Science of the Total Environment 485–486 (2014) 705–710
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Ethanol effects on the fate and transport of gasoline constituents in the UK Simon Firth a,⁎, Beate Hildenbrand b, Phil Morgan c a b c
Firth Consultants Ltd, Windsor House, Greville Road, Bristol, BS3 1LL, UK Energy Institute, 61 New Cavendish Street, London, W1G 7AR, UK The Sirius Group, Red Hill House, Hope Street, Chester, CH4 8BU, UK
H I G H L I G H T S • • • •
Monte Carlo analysis used to assess risk to groundwater from ethanol blended gasoline Compared predicted benzene/MtBE plume lengths with/without ethanol in gasoline Compared predicted number of abstractions at risk with/without ethanol in gasoline Results show similar risk to UK groundwater with and without ethanol in gasoline
a r t i c l e
i n f o
Article history: Received 25 June 2013 Received in revised form 14 November 2013 Accepted 27 February 2014 Available online 15 March 2014 Keywords: Fate and transport modelling Groundwater Ethanol Gasoline Environmental risk
a b s t r a c t In the UK, use of ethanol in fuel as a fuel oxygenate/fuel supplement is currently limited but could rise in an effort to meet the requirements of the European “Biofuels” Directive. This Energy Institute study focussed on the risk that accidental releases of ethanol blended gasoline (EBG) (i.e. gasoline containing 10% or less of ethanol) could pose to UK groundwater resources. Ethanol is miscible and highly biodegradable. As a result it tends to be strongly attenuated in the unsaturated zone and in groundwater and so does not, in itself, pose a significant risk to groundwater resources. However, it may lead to increased persistence of other gasoline constituents, particularly through alteration of geochemical conditions as a result of intensive biodegradation activity. A semi-probabilistic modelling exercise was undertaken to better understand the risks that use of EBG could pose to UK groundwater resources. Site investigation information from over 500 filling stations was used in combination with GIS data to predict the proportion of potable water supply wells that could potentially be impacted by benzene and MtBE, and estimate the length of benzene and MtBE plumes with and without the use of ethanol in gasoline. The results show that the use of EBG is likely to have a negligible effect on MtBE plumes. Some increase in benzene plume length is predicted, most notably in fissured aquifers, but increases in plume length of greater than 30% are predicted to be rare. A corresponding slight increase in risk to licensed potable water supply wells from benzene was predicted with the use of EBG but the percentage of wells at risk was still predicted to be small (0.13%), and in the context of the conservatism within the modelling, it was concluded that widespread use of EBG is unlikely to cause an increased risk to UK water resources. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Ethanol is broadly being used as both a fuel oxygenate in reformulated gasoline and as an alternative fuel/fuel supplement. Currently in the UK, the use of ethanol for this purpose is limited but could rise in an effort to meet the requirements of the European
⁎ Corresponding author at: Firth Consultants Ltd, Windsor House, Greville Road, Bristol, BS3 4QF, UK. E-mail addresses: simon@firthconsultants.co.uk (S. Firth), [email protected] (B. Hildenbrand), [email protected] (P. Morgan).
http://dx.doi.org/10.1016/j.scitotenv.2014.02.119 0048-9697/© 2014 Elsevier B.V. All rights reserved.
“Biofuels” Directive (EC, 2003). Whilst the use of biofuels has the potential to cut carbon dioxide emissions, the potential for environmental risks should not be discounted. In particular, experience from use of other fuel additives such as MtBE has demonstrated the importance of understanding risks to water resources from accidental releases of fuel. When considering the risks from ethanol-containing fuel it is important to distinguish between ethanol-blended gasoline (EBG) such as E5 or E10 (gasoline containing 5 or 10% v/v ethanol) and fuel grade ethanol (FGE), i.e. fuel containing 85% ethanol or more (including denatured ethanol). In the UK the vast majority of ethanol used as road transport fuel is sold as EBG and this situation is unlikely to change in the future. As a result, releases of ethanol-containing fuel will most commonly be
706
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
associated with leaks or spills of EBG from filling stations. The likelihood of such releases occurring is reducing with time as a result of widespread replacement or decommissioning of old infrastructure and other improved environmental measures. This paper presents work conducted for the Energy Institute to better understand the risks that use of ethanol as a gasoline constituent could have on UK water resources (Energy Institute, 2012). The paper gives a brief summary of literature on the fate and transport of ethanol in the subsurface and its effect on the persistence of other gasoline constituents, and then goes on to describe modelling work conducted to assess the risk that increased use of EBG could pose to UK groundwater resources. 2. The fate of ethanol in the subsurface and its effect on other gasoline constituents Ethanol is miscible with water and, relative to petroleum hydrocarbons such as benzene, has a low Henry's law constant, indicating that it has a relatively low volatility from the dissolved phase (Table 1). As a result, when released to the subsurface, the majority of ethanol will partition to the aqueous phase. For small volume releases above the water table this can result in a significant proportion of the released ethanol being retained in solution within pore water in the unsaturated zone, significantly reducing the impact it has on groundwater (Freitas and Barker, 2011; Lahvis, 2003). Ethanol-containing fuel released at or below the water table, or large volume releases that occur above the water table, can result in ethanol reaching groundwater and forming a dissolved-phase plume. Once in groundwater, ethanol is typically rapidly biodegraded and so is unlikely to create a risk to groundwater resources in itself. However ethanol's strong polarity and the relative ease with which it is biodegraded can affect the risk caused by other gasoline constituents. Ethanol's strong polarity means that it can increase the aqueous solubility of other gasoline constituents as a cosolvent. However, in practice, the concentration of ethanol required for cosolvency effects to become significant (N80,000 mg L−1, Heermann and Powers, 1998) is unlikely to occur other than in exceptional circumstances such as the groundwater source zone beneath a large volume release of FGE. Ethanol's strong polarity can also theoretically result in increased mobility of non-aqueous phase liquid (NAPL) hydrocarbons (McDowell et al, 2003), but again this effect is only ever likely to become significant for large spills of FGE itself. Tank-scale tests have shown that spills of FGE onto pre-existing NAPL have the potential to reduce capillary fringe thickness allowing subsequent re-distribution of NAPL closer to the water table (Yu et al, 2009). However, field evidence for this effect is lacking. Ethanol's rapid biodegradation in groundwater can affect the persistence of other gasoline constituents through alteration of geochemical conditions. Biodegradation of ethanol quickly uses up available electron acceptors and can lead to the onset of methanogenic conditions (Jiang et al, 2009; Cápiro et al, 2008; Dakhel et al., 2003). These changes can reduce the biodegradation rates of other gasoline constituents. Field studies have shown some evidence of a reduction in biodegradation rates of BTEX compounds resulting in plume elongation (MacKay et al., 2006; Freitas et al, 2011; Corseuil et al., 2011).
Various modelling studies have been conducted to assess the effect that ethanol in EBG could have on the persistence of benzene in groundwater (e.g. McNab et al., 1999; Shen et al., 2001; Molson et al., 2002; Gomez and Alvarez, 2009). The results from these studies vary significantly, with the predicted increase in benzene plume length ranging from 5% to 635%, although predicted increases of 30% to 100% are more typical (Energy Institute, 2012). All of these studies assume that the groundwater source zone contains a significant concentration (e.g. N3 mg L−1) of ethanol and that this persists for some time (e.g. greater than 1 year). However, as discussed above, attenuation in the unsaturated zone and ethanol's rapid biodegradation is likely to limit concentrations and longevity of ethanol in groundwater and this assumption is likely to be highly conservative for releases of EBG from gasoline retail sites. This is supported by experience from California, where EBG has been widely used since 2002. A review of groundwater monitoring data collected from 207 gasoline retail sites from that state showed that ethanol had rarely been detected in groundwater, with only 21 of 13,000 samples recording a concentration of ethanol greater than 1 mg L−1 (API, 2010). These previous modelling studies have also tended to focus on a single field study or hypothetical case, where one set of aquifer properties with one set of source zone concentrations have been modelled. They do not attempt to model the variability in benzene plume length caused by variability in source zone concentrations and aquifer properties. The modelling exercise presented in Section 3 attempts to predict the potential impact of use of EBG on UK groundwater resources as a whole, taking account of variability in parameters such as source zone dimensions and concentrations, aquifer properties and abstraction rates. 3. Predictive modelling A modelling study was undertaken to assess the potential for risk to groundwater resources in the UK from the use of EBG. This consisted of two components: (1) modelling the proportion of potable water supply wells in England and Wales potentially impacted by benzene and MtBE with and without the use of ethanol in gasoline; and (2) modelling the distribution of plume lengths of benzene and MtBE in UK groundwater with and without the use of ethanol in gasoline. The model approach and results for both components are discussed below. 3.1. Model approach 3.1.1. Modelling the proportion of potable water supply wells potentially impacted A simple capture zone analysis was undertaken using a GIS containing the locations of 10378 gasoline retail sites and 3616 licensed abstraction wells in England and Wales (note that the required information was not available for Scotland). This identified 2536 gasoline retail sites that could plausibly pose a risk to 1191 licensed abstraction wells, 679 of which are used for potable water supply. An ExcelTM spreadsheet model was then developed to estimate the concentrations of MtBE and benzene in each of these abstraction wells from hypothetical sources at the 2536 gasoline retail sites. The abstracted concentrations were calculated using the following methodology: 1. The mass flux of benzene, MtBE and ethanol leaving the source zone at each gasoline retail site were calculated using Eq. (1):
Table 1 Comparison of physico-chemical properties of ethanol and benzene. Physical property
Ethanol
Benzene
CAS number Chemical formula Molecular weight (g mol−1) Solubility in water @ 20 °C (mg L−1) Vapour pressure @ 20 °C (Pa) Dimensionless Henry's law constant Organic carbon partition coefficient (L kg−1)
64-17-5 C2H5OH 46.07 Miscible c 5900 c 2.1 × 10−4 d 1.58 e
71-43-2 C6H6 78.11 1780 a 10,000 b 0.116 a 67.6 a
a. Environment Agency, 2008; b. Lide, 2008; c. European Chemicals Bureau, 2000; d. Malcolm Pirnie Inc., 1998; e. Risk Assessment Information System, 2011.
M sz ¼ K:i:W:b:C sz
ð1Þ
Where,
Msz K i W
mass flux of contaminant leaving source zone (g d−1) hydraulic conductivity of groundwater bearing unit (m d−1) hydraulic gradient in groundwater bearing unit (m m−1) width of source zone (m)
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
b Csz
thickness of source zone (m) dissolved phase concentration of contaminant in source zone (mg L−1)
2. The mass fluxes of MtBE and benzene arriving at each abstraction well were then calculated using Eq. (2): −λ:t c
Mr ¼ Msz :e
ð2Þ
Where mass flux of contaminant arriving at receptor (g d−1) mass flux of contaminant leaving source zone (g d−1) contaminant degradation decay rate (d−1) contaminant transit time between source zone and receptor (d)
Mr Msz λ tc
Note that this method does not account for attenuation by dilution or dispersion, which are expected to be negligible relative to degradation within the aquifer. The method assumes a non-diminishing source. As discussed in the uncertainty analysis below, this assumption is conservative. The contaminant transit time from the source zone to the abstraction well is calculated from the groundwater transit time (tgw) multiplied by the retardation factor (Rf). The groundwater transit time is calculated using Eq. (3): t gw ¼
π:B:ne :r 2 Qw
ð3Þ
Where tgw B ne r Qw
groundwater transit time to the receptor (d) saturated aquifer thickness (m) effective porosity (dimensionless) distance to receptor (r) abstraction rate from well (m3 d−1)
This method is based on Darcy's law with the following assumptions: • the principal flow path is via the aquifer where abstraction occurs; • the drawdown in the abstraction well is negligible compared to the saturated thickness of the aquifer; • the aquifer is isotropic and homogenous; • dilution via infiltration is negligible; and • flow towards the wells is radial. The retardation factor is given using Eq. (4): Rf ¼ 1 þ
ρd :K oc :f oc ne
ð4Þ
Where ρb Koc foc
aquifer bulk density (g cm−3) contaminant organic carbon partition coefficient (L kg−1) fraction of organic carbon in aquifer (dimensionless)
Ethanol's effect on the persistence of MtBE and benzene was modelled using the USEPA ‘Footprint’ model approach (USEPA, 2008). This is a simplistic method that assumes that the biodegradation of other gasoline constituents does not occur where the concentration of ethanol is above a threshold concentration (assumed to be 3 mg L−1). This is a conservative assumption as field evidence suggests that the biodegradation of other gasoline constituents may be slowed but is unlikely to be prevented where ethanol is present (Mackay et al., 2006; Freitas et al, 2011; Corseuil et al., 2011). The model predicts the length of ethanol
707
plume dictated by the threshold concentration and calculates the remaining distance to the abstraction well. This distance (r) is used in Eq. (3) above to determine the available time for degradation of benzene and MtBE. Thus, the longer the ethanol plume, the less time there is available for benzene and MtBE degradation to occur and the higher the concentrations of these contaminants will be in the abstracted water at the well. 3. Each abstraction well in the model has one or more gasoline retail sites potentially contributing benzene and MtBE to it. The fluxes from these sites are summed to calculate the total mass fluxes of MtBE and benzene arriving at each well. The concentrations of MtBE and benzene in each well are then calculated by dividing these mass fluxes by the well’s licensed abstraction rate. Investigations at over 500 Energy Institute member sites in the UK have shown a large variability in the extent and magnitude of petroleum hydrocarbon contamination in soils and groundwater (Energy Institute, 2009). Whilst many sites have maximum concentrations in soil or groundwater at or below analytical detection limits, there are some where near saturation concentrations have been detected as a result of the presence of free-phase fuel within the subsurface. Given that source concentration data were not available for the 2536 filling stations considered, a probabilistic approach was adopted for the estimation of mass flux leaving each gasoline retail site (Eq. (1)). Probability distribution functions (PDFs) of groundwater source concentration, source zone dimensions and the product of hydraulic conductivity and gradient were used in a Monte-Carlo analysis to model the mass flux of benzene, ethanol and MtBE leaving each of the identified filling stations. The PDFs for source dimension, hydraulic properties and source concentrations of MtBE were based on data gathered from the 500 + Energy Institute member sites from a previous study (Energy Institute, 2009), whilst the PDFs of source concentration of benzene and ethanol were based on available data from the literature and expert judgement and assuming that EBG was in common usage in the UK. The parameters required for Eqs. (2), (3) and (4) were modelled deterministically. Best estimates of central tendency values were used for the biodegradation decay rates of benzene, MtBE and ethanol, based on literature sources. Aquifer maps were used in the GIS to identify likely aquifer properties (saturated aquifer thickness, fraction of organic carbon and effective porosity) for each abstraction well. The GIS was also used to calculate the distance from each gasoline retail site to its associated abstraction well. The Monte Carlo analysis was run for 2000 simulations. The percentages of abstraction wells with concentrations of benzene and MtBE above threshold concentrations of 1 μg L−1 for benzene (the UK drinking water standard) and 15 μg L−1 for MtBE (the WHO taste threshold) were calculated for each simulation and used to produce a distribution of these percentages for all simulations. This process was undertaken with and without ethanol to assess the impact the use of EBG could have on risk to abstraction wells. 3.1.2. Modelling the length of benzene and MtBE plumes A separate model was developed to assess the effect of ethanol on the plume lengths of benzene and MtBE in groundwater beneath gasoline retail sites. As with the abstraction well model, this model is part probabilistic and part deterministic. Plume length is defined by the distance from the groundwater source to the point down hydraulic gradient at which the contaminant concentration equals a threshold concentration and is calculated using the Domenico equation for steady state conditions (Eq. (5)). Again, this method assumes a non diminishing source, which is a conservative assumption. ( C x ¼ C O exp
x 2ax
( ) rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!) Sy 4λax Sz erf pffiffiffiffiffiffiffi erf pffiffiffiffiffiffiffi u 2 ay x 4 az x
1− 1 þ
ð5Þ
708
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
Table 2 Predicted occurrence of benzene and MtBE in licensed potable water supply wells with and without use of ethanol. Contaminant
Threshold concentration
1 μg L−1 15 μg L−1
Benzene MtBE
Average percentage of wells with concentration exceeding threshold value Without ethanol
With ethanol
0.03% 0.33%
0.13% 0.33%
Where concentration at distance ‘x’ downstream of source zone (mg L− 1) source zone concentration (mg L−1) distance downstream from source zone (m) longitudinal dispersivity coefficient (m) (assumed equal to 0.1x) transverse dispersivity coefficient (m) (assumed equal to 0.01x) vertical dispersivity coefficient (m) (assumed equal to 0.001x) decay rate (d−1) contaminant velocity (m d−1) source zone width (m) source zone thickness (m)
Cx Co x ax ay az λ u Sy Sz
Contaminant velocity was estimated using Eq. (6):
u¼
Ki ne :R f
ð6Þ
Where Ki
product of hydraulic conductivity and hydraulic gradient (m d− 1) aquifer effective porosity (dimensionless) contaminant retardation factor (dimensionless)
Table 4 Modelled plume lengths—fractured bedrock aquifers. Contaminant
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m) Median
75th percentile
90th percentile
95th percentile
4.2 4.4 6% 48 54 12%
41 46 12% 184 197 7%
138 157 14% 464 492 6%
236 280 19% 794 797 0%
of dispersion, and the model will therefore tend to overestimate plume extension. As with the abstraction well model, source dimensions and concentrations, hydraulic conductivity and gradient were modelled probabilistically. Contaminant degradation decay rate and Koc were modelled as single (deterministic) values for each contaminant. The modelling was undertaken for a range of aquifer types, with generic values of aquifer thickness, effective porosity and fraction of organic carbon assigned to each aquifer type consistent with the values used for the abstraction well modelling. 3.2. Model results 3.2.1. Proportion of potable water supply wells potentially impacted The predicted percentages of licensed potable abstraction wells in England and Wales with benzene and MtBE above the threshold concentrations with and without the use of ethanol in gasoline are presented in Table 2 below. This shows that the use of ethanol is not expected to have a significant effect on the occurrence of MtBE in abstractions wells, with a predicted 0.33% of abstraction wells having N 15 μg L− 1 MtBE with or without the use of EBG. The percentage of wells with N1 μg L−1 benzene is predicted to be slightly higher (0.13%) with the use of ethanol than without (0.03%).
The length of plume was found by varying distance ‘x’ iteratively until ‘Cx’ is equal to a threshold concentration. Again, threshold concentrations of 1 μg L−1 and 15 μg L−1 were chosen for benzene and MtBE, respectively. A threshold concentration of 3 mg L−1 was assumed for ethanol, which, as discussed above, is taken as the concentration above which degradation of benzene and MtBE is assumed not to occur. Thus, the plume length of ethanol is the predicted distance from the site that groundwater must travel before benzene and MtBE degradation can occur. This length is then added to the estimated benzene and MtBE plume lengths (without ethanol) to estimate the length of benzene and MtBE plumes in the presence of ethanol. This simplified approach is conservative because it ignores the fact that benzene and MtBE concentrations will reduce within the ethanol plume as a result
3.2.2. Benzene and MtBE plume lengths Tables 3 to 5 compare predicted median and upper percentile plume lengths for releases of gasoline with and without use of ethanol for three different aquifer types: (1) unconsolidated alluvial deposits (e.g. sand and gravels), (2) fractured bedrock aquifers (such as the Cretaceous Chalk, Jurassic Oolite and Magnesian Limestone), and (3) intergranular bedrock aquifers (such as the Lower Cretaceous Sands and Permian–Triassic Sandstone). The distributions of modelled plume lengths with and without ethanol are also compared graphically in Figs. 1 to 3. The model predicts a slight increase in median plume length for MtBE with use of ethanol in gasoline, but ethanol's effects on longer plumes is negligible. Given that the median plume lengths are relatively short (b50 m), ethanol's overall effect on risk from MtBE is likely to be negligible. This finding is consistent with the abstraction well modelling. Ethanol is predicted to have a greater influence on benzene plume length, with a 9% increase in median plume length for all aquifer types.
Table 3 Modelled plume lengths—alluvial aquifers.
Table 5 Modelled plume lengths—intergranular bedrock aquifers.
ne Rf
Contaminant
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m)
Contaminant
Median
75th percentile
90th percentile
95th percentile
0.11 0.12 9% 16 17 7%
1.8 2.2 23% 88 90 2%
12 16 31% 240 242 1%
40 53 33% 391 395 1%
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m) Median
75th percentile
90th percentile
95th percentile
0.37 0.40 9% 13 15 13%
5.8 6.3 8% 80 86 8%
39 44 12% 219 224 2%
80 89 12% 341 345 1%
1400 1200 1000 800 600 400 200 0
Without ethanol With ethanol
Frequency
Frequency
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
800 700 600 500 400 300 200 100 0
Benzene plume length (m)
709
Without ethanol With ethanol
MtBE plume length (m)
Fig. 1. Frequency distributions of modelled benzene and MtBE plume lengths in alluvium aquifers.
However, given that median benzene plume lengths are less than 5 m, ethanol's predicted effect on plume length can be considered negligible for the majority of sites. Ethanol's effect is more pronounced for the upper percentile benzene plumes, with approximately 12 to 33% increases in the 95th percentile plume lengths. The largest percentage increases occur for the alluvial aquifer type. This has the highest fraction of organic carbon content and consequently has the highest ratio of benzene to ethanol retardation coefficients. As a result, the non-retarded steady state plume of ethanol causes the greatest impact on overall benzene degradation for this aquifer type. In reality, for diminishing sources, the retardation contrast will typically result in the ethanol plume detaching from the benzene plume. Under such circumstances, ethanol's effect on the persistence of benzene will be reduced, i.e. the model likely over-predicts ethanol's effects on benzene plume lengths for alluvial aquifers. 3.3. Uncertainty analysis Key uncertainties in the modelling are discussed below.
800 700 600 500 400 300 200 100 0
3.3.2. Model parameters Sensitivity analysis was conducted to assess the effects of uncertainty in key model input parameters on model results. Parameters tested were the threshold concentration for ethanol, the PDF for the groundwater source concentration for ethanol and the biodegradation decay rates of benzene and ethanol. This analysis showed that uncertainty in these parameters caused relatively little variation in the model outputs. 4. Conclusion The modelling conducted for this study provides a useful indicator of the potential risks to UK groundwater resources from use of EBG. A probabilistic approach has been adopted to account for the high degree of variability between sites, especially with regard to source concentration. The results of this exercise indicate that the widespread use of EBG such as E5 or E10 is unlikely to cause a significant increase in risk to UK groundwater resources. For the majority of filling stations, plume lengths of benzene are unlikely to increase by more than 10%. Greater increases in plume length may occur for longer plumes in alluvial and fracture flow aquifers, but increases above 30% are predicted to be rare. Furthermore, field evidence suggests that any elongation of plumes will be temporary. The increase in benzene plume lengths is predicted to have some impact on the risk to licensed potable water supply wells. However,
Without ethanol With ethanol
Benzene plume length (m)
Frequency
Frequency
3.3.1. Model approach The model approach is based on the assumption that biodegradation of benzene and MtBE will not occur in groundwater where ethanol is present above a threshold concentration. Field studies of releases of EBG have shown that, whilst biodegradation rates of benzene may decline as a result of the onset of methanogenic conditions, biodegradation of benzene in the presence of ethanol does still occur. There is even evidence to suggest that, following depletion in concentrations of ethanol and its breakdown intermediates, benzene degradation rates may be enhanced due to increased microbial biomass (Corseuil et al., 2011). There is also evidence that the presence of ethanol itself may stimulate biodegradation of MtBE to tert-butyl alcohol (TBA) (Mackay et al., 2007). Thus, the assumption of no degradation of benzene and MtBE where ethanol is present above a threshold concentration is likely to be conservative and tend towards the over-prediction of risk. Another source of potential conservatism within the model is the assumption of steady state conditions. The groundwater source concentrations of ethanol, benzene and MtBE are assumed to remain constant
and not decrease with time. In reality, source concentrations will tend to decrease with time as contaminants are removed from the site by biodegradation, volatilisation and/or via advective transport in groundwater. In particular, persistent sources of ethanol are not expected due to its miscibility with water and rapid biodegradation. Field evidence shows that ethanol rarely persists in groundwater for more than a few months. The low retardation and short source duration of ethanol relative to hydrocarbons can also result in the ethanol plume detaching from the hydrocarbon plumes such that it no longer has any impact on hydrocarbon degradation in groundwater. This effect was observed in the controlled field releases of E10 and E95 at the Borden site in Canada (Freitas et al., 2011). The assumption of steady state conditions is therefore likely to be conservative and tend towards an over-prediction of risk.
800 700 600 500 400 300 200 100 0
Without ethanol With ethanol
MtBE plume length (m)
Fig. 2. Frequency distributions of modelled benzene and MtBE plume lengths in fissure aquifers.
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
Frequency
1200 1000
Without ethanol With ethanol
800 600 400 200 0
Benzene plume length (m)
Frequency
710
800 700 600 500 400 300 200 100 0
Without ethanol With ethanol
MtBE plume length (m)
Fig. 3. Frequency distributions of modelled benzene and MtBE plume lengths in intergranular bedrock aquifers.
although a slight increase is predicted, the predicted proportion of wells impacted is still small (0.13%). Furthermore, given that a number of conservative assumptions have been used for the modelling and that the frequency and magnitude of leaks from filling stations in the UK continues to decrease as a result of continuing improvements to infrastructure, the model results are likely to over-predict actual impacts from the future use of EBG. Thus, it is concluded that use of EBG is unlikely to cause a significant increase in risk to groundwater resources in the UK. Acknowledgments The authors would like to thank the Energy Institute's Soil, Waste and Groundwater Group who commissioned the work and, in particular, the steering group members for their valuable contribution to the work, namely Ruth Chippendale (Shell), Martyn Dunk (ExxonMobil), Chris Hughes (Chevron), Martyn Lambson (BP) and Fraser Will (Total). We are also grateful to Brent Stafford from Shell, Jim Barker from the University of Waterloo, Canada and Steve Thornton from University of Sheffield, UK for their assistance. References American Petroleum Institute (API). Management of ethanol-blended fuel releases: a conceptual model approach; 2010 [November]. Cápiro NL, Da Silva MLB, Stafford BP, Rixey WG, Alvarez PJJ. Microbial community response to a release of neat ethanol onto residual hydrocarbons in a pilot-scale aquifer tank. Environ Microbiol 2008;10:2236–44. Corseuil HX, Monier AL, Fernandes M, Schneider MR, Nunes CC, do Rosario M, et al. BTEX plume dynamics following an ethanol blend release: geochemical footprint and thermodynamic constraints on natural attenuation. Environ Sci Technol 2011;45:3422–9. Dakhel N, Pasteris G, Werner D, Hener P. Small-volume releases of gasoline in the vadose zone: impact of the additives MtBE and ethanol on groundwater quality. Environ Sci Technol 2003;37:2127–33. Energy Institute (EI). Reported and potential occurrence of ether oxygenates on water resources in the UK. London: Energy Institute; 2009. Energy Institute (EI). Ethanol effects on the fate and transport of gasoline constituents in the UK. 1st ed. London: Energy Institute; 2012 [July]. Environment Agency (EA). Compilation of data for priority organic pollutants for derivation of soil guideline values. Sci Rep 2008. [SC050021/SR7]. European Chemicals Bureau (ECB). Information on ethanol. ESIS database. European Chemicals Bureau; 2000.
European Council (EC). Directive 2003/30/EC of the European Parliament and of the Council of 8 May 2003 on the promotion of the use of biofuels or other renewable fuels for transport; 2003. Freitas JG, Barker JF. Oxygenated gasoline release in the unsaturated zone. Part 1: source zone behaviour. J Contam Hydrol 2011;126:153–66. [May 18]. Freitas JG, Doulatyari B, Molson JW, Barker JF. Oxygenated gasoline release in the unsaturated zone. Part 2: downgradient transport of ethanol and hydrocarbons. J Contam Hydrol 2011;125:70–85. Gomez DE, Alvarez PJJ. Modeling the natural attenuation of benzene in groundwater impacted by ethanol-blended fuels: effect of ethanol content on the lifespan and maximum length of benzene plumes. Water Resour Res 2009;45:W03409. http://dx.doi. org/10.1029/2008WR007159. Heermann SE, Powers SE. Modeling the partitioning of BTEX in water-reformulated gasoline systems containing ethanol. J Contam Hydrol 1998;34:315–41. Jiang Y-P, Chen Y-D, Zhu Y-N, Lan J-K. Transport and biodegradation of ethanol-blended monoaromatic hydrocarbons in groundwater. J Hydraul Eng 2009;2009–07. Lahvis MA. Evaluation of small-volume releases of ethanol-blended gasoline at UST sites. American Petroleum Institute Soil and Groundwater Research Bulletin, No. 19. Washington, D.C: American Petroleum Institute; 2003. [8 pp.]. Lide DR. CRC handbook of chemistry and physics. Eighty-eighth edition. Boca Raton, Florida: CRC Press; 2008. Mackay DM, de Sieyes NR, Einarson MD, Feris KP, Pappas AA, Wood IA, et al. Impact of ethanol on the natural attenuation of benzene, toluene, and o-xylene in a normally sulfate-reducing aquifer. Environ Sci Technol 2006;40:6123–30. Mackay D, de Sieyes N, Einarson M, Feris K, Pappas A, Wood I, et al. Impact of Ethanol on the Natural Attenuation of MTBE in a Normally Sulfate-Reducing Aquifer. Environ. Sci. Technol. 2007;41:2015–21. Malcolm Pirnie Inc. Evaluation of fate and transport of ethanol in the environment. Prepared for the American Methanol Institute; 1998. McDowell C, Buscheck T, Powers S. Behaviour of gasoline pools following a denatured ethanol spill. Ground Water 2003;41(6):746–57. McNab WS, Heerman SE, Dooher B. Screening model evaluation of the effects of ethanol on benzene plume lengths. In: Rice DW, editor. Health and Environmental Assessment of the Use of Ethanol as a Fuel Oxygenate, Report to the Governor of the State of California in Response to Executive Order D-5-99, Vol. 4, Potential Ground and Surface Water Impacts, vol. 4. 1999. Molson JW, Barker JF, Frind EO. Modeling the impacts of ethanol on the persistence of benzene in gasoline-contaminated groundwater. Water Resour Res 2002;38(1). Risk Assessment Information System (RAIS). Chemical Specific Factors. Available at: http://rais.ornl.gov/tools/profile.php. Shen H, Mravik S, Wilson J, Sewell G. Impact of ethanol on benzene migration; taken from Bioremediation of MtBE. In: Magar V, Gibbs J, O’Reilly K, Hyman M, Leeson A, editors. Alcohols and ethers; 2001. US Environmental Protection Agency (USEPA). FOOTPRINT. A Screening Model for Estimating the Area of a Plume Produced from Gasoline Containing Ethanol. Version 1. 0. EPA/600/R-08/058; 2008. [June 2008]. Yu S, Freitas JG, Unger AJA, Barker JF, Chatzis J. Simulating the evolution of an ethanol and gasoline source zone within the capillary fringe. J Contam Hydrol 2009;105:1–17.
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Ethanol effects on the fate and transport of gasoline constituents in the UK Simon Firth a,⁎, Beate Hildenbrand b, Phil Morgan c a b c
Firth Consultants Ltd, Windsor House, Greville Road, Bristol, BS3 1LL, UK Energy Institute, 61 New Cavendish Street, London, W1G 7AR, UK The Sirius Group, Red Hill House, Hope Street, Chester, CH4 8BU, UK
H I G H L I G H T S • • • •
Monte Carlo analysis used to assess risk to groundwater from ethanol blended gasoline Compared predicted benzene/MtBE plume lengths with/without ethanol in gasoline Compared predicted number of abstractions at risk with/without ethanol in gasoline Results show similar risk to UK groundwater with and without ethanol in gasoline
a r t i c l e
i n f o
Article history: Received 25 June 2013 Received in revised form 14 November 2013 Accepted 27 February 2014 Available online 15 March 2014 Keywords: Fate and transport modelling Groundwater Ethanol Gasoline Environmental risk
a b s t r a c t In the UK, use of ethanol in fuel as a fuel oxygenate/fuel supplement is currently limited but could rise in an effort to meet the requirements of the European “Biofuels” Directive. This Energy Institute study focussed on the risk that accidental releases of ethanol blended gasoline (EBG) (i.e. gasoline containing 10% or less of ethanol) could pose to UK groundwater resources. Ethanol is miscible and highly biodegradable. As a result it tends to be strongly attenuated in the unsaturated zone and in groundwater and so does not, in itself, pose a significant risk to groundwater resources. However, it may lead to increased persistence of other gasoline constituents, particularly through alteration of geochemical conditions as a result of intensive biodegradation activity. A semi-probabilistic modelling exercise was undertaken to better understand the risks that use of EBG could pose to UK groundwater resources. Site investigation information from over 500 filling stations was used in combination with GIS data to predict the proportion of potable water supply wells that could potentially be impacted by benzene and MtBE, and estimate the length of benzene and MtBE plumes with and without the use of ethanol in gasoline. The results show that the use of EBG is likely to have a negligible effect on MtBE plumes. Some increase in benzene plume length is predicted, most notably in fissured aquifers, but increases in plume length of greater than 30% are predicted to be rare. A corresponding slight increase in risk to licensed potable water supply wells from benzene was predicted with the use of EBG but the percentage of wells at risk was still predicted to be small (0.13%), and in the context of the conservatism within the modelling, it was concluded that widespread use of EBG is unlikely to cause an increased risk to UK water resources. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Ethanol is broadly being used as both a fuel oxygenate in reformulated gasoline and as an alternative fuel/fuel supplement. Currently in the UK, the use of ethanol for this purpose is limited but could rise in an effort to meet the requirements of the European
⁎ Corresponding author at: Firth Consultants Ltd, Windsor House, Greville Road, Bristol, BS3 4QF, UK. E-mail addresses: simon@firthconsultants.co.uk (S. Firth), [email protected] (B. Hildenbrand), [email protected] (P. Morgan).
http://dx.doi.org/10.1016/j.scitotenv.2014.02.119 0048-9697/© 2014 Elsevier B.V. All rights reserved.
“Biofuels” Directive (EC, 2003). Whilst the use of biofuels has the potential to cut carbon dioxide emissions, the potential for environmental risks should not be discounted. In particular, experience from use of other fuel additives such as MtBE has demonstrated the importance of understanding risks to water resources from accidental releases of fuel. When considering the risks from ethanol-containing fuel it is important to distinguish between ethanol-blended gasoline (EBG) such as E5 or E10 (gasoline containing 5 or 10% v/v ethanol) and fuel grade ethanol (FGE), i.e. fuel containing 85% ethanol or more (including denatured ethanol). In the UK the vast majority of ethanol used as road transport fuel is sold as EBG and this situation is unlikely to change in the future. As a result, releases of ethanol-containing fuel will most commonly be
706
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
associated with leaks or spills of EBG from filling stations. The likelihood of such releases occurring is reducing with time as a result of widespread replacement or decommissioning of old infrastructure and other improved environmental measures. This paper presents work conducted for the Energy Institute to better understand the risks that use of ethanol as a gasoline constituent could have on UK water resources (Energy Institute, 2012). The paper gives a brief summary of literature on the fate and transport of ethanol in the subsurface and its effect on the persistence of other gasoline constituents, and then goes on to describe modelling work conducted to assess the risk that increased use of EBG could pose to UK groundwater resources. 2. The fate of ethanol in the subsurface and its effect on other gasoline constituents Ethanol is miscible with water and, relative to petroleum hydrocarbons such as benzene, has a low Henry's law constant, indicating that it has a relatively low volatility from the dissolved phase (Table 1). As a result, when released to the subsurface, the majority of ethanol will partition to the aqueous phase. For small volume releases above the water table this can result in a significant proportion of the released ethanol being retained in solution within pore water in the unsaturated zone, significantly reducing the impact it has on groundwater (Freitas and Barker, 2011; Lahvis, 2003). Ethanol-containing fuel released at or below the water table, or large volume releases that occur above the water table, can result in ethanol reaching groundwater and forming a dissolved-phase plume. Once in groundwater, ethanol is typically rapidly biodegraded and so is unlikely to create a risk to groundwater resources in itself. However ethanol's strong polarity and the relative ease with which it is biodegraded can affect the risk caused by other gasoline constituents. Ethanol's strong polarity means that it can increase the aqueous solubility of other gasoline constituents as a cosolvent. However, in practice, the concentration of ethanol required for cosolvency effects to become significant (N80,000 mg L−1, Heermann and Powers, 1998) is unlikely to occur other than in exceptional circumstances such as the groundwater source zone beneath a large volume release of FGE. Ethanol's strong polarity can also theoretically result in increased mobility of non-aqueous phase liquid (NAPL) hydrocarbons (McDowell et al, 2003), but again this effect is only ever likely to become significant for large spills of FGE itself. Tank-scale tests have shown that spills of FGE onto pre-existing NAPL have the potential to reduce capillary fringe thickness allowing subsequent re-distribution of NAPL closer to the water table (Yu et al, 2009). However, field evidence for this effect is lacking. Ethanol's rapid biodegradation in groundwater can affect the persistence of other gasoline constituents through alteration of geochemical conditions. Biodegradation of ethanol quickly uses up available electron acceptors and can lead to the onset of methanogenic conditions (Jiang et al, 2009; Cápiro et al, 2008; Dakhel et al., 2003). These changes can reduce the biodegradation rates of other gasoline constituents. Field studies have shown some evidence of a reduction in biodegradation rates of BTEX compounds resulting in plume elongation (MacKay et al., 2006; Freitas et al, 2011; Corseuil et al., 2011).
Various modelling studies have been conducted to assess the effect that ethanol in EBG could have on the persistence of benzene in groundwater (e.g. McNab et al., 1999; Shen et al., 2001; Molson et al., 2002; Gomez and Alvarez, 2009). The results from these studies vary significantly, with the predicted increase in benzene plume length ranging from 5% to 635%, although predicted increases of 30% to 100% are more typical (Energy Institute, 2012). All of these studies assume that the groundwater source zone contains a significant concentration (e.g. N3 mg L−1) of ethanol and that this persists for some time (e.g. greater than 1 year). However, as discussed above, attenuation in the unsaturated zone and ethanol's rapid biodegradation is likely to limit concentrations and longevity of ethanol in groundwater and this assumption is likely to be highly conservative for releases of EBG from gasoline retail sites. This is supported by experience from California, where EBG has been widely used since 2002. A review of groundwater monitoring data collected from 207 gasoline retail sites from that state showed that ethanol had rarely been detected in groundwater, with only 21 of 13,000 samples recording a concentration of ethanol greater than 1 mg L−1 (API, 2010). These previous modelling studies have also tended to focus on a single field study or hypothetical case, where one set of aquifer properties with one set of source zone concentrations have been modelled. They do not attempt to model the variability in benzene plume length caused by variability in source zone concentrations and aquifer properties. The modelling exercise presented in Section 3 attempts to predict the potential impact of use of EBG on UK groundwater resources as a whole, taking account of variability in parameters such as source zone dimensions and concentrations, aquifer properties and abstraction rates. 3. Predictive modelling A modelling study was undertaken to assess the potential for risk to groundwater resources in the UK from the use of EBG. This consisted of two components: (1) modelling the proportion of potable water supply wells in England and Wales potentially impacted by benzene and MtBE with and without the use of ethanol in gasoline; and (2) modelling the distribution of plume lengths of benzene and MtBE in UK groundwater with and without the use of ethanol in gasoline. The model approach and results for both components are discussed below. 3.1. Model approach 3.1.1. Modelling the proportion of potable water supply wells potentially impacted A simple capture zone analysis was undertaken using a GIS containing the locations of 10378 gasoline retail sites and 3616 licensed abstraction wells in England and Wales (note that the required information was not available for Scotland). This identified 2536 gasoline retail sites that could plausibly pose a risk to 1191 licensed abstraction wells, 679 of which are used for potable water supply. An ExcelTM spreadsheet model was then developed to estimate the concentrations of MtBE and benzene in each of these abstraction wells from hypothetical sources at the 2536 gasoline retail sites. The abstracted concentrations were calculated using the following methodology: 1. The mass flux of benzene, MtBE and ethanol leaving the source zone at each gasoline retail site were calculated using Eq. (1):
Table 1 Comparison of physico-chemical properties of ethanol and benzene. Physical property
Ethanol
Benzene
CAS number Chemical formula Molecular weight (g mol−1) Solubility in water @ 20 °C (mg L−1) Vapour pressure @ 20 °C (Pa) Dimensionless Henry's law constant Organic carbon partition coefficient (L kg−1)
64-17-5 C2H5OH 46.07 Miscible c 5900 c 2.1 × 10−4 d 1.58 e
71-43-2 C6H6 78.11 1780 a 10,000 b 0.116 a 67.6 a
a. Environment Agency, 2008; b. Lide, 2008; c. European Chemicals Bureau, 2000; d. Malcolm Pirnie Inc., 1998; e. Risk Assessment Information System, 2011.
M sz ¼ K:i:W:b:C sz
ð1Þ
Where,
Msz K i W
mass flux of contaminant leaving source zone (g d−1) hydraulic conductivity of groundwater bearing unit (m d−1) hydraulic gradient in groundwater bearing unit (m m−1) width of source zone (m)
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
b Csz
thickness of source zone (m) dissolved phase concentration of contaminant in source zone (mg L−1)
2. The mass fluxes of MtBE and benzene arriving at each abstraction well were then calculated using Eq. (2): −λ:t c
Mr ¼ Msz :e
ð2Þ
Where mass flux of contaminant arriving at receptor (g d−1) mass flux of contaminant leaving source zone (g d−1) contaminant degradation decay rate (d−1) contaminant transit time between source zone and receptor (d)
Mr Msz λ tc
Note that this method does not account for attenuation by dilution or dispersion, which are expected to be negligible relative to degradation within the aquifer. The method assumes a non-diminishing source. As discussed in the uncertainty analysis below, this assumption is conservative. The contaminant transit time from the source zone to the abstraction well is calculated from the groundwater transit time (tgw) multiplied by the retardation factor (Rf). The groundwater transit time is calculated using Eq. (3): t gw ¼
π:B:ne :r 2 Qw
ð3Þ
Where tgw B ne r Qw
groundwater transit time to the receptor (d) saturated aquifer thickness (m) effective porosity (dimensionless) distance to receptor (r) abstraction rate from well (m3 d−1)
This method is based on Darcy's law with the following assumptions: • the principal flow path is via the aquifer where abstraction occurs; • the drawdown in the abstraction well is negligible compared to the saturated thickness of the aquifer; • the aquifer is isotropic and homogenous; • dilution via infiltration is negligible; and • flow towards the wells is radial. The retardation factor is given using Eq. (4): Rf ¼ 1 þ
ρd :K oc :f oc ne
ð4Þ
Where ρb Koc foc
aquifer bulk density (g cm−3) contaminant organic carbon partition coefficient (L kg−1) fraction of organic carbon in aquifer (dimensionless)
Ethanol's effect on the persistence of MtBE and benzene was modelled using the USEPA ‘Footprint’ model approach (USEPA, 2008). This is a simplistic method that assumes that the biodegradation of other gasoline constituents does not occur where the concentration of ethanol is above a threshold concentration (assumed to be 3 mg L−1). This is a conservative assumption as field evidence suggests that the biodegradation of other gasoline constituents may be slowed but is unlikely to be prevented where ethanol is present (Mackay et al., 2006; Freitas et al, 2011; Corseuil et al., 2011). The model predicts the length of ethanol
707
plume dictated by the threshold concentration and calculates the remaining distance to the abstraction well. This distance (r) is used in Eq. (3) above to determine the available time for degradation of benzene and MtBE. Thus, the longer the ethanol plume, the less time there is available for benzene and MtBE degradation to occur and the higher the concentrations of these contaminants will be in the abstracted water at the well. 3. Each abstraction well in the model has one or more gasoline retail sites potentially contributing benzene and MtBE to it. The fluxes from these sites are summed to calculate the total mass fluxes of MtBE and benzene arriving at each well. The concentrations of MtBE and benzene in each well are then calculated by dividing these mass fluxes by the well’s licensed abstraction rate. Investigations at over 500 Energy Institute member sites in the UK have shown a large variability in the extent and magnitude of petroleum hydrocarbon contamination in soils and groundwater (Energy Institute, 2009). Whilst many sites have maximum concentrations in soil or groundwater at or below analytical detection limits, there are some where near saturation concentrations have been detected as a result of the presence of free-phase fuel within the subsurface. Given that source concentration data were not available for the 2536 filling stations considered, a probabilistic approach was adopted for the estimation of mass flux leaving each gasoline retail site (Eq. (1)). Probability distribution functions (PDFs) of groundwater source concentration, source zone dimensions and the product of hydraulic conductivity and gradient were used in a Monte-Carlo analysis to model the mass flux of benzene, ethanol and MtBE leaving each of the identified filling stations. The PDFs for source dimension, hydraulic properties and source concentrations of MtBE were based on data gathered from the 500 + Energy Institute member sites from a previous study (Energy Institute, 2009), whilst the PDFs of source concentration of benzene and ethanol were based on available data from the literature and expert judgement and assuming that EBG was in common usage in the UK. The parameters required for Eqs. (2), (3) and (4) were modelled deterministically. Best estimates of central tendency values were used for the biodegradation decay rates of benzene, MtBE and ethanol, based on literature sources. Aquifer maps were used in the GIS to identify likely aquifer properties (saturated aquifer thickness, fraction of organic carbon and effective porosity) for each abstraction well. The GIS was also used to calculate the distance from each gasoline retail site to its associated abstraction well. The Monte Carlo analysis was run for 2000 simulations. The percentages of abstraction wells with concentrations of benzene and MtBE above threshold concentrations of 1 μg L−1 for benzene (the UK drinking water standard) and 15 μg L−1 for MtBE (the WHO taste threshold) were calculated for each simulation and used to produce a distribution of these percentages for all simulations. This process was undertaken with and without ethanol to assess the impact the use of EBG could have on risk to abstraction wells. 3.1.2. Modelling the length of benzene and MtBE plumes A separate model was developed to assess the effect of ethanol on the plume lengths of benzene and MtBE in groundwater beneath gasoline retail sites. As with the abstraction well model, this model is part probabilistic and part deterministic. Plume length is defined by the distance from the groundwater source to the point down hydraulic gradient at which the contaminant concentration equals a threshold concentration and is calculated using the Domenico equation for steady state conditions (Eq. (5)). Again, this method assumes a non diminishing source, which is a conservative assumption. ( C x ¼ C O exp
x 2ax
( ) rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!) Sy 4λax Sz erf pffiffiffiffiffiffiffi erf pffiffiffiffiffiffiffi u 2 ay x 4 az x
1− 1 þ
ð5Þ
708
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
Table 2 Predicted occurrence of benzene and MtBE in licensed potable water supply wells with and without use of ethanol. Contaminant
Threshold concentration
1 μg L−1 15 μg L−1
Benzene MtBE
Average percentage of wells with concentration exceeding threshold value Without ethanol
With ethanol
0.03% 0.33%
0.13% 0.33%
Where concentration at distance ‘x’ downstream of source zone (mg L− 1) source zone concentration (mg L−1) distance downstream from source zone (m) longitudinal dispersivity coefficient (m) (assumed equal to 0.1x) transverse dispersivity coefficient (m) (assumed equal to 0.01x) vertical dispersivity coefficient (m) (assumed equal to 0.001x) decay rate (d−1) contaminant velocity (m d−1) source zone width (m) source zone thickness (m)
Cx Co x ax ay az λ u Sy Sz
Contaminant velocity was estimated using Eq. (6):
u¼
Ki ne :R f
ð6Þ
Where Ki
product of hydraulic conductivity and hydraulic gradient (m d− 1) aquifer effective porosity (dimensionless) contaminant retardation factor (dimensionless)
Table 4 Modelled plume lengths—fractured bedrock aquifers. Contaminant
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m) Median
75th percentile
90th percentile
95th percentile
4.2 4.4 6% 48 54 12%
41 46 12% 184 197 7%
138 157 14% 464 492 6%
236 280 19% 794 797 0%
of dispersion, and the model will therefore tend to overestimate plume extension. As with the abstraction well model, source dimensions and concentrations, hydraulic conductivity and gradient were modelled probabilistically. Contaminant degradation decay rate and Koc were modelled as single (deterministic) values for each contaminant. The modelling was undertaken for a range of aquifer types, with generic values of aquifer thickness, effective porosity and fraction of organic carbon assigned to each aquifer type consistent with the values used for the abstraction well modelling. 3.2. Model results 3.2.1. Proportion of potable water supply wells potentially impacted The predicted percentages of licensed potable abstraction wells in England and Wales with benzene and MtBE above the threshold concentrations with and without the use of ethanol in gasoline are presented in Table 2 below. This shows that the use of ethanol is not expected to have a significant effect on the occurrence of MtBE in abstractions wells, with a predicted 0.33% of abstraction wells having N 15 μg L− 1 MtBE with or without the use of EBG. The percentage of wells with N1 μg L−1 benzene is predicted to be slightly higher (0.13%) with the use of ethanol than without (0.03%).
The length of plume was found by varying distance ‘x’ iteratively until ‘Cx’ is equal to a threshold concentration. Again, threshold concentrations of 1 μg L−1 and 15 μg L−1 were chosen for benzene and MtBE, respectively. A threshold concentration of 3 mg L−1 was assumed for ethanol, which, as discussed above, is taken as the concentration above which degradation of benzene and MtBE is assumed not to occur. Thus, the plume length of ethanol is the predicted distance from the site that groundwater must travel before benzene and MtBE degradation can occur. This length is then added to the estimated benzene and MtBE plume lengths (without ethanol) to estimate the length of benzene and MtBE plumes in the presence of ethanol. This simplified approach is conservative because it ignores the fact that benzene and MtBE concentrations will reduce within the ethanol plume as a result
3.2.2. Benzene and MtBE plume lengths Tables 3 to 5 compare predicted median and upper percentile plume lengths for releases of gasoline with and without use of ethanol for three different aquifer types: (1) unconsolidated alluvial deposits (e.g. sand and gravels), (2) fractured bedrock aquifers (such as the Cretaceous Chalk, Jurassic Oolite and Magnesian Limestone), and (3) intergranular bedrock aquifers (such as the Lower Cretaceous Sands and Permian–Triassic Sandstone). The distributions of modelled plume lengths with and without ethanol are also compared graphically in Figs. 1 to 3. The model predicts a slight increase in median plume length for MtBE with use of ethanol in gasoline, but ethanol's effects on longer plumes is negligible. Given that the median plume lengths are relatively short (b50 m), ethanol's overall effect on risk from MtBE is likely to be negligible. This finding is consistent with the abstraction well modelling. Ethanol is predicted to have a greater influence on benzene plume length, with a 9% increase in median plume length for all aquifer types.
Table 3 Modelled plume lengths—alluvial aquifers.
Table 5 Modelled plume lengths—intergranular bedrock aquifers.
ne Rf
Contaminant
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m)
Contaminant
Median
75th percentile
90th percentile
95th percentile
0.11 0.12 9% 16 17 7%
1.8 2.2 23% 88 90 2%
12 16 31% 240 242 1%
40 53 33% 391 395 1%
Benzene no ethanol Benzene with ethanol % increase MtBE no ethanol MtBE with ethanol % increase
Plume length (m) Median
75th percentile
90th percentile
95th percentile
0.37 0.40 9% 13 15 13%
5.8 6.3 8% 80 86 8%
39 44 12% 219 224 2%
80 89 12% 341 345 1%
1400 1200 1000 800 600 400 200 0
Without ethanol With ethanol
Frequency
Frequency
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
800 700 600 500 400 300 200 100 0
Benzene plume length (m)
709
Without ethanol With ethanol
MtBE plume length (m)
Fig. 1. Frequency distributions of modelled benzene and MtBE plume lengths in alluvium aquifers.
However, given that median benzene plume lengths are less than 5 m, ethanol's predicted effect on plume length can be considered negligible for the majority of sites. Ethanol's effect is more pronounced for the upper percentile benzene plumes, with approximately 12 to 33% increases in the 95th percentile plume lengths. The largest percentage increases occur for the alluvial aquifer type. This has the highest fraction of organic carbon content and consequently has the highest ratio of benzene to ethanol retardation coefficients. As a result, the non-retarded steady state plume of ethanol causes the greatest impact on overall benzene degradation for this aquifer type. In reality, for diminishing sources, the retardation contrast will typically result in the ethanol plume detaching from the benzene plume. Under such circumstances, ethanol's effect on the persistence of benzene will be reduced, i.e. the model likely over-predicts ethanol's effects on benzene plume lengths for alluvial aquifers. 3.3. Uncertainty analysis Key uncertainties in the modelling are discussed below.
800 700 600 500 400 300 200 100 0
3.3.2. Model parameters Sensitivity analysis was conducted to assess the effects of uncertainty in key model input parameters on model results. Parameters tested were the threshold concentration for ethanol, the PDF for the groundwater source concentration for ethanol and the biodegradation decay rates of benzene and ethanol. This analysis showed that uncertainty in these parameters caused relatively little variation in the model outputs. 4. Conclusion The modelling conducted for this study provides a useful indicator of the potential risks to UK groundwater resources from use of EBG. A probabilistic approach has been adopted to account for the high degree of variability between sites, especially with regard to source concentration. The results of this exercise indicate that the widespread use of EBG such as E5 or E10 is unlikely to cause a significant increase in risk to UK groundwater resources. For the majority of filling stations, plume lengths of benzene are unlikely to increase by more than 10%. Greater increases in plume length may occur for longer plumes in alluvial and fracture flow aquifers, but increases above 30% are predicted to be rare. Furthermore, field evidence suggests that any elongation of plumes will be temporary. The increase in benzene plume lengths is predicted to have some impact on the risk to licensed potable water supply wells. However,
Without ethanol With ethanol
Benzene plume length (m)
Frequency
Frequency
3.3.1. Model approach The model approach is based on the assumption that biodegradation of benzene and MtBE will not occur in groundwater where ethanol is present above a threshold concentration. Field studies of releases of EBG have shown that, whilst biodegradation rates of benzene may decline as a result of the onset of methanogenic conditions, biodegradation of benzene in the presence of ethanol does still occur. There is even evidence to suggest that, following depletion in concentrations of ethanol and its breakdown intermediates, benzene degradation rates may be enhanced due to increased microbial biomass (Corseuil et al., 2011). There is also evidence that the presence of ethanol itself may stimulate biodegradation of MtBE to tert-butyl alcohol (TBA) (Mackay et al., 2007). Thus, the assumption of no degradation of benzene and MtBE where ethanol is present above a threshold concentration is likely to be conservative and tend towards the over-prediction of risk. Another source of potential conservatism within the model is the assumption of steady state conditions. The groundwater source concentrations of ethanol, benzene and MtBE are assumed to remain constant
and not decrease with time. In reality, source concentrations will tend to decrease with time as contaminants are removed from the site by biodegradation, volatilisation and/or via advective transport in groundwater. In particular, persistent sources of ethanol are not expected due to its miscibility with water and rapid biodegradation. Field evidence shows that ethanol rarely persists in groundwater for more than a few months. The low retardation and short source duration of ethanol relative to hydrocarbons can also result in the ethanol plume detaching from the hydrocarbon plumes such that it no longer has any impact on hydrocarbon degradation in groundwater. This effect was observed in the controlled field releases of E10 and E95 at the Borden site in Canada (Freitas et al., 2011). The assumption of steady state conditions is therefore likely to be conservative and tend towards an over-prediction of risk.
800 700 600 500 400 300 200 100 0
Without ethanol With ethanol
MtBE plume length (m)
Fig. 2. Frequency distributions of modelled benzene and MtBE plume lengths in fissure aquifers.
S. Firth et al. / Science of the Total Environment 485–486 (2014) 705–710
Frequency
1200 1000
Without ethanol With ethanol
800 600 400 200 0
Benzene plume length (m)
Frequency
710
800 700 600 500 400 300 200 100 0
Without ethanol With ethanol
MtBE plume length (m)
Fig. 3. Frequency distributions of modelled benzene and MtBE plume lengths in intergranular bedrock aquifers.
although a slight increase is predicted, the predicted proportion of wells impacted is still small (0.13%). Furthermore, given that a number of conservative assumptions have been used for the modelling and that the frequency and magnitude of leaks from filling stations in the UK continues to decrease as a result of continuing improvements to infrastructure, the model results are likely to over-predict actual impacts from the future use of EBG. Thus, it is concluded that use of EBG is unlikely to cause a significant increase in risk to groundwater resources in the UK. Acknowledgments The authors would like to thank the Energy Institute's Soil, Waste and Groundwater Group who commissioned the work and, in particular, the steering group members for their valuable contribution to the work, namely Ruth Chippendale (Shell), Martyn Dunk (ExxonMobil), Chris Hughes (Chevron), Martyn Lambson (BP) and Fraser Will (Total). We are also grateful to Brent Stafford from Shell, Jim Barker from the University of Waterloo, Canada and Steve Thornton from University of Sheffield, UK for their assistance. References American Petroleum Institute (API). Management of ethanol-blended fuel releases: a conceptual model approach; 2010 [November]. Cápiro NL, Da Silva MLB, Stafford BP, Rixey WG, Alvarez PJJ. Microbial community response to a release of neat ethanol onto residual hydrocarbons in a pilot-scale aquifer tank. Environ Microbiol 2008;10:2236–44. Corseuil HX, Monier AL, Fernandes M, Schneider MR, Nunes CC, do Rosario M, et al. BTEX plume dynamics following an ethanol blend release: geochemical footprint and thermodynamic constraints on natural attenuation. Environ Sci Technol 2011;45:3422–9. Dakhel N, Pasteris G, Werner D, Hener P. Small-volume releases of gasoline in the vadose zone: impact of the additives MtBE and ethanol on groundwater quality. Environ Sci Technol 2003;37:2127–33. Energy Institute (EI). Reported and potential occurrence of ether oxygenates on water resources in the UK. London: Energy Institute; 2009. Energy Institute (EI). Ethanol effects on the fate and transport of gasoline constituents in the UK. 1st ed. London: Energy Institute; 2012 [July]. Environment Agency (EA). Compilation of data for priority organic pollutants for derivation of soil guideline values. Sci Rep 2008. [SC050021/SR7]. European Chemicals Bureau (ECB). Information on ethanol. ESIS database. European Chemicals Bureau; 2000.
European Council (EC). Directive 2003/30/EC of the European Parliament and of the Council of 8 May 2003 on the promotion of the use of biofuels or other renewable fuels for transport; 2003. Freitas JG, Barker JF. Oxygenated gasoline release in the unsaturated zone. Part 1: source zone behaviour. J Contam Hydrol 2011;126:153–66. [May 18]. Freitas JG, Doulatyari B, Molson JW, Barker JF. Oxygenated gasoline release in the unsaturated zone. Part 2: downgradient transport of ethanol and hydrocarbons. J Contam Hydrol 2011;125:70–85. Gomez DE, Alvarez PJJ. Modeling the natural attenuation of benzene in groundwater impacted by ethanol-blended fuels: effect of ethanol content on the lifespan and maximum length of benzene plumes. Water Resour Res 2009;45:W03409. http://dx.doi. org/10.1029/2008WR007159. Heermann SE, Powers SE. Modeling the partitioning of BTEX in water-reformulated gasoline systems containing ethanol. J Contam Hydrol 1998;34:315–41. Jiang Y-P, Chen Y-D, Zhu Y-N, Lan J-K. Transport and biodegradation of ethanol-blended monoaromatic hydrocarbons in groundwater. J Hydraul Eng 2009;2009–07. Lahvis MA. Evaluation of small-volume releases of ethanol-blended gasoline at UST sites. American Petroleum Institute Soil and Groundwater Research Bulletin, No. 19. Washington, D.C: American Petroleum Institute; 2003. [8 pp.]. Lide DR. CRC handbook of chemistry and physics. Eighty-eighth edition. Boca Raton, Florida: CRC Press; 2008. Mackay DM, de Sieyes NR, Einarson MD, Feris KP, Pappas AA, Wood IA, et al. Impact of ethanol on the natural attenuation of benzene, toluene, and o-xylene in a normally sulfate-reducing aquifer. Environ Sci Technol 2006;40:6123–30. Mackay D, de Sieyes N, Einarson M, Feris K, Pappas A, Wood I, et al. Impact of Ethanol on the Natural Attenuation of MTBE in a Normally Sulfate-Reducing Aquifer. Environ. Sci. Technol. 2007;41:2015–21. Malcolm Pirnie Inc. Evaluation of fate and transport of ethanol in the environment. Prepared for the American Methanol Institute; 1998. McDowell C, Buscheck T, Powers S. Behaviour of gasoline pools following a denatured ethanol spill. Ground Water 2003;41(6):746–57. McNab WS, Heerman SE, Dooher B. Screening model evaluation of the effects of ethanol on benzene plume lengths. In: Rice DW, editor. Health and Environmental Assessment of the Use of Ethanol as a Fuel Oxygenate, Report to the Governor of the State of California in Response to Executive Order D-5-99, Vol. 4, Potential Ground and Surface Water Impacts, vol. 4. 1999. Molson JW, Barker JF, Frind EO. Modeling the impacts of ethanol on the persistence of benzene in gasoline-contaminated groundwater. Water Resour Res 2002;38(1). Risk Assessment Information System (RAIS). Chemical Specific Factors. Available at: http://rais.ornl.gov/tools/profile.php. Shen H, Mravik S, Wilson J, Sewell G. Impact of ethanol on benzene migration; taken from Bioremediation of MtBE. In: Magar V, Gibbs J, O’Reilly K, Hyman M, Leeson A, editors. Alcohols and ethers; 2001. US Environmental Protection Agency (USEPA). FOOTPRINT. A Screening Model for Estimating the Area of a Plume Produced from Gasoline Containing Ethanol. Version 1. 0. EPA/600/R-08/058; 2008. [June 2008]. Yu S, Freitas JG, Unger AJA, Barker JF, Chatzis J. Simulating the evolution of an ethanol and gasoline source zone within the capillary fringe. J Contam Hydrol 2009;105:1–17.
