Environ. Sci. Technol. 1999, 33, 1726-1731
Effect of Flow Rate Changes and Pulsing on the Treatment of Source Zones by in Situ Air Sparging PAUL C. JOHNSON,* AMARJYOTI DAS, AND CRISTIN BRUCE Department of Civil and Environmental Engineering, Arizona State University, Tempe, Arizona 85268-5306
Laboratory-scale two-dimensional aquifer physical model studies were conducted to qualitatively assess how changes in air injection rate, air injection pulsing, and chemical type affect the rate and extent of removal by in situ air sparging. In this work, the treatment of immiscible-phase source zones in coarse media has been simulated. To provide a basis for comparison, equivalent unsaturated-soil soil vapor extraction-like simulation experiments were also conducted. Results suggest that, (i) initially, removal occurs from within saturated zone air flow channels at rates similar to those observed for soil vapor extraction, (ii) during this initial period, removal rates are proportional to air flow rate and equilibrium chemical vapor concentrations, (iii) while increased air injection rates improve volatilization rates in the short-term, the long-term cumulative removal efficiency may not be affected unless the saturated zone air flow distribution changes significantly with flow rate, (iv) pulsing the air injection can improve the long-term cumulative removal efficiency, and (v) while short-term removal efficiency improves with increasing flow rate and vapor pressure, the long-term removal efficiency appears to improve with increasing solubility.
Introduction In situ air sparging (IAS) generally involves the injection of air into an aquifer in order to volatilize and biodegrade contaminants trapped beneath the water table or within the capillary zone (1). Applications include the treatment of immiscible-phase hydrocarbon source zones, dissolved contaminant plumes, and the use of IAS as a barrier to dissolved contaminant plume migration (2-4). While air is generally the injected gas, other vapors may be mixed with the injection air stream to promote biodegradation; for example, butane might be blended with the air to promote cometabolic degradation of a chlorinated solvent. In situ air sparging has traditionally been used in conjunction with soil vapor extraction, where the chemical vapors liberated by the injected air are collected and treated by the soil vapor extraction system. Reports from field applications indicate that in situ air sparging has been very effective at some sites, and less effective at others (2, 3). It is difficult, however, to determine if these results are due to different site conditions (hydrogeologic properties, contaminant type, etc.) or to varied design practices and operating conditions. It is not clear from the mostly anecdotal reports how changes in operating * To whom correspondence should be addressed. Phone: 602965-9115; Fax: 602-965-0557; E-mail:
[email protected]. 1726
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 10, 1999
conditions affect system performance. Furthermore, our ability to anticipate how process and design changes might affect system performance is limited at this time, except in a very general sense. There are a number of routes one can follow to gain a better understanding of how process changes might impact in situ air sparging performance. Field-scale studies (5-9), physical model experiments (10-14), theoretical analyses (11, 15, 16), and numerical simulations are options. Each has advantages and limitations, is complementary to the others, and is integral to developing a complete understanding. In this work, physical model studies were conducted because (a) field-scale studies often have long time scales (weeks to months) and costs are high, so that the range of conditions that can be practically studied is limited; (b) proven numerical simulators have yet to be developed; and (c) relative to actual field studies, physical models are more easily monitored and characterized, and the experiments are of much shorter duration (days to weeks, as opposed to months to years in the field). Physical model studies have been used in the past to better understand in situ air sparging performance. Ji et al. (10, 11) conducted flow visualization studies in a 2 ft × 2 ft × 1 in. (approximately) transparent tank. The tank was packed with various sizes of glass beads and several model geologies were simulated. The goal of the study was to observe how injected air distributes itself beneath the water table, and how this distribution is affected by particle size, stratigraphy, and air injection flow rate. The reader is referred to the original publication for excellent pictures of the air distributions observed during their studies. Of particular importance for this work is their visual observation of changes in air distribution with increasing flow rate. Starting at very low air injection rates, increases in air injection rate corresponded to increases in air flow channels and zone of air flow. At higher air flow rates, however, increases in air flow rate did not cause substantial changes in the size of the zone of air flow. This was also observed by Rutherford and Johnson (12, 13) who used a laboratory-scale physical model to study the effects of process changes on oxygen mass transfer rates. Of note for this work were their observations that increases in air injection rate did not increase oxygenation rates at the higher flow rates (in fact the opposite was true), and pulsing the air injection did not seem to enhance the time-averaged oxygen transfer rate either. Physical models may also be of a very large scale. For example, Bruce et al. (14) report on the use of two large-scale aquifer physical models measuring 3 m × 0.3 m × 10 m and 2 m × 2 m × 6 m (approximately) for MTBE treatability studies. It should be noted that there are limitations to the use of physical models. In the absence of validated predictive models, quantitative results from physical model studies are not easily extrapolated to any real field setting. Thus, one must be cautious when extrapolating physical model results to field applications. Here, it is our hypothesis that, if the experiment is well planned, then qualitative trends observed in the physical model studies should extend to field sites. For example, if one were to determine in a physical model that increases in air injection rate improve performance, it would be assumed that changes in the air injection rate would also improve performance at a field site having characteristics and operating conditions representative of the physical model study. The work described here focuses on the removal of immiscible-phase hydrocarbons from relatively homogeneous and coarse soils. These physical model experiments 10.1021/es9807688 CCC: $18.00
1999 American Chemical Society Published on Web 04/10/1999
FIGURE 1. Schematic of aquifer physical model. were designed to yield a better understanding of the use of air sparging for source zone treatment and how changes in air injection rate, air injection pulsing, and chemical type might affect performance.
Experimental Apparatus The two-dimensional aquifer physical model used in this work and shown in Figure 1 has been described in detail previously (12, 13). It is 2.4 m (8 ft) long, 1.2 m (4 ft) tall, and 5.1 cm (2 in) wide and is constructed primarily from Plexiglas in order to facilitate visualization of the air flow field. A compressed air supply line is connected to a port located at the bottom of the physical model. This port is connected to a diffuser located 1.2 m from either side of the tank and approximately 10 cm above the bottom of the tank. The air diffuser is constructed from a short section of 2.54 cm (1 in.) diameter PVC pipe that had been manually perforated with a hand drill. A flow meter (rotameter) and pressure regulator are used to measure and regulate the influent air stream. Most of the outlet air from the physical model flows directly to a flow meter and then to a carbon bed before discharge to the atmosphere. As shown in Figure 1 and discussed elsewhere (12, 13), the physical model is constructed to allow the user to impose a horizontal water flow; however, the experiments reported here were conducted under conditions of no net bulk flow across the tank. A slip-stream of off-gas is drawn through a 1 mL sample loop and gas sampling valve connected to a gas chromatograph (GC) equipped with a flame ionization detector (SRI Instruments, Torrance, CA model 9300B). A software-driven data acquisition system (PeakSimple for Windows-SRI Instruments, Torrance, CA) is used to automate gas sample analysis and data storage. Separation is achieved by a 15 m MXT-1 column (Restek Corporation, Bellefonte, PA). Off-gas sample analyses were conducted every 4 min during each of the experiments described here. For this work, the aquifer model was packed with 1 mm diameter glass beads (Dragonite grinding media by Jaygo) (12). For reference, IAS applications have been reported for
media ranging from silts to coarse sands and gravels; thus, this model media would be representative only of the more coarser media IAS applications. For most of the experiments discussed here, liquid hydrocarbon is introduced to the physical model through a small diameter injection tube that passes through the physical model’s top center and ends at a point roughly 100 cm from the bottom of the physical model (≈20 cm below the top).
Experimental Protocol Three types of experiments were conducted in this study. All used the same granular media, process flow conditions, chemicals, and the source zone creation procedure described below. The first category of experiments was designed to simulate remediation by in situ soil vapor extraction. Here, the immiscible-phase hydrocarbon source zone was located above the water table in unsaturated media, rather than below the water table in water-saturated media as in the air sparging experiments. As many practitioners have experience with in situ soil vapor extraction, these experiments were conducted to provide a baseline against which equivalent in situ air sparging experiments could be compared. The second and third categories of experiments were designed to simulate removal by in situ air sparging under steady and pulsed air injection, respectively. The experimental protocol included the following steps: (1) the gas chromatograph was first calibrated using Tedlar bags containing known concentrations sampled through the aquifer physical model sample loop; (2) the water table was raised to a level 5 cm below the liquid hydrocarbon injection tube outlet, with the injection tube outlet being ∼100 cm from the bottom of the physical model; (3) 50 mL of selected hydrocarbon was injected into the injection tube with a syringe; (4) the fluid was allowed to spread across the water table for ∼10 min; (5) the water table was lowered to a height of about 0.3 m (1 ft) from the bottom of the tank by draining water from the bottom ports; (6) for air sparging simulations, the water table was raised back to the original level used in step (2); otherwise, for soil vapor extraction simulations, the VOL. 33, NO. 10, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1727
TABLE 1. Properties of Chemicals Used in the Experimentsa compound
G (g/mL)
Mw (g/mol)
HI
bp (°C)
SI (mg/L)
Pv (atm)
C vI (mg/L)
octane hexane
0.7028 0.6603
114 86
93 43
126 69
0.7 13
0.014 0.16
66 572
a Verschueren (1977). F ) density; bp ) boiling point at 1 atm absolute pressure; Mw ) molecular weight; Pv) pure component vapor pressure at 20 °C; HI ) CvI /S; HI ) dimensionless Henry’s Law constant; SI ) solubility at 20 °C; CvI ) saturated vapor concentration at 20 C (mg/L).
water level was maintained at 0.3 m (1 ft) from the bottom of the tank; (7) automated GC vapor sampling and analysis was initiated; (8) air injection was initiated at the predetermined flow rate and flow conditions and flow and sampling are continued until the end of the experiment; (9) in order to complete the mass balance, at some predetermined time (generally ∼3600 min), the water table was lowered to expose all hydrocarbon-contaminated media and air injection and sampling/GC analysis was continued until vapor concentrations declined to about 0.01 mg/L. These experiments were conducted for two chemicals (octane and hexane), two steady air injection rates (1.25 and 10 L/min), and one pulsed flow condition (10 L/min; 200 s on and 20 s off). In addition, experiments were conducted under exposed and submerged-source conditions as discussed above. Octane and hexane were selected as their vapor pressures and solubilities fall within the range of properties for many chemicals targeted for in situ air sparging applications. Chemical and physical properties of these chemicals are given in Table 1. Hexane has the greater saturated vapor concentration and aqueous solubility, while octane has a greater Henry’s Law constant. The initial hydrocarbon volume was chosen to ensure that, after spreading on the water table, all of the hydrocarbon would remain within the air flow field. The air injection flow rates were selected based on the work of Rutherford and Johnson (12, 13), which used the same physical model and media. In this flow rate range, the air distribution was roughly conical above the air injection point based on visual observations of increased air saturation in this region. The gross features of the flow field were very similar to those shown in pictures presented by Ji et al. (10) for homogeneous media and medium to high flow rates. Upon close inspection, preferential flow channels were observed within the zone of desaturated media. Between air injection flow rates of 1 and 10 L/min, changes in air flow rate did not substantially change the air flow distribution based on visual observations; however, the overall water saturation and permeability to water flow did decrease with increasing air flow rate within the air flow zone (12, 13). The pulsing frequency (200 s on, 20 s off) was based on visual observation of aquifer relaxation following the start and end of air injection, and the on-off durations were chosen to be slightly longer than the times for dissipation of the water mound at the start of air injection, or recovery of the water trough formed when the air was turned off. With respect to the initial distribution of contaminant, it was visually observed to be present as an immiscible residual hydrocarbon phase. On the basis of equilibrium partitioning calculations,
