Scenario Analysis of the Impact on Drinking Water Intakes from Bromide in the Discharge of Treated Oil and Gas Wastewater

Weaver, James W.; Xu, Jie; Mravik, Susan C.

doi:10.1061/(ASCE)EE.1943-7870.0000968

Open access

Technical Papers

Aug 13, 2015

Scenario Analysis of the Impact on Drinking Water Intakes from Bromide in the Discharge of Treated Oil and Gas Wastewater

Authors: James W. Weaver, A.M.ASCE [email protected], Jie Xu [email protected], and Susan C. Mravik [email protected]Author Affiliations

Publication: Journal of Environmental Engineering

Volume 142, Issue 1

https://doi.org/10.1061/(ASCE)EE.1943-7870.0000968

PDF

Abstract

Elevated levels of bromide have been shown to contribute to increased formation of disinfection byproducts (DBPs). Both produced water from unconventional oil and gas wells, which are hydraulically fractured using high volumes of fluids, and produced water from conventional oil and gas wells, which are also typically hydraulically fractured but with lower volumes of fluids, can contain high levels of bromide. If these produced waters are treated in conventional commercial wastewater treatment plants, bromide may not be removed from the effluent and is discharged to receiving water bodies. Elevated bromide levels at drinking water plant intakes is a concern for public health reasons if elevated bromide levels cause elevated levels of DBPs. This study used data from commercial wastewater treatment plants and river flow data in western Pennsylvania to construct generic discharge scenarios that illustrate the potential impacts from disposal of five classes of water that were developed from flowback and produced water bromide concentrations. Months with the historical high and low flows in the Allegheny River (Pennsylvania) and Blacklick Creek (Pennsylvania) were chosen for simulation, and treatment plant discharge rates were set at 100, 50, 33, and 25% of the permitted value for the purpose of varying the mass loading. Steady-state simulation results showed the highest probably of impact, defined as concentrations above target levels of 0.02 and

0.10 mg / L

, for produced water in the creek at both high and low flows (100%), and produced water in the river at low flows (

> 75 %

). High probability of impact (

> 50 %

) occurred in the river at low flows and all flows in the creek with treated mixed/flowback water discharge. Modeled reduction in the effluent discharge rate reduced downstream impacts proportionally. Transient simulation showed that transient peak concentrations may exceed time-averaged concentration by up to a factor of four when mixing conditions are met.

Introduction

The use of high-volume hydraulic fracturing and horizontal drilling has opened up low-permeability shale, including the Marcellus Shale, and tight sand reservoirs for oil and gas extraction (Vengosh et al. 2014). Beginning in 2005, over 7,000 shale gas wells were added to the more than 34,000 producing oil and gas wells in Pennsylvania (Brantley et al. 2014; Vidic et al. 2013; Vengosh et al. 2014). In hydraulic fracturing, water is used to raise the downhole pressure above the level required to exceed the fracture pressure of the formation (Gregory et al. 2011; Entrekin et al. 2011). In addition to water, chemicals are added for various purposes (friction reduction, iron control, bacterial control, oxygen scavengers, and others) and proppants are used to maintain porosity as pressure is reduced (Vidic et al. 2013). After hydraulic fracturing, pressure is released, and some fraction of the injected fluid flows from the well as produced water. In the early stages, usually the first few weeks after hydraulic fracturing is complete, this produced water is called flowback (Gregory et al. 2011). As time goes on, the water flowing from the well resembles more and more the formation water and is called produced water (Haluszczak et al. 2013). Depending on the formation, produced water can have high concentrations of total dissolved solids, chloride, bromide, and other constituents. Marcellus Shale produced water has been reported as being among the highest in salinity in the United States (Vidic et al. 2013; Vengosh et al. 2014). Historically, flowback and other produced water were treated as waste materials, although reuse is possible under appropriate circumstances (Boschee 2014; Vengosh et al. 2014). Most wastewater from oil and gas production in the United States has been disposed of in injection wells (Clark and Veil 2009), although this option is limited for western Pennsylvania due to limited investment in disposal well construction. Some wastewater from oil and gas operations has been treated in both publically owned treatment works (POTWs) and commercial wastewater treatment plants (CWTPs) in western Pennsylvania. The practice of disposing of wastewater from oil and gas operations in POTWs was reduced in response to a request from the Pennsylvania Department of Environmental Protection (Ferrar et al. 2013; U.S. EPA 2011a). Some CWTPs in this area were designed to treat and permitted to treat this wastewater, but they did not usually remove total dissolved solids including bromide (Ferrar et al. 2013; Soeder and Kappel 2009; Wilson et al. 2013).

Disinfection Byproducts

Under the Stage 1 Disinfectants and Disinfection Byproducts Rule, the United States EPA established maximum contaminant levels (MCLs) for total trihalomethanes (TTHM) at

0.080 mg / L

, total haloacetic acids at

0.060 mg / L

, and bromate (

{BrO}_{3}^{-}

) at

0.010 mg / L

(U.S. EPA 2001). Maximum contaminant level goals (MCLG) were set for brominated TTHMs: zero for bromodichloromethane, bromoform, and bromate; and

0.06 mg / L

for dibromochloromethane. Halogens reacting with organic matter from source water are the cause of disinfection byproducts (U.S. EPA 2001). Bromide, pH, concentration, and reactivity of organic matter, free chlorine concentration, temperature, contact time, and other factors influence disinfection byproduct formation in treated drinking water (Krasner et al. 1989; Sohn et al. 2006; Obolensky and Singer 2008; Plewa et al. 2008; Roccaro et al. 2008; Brown et al. 2011; Sharma et al. 2014). Elevated bromide in source water has the potential to influence the formation of brominated DBPs and can increase a suite of recently identified unregulated compounds, which include halonitromethanes, haloamides, haloacetronitriles, and other classes of compounds (Krasner 2006; Richardson et al. 2007; Pressman et al. 2010). Increased bromide concentrations can also shift the TTHM production to favor brominated compounds (McClain et al. 2002; Sohn et al. 2006; Richardson et al. 2007; States et al. 2013), which are generally viewed as being more toxic than chlorinated compounds (McTigue et al. 2014; Richardson et al. 2007).

Bromate can be produced in treated drinking water by ozonation or chlorine dioxide treatment. Bromate concentrations above the MCL were reported for samples that had bromide above

0.050 mg / L

., ozone doses above

1.5 mg / L

, alkalinity above 60 mg

{CaCO}_{3} / L

, or TOC above

2 mg / L

(Moll and Krasner 2002).

To support regulatory development, on May 14, 1996, the U.S. EPA promulgated an information collection rule (ICR) for cryptosporidium, giardia, viruses, and disinfection byproducts in drinking water, which required monitoring at large public water systems (Federal Register 1996). DBP precursors, specifically bromide and total organic carbon (TOC), were also included in the required monitoring. Bromide and other inorganic ions were analyzed by U.S. EPA method 300.0, with a method detection limit of

0.010 mg / L

for bromide (U.S. EPA 1996). A method reporting limit (MRL) of

0.020 mg / L

for bromide was selected so that (1) most laboratories could meet the precision and accuracy criteria under normal operating conditions; (2) most samples would have concentrations above the MRL; and (3) values less than the MRL were assumed not to be of health significance (Fair et al. 2002).

Various researchers have reported that low bromide concentrations can cause the formation of brominated DBPs in finished drinking waters, although a specific threshold level has not been identified (Cooper et al. 1985; Chang et al. 2001; Obolensky and Singer 2008; Brown et al. 2011). McGuire and Graziano (2002) summarized and analyzed finished water TTHM and source water bromide concentrations from the ICR data, finding that mean TTHM concentrations increased with increasing TOC levels for all source water bromide concentrations between the MRL of 0.020 and

0.10 mg / L

. McGuire and Graziano (2002) attributed reduced TTHM concentrations above

0.10 mg / L

bromide and

3 mg / L

TOC to treatment plants taking steps (with associated costs) to control TTHM formation. Bromide in its unreacted form has no known ingestion health effects and is thus unregulated (McTigue et al. 2014). Consequently, two target levels were selected for discussing the simulation results that follow. The first target level was selected from the ICR criteria listed previously, principally that most laboratories could routinely meet precision and accuracy requirements above

0.02 mg / L

, and, further, the McGuire et al. (2002) report of increased DBP concentrations with increasing TOC above

0.02 mg / L

. The second target level follows a concern for increased DBP formation reported by some water utilities for bromide at

0.10 mg / L

or higher (Bonacquisti 2006; States et al. 2013).

Bromide in Oil and Gas Wastewater

Bromide in Marcellus Shale produced water originates from evaporated seawater (Haluszczak et al. 2013). Water injected for hydraulic fracturing is typically low in TDS and bromide, particularly when freshwater is used as the base liquid. Because of contact with formation water and production of formation water (Haluszczak et al. 2013), the salinity and bromide concentration of flowback water typically increases with time.

Wilson et al. (2013) found no statistical differences in median bromide concentration in produced water from Marcellus Shale and conventional gas wells. Further, they found no significant difference in median bromide concentration between oil and gas produced water (Marcellus shale gas wells, conventional gas wells, and conventional oil wells) and treated Pennsylvania brine plant effluent. These two results suggest that the produced water from Marcellus Shale gas wells is not distinguishable on the basis of median bromide concentration from that of conventional wells and that the treatment plants do not remove bromide from the wastewater. Similarly, Warner et al. (2013) reported mean bromide for produced water from the Lower Devonian, Upper Devonian, and Marcellus formations to be 1,283, 787, and

744 mg / L

, respectively, while treated CWTP discharge averaged

643 mg / L

with standard deviation of

201 mg / L

. Hayes (2009) and Haluszczack et al. (2013) found similar ranges for flowback:

16 - 1,190 mg / L

, and nondetect to

613 mg / L

, respectively. Ferrar et al. (2013) sampled effluent from two POTWs and one CWTP (Outfall B discussed subsequently) before and after the state of Pennsylvania requested treatment plants stop accepting shale gas wastewater (PA DEP 2011). Effluent concentrations dropped from the POTWs. In particular, the bromide concentrations dropped to

0.1 mg / L

or lower (with the exception of one early time data point of

0.43 mg / L

at one plant). For the CWTP, the initial bromide effluent concentration range (

973 - 1,100 mg / L

) was reduced to

602 - 744 mg / L

in late 2011.

Bromide in Pennsylvania Surface Waters

Concentrations of bromide in surface water from Pennsylvania counties with hydraulically fractured shale-gas wells ranged from 0.01 to

100 mg / L

from mid-2009 through late 2011 (Vidic et al. 2013). Most concentrations were between 0.03 and

2 mg / L

, and the higher values occurred in 2011. Bromide was detected at a concentration of

75 mg / L

near the outfall of a CWTP in Pennsylvania. Mean bromide concentrations decreased from

138 mg / L

within 20 m of the plant to

0.52 mg / L

at a distance beyond 300 m downstream (Hladik et al. 2014).

In the Monongahela River, high levels of total dissolved solids and sulfate were reported in 2008 and corresponded to increased levels of brominated DBPs at drinking water plants (Wilson and Van Briesen 2013). Wilson and Van Briesen (2013) found that low-flow conditions in 2010 were associated with increased bromide concentration at drinking water intakes for six drinking water plants. In 2012, discharges from shale gas wastewater treatment plants were reduced, but similar low-flow conditions did not produce elevated bromide concentrations. Using their estimated loadings, Wilson and Van Briesen (2013) predicted some bromide concentrations would exceed

0.010 mg / L

over the recorded range of flows, but the concentrations would be reduced by orders-of-magnitude when flows were above minimum levels.

In response to 2010 increases in DBPs in Pittsburgh’s finished drinking water, States et al. (2013) sampled water in the Allegheny River to determine background bromide levels and to determine sources of additional bromide. They determined that the likely sources were coal-fired power plants, steel mills, Marcellus Shale wastewater-treatment plants and coal mine drainage, and associated the 2011 elevated bromide levels with low-flow conditions in the river. They attributed increases in bromide concentration at the drinking water intakes to discharges from three CWTPs and coal-fired power plants, with the CWTPs accounting for roughly 50% of the increase.

The purpose of this work is to estimate, in a generic sense, downstream bromide concentrations based on scenarios developed from stream flow and CWTP effluent data. The scenarios contained two major elements. First, the available data from CWTPs were used to generate concentration frequency distributions for produced water with and without Marcellus Shale water, flowback, a mixture of flowback other produced water, and a lower bromide water. Under separate scenarios the release of each type of water from the CWTP locations was modeled to determine the impact of various operations on the receiving water body. CWTPs are not the only source of bromide in the watersheds where they operate; coal mine drainage and coal-fired power plant wastes also contribute bromide to these watersheds. The simulations used only the CWTP discharge to isolate the potential impact from disposal of treated oil and gas production wastewater. Second, the historical flow record for the Allegheny River and Blacklick Creek were used to simulate both low-flow and high-flow conditions. Variations in all the parameters were examined using Monte Carlo methods. The scenarios were simulated by steady-state flows and discharges, and because the plants might not operate continuously, with pulsed inputs of treated effluent.

Data Sources

Data to define the scenarios were compiled from effluent discharges from CWTPs in western Pennsylvania, receiving river flows at the point of discharge, and data on flow accretion with downstream travel distance. The reported plant effluent discharges and USGS gauge information on flows in the Allegheny River and Blacklick Creek are given in Table 1. The frequency distribution of flows during the average low-flow and high-flow months (August and March) were used in the scenarios that follow.

Table 1. Permitted Discharges for All Outfalls Reporting Data and USGS Gauge Data Used to Develop Scenarios

Outfall permit data		USGS gauge	USGS gauge data
Outfall	Permitted discharge (MGD)		Number of months of record	Minimum monthly discharge $m^{3} / s$ (date)	Maximum monthly discharge $m^{3} / s$ (date)	Observed receiving-body discharge distributions ( $m^{3} / s$ )
Outfall	Permitted discharge (MGD)		Number of months of record	Minimum monthly discharge $m^{3} / s$ (date)	Maximum monthly discharge $m^{3} / s$ (date)	Month	Minimum	Median	Maximum
B	0.155	03042000	720	0.82 ( $9 / 1998$ )	45.7 ( $3 / 1967$ )	August	1.0	2.8	16.6
						All	0.82	8.4	45.7
						March	6.20	21.1	45.7
C/E	0.045/0.018	03038000	900	0.17 ( $9 / 1952$ )	37.9 ( $3 / 1994$ )	September	0.17	1.4	30.6
						All	0.17	6.2	37.9
						March	2.4	16.1	37.9
D	0.3	03025500	863	16.1 ( $10 / 1963$ )	1370 ( $3 / 1945$ )	August	21.0	64.6	391
						All	16.1	272	1,370
						March	181	552	1,370

Commercial Treatment Plant Effluent

Five CWTPs voluntarily submitted effluent data, which included the discharge rate and bromide concentration from six outfalls to surface waters in western Pennsylvania for the period from February to December 2011 (U.S. EPA 2011e), and are denoted A–F in the following discussion. Three of the plants also submitted additional data in response to information requests issued by EPA pursuant to Section 308 of the Clean Water Act (U.S. EPA 2011c, b, d). The NPDES permits for these plants required monitoring and reporting of bromide concentrations, but did not impose an effluent limit. After September 2011, the plants stopped treating Marcellus Shale produced water, although they continued to accept other oil and gas produced water (Hart Resources 2011). Subsequent administrative settlements with EPA require the installation of controls what will reduce effluent concentrations from these plants (U.S. EPA 2013); therefore the bromide frequency distributions described subsequently represent historic practices.

Laboratory analysis reports were included with the data and identified the analytical methods, detection limits, dilution factors, and other information. Taken together this information was used to assess the quality of the data. Generally one sample was collected and analyzed per month, however, in some cases up to four samples were collected and analyzed.

The bromide concentrations from the outfalls ranged from 3.4 to

8,290 mg / L

. As no single outfall matched this span of observed bromide concentrations, frequency distribution curves were developed for the individual outfalls (Fig. 1). Plant A had the three lowest effluent bromide concentrations in the dataset, but supplied no information on the volumetric discharge (Fig. 1). One of the plants used separate treatment processes for flowback water (Outfall C) and other produced water (Outfall E) and monitored each effluent separately (U.S. EPA 2011d). This plant then combined Outfalls C and E with a stipulated maximum amount of stormwater prior to discharging to the receiving creek. For this paper, the concentration of the combined discharge was estimated as the flow-weighted average for the bounding cases of no stormflow and the maximum permitted stormflow (

59.1 m^{3} / d

or 15,196 gal. per day). The resulting estimated concentrations of bromide from the combined outfalls range from 325 to

1,970 mg / L

with no contribution from stormflow, and

260 - 1,590 mg / L

with the maximum permitted stormflow (Fig. 1). The concentrations from this plant’s two outfalls and literature data on flowback and produced water were used as described in a subsequent section to define characteristic effluent concentrations for flowback, other produced water and mixed Marcellus Shale and non-Marcellus Shale produced water.

Fig. 1. Bromide concentration frequency from outfalls with NPDES-reported data. The composite frequency distribution from all plants is shown as a solid line without symbols. Outfall A (dot-dash) data used to represent lower bromide effluent. Outfall C (triangles) and Outfall E (squares) discharged treated flowback and produced water, respectively. These two were monitored separately but were combined at the point of discharge, with estimates of either no storm water (short dash) or with a design maximum of $59.1 m^{3} / s$ (long dash). Outfall D (circles) discharged similar concentrations to the combined flowback and produced water of Outfalls C and E. Outfall B (diamonds) discharged similarly to the treated produced water from Outfall E. Outfall F (dot-dot-dash) represents similar effluent as Outfall C and was not further used in simulation

The produced water outfall (E) had reported effluent bromide concentrations ranging from 766 to

6,630 mg / L

. The highest and lowest concentrations, however, occurred after the plant stopped accepting Marcellus Shale produced water in September 2011. Outfall B bromide concentrations covered a similar range, but the two highest values (4,620 and

8,290 mg / L

) for this outfall also were reported after September 2011. Fig. 1 shows the bromide frequency distribution curves for data collected prior to voluntary removal of Marcellus wastewater, while the entire data set is shown in Fig. S1. The bromide concentration from the flowback water treatment train (Outfall C) ranged from 63 to

373 mg / L

and was considered to represent flowback concentrations. The range of bromide concentrations from Outfall D (

91 - 1,040 mg / L

) indicated treatment of a mixture of flowback and other produced water, based on comparison with the estimated combined effluent from Outfalls C and E (Fig. 1).

From these data, five categories of effluent bromide concentrations were developed and used in the following simulations (Table 2). Because of the use of hydraulic fracturing in the Marcellus Shale, continuing conventional oil and gas production in Pennsylvania, and the similarity of produced water from conventional and unconventional production wells, CWTP effluent may represent a mixture of produced water from conventional and unconventional hydraulically fractured formations. This situation was simulated for the complete produced water effluent, denoted by Produced-C (Table 2). Given the sources of data available for this study, differentiating between conventional and unconventional produced water is not generally possible. For produced water treated after the exclusion of the Marcellus Shale produced water, however, the effluent was denoted Produced-ME. The range of concentration for the mixed flowback and other produced water, as based on the combined Outfall C and E data, overlaps the range for Marcellus Shale flowback reported by Hayes (2009) (

16 - 1,190 mg / L

). Thus the concentration range denoted Mixed/Flowback represents the mixed flowback and other produced water of the outfalls and the flowback reported by Hayes (2009). Lower Flowback denotes effluent concentration from Outfall C which was designed to treat flowback (

63 - 373 mg / L

). Lower Bromide denotes the bromide concentrations reported for Outfall A (

3.4 - 19.7 mg / L

).

Table 2. Five Categories of Bromide Concentration Input Used in Model Simulations

Data source	Short name	Effluent bromide concentration ( $mg / L$ )
Data source	Short name	Minimum	Median	Maximum
Complete Outfall B	Produced-C	1,100	2,370	8,290
Outfall B prior to 9/2011, Marcellus exclusion	Produced-ME	1,110	2,360	3,600
Outfall D/Hayes (2009) flowback	Mixed/flowback	82.8	530	1,040
Outfall C	Lower flowback	63	146	373
Outfall A	Lower bromide	3.4	4.6	19.7

Flow Accretion

Because the simulations were intended to determine conditions where the discharge of treated oil and gas production wastewater might increase bromide concentrations at downstream water treatment plant intakes, the change in flow along the rivers was simulated from tracer study data. Results from dye-release experiments with measured flows between sampling stations were used to determine the increase or decrease in flow per meter of reach length (Jobson 1996; Nordin and Sabol 1974). Data from a total of 102 dye-release experiments were used to generate normalized increments of discharge per meter of flow distance, which varied with discharge (Fig. S2). Lower average flows had greater potential for change than higher average flows. Thus, at apparent breaks on the scatter plot shown in Fig. S2, the changes in flow were divided into three groups: less than

25 m^{3} / s

,

25 - {140 m}^{3} / s

, and greater than

140 m^{3} / s

(Fig. 2).

Models

Both steady-state and transient releases were simulated. Steady-state simulations were developed to present an upper-bound release. To augment the steady-state release scenarios, transient scenarios were developed with pulsed inputs to the receiving body.

Steady-State Model

The steady-state model is based on mass and volume flow balance in a river network. Because of its conservative nature, bromide is treated as a tracer. Because flows vary along the river network, the river was divided into reaches with different lengths and flows. The concentration in each reach is determined by applying a dilution factor in proportion to the ratio of upstream to downstream flows. If the stream gains flow, concentrations are reduced; losing river reaches remove mass from the river without altering concentration (e.g., Asher et al. 2006).

To capture the variation in flows and bromide concentrations, Monte Carlo simulation with 10,000 individual model runs was used. Inputs were generated from the frequency distributions for the flow, effluent bromide concentration, and flow accretion as described previously. A single value of each input parameter was drawn from the input frequency distributions and used in a model run. With a sufficient number of individual runs of the model, the entire distribution of each input variable is included. Each simulation represents steady-flow and transport with complete mixing from the source to the receptor. Latin Hypercube Sampling is used to reduce the number of individual model runs needed (McKay et al. 1979), as it forces the extreme ends of the distributions to be included in the simulations. Regardless, tests of the model showed that 10,000 simulations were needed for adequate representation of the entire input frequency distributions.

Transient Model

A semi-analytical, one-dimensional model was used for transient transport analysis. Each individual release to the network is broken into a series of small mass inputs, where each is represented by a triangular concentration distribution (Kilpatrick 1993; Kilpatrick and Taylor 1986). The triangular distributions are summed to represent the downstream concentration distribution in the river. The regression formulas developed by Jobson (1996, 1997) described in the appendix are used to parameterize the model with approximate travel times and longitudinal spreading. These data are drawn from tracer experiments conducted around the United States and mostly represent humid climates. This compilation was chosen because the range of flows contained in the dataset cover the expected flows in the scenarios, the data are mostly drawn from humid zones in the U.S., and a regression formula for peak velocity was presented (Jobson 1996, 1997).

Two modifications to the approach developed by Kilpatrick (1983) were made: allowance for properties to vary from reach-to-reach in river networks and inclusion of uncertainty analysis. First, because the regression equations were originally developed for application to a uniform reach, a procedure was implemented to allow the calculations to be performed over a series of connected reaches (Figs. S3 and S4). The time-varying concentration is determined for the downstream end of the reach. The resulting set of concentrations is applied as input to the next downstream reach. Because the flows can differ between the reaches, the input concentrations are adjusted by the ratio of the flow in the two reaches

c_{down} (t) = \frac{Q_{up}}{Q_{down}} c_{up} (t)

(1)

where

c (t)

are the time-varying concentrations (

M / L^{3}

);

Q

are the discharges (

L^{3} / T

) and the subscripts refer to up and down stream reaches, respectively. When the upstream and downstream flow rates are different, this constraint imposes a discontinuity in concentration as the downstream concentrations are abruptly decreased to the required levels. This suggests that the river should be broken into as many segments as reasonable, so that the difference in flows between segments is about 10%.

Second, uncertainty analysis is needed because of variability in parameters, including scatter in the data underlying the regressions. The parameters of the expression for peak velocity are developed into a frequency distribution (Figs. S5 and S6). The model was applied deterministically to tracer concentration data from the Nordin and Saybol (1974) tabulation (Fig. 6 of the Appendix, Tables S1 and S2) and shown to match the tracer data and a calibrated water quality analysis simulation package (WASP) simulation (Ambrose et al. 1983, 2009). Uncertainty analysis was used for five additional experiments from Nordin and Saybol (1974) to demonstrate that the experimental data lie within the uncertainty bounds of the model for data used in developing the regressions (Table 7 of the Appendix and Figs. S7–S10). A tracer experiment not used in developing the regressions was simulated with similar result (Table 8 of the Appendix and Figs. S11 and S12). Regardless of these comparisons, the application of the regression formulas is an extrapolation as the specific waterways used in the scenarios were not included in the dataset. The purpose of using this approach is to estimate transport in hypothetical scenarios where, by the nature of the scenario analysis, calibration data are unavailable (Table 9).

Yotsukura and Cobb (1972) developed an optimal distance,

L_{o}

, for transverse mixing based on the average velocity, width and the transverse mixing coefficient from

L_{o} = K \frac{v W^{2}}{E_{z}}

(2)

where

K

= coefficient depending on the configuration of the discharge;

v

= mean velocity (L/T);

W

= channel width; and

E_{z}

= transverse mixing coefficient (

L^{2} / T

). Fischer et al. (1979) and Kilpatrick and Wilson (1989) presented values of

K

for edge (0.4), centerline (0.1), center of each half of flow (0.025), and center of each third of flow (0.011). Prior to

L_{o}

, the movement of the tracer cloud does not represent the average flow in the river and concentration distributions are skewed (Yotsukura and Cobb 1972; Fischer et al. 1979; Martin and McCutcheon 1999).

Scenarios

Construction of the steady-state scenarios began with the monthly discharge data. Based on the entire period of record, the median flow in Blacklick Creek was

8.4 m^{3} / s

, and the median flow in the Allegheny River was

272 m^{3} / s

. These were chosen for scenarios because they have two orders-of-magnitude difference in median flow, and both had permitted discharges from the outfalls (B and D) discussed previously. Variation in river/creek flow was included by simulating the entire period of record for the months with the highest (March) and lowest mean flows (August). Comparing the results for these two months gives the expected bounding bromide concentration values. The incremental flow accretion frequency curve (Fig. 2) was applied to each simulation so that all reaches gained or lost flow, based on the random selection from the flow accretion frequency distribution. The CWTP data were used to generate operational definitions for types of water (Table 2). Concentration distributions for these waters and the permitted discharge rates were used in the scenarios. Because the effluent discharge for each outfall was fixed at its permitted level, a separate set of simulations were performed in which the bromide concentration Mixed/Flowback with its range of bromide concentration from 82 to

1,040 mg / L

, was used along with 100, 50, 33, and 25% of the permitted discharge rate to vary the mass loading. A summary of all scenarios is given in Table S3.

To augment the steady-state results, transient scenarios were constructed for releases that lasted for 24, 12, 8, and 4 h per day during a five-day working week. The effluent discharge rate was again fixed at the level permitted for the outfall, and the effluent bromide concentrations were chosen to represent Produced-ME, Mixed/Flowback, and Lower Flowback using the same concentration frequency distributions as the steady-state simulations. The scenarios were developed for releases to the Allegheny River and transport of 127 km from Franklin to Kittanning, Pennsylvania (Tables 3, S4, and S5). Simulations were made for both the 5th and 95th percentile bromide concentrations determined from the produced water input concentration data (Produced-ME). USGS gauge data on mean monthly flows were used to characterize the flow at Franklin, Parker, and Kittanning, Pennsylvania. Flows were estimated for a former station with incomplete data at Rimer, Pennsylvania, by linear interpolation of the frequency distributions at Parker and Kittanning. Similar to the steady-state simulations, variability in river flow, travel time, and dispersion were directly simulated by Monte Carlo-Latin Hypercube simulation for the low-flow and high-flow months (August and March). The flows at each location along the length of the river were drawn from their respective frequency distributions, based on a randomly selected cumulative probability from 0.0 to 1.0 applied to each station.

Table 3. Input Parameters for the Transient Simulations in the Allegheny River

Location	USGS gauge	River mile (km upstream from Pittsburgh)	Upstream drainage area ( ${km}^{2}$ )	Slope (m/m)	Mean annual discharge ( $m^{3} / s$ )	Number of months of record
Franklin, Pennsylvania	3025500	200.2	15,493	$5.66 \times 10^{- 4}$	313	864
Parker, Pennsylvania	3031500	134.2	19,868	$5.11 \times 10^{- 4}$	393	960
Rimer, Pennsylvania	3033000	96.9	21,730	$1.11 \times 10^{- 4}$	432^a	N/A
Kittanning, Pennsylvania	3036500	73.7	23,240	$7.67 \times 10^{- 4}$	456	996

a

Estimated in proportion to the distance between Kittinning, Pennsylvania, and Parker, Pennsylvania.

Results and Discussion

Steady-State Simulations

The modeled effluent concentrations for each type of water (Table 2) produce characteristic ranges of concentration at downstream locations for the river scenario (Fig. 3, Tables 4–6). In the figure each result is represented by the median concentration from the 10,000 Monte Carlo runs for the mean low-flow (August) and mean high-flow (March) months of the year. The initial concentrations reflect potential end of pipe dilution based on the river or stream flow (Table 4). These concentrations are not likely to be achieved immediately because of initial mixing effects (see discussion to follow), but provide a metric that relates magnitude of the effluent discharge and receiving-body flows. The difference in flows between the low-flow and high-flow months results in an order of magnitude range of bromide concentration over the year for a specified effluent type (Table 4 and Fig. 3). The median simulated in-stream bromide concentration decreased in proportion to the decrease in median effluent concentration (Table 2), so, as expected, greater potential impacts occur with higher input levels of bromide. The impact of including the non-Marcellus data points in the produced water effluent (Produced-ME), was mostly to influence the maximum (4.97 versus 2.24 for low flow in the river) and upper end of the frequency distribution curve, because the non-Marcellus data points were only the two highest values. As a measure of the spread of the distributions the interquartile range divided by the median was fairly consistent for each group of simulations (Table 4).

Fig. 3. Median bromide concentrations in steady-state river scenario simulation results for four effluents: Produced-ME, Mixed/Flowback, Lower Flowback, and Lower Bromide. Produced-C omitted as the medians over plot the Produced-ME results. The concentration ranges are indicated between low values in the high-flow month (March) and high values in the low-flow month (August). Light shading indicates concentrations between 0.02 and 0.1 mg/L; dark shading indicates concentrations below 0.02 mg/L

Table 4. Modeled Bromide Concentrations Considering Mixing of Effluent with Initial Flow Only

Type^a	Minimum	Percentiles					Maximum	IQ/median^b
Type^a	Minimum	5th	25th	Median	75th	95th	Maximum	IQ/median^b
River scenario by effluent type, low flow^c
Produced-C	$3.65 \times 10^{- 2}$	$8.88 \times 10^{- 2}$	$2.18 \times 10^{- 1}$	$3.70 \times 10^{- 1}$	$5.78 \times 10^{- 1}$	$1.27 \times 10^{0}$	$4.97 \times 10^{0}$	0.97
Produced-ME	$3.87 \times 10^{- 2}$	$9.16 \times 10^{- 2}$	$2.32 \times 10^{- 1}$	$3.64 \times 10^{- 1}$	$5.22 \times 10^{- 1}$	$1.09 \times 10^{0}$	$2.24 \times 10^{0}$	0.80
Mixed/flowback	$3.45 \times 10^{- 3}$	$2.11 \times 10^{- 2}$	$5.64 \times 10^{- 2}$	$9.18 \times 10^{- 2}$	$1.53 \times 10^{- 1}$	$3.16 \times 10^{- 1}$	$6.33 \times 10^{- 1}$	1.06
Lower flowback	$2.17 \times 10^{- 3}$	$5.06 \times 10^{- 3}$	$1.24 \times 10^{- 2}$	$2.30 \times 10^{- 2}$	$3.92 \times 10^{- 2}$	$8.72 \times 10^{- 2}$	$2.15 \times 10^{- 1}$	1.17
Lower bromide	$1.16 \times 10^{- 4}$	$2.18 \times 10^{- 4}$	$5.81 \times 10^{- 4}$	$9.57 \times 10^{- 4}$	$1.98 \times 10^{- 3}$	$4.76 \times 10^{- 3}$	$1.19 \times 10^{- 2}$	1.47
River scenario by effluent type, high flow^c
Produced-C	$1.06 \times 10^{- 2}$	$2.16 \times 10^{- 2}$	$3.77 \times 10^{- 2}$	$5.44 \times 10^{- 2}$	$8.45 \times 10^{- 2}$	$1.48 \times 10^{- 1}$	$4.39 \times 10^{- 1}$	0.86
Produced-ME	$1.05 \times 10^{- 2}$	$2.33 \times 10^{- 2}$	$3.92 \times 10^{- 2}$	$5.37 \times 10^{- 2}$	$7.93 \times 10^{- 2}$	$1.12 \times 10^{- 1}$	$2.39 \times 10^{- 1}$	0.75
Mixed/flowback	$8.71 \times 10^{- 4}$	$5.53 \times 10^{- 3}$	$9.51 \times 10^{- 3}$	$1.45 \times 10^{- 2}$	$2.04 \times 10^{- 2}$	$3.54 \times 10^{- 2}$	$7.05 \times 10^{- 2}$	0.75
Lower flowback	$6.03 \times 10^{- 4}$	$1.21 \times 10^{- 3}$	$2.19 \times 10^{- 3}$	$3.47 \times 10^{- 3}$	$5.55 \times 10^{- 3}$	$9.61 \times 10^{- 3}$	$2.27 \times 10^{- 2}$	0.97
Lower bromide	$3.25 \times 10^{- 5}$	$5.58 \times 10^{- 5}$	$9.12 \times 10^{- 5}$	$1.52 \times 10^{- 4}$	$2.83 \times 10^{- 4}$	$5.91 \times 10^{- 4}$	$1.32 \times 10^{- 3}$	1.26
Creek scenario by effluent type, low flow^c
Produced-C	$4.49 \times 10^{- 1}$	$1.01 \times 10^{0}$	$2.74 \times 10^{0}$	$5.25 \times 10^{0}$	$9.10 \times 10^{0}$	$1.59 \times 10^{1}$	$4.86 \times 10^{1}$	1.21
Produced-ME	$4.58 \times 10^{- 1}$	$1.07 \times 10^{0}$	$2.80 \times 10^{0}$	$5.25 \times 10^{0}$	$8.75 \times 10^{0}$	$1.23 \times 10^{1}$	$2.16 \times 10^{1}$	1.13
Mixed/flowback	$4.11 \times 10^{- 2}$	$2.48 \times 10^{- 1}$	$6.94 \times 10^{- 1}$	$1.38 \times 10^{0}$	$2.17 \times 10^{0}$	$4.00 \times 10^{0}$	$6.65 \times 10^{0}$	1.07
Lower flowback	$2.62 \times 10^{- 2}$	$5.62 \times 10^{- 2}$	$1.64 \times 10^{- 1}$	$3.21 \times 10^{- 1}$	$5.65 \times 10^{- 1}$	$1.05 \times 10^{0}$	$2.18 \times 10^{0}$	1.25
Lower bromide	$1.39 \times 10^{- 3}$	$2.41 \times 10^{- 3}$	$7.58 \times 10^{- 3}$	$1.48 \times 10^{- 2}$	$2.65 \times 10^{- 2}$	$6.54 \times 10^{- 2}$	$1.20 \times 10^{- 1}$	1.28
Creek scenario by effluent type, low flow^c
Produced-C	$1.66 \times 10^{- 1}$	$2.96 \times 10^{- 1}$	$5.11 \times 10^{- 1}$	$7.58 \times 10^{- 1}$	$1.12 \times 10^{0}$	$2.35 \times 10^{0}$	$8.42 \times 10^{0}$	0.81
Produced-ME	$1.70 \times 10^{- 1}$	$3.38 \times 10^{- 1}$	$5.45 \times 10^{- 1}$	$7.46 \times 10^{- 1}$	$1.04 \times 10^{0}$	$1.86 \times 10^{0}$	$3.69 \times 10^{0}$	0.66
Mixed/flowback	$1.40 \times 10^{- 2}$	$7.73 \times 10^{- 2}$	$1.33 \times 10^{- 1}$	$1.94 \times 10^{- 1}$	$2.95 \times 10^{- 1}$	$5.18 \times 10^{- 1}$	$1.12 \times 10^{0}$	0.84
Lower flowback	$9.43 \times 10^{- 3}$	$1.68 \times 10^{- 2}$	$2.97 \times 10^{- 2}$	$4.90 \times 10^{- 2}$	$7.68 \times 10^{- 2}$	$1.43 \times 10^{- 1}$	$3.84 \times 10^{- 1}$	0.96
Lower bromide	$5.15 \times 10^{- 4}$	$7.93 \times 10^{- 4}$	$1.28 \times 10^{- 3}$	$2.09 \times 10^{- 3}$	$4.01 \times 10^{- 3}$	$7.95 \times 10^{- 3}$	$2.13 \times 10^{- 2}$	1.30

Note: Concentrations between 0.02 and $0.10 mg / L$ are bolded; concentrations of $0.02 mg / L$ or less are italicized.

a

Effluent bromide concentration ranges defined in Table 2.

b

IQ/median = interquartile range/median.

c

High-flow and low-flow conditions are simulated with the historical frequency distributions for the months with the highest and lowest average flows, respectively.

Table 5. Modeled Bromide Concentrations at 50 km Downstream from Source

Type^a	Minimum	Percentiles					Maximum	IQ/median^b
Type^a	Minimum	5th	25th	Median	75th	95th	Maximum	IQ/median^b
River scenario by effluent type, low flow^c
Produced-C	$4.76 \times 10^{- 3}$	$5.47 \times 10^{- 2}$	$1.72 \times 10^{- 1}$	$3.34 \times 10^{- 1}$	$5.52 \times 10^{- 1}$	$1.26 \times 10^{0}$	$4.97 \times 10^{0}$	1.14
Produced-ME	$5.34 \times 10^{- 3}$	$5.50 \times 10^{- 2}$	$1.76 \times 10^{- 1}$	$3.31 \times 10^{- 1}$	$5.10 \times 10^{- 1}$	$1.09 \times 10^{0}$	$2.24 \times 10^{0}$	1.01
Mixed/flowback	$9.03 \times 10^{- 4}$	$1.28 \times 10^{- 2}$	$4.37 \times 10^{- 2}$	$8.39 \times 10^{- 2}$	$1.47 \times 10^{- 1}$	$3.16 \times 10^{- 1}$	$6.33 \times 10^{- 1}$	1.23
Lower flowback	$2.89 \times 10^{- 4}$	$2.97 \times 10^{- 3}$	$1.01 \times 10^{- 2}$	$2.01 \times 10^{- 2}$	$3.77 \times 10^{- 2}$	$8.72 \times 10^{- 2}$	$2.15 \times 10^{- 1}$	1.37
Lower Bromide	$1.62 \times 10^{- 5}$	$1.31 \times 10^{- 4}$	$4.73 \times 10^{- 4}$	$8.59 \times 10^{- 4}$	$1.87 \times 10^{- 3}$	$4.76 \times 10^{- 3}$	$1.19 \times 10^{- 2}$	1.63
River scenario by effluent type, high flow^c
Produced-C	$1.31 \times 10^{- 3}$	$5.85 \times 10^{- 3}$	$1.63 \times 10^{- 2}$	$3.39 \times 10^{- 2}$	$6.90 \times 10^{- 2}$	$1.29 \times 10^{- 1}$	$4.39 \times 10^{- 1}$	1.55
Produced-ME	$1.31 \times 10^{- 3}$	$6.12 \times 10^{- 3}$	$1.65 \times 10^{- 2}$	$3.25 \times 10^{- 2}$	$6.76 \times 10^{- 2}$	$1.10 \times 10^{- 1}$	$2.39 \times 10^{- 1}$	1.58
Mixed/flowback	$1.18 \times 10^{- 4}$	$1.48 \times 10^{- 3}$	$4.24 \times 10^{- 3}$	$8.82 \times 10^{- 3}$	$1.70 \times 10^{- 2}$	$3.46 \times 10^{- 2}$	$7.05 \times 10^{- 2}$	1.45
Lower flowback	$7.77 \times 10^{- 5}$	$3.37 \times 10^{- 4}$	$9.68 \times 10^{- 4}$	$2.17 \times 10^{- 3}$	$4.12 \times 10^{- 3}$	$9.11 \times 10^{- 3}$	$2.27 \times 10^{- 2}$	1.45
Lower bromide	$4.05 \times 10^{- 6}$	$1.52 \times 10^{- 5}$	$4.40 \times 10^{- 5}$	$1.05 \times 10^{- 4}$	$1.87 \times 10^{- 4}$	$5.53 \times 10^{- 4}$	$1.32 \times 10^{- 3}$	1.37
Creek scenario by effluent type, low flow^c
Produced-C	$5.59 \times 10^{- 2}$	$2.46 \times 10^{- 1}$	$1.20 \times 10^{0}$	$3.36 \times 10^{0}$	$7.55 \times 10^{0}$	$1.43 \times 10^{1}$	$4.86 \times 10^{1}$	1.89
Produced-ME	$6.02 \times 10^{- 2}$	$2.54 \times 10^{- 1}$	$1.22 \times 10^{0}$	$3.25 \times 10^{0}$	$7.52 \times 10^{0}$	$1.20 \times 10^{1}$	$2.16 \times 10^{1}$	1.94
Mixed/flowback	$8.36 \times 10^{- 3}$	$6.22 \times 10^{- 2}$	$3.06 \times 10^{- 1}$	$8.47 \times 10^{- 1}$	$1.88 \times 10^{0}$	$3.89 \times 10^{0}$	$6.65 \times 10^{0}$	1.86
Lower flowback	$3.38 \times 10^{- 3}$	$1.42 \times 10^{- 2}$	$7.05 \times 10^{- 2}$	$2.10 \times 10^{- 1}$	$4.41 \times 10^{- 1}$	$1.02 \times 10^{0}$	$2.18 \times 10^{0}$	1.76
Lower bromide	$1.77 \times 10^{- 4}$	$6.38 \times 10^{- 4}$	$3.45 \times 10^{- 3}$	$9.94 \times 10^{- 3}$	$1.98 \times 10^{- 2}$	$6.14 \times 10^{- 2}$	$1.20 \times 10^{- 1}$	1.65
Creek scenario by effluent type, high flow^c
Produced-C	$2.09 \times 10^{- 2}$	$1.02 \times 10^{- 1}$	$2.94 \times 10^{- 1}$	$5.64 \times 10^{- 1}$	$1.01 \times 10^{0}$	$2.25 \times 10^{0}$	$8.42 \times 10^{0}$	1.27
Produced-ME	$2.23 \times 10^{- 2}$	$1.07 \times 10^{- 1}$	$2.95 \times 10^{- 1}$	$5.57 \times 10^{- 1}$	$9.57 \times 10^{- 1}$	$1.86 \times 10^{0}$	$3.69 \times 10^{0}$	1.19
Mixed /flowback	$1.86 \times 10^{- 3}$	$2.60 \times 10^{- 2}$	$7.32 \times 10^{- 2}$	$1.43 \times 10^{- 1}$	$2.56 \times 10^{- 1}$	$5.18 \times 10^{- 1}$	$1.12 \times 10^{0}$	1.27
Lower flowback	$1.18 \times 10^{- 3}$	$6.00 \times 10^{- 3}$	$1.74 \times 10^{- 2}$	$3.39 \times 10^{- 2}$	$6.57 \times 10^{- 2}$	$1.39 \times 10^{- 1}$	$3.84 \times 10^{- 1}$	1.43
Lower bromide	$6.89 \times 10^{- 5}$	$2.74 \times 10^{- 4}$	$7.81 \times 10^{- 4}$	$1.60 \times 10^{- 3}$	$3.19 \times 10^{- 3}$	$7.70 \times 10^{- 3}$	$2.13 \times 10^{- 2}$	1.51

Note: Concentrations between 0.02 and $0.10 mg / L$ are bolded; concentrations of $0.02 mg / L$ or less are italicized.

a

Effluent bromide concentration ranges defined in Table 2.

b

IQ/median = interquartile range/median.

c

High-flow and low-flow conditions are simulated with the historical frequency distributions for the months with the highest and lowest average flows, respectively.

Table 6. Modeled River Bromide Concentrations at 100 km Downstream from Source

Type^a	Minimum	Percentiles					Maximum	IQ/median^b
Type^a	Minimum	5th	25th	Median	75th	95th	Maximum	IQ/median^b
River scenario, low flow^c
Produced-C	$6.14 \times 10^{- 4}$	$3.13 \times 10^{- 2}$	$1.36 \times 10^{- 1}$	$3.08 \times 10^{- 1}$	$5.44 \times 10^{- 1}$	$1.26 \times 10^{0}$	$4.97 \times 10^{0}$	1.33
Produced-ME	$7.03 \times 10^{- 4}$	$3.04 \times 10^{- 2}$	$1.33 \times 10^{- 1}$	$3.09 \times 10^{- 1}$	$5.07 \times 10^{- 1}$	$1.09 \times 10^{0}$	$2.24 \times 10^{0}$	1.21
Mixed/flowback	$1.88 \times 10^{- 4}$	$8.05 \times 10^{- 3}$	$3.41 \times 10^{- 2}$	$7.95 \times 10^{- 2}$	$1.44 \times 10^{- 1}$	$3.16 \times 10^{- 1}$	$6.33 \times 10^{- 1}$	1.39
Lower flowback	$3.63 \times 10^{- 5}$	$1.68 \times 10^{- 3}$	$8.05 \times 10^{- 3}$	$1.87 \times 10^{- 2}$	$3.70 \times 10^{- 2}$	$8.72 \times 10^{- 2}$	$2.15 \times 10^{- 1}$	1.55
Lower bromide	$2.07 \times 10^{- 6}$	$8.18 \times 10^{- 5}$	$3.90 \times 10^{- 4}$	$8.02 \times 10^{- 4}$	$1.81 \times 10^{- 3}$	$4.76 \times 10^{- 3}$	$1.19 \times 10^{- 2}$	1.77
River scenario, high flow^c
Produced-C	$1.62 \times 10^{- 4}$	$1.49 \times 10^{- 3}$	$7.06 \times 10^{- 3}$	$2.08 \times 10^{- 2}$	$5.80 \times 10^{- 2}$	$1.24 \times 10^{- 1}$	$4.39 \times 10^{- 1}$	2.44
Produced-ME	$1.64 \times 10^{- 4}$	$1.57 \times 10^{- 3}$	$7.03 \times 10^{- 3}$	$1.94 \times 10^{- 2}$	$5.84 \times 10^{- 2}$	$1.09 \times 10^{- 1}$	$2.39 \times 10^{- 1}$	2.65
Mixed/flowback	$1.58 \times 10^{- 5}$	$3.79 \times 10^{- 4}$	$1.79 \times 10^{- 3}$	$5.20 \times 10^{- 3}$	$1.53 \times 10^{- 2}$	$3.42 \times 10^{- 2}$	$7.05 \times 10^{- 2}$	2.60
Lower flowback	$1.00 \times 10^{- 5}$	$8.56 \times 10^{- 5}$	$4.17 \times 10^{- 4}$	$1.32 \times 10^{- 3}$	$3.53 \times 10^{- 3}$	$8.95 \times 10^{- 3}$	$2.27 \times 10^{- 2}$	2.36
Lower bromide	$5.00 \times 10^{- 7}$	$3.70 \times 10^{- 6}$	$2.03 \times 10^{- 5}$	$6.53 \times 10^{- 5}$	$1.64 \times 10^{- 4}$	$5.28 \times 10^{- 4}$	$1.32 \times 10^{- 3}$	2.20

Note: Concentrations between 0.02 and $0.10 mg / L$ are bolded; concentrations of $0.02 mg / L$ or less are italicized.

a

Effluent bromide concentration ranges defined in Table 2.

b

IQ/median = interquartile range/median.

c

High-flow and low-flow conditions are simulated with the historical frequency distributions for the months with the highest and lowest average flows, respectively.

Table 7. Modeled Bromide Concentrations at 50 km Downstream from Source of Mixed/Flowback Effluent with Reduced Discharges: One-Half, One-Third, and One-Fourth

Type^a	Minimum	Percentiles					Maximum	IQ/median^b
Type^a	Minimum	5th	25th	Median	75th	95th	Maximum	IQ/median^b
River scenario by discharge, low flow^c
Full	$9.03 \times 10^{- 4}$	$1.28 \times 10^{- 2}$	$4.37 \times 10^{- 2}$	$8.39 \times 10^{- 2}$	$1.47 \times 10^{- 1}$	$3.16 \times 10^{- 1}$	$6.33 \times 10^{- 1}$	1.23
Half	$4.22 \times 10^{- 4}$	$6.23 \times 10^{- 3}$	$2.23 \times 10^{- 2}$	$4.22 \times 10^{- 2}$	$7.44 \times 10^{- 2}$	$1.59 \times 10^{- 1}$	$3.20 \times 10^{- 1}$	1.23
Third	$2.84 \times 10^{- 4}$	$4.22 \times 10^{- 3}$	$1.47 \times 10^{- 2}$	$2.75 \times 10^{- 2}$	$5.01 \times 10^{- 2}$	$1.06 \times 10^{- 1}$	$2.09 \times 10^{- 1}$	1.29
Fourth	$3.65 \times 10^{- 4}$	$3.14 \times 10^{- 3}$	$1.10 \times 10^{- 2}$	$2.08 \times 10^{- 2}$	$3.70 \times 10^{- 2}$	$8.07 \times 10^{- 2}$	$1.65 \times 10^{- 1}$	1.25
River scenario by discharge, high flow^c
Full	$1.18 \times 10^{- 4}$	$1.48 \times 10^{- 3}$	$4.24 \times 10^{- 3}$	$8.82 \times 10^{- 3}$	$1.70 \times 10^{- 2}$	$3.46 \times 10^{- 2}$	$7.05 \times 10^{- 2}$	1.45
Half	$1.39 \times 10^{- 4}$	$7.28 \times 10^{- 4}$	$2.11 \times 10^{- 3}$	$4.43 \times 10^{- 3}$	$8.52 \times 10^{- 3}$	$1.72 \times 10^{- 2}$	$3.54 \times 10^{- 2}$	1.45
Third	$5.34 \times 10^{- 5}$	$4.87 \times 10^{- 4}$	$1.39 \times 10^{- 3}$	$2.92 \times 10^{- 3}$	$5.69 \times 10^{- 3}$	$1.17 \times 10^{- 2}$	$2.36 \times 10^{- 2}$	1.47
Fourth	$3.26 \times 10^{- 5}$	$3.75 \times 10^{- 4}$	$1.04 \times 10^{- 3}$	$2.18 \times 10^{- 3}$	$4.29 \times 10^{- 3}$	$8.69 \times 10^{- 3}$	$1.90 \times 10^{- 2}$	1.49
Creek scenario by discharge, low flow^c
Full	$8.36 \times 10^{- 3}$	$6.22 \times 10^{- 2}$	$3.06 \times 10^{- 1}$	$8.47 \times 10^{- 1}$	$1.88 \times 10^{0}$	$3.89 \times 10^{0}$	$6.65 \times 10^{0}$	1.86
Half	$2.52 \times 10^{- 3}$	${3.13 \times 10}^{- 2}$	$1.54 \times 10^{- 1}$	$4.28 \times 10^{- 1}$	$9.39 \times 10^{- 1}$	$1.89 \times 10^{0}$	$3.21 \times 10^{0}$	1.83
Third	$2.11 \times 10^{- 3}$	$2.02 \times 10^{- 2}$	$1.05 \times 10^{- 1}$	$2.85 \times 10^{- 1}$	$6.25 \times 10^{- 1}$	$1.26 \times 10^{0}$	$2.18 \times 10^{0}$	1.82
Fourth	$1.78 \times 10^{- 3}$	$1.52 \times 10^{- 2}$	$7.77 \times 10^{- 2}$	$2.14 \times 10^{- 1}$	$4.66 \times 10^{- 1}$	$9.61 \times 10^{- 1}$	$1.63 \times 10^{0}$	1.81
Creek scenario by discharge, high flow^c
Full	$1.86 \times 10^{- 3}$	$2.60 \times 10^{- 2}$	$7.32 \times 10^{- 2}$	$1.43 \times 10^{- 1}$	$2.56 \times 10^{- 1}$	$5.18 \times 10^{- 1}$	$1.12 \times 10^{0}$	1.27
Half	$1.07 \times 10^{- 3}$	$1.31 \times 10^{- 2}$	$3.69 \times 10^{- 2}$	$7.21 \times 10^{- 2}$	$1.26 \times 10^{- 1}$	$2.57 \times 10^{- 1}$	$5.59 \times 10^{- 1}$	1.24
Third	$9.72 \times 10^{- 4}$	$8.59 \times 10^{- 3}$	$2.47 \times 10^{- 2}$	$4.86 \times 10^{- 2}$	$8.43 \times 10^{- 2}$	$1.71 \times 10^{- 1}$	$3.64 \times 10^{- 1}$	1.23
Fourth	$5.16 \times 10^{- 4}$	$6.41 \times 10^{- 3}$	$1.84 \times 10^{- 2}$	$3.63 \times 10^{- 2}$	$6.39 \times 10^{- 2}$	$1.29 \times 10^{- 1}$	$2.82 \times 10^{- 1}$	1.25

Note: Concentrations between 0.02 and $0.10 mg / L$ are bolded; concentrations of $0.02 mg / L$ or less are italicized.

a

Full = permitted discharge; Half = half the permitted discharge; Third = one-third the permitted discharge; Fourth = one-fourth the permitted discharge.

b

IQ/median = interquartile range/median.

c

High-flow and low-flow conditions are simulated with the historical frequency distributions for the months with the highest and lowest average flows, respectively.

At a distance of 50 km downstream, the simulated median bromide concentrations were reduced due to flow accretion (Fig. 3 and Table 5). The reduction is proportional to the increase in median flow:

79.7 - {88.2 m}^{3} / s

during low flow, and

553 - {934 m}^{3} / s

for high flow in the river; and

2.8 - {4.8 m}^{3} / s

during low flow, and

21.1 - {26.3 m}^{3} / s

during high flow in the creek. The initial and 50 km maximum simulated concentrations were the same, because the flow accretion data included flow loss in losing reaches. In these cases, mass is lost from the river and the concentration remains the same as at the discharge point. Blacklick Creek discharges to the Conemaugh River 48 km downstream from the USGS gauge, so the creek results end approximately at this location. Further reduction in simulated median bromide concentrations occurred in the river at 100 km (Table 6). These simulated concentration reductions are those that would occur with the nationwide flow accretion data, tributary inflows would increase the potential for dilution and hence concentration reduction. The width of the simulated bromide concentration distributions increased downstream (as measured by the interquartile range divided by the median), because by adding the randomized flow accretion, the variability of the results increased. Because the mean low-flow (August) data for both the creek and river did not include the lowest flows on record, the maximum bromide concentrations for the lowest recorded flow would be higher than shown in Tables 4–6. As for the Produced-C simulations, a higher result within the 10,000-run Monte Carlo would have a small impact on the frequency distribution results.

Simulated bromide concentrations above the two target levels of

0.02 mg / L

, where most laboratories could meet precision and accuracy requirements, and

0.10 mg / L

, where DBP formation could be a concern, are indicated by bold and italic type in Tables 4–6. For the initial flow dilution (Table 4) and low flows in the creek and the river, only the Lower Bromide inputs do not exceed both levels. Bromide concentrations from both Produced-C and Produced-ME simulations under low-flow conditions in the river fell below

0.10 mg / L

in less than 25% of the simulations. For high river flows, the number increases to more than 75%. For all flows in the creek, both Produced-ME and Produced-C simulations results exceeded both target levels. For Mixed/Flowback water effluent, approximately 50% exceeded

0.10 mg / L

under low-flow conditions in the river and none for high flow. The creek simulation with its generally lower median flow had many more results exceeding the two specified levels. The Lower Bromide simulation almost always resulted in bromide concentrations below both thresholds (except for the low flow maximum in the creek). At 50 km downstream from the source, dilution increased the frequency of results with bromide concentration below the thresholds (Table 5). The qualitative nature of these results are similar to the predictions made by Wilson and Van Briesen (2013) and the observed drinking-water intake concentrations observed by States et al. (2013), in that high flow rates were associated with lower bromide concentrations.

Mass loading impacts the predicted river concentration through the effluent discharge and the bromide concentration. Scenarios were constructed to assess the impact of reducing the effluent discharge to 50, 33, and 25% of the permitted level. Mixed/Flowback water was used for the effluent concentration. The reduction in loading directly caused a proportional reduction in predicted downstream bromide concentrations (Table 7). The median bromide concentration for each case fell below

0.10 mg / L

for the river (Fig. 4). For the creek simulations, the median bromide concentrations were all above

0.10 mg / L

for low flow (Table 7), and below

0.10 mg / L

under high-flow conditions if the effluent discharge rate was reduced below one-half (Table 7).

Fig. 4. River scenario simulation results for mixed/flowback effluent at the full, half, one-third, and one-fourth the permitted level. The concentration ranges are indicated between low values in the high-flow month (March) and high values in the low-flow month (August). Light shading indicates concentrations between 0.02 and 0.1 mg/L; dark shading indicates concentrations below 0.02 mg/L

Transient Simulation

Average, predicted bromide concentrations from Produced-ME effluent were reduced for discharges that lasted less than 24 h (Fig. 5). Comparing against the 24-h simulation results demonstrate that the median, predicted, time-averaged bromide concentration decreases for each reduced duration because of flow accretion from Franklin to Kittanning, Pennsylvania. Low-flow and high-flow months differ in concentration by approximately a factor of 6. Reducing the pulse duration reduces the time-averaged bromide concentration at the downstream locations by factors that are close to the ratio of pulse time to 24 h (Fig. 6 and Table S6).

Fig. 5. Median downstream river concentrations for 24-h, 12-h, 8-h, and 4-h release transient scenarios based on the (a) 95th percentile and (b) 5th percentile for Produced-ME effluent

Fig. 6. Monocacy River comparison between the tracer data, semi-analytical model result, and WASP model result for a tracer cloud passing four measuring stations

Table 8. Characteristics of Five Rivers Used for Test Simulations, with a Range of Discharges

Sample station	Distance (km)	Discharge ( $m^{3} / s$ )	Time since injection (h)	Simulation frequency of time	Observed concentration ( $mg / L$ )	Conservative concentration ( $mg / L$ )	Simulation frequency of concentration
Antietam Creek
1	2.6	1.16	3.2	0.3459	0.0921	0.1057	0.4263
2	9.7	1.19	13	0.3713	0.0292	0.0386	0.494
3	21.6	1.64	43	0.6553	0.0049	0.0113	0.2669
4	29.6	1.78	67	0.7186	0.0022	0.007	0.1709
Tangiphoa River
1	8.2	5.8	9.8	0.5801	0.1241	0.1048	0.4061
2	18.0	9.8	21.5	0.6035	0.0428	0.0366	0.3689
3	41.5	11.9	35.4	0.3447	0.0231	0.0221	0.6664
4	55.4	12.4	45.2	0.3065	0.0184	0.0194	0.7122
5	71.0	14.4	57.2	0.2782	0.0142	0.0136	0.6876
6	82.1	17.8	66.2	0.2651	0.0097	0.0108	0.7359
7	94.0	18.7	77.2	0.2626	0.0088	0.0101	0.793
Red River
1	5.8	230	2.5	0.0482	0.0214	0.0289	0.691
2	75.6	245	33.3	0.0681	0.004	0.0054	0.9918
3	132.8	249	59.8	0.0586	0.0025	0.0036	0.991
4	193.1	249	92.8	0.0857	0.0015	0.0026	0.953
Wind-Bighorn River
1	9.2	233	1.4	0.1808	0.0117	0.0105	0.7994
2	32.7	235	4.9	0.1746	0.0058	0.0051	0.955
3	50.4	235	8.7	0.2482	0.0029	0.0035	0.8765
4	75.3	224	13.5	0.2638	0.002	0.0026	0.7654
5	99.9	215	18	0.2965	0.0015	0.0022	0.742
6	141.9	218	26	0.2872	0.0012	0.0021	0.969
7	181.4	255	33.2	0.2666	0.0008	0.0016	0.989
Mississippi River
1	54.7	6,820	10.3	0.2495	0.0037	0.0038	$> 1$
2	96.6	6,820	17.8	0.1998	0.0017	0.002	0.9867
3	117.5	6,820	23	0.2391	0.001	0.0009	0.4475
4	294.5	6,820	54.3	0.2099	0.0003	0.0006	0.681

Note: The range of discharges is from $1.16 m^{3} / s$ (41 cfs) to $294.5 m^{3} / s$ (241,000 cfs); the simulation frequency is the cumulative frequency from the Monte Carlo results corresponding to the observed time or concentration.

Table 9. Yellowstone River Release Characteristics and Results

Location	Distance from injection point (km)	Streamflow ( $m^{3} / s$ )	Time-to-peak (h)	Simulation frequency of time	Peak concentration ( $μg / L$ )	Simulation frequency of concentration
Lockwood Bridge injection (68 L)
Lockwood Bridge	0	99.1	0	—	—	—
Huntley Bridge	19.3	98.8	5.39	0.0069	30.4	0.9929
Pompeys Bridge	51.5	101.9	15.7	0.0166	8.9	0.7748
Custer Bridge	97.4	106.8	32.6	0.0426	4.63	0.6462
Myers Bridge	124.4	184.6	42	0.0594	2.27	0.5625
Forsyth Bridge	196.0	186.9	65.7	0.0838	2.05	0.7068
Myers Bridge injection (21 L)
Myers Bridge	0.0	191.1	0	—	—	—
Forsyth Bridge	71.6	195.1	22.3	0.1151	1.52	0.948
Cartersville Dam	73.4	195.1	23.1	0.1193	1.42	0.939
Rosebud Bridge	95.1	194.3	30	0.106	1.22	0.9357
Cartersville Dam injection (33 L)
Cartersville Dam	0.0	194.3	0	—	—	—
Rosebud Bridge	21.7	194.3	5.9	0.0418	9.27	0.9916
Fort Keogh Bridge	82.9	207.3	25.5	0.0995	2.08	0.8189
Kinsey Bridge	105.9	208.1	32.2	0.0878	1.88	0.8172
Miles City Bridge injection (51.5 L)
Miles City Bridge	0.0	210.1	0	—	—	—
Kinsey Bridge	19.0	211.5	4.98	0.0262	26.2	$> 1$
Calypso Bridge	62.4	214.4	16.7	0.0042	4.18	0.8558
Fallon Bridge	91.4	209.0	25.2	0.0028	2.82	0.6927
Glendive Bridge	143.4	211.8	42.4	0.0016	1.63	0.5344

Note: The simulation frequency is the cumulative frequency in the Monte Carlo results corresponding to the observed time or concentration.

For discharges at the river edge at Franklin, the estimated initial mixing distance is greater than the distance to Kittanning (Table S7). Discharging the effluent over the width of the river would be necessary for the initial mixing distance to be exceeded at the downstream stations, according to the criteria of Eq. (2). Where this condition is met, the transient model results indicated that the ratio of the median peak concentration to the median time-averaged concentration varied from 1.76 to 3.57 (Tables S8–S11). For most of the medians the ratio increased with decreasing input pulse duration, but peak concentration in pulse scenarios could be as much as a factor of four higher than the time-averaged level.

Conclusions

Using data from several CWTPs and actual receiving streams, the simulations showed that elevated downstream bromide concentrations (above the target levels of 0.02 and

0.10 mg / L

) can result from discharging each of the five categories of treated wastewater (Table 2). The degree to which a subsequent impact occurs at a drinking water intake depends on the characteristics of the CWTP effluent and the receiving stream. The potential for these impacts is greatest for the creek scenario and low-flow conditions in the river. Bromide concentrations at the drinking water intake depend on the volume, timing, and concentration in the effluent; the effluent discharge rate; the flow in the receiving body; and the distance from the wastewater discharge. Simulated median downstream bromide concentrations were reduced in proportion to reduction in effluent discharge in steady-state simulations, and for time-averaged concentrations from transient pulse loadings. Peak transient concentrations were found to be a factor of four higher than time-average concentrations. Although the focus of this paper was on single CWTPs, a watershed could contain other bromide sources including coal mine drainage and coal-fired electric plant effluent. The impact inside the drinking water treatment plant depends on the influent bromide concentration from all sources and a number of other factors, including the influent TOC concentration and the disinfection processes. The results from this paper suggest mitigation measures that can help meet the needs of drinking water plants: reduction in bromide mass released by limiting effluent through lower discharge limits or pulsed inputs, treatment of lower-bromide wastewater, or limiting discharge to high-flow conditions.

Supplemental Data

A discussion of initial mixing, Tables S1–S11, and Figs. S1–S12 are available online in the ASCE Library (www.ascelibrary.org).

Supplemental Materials

File (supplemental_data_ee.1943-7870.0000968_weaver.pdf)

Download
1.68 MB

Appendix. Transient Model Details

Kilpatrick (1993) developed a representation of concentration distributions based on superposition of a series of triangular distributions. To compare data from rivers of diverse sizes and injections of various amounts, the data are normalized by the mass of injection, flow rate, and mass lost to sorption or degradation. A unit concentration,

C_{u} (T^{- 1})

, is determined from (Kilpatrick and Taylor 1986)

C_{u} = 1 \times 10^{6} \frac{C}{R_{r}} \frac{Q}{M_{i}}

(3)

where

C

= concentration (

M / L^{3}

);

R_{r}

= recovery ratio (dimensionless);

Q

= stream discharge (

L^{3} / T

); and

M_{i}

= mass injected (M), and the factor of

1 \times 10^{6}

is included for convenience. The recovery ratio is defined as the mass passing a cross section to the mass injected. Although called a concentration, the unit concentration is actually the partially nondimensional mass flux (mass flux per unit mass of injected solute), which retains the time unit in the denominator. The remaining variable, the longitudinal dispersion, is assumed to be comparable from one river to another (Jobson 1996, 1997).

Jobson (1996, 1997) developed empirical data from tracer studies to estimate travel time and longitudinal dispersion in rivers and streams. Because many of the experiments date back to the 1960s, limitations of the data include inaccuracy in the location sampling stations and timing of peak concentrations, and concentration data potentially collected within initial mixing zones. Such errors contribute to the uncertainties associated with the regression formulas subsequently presented.

Jobson (1996, 1997) used data from 60 rivers, 109 tracer injections, and 422 cross sections to develop the regression equations. The river discharges ranged from a mean annual discharge of

1.3 m^{3} / s

in a small creek to

11,000 m^{3} / s

in the Mississippi River. The slopes ranged from

36.0 m / km

in the creek to

0.01 m / km

in the Mississippi River.

From the 422 cross sections that had data on annual mean flow, Jobson found that the peak unit-concentration was represented by

C_{up} = 857 T_{p}^{- 0.760 {(Q / Q_{a})}^{- 0.079}}

(4)

which includes the travel time for the peak concentration,

T_{p} (T)

, the ratio of river discharge,

Q (L^{3} / T)

, to mean annual river discharge

Q_{a} (L^{3} / T)

. Parameters that influence the transport time include the drainage area,

D {(L}^{2})

, the reach’s slope,

S (L / L)

, the mean annual discharge,

Q_{a}

, (

L^{3} / T

), and the discharge at time of measurement,

Q

, (

L^{3} / T

). A regression formulation for the velocity of the peak concentration,

V_{p}

, (

m / s

) were formed from 939 data points and these parameters

V_{p} = 0.094 + 0.0143 {(D^{'})}^{0.919} {(Q^{'})}^{- 0.469} S^{0.159} \frac{Q}{D}

(5)

where

D^{'}

= dimensionless drainage area defined by

D^{'} = D^{0.125} \sqrt{g} / Q_{a}

;

g

= acceleration of gravity (

L / T^{2}

); and

Q^{'}

= dimensionless discharge defined by

Q / Q_{a}

. Discharge

Q

is expressed in

m^{3} / s

, and drainage,

D

, in

m^{2}

for this regression equation.

The time for the leading edge of the concentration distribution,

T_{l}

(hours), was found to be highly correlated to the peak arrival time, following

T_{l} = 0.890 T_{p}

(6)

The time for the trailing edge of the concentration distribution,

T_{d 10}

(hours), was estimated from

T_{d 10} = \frac{2 \times 10^{6}}{C_{up}}

(7)

and is based on the assumptions that the area under the unit concentration curve is

1 \times 10^{6}

and that half of the mass lies between the peak concentration and a point where the concentration is 10% of the peak value. This assumption effectively truncates trailing concentration distributions. In later work, localized travel time equations were developed for streams in Arkansas (Funkhouser and Barks 2004), Iowa (Christiansen 2009), and the Susquehanna, Delaware, and Lehigh River basins (Reed and Stuckey 2002).

To incorporate variability in the regression formulas into the Monte Carlo analysis the following procedure is used. For the travel time and concentration input, a frequency distribution was developed from the empirical data and the Jobson regression formulas [Jobson 1996; Eq. (5)]. The equations have the form

V_{p} = A + B \hat{x}

(8)

where

A

and

B

are constants and

\hat{x}

is

{(D^{'})}^{0.919} {(Q^{'})}^{- 0.469} S^{0.159} (Q / D)

as defined previously. For each choice of

A

and

B

, the cumulative probability that the experimental peak velocity did not exceed the Eq. (8) estimate was determined. The resulting set of cumulative probabilities (Figs. S4 and S5) is used to represent the variability in the regression formulas in the Monte Carlo simulations described below.

Test Problems and Model Results

An experiment conducted in the Monocacy River was used for the first test problem. Input parameters for the semi-analytical model were taken from Nordin and Sabol (1974) and Jobson (1996), while the estimated conservative concentrations were taken from Nordin and Sabol (1974). The experiment consisted of an instantaneous injection of 1.908 kg of dye, which was then observed passing four sampling stations located between 10.3 and 34.3 km downstream. Data were collected only when the tracer cloud was beyond the initial mixing distance as defined by Fischer et al. (1979) (Table S1).

The semi-analytical model was run and did not include adjustment of parameters to achieve calibration (Fig. 6). The uncalibrated semi-analytical model results closely match the peak arrival times of the tracer experiment (Table S2). In the model, the peak concentrations tended to be higher and the distributions narrower than in the experiment. The agreement between the semi-analytical model and the tracer data results from the Monocacy River parameters falling near the Jobson regression line. The WASP model (Ambrose et al. 1983, 2009) was calibrated to data from the June 7, 1968, Monocacy River tracer experiment (Nordin and Sabol 1974) by adjusting the number of segments and Manning’s

n

. These results show that for a tracer experiment where the travel time is well estimated by the Jobson formulas, the uncalibrated semi-analytical model and the calibrated WASP model can each reproduce the tracer data.

Five additional tracer experiments were chosen to cover a wide range of flows within the Nordin and Sabol (1974) dataset (Table 8). While the Monocacy River showed an exceptional fit to the semi-analytical model and served as an ideal test case, tracer data is expected to lie only within the bounds of the modeled frequency distributions, because of scatter in the regression formulas. The semi-analytical model was run 10,000 times to include sufficient numbers of extreme values of the input frequency distributions. Each tracer experiment was simulated by treating only the parameters of the regression formula as random; all other parameters were set to the observed values (Table 8). After running the model, output frequency distributions were developed for the time-to-peak and peak concentration. These curves typically show an order-of-magnitude variation in the output. For each sampling station, the observed time-to-peak fell within the bounds of the frequency distribution (Table 8 and Figs. S7 and S8), as did all but one observed peak concentration (Table 8, Fig. S10). The peak concentration at the first sampling station of the Mississippi River experiment was 2.6% above the highest concentration on the frequency curve, which might have occurred because of incomplete mixing.

Four dye releases were made to the Yellowstone River using slug injections of a 20% solution of rhodamine dye (McCarthy 2009). Flow, elevation, and drainage area were available at four USGS gauges that spanned the length of the experimental reaches. Parameters for 12 sampling locations were interpolated from this gauge data. The releases were made at four injection points along the river (Lockwood Bridge, Myers Bridge, Cartersville Dam, and Miles City Bridge), and samples were collected at 3–5 locations downstream of the injection points (Table 9). Samples were collected when the dye clouds had passed the initial mixing distance. The time-to-peak (Fig. S11) and peak concentrations (Fig. S12) were plotted against the frequency distributions for 10,000 runs of the semi-analytical model. Each of the time-to-peak values plotted within the frequency distribution, while the peak concentration at the first sampling station downstream from Lockwood Bridge exceeded the maximum from the model. Again, this may be due to incomplete mixing at the first sampling location.

Acknowledgments

The authors wish to thank Angela McFadden, Amy Bergdale, and Chad Harsh of U.S. EPA for facilitating access to NPDES permit data. The authors acknowledge valuable input from Richard Lowrance, Stephen Kraemer, and Stig Regli, U.S. EPA; Kay Pinley, U.S. EPA grantee; Ashley McElmurry, contractor to U.S. EPA, WASP simulation results from Chris Knightes, U.S. EPA and Bob Ambrose. Clinton Hittle of USGS supplied unpublished USGS flow-discharge data for mixing zone estimation. The authors thank several internal EPA reviewers and three anonymous journal reviewers for their comments, which improved the quality of the paper. The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.

References

Ambrose, R. B., Hill, S. I., and Mulkey, L. A. (1983). “User’s manual for the chemical transport and fate model (TOXIWASP), version 1.”, U.S. Environmental Protection Agency, Athens, GA.

Google Scholar

Ambrose, R. B., Martin, J. L., and Wool, T. A. (2009). “WASP7 stream transport model theory and user’s guide.”, U.S. Environmental Protection Agency, Athens, GA.

Google Scholar

Asher, W., Bender, D. A., Sogorski, J. S., and Bartholomay, R. C. (2006). “StreamVOC—A deterministic source-apportionment model to estimate volatile organic compound concentration in rivers and streams.”, U.S. Geological Survey, Reston, VA.

Google Scholar

Bonacquisti, T. P. (2006). “A drinking water utility’s perspective on bromide, bromated and ozonation.” Toxicology, 221(2-3), 145–148.

Crossref

Google Scholar

Boschee, P. (2014). “Produced and flowback water recycling and reuse.” Oil Gas Facil., 17–21.

Google Scholar

Brantley, S. L., et al. (2014). “Water resource impacts during unconventional shale gas development: The Pennsylvania experience.” Int. J. Coal Geol., 126, 140–156.

Crossref

Google Scholar

Brown, D., Bridgeman, J., and West, J. R. (2011). “Predicting chlorine decay and THM formation in water supply systems.” Rev. Environ. Sci. Biotechnol., 10(1), 79–99.

Crossref

Google Scholar

Chang, E. E., Lin, Y. P., and Chiang, P. C. (2001). “Effects of bromide on the formation of THMs and HAAs.” Chemosphere, 43(8), 1029–1034.

Crossref

Google Scholar

Christiansen, D. E. (2009). “Dye tracer tests to determine time-of-travel in Iowa streams, 1990–2006.” Data Series 444, U.S. Geological Survey, Reston, VA.

Google Scholar

Clark, C. E., and Veil, J. A. (2009). “Produced water volumes and Mmanagement practices in the United States.”, U.S. Dept. of Energy, National Energy Technology Laboratory, Oakridge, TN.

Google Scholar

Cooper, W. J., Zika, R. G., and Steinhauer, M.S. (1985). “Bromide-oxidant interactions and THM formation: A literature review.” Am. Water Works Assoc. J., 116–121.

Crossref

Google Scholar

Entrekin, S., Evans-White, M., Johnson, B., and Hagenbuch, E. (2011). “Rapid expansion of natural gas development poses a threat to surface waters.” Front. Ecol. Environ., 9(9), 503–511.

Crossref

Google Scholar

Fair, P. S., Sorrell, R. K., Karapondo, M. S., Bohl, B. J., Frebis, C. P., and Feige, M. A. (2002). “Quality of ICR monitoring data.” Information collection rule data analysis, M. J. McGuire, J. L. McLain, and A. Obolensky, eds., American Water Works Association Research Foundation, Denver, 39–66.

Google Scholar

Federal Register. (1996). “National primary drinking water regulations: Monitoring requirements for public drinking water supplies: Cryptosporidium, giardia, viruses, disinfection byproducts, water treatment plant data and other information requirements.” 61(94), 24354–24388.

Google Scholar

Ferrar, K. J., Michanowicz, D. R., Christen, C. L., Mulcahy, N., Malone, S. L., and Sharma, R. K. (2013). “Assessment of effluent contaminants from three facilities discharging Marcellus shale wastewater to surface waters in Pennsylvania.” Environ. Sci. Technol., 47(7) 3472–3481.

Crossref

Google Scholar

Fischer, H. B., List, E. J., Koh, R. C. Y., Imberger, J., and Brooks, N. H. (1979). Mixing in Inland and coastal waters, Academic Press, New York, 483.

Google Scholar

Funkhouser, J. E., and Barks, C. S. (2004). “Development of a travel time prediction equation for streams in Arkansas.”, U.S. Geological Survey, Reston, VA.

Google Scholar

Gregory, K. B., Vidic, R. D., and Dzombak, D. A. (2011). “Water management challenges associated with the production of shale gas by hydraulic fracturing.” Elements, 7(3), 181–186.

Crossref

Google Scholar

Haluszczak, L. O., Rose, A. W., and Kump, L. R. (2013). “Geochemical evaluation of flowback brine from Marcellus gas wells in Pennsylvania, USA.” Appl. Geochem., 28, 55–61.

Crossref

Google Scholar

Hart Resources. (2011). “Letter to customers.” 〈http://www.epa.gov/region3/marcellus_shale/pdf/hart-letters/2customers.pdf〉 (Sep. 19, 2011).

Google Scholar

Hayes, T. (2009). “Sampling and analysis of water streams associated with the development of Marcellus Shale gas.” Marcellus Shale Coalition Report, Des Plaines, IL.

Google Scholar

Hladik, M. L., Focazio, M. J., and Engle, M. (2014). “Discharges of produced waters from oil and gas extraction via wastewater treatment plants are sources of disinfection byproducts to receiving streams.” Sci. Total Environ., 466-467, 1085–1093.

Crossref

Google Scholar

Jobson, H. E. (1996). “Prediction of travel time and longitudinal dispersion in rivers and streams.”, U.S. Geological Survey, Reston, VA.

Google Scholar

Jobson, H. E. (1997). “Predicting travel time and dispersion in rivers and streams.” J. Hydraul. Eng., 971–978.

Crossref

Google Scholar

Kilpatrick, F. A. (1993). “Simulation of soluble waste transport and buildup in surface waters using tracers.” Chapter A20, Application of hydraulics, Techniques of Water-Resources Investigations of the U.S. Geological Survey, Denver.

Google Scholar

Kilpatrick, F. A., and Taylor, K. R. (1986). “Generalization and applications of tracer dispersion data.” Water Resour. Bull., 22(4), 537–548.

Crossref

Google Scholar

Kilpatrick, F. A., and Wilson, J. F. (1989). “Measurement of time of travel in streams by dye tracing.” Chapter A9, Application of hydraulics, Techniques of Water-Resources Investigations of the U.S. Geological Survey, Denver.

Google Scholar

Krasner, S. W., et al. (2006). “Occurrence of a new generation of disinfection byproducts.” Environ. Sci. Technol., 40(23), 7175–7185.

Crossref

Google Scholar

Krasner, S. W., McGuire, M. J., Jacangelo, J. G., Patania, N. L., Reagan, K. M., and Aieta, E. M. (1989). “The occurrence of disinfection by-products in US drinking water.” J. AWWA, 41–53.

Crossref

Google Scholar

Martin, J. F., and McCutcheon, S. C. (1999). Hydrodynamics and transport for water quality modeling, CRC Press, Boca Raton, FL, 816.

Google Scholar

McCarthy, P. (2009). “Travel times, streamflow velocities, and dispersion rates in the Yellowstone River, Montana.”, U.S. Geological Survey, Reston, VA.

Google Scholar

McClain, J. L., Obolensky, A., and Shukairy, H. M. (2002). “Disinfection by-product speciation.” Information collection rule data analysis, M. J. McGuire, J. L. McLain, and A. Obolensky, eds., American Water Works Association Research Foundation, Denver, 141–168.

Google Scholar

McGuire, M. J., and Graziano, N. (2002). “Trihaolmethanes in U.S. drinking water—NORS to ICR.” Information collection rule data analysis, M. J. McGuire, J. L. McLain, and A. Obolensky, eds., American Water Works Association Research Foundation, Denver, 81–110.

Google Scholar

McKay, M. D., Beckman, R. J., and Conover, W. J. (1979). “A comparison of three methods for selecting values of input variables in the analysis of output from a computer code.” Technometrics, 21(2), 239–245.

Crossref

Google Scholar

McTigue, N. E., Cornwell, D. A., Graf, K., and Brown, R. (2014). “Occurrence and consequences of increase bromide in drinking water sources.” J. Am. Water Works Assoc., 106(11), E492–E508.

Crossref

Google Scholar

Moll, D. M., and Krasner, S. W. (2002). “Bromate occurrence in disinfected water.” Information collection rule data analysis, M. J. McGuire, J. L. McLain, and A. Obolensky, eds., American Water Works Association Research Foundation, Denver, 225–250.

Google Scholar

Nordin, C. F., and Sabol, G. V. (1974). “Empirical data on longitudinal dispersion in rivers.”, U.S. Geological Survey, Lakewood, CO.

Google Scholar

Obolensky, A., and Singer, P. C. (2008). “Development and interpretation of disinfection byproduct formation models using the Information Collection Rule database.” Environ. Sci. Technol., 42(15), 5654–5660.

Crossref

Google Scholar

PA DEP. (2011). “Letter from the Pennsylvania department of environmental protection to the EPA regional administrator.” 〈http://www.epa.gov/region3/marcellus_shale/pdf/letter-padep-to-epa7-26-11.pdf〉 (Jul. 26, 2011).

Google Scholar

Plewa, J. J., et al. (2008). “Occurrence, synthesis and mammalian cell cytotoxicity and genotoxicity of haloacetamides: An emerging class of nitrogenous drinking water disinfection byproducts.” Environ. Sci. Technol., 42(3), 955–961.

Crossref

Google Scholar

Pressman, J. G., et al. (2010). “Concentration, chlorination, and chemical analysis of drinking water for disinfection by product mixtures health effects research: U.S. EPA’s four lab study.” Environ. Sci. Technol., 44(19), 7184–7192.

Crossref

Google Scholar

Reed, L. A., and Stuckey, M. H. (2002). “Prediction of velocities for a range of streamflow conditions in Pennsylvania.”, U.S. Geological Survey, New Cumberland, PA.

Google Scholar

Richardson, S. D., Plewa, M. J., Wagner, E. D., Schoeny, R., and DeMarini, D. M. (2007). “Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: A review and roadmap for research.” Mutat. Res., 636(1–3), 178–242.

Crossref

Google Scholar

Roccaro, P., Chang, H.-S., Vagliasinde, F. G. A., and Korshin, G. V. (2008). “Differential absorbance study of effects of temperature and chlorine consumption and formation of disinfection by-products in chlorinated water.” Water Res., 42(8–9), 1879–1888.

Crossref

Google Scholar

Sharma, V. K., Boril, R. Z., and McDonald, T. J. (2014). “Formation and toxicity of brominated disinfection byproducts during chlorination and chloramination of water: A review.” J. Environ. Sci. Health Part B, 49(3), 212–228.

Crossref

Google Scholar

Soeder, D. J., and Kappel, W. M. (2009). “Water resources and natural gas production from the Marcellus Shale.”, U.S. Geological Survey, Baltimore, MD.

Google Scholar

Sohn, J., Amy, G., and Yoon, Y. (2006). “Bromide ion incorporation into brominated disinfection by-products.” Water Air Soil Pollut., 174(1–4), 265–277.

Crossref

Google Scholar

States, S., et al. (2013). “Marcellus Shale drilling and brominated THMs in Pittsburgh, PA, drinking water.” J. Am. Water Works Assoc., 105, E432–E448.

Crossref

Google Scholar

U.S. EPA. (1996). “Reprints of EPA methods for chemical analyses under the information collection rule.”, Office of Water, Cincinnati.

Google Scholar

U.S. EPA. (2001). “Stage 1 disinfectants and disinfection byproducts rule.”, Office of Water.

Google Scholar

U.S. EPA. (2011a). “Clean water act section 309 administrative orders.” Philadelphia, 〈http://www.epa.gov/region03/marcellus_shale/index.html〉 (Feb. 9, 2015).

Google Scholar

U.S. EPA. (2011b). “In the matter of Pensylvania Brine Treatment, Inc—Josephine Facility, respondant’s final response.”, Philadelphia.

Google Scholar

U.S. EPA. (2011c). “In the matter of Pensylvania brine treatment, Inc—Franklin facility, respondant’s final response.”, Philadelphia.

Google Scholar

U.S. EPA. (2011d). “In the matter of hart resource technologies, Inc—Creekside Facility, respondant’s final response.”, Philadelphia.

Google Scholar

U.S. EPA. (2011e). “Voluntary sampling data from oil and gas wastewater treatment facilities.” Philadelphia, 〈http://www.epa.gov/region03/marcellus_shale/index.html〉 (Jul. 29, 2015).

Google Scholar

U.S. EPA. (2013). “Administrative order for compliance on consent EPA.”, Philadelphia.

Google Scholar

Vengosh, A., Jackson, R., Warner, N., Darrah, T., and Kondash, A. (2014). “A critical review of the risks to water resources from unconventional shale gas development and hydraulic fracturing in the United States.” Environ. Sci. Technol., 48(15), 8334–8348.

Crossref

Google Scholar

Vidic, R. D., Brantley, S. L., Vandenbossche, J. M., Yoxtheimer, D., and Abad, J. D. (2013). “Impact of shale gas development on regional water quality.” Science, 340(6134), 1235009-1–1235009-9.

Crossref

Google Scholar

Warner, N. R., Christie, C. A., Jackson, R. B., and Wengosh, A. (2013). “Impacts of shale gas waste water disposal on water quality in western Pennsylvania.” Environ. Sci. Technol., 47(20), 11849–11857.

Crossref

Google Scholar

Wilson, J. M., and Van Briesen, J. M. (2013). “Source water changes and energy extraction activities in the Monongahela River, 2009–2012.” Environ. Sci. Technol., 47(21), 12575–12582.

Crossref

Google Scholar

Wilson, J. M., Wang, Y., and VanBriesen, J. M. (2013). “Sources of high total dissolved solids to drinking water supply in southwestern Pennsylvania.” J. Environ. Eng., B4014003.

Google Scholar

Yotsukura, N., and Cobb, E. B. (1972). “Transverse diffusion of solutes in natural streams.”, U.S. Geological Survey, Washington, DC.

Google Scholar

Information & Authors

Information

Published In

Journal of Environmental Engineering

Volume 142 • Issue 1 • January 2016

Copyright

This work is made available under the terms of the Creative Commons Attribution 4.0 International license, http://creativecommons.org/licenses/by/4.0/.

History

Received: Jul 18, 2014

Accepted: Mar 25, 2015

Published online: Aug 13, 2015

Published in print: Jan 1, 2016

Discussion open until: Jan 13, 2016

Authors

Affiliations

James W. Weaver, A.M.ASCE [email protected]

Research Hydrologist, United States Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Ground Water and Ecosystem Restoration Division, 919 Kerr Research Dr., Ada, OK 74820 (corresponding author). E-mail: [email protected]

View all articles by this author

Jie Xu [email protected]

Student Services Contractor, United States Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Ground Water and Ecosystem Restoration Division, 919 Kerr Research Dr., Ada, OK 74820. E-mail: [email protected]

View all articles by this author

Susan C. Mravik [email protected]

Soil Scientist, United States Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Ground Water and Ecosystem Restoration Division, 919 Kerr Research Dr., Ada, OK 74820. E-mail: [email protected]

View all articles by this author

Metrics & Citations

Metrics

Citations

Download citation

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Abstract

Introduction

Disinfection Byproducts

Bromide in Oil and Gas Wastewater

Bromide in Pennsylvania Surface Waters

Data Sources

Commercial Treatment Plant Effluent

Flow Accretion

Models

Steady-State Model

Transient Model

Scenarios

Results and Discussion

Steady-State Simulations

Transient Simulation

Conclusions

Supplemental Data

Supplemental Materials

Appendix. Transient Model Details

Test Problems and Model Results

Acknowledgments

References

Information

Published In

Copyright

History

Authors

Affiliations

Metrics

Citations

Download citation

Cited by

Figures

Other

Share

Copy the content Link

Share with email

Share

Request Username

Create a new account

Change Password

Password Changed Successfully

Verify Phone

Congrats!