Naval Air Station (NAS) Lemoore Site 17 JP-5 Product Recovery

The NAS Lemoore Site 17 product recovery system was constructed to extract a plume of JP-5 fuel that remained after previous efforts at product recovery were only partially successful.  The system was designed to extract mobile free product, residual product and vapor-phase product from the saturated and unsaturated zones.  The system was operated in multiphase extraction (MPE) mode for 28 months until the mobile free product had been removed, and was subsequently converted to bioventing mode for 10 months (as of May 2006) to stimulate aerobic biodegradation and removal of vapor-phase hydrocarbons.  When system operation began in January 2003, the average apparent thickness of free product observed in the 89 monitoring wells was approximately one foot and some wells contained more than three feet of product.

Recent product thickness measurements at Site 17 indicate that all of the 89 wells at the site now meet the California cleanup standard for free product of a sheen (less than 0.01 foot apparent thickness).  The Site 17 MPE system recovered more than 28,000 gallons of free product while it was operated in MPE mode from January 2003 to May 2005.  The system is now operating in bioventing mode, which requires significantly less energy and expenditure on operations and maintenance (O&M). 

Site Background

A leak was discovered in an underground 16-inch JP-5 jet fuel pipeline in 1987.  Excavating and pumping free product that was accumulating in the excavations recovered approximately 138,000 gallons of free product.  In 1994, a steam injection/vapor extraction (SIVE) system operated at the site for 3 months by another contractor.  The SIVE system recovered an additional 80,000 gallons of product, until the site appeared to have been successfully remediated. 

In 1997, Tetra Tech EM Inc. (Tetra Tech) began to monitor the site in preparation for remediation and site closure through monitored natural attenuation (MNA).  During monitoring, Tetra Tech found that free product had re-accumulated in a number of wells at the site.  It is not clear whether the free product re-accumulated because (1) the SIVE system footprint did not cover the entire plume, (2) vapor recovery was incomplete and JP-5 condensed in the vadose zone, or (3) free product moved laterally from areas that were not monitored in response to fluctuations in the water table, or a combination of these factors.

Pilot Testing

Tetra Tech conducted pilot testing of product skimming and multiphase extraction at the site.  Conventional product skimming rates were extremely low (less than 1 gallon per day) because of the moderately low hydraulic conductivity in the silty sand aquifer and the higher viscosity of JP-5 compared with gasoline.  MPE pilot testing demonstrated much higher product recovery rates, ranging from 1 to 8 gallons per minute.

MPE Configuration

The MPE system used at the site employed high-vacuum liquid ring pumps to draw soil vapor, free product, and groundwater from a single drop tube placed in wells screened across the water table. 

Initial operation began with the drop tube tip placed at the oil/water interface in the extraction well.  Initially, extracted liquid contained nearly 10 percent product, but this level quickly dropped to less than 1 percent.  Typical vacuum applied to each of the 35 extraction wells was 20 to 25 inches mercury and a vapor flow rate of 8 to 12 standard cubic feet per minute (scfm).  Typical liquid flow rates were 0.5 to 0.75 gallons per minute (gpm) per extraction well because of the low hydraulic conductivity of the site. 

Site 17 employed 35 extraction wells placed at 35-foot centers, based on a radius of influence (ROI) of 20 feet.  Pilot testing indicated a vacuum ROI of 20 feet and a groundwater drawdown ROI of 20 to 50 feet.

Extracted Liquid Treatment

Extracted liquid was treated using a treatment train consisting of sedimentation, oil/water separation, coagulation/flocculation, dissolved air floatation, and granular activated carbon (GAC) absorption.  Treated effluent is discharged to an infiltration gallery 1,000 feet from the site.

Treatment Problems

Two liquid treatment problems were encountered:  emulsification and biofouling.  Emulsification caused by shearing of the JP-5 during the extraction process created a mechanical oil-in-water emulsion, despite the use of low-shear pumps.  Colloidal clay and water high in total dissolved solids (TDS) at the site caused the emulsion to be stable indefinitely.  Thus, the emulsified oil droplets (ranging in size from 0.01 to 2 microns), were too small and neutrally buoyant to be removed by gravity separation and too large to be removed efficiently by GAC adsorption.  As a result, coagulant and flocculant addition was necessary to agglomerate the emulsified fuel.

Biological growth in the recovered fuel clogged the coalescing media in the oil/water separator.  Commercially available fuel biocides, such as Biobor and Kathon FP, were applied but failed to control the problem.  Monthly cleaning of the oil/water separator media was necessary to maintain its proper function.

Vapor Treatment

Vapor treatment was accomplished by a catalytic oxidizer.  Hydrocarbon concentrations started at 1,200 parts per million by volume (ppmv) and decreased steadily to 150 ppmv.  Despite use of knockout drums, demisters, and coalescing filters, carryover of seal oil from the liquid ring pumps was initially a problem as it coated the catalyst beds, causing system shutdowns.  Increasing the preheat temperature of the catalytic oxidizer from 550 to 700 degrees Fahrenheit solved the problem, and catalytic oxidizer consistently treated the offgas to below 2 ppmv. 

Product Recovery Effectiveness

The MPE system recovered 28,000 gallons or 99.8 percent of the free product at the site within 28 months of operation.  Guidance documents on free product recovery by the American Petroleum Institute (API) indicate that such recovery rates are unlikely given the moderate hydraulic conductivity and moderate viscosity of weathered JP-5 (API 2004.  Interactive LNAPL Guide.  Version 2.0, August). 

However, experience at NAS Lemoore demonstrates that light nonaqueous phase liquid (LNAPL) recovery via MPE can be successful over a wider range of fuels and at lower permeability sites than is suggested by API and other researchers.

The effectiveness of the Site 17 system can be attributed to the high vacuum applied to the extraction wells, which overcomes the capillary forces that would otherwise prevent free product from moving.  Additionally, the extraction of soil vapor removes contaminants via soil vapor extraction (SVE) and promotes biodegradation of residual product in the vadose zone.

Unlike vacuum enhanced skimming, which causes the water table to mound at each extraction point, MPE creates a cone of depression, allowing free product to flow toward the well.  Furthermore, within the cone of depression, MPE can remove residual product that is trapped below the water table via SVE and bioventing. 

Confirmation Sampling Program

Eleven additional piezometers were installed in areas where extraction or monitoring well coverage was sparse to reduce the potential that free product migrated or existed outside of the treatment system footprint.  Free product was found at only three of the 11 new piezometers, ranging in apparent thickness from 0.01 to 1.4 feet.  However, all of these wells were absent of free product by the end of site remediation. 

Additionally, concern existed that free product may still be present between the extraction wells.  Therefore, six additional borings were completed between four extraction wells.  Soil samples were analyzed and found not only to be without free product, but also to have low concentrations of residual product, confirming the effectiveness of the extraction wells throughout each well’s 20-foot radius of influence. 

Bioventing Phase

By April 2005, it became evident that free product existed only in small pockets with very small apparent product thicknesses (less than 0.25 foot), which did not justify the O&M expense of operating the system in MPE mode.  Furthermore, the rate of free product recovery had decreased to less than 10 gallons per month in the liquid phase. 

The system was designed so that it could operate in bioventing mode.  In this configuration, the drop tubes are removed from each extraction well so that only vapor is drawn from the well.  Additionally, the high-vacuum liquid ring pumps are turned off and only the catalytic oxidizer booster blower is operated, reducing costs for O&M and electricity.  Typical bioventing rates were 2 to 4 scfm per well at 50 inches water column vacuum.  Oil absorbent socks were placed in all wells that had visible sheens.  The socks were changed out monthly and were generally found to contain very little product.

During bioventing, soil gas measurements of carbon dioxide, oxygen, and volatile organic compounds (VOCs) (using a photoionization detector) were collected from sealed wells at the site.  From the start of bioventing, oxygen levels increased by 55 percent and VOC levels decreased by 87 percent.  More importantly, average apparent thicknesses of free product at the site decreased from 0.006 feet to 0 feet.  Although bioventing is not directly effective at removing free product, it is effective in removing thin layers of free product via several mechanisms.  First, bioventing promotes biodegradation of residual free product—this newly remediated soil then has capacity to adsorb and trap free product as residual product, which is more amenable to biodegradation.