Bioventing revisited: efficacy of enhanced biodegradation for sites with mobile LNAPL

Bioventing is a remediation technology that enhances aerobic biodegradation of petroleum-affected soil in the vadose zone by introducing oxygen to the subsurface. Bioventing was historically considered effective for decreasing petroleum hydrocarbons concentrations in soil but discounted for sites where mobile light non-aqueous phase liquid (LNAPL) accumulated in wells. While the science behind bioventing has not changed, the conceptual understanding of LNAPL depletion processes and framework for evaluating the efficacy of LNAPL remediation technologies has changed markedly since the 1990s. This shift leads to a new perspective on the utility and effectiveness of bioventing compared to other LNAPL remedial alternatives.

A case study is used to illustrate that mass depletion by bioventing often outperforms hydraulic recovery over time. Biodegradation processes enhanced by bioventing deplete LNAPL constituents in mobile and residual LNAPL in the LNAPL smear zone, which means that bioventing can address a larger mass of LNAPL and may induce a beneficial composition change. Hydraulic LNAPL recovery technologies only access the mobile LNAPL and do not induce a composition change. Furthermore, for low LNAPL recoverability (transmissivity), bioventing typically uses less energy and produces less waste per unit mass of hydrocarbon removed than hydraulic recovery, making bioventing a more sustainable remedial technology.

Advances in our understanding of the biodegradation of light non-aqueous phase liquid (LNAPL) and a review of LNAPL mass removal rates through bioventing indicate that bioventing can be an effective approach for addressing both LNAPL saturation (bulk removal of LNAPL) and composition (reducing contaminants of concern) concerns (ITRC 2018a; Gaito et al. 2019). Often the default remedial approach for addressing mobile LNAPL (i.e. free product) to meet regulatory mandates is hydraulic recovery (for the purpose of this Technical Note, absorbent socks, manual bailing, passive skimmers or periodic, short-term vacuum truck extraction events are not considered hydraulic recovery technologies, as noted in ITRC 2018a). Conventional thought suggested that biodegradation of mobile LNAPL did not occur at significant rates (US EPA 1994), and hydraulic recovery was the most effective technology to meet regulatory mandates. Bioventing is a relatively mature remediation technology that stimulates aerobic biodegradation of petroleum-affected soil in the vadose zone by introducing oxygen to the subsurface. Bioventing was established in the late 1980s and early 1990s, and was successfully applied at more than 140 US Department of Defense sites to decrease the total petroleum hydrocarbon (TPH) content in soil to acceptable levels (Leeson and Hinchee 1996a; AFCEE 2004). Bioventing can be performed using active or passive techniques (ESTCP 2004). In active configurations, vadose-zone soils are aerated by either pumping air into the vadose zone or by extracting soil gas, thereby drawing air flow from the atmosphere into the vadose zone, similar to in situ soil vapour extraction (SVE). The decision to use air injection, soil gas extraction or a combination of the two often depends on site-specific conditions. For example, vapour extraction may be the preferred configuration if there are nearby buildings or buried utilities in order to reduce the potential for uncontrolled vapour migration. However, if safe and feasible, an air-injection configuration may be preferred to eliminate the need for off-gas treatment. In addition, air injection can be more effective for addressing petroleum-hydrocarbon-impacted soil within the capillary fringe, which may be inaccessible using extraction techniques due to upwelling of the water table under an applied vacuum (Leeson and Hinchee 1996a, 19b).

While TPH in soil at levels above 18 (diesel) or 106 mg kg⁻¹ (gasoline) is an indicator of LNAPL in soil (Brost and DeVaull 2000), bioventing has not traditionally been considered an appropriate technology for mobile LNAPL. The limited application of bioventing for bulk LNAPL remediation may, in part, be attributed to concerns over the chemotoxic effect on microorganisms in the presence of LNAPL raised in a technical document for screening of remediation technologies (US EPA 1994, chapter III, p. 17):

In general, concentrations of petroleum hydrocarbons in excess of 25 000 ppm [mg kg⁻¹], or heavy metals in excess of 2500 ppm [mg kg⁻¹], in soils are considered inhibitory and/or toxic to aerobic bacteria.

Although no rationale for this aerobic inhibition threshold of 25 000 mg kg⁻¹ of petroleum hydrocarbons was provided, since that time academic research has shown that native microbial communities are equipped with diverse metabolic capabilities that allow them to thrive in LNAPL source zones (Bekins et al. 1999; Whyte et al. 1999; Galperin and Kaplan 2011). In addition, the US Army Corps of Engineers’ technical manual on soil vapour extraction and bioventing, which is widely used by practitioners as design guidance, noted that bioventing is not cost-effective for mobile LNAPL (US ACE 2002, pp. 5–24) and suggests integration with free product recovery systems:

In general, soil vapor extraction or biovent systems are not economical for the removal of significant amounts of free product. Soil vapor extraction has, however, been used successfully to remediate thin (less than 0.15 metre) lenses of volatile LNAPL, such as gasoline. Many SVE systems are operated in conjunction with a groundwater and/or free product recovery system.

This position was flawed in that it did not consider LNAPL distribution within the smear zone above the water table nor the effect of enhanced biodegradation on hydrocarbon mass loss and compositional changes. Furthermore, the position did not consider the value of bioventing compared to hydraulic recovery when LNAPL recoverability is low. It is logical that the cost of LNAPL recovery on a volumetric basis (price per volume recovered) for a hydraulic LNAPL recovery system will be lower if the starting LNAPL transmissivity is measured in the range of single digits up to tens of m²/day than for starting LNAPL transmissivities measured in tenths of m²/day or less. In contrast, the cost for bulk LNAPL destruction via bioventing is not sensitive to LNAPL transmissivity. More recent literature on LNAPL remediation technologies categorizes bioventing as a cost-effective remediation approach compared to other LNAPL recovery technologies (ITRC 2018a). In addition, bioventing is considered a compositional change remediation technology (CL:AIRE 2014; ITRC 2018a), preferentially depleting certain hydrocarbon compounds over others, particularly volatile organic compounds (VOCs). Compositional change can directly address dissolved- and vapour-phase concerns associated with the LNAPL source.

Site owners and practitioners have recently reassessed the applicability of bioventing for petroleum LNAPL remediation. The catalyst for rethinking bioventing is a new understanding of the significance of natural source zone depletion (NSZD) processes for degrading LNAPL mass over time (Garg et al. 2017; Smith et al. 2022). In some cases, NSZD rates of the order of 10 000 l of petroleum degradation per hectare per year have been shown, which deplete LNAPL at higher rates than active LNAPL pumping systems can achieve (Sale et al. 2018). Bioventing is complementary to NSZD in that bioventing effectively enhances the rate of NSZD; NSZD occurs at a higher rate under aerobic conditions than anaerobic conditions that are typically found in the vicinity of petroleum hydrocarbons in natural systems. Given that finding, bioventing has the potential to match or exceed hydraulic recovery rates at many sites where the LNAPL smear zone can be converted from anaerobic to aerobic by the addition of oxygen to the system.

While the implementation methods need to be accounted for, bioventing, either through injection (no off-gas to be treated) or extraction (often requiring off-gas treatment), is also favourable compared to hydraulic recovery (active skimming and pumping) when comparing the resources (e.g. equipment, energy, waste stream, and health and safety) required to implement both technologies. The number of wells required for effective bioventing is affected by the depth to LNAPL impacts. Bioventing is not typically applied for LNAPL impacts less than 3 m below ground surface (m bgs) unless low-permeability surface materials (e.g. asphalt, concrete or silt- and clay-rich soils) are present because surface leakage limits the bioventing radius of influence. Injection bioventing becomes more efficient with greater depth because higher air-injection flow rates can be applied without injected air short-circuiting and discharging to surface; this efficiency allows for larger spacing between bioventing injection wells (potentially larger spacing for extraction bioventing systems). Conversely, hydraulic recovery well spacing does not change as a function of depth but the energy input required to lift recovered liquids to the surface does increase.

Overall, at depths of 3 m or more, bioventing typically requires the installation of fewer wells, has lower maintenance requirements because there are no wetted parts, is less sensitive to fluctuating water-table elevations and does not generate a waste stream, as the LNAPL is depleted in the subsurface. LNAPL recoverability is a key consideration; higher LNAPL transmissivity/recoverability favours hydraulic recovery. However, in cases where LNAPL mass depletion by bioventing is equal to or exceeds mass depletion by hydraulic recovery, bioventing is a more sustainable technology.

A bioventing field study was implemented at a former refinery in the US Midwest to evaluate the application of bioventing to address mobile and residual LNAPL from historical releases of gasoline and middle distillate fuels at the site. The site geology consists of silt and clay overbank deposits to c. 1–3 m bgs, underlain by sand. The upper 50 ft of sand includes channel and point-bar sediments with fining-upward sequences, while the lower sands generally comprise higher-permeability, medium- to coarse-grained outwash sands. Groundwater is c. 9 m bgs in the upper sand unit, with typical seasonal fluctuations of between 7.5 and 9.5 m bgs. The smear zone extends from 3 to 6 m above the water table and as deep as 12 m below the water table, all within the sand unit. The depth of the LNAPL smear zone below the current water table is due to the historical operation of groundwater production wells at the site. These hydrogeological conditions are generally favourable for bioventing. The petroleum hydrocarbon impacts occur within permeable, sandy soils, which allow movement of air and delivery of oxygen to the target portion of the unsaturated zone. The presence of low-permeability silt and clay overbank deposits at the ground surface acts to limit vertical air leakage and extend the bioventing zone of influence (Beckett and Huntley 1994). In addition, the 2 m water-table fluctuations redistribute LNAPL across the smear zone, replenishing LNAPL within the zone of water-table fluctuation, and increasing the mass available to enhanced aerobic biodegradation from bioventing.

Historically, the site remediation actions were focused on LNAPL and groundwater pumping to recover LNAPL mass and hydraulically control impacted groundwater. LNAPL recovery rates at the start of system operations were of the order of 13 800 l/day but the system efficiency decreased over time due to decreases in LNAPL saturation resulting from LNAPL recovery and LNAPL mass destruction by NSZD. By 2014, the LNAPL recovery rate decreased by 97% and the system operations were terminated in 2017 (Fig. 1).

While historical recovery efforts substantially depleted the mobile LNAPL mass, LNAPL still accumulates in wells, and the remaining mobile and residual LNAPL continues to be a source of benzene (and other petroleum constituents) to groundwater.

The ITRC published an LNAPL management technical guidance manual that identifies concerns associated with LNAPL in the environment and a process to establish remedial goals to address these concerns. The ITRC guidance categorizes LNAPL remedial goals as saturation-based goals, composition-based goals and aesthetic-based goals (ITRC 2018a). For this case study, the LNAPL pumping system was directed at achieving a saturation-based goal of removing LNAPL to the maximum extent practicable. While the LNAPL recovery system significantly decreased the LNAPL mass at the site, the recovery action did not materially change the benzene concentrations in groundwater. Concentrations of benzene in groundwater recovered from two groundwater pumping wells (Well A and Well B) from 1994 through to 2014 are shown in Figure 2. Over a period of 20 years, during which nearly 25 × 10⁶ l of LNAPL were removed (Fig. 1), average benzene concentrations decreased by c. 50%. Benzene concentrations remain approximately one–two orders of magnitude greater than maximum contaminant levels (MCLs) (0.005 mg l⁻¹).

Therefore, the next phase of remedial action is focused on satisfying an LNAPL composition-based goal of decreasing the mass fraction of benzene in the LNAPL to improve groundwater quality. The change from a saturation-based goal to a composition-based goal requires a different approach to that for the bulk LNAPL removal programme that was used to satisfy the saturation goal. A bioventing field test was performed to evaluate enhanced biodegradation for both the LNAPL composition change and the bulk LNAPL mass recovery.

The bioventing field study used multiple air-injection and monitoring wells to fully assess the system's performance (Fig. 3). The wells installed to support the field test included:

• Six air-injection wells installed in three nests of shallow and deep injection well pairs to target remediation of LNAPL in the shallow vadose zone and the deep vadose zone near the water table. The shallow injection wells were screened from c. 4.5 to 7.6 m bgs, and the deeper air-injection wells were screened across the water table (c. 9 m bgs) from c. 7.6 to 10.7 m bgs.

• Four nested vapour monitoring points (VMPs) installed at different radial distances from the air-injection wells for measuring changes in the soil gas pressure and composition (i.e. concentrations of O₂, CO₂, CH₄ and volatile petroleum hydrocarbons) in response to air injection.

• Each VMP consisted of two boreholes, each with four nested probe depths, but shown as a single location in Figure 3 for simplicity. Nested VMPs were constructed using polytetrafluoroethylene (PTFE) tubing and stainless-steel AMS probe tips packed in 15 cm sand intervals, and separated from one another by layers of hydrated bentonite. The approximate mid-point depths of the sampling intervals were 3.0, 4.6, 6.1, 6.7, 7.3 7.9, 8.5, and 9.1 m bgs.

• Three piezometers to monitor the effects of the system on shallow groundwater composition, including changes in concentrations of dissolved petroleum hydrocarbons, dissolved gases and potential accumulation of mobile LNAPL in wells in the test area.

• One deep groundwater monitoring well to evaluate the effects of the system on the groundwater composition to the lowermost depth at which LNAPL was measured (c. 10.7–12.2 m below the water table).

A rotary vane air pump was used to push atmospheric air into the bioventing air-injection wells. Automated valves were used to alternate air flow between the three shallow and three deep bioventing air-injection wells. The air from the pump was split evenly between the three wells using rotameters equipped with flow control valves. At system start-up in April 2017, the air-injection rate was c. 25 Sm³ h⁻¹ (where S is standard) or 8.5 Sm³ h⁻¹ per well. The air flow was increased to c. 40 Sm³ h⁻¹ (c. 13.4 Sm³ h⁻¹ per well) in November 2017. Air-injection flow rates and the oxygen concentrations in soil gas measured at nested VMP MP1 during the initial year of operation are included in Figure 4. The soil gas data show that the oxygen content in soil gas was near atmospheric oxygen content (c. 20.5% at the site elevation) while the air pump was operating, and oxygen was generally consumed during system shutdown periods within 1–2 days. Respiration tests were completed during shutdown periods to measure the rates of oxygen utilization (i.e. the rate at which the concentration declined when oxygen was no longer being supplied by the bioventing system) at the nested VMP locations. Oxygen utilization rates measured during the respiration tests were used to estimate the rates of hydrocarbon depletion based on stoichiometric relationships for aerobic oxidation of hydrocarbon compounds as described by Leeson and Hinchee (1996b).

LNAPL mass can be depleted by hydraulic LNAPL recovery, NSZD and bioventing. However, it is difficult to directly compare the technologies because LNAPL mass recovery results are reported in different units. In literature studies, bioventing is often reported as the change in total petroleum hydrocarbon concentrations in soil over time; LNAPL recovery as LNAPL volume produced over time per pumping well or for a well field; and NSZD as LNAPL mass depleted per unit area per unit time. A direct comparison of LNAPL depletion rates by technology found in published studies is presented in Figure 5. The disparate LNAPL depletion rates were converted to equivalent units of volume depleted (litres) per unit area (hectare) per unit time (years) to provide a consistent basis for comparison. (The literature values for NSZD represent estimates from numerous literature case studies, including Lundegard and Johnson (2006), Garg et al. (2017), Concawe (2020), Smith et al. (2021), CRC CARE (2020) and Kulkarni et al. (2020). The representative median NSZD rate for the bioventing case study site was estimated based on methane production rates measured during respiration testing events, when the system was temporarily turned off.)

The literature ranges in Figure 5 overlap and span several orders of magnitude; however, the middle 50% of the ranges indicates that bioventing is generally more effective than NSZD and LNAPL skimming from a mass reduction perspective. This outcome needs to be qualified by the fact that for bioventing to be successful, there must be some LNAPL-affected soil in the vadose zone; bioventing is not effective for confined LNAPL. The estimated initial skimming rate for the case study site (63 000 l ha⁻¹ a⁻¹) is comparable to the median site bioventing LNAPL depletion rate (72 000 l ha⁻¹ a⁻¹). While the initial LNAPL depletion rates are similar, bioventing and LNAPL skimming utilize different remedial mechanisms acting on different portions of the LNAPL. LNAPL skimming and other hydraulic recovery technologies target only the mobile LNAPL, and recovery rates that can be achieved with these technologies typically decline in direct proportion to the remaining mobile LNAPL present in the zone of capture of the system (Sale 2001). Bioventing, on the other hand, targets both mobile and residual LNAPL in the smear zone. Bioventing rates will also decline over time, as the mass of aerobically biodegradable hydrocarbons in the treatment zone becomes rate-limiting.

A comparison of cumulative recovery for LNAPL skimming recovery and bioventing is presented in Figure 6. The cumulative LNAPL depletion from bioventing is derived from field respiration test data collected after operating the bioventing system for 2, 3, 7, 10 and 14 months. The cumulative LNAPL depletion was integrated using the trapezoid rule between measurement events (Fausett 1999). The LNAPL skimming recovery performance was simulated using the API LNAPL Distribution and Recovery Model (Charbeneau 2007). The skimming simulations were run using inputs based on site-specific soil and LNAPL physical properties with an initial mobile LNAPL interval of 0.15 m and an initial LNAPL transmissivity of 0.1 m²/day, which is approximately representative of the maximum gauged LNAPL thickness conditions observed during periods of low water-table elevation in the bioventing study area. Two LNAPL skimming well configurations were modelled and compared to the performance of the bioventing system: one with 16 skimming wells installed in each acre of LNAPL impacts and another with one skimming well per acre.

In the LNAPL skimming simulation presented in Figure 6, the LNAPL recovery rate declines exponentially over time as the volume of recoverable LNAPL in the formation is depleted. While the LNAPL depletion rates for bioventing vary over the 14 months of operation shown in Figure 6, there is no evidence of a consistent decline in depletion rate over time. At sites where declining respiration rates were documented over the initial year of operation, the decline was attributed to the placement of respiration monitoring points in areas where LNAPL was not present (Leeson and Hinchee 1996a). The sustained LNAPL depletion rates observed in the case study are consistent with other case studies in the literature (ITRC 2018a) and the experience of the authors at several other bioventing applications where mobile LNAPL is present within the targeted treatment zone (Gaito et al. 2019).

The analysis presented in Figure 6 compares idealized LNAPL skimming to the performance of the pilot-scale bioventing system. In the simulation, LNAPL is gradually recovered from a consistent mobile LNAPL interval without the influence of water-table changes on LNAPL recoverability. In reality, changes in water-table elevation have a dramatic effect on the hydraulic recovery at the case study site. LNAPL only accumulates in wells during periods of low water-table elevation. During periods of higher groundwater elevation, LNAPL is present as an immobile discontinuous phase and cannot be recovered via skimming. A comparison of cumulative LNAPL skimming recovery and bioventing performance incorporating the influence of water-table fluctuations is included in Figure 7. In the analysis, LNAPL transmissivity and skimming recovery rates are related to fluid-level gauging data from piezometer PZ-2 during the bioventing study. The analysis indicates that LNAPL cannot be hydraulically recovered for more than half of the time due to seasonal water-table fluctuations. Bioventing rates also change in response to water-table fluctuations. However, bioventing continues to act on residual LNAPL in the vadose zone during periods of higher water-table elevation.

Bioventing differs from LNAPL hydraulic recovery in that LNAPL depletion is not uniform for the compounds that make up petroleum LNAPLs. Over time, bioventing changes the chemical make-up of the petroleum LNAPL because monoaromatic (e.g. BTEX) and straight-chain hydrocarbon compounds are removed preferentially and more quickly than highly branched and multi-cyclical petroleum compounds such as isoprenoids and polyaromatic hydrocarbons (AFCEE 2004). This is due to both the volatility of specific constituents and biodegradability (Speight and Arjoon 2012). Bioventing using an air-injection configuration is conceptualized to increase biodegradation in pore spaces within LNAPL-affected soil, as well as areas beyond the LNAPL body as volatile LNAPL constituents are transported away from the source area to other pore spaces where microbial communities and additional electron acceptors exist. Such transport essentially increases the size of the bioreactor (Leeson and Hinchee 1996b). Where applicable, the transport of vapour must be managed by varying the flow rate and depth of oxygen delivery so as not to create a potential vapour intrusion concern for receptors that may be within the zone of influence of the bioventing system. Air-injection bioventing should not be implemented where the protection of receptors cannot be achieved. Alternatively, bioventing using a vapour-extraction configuration may be considered to eliminate the potential for vapour migration. As bioventing remediation progresses, the mass fraction of the monoaromatic and aliphatic compounds in the LNAPL decreases relative to the less degradable compounds. The decreased concentrations of these aromatic and aliphatic compounds in the LNAPL results in a proportional decrease in the effective solubility of the constituents in groundwater (Huntley and Beckett 2002). For this reason, documents such as ITRC (2018a) identify bioventing and other biodegradation mechanisms as inducing compositional change (i.e. the preferential removal of more biodegradable components) to the LNAPL source.

To assess the LNAPL composition change over time, soil and groundwater samples were collected for benzene and total petroleum hydrocarbons (TPHs) during the bioventing field test. The samples were collected before bioventing system start-up and after approximately 1 year of operation. Groundwater samples were collected in January 2017 (system start-up) and July 2018 (after 1 year of operation) from three piezometers (A3PZ-N, A3PZ-C and A3PZ-S) screened in the upper 1.5–3 m of groundwater. Soil characterization included 12 soil samples collected in September 2016 and eight collected in April 2018. All of the soil samples were collected from the treatment area within the historical LNAPL smear zone, from depths of c. 3–9 m bgs. Before and after soil samples were not co-located.

Average concentrations of benzene, gasoline range organics (GRO: C₆–C₁₀ hydrocarbons) and diesel range organics (DRO: C₁₀–C₂₈ hydrocarbons) in soil and groundwater measured prior to system start-up and after 1 year of operation are presented in Figure 8. Additional details, including the range of concentrations measured in soil and groundwater at start-up and after 1 year of operation, are presented in Table 1. The small soil and groundwater sampling dataset, comprising two sampling events, decreases confidence in the interpretation. While it is possible that some of the observed changes in concentrations measured prior to start-up and after 1 year of operation may be, in part, related to factors other than system operation, such as water-table elevation changes (Kehew and Lynch 2011), the results presented for this case study are generally consistent with the performance of other bioventing systems in the literature with respect to changes in the composition of petroleum hydrocarbon impacts. The sample data showed that average benzene concentrations in soil decreased by more than 63% in the first year of operating the bioventing system, and average benzene concentrations in groundwater decreased by c. 30% (Fig. 8). The benzene decrease in soil is somewhat lower than findings of several bioventing evaluations described in the literature. For example, AFCEE (2004) evaluated bioventing performance from 145 sites and found that, on average, benzene decreased 97% within the first year of operation. However, mobile LNAPL was not present in significant quantities at many of the AFCEE (2004) sites. At the case study site, LNAPL is redistributed with changes in water-table elevation, replenishing LNAPL at the base of the vadose zone. Further, depletion of LNAPL in the vadose zone can potentially draw LNAPL from the saturated zone into the vadose zone via capillary processes (Sale et al. 2018), making additional LNAPL mass accessible to enhanced aerobic biodegradation from bioventing.

The concentration of GRO (C₆–C₁₀ hydrocarbons) in soil decreased by c. 32%, and a decrease of c. 35% was observed in the groundwater. The concentration of DRO (C₁₀–C₂₈ hydrocarbons) increased by nearly a factor of 3 in both the soil and groundwater after the first year of bioventing. The increase in dissolved phase DRO may be the result of biodegradation contributing more soluble polar metabolite compounds to the groundwater (Zemo et al. 2017; ITRC 2018b; O'Reilly et al. 2021). The DRO increase in soil is likely to be the result of changes in the water-table elevation. Groundwater elevations were c. 1.2 m higher during collection of the before sample; the sampling intervals were below the water table and mobile LNAPL interval, which moves vertically with water-table fluctuations (Sale et al. 2018). The fluid fluctuation does not represent an overall change in mass but a shift in where the LNAPL signal strength is highest.

While the samples represent the same LNAPL within the system at different overall magnitudes, there was a clear shift in the LNAPL composition. Both benzene and the GRO fraction of the LNAPL decreased during the study. Benzene and GRO are the LNAPL components that are expected to be preferentially depleted; benzene and GRO compounds are more volatile and soluble, and have higher aerobic biotransformation rates than petroleum compounds in the DRO fraction.

Green remediation considers the environmental impact of remediation activities implemented to maximize the benefit of remediation by reducing resource consumption during implementation. In this case study, greenhouse gas (GHG) emissions were selected as a relevant metric to be compared for bioventing and LNAPL skimming. The SiteWise^TM environmental footprint tool was used to estimate the GHG emissions for 1 year of system operation. (SiteWise^TM calculations inputs: the GHG emissions were calculated for system operation only. Systems were assumed to cover a 1 acre area and consisted of 15 bioventing wells or 16 LNAPL recovery wells. Bioventing assumes use of one 7460 W (kg m² s⁻³) blower, operating continuously for 1 year. Skimming assumes use of 16, 280 W pumps, operating for c. 46% of the year, as shown in Figure 7. Installation and waste removal were not considered as part of this assessment.) GHG emissions, assumed to remain constant over the duration of system operation, were divided by the LNAPL depleted during the year to calculate a GHG ratio (GHG emissions rate (mass/time)/LNAPL depletion rate (mass/time)). The GHG ratio is an indicator of the efficiency of the remedial system in terms of GHG emissions; the higher the GHG ratio, the less efficient the remedial system. Figure 9 presents the LNAPL depletion rate and GHG ratio for the bioventing and LNAPL skimming systems. Figure 9 shows that under the modelled configuration, the bioventing system depletes more LNAPL mass (mass-depletion rates: black dashed line, bioventing; black solid line, LNAPL skimming), initially and over the modelled operating period. While the efficiency of the two systems (GHG ratio: blue dashed line, bioventing; blue solid line, LNAPL skimming) were initially equivalent, the decreasing LNAPL recovery rates of the skimming system indicates that long-term operation of the bioventing system is more efficient and sustainable.

The literature and case study data show that bioventing is effective for LNAPL remediation when there is an LNAPL smear zone suitable for bioventing: that is, LNAPL is not confined and the permeability of the LNAPL-affected soil facilitates soil gas exchange. The data analysis in this study yielded the following findings:

• Literature studies and the case study indicate that bioventing can match or exceed LNAPL mass recovery achieved by LNAPL hydraulic recovery.

• Bioventing is effective for modifying LNAPL composition, demonstrated by the finding that benzene and GRO were preferentially depleted in the case study.

• Bioventing requires less infrastructure, maintenance and waste handling, and is preferential to LNAPL hydraulic recovery when considering the environmental footprint associated with remedy implementation.

Bioventing should be considered as a remediation option at LNAPL sites as an alternative to, or complement to, hydraulic recovery. In cases where a hydraulic recovery system is already operating, field testing of bioventing is recommended to determine an efficient transition point from hydraulic recovery to bioventing, which may occur much earlier in the operational life cycle of the recovery system than historically assumed.

The authors would like to thank the members of the Bioventing Forum, a collection of industry and regulators including the API Soil and Groundwater Technical Group, EPA, and various state regulators, for their comments and contributions. The paper was improved by review comments from Tom Fox, CO Division of Oil & Public Safety, Denver, CO. The authors would like to thank the reviewers for their valuable comments, which helped improve the quality of this manuscript. The work reflects the views of the authors and may not reflect the views or policy of the members of the Bioventing Forum.

JJS: formal analysis (equal), writing – original draft (equal), writing – review & editing (supporting); STG: formal analysis (equal), writing – original draft (equal), writing – review & editing (supporting); BWK: formal analysis (supporting), writing – original draft (equal), writing – review & editing (lead).

This work was partially funded by the API Soil and Groundwater Technical Group.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The datasets generated and/or analysed during the current study are not publicly available because the authors do not maintain public data repositories but are available from the corresponding author on reasonable request.