MTBE: Is A Little Bit OK? Presented at the National Ground Water Association's 2002 Petroleum Hydrocarbons and Organic Chemicals in Ground Water®: Prevention, Assessment, and Remediation, November 6-8, Atlanta, Georgia, pp 206-219 James W. Weaver Ecosystems Research Division National Exposure Research Laboratory United States Environmental Protection Agency Athens, Georgia 30605 Matthew C. Small Underground Storage Tank Program Office Region 9 United States Environmental Protection Agency San Francisco, California 94105 Abstract Methyl tertiary butyl ether (MTBE) has been used as a gasoline additive to serve two major purposes. First, MTBE was used as an octane-enhancer to replace organic lead, beginning in about 1979. Beginning in about 1992, MTBE was also used as a fuel oxygenate additive to meet requirements of the Clean Air Act Amendments (CAAA) of 1990. Generally, the amount of MTBE used for octane enhancement was lower than that required to meet CAAA requirements. An unintended consequence of using MTBE to address air quality issues has been widespread groundwater contamination. The decision to use certain amounts of MTBE or other chemicals as gasoline additives is the outcome of economic, regulatory, policy, political, and scientific considerations. Decision makers ask questions such as "How do ground water impacts change with changing MTBE content? How many wells would be impacted? and What are the associated costs?" These questions are best answered through scientific inquiry, but many different approaches could be developed. Decision criteria include time, money, comprehensiveness, and complexity of the approach. Because results must be communicated to a non-technical audience, there is a trade-off between the complexity of the approach and the ability to convince economists, lawyers and policy makers that the results make sense. The questions on MTBE content posed above were investigated using transport models, a known release scenario and varying gasoline compositions. A set of simulations was performed that assumed 3% (octane enhancement) and 11% (CAAA) MTBE in gasoline. The results were that ground water concentrations would be reduced in proportion to the reduction of MTBE in the fuel. Plume lengths, though, would not be proportionately reduced. One implication of these results was that the concentrations would be reduced, but the number of impacted wells would remain similar. Because simulations included emplacement of the gasoline, dissolution from contact with flowing ground water and transient transport in the aquifer, a common sense explanation of the results was difficult to construct. A simpler model was then used for the purpose of explaining to policy makers why the plume length reductions were less than proportionate to the reduction of the amount of MTBE. The model was simple enough (one-dimensional, steady state, constant source concentration) so that the effect of each term of the transport equation on plume length could be easily shown. The weight of evidence from using multiple models, direct explanations from the transport equation, and field observation, should provide a sufficient basis for policy makers to understand scientifically how gasoline composition affects ground water impacts. ------- Comparative Modeling Approach The evaluation of the impacts of 3% vs 11% MTBE in gasoline were conducted from the following approach. Effective solubilities of MTBE were estimated from Raoult's Law. The resulting concentrations were either used directly as the boundary condition of a steady state model or as an oil-water partition coefficient in the Hydrocarbon Spill Screening Model (Weaver et al., 1994). Each of these two models captured different phenomena and contribute understanding of the role of source concentration to ground water impacts. Each model also contains a number of limitations and universal applicability of these models to all field sites is not claimed. All impacts of MTBE are based, however, on a release scenario that is compatible with the assumptions underlying the model in use. Further, and to minimize the effects of simplifications in the models, the results are given as comparisons between the effects of MTBE content of 3% or 11%. Since each model correctly simulates the scenario upon which it is based, the comparison between the two MTBE contents is valid within the context of the modeled scenario. Effective Solubility Chemical handbooks and databases list measured solubilities of chemicals in water (Montgomery, 2000). Generally these data are limited and contradictory in that for most chemicals there are differing reported solubilities measured at a given temperature, and often data are only available for one temperature. For example, MTBE solubilities of 36,200 mg/L, 48,000 mg/L and 51,000 mg/L have been reported for measurements made at 25°C (www.dep.state.pa.us/phvsicalproperties/Default.htmY Discrepancies such as this are generally attributed to the difficulty of measuring the solubilities of hydrophobic organic chemicals (MacKay and Boethling, 2000). Other factors can influence concentrations observed at field sites. For example, at a site with a release of neat MTBE (i.e., liquid MTBE that was not mixed with gasoline) in Southeast Texas, the highest observed MTBE concentration roughly 10 years after the release was 12,900 mg/L (Wilson, 2000). The release occurred in a tight clay where no significant MTBE plume formed. This concentration was roughly l/3rd to 1/4111 the reported MTBE solubility. The lower concentration illustrates that theoretical solubilities may not be observed in the field. Reasons stem from the timing of sampling relative to the release and the relative position of well screens in both the vertical and horizontal. When a monitoring well is screened across a interval with varying concentration and hydraulic conductivity, the mass entering the well from each different elevation also varies. The resulting contaminant concentration that would be observed in the well bore depends on the relative magnitudes of these contributions, among other factors. When pumped, the well bore concentration is expected to be less than the peak concentration actually present in the formation. Gbnerally, the longer the screen the more averaging occurs as more variability is encountered in the formation. Screen placement also can contribute to reduction of observed concentrations. If part of the screen is in contract with clean water, then upon pumping this water is mixed with the contaminated water, again yielding a concentration lower than in the formation within the bounds of the plume. The effects of screen length and vertical placement are illustrated in the average borehole concentration calculator found at the EPA web site at http://www. epa. gov/athens/onsite. The calculator shows that the measured concentration can range from 0.0 to somewhat lower than the actual peak concentration, depending on each of the factors described above. When present as part of a fuel mixture the pure compound MTBE solubility given above will not be observed in ground water samples. The actual aqueous MTBE concentrations resulting from contact of water with a fuel depend upon the solubility of MTBE, the amount of MTBE in the fuel, the activity coefficient of MTBE (assumed equal to 1.0 in the following analysis) and the composition of the fuel. Raoult's law has been found to adequately approximate these so-called effective solubilities for gasoline containing MTBE (Cline et al., 1991). Since petroleum fuels are largely composed of aromatics and aliphatic hydrocarbons, the dependence on composition is minimal so that detailed compositional data are not necessary for approximating effective solubilities. Figure 1 shows the variation of theoretical effective solubility with fraction MTBE in gasoline for assumed solubilities of 48300 mg/1 and 36200 mg/1 and activity coefficients equal to 1.0. For these cases the reduction ------- of MTBE in gasoline from 11% to 3% results in a reduction of effective MTBE solubility by 72.3 and 72.1%, respectively. The reduction is in direct proportion to the reduction of MTBE in gasoline (Table 1). Reducing MTBE content in gasoline, then, is expected to result in a directly proportionate reduction in MTBE concentrations in ground water resulting from fuel releases. 8000 o» m 6000 -*—¦ 4000 o CO 2000 _> c> CD in A A MTBE Solubility = 48300 mg/l ~ ~ MTBE Solubility - 36200 mg/l 0 5 10 15 Fraction MTBE in Gasoline Figure 1. Estimated MTBE effective solubility as a function of the MTBE content in gasoline for MTBE water solubilities of 48,300 mg/l and 36,200 mg/l. ------- Table 1 MTBE Effective Solubilities Percent MTBE in Gasoline Unitary Activity Coefficients (y = 1.0) Assumed Solubility at 25°C 48300 mg/1 36200 mg/1 1 575 431 3 1720 1280 5 2860 2120 7 3990 2960 9 5110 3760 11 6220 4590 Steady State Model A form of the mass balance equation for transport of solutes in ground water is given by dc dc d2c d2c d2c R— — —v h D —y+ D —— + D —5— Ac (1) j x ^ x 2 y L z * l v y ot ox ox oy oz where R is the retardation factor [dimensionless], c is the concentration [M/L3]; t represents time [T]; x, y, and z are the three cartesian coordinate directions [L]; vx, is the x-direction seepage velocity [L/T], Dx, Dy and Dx are the three components of dispersion [L2/T], and A is a first order loss coefficient [T"1]. This form of the transport equation is based on the assumption that the dispersion constants are independent of time and space, and that ground water flow is one-dimensional. This equation is significant because only when the ground water flow is assumed one-dimensional, steady and uniform can an analytical solution be obtained. Plume Length The transport equation is a linear partial differential equation that is solved for concentration. Measures of plume fength may be derived from, but are not direct results of the solution. Plume lengths represent the interaction of the transport and loss processes, the boundary conditions and the aquifer geometry To express plume lengths in a simple way, consider the steady-state, one-dimensional transport equation: n dc d2c . 0 = -vx— + Dx—- - Ac (2) dx dc2 subject to the boundary conditions c(0) = co dc , . —(oo) = 0 dx The analytical solution gives the concentration at any point from (van Genuchten and Alves, 1982): ------- c = co exp 2D 1 - 1 + 4 ID \XY (3) The scenario governed by this equation consists of an infinite source at concentration c0, that is transported in only one direction in a uniform aquifer. By considering steady state conditions, a bounding case is developed. This simplified case was used because it is possible to solve explicitly for the plume length, x, associated with any specified concentration, c. Thus, 2 Z) In r \ c. x = V/ c O J r 4 AD ^ 1 + — V V J Yi (4) Since the concentration of interest is the MTBE drinking water advisory of 20 i_ig/l the concentration, c, of interest is 0.02 mg/1. The dispersion constant is conventionally expressed as a product of the dispersivity, a, and seepage velocity, v. Incorporating the Xu and Eckstein (1995) regression equation for dispersivity as a function of plume length, a = 0.83[ logio(Lp)]2 414 = f(x), gives an implicit expression for the plume length (x = Lp), that incorporates scale-dependent dispersivity. f 1 - ( 4 ID 1 + — = 2/(x)vln( 0- 1.02/ (5) This equation can be solved numerically to determine the plume length associated wth the 0.02 mg/1 concentration level, given the initial concentration, seepage velocity and half-life. The solution for steady state plume length has been implemented as a JavaScript "calculator" and can be run from the EPA web site: http://www.epa.gov/athens/software/trainingAVebCourse/part-two/onsite. The calculator includes a iterative solution of the nonlinear (f(x)) form of the equation (equation 5), and as a QA check a direct solution of the fixed dispersivity version (equation 4). Equation 4 can also be used with a hand held calculator to verify the results given below. The Steady State Plume Length Calculator was run to evaluate the effect of changing source concentration on plume length for the parameter sets given in Table 2. ------- Table 2. Model parameters Parameters Scenario Hydraulic Conductivity, Ks (ft/d) 10, 25, 50, 75, 100 Gradient 0.002 Porosity 0.25 Half-Life (days) 730, 1825 Down gradient Concentration (to define plume extent) 20 i_ig/l The source concentrations in the model runs were selected to represent fresh gasoline containing 11% and 3% MTBE. Table 1 gives the associated MTBE effective solubilities as 6220 mg/1 and 1720 mg/1. Because of expected variability in MTBE content in gasoline, and to cover a wider range of possibilities, source concentrations of 1000 mg/1 and 333 mg/1 were also simulated. The model results are summarized in Table 3. Table 3. Steady State One-Dimensional Plume Lengths Scenario Half-Life 730 days 1825 days c0 (mg/1) c0 (mg/1) c0 (mg/1) c0 (mg/1) 333 1000 1720 6220 333 1000 1720 6220 Ks = 10 ft/d 69.47 77.70 81.78 91.50 171.3 191.2 201 224.5 Ks = 25 ft/d 440.0 491.5 517.0 577.6 1071 1194 1255 1400 Ks = 50 ft/d 1750 1953 2053 2291 4253 4741 4981 5552 Ks = 75 ft/d 3911 4362 4585 5115 9522 10610 11150 12420 Ks= 100 ft/d 6917 7712 8105 9038 16870 18790 19740 21990 The results in Table 3 show that the plume length increases with increasing source concentration, hydraulic conductivity and half-life. When the conductivity was increased by a factor of 10 (from 10 ft/d to 100 ft/d) there was an almost proportional increase in plume length. Because the gradient was not varied, increasing the hydraulic conductivity results directly in an increase in ground water velocity. With higher velocity/conductivity, the plume expands farther from its source before concentrations are reduced to the concentration of 0.02 mg/1, used to define the plume. Increased source concentration is directly related to increased mass in the aquifer. By increasing the mass the plume lengths increased until the gain in mass was balanced by reduction in concentration. Plume lengths also increase with half-life because long half-lives mean that losses (degradation) are lowered Figure 2 shows that for two half-lives (730 days and 1825 days) the plume lengths differ by relatively small amounts for 3% or 11% MTBE in fuel. It also shows the consistency of this result for various values of hydraulic conductivity. By generating the solution for pairs of source concentrations the effect of a reduction in source concentration can be shown. For each pair of concentrations (6220 mg/1 and 1720 mg/1; 1000 mg/1 and 333 mg/1) the ------- reduction is about 73%. The resulting variation in plume lengths was only about 10% (10.2% to 10.6%) for each pair of simulations. This shows that the relative variation of about 10% in plume length is insensitive to the source concentration in the steady one-dimensional model. The steady state plume lengths represent the maximum extent of contamination in a fully developed plume, because of the assumed infinite-duration source. Although not included in this approach, time is required to achieve the steady state. Time is required for both transport in the aquifer and the release of a sufficient mass of contaminant. Before the steady state is achieved, or if it is not achieved, the plume is transient, i.e., concentrations in the aquifer are changing with time. During this period the plume length is shorter than that given by the steady state solution, and changes with time. 100 >. "> o 3 ¦o «Z o O o "5 m x5 >. ¦A Half Life = 730 days, 9% MTBE ¦~Half Life = 730 days, 11% MTBE ¦O Half Life = 1 025 days, 3% MTBE ¦V Half Life = 1B25days, 11% MTBE 5000 10000 15000 20000 25000 Plume Length (it) Figure 2. The effect of increased half-life (1825 days versus 730 days) on plume length. Plumes are longer with greater half-life, but as shown in Table 3, the plume lengths still vary by 10% for a given half-life. ------- Real leaks from USTs have finite sources, thus a true steady state cannot be maintained. However, if the mass is sufficient, the steady -state plume length may be achieved before the source is exhausted. Prior to this time, the plume length will be shorter than the steady state plume. In cases where the source mass is insufficient to reach the steady state distance, the plume is always shorter. In transient transport, the plume length depends on the transport parameters, as before, as well as time and the rate of release of mass from the source. The following sections address transient conditions and two-dimensional aquifer transport. Fuel Release Simulations Releases of 2000 gallons of gasoline containing 11% or 3% MTBE were simulated. The model included flow of the specified volume of fuel, dissolution of MTBE from the gasoline and transport to receptors in an aquifer. The model used was the Hydrocarbon Spill Screening Model (HSSM, Weaver et al., 1994 and Charbeneau et al., 1995). This model simulates the release of fuel and its flow to the water table, the evolution of a lens as fuel continues to be added to the subsurface, the time-dependent mass flux to the aquifer and resulting contaminant concentrations in the aquifer. The model is described in a user's guide (Weaver et al., 1994) and a background document describing the theory underlying the model (Charbeneau et al., 1995). A review of the application of the model to data from typical underground storage tank sites and the release at East Patchogue, New York was given by Weaver and Charbeneau (2000) and Weaver (1996). Because HSSM simulates the emplacement of the fuel, the contaminant source to the aquifer changes over time (is "transient") and it can reflect the nature of the fuel loading to the subsurface. For example, during a catastrophic release, the entire volume of fuel is emplaced over a short time. Dissolution from the fuel quickly increases to a maximum before being reduced relatively slowly as the contaminant is exhausted from the fuel. Chronic releases may represent situations where releases occur over several years, as might happen during an undetected release. These release of mass also increases to a maximum, but because of replenishment of the source a plateau of sorts may be reached. Once the leak is eliminated the mass flux can decrease similarly to the catastrophic release case. In either situation, the input of mass to the aquifer follows a transient pattern. This pattern depends on the hydraulics, hydrology and geology as well as the rate and duration of release. Transient Model Results An averaged measure of concentration was used for generating comparisons between transient simulation results. For receptors located at selected distances from the source, the curve of concentration versus time (called a breakthrough curve) can be averaged as follows where c' is the time-averaged concentration, % and t<, are the beginning and ending times for concentrations above a specified threshold, respectively. Simplifying this equation gives the average concentration and its duration u c'At = c'(te - tb) = jcdi (7) % This measure of contaminant level ( c' At ) will be called the integrated exposure level and will be used to compare contaminant levels resulting from different source inputs. This representation of the HSSM results as an average concentration and duration is similar in concept to approaches used in risk analysis (U.S. EPA, 1989). ------- Six HSSM runs were made using two scenarios. Three comparisons between gasoline with 3% and 11% MTBE were made from the results. The first scenario used parameters from Long Island and was chosen because the HSSM model has been successfully applied to the East Patchogue site (Weaver, 1996 and Weaver and Charbeneau, 2000). At this site and other similar sites on Long Island (Weaver et al., 1999) field data suggested that biodegradation of MTBE was limited, both by study of field data and from calibration of the model. Thus a half-life of 10 years was chosen to represent a low rate of degradation. The Long Island scenario was also characterized by a high ground water flow velocity. The impact of these two assumptions on the contaminant was to produce simulated plumes of large extent that became detached from their sources as the MTBE is depleted from the gasoline. The concentrations, durations and extents are compared below for simulations that differ only in the MTBE content of the gasoline. To provide a contrasting conceptualization, a generic alluvial scenario was simulated with both 10 year and 2 year half-lives. The hydraulic conductivity of the alluvial scenario was 1/10th that of the Long Island scenario. The 2 year half-life was included because, in contrast to the conditions on Long Island, degradation of MTBE might occur at higher rates in other hydrogeologic settings (Wilson, 2000). Long Island Scenario Simulation Tables 4 through 6 show results from an HSSM run for the Long Island Scenario for a catastrophic release of 2000 gallons of gasoline. Results are given for locations ranging from 100 feet to 6000 feet from the source. For each location, the peak concentration, its arrival time, pulse duration, average concentration, integrated exposure level, plume length and maximum contamination extent are given in the tables. These values are given for simulations with bothll% and 3% MTBE fuel contents. The percent difference between the two simulations is given and denoted by %A, which is calculated from [value, - valuev, I . , %A = 100-i — ^ (8) valuell% where valueno/o and value3»/o are the results from the 11% and 3% MTBE simulations, respectively. Each of the model results was based only on concentrations above 20 fig/1. This level was chosen to correspond to the lower limit of EPA drinking water advisory. All of the peak concentration arrival times in Table 4 are close together and represent essentially the same value. The peak arrival times differ by -1% to 0.72% between the two simulations. The differences were caused by the numerical procedure that is used to determine the peak concentration. The model only requires the peak to be found to within a certain tolerance, which can result in a several-day-long target period. The corresponding peak concentrations dropped by about 73% for each location when MTBE was reduced from 1 1% to 3%. This reduction was very close to the MTBE reduction assumed for the gasoline (72.7% = 100 x (0.11 - 0.03)/0.11 ). Table 5 shows the comparison between the duration of concentration above 20 i_ig/l at each location and the average concentration over that duration. The simulated duration differences are less than proportionate to the change in fuel MTBE concentration since they only range from 9% to 19%. The average pulse concentration over this time period calculated using equation 6 gave a range of difference in concentration of 67.0% to 70.6%. Although not as close as the reduction in peak concentration values, these are still close to the proportion of fuel MTBE reduction. The differences between these values and the proportionate decrease in MTBE content are small and most likely related to numerical errors inherent in evaluating equation 6. These results indicate that a proportionate reduction in fuel MTBE content propagates through both the peak and average pulse concentrations. By combining pulse-duration and average concentration into an "integrated exposure level" (equation 7), the overall change in exposure can be assessed. Table 5 gives the integrated exposure levels and the percent difference between simulations with 1 P/o and 3% MTBE in gasoline at each location. The integrated exposure level dropped by about 73% (Table 5, column 10), somewhat more than the 67% to 71% reduction in average ------- concentration alone (Table 5, column 7). The reduction in pulse duration accounts for the remaining 3% to 6% of change in exposure. Clearly, the reduction in average pulse concentration dominates the change in simulated exposure at the selected locations. Table 4 Long Island (3650 day half-life) Catastrophic Release Distance from Source Arrival Time Peak Concentration (ft) (Days) (mg/1) 11% 3% %A 11% 3% %>A 100 24.3 24.2 0.41 1320 360 72.7 250 77.3 77.4 -0.13 381 104 72.7 500 186 185 0.54 149 40.7 72.7 750 299 302 -1.00 89.4 24.4 72.7 1000 420 420 0.00 62.9 17.1 72.8 1500 660 658 0.30 38.5 10.5 72.7 2000 898 898 0.00 27.1 7.39 72.7 3000 1380 1370 0.72 16.3 4.42 72.9 4000 1860 1850 0.54 11.1 3.00 73.0 5000 2330 2330 0.00 8.07 2.18 73.0 6000 2810 2805 0.18 6.13 1.66 72.9 ------- Table 5 Long Island (3650 day half-life) Catastrophic Release Distance from Source (ft) MTBE Content Pulse Duration (Days) 11% 3% %A Average Concentration (mg/1) 11% 3% %A Integrated Exposure Level (days mg/1)/1000 11% 3% % A 100 780 633 18.9 89.8 29.6 67.0 70.0 18.7 73.2 250 934 788 15.6 45.9 14.5 68.4 42.9 11.4 73.3 500 1110 963 13.2 26.3 8.14 69.0 29.2 7.84 73.1 750 1260 1110 11.9 18.3 5.57 69.6 23.1 6.18 73.2 1000 1380 1230 10.9 13.9 4.21 69.7 19.2 5.18 73.0 1500 1640 1430 12.8 9.19 2.79 69.6 15.1 3.99 73.5 2000 1790 1600 10.6 6.88 2.05 70.2 12.3 3.28 73.4 3000 2070 1880 9.18 4.39 1.30 70.4 9.09 2.44 73.1 4000 2320 2110 9.05 3.10 0.91 70.6 7.19 1.92 73.2 5000 2530 2310 8.70 2.31 0.68 70.5 5.84 1.57 73.1 6000 2710 2470 8.86 1.80 0.53 70.6 4.88 1.31 73.2 Table 6 shows plume lengths and the maximum plume extent for a period of 9 years after the end of the release. The simulation results showed that at early times the MTBE plume was attached to the source. Later the plumes detached from the sources and migrated as pulses in the aquifer. The leading edge of the plumes or pulses marks the maximum extent of the contamination at or above the threshold concentration of 20 fig/1. Thus all locations closer to the source than the maximum extent of the plume would have seen MTBE contamination at levels of 20 i_ig/l or higher at some time during the episode. For the Long Island simulations, the maximum plume extent differed by 3.25% to 4.51% over the nine year simulation period (Table 6, column 9). In this case the maximum difference decreased with time, indicating that the 11% and 3% cases were becoming closer together. In contrast, the difference in plume (or pulse) length increased to from 4.51% to 14.37% over the same time period (Table 6, column 8). Because the leading edges were becoming closer together, the increasing difference between plume lengths was due to the difference in the trailing edges. Note that before the plume detached the difference in plume length was the same as the difference in the plume extent. ------- Table 6 Long Island Scenario Plume Lengths* Y ears After Release 11% MTBE Plume dimensions relative to the source 3% MTBE Plume dimensions relative to the source % Difference in Plume Length % Difference in Maximum Extent Trailing Edge (ft) Leading Edge (ft) Length (ft) Trailing Edge (ft) Leading Edge (ft) Length (ft) 1 - 1706 1706 - 1629 1629 4.51 4.51 2 108 2809 2701 347 2692 2345 13.18 4.17 3 590 3813 3223 888 3664 2776 13.87 3.91 4 1168 4767 3599 1482 4589 3107 13.67 3.73 5 1780 5690 3910 2105 5485 3380 13.55 3.60 6 2413 6592 4179 2751 6363 3612 13.56 3.47 7 3060 7475 4415 3412 7221 3809 13.73 3.40 8 3719 8343 4624 4085 8067 3982 13.88 3.31 9 4387 9201 4814 4780 8902 4122 14.37 3.25 * Both leading and trailing edges of the plume are determined at points where the simulated MTBE concentrations are 0.02 mg/1. Alluvial Scenario Simulation The results for the alluvial scenario, using degradation half-lives of 2 years and 10 years were similar to the Long Island simulations except that the plume lengths were shorter than for the Long Island scenario. Longer simulation periods were used to compare plume length and extent, because the shorter plumes stayed closer to the source over long periods of time. As before, the peak arrival times are close together and the peak and average concentrations are in proportion to the reduction of MTBE in gasoline for simulations with both half-lives. Differences in pulse duration were similar for the two cases, where the range of values was about 6% to 9% for each. The integrated exposure level (duration times average concentration) was reduced by about 73% in each case, again dominated by the reduction in average concentration. Reduction in pulse duration accounted for up to 3% of the this total difference. Larger differences between the two fuel contents appeared for the plume lengths than in the Long Island scenario. For the 10 year half-life simulation, the maximum difference was 8.86%, while that for the 2 year half-life was 21.0%. For the 2 year half-life simulation, the maximum extent of contamination was smaller in magnitude but showed the greater difference (21.0%) between the differing MTBE contents. Presumably this result was due to the impact of increased biodegradation. Summary of Modeling Results For every case simulated, the reduction in the peak and average pulse concentrations were almost directly proportional to the reduction in MTBE content in the source (Table 7). These results show that the reduction appears in both the peak and average concentration observed within the aquifers. This universal reduction in ------- concentration with reduced MTBE content also dominated the response of the integrated exposure levels (pulse duration times average pulse concentration), because the effect of shortened duration reduces the exposure level by only a small amount. The maximum plume extent was reduced by a only a few percent (3% to 6%) showing that the plumes reach similar distances regardless of the MTBE content of the gasoline. Table 7 Summary of Percent Differences in Plume Measures Scenario Concentration Pulse Duration Exposure Impact (Duration x Average Cone.) % Length Peak Average Plume/ pulse length % Maximum Extent % MTBE Half-Life (yr) Release Type % % % Long Island 10 catastrophic 72.7 - 73.0 67.0 70.6 18.8-8.7 73.0 -73.5 4.51- 14.37 4.51- 3.25 Alluvial 10 catastrophic 72.7 - 72.8 69.8 70.8 9.34-6.11 72.6 - 72.9 4.15 - 8.65 4.15 - 3.23 2 catastrophic 72.6 - 72.8 67.1 71.1 9.78-5.81 69.8 - 72.8 4.47- 21.00 4.47- 5.94 Conclusions The comparative modeling study showed that reducing MTBE concentrations in gasoline resulted in nearly proportionate reductions in peak and average MTBE concentration in ground water. As a consequence, exposures to MTBE contamination would be reduced due to the reduction in concentration. These reductions would be nearly proportionate to the reduction of MTBE in gasoline. Plume lengths, however, as determined by both a steady-state model and the Hydrocarbon Spill Screening Model (HSSM) were not reduced proportionately. In some cases the reduction in extent was relatively small, so that many of the same receptors would be impacted by plumes originating with 11% or 3% MTBE in gasoline. Disclaimer This paper has been reviewed in accordance with the U.S. Environmental Protection Agency's peer and administrative review policies and approved for presentation and publication. References Charbeneau, R.J., J.W. Weaver, and B.K. Lien, 1995, The Hydrocarbon Spill Screening Model (HSSM) Volume 2: Theoretical Background and Source Codes, US EPA, EPA/600/R-94/039b. Cline, P. V., J.J. Delfino, and P.S.C. Rao, 1991, Partitioning of aromatic constituents into water from gasoline and other complex solvent mixtures, Environmental Science and Technology, 23, 914-920. MacKay D., and R.S. Boethling, 2000, Handbook of Property Estimation Methods for Chemicals, lewis Publishers, Boca Raton, Florida, 481pp. United States Environmental Protection Agency, 1989, Risk Assessment Guidance for Superfund Volume 1, Human Health Evaluation Manual (Part A) Interim Final, US EPA, EPA/540/1-89/002. ------- van Genuchten, M. T., and J. Alves, 1982, Analytical Solutions of the One-Dimensional Convective-Dispersive Solute Transport Equation, USDA Agricultural Research Service, Technical Bulletin 1661. Weaver, J.W., and R. J. Charbeneau, 2000, A Screening Approach to Fuel Hydrocarbon Spill Assessment, Chapter for "Subsurface Remediation: Modeling and Design for Organic Contaminants, ASCE Monograph, in press, February 2000. Weaver, J.W., R.J. Charbeneau, J.D. Tauxe, B.K. Lien, and J.B. Provost, 1994, The Hydrocarbon Spill Screening Model (HSSM) Volume 1: User's Guide, US EPA, EPA/600/R-94/039a. Weaver, J.W, 1996, Application of the Hydrocarbon Spill Screening Model to Field Sites, Procedings of Non- Aqueous Phase Liquids (NAPLs) in Subsurface Environment: Assessment and Remeidation, ed. L. Reddi, American Society of Civil Engineers, Washington, D.C., November 12-14, pp. 788-799, (see http: //www. ep a. gov/ada/patcho sue. html). Weaver, J.W., J.E. Haas, C.B. Sosik, 1999, Characteristics of Gasoline Releases in the Water Table Aquifer of Long Island, Proceedings of 1999 Petroleum Petroleum Hydrocarbons and Organic Chemicals in Ground Water, API/NGWA, Houston, Texas, November 17-19, pp. 262. Weaver, J.W. and C. Sosik, 2000, OnSite Site Assessment Guide, US. EPA, in preparation. Wilson J.T., 2000, Personal Communication. Xu, M., and Y. Eckstein, 1995, Use of weighted least-squares method in evaluation of the relationship between dispersivity and scale, Groundwater, 33(6), 905-908. ------- |