\ I / United States Environmental Protection Agency Municipal Environmental Research Laboratory Cincinnati OH 45268 Research and Development EPA-600/S2-84-112 Sept. 1984 Project Summary Predicting Toxic Waste Concentrations in Community Drinking Water Supplies: Analysis of Vulnerability to Upstream Industrial Discharges James A. Goodrich and Robert M. Clark A study was conducted to predict toxic waste concentrations in com- munity drinking water supplies along the Ohio and Kanawha Rivers between Charleston, West Virginia, and Cincin- nati, Ohio, using QUAL-II, a water quality simulation model. Specifically, a toxics screening method was devel- oped that can close the potential gap in the loop between water pollution control and water consumption. The project was a response to the lack of methods for identifying and assessing communities whose water supplies were vulnerable to excessive chloroform and synthetic organics resulting from industrial pollution and urban and agricultural runoff. The most important factors to consider in identifying vulnerable communities are potency and persistence of the pollutants, amount and timing of discharge of pollutants, storage times of utilities, and relative location of point sources and community intakes. This Project Summary was developed by EPA's Municipal Environmental Research Laboratory, Cincinnati, OH, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction On Februarys, 1978, the Environmental Protection Agency (EPA) proposed a regulation designed to protect the public health from organic chemical contam- inants in drinking water (Federal Regis- ter, "Environmental Protection Agency Interim Primary Drinking Water Regula- tions," Vol. 43, No. 28, Thursday, Febru- ary 9, 1978, pp. 5756-5780). Originally, the regulation consisted of two major parts: A. A maximum contaminant level (MCL) that initially required water treatment systems serving popula- tions greater than 75,000 to reduce trihalomethanes (TTHM) to 0.10 mg/L (100 parts per billion), and B. A treatment technique regulation that required water systems serving populations greater than 75,000 to use GAC in the treatment process to remove synthetic organic chemi- cals unless a variance was granted. Part A was promulgated as proposed on November 29, 1979; but Part B has been cancelled as proposed because of the wide concern over the impact of this regulation, especially regarding the procedure granting variances. Are those utilities within a specified number of miles downstream from industrial dis- chargers vulnerable and subject to the regulation? Or should utilities be granted a variance if fewer than a certain number of dischargers exist upstream from its intakes? This project uses a case study to answer these questions and to close the gap that potentially exists in this ------- loop between water pollution control and water consumption. Through the use of the methodology developed here, this study predicts which communities in the case study area will be vulnerable to (1) background levels of typical daily discharges, and (2) large, unexpected spills. Thus the report describes who is at risk, their relative levels of risk, and the main factors contributing to the risk. QUAL-II, a water quality simulation model, was used to bring together the diverse elements of mathematical model- ing, fluid dynamic's, organic chemistry, and geography to create an interactive systems analysis approach that can have an impact on public policy in drinking water. Though QUAL-II is less flexible than other models in simulating various flow scenarios and less sophisti- cated in modeling dozens of built-in parameters and biological and chemical transformations, the model exhibits a spatial organization that simplifies thinking and highlights critical variables such as the relative locations of utilities and dischargers and the time of travel. Procedures Determining the Case Study Area Figure 1 presents the case study area, and Figure 2 schematically represents the waste loads, tributaries, and junctions involved in the water quality modeling. The contaminants were routed approxi- mately 200 miles at various flow scenarios to account for seasonal variations inflow. The Kanawha River averages 25,000 cubic feet per second (cfs), and the Ohio River, 125,000 cfs. Time of travel during average flow is approximately 4.1 days through the case study area. Identifying Existing Point Sources and Communities The next step in the analysis was to provide an inventory and description of existing point sources and communities. To make a complete and thorough analysis of a community's vulnerability to water pollution, nonpoint sources of pollution as well as the point sources should be considered. But tremendous gaps often exist in the land use data, especially in watersheds involving various states and regional authorities, and thus any attempts to model runoff water quality are pre-empted. To simplify the analysis, only industrial dischargers and their wastes are considered in this paper, though various techniques could be used to model both municipal discharges and nonpoint run off. Calculating Waste Stream Data for Industrial Dischargers Typical waste stream data for each industry type was needed next. For this analysis, each point source was assigned a discharge value in parts per billion based on the best available control technology (BACT) for the relevant contaminants coming from each indus- trial production process. For modeling purposes, a general idea of a pollutant's persistence in the stream is required. Based on extensive calculations and literature reviewed, disappearance rates to account for processes such as volatiliz- ation were assigned to each pollutant (Maybey, W.R., et al., "Aquatic Process Data for Organic Priority Pollutants," Final Draft Report, Michael W. Slimak, Project Officer, Monitoring and Data Support Division, Off ice of Water Regula- tions and Standards, U.S. Environmental Protection Agency, Washington, DC, July, 1981, pp. 409-434.) Determining Potential Impact on Public Health To assess the potential impact on public health, simulated pollutant con- centrations for each utility are compared with the Water Quality Criteria, which only suggest at this time the concentra- tions of various pollutants that could be harmful to human health. The Criteria take into account toxicity, carcinogenicity, or organolepticity (taste and odor) of the pollutants (Personal communication with Dr. Christopher T. DeRosa, EPA, Environ- mental Criteria Assessment Office, 1982). The Water Quality Critera for toxicity and taste and odor indicate the general population would be affected by consum- ing water with a pollutant reaching the guideline concentration for example, 3,770 fjg/L of cyanide (a toxic), or 300 yug/L phenol (an odor-causing agent). A tenfold buffer is incorporated in the toxic and organoleptic standards to take into account the more sensitive or susceptible consumers such as those who are very young, old, pregnant, or ill. Thus segments of a population could be possibly affected by 377 yug/L and 30yug/L of cyanide and phenol, respectively. The carcinogenic data are estimates of incremental risks associated with expo- sures from suspected carcinogens in drinking water. For example, a person is assumed to be at the 0.00001 risk level of developing cancer in his or her lifetime by drinking 2 L of water with 6.6 yug/L of benzene daily. The only no-risk level for carcinogens is zero concentration. Risk is assumed to be linear, but promoters and synergism among the pollutants could actually increase the risk levels. This analysis is to be undertaken for daily dischargers of industrial wastes. In the event of a spill or large accidental discharge, information regarding the storage time for each utility is also necessary. Storage time indicates how long a utility could operate if the intakes were closed to prevent the high concen- tration of pollutants from entering the system. This factor affects the vulnerability of a utility. Creating Various Flow Scenarios To assess the vulnerability of commun- ities to daily discharges of toxic wastes, three scenarios were created to account for variations in flow. With the use of QUAL-II, the applicable priority poll utants discharged in the case study area (81 out of the 129 priority pollutants) were simulated at average, high, and low flows. Average flow was set at 125,000 cfs, high flow was 220,000 cfs, and low flow was 35,000 cfs. As the volume of water increases, so does the mean velocity in the river channel, thus reducing the time of travel. For toxic and organoleptic pollutants, the flow scenarios are critical in deter- mining whether a Water Quality Criteria has been exceeded. Because vulnerability to carcinogens is evaluated over years of exposure to pollutants in the drinking water, carcinogenic risk levels were initially estimated only at average flows. In the event of a spill, however, the flow characteristics can be important even to carcinogens, since very high concentra- tions of carbon tetrachloride, for example, can have an acute health effect. In addition, regulatory agencies may want to use higher- or lower-than-average flows to increase safety factors. As will be demonstrated later, and contrary to the conventional wisdom, low flows do not necessarily exhibit higher pollutant concentrations than high flows because of decreased dilution. Utilities can be vulnerable to different pollutants at different flows. Results and Discussion Organoleptic Pollutants Only 2 of the 11 organoleptic pollutants simulated exceeded a level at which sensitive consumers could be affected by ------- Hamilton Co. Legend State Boundary County Boundary ^^^ River U.S. Highway ^~~ Direction of Flow g s is miles Figure 1. Case study area. KENTUCKY WEST VIRGINIA taste and odor problems. Those two pollutants are 2-chlorophenol and 2,4- dichlorophenol. Only during the low-flow scenario were these pollutants of any concern to the utilities. Though taste and odor problems are not dangerous, they often create greater response than do reports of possible carcinogens. Histori- cally, aesthetic considerations have often been the basis for regulation rather than public health concerns. Toxic Pollutants Toxic pollutants that exceeded the Water Quality Criteria values at the utility intakes are listed in Table 1. During low- flow periods, the sensitive consumers of all the utilities would have been affected by cadmium. Ashland consumers would also be affected by mercury, and Greenup would have to deal with lead and chromi- um in addition to cadmium. The simulated concentrations of mercury exceeded the criteria guideline for all consumers at Ironton, Greenup, Portsmouth, Maysville and Cincinnati during low flow. Because of greater dilution of the pollutants at high flow, only mercury remains a con- cern to some of the utilities. During a high-flow period, mercury would affect only the sensitive consumers for the five utilities mentioned above. At average flow, cadmium and mercury exceed the guidelines for sensitive consumers at the same utilities except for Greenup. Through- out this analysis, mercury remains a problem at various levels for all flow scenarios. Simulated cadmium, lead, and chromium concentrations would affect sensitive consumers only during low and average flow scenarios. Carcinogenic Pollutants Analyses of carcinogenic pollutants proceed differently from the previous analyses for toxic and organoleptic pollutants. When assessing risks posed by carcinogens, no single value signifies that a health hazard exists for each pollutant. Rather, the Water Quality Criteria describe the carcinogenicity at the 1 x10~5 risk level. These values can be used to calculate a community's expected death rate per pollutant per year. Table 2 ranks the utilities from most to least vulnerable, their river mile location, and the expected annual deaths attributable to carcinogens. Table 3 summarizes the vulnerability for different flow rates. As expected, the overall number of expected deaths are lower at high flow because of the greater dilution of the pollutants at 220,000 cfs. Only Huntington and Ashland reverse positions in vulner- ability at high flow. However, a few carcinogens do have downstream risk levels at high flow that exceed those at average and low flow. Through the total expected deaths are lower at high flow than at average and low flows, a few individual pollutants exhibit higher concentrations because of their disap- ------- Main Stem * Headwater Utility A Intake ± N : Utility D Intake J -» Utility F Intake *. -> 0 30 miles Utility H Intake «- L Ei Figure 2. Schema pearance rates, all average and low fl time of travel. This usefulness of mode parameters, since masked at various total expected nurr lated. Theyexpec Maysville anaCinci are greater than at Huntington, Ashla exhibit the same res pollutants that hav rates and come fror The decreased time has not allowed disappear. Thus util are at higher risk du and consequently a low flow. Utilities c would exhibit the e highest risk at Iow1 high flow, regardU rates. A great deal o between the facto concentration dis pearance rate. Fig strate the variabilit trations at utilities 1 between flow, time pearance rate. ) VJ® ^riributary ~?r Headwater J g Utility B Intake -» Utility C Intake | Utility E Intake Direction of UFlow + Utility G Intake O Headwater (^Junction -t\Point Source Loads ^Withdrawals A id tic of the study area. owing more decay at ow during the longer > result points to the ling the water quality many pollutants are flow scenarios by the ber of deaths calcu- ted death rates at nnati during high flow low flow. Utilities at id, and Ironton also ult for a few individual e high disappearance n the Kanawha River. of travel at high flow for the pollutant to ties well downstream ring high flow periods re at lower risk during ilosest to the outfalls xpected risk levels low and lowest risk at jss of disappearance f sensitivity exists rs of flow, pollutant :harged, and disap- iires 3 and 4 demon- y of pollutant concen- Dased on the tradeoffs > of travel, and disap- 4 Table 1 . Summary of Toxic Pollutants Exceeding Health Guidelines A Average Flow High Flow Low Flow 70% 700% 70% 700% 70% 700% Utility Level Level Level Level Level Level Gallipolis, None None None None Cadmium None Ohio Huntington, None None None None Cadmium None West Virgina Ashland. None None None None Cadmium None Kentucky Mercury Ironton, Cadmium None Mercury None Cadmium Mercury Ohio Mercury Greenup, Cadmium None Mercury None Cadmium Mercury Kentucky Lead Chromium Portsmouth, Cadmium None Mercury None Cadmium Mercury Ohio Mercury Maysville, Cadmium None Mercury None Cadmium Mercury Kentucky Mercury Cincinnati, Cadmium None Mercury None Cadmium Mercury Ohio Mercury Table 2. Vulnerability of Utilities to Carcinogenic Pollutants at Average Flow Vulnerability Downstream Expected Number of Utility Rank Order River Mile Cancer Deaths/ 1 00,000* Greenup 1 5 334.7 6.47 Portsmouth 2 6 355.5 5.01 Maysville 3 7 408.4 2.99 M Cincinnati 4 8 462.8 1.56 Ironton 5 4 327.0 0.20 Huntington 6 2 304.3 0.19 Ashland 7 3 319.6 0.18 Gallipolis 8 1 265.8 0.02 ^Calculated rates. Table3. Vulnerability of Utilities to Carcinogenic Pollutants at Various Flow Rates Expected Death Expected Death Expected Death Rate at Rate at Rate at Ut/lity Average Flow High Flow Low Flow Greenup 6.47(1)* 3.71(1) 19.77(1) Portsmouth 5.01(2) 3.14(2) 10.42(2) Maysville 2.99(3) 2.36(3) 2.35(3) Cincinnati 1.56(4) 1.50(4) 0.56(4) Ironton 0.20(5) 0.13(5) 0.46(5) Hungtinton 0.19(6) 0.04(7) 0.44(6) Ashland 0.18(7) 0.11(6) 0.40(7) Gallipolis 0.02(8) 0.01(8) 0.07(8) * Figures in parentheses indicate ranking. In those two figures, chlorobenzene disappearance to occur. As the time of and nitrobenzene are identically dis- travel increases, so does the amount of charged from the same industries. disappearance (thus the low concentra- However, chlorobenzene exhibits a high tions of pollutant downstream during low disappearance rate of 0.55/day, compared flow). Figure 4 reinforces this concept with 0.05/day for nitrobenzene. In Figure because the curves do not cross over at 3, one can see howthe low-flow pollutant various flow rates with the low disap- concentrations fall far belowthe average- pearance rate. The concentration of and high-flow levels. The speed at which nitrobenzene at each utility owes most of^ chlorobenzene travels downstream at its decay to the dilution, not to it^l higher flows does not allow for much disappearance rate hence the almost^ ------- 700 -, 10- O Average Flow D High Flow A Low Flow I 250 300 350 400 Ohio River Mile Point 450 500 Figure 3. Chlorobenzene concentrations at various flow rates. 100 H ^. 10- .1 1 § i o o.i- 250 300 350 400 Ohio River Mile Point 450 500 Figure 4. Nitrobenzene concentrations at various flow rates. parallel curves \n Figure 4 exhibiting the expected relationships. The low-flow curve begins to drop a bit more quickly than the high- and average-flow curves around the 450-mile point as the low disappearance rate begins to have an effect. Spill Events Accidental discharges occur in every conceivable place and manner. QUAL-II was used to route a 1-day, 60-ton spill through the case study area from its entry into the large tributary. The trade-offs between flow, magnitude of the spill, and the pollutant's disappearance rate are critical to downstream concentrations as in the daily discharge analysis. Table 4 lists the simulated concentrations of a conservative pollutant as it travels downstream at low flow. The peak concentration is not the only important statistic to be concerned with in a spill event. The length of time it takes for a spill to pass the intakes is also vital to a community's welfare. A slow-passing spill, though of lower concentration, may pose a larger problem to a utility with limited storage capacity than a very high concentration of a pollutant that passes quickly. At high flow, the spill would take only 2 days to pass Gallipolis. Average- and low- flow scenarios would require 5.5 and 14.32 days, respectively. Gallipolis has approximately 2 days of storage available and could close the intakes and not be harmed during high flow. However, at average and lowf lows, the spill requires a longer time to pass, and Gallipolis officials would need immediate and accurate information regarding the discharge to be able to decide when to close the intakes and reduce the exposure to the pollutant. Proper timing of the closure during the peak of the curve cou Id reduce the health risk immensely. A worst-case scenario would include a nondisappearing, highly toxic pollutant discharged during a low-flow period. Such a case would not be a total disaster, however, since the slow time of travel would allow ample time for downstream utilities to take precautionary measures, possibly even altering treatment techni- ques temporarily to mitigate the health hazard should storage volumes be inadequate to serve the community. Emergency conservation and public education of the situation could be instituted to stretch available supplies. A utility would be wise to have such a contingency plan developed to ensure quick and accurate implementation. ------- Conclusions This study examines a functional region that serves as the source of drinking water for more than 1.1 million people, even though it is only a portion of a watershed. The potential risk posed to communities results from the gap that exists between public health considera- tions and water pollution control stra- tegies. The main contribution of this research has been the development of an interactive systems analysis approach that can affect public policy on drinking water. Recent investigations suggest that much techno- logical manipulation of the environment produces new hazards and ameliorates old ones, and that effective means for coping with these events call for a sensitive understanding of natural phe- nomena as altered by man's actions. To study these interactions, it was necessary to collect for the first time a myriad of data and procedures and place them in an areal framework that can address prob- lems between man and his environment. The water quality simulation model, QUAL-II, was the mechanism that brought together diverse elements of mathemati- cal modeling, fluid dynamics, epidemio- logy, organic chemistry, and geography. In addition, QUAL-II has traditionally modeled only the typical parameters such as BOD, DO, temperature, etc. In this analysis, QUAL-II was used to go a step further in simulating toxic pollutants. First order decay coefficients were calculated from other sources and inserted into the model to estimate the fate of priority pollutants. Thus the issue of vulnerability is not a clear-cut matter of looking for the most downstream utility or simulating pollu- tants at an average flow. Very detailed information on flow probabilities, pollu- tant characteristics, industrial discharges, and location are needed. Table4. Priority Pollutant Spill Simulation (60-ton spill at low flow) Utility Gallipolis Huntington Ashland Ironton Greenup Portsmouth Maysville Cincinnati Arrival Time (days) 2.67 4.51 5.34 5.84 6.34 735 1035 14.00 Leave Time (days) 16.99 19.98 21.14 21.97 22.64 24.13 29.25 34.50 Days with Contamina- tion 14.32 1547 1580 16.13 1630 16.78 1890 20.50 Peak Day 7.51 10.02 11.18 11.85 12.51 13.84 1749 21.97 Peak Concentration (V9/LJ 28892 209.29 17335 169.41 165.84 149.12 136.37 111.39 The EPA authors James A. Goodrich and Robert M. Clark are with the Municipal Environmental Research Laboratory, Cincinnati, OH 45268. The complete report, entitled "Predicting Toxic Waste Concentrations in Community Drinking Water Supplies: Analysis of Vulnerability to Upstream Industrial Discharges," (Order No. PB 84-206 531; Cost: $14.50, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA authors can be contacted at: Municipal Environmental Research Laboratory U.S. Environmental Protection Agency Cincinnati, OH 45268 *USGPO: 1984-759-102-10670 ------- United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 15 Official Business Penalty for Private Use $300 0000 HsiVJLK PNiiTftTlOfM UN 5 LIdHAKY AGENLY iJ iL oOb<)<4 ------- |