In accordance with MADEP requirements, and to further assess the extent of the plume's vertical migration, three groundwater monitoring wells, MW-1 through MW-3, were installed in the area of the release. To reduce the possibility of puncturing vadose zone confining layers (if any were present), or introducing drilling fluids to the subsurface (which could enhance vertical migration of the plume), the wells were installed in cased boreholes using a dual air rotary drill rig. Split spoon samples collected during the installation of these wells provided supporting information regarding subsurface conditions to a depth of 50 feet, and new information regarding the subsurface conditions below 50 feet. In general, as indicated previously, the subsurface soil conditions at the CCAFS consist of fine to coarse sand with little silt to a depth of approximately 220 feet, the depth at which the boreholes were terminated. The surface of the unconfined aquifer was encountered at a depth of approximately 210 feet. Based on field monitoring of samples collected from well MW-1, installed in the area of the release, and the information collected during the installation of wells B-1 through B-5, we estimated that, at the time, the plume encompassed the area shown in Figure 2, and had migrated to a depth of approximately 75 feet in the area of the release.

The purpose for the bioventing pilot test was threefold: (1) to assess whether the existing vadose zone monitoring wells could be incorporated into the system design, and/or whether the installation of additional extraction wells would be required; (2) to develop a calibration database for the numerical model which would be used to identify the range of flow and vacuum requirements needed to effectively aerate the volume of soil within the vadose zone plume, and; (3) to assess the potential need for air pollution control equipment.

The pilot test was performed using GZA's trailer-mounted mobile pilot testing system which consists of a regenerative blower and a propane-fueled catalytic incinerator. Ten vacuum monitoring probes constructed of 1/2-inch diameter black-iron pipe were installed at depths ranging from five to twenty feet at locations within 35 feet of the spill area. Each of the probes were fitted with vacuum-tight PVC caps equipped with barbed compression fittings. When not used as extraction wells, wells B-1, B-2, and B-4 were equipped with similar wellhead fittings and used as monitoring points.

Seven separate pilot tests were conducted at flow rates ranging from 50 to 150 standard cubic feet per minute (scfm) and corresponding vacuums of 7 to 29 inches of water, at the two designated extraction wells, B-1 and B-2, over the course of a two-day period. During each of these three-hour tests, vacuums in the probes were monitored using magnahelic gauges, volatile organic compound (VOC) concentrations were monitored at the intake and exhaust of the incinerator using portable field instruments, and additional air samples were collected in teflon bags for subsequent laboratory analysis.

The results of the pilot tests indicated that:

• the possible aeration zones of the 2-inch diameter extraction wells extended beyond the estimated limits of the plume - therefore they could be incorporated into the final design of the system, thereby limiting the need for the installation of additional shallow extraction wells;
• the extracted soil gas would contain at least during the initial phases of remediation, approximately 150 µg/1 (or, assuming an average molecular weight of 92 g/mole, approximately 37 parts per million (ppm)) of total non-methane hydrocarbons (TNMHC). Because this concentration equates to a mass emission rate of approximately 6.7 pounds of TNMHC per operating scfm per year, the system would exceed the Massachusetts Air Pollution Control Regulations TNMHC mass emission criteria of 1 ton per year at operating flow rates of greater than approximately 400 scfm. Therefore, depending on the design flow rate of the full-scale system, air pollution control measures may be required, and;
• measurable vacuums were detected in the vacuum monitoring probes installed approximately 35 feet from the extraction wells. These results were used as a calibration database for the three-dimensional flow model.

Three Dimensional Numerical Flow Modeling

To estimate the flow and vacuum requirements of the full-scale system, GZA developed a 50,000 sq. ft., 19,800 node finite difference numerical flow model of the site based on MODFLOW groundwater code developed by M.G. McDonald and A.W Harbaugh of the U.S.G.S. The MODFLOW code is particularly well-suited for modeling the flow of air through porous media due to its flexibility, modular design, its 3D capabilities, and the variety of boundary and miscellaneous fluid source and sink options it contains. While the resultant model is based on several simplifying assumptions of the true soil gas flow regime, particularly those regarding the relative incompressibility of air at normal bioventing system operating vacuums, it can provide a reasonable approximation of this regime.

The input parameters for the CCAFS flow model were initially developed based on the information obtained during the installation of the boring logs B-1 through B-5. This model was run several times to assess potential flow and vacuum rates for the initial pilot test. Once the pilot test was completed, several stress scenarios were developed to model the individual pilot test scenarios. For each of these scenarios, pneumatic transmissivity, conductance, and boundary condition parameters were adjusted until the head and flow outputs from the model matched those observed in the field. Once the calibrated model was developed, several full-scale extraction scenarios were modeled to identify combinations of existing and proposed well locations and extraction flow rates and vacuums which would (1) create an aeration zone which extended beyond the identified extent of contamination, and (2) remediate (through the combined processes of biodegradation and volatilization) the volatile fraction of the fuel oil within one year.

The results of this modeling effort indicated that several different combinations of well locations and system operating capacities would meet the remedial goals. Therefore, in an effort to reduce the marginal costs associated with the implementation of the full scale system, the system was specified such that the subsurface extraction system would consist of four extraction wells three existing vadose zone wells (B-1, B-2, and B-4) and one new, deep extraction well (EW-2), from which soil gas would be extracted at a total flow rate of 1,500 scfm and a total system vacuum of 7 in. of Hg. (gauge). Vacuum contours developed from the model results of this scenario at a depth of 50 feet below the ground surface are shown in Figure 3. Based on these results, it was expected that the system would develop vacuums of up to 0.5 in. of Hg. at the outer edges of the plume, and the selected system configuration would easily deliver oxygen to the fuel oil contaminated soils.

The design of the full-scale bioventing system incorporated features which (1) addressed the design parameters established during the additional investigation and modeling tasks, (2) allowed for inherent flexibility and on-going monitoring capabilities, and (3) addressed logistical issues associated with operating the system at a high security military installation.

Based on the results of the pilot testing and modeling programs, the system was designed to extract approximately 1,500 scfm from four separate extraction wells at a total vacuum of 10 in. of Hg. (gauge) (including 3 in. of Hg. For piping losses). The system consisted of the three existing extraction wells (B-1, B-2, and B-4), one new extraction well, EW-2, which would be installed to a depth of approximately 90 feet below the ground surface, three multi-level vacuum and VOC monitoring well, and the process components installed in two mobile trailers (Figure 4). The process components consisted of: a multistage centrifugal blower, an air/water separator, an in-line air heater, two 3,000 pound activated carbon adsorption units, an air monitoring system, and associated piping, electrical, mechanical, and instrumentation equipment. Important aspects of the system design included:

• Subsurface Equipment. As previously mentioned, three of the existing monitoring wells were used as extraction wells in the full-scale system. To address contamination below 50 feet, a four-inch diameter extraction well, EW-2, was installed to a depth of 90 feet. The well was installed deeper than the identified depth of contamination to account for vertical migration of the plume which may have occurred between the time the well was installed and the time the system was brought online. In addition to this deep extraction well, three vapor monitoring wells (VM-1, VM-2, and VM-3) were installed. These wells consisted of two foot sections of 1/2-inch diameter PVC screen which were installed at three discrete depths (i.e., l5 feet, 35 feet, and 70 feet below the ground surface) in an uncased borehole. The monitoring points were connected to the surface via 1/4-inch copper tubing. To reduce the possibility that the borehole would become a preferential air flow pathway, the area between the monitoring points within a borehole was backfilled with a dry sand and bentonite powder seal. The dry seal reduced the possibility that the extraction air flow would shrink the seal and create preferential air flow pathways. The vacuums and TNMHC concentrations in the subsurface are monitored periodically to assess current system operating conditions and provide information used to modify operating parameters.

• Air pollution control system. Based on the results of the pilot test, which indicated the likelihood that TNMHC emissions from the system would exceed 1 ton per year, air pollution control equipment were included in the system. Engineering analysis of the potential costs associated with air pollution control indicated that, due primarily to the high required flow rate and relatively low expected VOC concentration (and, therefore, BTU content) of the extracted soil gas, activated carbon adsorption would be more cost effective that other air pollution control systems (e.g., incineration). To increase the efficiency of the activated carbon system: (1) the adsorption units were installed on the vacuum-side of the blower (because the expected temperature of the blower discharge would be approximately 175 degrees F, installing the adsorption units at the discharge of the blower would severely reduce the adsorptive capacity of the activated carbon); (2) an inline heater was installed upstream of the adsorption units; the purpose of the heater would be to slightly raise the temperature of the influent air stream to above the dew point, thereby reducing the possibility that water vapor would condense in the pores of the activated carbon (water condensation in vapor-phase activated carbon beds can reduce the overall TNMHC adsorption capacity of the beds), and; (3) the adsorption units were piped such that they could be operated in an alternating "lead-lag" configuration; once one bed is "exhausted", the second bed is reconfigured as the first, and the "exhausted" bed is replaced with fresh activated carbon and operated as the second canister. In this manner, the full capacity of each bed is used prior to regeneration.

• An automatic computer-based data acquisition and control system (DACS). To facilitate operation of the system at the high security facility, GZA developed a C-based DACS specifically for the CCAFS project. The system is capable of monitoring upwards of 32 digital and analog instrument signals from level switches, pressure, flow and temperature transducers, and the air monitoring system, logging the information digitally, controlling process components based on user-input operational parameters, and allowing for remote data collection and real-time monitoring. If certain shutdown criteria are exceeded, or if the system detects a power failure, the DACS will execute appropriate control actions, and notify GZA's office, via a telephone of the condition. For example, approximately hourly, air samples from eight inline sampling locations are collected and analyzed by the PID and the results of these analysis are transmitted to the DACS. If TNHMC concentrations in the samples collected between or after the activated carbon units indicates that either breakthrough or exhaustion of the beds has occurred, the system will notify us of this condition. If TNMHC concentrations in the stack gas exceed those specified in the system's air pollution control registration, the system will shut down and notify us that the discharge criteria has been exceeded. The use of the DACS greatly reduced the need for frequent operation and maintenance visits and, therefore, reduced the overall cost of the project.

Once the design of the system was approved by the Air Force and the MADEP, the system was constructed on a fast-track basis in approximately one month by GZA Remediation, Inc., GZA's fabrication subsidiary, installed at the site, and placed online.

SYSTEM PERFORMANCE

To assess the effectiveness of the system in remediating subsurface soils via both volatilization and biodegradation, several operational parameters have been monitored on an ongoing basis during the first 50 days of system operation. These parameters include:

• Vacuums in the extraction and vapor monitoring wells;
• TNMHC concentrations in soil gas samples collected from the individual vapor monitoring probe sample locations, and;
• TNMHC and carbon dioxide concentrations in the extracted soil gas.

Subsurface Vacuums

The effective aeration zone of the bioventing system has been assessed by monitoring the vacuums in both the vapor monitoring and extraction wells. Measured operating vacuum contours for a 50 foot depth below the ground surface are shown in Figure 5. These contours indicate that, although the radial influence of the system is not as large as the modeling efforts indicated it might be, it is more than sufficient to promote aeration of the subsurface soils impacted by the fuel oil. For example, measured vacuums in vapor monitoring well VM-3, which was installed at the outer edge of the plume, have been in the range of 0.2 to 0.3 in of Hg (rather than in the range of 0.6 to 0.7 inches of mercury, as predicted by the model) indicating that the effective flow field of the system extends well beyond this range. Based on a comparison of actual measured vacuums versus modeled vacuums, we have concluded that the difference between the model results and actual field measurements is due to both a slight overestimation of both the horizontal and vertical pneumatic conductivity of the subsurface soils, and the inability of the model to simulate the head losses associated with the higher, possibly turbulent air flow through the soils in the immediate area of the extraction wells. Recently, we have begun investigating the limitations associated with using MODFLOW to model high flow vapor extraction systems, and will present the results of this investigation in later papers.

TNMHC concentrations in the subsurface soils have been monitored by collecting soil gas samples from the individual sampling probes installed in each of the vapor monitoring wells. As previously mentioned, vapor monitoring probes were installed at three depths, 15 feet (shallow-s), 35 feet (middle-m), and 70 feet (deep-d) below the ground surface, in each of the vapor monitoring wells, VM-1 through VM-3. A comparison of measurements made prior to the installation of the bioventing system and measurements made after approximately 50 days of system operation (Figure 6) indicates that in most of the vapor monitoring probes, particularly those located at the shallow and deeper depths, the reduction of TNMHC concentrations has been as high as 90 percent. TNMHC concentrations have increased slightly in the middle sampling locations of vapor monitoring wells VM-1 and VM-2 due, most likely, to the slight downward and horizontal migration of the fuel oil at shallow depths to the deep extraction well, EW-2, which is screened from 50 to 90 feet below the ground surface.

The combined results of monitoring both vacuums and TNMHC concentrations in the vacuum monitoring probes indicates that (1) the effective aeration zone produced by the bioventing appears to be sufficiently supplying air (and, therefore, oxygen) to the fuel oil contaminated soils, and (2) the introduction of oxygen to the subsurface and stripping of the contaminants has effectively remediated a significant percentage of the volatile fraction of the fuel oil contamination.

The overall effectiveness of the bioventing system can be tracked by monitoring TNMHC and carbon dioxide concentrations in the extracted soil gas. TNMHC concentrations can be directly equated to the total mass of volatile hydrocarbons volatilized. Carbon dioxide, which is produced as a by-product of aerobic biodegradation processes, can be indirectly equated, via stoichiometric calculations, to an estimate of the total mass of both volatile and semi-volatile hydrocarbons biodegraded. For the CCAFS project, TNMHC concentrations are measured and recorded continuously by the continuous air monitoring instrument and DACS. Carbon dioxide concentrations are measured by a portable infrared monitoring instrument during periodic monitoring events.

Since the system was placed online, TNMHC concentrations samples collected from each extraction well have remained relatively stable, with the exception of an initial spike during the first 30 minutes of system startup. TNMHC concentrations in the total extracted soil flow rate of 1,200 to 1,500 scfm have been approximately 20 ppm, slightly lower than the expected (based on the pilot test results) 37 ppm. We believe that this difference is due primarily to the facts that because (1) the effective aeration zone produced by the system extends beyond the limits of identified subsurface contamination, and; (2) the air velocities through the pores of the subsurface soils are higher than those induced by the lower flow pilot test, the equilibration time between the clean air drawn through the subsurface and the fuel oil contamination is lower than in the pilot test and, therefore, TNMHC concentrations in the extracted soil gas are lower. Based on the results, and assuming the specific gravity of the product is 0.8, we estimate that the CCAFS bioventing system has volatilized an average of approximately 0.75 to 1 gallon of hydrocarbons per day for the first 50 days of system operation.

Carbon dioxide concentrations in the soil gas are generally measured as a percentage of the total extracted soil gas volume. Ambient air generally contains approximately 0.03 percent carbon dioxide; C02 concentrations in the soil gas extracted from clean or background soils may contain up to 0.5 percent C02, depending on the total organic carbon (TOC) content of the soils. Due to the sandy nature of the soils at the CCAFS, TOC concentrations are relatively low, and, correspondingly, C02 concentrations in soil gas extracted from "clean" soils should be less than 0.2 percent. During the initial startup of the CCAFS bioventing system, C02 concentrations in the extracted soil gas were approximately 1 percent due the presence of residual C02 in the pores of the subsurface soils resulting from low-level anaerobic and aerobic biodegradation of the fuel oil (C02 is a also a byproduct of anaerobic biodegradation). After approximately 6 hours, C02 concentrations in the extracted soil gas began to decrease to approximately 0.5 percent (i.e., background). However, after approximately 1 day of operation, C02 concentrations began to increase rapidly until, after approximately one week of system operation, C02 concentrations in the total extracted soil gas flow leveled off at about 1 percent. Monitoring of each of the individual extraction wells indicates that C02 concentrations in wells B-l, B-2, and B-5 are approximately 0.75 percent, while C02 concentrations in EW-2, the deeper extraction well installed directly in the area of the release, are approximately 2 percent. Based on the results of these periodic monitoring events, and conservatively assuming that the background C02 concentration in the extracted soil gas is 0.5 percent, we estimate that the system is biodegrading approximately 40 to 50 gallons of hydrocarbons per day during the first 50 days of system operation.

The combined effects of volatilization and biodegradation are illustrated in Figure 7. After 50 days of operation, a total of approximately 2,200 gallons of hydrocarbons have been remediated by the CCAFS bioventing system approximately 2 percent, or 45 gallons, have been volatilized from the subsurface, and approximately 98 percent, or 2,150 gallons, have been biodegraded. At this point, TNMHC extraction and C02 evolution rates have been fairly steady. We fully anticipate that as the system continues to operate, TNMHC extraction rates will decrease, and biodegradation will become the overriding remedial mechanism.

In response to an accidental release of 11,000 gallons of virgin No. 2 fuel oil at the CCAFS in Sagamore, Massachusetts, the Air Force worked closely with the MADEP to coordinate the installation of a remedial system designed to remediate the most mobile fraction of the contaminant before it reached the underlying sole source drinking water aquifer. Ironically, the high permeability of the sandy soils at the site which allowed the contaminants to quickly migrate vertically to a depth of approximately 75 feet also greatly enhances the effectiveness of the selected remedial technique - bioventing. With the input of both the Air Force and the MADEP, GZA performed a streamlined subsurface investigation program consisting of additional boring installations, pilot testing, and numerical flow field modeling. To reduce the overall cost of the project, several issues were considered when developing the final system design: (1) existing borings installed during emergency response activities were used as extraction wells, (2) the air pollution control system, which consists of activated carbon canisters rather than an incinerator due to the relatively low BTU content (and, therefore, high supplemental fuel requirement) of the extracted soil gas, is installed in a "lead-lag" configuration on the vacuum-side of the blower and is equipped with an inline air heater to increase carbon efficiency, (3) vapor/vacuum monitoring wells were installed during the installation of the deep extraction wells - information collected from these monitoring points is used to adjust system operational parameters, and; (4) the system is equipped with a computer-based DACS which completely monitors the system and reduces overall operation and maintenance requirements.

After 50 days of operation, the system has remediated over 2,200 gallons of hydrocarbons - 2 percent has been removed via volatilization, and 98 percent has been remediated via biodegradation. TNMHC concentrations in the subsurface have decreased as much as 90 percent in some areas. Periodic monitoring of vacuums in the subsurface indicate that effective aeration zones extend well beyond the identified limits of contamination.