In biodegradation, both anaerobic and aerobic microorganisms present in soil and sediments decompose and metabolize PCBs. Biodegradation rates are highly variable because they depend on a number of factors including: the degree of chlorination, concentration of PCBs, types of microbial populations present, and the available nutrients and temperature in the subsurface (USDHHS, 1995). Generally, microbial degradation of the lower chlorinated biphenyls has been found to occur at a faster rate than the higher chlorinated biphenyls, but the process can be enhanced by the addition of pre-exposed microbial populations. Humans can be exposed to PCBs by the ingestion of contaminated food, inhalation or dermal contact with contaminated media. Since 1985, when PCBs were restricted to sealed systems, ingestion has become the most significant route of exposure to the general population while inhalation and dermal contact are associated more with occupational exposure. Food can become contaminated with PCBs as a result of accidental spills, equipment malfunctions, and from contaminated food packaging. Currently, the primary source of PCB ingestion is through the consumption of contaminated fish (USDHHS, 1995). 3 Fish uptake PCBs in water through their gills and through the food chain by consumption of contaminated aquatic organisms. Once PCBs are absorbed into the bloodstream they accumulate primarily in fatty tissues where they have the ability to biomagnify, or increase in concentration, as the compound is transferred through the food chain. In humans and other mammals, PCBs accumulate in the gastrointestinal tract, adipose tissue and skin. Opinions vary as to the precise health risks PCBs pose, however they are classified as probable human carcinogens and several studies suggest they can enhance the carcinogenicity of other chemicals. Most information regarding health effects of PCBs in humans is generated from occupational exposure studies. Currently, there is no conclusive evidence linking exposure of PCBs to cancer in humans. Individuals who have been exposed to PCBs have experienced symptoms such as chloracne, jaundice, numbness and swelling of limbs, spasms, hearing and vision problems, increased eye discharges, and gastrointestinal disorders (USEPA, 1980). Epidemiological studies indicate the major toxic effect in animals appears to be liver damage. Other effects include stomach, thyroid and kidney damage and immunosuppressive effects. Further laboratory testing has shown that PCBs are fetotoxic in rats, monkeys, minks and rabbits. 1.3 STUDY AREA DESCRIPTION The Ohio River is 981 miles long, starting at the confluence of the Allegheny and Monongahela Rivers in Pittsburgh, PA and ending in Cairo, IL where the Ohio flows into the Mississippi River. The TMDL discussed in this report is for the portion of the Ohio River that begins at the Pennsylvania and West Virginia border near Chester, WV at Ohio River Mile (ORM) 40.0, and extends downstream for 277 river miles to the border between Kentucky and West Virginia near Kenova, WV at ORM 317.1 (Figure 1-1). Along this stretch, the Ohio River forms the border between Ohio and West Virginia. The Ohio River Basin upstream of the TMDL segment drains approximately 23,300 square miles and includes three major tributary sub-basins (i.e., drainage area greater than 1,000 square miles) ­ the Allegheny, Monongahela, and Beaver Rivers. Within the TMDL study area, five major tributaries enter the Ohio River. These tributaries are the Muskingum, Little Kanawha, Hocking, Kanawha, and Guyandotte Rivers, and their confluences are at Ohio River mile points 172.2, 184.6, 199.3, 265.7, and 305.2, respectively. The Ohio River at the downstream end of the study area has a drainage area of approximately 56,000 square miles. 4 Figure 1-1. Map of the TMDL study area. Beaver Allegheny PA
Pittsburgh Ohio River OH
Muskingum Flow Wheeling Monongahela Marietta Hocking
Parkersburg WV
Little Kanawha Gallipolis Ohio River Kanawha Big Sandy Huntington Guyandotte TMDL Segment KY 5 2.0 TMDL DEVELOPMENT
2.1 APPLICABLE WATER QUALITY STANDARDS A TMDL must be designed to meet the applicable water quality standards. Water quality standards for both West Virginia and Ohio must be considered in the development of this TMDL since the Ohio River forms the boundary between the two states. The water quality criteria established in ORSANCO's Pollution Control Standards (2000) also apply to the Ohio River, and must be considered in the TMDL. Pennsylvania's water quality standards must be considered with respect to establishing load allocations to the tributaries entering the Ohio River upstream of the TMDL segment. West Virginia's human health water quality standard for PCBs is 0.044 nanograms/liter (ng/L) (WV 46CSR1) and is applied at all times when river flows are equal to or greater than the minimum seven consecutive day drought flow with a ten year return frequency (7Q10) (WV 46-1-7.2b). This value is established at a cancer risk level (CRL) of 10-6 or one additional cancer case per 1,000,000 individuals. The current human health water quality standard for PCBs in Ohio is 0.79 ng/L. Ohio, however, recently proposed changes to their water quality standards. The new standard, which will go into effect in November 2002, is 1.7 ng/L. The standard is applied at onetenth the harmonic mean flow, and represents a CRL of 10-5, or one additional cancer case per 100,000 individuals. ORSANCO's human health water quality criterion for PCBs is established as the criteria published by the USEPA pursuant to Section 304(a) of the Clean Water Act. USEPA's water quality criterion, and therefore ORSANCO's criterion, is 0.17 ng/L. As PCBs are considered a carcinogen, this value is applied at the harmonic mean flow, and represents a CRL of 10-6, or one additional cancer case per 1,000,000 individuals. The human health water quality standard for Pennsylvania is 0.044 ng/L. This value is applied at harmonic mean flows due to the carcinogenic nature of PCBs. This standard represents a CRL of 10-6, or one additional cancer case per one million individuals. The Pennsylvania standard does not apply to the Ohio River TMDL segment, however, this standard was used as a basis for determining tributary allocations for streams that enter the Ohio River upstream of the TMDL segment. The West Virginia water quality standard of 0.044 ng/L is more restrictive than that of Ohio and ORSANCO. The West Virginia standard, being more protective of human health, has therefore been used to establish the TMDL endpoints within the TMDL segment. This numeric endpoint identifies the in-stream concentration at which all designated uses of the Ohio River will be attained. The endpoint also provides the basis for calculating the allowable PCB loadings in the Ohio River, and determining the load reductions necessary to meet water quality standards. 6 2.2 CRITICAL CONDITION AND SEASONALITY Concurrent with the selection of a numeric endpoint, the environmental conditions that will be used to calculate the allowable loads must be defined. TMDLs generally are designed around the concept of "critical condition." The critical condition is the set of environmental conditions, if controls are designed to protect, that will ensure attainment of standards for all other conditions. Because PCBs are considered carcinogenic and human health criteria for carcinogens are derived assuming a lifetime exposure, PCB human health criteria thus apply to ambient water concentrations averaged over a human lifetime (approximately 70 years). Harmonic mean flow has been used in other TMDLs to best represent the averaging of hydrologic conditions over a long period of time. As a result, harmonic mean flow has been selected as the hydrologic condition that this TMDL will be based on. Table 2-1 presents the established harmonic mean flows for the Ohio River in the TMDL segment. Table 2-1. Harmonic mean flow values for the Ohio River within the TMDL study area (ORSANCO, 2000). Ohio River Segment (river miles) 40.0 ­ 161.7 161.7 ­ 237.5 237.5 ­ 279.2 279.2 ­ 305.2 305.2 ­ 317.1 Harmonic Mean Flow (feet3/second) 20,500 24,500 26,000 34,500 35,900 Also, while the West Virginia water quality standard applies to the 7Q10 low flow condition, establishing the critical condition at the harmonic mean flow is considered protective of the WV standard since instream PCB concentrations have been found to increase with stream flow. Because of this positive correlation between stream flow and concentrations, load reductions established at the harmonic mean flow will result in instream concentrations below the WV standard at lower flow conditions. Seasonality also must be considered in the TMDL development process. Simply stated, seasonality, in the context of a TMDL, refers to the natural variations of environmental conditions that affect pollutant concentrations. Stream flow is the most important environmental condition to consider for PCBs. On the Ohio River, periods of high flow conditions generally occur during the early spring months, while low flow seasonally occurs in late summer or early fall. In-stream concentrations of PCBs are directly affected by stream flow. In cases where point sources dominate, concentrations will be greatest during drought conditions due to less water for dilution. Conversely, PCB loads from non-point sources are greatest during rainy, high flow periods due to increased loadings from overland runoff of contaminated soils and resuspension of contaminated sediments from the river bottom. 7 While significant variations in concentrations of PCBs have been observed in the Ohio River, seasonality is inherently accounted for through use of the harmonic mean flow as the critical condition. Harmonic mean flow provides a representative long term average, that is consistent for use with a human health standard based on a lifetime exposure, as is the case for PCBs and all carcinogens. 2.3 TOTAL MAXIMUM DAILY LOAD CALCULATION The total maximum daily load (TMDL) represents maximum pollutant loading a water body can assimilate at the critical flow condition and still meet the applicable water quality standard (WQS). The TMDL can be calculated using the following equation: TMDL = Qriv x CWQS Where: TMDL = Total Maximum Daily Load to meet water quality standard (mass/time) Qriv = River flow (volume/time) CWQS = Water Quality Standard (mass/volume) The allowable PCB loads for the Ohio River within the TMDL study area are presented in Table 2-2. These loads represent the total maximum daily loads allowed for each river segment at harmonic mean flow using the applicable water quality standard of 0.044 ng/L. Table 2-2. Total maximum daily PCB loads to meet WQS for segments of the Ohio River. Ohio River Segment (river miles) 40.0 ­ 161.7 161.7 ­ 237.5 237.5 ­ 279.2 279.2 ­ 305.2 305.2 ­ 317.1 Harmonic Mean Flow (feet3/second) 20,500 24,500 26,000 34,500 35,900 PCB TMDL (grams/day) 2.207 2.637 2.799 3.714 3.865 2.4 EXISTING OHIO RIVER PCB LOADS ORSANCO utilized an innovative sampling technique referred to as high-volume water sampling to quantify ultra low-level concentrations of PCBs in the Ohio River, major tributaries and discharges. This sampling method involves filtering a large volume of water (1,000 liters) in order to collect a sufficient amount of PCBs, such that it can be detected by existing high-resolution analytical methods. This process is accomplished by first drawing the river water through glass fiber filters that separate and collect suspended solids. The filtered water then passes through stainless steel columns filled with a hydrophobic resin (XAD2) that extracts the PCBs present in the dissolved phase. The filters and columns then are analyzed separately to quantify PCB levels in both the 8 particulate and dissolved phases. While this methodology is not yet approved by EPA, it provides a way to quantify low levels of PCBs. High-volume water samples were collected at nine Ohio River sampling locations within the TMDL segment, and at four sites on the Ohio River in Pennsylvania, upstream of the TMDL segment. Each site was sampled at least twice, with some targeted sites sampled up to five times. Sampling was conducted at different river flow conditions to evaluate the range of PCB concentrations present in the river. Samples were analyzed by USEPA Method 1668A for all 209 PCB congeners. Figure 2-1 graphically illustrates the Ohio River high-volume sampling results for total PCBs (sum of dissolved and particulate phases combined). In some cases only the dissolved or particulate portions of the samples were available for analysis. Data for these samples are not included in the graph; however, these results are provided in the complete high-volume water sampling data summary in Appendix A. Figure 2-1. Observed Ohio River PCB concentrations. 5 4 PCBs (ng/L) 3 2 1 0 0 50 100 150 200 250 300 350 Ohio River Mile
These results clearly indicate that current PCB levels in the Ohio River exceed West Virginia's water quality standard of 0.044 ng/L for all conditions sampled. The variability in the observed concentrations also indicates that loadings can fluctuate significantly with changes in stream flow. As mentioned earlier in this report, the allowable PCB loading calculations for this TMDL are based on a critical condition established at harmonic mean flow. To compare allowable loads to existing loads in the Ohio River, and to quantify necessary reductions to meet water quality standards, current in-stream concentrations and loadings at harmonic mean flow were predicted using the high-volume sampling data collected over a range of flow conditions. Applicable Water Quality Standard 9 In-stream PCB concentrations and loadings were calculated only for sites within the TMDL segment that were sampled on at least three occasions. Data for each site were evaluated separately to select the best method to predict the concentration at harmonic mean flow. Concentrations were estimated for the seven Ohio River sampling points included in Table 2-3. For all sites except ORM 207.7, a clear positive correlation between stream flow and PCBs in the particulate phase was observed. This type of correlation is expected since PCBs strongly bind to particulates in the water column, and suspended solids concentrations increase with stream flow due to an increase in soil erosion and resuspension of bottom sediments. Dissolved phase PCB results, in general, did not show a correlation with stream flow except at ORM 129. At this site, there was a positive correlation between dissolved PCB concentrations and flow. Based on the observed correlations mentioned above, total PCB concentrations at harmonic mean flow were calculated for each site except ORM 129.0 and ORM 207.7 based on the average dissolved concentration plus the estimated value generated from a linear regression between stream flow and observed particulate phase concentrations (see Table 2-3). At ORM 129.0, both the particulate and dissolved phase concentrations indicated a direct relationship with stream flow, therefore the total concentration at this site was determined by a linear regression between flow and total PCB concentration. No relationship between flow and PCB levels was found for sampling data collected at ORM 207.7. Using a simple mean value for the total PCB concentration was considered for this site, however, the mean concentration value, which would be applied at harmonic mean flow, was less than the concentration observed at lower stream flows. It was deemed that a more conservative approach should be applied to ensure that the water quality standard will be attained provided that the reductions called for in this TMDL are met. Ultimately, the concentration used for ORM 207.7 was established at the single highest total PCB concentration measured at flows less than the harmonic mean flow. This value of 2.44 ng/L is more conservative, and therefore more protective of human health, than the mean concentration of 1.84 ng/L. Using the predicted in-stream concentrations at harmonic mean flow, PCB loads were calculated for each sampling location (see Table 2-4). The greatest daily PCB load for the seven sampling locations (229.1 grams/day) occurred at the upstream TMDL boundary at the Pennsylvania/West Virginia border (ORM 40.0). This loading exceeds the allowable load by more than two orders of magnitude. The PCB load generally decreases as you move downstream, with the most significant decrease in load (approximately 35% decrease) occurring between ORM 40.0 and ORM 129.0. The one exception is between ORM 129.0 and ORM 175.1, which saw a slight increase in load from 148.6 grams/day to 152.0 grams/day (2% increase). Overall, the load at the most downstream site (ORM 302.9) is less than half of the existing load at the upstream boundary (ORM 40.0). This significant natural load reduction is likely the result of settling of contaminated particulate matter. 10 It should be pointed out that the existing load estimated in this report for ORM 40 is significantly less than that presented in the Ohio River PCB TMDL completed by Pennsylvania for the upper 40 miles of the Ohio River (2001). At the time the Pennsylvania TMDL was completed, no high volume sampling data was available for the Ohio River and water column concentrations were extrapolated using fish tissue sampling results. This extrapolation resulted in an estimated water column concentration of 45.77 ng/L, the predicted concentrations based on actual water column analytical data presented in this report is 4.57 ng/L. This represents an order of magnitude difference in predicted existing load values. Using the river data collected by ORSANCO provides a more accurate estimate of the existing PCB load at mile point 40 of 229.08g/day rather than the 2292 g/day prediction derived from extrapolation of fish tissue results. Table 2-3. Predicted PCB concentrations at harmonic mean flow. Sampling Location (Ohio River Mile) ORM 40.0 ORM 129.0 ORM 175.1 ORM 207.7 ORM 264.0 ORM 281.5 ORM 302.9 Harmonic Mean Flow (feet3/second) 20,500 20,500 24,500 24,500 26,000 34,500 34,500 Predicted PCB Concentration (ng/L) 4.57 2.96 2.54 2.44 2.27 1.49 1.31 Method for Estimation of Total PCB Concentration Linear regression for particulate + average dissolved Linear regression for total PCBs Linear regression for particulate + average dissolved Highest observed concentration at < harmonic mean flow Linear regression for particulate + average dissolved Linear regression for particulate + average dissolved Linear regression for particulate + average dissolved 2.5 NECESSARY LOAD REDUCTIONS Comparing existing loads to allowable loadings, the load reductions necessary to meet the applicable water quality standard of 0.044 ng/L were established. Necessary load reductions for the Ohio River ranged between 96.6 ­ 99.0 percent, with the greatest reductions needed at the upstream TMDL boundary (ORM 40.0). Table 2-4 presents the loading information, along with the necessary reductions to meet standards. 11 Table 2-4. Ohio River load reductions necessary to meet water quality standards. Sampling Location (river mile) ORM 40.0 ORM 129.0 ORM 175.1 ORM 207.7 ORM 264.0 ORM 281.5 ORM 302.9 Existing Load (g/day) 229.080 148.636 152.013 146.017 144.206 125.972 110.500 Maximum Allowable Load (g/day) 2.207 2.207 2.637 2.637 2.799 3.714 3.714 Load Reduction (g/day) 226.873 146.429 149.376 143.380 141.407 122.258 106.786 % Reduction Necessary 99.0 98.5 98.3 98.2 98.1 97.1 96.6 2.6 MARGIN OF SAFETY To account for any uncertainties associated with the TMDL analysis, a margin of safety (MOS) must be incorporated into TMDL calculations. The MOS can either be implicit (e.g., use of conservative assumptions in the TMDL analysis) or explicit (e.g., expressed as a percentage of the total allowable loading held in reserve as a safety factor). For the TMDL discussed in this report, the MOS is implicitly incorporated through conservative assumptions. The two areas where conservative assumptions are applied to provide a MOS are 1) mass is assumed to be completely conserved as it passes through the study area, and 2) the existing Ohio River and tributary loadings, and therefore necessary load reductions, are estimated using a conservative approach to ensure that the applicable water quality standard is met. For the Ohio River, the existing loads established in Sections 2.4 and 2.5 were estimated based on the either a linear regression between concentration and stream flow or the highest observed concentrations observed at stream flows less than the harmonic mean flow. The higher of the two values generated by these methods was used to establish the current Ohio River loads. Unlike the main stem Ohio River data, the tributary results did not show a clear correlation with stream flow. As a result, the linear regression method used to estimate Ohio River concentrations at harmonic mean flow could not be applied to the tributaries. Instead, a combination of two methods was used to provide estimates of the concentrations at harmonic mean flow that were both reasonable and conservative with respects to protecting human health. For each tributary, the average total concentration was compared to the highest concentration observed at stream flows less than the harmonic mean flow. The higher of the two values was selected as the estimated concentration at harmonic mean flow for loading calculations. This conservative method was applied to ensure that the water quality standard would be attained provided that the reductions called for in this TMDL are achieved. 12 3.0 INDUSTRIAL AND MUNICIPAL SOURCE ASSESSMENT
3.1 METHODS FOR SOURCE IDENTIFICATION
Sampling was conducted by ORSANCO as part of the Ohio River Watershed Pollutant Reduction Program to quantify current levels of PCBs in ambient air, water, sediment, and fish tissue within the TMDL study area. In addition to establishing the current concentrations present in the environment, the analytical results were reviewed to identify "hot spots" of contamination, and potentially identify PCB sources. The investigation of sites where PCB hot spots were detected is listed in the following section. ORSANCO conducted an extensive search to identify potential sources within the upper portion of the Ohio River for PCB loadings. Industrial and municipal sources were identified using several different databases, agencies, and contacts. The National Priorities List (NPL) search was done using the Right to Know (RTK) Network database to identify all CERCLA (Comprehensive Environmental Response, Compensation, and Liability Act) facilities in each state within the Ohio River Basin. The list of sites then was reduced to only those facilities on the final NPL or proposed for the final NPL. The sites in the watershed were then investigated on EPA's CERCLIS (Comprehensive Environmental Response, Compensation and Liability Information System) database website to identify chemicals of concern (COCs). COCs were taken from the official Record of Decision (ROD) when available or from the EPA region's superfund site descriptions. Since a large amount of information was returned from the search, the final NPL list contained in Appendix B contains only information on NPL sites listed for PCBs in Ohio River counties relating to Ohio River miles zero to 317. Counties in the watershed were queried to generate a list of facilities reporting releases of PCBs and the quantities released. This search was conducted by using USEPA's Toxic Release Inventory (TRI) and the RTK database. The years 1988 through 1998 for each state were queried. State agencies for Pennsylvania, West Virginia, and Ohio were also contacted. State agencies ran queries within the Permit Compliance System (PCS) to yield returns on facilities that have National Pollutant Discharge Elimination System (NPDES) permits that require monitoring for PCBs. State contacts provided such information to ORSANCO through telephone conversations and documents via email. NPDES searches were conducted using USEPA's Envirofacts Warehouse database and onsite at ORSANCO, using the filed NPDES permits for Ohio River dischargers. A complete listing of all of sites identified through these searches is provided in Appendix B. During the source assessment, weaknesses were found within the databases used for identification of possible sources of PCB loading. Insufficient data within TRI, such as pathways of releases and quantities released, provided problems in assessing the potential impacts of releases to the Ohio River. ORSANCO made every effort to obtain the best and most complete source information available, however, there are gaps in the data regarding the sources due to the limitations and incompleteness of the databases searched. 13 Databases and Agencies utilized in Source Investigation · Toxic Release Inventory (USEPA) · Right To Know Network · Permit Compliance System (USEPA and States of Pennsylvania, West Virginia, and Ohio) · USEPA's Envirofacts Warehouse database · ORSANCO NPDES permit files cataloged onsite · Comprehensive Environmental Response, Compensation, and Liability Information System (CERCLIS) · Individual federal and state agency personnel 3.2 SITE DISCUSSIONS
The information regarding sites discussed below is based on the multimedia sampling results collected by ORSANCO, and facility information found within the source assessment discussed in Section 3.1. The facilities named below are either those that have confirmed, past PCB contamination problems, or those facilities in which highvolume water sampling results revealed the presence of PCBs in their effluents. Further investigation is warranted prior to any recommended action at any of these locations. Appendix B includes a listing of sources identified through searches of the TRI, NPL, and PCS databases. Also provided in Appendix C, is an inventory of potential PCB sources identified based on general industry type. This list includes industries simply associated with the use of PCBs, and therefore, many of these facilities may not be actual PCB sources. 3.2.1 Ohio River Mile Point 3.3 Elevated levels of PCBs were detected in a sediment sample taken at Ohio River mile point 3.3. One high-volume water sampling event was conducted on ALCOSAN's effluent, which is a 200 million gallon per day (MGD) sewage treatment plant located at mile 3.1, in order to quantify potential PCB loadings from the plant. A total concentration of 6.4 ng/L was measured for the single sampling event. Applying this concentration to the plants design discharge capacity of 200 MGD, the potential PCB load from this facility is 4.9 g/day. The allowable PCB load for this section of the Ohio River at harmonic mean flow is 1.6 g/day. Based on this information, further sampling of the river and sediments in this area may be warranted, in addition to further upstream sampling, sampling of the ALCOSAN system, and public and industrial water supplies tributary to the ALCOSAN system. It is possible that the detection of PCBs in ALCOSAN effluent is simply due to pass through of upstream river concentrations from water supply systems tributary to the ALCOSAN system since river concentrations in this area were found to be between 1 and 5 ng/L. The former Allis Chalmers site is located in Pittsburgh, Allegheny County, PA, on the north bank of the Ohio River across from Brunot Island. During the 1970s, USEPA conducted an investigation and documented that a 30,000-gallon vault of PCBs was at this site. The PCB TMDL report completed by Pennsylvania Department of 14 Environmental Protection (PA DEP) for the Pennsylvania stretch of the Ohio River reported that, based on information provided by USEPA, the vault of PCB contaminated oil at this site has been removed. The report also indicated that there is no evidence to suggest this site is currently a source of PCB contamination in the Ohio River basin. However, some of the PCBs contaminating Ohio River sediments could be the result of past releases from the former Allis Chalmers facility. 3.2.2 Ohio River Mile Point 36.3 Elevated levels of PCBs were detected in a sediment sample taken at Ohio River mile point 36.3. Such results suggest potential sources within the vicinity. A steel manufacturing facility, power generating facility, and a petroleum terminal, all reside in close proximity to the site where the sample was collected. Unfortunately, no information was found through database searches that points to potential sources of the PCB contamination. 3.2.3 Ohio River Mile Point 71.4 Elevated levels of PCBs (highest levels found by ORSANCO) were detected within sediment collected at Ohio River mile point 71.4. These results suggest potential localized sources. The sample was collected at the downstream edge of a large barge mooring area adjacent to a large steel making facility in Steubenville, Ohio. While the location of the sampling point suggests this facility as a possible source, other nearby potential sources are located upstream of the contaminated sediment. Another large steel manufacturing facility is located 1.4 miles upstream and a wastewater treatment plant is 0.9 miles upstream of the site. 3.2.4 Ohio River Mile Point 122.9 Elevated levels of PCBs were detected within the sediment sample taken at Ohio River mile point 122.9. Such results suggest potential nearby sources. Data obtained from the NPL search and the CERCLIS database regarding ORMET (NPDES permit number OH0010855) indicate that this facility may potentially be the source of elevated PCB levels in the sediment sample taken at this site. The sample was removed directly in front of what once was a backwater drainage ditch for Outfall 004 at the facility. During Superfund cleanup, this area's sediment was sampled and showed the highest contamination at the facility. The facility was placed on the NPL with confirmed Aroclor 1248 contamination. Since cleanup, the drainage/backwater area from former Outfall 004 has been bermed and closed off from public access from the river with fencing. Upstream of Ohio River mile point 122.9, other potential sources include industrial chemical and metal coating facilities. However, no information was found through database searches that points to other potential sources of PCB contamination. 3.2.5 Monongahela River Mile Point 2.6 Elevated levels of PCBs were detected in a sediment sample taken at Monongahela River mile point 2.6. Such results suggest potential sources within the vicinity. A large steel making facility and a petroleum company both reside upstream of the sample location, however, no conclusive data exists to determine the source of the PCB contamination. 15 3.2.6 Great Kanawha River Mile Point 44.0 High-volume water sampling was conducted on the effluent of the Nitro Wastewater Treatment Plant (WWTP) located on the Great Kanawha River at mile 44.0. The Great Kanawha River enters the Ohio River near Point Pleasant, West Virginia at Ohio River mile 265.7. The results for the single sampling event indicated a total PCB concentration of 4.6 ng/L. Applying this concentration to the plants design capacity of 1.25 MGD, the plants potential PCB load to the Kanawha River is 0.022 g/day. 3.3 POTENTIAL SOURCES IDENTIFIED IN PA OHIO RIVER PCB TMDL In addition to the sites referred to above, several other sites along the Ohio River were identified in the Ohio River PCB TMDL completed by Pennsylvania DEP in 2001. These include the Breslube-Penn site, the former H. K. Porter site, the Texas Eastern Holbrook Compressor Station, and the Ohio River Park. The former Allis Chalmers site was also referred to in the PA PCB TMDL, which was previously discussed in Section 3.2.1, Ohio River Mile 3.3 discussion. The Breslube-Penn site is located in Coraopolis, Allegheny County, PA. The site is situated along Montour Run, a tributary that enters the Ohio River at mile 9.7. The facility historically operated as a solvent recovery and oil recycling facility, and currently is inactive. The PA PCB TMDL stated that elevated levels of PCBs had been found in soil and groundwater at a soil staging area and filter cake area, where soil and filter cake wastes from previous remedial activities had been stockpiled on site. The report indicates that sampling of this area, revealed an average PCB concentration of 52 mg/kg. The site may be an existing source of PCBs to the Ohio River through continuous contaminated soil erosion; however, there is insufficient data to quantify any contributions. The former H. K. Porter site is located in Hopewell Township, Beaver County on Shouse Run. Shouse Run is a tributary to the Ohio River, entering the Ohio at river mile 14.8. PCB concentrations in the soil are documented to be as high as 130 mg/kg; however, no PCBs were detected in Shouse Run. This site is being addressed under Pennsylvania's Hazardous Sites Cleanup Act (HSCA) program. The former H. K. Porter Drum Dump Site is located on approximately 17.5 acres of property situated 0.25 miles west of the Ohio River and adjacent to State Route 51 in Hopewell Township. Shouse Run transects the property and is located at the end of the disposal area, which contained between 1,500 and 2,000 rusted 55-gallon drums containing various hazardous wastes. Analytical results from soils and wastes collected from October 1990 through January 1993 revealed the presence of lead and PCBs at elevated concentrations. In 1991, H. K. Porter excavated approximately 7,875 tons of non-hazardous wastes and 4,260 tons of hazardous wastes from the disposal area. In the late 1990s, Pennsylvania Department of Environmental Protection (PA DEP) conducted additional cleanup activities under HSCA, including the excavation and off-site disposal of about 50,000 cubic yards of hazardous waste. A soil cover was then installed and the entire site was revegetated. The site does not represent a current source of contaminated soil erosion to the Ohio River; however, past releases may have contributed to the sediment contamination in the Ohio River. 16 The Texas Eastern Holbrook Compressor Station (NPDES permit number PA0216593) is located in Richmond Hill, Greene County. This site was an historic non-point source of PCBs in the Ohio River watershed. A statewide Consent Order and Agreement (CO&A) required Texas Eastern to remove PCB contaminated soil, and to collect and treat contaminated groundwater. The facility currently discharges treated groundwater to Dunkard Fork Creek, a tributary of Wheeling Creek, which enters the Ohio River at mile 90.8. The NPDES permit allows for an average monthly concentration of 1.87 ng/L. Based on the plants design discharge capacity of 0.0489 MGD, the allowable daily load for this facility is 0.0003 grams. The Ohio River Park site is located approximately ten miles downstream of Pittsburgh, PA on the western end of Neville Island, within the Ohio River. This site has a NPL status of final. Remedial actions have been completed under CERCLA and a sports complex has been developed on the site, therefore, covering any remaining contaminated soil that could serve as a potential non-point source of PCB to the Ohio River. 3.4 GENERAL DISCUSSION ON PUBLICLY OWNED TREATMENT WORKS ORSANCO sampled effluents at the ALCOSAN Wastewater Treatment Plant (WWTP) (Ohio River mile 3.1) and the Nitro, West Virginia WWTP (Great Kanawha River mile 44.0). Initially, sampling was conducted at these sites to evaluate the possibility that POTWs in general discharge dioxin. These sites were not targeted based on any known contamination. ALCOSAN was selected simply because it is the largest POTW on the Ohio River. The Nitro plant was sampled because the facility receives wastes from several potential and confirmed dioxin sources. Since dioxins were found in samples taken at both ALCOSAN and Nitro WWTPs, ORSANCO elected to analyze the samples for PCBs as well. PCBs were detected in the high-volume water samples collected at both facilities. Similar results were found in a sample collected at another major POTW downstream of the TMDL study area. These results suggest that POTWs in general may be sources of PCBs to the Ohio River. It should be noted, however, that there is no information suggesting that POTWs create new PCBs. Potential sources of PCBs to these facilities include industrial sources, runoff from contaminated sites and other land-based runoff and the numerous water supply systems tributary to them which withdraw their water from the river representing pass through of existing PCB loads and resulting in no net increase in PCB levels in the river above those upstream of the discharges. Using the average concentration observed at the three facilities mentioned above, a gross estimate of the potential loading from all municipal wastewater treatment plants that directly discharge to the Ohio River within the TMDL study area was calculated. This estimate was based on the facilities design flow capacity, and an average concentration of 6.14 ng/L. Based on this calculation, 7.2 grams/day may potentially be entering the river from WWTPs between Ohio River miles 0.0 to 317.1. This loading represents 6.5 percent of the current Ohio River load measured at Ohio River mile 302.9. This load is also represents 186 percent of the allowable load at the downstream end of the TMDL segment. 17 4.0 ENVIRONMENTAL SOURCES
4.1 SEDIMENT Samples of Ohio River and tributary bottom sediments were collected from the confluence of the Allegheny and Monongahela Rivers to ORM 317 during low flow conditions in August and September of 2001. Bottom sediment was collected approximately every five miles on the main stem, at 26 targeted sites, and from each major tributary of the study area. Ninety-two bottom sediment samples were collected in all, nine of those duplicates, at a total of 83 sites. The purpose of the sediment survey was to characterize Ohio River bottom sediments from Pittsburgh through the TMDL study area. The survey was also intended to address water-column PCB loads resulting in part from resuspension of contaminated sediments. A secondary goal was to identify previously unknown "hot spots" or areas with significant PCB contamination. 4.1.1 Method Ohio River and tributary sediments were collected using the ORSANCO Standard Operating Procedure for Collection of Bottom Sediments. Samples were collected from a boat using a Petite Ponar clamshell-style sample dredge. Sediment samples were sieved in the field to remove particles larger than 2mm. Twenty-six targeted sample sites were selected based on their proximity to sites listed on the final NPL, TRI, or state agency records of contaminated sites. These samples were taken below outfalls of industrial sites or at the mouths of creeks draining the properties of interest. 4.1.2 Sediment Data and Results Eighty-three bottom sediment samples were collected in ten tributaries and 73 locations on the main stem of the Ohio River. In addition to PCBs, the samples were analyzed for dioxins and furans, chlordane, total organic carbon (TOC), and particle size composition. Results for total PCBs, TOC, and particle size are presented in a tabular format in Appendix B. 4.1.2.1 Polychlorinated Biphenyls (PCB) Analysis Total PCB data for the Ohio River sediment collected indicates widespread, low-level PCB contamination in the environment, as well as several areas of higher concentration zones of PCB contamination. Two locations not previously identified by the database investigation of sources were found to have significant PCB contamination in sediments, and Ohio River sediment contamination from several sites with documented PCB contamination was confirmed. Laboratory analysis for all 209 polychlorinated biphenyl congeners was done using USEPA method 1668A for High Resolution Gas Chromatography/High Resolution Mass Spectrometry (USEPA, 1999). Detection limits for this set of sediment samples ranged 18 from 1x10-6 to 1x10-5 parts per million (ppm). For simplicity, all total PCB values reported in this text are in ppm. 4.1.2.1.1 Sediment Quality Criteria Although specific sediment quality criteria for total PCBs has not been standardized for the Ohio River, The Incidence and Severity of Sediment Contamination In Surface Waters of the United States (EPA 823-R-97-006), also known as The National Sediment Inventory (NSI), includes multiple PCB screening levels for the protection of consumers. These values are based on the theoretical bioaccumulation potential (TBP) and cancer risk levels from the primary route of human exposure to contaminated sediment, consumption of fish. Screening levels are guidelines for analysis of sediment quality data; they have no applicability as regulatory criteria. The NSI calculated a 0.0