|
In Summer 2005, U.S. Army Corps of Engineers (USACE), Seattle District, worked with the TPA-CKY joint venture company to use the Triad approach to communicate and manage uncertainty associated with the removal of lead contaminated soil at the Evergreen Former Infiltration Range, Fort Lewis, Washington. Real time field portable X-ray Fluorescence (FPXRF) data were used to confirm completion of removal and determine the overall variability of the site lead concentrations. USACE and the customer, Fort Lewis Public Works (FLPW), used the FPXRF data to determine if soil in excess of the initial volume estimates required excavation. This information was also provided to Washington State Department of Ecology (WSDOE) in real time to maintain regulator involvement. Using Triad-based collaborative data sets and decision logic, the variability from the FPXRF data was also used to determine completeness of removal and the number of fixed laboratory data sets required to statistically confirm compliance with cleanup objectives.
This project is a follow-on removal action to a Triad-based characterization study that occurred at the former infiltration range in 2003. The characterization study is described in another Triad Profile at this website, entitled Use of Field Portable X-Ray Fluorescence (FPXRF) and the Triad Approach To Investigate the Extent of Lead Contamination at a Small Arms Training Range, Fort Lewis, WA.
| Site Name | Evergreen Former Infiltration Range |
| Location | Fort Lewis, WA |
| Site Type |
Small Arms Firing Range |
| Project Lead Organization | Fort Lewis Public Works (FLPW) |
| Project Lead Type |
U.S. Army Lead |
| Regulatory Lead Program |
State Remedial |
| Triad Project Status | Field Program Completed |
| Reuse Objective Identified |
Yes |
| Proposed Reuse: |
Military barracks |
Evergreen Former Infiltration Range was used for troop training under live fire in the 1950s and 1960s. Fixed-position machine guns firing into an impact berm provided this live fire training. The impact berm was set back approximately 300 feet from the firing discharge area. The berm was a constructed earthen bank approximately 40 feet high and 330 feet long. As part of the Fort Lewis Agreed Order with the WSDOE, FLPW tasked USACE, Seattle District, with determining the nature and extent of contamination on the range and with the removal of contaminated soil.
In 2003, USACE used the Triad approach to expedite site characterization of contaminated soil. This characterization is described in a separate profile.Lead contamination was determined to be present at elevated levels on the front and back side of the impact berm. The maximum detected concentration was 62,500 milligrams per kilogram (mg/kg). Antimony and copper were also detected, but only when lead was above the action level of 250 mg/kg. The majority of the berm soil also contained bullet fragments and failed the Toxicity Characteristic Leaching Procedure (TCLP) criteria, i.e., the material was designated as a Resource Conservation and Recovery Act (RCRA) hazardous waste for leachability of lead above 5 milligrams per liter (mg/L). Depths of contamination ranged from 0-7 feet below ground surface (bgs) depending on location on the berm.
The Evergreen Former Infiltration Range Interim Cleanup Action was expedited ahead of the Agreed Order Cleanup Action Plan to accommodate the "Whole Barracks Renewal" Military Construction Project planned for this area. The former infiltration range needed to be cleaned up by September 2005 to allow construction of a new military barracks to commence. A cleanup action level of 250 mg/kg for lead was selected based on the State Model Toxics Control Act (MTCA) Method A goal for unrestricted site use. Using this cleanup criterion, approximately 5,000 cubic yards (CY) of lead-impacted soils were estimated as requiring removal.
Contract plans and specifications were then developed by USACE using a performance based contracting approach. Performance criteria included removal of lead to achieve cleanup criteria of 250 mg/kg under the MTCA guidance. Performance of bullet removal was specified by requiring treated soil portions to contain <0.1% bullet. Treated soil was required to also meet the Federal RCRA hazardous waste criteria. Recycling of the bullet waste stream was also encouraged.
The contract for the cleanup was awarded to TPA- CKY, an 8(a) contractor with the USACE, in late 2004. Construction began in May 2005 and was completed in July 2005. A power screen was used to generate three waste streams: greater than 1 ½ inch, between 1 ½ and 7/16 inches, and less than 7/16 inch. The 1 1/2 inch plus size gravel waste stream (about 1/3 of the total volume) was clean and was left on site. Bullet fragments in the 1 ½ to 7/16 inch waste stream had enough steel that a magnet could be used to remove the fragments. The fragments were sent to a recycling facility and the remaining material left on-site. Soils passing the 7/16 inch screen were treated with 4% EnviroBlend® which was chosen based on a feasibility study to reduce leachability conducted by the USACE Engineer Research and Development Center (ERDC).The treated soil was then hauled to an active range on the installation and used to construct berms. The berms were shaped per Fort Lewis Range Control specifications and hydro-seeded. Physical location of the berms were surveyed using a geographic information system (GIS) and will be retained in the Fort Lewis Public Works Master Planning documents. Compliance of soil remaining in place was measured real time using the FPXRF and a statistically-based approach.
As discussed above, the primary contaminant at the site was lead. For the purposes of this remediation, the entire site area was considered potentially available for exposure via direct contact with humans. For soil cleanup levels based on human direct contact, the point of compliance was the upper 15 feet of soil throughout the site. A site-specific ecological risk assessment was deferred until after the cleanup.
The primary conclusions from using Triad approach on this cleanup action are as follows:
The overall project objective was to reduce lead soil concentrations from a former impact berm at the former infiltration range to below 250 mg/kg. Compliance with the cleanup goal was measured in accordance with the WSDOE MTCA that requires:
To achieve this objective, soils above this cleanup criterion were excavated. To allow for placement at an active range on Fort Lewis, the remediation also required removal of bullet fragments from the soil, and stabilization of the remaining waste stream to pass lead leachability testing in accordance with the TCLP. Compliance with these secondary objectives were measured by the following requirements:
The systematic project planning process was utilized to define overall quality objectives and the associated sampling and analytical strategy. Early development of the conceptual site model (CSM) during site investigation allowed for rapid decisions on the cleanup actions for the site. Uncertainties were handled with the decision endpoints in mind. The systematic planning approach also allowed for integrated involvement with the regulator and other stakeholders, resulting in shorter review times and quicker completion of the project.
Using a FPXRF made dynamic work strategies possible by gathering, interpreting, and sharing data fast enough to support real-time decisions. Decisions on whether to excavate additional materials were made - often in a matter of minutes - based on the real time data. If there were questions about a certain area of the site, the area in question could be sampled fast enough to support real-time decisions.
The FPXRF also allowed the project team to gather a large data set to statistically determine cleanup and reduce data uncertainty at a lower cost than traditional off-site laboratory methods. A total of 410 soil samples were analyzed for the site using FPXRF, including 359 samples from the excavation area.
The overall uncertainty in data used for compliance monitoring was lowered using this large data set. For example, precision (replicate) samples were collected at locations where data showed concentrations near the cleanup level. These data were used to confirm that adequate sample preparation and homogenization occurred. The data were also used to determine if actions were required in areas where results were near the cleanup level. When results indicated matrix heterogeneity in a given area, the project team was able to increase the sampling density, at minimal cost impact, in those areas where incorrect decisions would be risky from a protectiveness aspect.
Following cleanup, the large number of FPXRF data were used to calculate the mean and standard deviation of the site. These summary statistics were in turn used to calculate the number of collaborative off-site laboratory samples required to verify the FPXRF data and confirm site cleanup. The off-site laboratory used an ICP method for the collaborative data set in accordance with EPA Method 6010.
No formal cost comparison was performed for the project as a whole because comparative costs for a "traditional" cleanup were difficult to estimate. The project team estimated an analytical cost savings of approximately $5,000. In addition, the Triad approach helped to reduce the overall cleanup cost by saving time and money in the following ways:
Systematic planning strategies were used throughout the design, planning, and implementation phases of the soil removal. During the design phase, systematic planning was used to select the remedial alternative that best suited the project. The factors that drove selection the most were required time of completion, future site use, cost of cleanup, the current CSM, and the desire from FLPW to reduce long term monitoring requirements. The site required cleanup by September 2005. Ecological concerns were deferred to a later time since this would take time for evaluation. Since the future site use was for a barracks, a residential cleanup level protective of human health was chosen. The previous characterization indicated that approximately 5,000 CY would require remediation. Costs were controlled by utilizing active ranges as the point of disposal. The final remedial alternative, which was selected based on these factors, was excavation, bullet removal, stabilization of remaining soil, and hauling to active ranges for disposal. Details on alternative selection are presented in the Interim Cleanup Action Plan (see other profile).
Once the remedial action was selected, decision quality needs were evaluated based on the decision end points and on the control of uncertainty around these endpoints. The Remedial Action Management Plan (RAMP) was produced as a team effort among all parties to document the decision quality and uncertainty management strategies selected. A series of systematic planning meetings were also held to discuss project performance monitoring objectives.
Strategies for uncertainty management were required at several stages of the cleanup. Since a DMA had been previously conducted for the FPXRF at this site during the site characterization phase (see other profile)., uncertainties solely concerning the use of the FPXRF were significantly reduced for this effort. The primary uncertainties remaining for the FPXRF technology included the accuracy of the method near the action level due to the heterogeneity of the soil and the accuracy of the FPXRF readings to represent the final site condition. These two issues were addressed with the analytical strategy. Replicates were analyzed for samples near the action level to assess precision and determine if heterogeneity was great enough to impact site decisions. In addition, fixed laboratory samples were submitted once the site was deemed clean to provide a final confirmation of the FPXRF data.
Decision quality was key to designing the sampling strategy used to verify soil removal. The FPXRF was selected in part to help obtain a large number of samples for the site. This large data set helped to reduce the overall uncertainty in the final site condition. After excavation of individual designated excavation areas, a grid system was established over the remediation site. Each grid was approximately 30' by 30' square, and was further divided into nine subgrids. The grid dimension was based on excavation efficiency and reflects a balance between representation of the remediation area and a logical minimum response to discovery of additional contamination. Discrete bag samples were collected randomly from five of the nine subgrids.
Post excavation confirmatory sampling consisted of submitting archived FPXRF samples for analysis of lead at a fixed laboratory using ICP. The number of samples requiring fixed laboratory analysis was determined based on the distribution of the FPXRF data. The FPXRF data was used to calculate the standard deviation and mean for the site. This information was then used with the accepted levels of uncertainty (that is, statistical confidence and significance) to determine the necessary number of samples required to determine the site was clean.
Decision quality was also required for the bullet removal and treatment technology. A bench test was performed to determine the optimum screen design and the selection of and dosage of fixation agents. Considerations included effectiveness, ease of mixing, and cost. To ensure removal of bullets from the fines to be later treated, the contractor obtained sieve tests at a frequency of 5 kilograms (kg) for every 1 ton screened.
Prior to the remediation, a treatability study was conducted by USACE - ERDC. Based on this study, the treatment technology and amount were selected to be approximately 4 percent EnviroBlend® additive to soil. Decision quality for treatment was measured by collecting a 30-point composite sample from each 100 CY stockpile for TCLP lead.
The regulator (WSDOE) and the customer (FLPW) were involved as active members of the team throughout all phases of this project. WSDOE and FLPW approved sampling plans, were updated on and approved results of the DMA, reviewed the Site Investigation Report, the sampling data (for the grids and for site as a whole), the treatability study, and the RAMP. During remediation, they attended several site meetings and were updated regularly on the construction progress. In addition, FLPW approved of recommendations for additional excavations as needed in the field.
As previously discussed, a 30' by 30' square grid system was established over the remediation site after the initial excavation was complete. FPXRF field testing was performed on soil samples collected from each grid after excavation of the delineated excavation areas. The FPXRF data was used to determine additional areas of sampling and excavation.
The collected data were summarized in spreadsheets and maps. These were provided on a project website as created along with construction quality control reports. USACE used this information to determine if the project was meeting contract specifications. Modifications to the sampling plan were addressed in conference calls with the project team and USACE as part of a dynamic work strategy. Since an FPXRF was employed for most of the duration of operations, the preliminary analytical results for soil were usually available within 24 hours of sample collection on the project website. Although the data were posted on the website, this website was not utilized by the regulator and customer. Instead, they preferred an email presenting information on a periodic basis that also called out the major results and decisions that should be focused on. Also, toward the end of the project, decisions were made via phone calls based on data that had just been collected.
Achievement of cleanup levels was measured in accordance the project objectives listed previously in this profile. Since one of the cleanup criteria was that no sample should exceed 500 mg/kg, these areas were immediately excavated. Depending on the distribution of data, areas adjacent to the hot spots were handled in one of the following ways:
In some cases, hot spot areas required more than one excavation layer to remove all soil that contained lead above 500 mg/kg. For grids that were over-excavated and resampled, the new sample data were used to replace excavated sample results. All data from a given grid were used to recalculate sample means and 95% UCL values for information purposes.
Once the FPXRF sample set met the three cleanup criteria established in the project objectives, the operations proceeded to the post excavation confirmatory sampling. Post excavation confirmatory sampling consisted of submitting archived FPXRF cup samples for analysis of lead at a fixed laboratory using ICP. Based on the standard deviation and mean for the FPXRF data at the site, the number of samples required to determine the site was clean was calculated to be 32.
The majority of the activities performed during the remediation were field based. Confirmation of bullet fragment removal was conducted real time. All initial analysis of soil samples was conducted in the field on bulk soil samples using the FPXRF. Additional collaborative data included FPXRF cup subsampling, analysis, and shipment of cup samples to a local laboratory for Total Lead analysis by ICP. Treated soil samples were submitted to the fixed laboratory for analysis of TCLP.
The data was evaluated as it was generated by both field and office personnel using the decision logic presented above.
TQRS pending
Quality control (QC) was measured via the data quality indicators of accuracy, precision, representativeness, completeness, and comparability. The data quality was summarized in a Data Quality Review that was submitted as part of the Draft Site Investigation Report. The frequency of collection and analysis of field and laboratory QC samples met the Quality Assurance Project Plan (QAPP).
FPXRF accuracy was established using low, medium, and high concentration calibration check standards. Calibration verification checks were conducted at the beginning and end of each day and after every 20 samples. The percent difference (%D) was less than 20 percent. Equipment blanks were also taken to verify absence of meter signal when lead is known not to be present in the sample. Accuracy of ICP results was established via standard laboratory QC procedures. Matrix spikes (MS) and blank spike recoveries were evaluated to be within 85 to 115 percent. Two MS recoveries were above the acceptance criteria.
FPXRF precision was measured through precision samples (a sample that was analyzed seven times in replicate). The relative standard deviation (RSD) of the replicates was required to be below 20 percent. Four of the 19 samples had RSDs above this value. Precision was also evaluated against whether data variability impacted site decisions when comparing initial values to average values from the precision results. Six precision samples had average values that resulted in different decisions being made (above 250 mg/kg indicated removal, below 250 mg/kg did not) relative to the initial sample measurement. However, of those six areas represented by the precision samples, two areas were excavated due to proximity with elevated areas of lead contamination. The remaining four areas were not removed. However, the average results were used to determine site cleanup so this did not impact the overall site decision.
A total of 27 FPXRF field duplicate samples were collected. The quality assurance (QA) criterion was that the relative percent difference (RPD) between field duplicates be less than 50 percent. All but 4 samples met this criterion. However, both results from the duplicate pairs in these four samples resulted in the same removal decision. Three off-site laboratory duplicates for lead were also collected, and the RPDs for these duplicates were above the 35% requirement. In two cases, one of the duplicate pairs had non detected results.
Representativeness was ensured by selecting sample locations properly, collecting a sufficient number of samples to accurately reflect conditions at the site, and collecting a sufficient volume of sample material to complete the analyses and minimize bias or errors associated with sample particle size and heterogeneity.
Comparability was measured by taking split FPXRF cup samples and submitting them for both ICP and FPXRF analysis. A correlation analysis was performed on these samples between FPXRF and laboratory lead results to evaluate data comparability. A linear regression correlation coefficient (r) of 0.92 was obtained. The mean ratio of ICP result to dried soil FPXRF result was 1.07.
A project website contained daily and weekly reporting of data and analytical results. Contents of the website included daily construction quality reports, map of sample locations, excavation logs, survey data, and spreadsheets of FPXRF, ICP, TCLP, and sieve test results. USACE used this information to determine if the project was meeting contract specifications. The website data management process ensured that the progress was accurately tracked and that construction operations were in compliance with regulatory standards.
To determine compliance with the 95% UCL cleanup criteria, the 95% UCL was calculated using EPA's ProUCL software.
206-764-3322
kymberly.c.takasaki@usace.army.mil 206-766-6442 emile.pitre@usace.army.milTo update this profile, contact Cheryl T. Johnson at Johnson.Cheryl@epa.gov or (703) 603-9045.
