Environ. Sci. Technol. 2006, 40, 2304-2315
Simulating the Effect of Aerobic Biodegradation on Soil Vapor Intrusion into Buildings: Influence of Degradation Rate, Source Concentration, and Depth LILIAN D. V. ABREU AND PAUL C. JOHNSON* Department of Civil and Environmental Engineering, Ira A. Fulton School of Engineering, Arizona State University, Tempe, Arizona 85287
Steady-state vapor intrusion scenarios involving aerobically biodegradable chemicals are studied using a threedimensional multicomponent numerical model. In these scenarios, sources of aerobically biodegradable chemical vapors are placed at depths of 1-14 m beneath a 10 m × 10 m basement or slab-on-grade construction building, and the simultaneous transport and reaction of hydrocarbon and oxygen vapors are simulated. The results are presented as Johnson and Ettinger attenuation factors R (predicted indoor air hydrocarbon concentration/source vapor concentration), and normalized contour plots of hydrocarbon and oxygen concentrations. In addition to varying the vapor source depth, the effects of source concentration (2-200 mg chemical/L vapor) and oxygen-limited first-order reaction rates (0.018-1.8 h-1) are studied. Hydrocarbon inputs were specific to benzene, although the relevant properties are similar to those for a range of hydrocarbons of interest. Overall, the results suggest that aerobic biodegradation could play a significant role in reducing vapor intrusion into buildings (decreased R-values) relative to the no-biodegradation case, with the significance of aerobic biodegradation increasing with increasing vapor source depth, decreasing vapor source concentration, and increasing first-order biodegradation rate. In contrast to the no-biodegradation case, differences in foundation construction can be significant in some settings. The significance of aerobic biodegradation is directly related to the extent to which oxygen is capable of migrating beneath the foundation. For example, in the case of a basement scenario with a 200 mg/L vapor source located at 3 m bgs, oxygen is consumed before it can migrate beneath the foundation, so the attenuation factors for simulations with and without aerobic biodegradation are similar for all firstorder rates studied. For the case of a 2 mg/L vapor source located at 8 m bgs, the oxygen is widely distributed beneath the foundation, and the attenuation factor for the biodegradation case ranges from about 3 to 18 orders-ofmagnitude less than that for the no-biodegradation case.
Introduction The migration of chemical vapors from contaminated soils and groundwater through the subsurface, across foundations, * Corresponding author
[email protected]. 2304
9
phone:
(480)
965-9115;
e-mail:
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006
FIGURE 1. Comparison of model output and field data for a range of soil gas profile types as characterized by Roggemans et al. (37). Input nomenclature is given in Table 1. and into buildings has been the focus of a number of studies (1-34). It has also recently been a topic of interest to regulators who must make decisions concerning response actions and cleanup criteria at contaminated sites (35). To date, the link between subsurface contamination and impacts to indoor air quality has been established in the literature (1, 3, 10, 11, 18-20, 27, 29, 30, 32), and the development and use of predictive tools relating the magnitude of the impact of the source vapor concentration, distance from the source to the building, building characteristics, soil properties, and chemical properties has been reported (2, 6, 8-18, 24, 25, 28-34). The reliability of these tools for site-specific application, however, has not been well established due to a lack of data from well-studied field sites. Johnson et al. (30) and Hers et al. (32) summarize observations from field sites, and compare site-specific attenuation factors with those estimated from the Johnson and Ettinger (8) screening-level model. These analyses provide insight to the likely range of values for the Johnson and Ettinger attenuation factor R, which is defined as the ratio of the indoor air concentration to the soil gas concentration at a depth. For example, sub-slab attenuation factors (indoor air concentration/soil gas concentration immediately below a foundation) in the 0.001-0.01 range have been reported, and values of 0.0001 and smaller have been reported for deeper depths. Chlorinated solvents (e.g., TCE, PCE) were the chemicals of concern at most of the well-studied sites, and, because of 10.1021/es051335p CCC: $33.50
2006 American Chemical Society Published on Web 02/17/2006
TABLE 1. Input Parameters Used in Generating Figures 1-11, S3-S5, and Tables S3-S4a building/foundation parameters length: 10 m width: 10 m depth in soil: 2.0 m (basement type) 0.2 m (slab-on-grade type) foundation thickness (dck): 0.15 m enclosed space volume (Vb): 174 m3 air exchange rate (Aex): 0.5 h-1 crack width (wck): 0.001 m total crack length: 39 m crack location: perimeter disturbance pressure (pindoor): 5 Pa soil properties soil bulk density (Fb): 1700 kg/m3 mass fraction of organic carbon in the soil (foc): 0.001 kg oc/kg soil moisture-filled porosity (Φw): 0.07 m3water/m3soil total soil porosity (ΦT): 0.35 m3voids/m3soil soil permeability to soil gas flow (Kg): 10-11 m2 soil domainb dimensions in (x,y,z) directions 24 m × 24 m x (1-14 m depths) discretizationc (see sample grid in Figure 1) number of grid nodes: 3 × 104 spacing in fine grid area: 0.05-0.2 m spacing in coarse grid area: 0.4-2.0 m algorithm parameters numerical scheme: implicit disturbance pressure subroutine: variable time step range: 0.001 s - 2 h percent change allowed/time step: 10% concentration subroutine: variable time step range: 80 s - 100 h percent change allowed/time step: 5-55% hydrocarbon vapor source properties location: base of vadose zone source size: entire domain footprint hydrocarbon properties molecular diffusion coefficient in air (Dia): 3.17 × 10-2 m2/h molecular diffusion coefficient in water (Diw): 3.53 × 10-6 m2/h overall effective diffusion coefficient for transport in the porous media (Di): 3.73 × 10-3 m2/h overall effective diffusion coefficient for transport in the crack (Dick): 3.17 × 10-2 m2/h Henry’s Law constant (Hi): 0.228 m3water/m3vapor sorption coefficient of hydrocarbon to organic carbon (Koc,i): 61.7 kg/kg oc atmospheric concentration: 0.0 oxygen properties molecular diffusion coefficient in air (Dia): 7.2 × 10-2 m2/h molecular diffusion coefficient in water (Diw): 8.7 × 10-6 m2/h overall effective diffusion coefficient for transport in porous media (Di): 8.48 × 10-3 m2/h overall effective diffusion coefficient for transport in the crack (Di,ck): 7.2 × 10-2m2/h Henry’s Law constant (Hi): 31.6 m3water/m3vapor sorption coefficient of oxygen to organic carbon (Koc,i): negligible, assumed 0 kg/kg oc ratio of oxygen to hydrocarbon consumed (rko): 3 kg oxygen/kg hydrocarbon threshold concentration (Cmin og ): 13.7 mg/Lvapor atmospheric concentration (Catm og ): 279 mg/Lvapor others dynamic viscosity of air (µg): 0.0648 kg/m/h a Unless otherwise noted in the text or figures. b The symmetrical scenario domain includes only a quarter of the building footprint in the simulation. c Note that although the grid size and distribution vary over a wide range, the discretization around the building crack area is the same for all the simulated scenarios.
that, conclusions regarding R-value ranges may be specific to chlorinated and other more-recalcitrant chemicals. Considerably less comprehensive quantitative data are available for petroleum spill sites but anecdotal observations have led some to hypothesize that petroleum hydrocarbon vapor
attenuation is much more significant than that for chlorinated solvent vapors. The hypothesis is based on the arguments that petroleum hydrocarbon degrading aerobic organisms are relatively ubiquitous in soils, the atmosphere is a steady source of oxygen for the subsurface, and significant attenuVOL. 40, NO. 7, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2305
FIGURE 2. Normalized steady-state soil gas concentration distribution for oxygen and hydrocarbon undergoing aerobic biodegradation with a first-order rate λ ) 0.18 h-1 and 20, 100, and 200 mg/L source concentrations located 8 m bgs and underneath a basement foundation. Hydrocarbon and oxygen contours are normalized to source and atmospheric concentrations, respectively. ation of petroleum hydrocarbon vapors via aerobic biodegradation has been observed in natural settings away from buildings (36-38) and in controlled experimental settings (39-47). There is a need to better understand the role that aerobic biodegradation plays in attenuating vapor intrusion to buildings. Many petroleum spill sites (e.g., service stations) are located in close proximity to buildings and properties overlying petroleum residuals in soil that are being redeveloped (e.g., brownfield sites, residential neighborhoods built above old oil production fields, etc.). Our current understanding of biodegradation’s role is limited because of a lack of pathway-specific data from petroleum-impacted sites (e.g., soil gas profiles beneath buildings were reported in only a few studies (18, 26, 38)), and because the modeling to date has not adequately addressed these scenarios. For example, Johnson et al. (25) and Hers et al. (26) incorporated first-order kinetic expressions in the Johnson and Ettinger Model; however, these mathematical formulations are of limited use beyond the fitting of field data because they fail to account for oxygen transport and depletion. Some have modeled the simultaneous transport and reaction of hydrocarbon and oxygen vapors (36, 48) under diffusiondominated conditions and open-surface boundary condi2306
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006
tions, but these models do not describe the movement of oxygen from the atmosphere to beneath a building by combined diffusive and advective mechanism. This work utilizes the published Abreu and Johnson threedimensional, multicomponent, numerical model (33, 34) to examine the role that aerobic biodegradation might play in determining the vapor attenuation factor R at sites impacted with aerobically biodegradable chemicals. More specifically, the significance of aerobic biodegradation, relative to the no-biodegradation case, is studied as a function of vapor source concentration, kinetic parameters (i.e., first-order degradation rate), and building construction descriptors (basement and slab-on-grade scenarios) for homogeneous soil settings in which areally extensive vapor sources are placed underneath buildings at depths ranging from 1 to 14 m. This work is critical to the development of practicable regulations for the assessment and management of vapor intrusion at petroleum-impacted sites. Field data suggest significant attenuation increases for petroleum hydrocarbons relative to nondegrading chemicals at some sites, but the data are insufficient to define the range of site conditions where aerobic biodegradation is and is not significant. In the absence of a better understanding, petroleum-impacted sites will be assessed in the same manner as nondegrading
FIGURE 3. Normalized steady-state soil gas concentration distribution for oxygen and hydrocarbon undergoing aerobic biodegradation with a first-order rate λ ) 0.18 h-1 and 20, 100, and 200 mg/L source concentrations located 8 m bgs and underneath a slab-on-grade foundation. Hydrocarbon and oxygen contours are normalized to source and atmospheric concentrations, respectively. chemicals, and consequently a number of low-risk sites might be assessed as having unacceptable risks.
Abreu and Johnson Numerical Model The development and use of the numerical model is described in detail in Abreu and Johnson (33) and Abreu (34). In brief, the numerical model simultaneously solves equations for the soil gas pressure field (from which the advective flow field is computed), advective and diffusive transport and reaction of multiple chemicals, flow and chemical transport through foundation cracks, chemical mixing indoors, and boundary conditions at ground surface, at the foundation walls and cracks, at the lower boundary, and at lateral edges of the model domain (see summary of equations in Tables S1 and S2 in the Supporting Information). Time-varying atmospheric and indoor pressures are allowed, but steady indoor-outdoor pressure differentials were studied in this work. Inputs to the model include geometry descriptors (building footprint, foundation depth, crack locations and widths, source depth, etc.), chemical properties, kinetic parameters, the indoor-outdoor pressure differential, oxygen concentration at ground surface, and the chemical vapor concentration(s) at the vapor source. The model does not account for changes in soil gas pressure that result from changes in the total number of
gaseous moles during aerobic biodegradation reactions. Typical hydrocarbon aerobic biodegradation reaction stoichiometry results in a net decrease in gaseous moles and consequently a net decrease in the sum of the partial pressures of those components in the soil gas near the zone of reaction. In theory, this could cause an under-pressurization relative to the atmosphere and an enhancement in downward advective transport of oxygen. The significance of this would vary depending on the scenario simulated. For example, the significance of this would be lowest in cases where hydrocarbon source concentrations and reaction rates were sufficiently low that oxygen is not significantly depleted in the subsurface, and greatest in cases where hydrocarbon concentrations and reaction rates are high enough to cause depletion of the oxygen in the subsurface. This issue was not assessed from a theoretical perspective in this work, but was assessed empirically by comparison of model results with available soil gas profiles. As discussed below, model input compared favorably with field data for reasonable site-specific inputs, even for cases where oxygen was depleted in the subsurface. Thus, this comparison suggests that this phenomenon was not significant, at least for these conditions. Details on the numerical method are given in Abreu (34). Variable grid spacing is critical to the solution of this problem VOL. 40, NO. 7, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2307
as the physical dimensions range from the 10-3 m crack width to the 100 m full domain scale width. In general, the finer grid spacing is installed near the foundation cracks, domain boundaries, and chemical vapor source boundaries (see Figure S1, Supporting Information). Mass balance checks are performed and the results are used to refine the numerical grid on a problem-specific basis. Air flow mass balances are typically in the range of 93-99% and individual chemical mass balances are >98%. To assess the ability of the model to anticipate soil gas distributions for aerobic biodegradation scenarios, model results were compared with soil gas profiles from sites discussed in the Roggemans et al. (37) compilation. Figure 1 presents this comparison for three sites with distinctly different soil gas profiles; hydrocarbon and oxygen profiles are normalized in each to the hydrocarbon source and atmospheric oxygen concentrations, respectively. Key model inputs are also summarized in each figure (with corresponding nomenclature defined in Table 1). The Roggemans et al. (37) and Lahvis et al. (49) plots correspond to soil gas profiles beneath open surfaces, while the Laubacher et al. (38) soil gas profile was sampled beneath a basement with dimensions 10 m × 4 m × 1.5 m deep. The model was not rigorously calibrated to any of these site data sets; instead, reasonable site-specific inputs were used. As can be seen, the simulations reasonably anticipate the key features of the field data for these sites.
FIGURE 4. Influence of soil vapor source concentration on the vapor attenuation coefficient (r) for basement scenarios, three values of first-order aerobic biodegradation rate (λ), and vapor sources located at depths of 3 and 8 m bgs.
Conditions Simulated All simulations presented here correspond to steady-state scenarios in which the soil properties are homogeneous and a very large vapor source is located at some depth beneath the building foundation. Each simulation represents a unique combination of building construction, source depth, source concentration, and first-order biodegradation rate. Model inputs common to all simulations are summarized in Table 1, while the scenario-specific values are shown in the legends and captions of the figures discussed below. The scenarios studied are symmetrical in relation to the building foundation center line in both the x- and ydirections; therefore, the model domain was chosen to be one-quarter of the full simulation. Two building foundation geometries are modeled: basement and slab-on-grade constructions with perimeter cracks (see Figure S2, Supporting Information). Reasonable inputs for the building dimensions, foundation thickness, crack widths, air exchange rate, and building under-pressurizations are based on values reported in the literature (1-6, 8, 17, 18, 23, 31). The chemical property inputs are specific to benzene, but the inputs most relevant to steadystate simulations (diffusion coefficients in air and water) are representative of a wide range of chemicals of interest. The source vapor concentration is representative of total hydrocarbon concentrations measured at weathered gasoline spill sites (37). Soil properties (permeability, total porosity, and moisture content) are representative of fine- to coarsegrained soil types. Biodegradation kinetic parameters are based on values reported in the literature (39-42, 44-47). These simulations incorporate an oxygen concentration threshold of 1% v/v that must be exceeded for aerobic biodegradation to occur, as suggested by field soil gas profiles (37); however, the results are very similar to those for the case of a zero oxygen threshold concentration.
Results and Discussion Three-dimensional model output is presented below as twodimensional contour plots on vertical cross-sections through the center of the building. Hydrocarbon concentration contour plots are normalized to the source zone vapor 2308
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006
FIGURE 5. Influence of soil vapor source concentration on the vapor attenuation coefficient (r) for slab-on-grade scenarios, three values of first-order aerobic biodegradation rate (λ), and vapor sources located at depths of 3 and 8 m bgs. concentration, and oxygen concentration contour plots are normalized to atmospheric oxygen concentrations. In addition, the vapor attenuation factor (R) is calculated for each simulation. For reference, R is defined as follows:
R)
Cindoor ig Csource ig
(1)
Soil Gas Pressure Field and Air Flow Rate into the Building. Sample normalized disturbance pressure fields for basement and slab-on-grade scenarios as well as tabulated
FIGURE 6. Normalized steady-state soil gas concentration distributions for basement scenarios with oxygen and hydrocarbon undergoing aerobic biodegradation with first-order rates λ ) 0.018, 0.18, and 1.8 h-1 and a 200 mg/L vapor source located at a depth of 5 m bgs. Hydrocarbon and oxygen contours are normalized to the source and the atmospheric concentration, respectively. values of the calculated pressure-induced soil gas flow rates for soil conditions studied here are presented in Figures S3 and S4 and Tables S3 and S4 (Supporting Information). In brief, the disturbance pressure fields suggest air flow primarily traveling down from ground surface and then to the foundation crack, with flow beneath the foundation becoming more pronounced as the depth to groundwater increases. The pressure-induced soil gas flow rates from the subsurface and into the buildings predicted for the scenarios discussed below range from about 3 to 5 L/min, with flow rates generally being larger for slab-on-grade constructions and deeper depths to groundwater. Effect of Source Concentration. Soil gas concentration contour plots are shown in Figures 2 and 3 for basement and slab-on-grade scenarios, respectively. Hydrocarbon vapor source concentrations of 20, 100, and 200 mg/L are presented for source depths of 8 m and a first-order biodegradation rate of λ ) 0.18 h-1. The lower end of this concentration spectrum might be encountered at sites where hydrocarbonimpacted groundwater is the vapor source, while the upper end of the concentration spectrum is more representative of sites with gasoline-impacted soils (37). These source concentrations were selected for presentation because soil gas concentration profiles for source concentrations 200 mg/L generally resemble the 200 mg/L profiles. The 100 mg/L source concentration contour plots provide an intermediate point between the two. While the modeling domain was 24 m wide from the plane of symmetry (center of the building), all contour plot figures show only the soil gas distributions out to a lateral distance of 12 m. This distance was chosen for presentation purposes
because it was sufficiently large to show soil gas profiles beneath foundations, as well as the “far-field” soil gas profiles representative of regions unaffected by foundations. For the 20 mg/L source concentration, Figures 2 and 3 show oxygen penetration throughout much of the subsurface, and about a six order-of-magnitude reduction in hydrocarbon vapor concentrations between 7 and 2 m bgs. The hydrocarbon soil gas profiles show little effect of the foundation, as they are similar beneath the foundation and far from the foundation. Predicted soil vapor intrusion attenuation factors (R) are 0.05 and hydrocarbon vapors accumulate at 20-90% of the source concentration. For the basement scenario, soil gas profiles and R-values are relatively unaffected by changes in first-order biodegradation rate due to insufficient oxygen concentrations beneath the foundation. For the slab-on-grade scenarios (Figure 7), oxygen partially penetrates the region beneath the foundation. Normalized oxygen concentrations, above the 0.05 threshold for biodegradation, are found in the region surrounding the perimeter crack; thus, biodegradation occurs in this region and hydrocarbon concentrations entering the enclosed space are reduced relative to the no-biodegradation case. As a result, R decreases by about five orders-of-magnitude as the firstorder degradation rate increases from 0.018 to 1.8 h-1. The hydrocarbon accumulation beneath the foundation is also less than that in the basement scenarios, being only about 10-40% of the source concentration. Effect of Source Depth. Figures 8 and 9 show normalized soil gas contours for a 200 mg/L hydrocarbon source located at depths of 1-14 m bgs, and a first-order biodegradation rate of 0.18 h-1. Figure 8 shows basement scenarios with sources at depths of 3, 5, 8, and 14 m bgs. Figure 9 shows slab-on-grade scenarios with source at depths of 1, 3, 5, and 8 m bgs. The depths were chosen so that the distance from foundation to hydrocarbon vapor source would be comparable between Figures 8 and 9. 2312
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006
For the 200 mg/L source located at depths of 1, 3, 5, 8, and 14 m bgs, eq 2 anticipates the far-field depth of oxygen penetration to be about 0.5, 1.5, 2.5, 4.0, and 7.0 m bgs, respectively. This is in good agreement with Figures 8 and 9. The near-field oxygen concentrations are affected by the presence of the building at all depths, but the building influence decreases as the source depth increases. On the basis of the results in Figures 2 and 3, we can anticipate that at some depth the effect of the building is insignificant and that the depth at which this occurs decreases as the source vapor concentration decreases. For the basement scenario and source depths 8 m bgs, oxygen begins penetrating beneath the foundation and this is reflected in a decrease in R-values by about 5 orders-of-magnitude (λ ) 0.18 h-1) as the source depth increases from 8 to 14 m bgs. For the slab-on-grade scenarios depicted in Figure 9, oxygen penetration below the foundation starts to be observed when the hydrocarbon source is located at a depth of about 3 m bgs. For sources at depths of 8 m bgs or deeper, normalized oxygen concentrations greater than 0.05 are observed throughout the sub-slab region, and R-values
For the basement scenario (Figure 10), source depths
