AWI Final Report Treatability Studies DNAPL Contaminated Soils.pdf

"We are committed to serving our clients by exceeding their expectations."

Geotechnical Construction Monitoring Dam Engineering Geoscience Environmental

510 East Gay Street West Chester, PA 19380

Phone: (610) 696-6066 Fax: (610) 696-7771 www.schnabel-eng.com

March 8, 2010

Mr. Peter Pellissier, P.E.

Senior Project Manager EA Engineering, Science and Technology, Inc.

11019 McCormick Road Hunt Valley, MD 21031

Subject: Final Report, Treatability Studies for West Side DNAPL- Contaminated Soils

Atlantic Wood Industries (AWI) Superfund Site Portsmouth, Virginia (Schnabel Reference 09150034)

Dear Mr. Pellissier:

Schnabel Engineering, LLC (Schnabel) was retained by EA Engineering, Science and Technology, Inc. (EA) to perform a stabilization/solidification (S/S) treatability study on the West Side dense non-aqueous phase liquid (DNAPL) contaminated soils at the AWI Superfund Site per the Request for Proposal 09 502 dated April 8, 2009, and contracted on May 14, 2009. This submittal constitutes the Technical Memorandum required by Task 7.4D of the Scope of Work (SOW).

EXECUTIVE SUMMARY

Schnabel completed an S/S study, with the main aspects of the project being the development of mix designs and an evaluation of S/S for full-scale implementation. Schnabel collected three discrete soil samples to establish an environmental baseline based on the sands (two samples) underlying the Historic Disposal Area (HDA), as well as the upper fill materials (one sample) which serves as the DNAPL source zone. From the soil cuttings collected from two adjacent borings to depths of approximately 26 feet (ft) below the ground surface (bgs), Schnabel created a series of laboratory composites based on the upper fill materials blended with the deeper silty, sandy clays. To conservatively simulate the potential contaminant downdrag effects during in-situ S/S, the fill and underlying soils were mixed using a ratio of 1:2 to reflect their relative thicknesses versus the total anticipated treatment depth.

Six mix designs (M-1 to M-6) were prepared using combinations of cement, bentonite and granular activated carbon (GAC) in the total dose range of 8 to 13 percent. The mix designs were tested for their unconfined compressive strength (UCS) at 7 and 28 days, and select mix designs were analyzed for hydraulic conductivity at 28 days. The UCS values for all mix designs were well above those required for site geotechnical considerations (>50 psi) and the hydraulic conductivities of M-2 and M-6 (with low and high GAC doses, respectively) ranged from 9.6x10-9 to 7.5x10-8 cm/s, well below the typical criteria for impermeable media such as landfill liners.

The synthetic precipitation leaching procedure (SPLP) results from the field samples, laboratory composite samples, and the 28-day cured mix designs all indicated that pentachlorophenol (PCP) occurred at concentrations high enough (1 to 14 milligrams per liter [mg/L]) to be identified as the main component of a

Project 09150034 / March 8, 2010 Page 2 Schnabel Engineering, LLC potential DNAPL. The complex nature of the DNAPL in combination with the reagents (cement, etc.) created numerous analytical interferences which meant that the totals values for PCP were not necessarily accurate and also raised the reporting limits of the SPLP testing for the mix design and other samples to above the numerical groundwater criteria identified in the Record of Decision (ROD) Table 7 for the site (USEPA 2007). Despite these issues, the SPLP results for M-2 through M-6 (all of which contained GAC) nevertheless achieved the numerical ROD criteria except for PCP and benzo(a)pyrene, the latter of which was non-detect but at a reporting limit five times the ROD criteria. Thereafter, subsequent efforts of this treatability study mainly focused on reducing the migration of PCP, and understanding the post-remediation fate and transport of the DNAPL (if any).

The current round of mix designs has shown that it is possible to meet the ROD-mandated performance criteria except for PCP. With respect to PCP, the mix designs were shown to reduce the dissolved phase mobility of PCP by encapsulating it in a geotechnically-strong (see Section 5.1), low permeability (Section 5.2), and diffusion-controlled (Section 7) soil-cement mass. In the GAC-containing mix designs, the PCP effective diffusion coefficient was lowered by 400 times compared to the PCP diffusion in bulk solution (Section 7). However, the PCP leaching results were still in excess of the ROD performance criteria (0.001 mg/L).

When looking beyond the ROD to other semi-volatile organic compounds (SVOCs), it was not possible to rule out the potential existence of a DNAPL persisting within the stabilized mass following treatment with the current round of mix designs. To resolve this issue, the potential addition of organophilic clay (OC) to the current mix designs up to a 4% dose level was empirically shown to eliminate the existence of any remnant DNAPL. Also, studies have indicated that OC doses on the order of 2% to 7% have significantly reduced PCP concentrations from its solubility limit to 0.2 to 0.04 mg/L depending on the soil matrix, initial concentrations, and total reagent doses (Appendix F). While OC showed significant reductions in PCP leaching, none of the reviewed literature identified a specific OC dose capable of achieving PCP <0.001 mg/L. Nevertheless, it appears that OC at small doses can serve the dual function of eliminating the DNAPL phase in the stabilized mass based on mass-volume relationships and, through sorption, significantly lower the dissolved concentrations of numerous SVOCs, including PCP, which leads to even lower contaminant migration (through greater retardation coefficients).

Therefore, our recommendation is to conduct a second round of treatability tests, specifically targeting additional SVOC and PCP immobilization. The literature review indicated that two main types of reagent are effective at immobilizing organic contaminants of concern (COCs): granular activated carbon (GAC) and OC.

The current cost of both reagents is on the order of $1.25/lb (1% dose~ $42/CY) making them expensive to use. GAC and OC also serve different purposes: GAC is used to immobilize strongly hydrophobic organic COCs, whereas modified clays such as OC appear to be more effective for the moderately soluble organic COCs such as PCP, naphthalene, etc., whose solubility limits are on the order of 0.01 to 10,000 mg/L.

However, OC sorption processes are essentially irreversible, and it is anticipated to further reduce the mobility of the SVOCs at the AWI site. Thus, in this context, it appears possible to reduce or eliminate the current GAC dose, and emphasize OC addition. However, since it is not clear which will perform better under the semi-dynamic leaching test conditions, there is value in verifying through further treatability testing their performance individually and in combination at low doses (the primary range of interest).

Based on the favorable performance of Mix 2 (M-2) from the first round of the treatability tests, we will use it as a base mix (less its 0.1% GAC) to evaluate the relative impacts of substantially higher doses of organic carbon. Based on the results of the DNAPL stability analysis, preference is given to OC, but increasing GAC will also be assessed, both as a stand-alone reagent and in combination with OC. However, a sufficient (total and relative) dose of cement must be maintained to still achieve a UCS>50 psi and a hydraulic conductivity value less than 10-7 cm/s. Accordingly, one proposed mix design will have an increased Portland-Slag dose to compensate for the elevated OC content and to maintain the structural performance of the mix design. Given

Project 09150034 / March 8, 2010 Page 3 Schnabel Engineering, LLC these constraints (and the discussion provided in Appendix G.3), we propose the following six mix designs for the second round of treatability testing:

MIX

ID

Total Reagent

Dose

Portland-Slag Dose

Bentonite Powder

Dose

GAC

Dose

OC

Dose Basis

7 9.5 8 0.5 0 1 Low OC dose

8 9.5 8 0.5 1 0 Mid GAC Dose

9 10.5 8 0.5 0 2 Mid OC dose

10 10.5 8 0.5 1 1 Mid GAC/Low OC dose

11 12.5 8 0.5 0 4 High OC dose

12 16.5 12 0.5 0 4 Increased cement, High OC dose

All data in wt%

The proposed mix designs 7 to 12 will be subjected to a similar suite of tests at 28 days (UCS, hydraulic conductivity, leaching) as the prior round with the focus being SVOCs, not just PCP. The addition of OC is viewed to provide a more enhanced performance than that demonstrated by this study, and its dosage should be consistent with site goals and balanced against the total cost of full-scale implementation. Environmental testing will emphasize totals and semi-dynamic leach testing. SPLP testing method will be conducted to handle benchmarking with other treatability studies and field projects.

CLOSURE

We appreciate the opportunity to have worked with you on this project and look forward to the possibility of working on future phases of work at this challenging site. If you have any questions, please contact me at

(610) 696-6066.

Sincerely, SCHNABEL ENGINEERING, LLC

Dennis G. Grubb, Ph.D., P.E.

Senior Associate

DGG:RAD:hcf

Final Report

Treatability Studies for West Side DNAPL-Contaminated Soils

Atlantic Wood Industries (AWI) Superfund Site Portsmouth, Virginia

Project 09150034 March 8, 2010

Project 09150034 / March 8, 2010 i Schnabel Engineering, LLC

TABLE OF CONTENTS

1.0 INTRODUCTION

1.1 Remedial Action Objectives (RAOs)

1.2 Groundwater Cleanup Criteria

1.3 Performance Standards for S/S

2.0 TREATABILITY STUDY SCOPE

2.1 Background and Context

2.2 Treatability Study Approach

3.0 FIELD SAMPLE COLLECTION

4.0 PREPARATION OF LABORATORY COMPOSITE SOIL

4.1 Soil-Cement Mix Design

5.0 MIX DESIGN (SOIL-CEMENT) RESULTS

5.1 Unconfined Compressive Strength Testing

5.2 Hydraulic Conductivity Testing

6.0 ENVIRONMENTAL RESULTS

6.1 General Remarks

6.2 Field Controls and Laboratory Composites (S-1 to S-3; L-1 to L-3, Tables 2 and 3)

6.3 Mix Designs (M-1 to M-6) SPLP Data (Tables 4 to 7)

6.4 PCP Analytical Data Issues

7.0 SEMI-DYNAMIC LEACHING TEST RESULTS (ASTM C1308)

8.0 DNAPL STABILITY EVALUATION

9.0 ENVIRONMENTAL DATA SUMMARY

10.0 OTHER TREATABILITY STUDY COMPONENTS

10.1 Geotechnical Stability Analyses

10.2 Economic and Implementation Issues

10.3 Changes to Scope

11.0 CONCLUSIONS & RECOMMENDATIONS

12.0 LIMITATIONS

Project 09150034 / March 8, 2010 ii Schnabel Engineering, LLC

List of Acronyms Tables Photos Figures

Appendices:

Appendix A: Geotechnical Site Investigation Report Appendix B: Geotechnical Testing Results – Mix Designs Appendix C: Semi Dynamic Leaching Test (ASTM C1308) Results Appendix D: Geotechnical Site Stability Evaluation Appendix E: AWI ISR Process White Paper Appendix F: Limited Literature Review Appendix G: References and DNAPL Stability Evaluation Appendix H: Environmental Analytical Testing Results

Project 09150034 / March 8, 2010 iii Schnabel Engineering, LLC

List of Acronyms

ANS American Nuclear Society ASTM American Society for Testing Materials AWI Atlantic Wood Industries BGS Below Ground Surface COC Constituent of Concern CY Cubic Yards De Effective Diffusivity (cm2/s) DI De-ionized DNAPL Dense Non-Aqueous Phase Liquid EA EA Engineering, Science and Technology ERA Environmental Risk Assessment foc Fraction (soil) organic matter content (%) GAC Granular Activated Carbon HDA Historic Disposal Area HHRA Human Health Risk Assessment k Hydraulic Conductivity (cm/s) LX Leachability Index MCL Maximum Contaminant Level MGP Manufactured Gas Plant OC Organophilic Clay PAH Polycyclic Aromatic Hydrocarbons PCP Pentachlorophenol PPL Priority Pollutant List QA/QC Quality Assurance and Quality Control RAO Remedial Action Objective ROD Record of Decision SPT Standard Penetration Test SPLP Synthetic Precipitation Leaching Procedure S/S Stabilization/Solidification SVOC Semivolatile Organic Compounds TCLP Toxicity Characteristic Leaching Procedure UCS Unconfined Compressive Strength USEPA US Environmental Protection Agency VOC Volatile Organic Compounds

Project 09150034 / March 8, 2010 Page 1 Schnabel Engineering, LLC

1.0 INTRODUCTION

The Atlantic Wood Industries (AWI) site is located along the west shore of the Southern Branch of the Elizabeth River, east of the U.S. Navy’s Paradise Creek Landfill. The site is south of the Norfolk Naval Shipyard and north of the Navy’s Southgate Annex and the Portsmouth Public School District offices. Elm Avenue (Virginia Route 337) is located along the northern perimeter of the site. The site is contaminated from wood treatment operations, storage of treated wood, and disposal of wastes. The soils and groundwater on this site are contaminated with a number of compounds including creosote, polycyclic aromatic hydrocarbons (PAHs), pentachlorophenol (PCP), dioxins/furans, and heavy metals. The Historic Disposal Area (HDA) has dense non-aqueous phase liquid (DNAPL) contamination from the ground surface to approximately 27 feet (ft) below ground surface (bgs). The contaminated soil volume was originally estimated to be approximately 7,200 cubic yards (CY) (United States Environmental Protection Agency [USEPA], 2007), but based on additional characterization conducted by USEPA in 2008/2009, the current volume estimate approaches 45,000 CY.

The Record of Decision (ROD) for the AWI Site (USEPA, 2007) indicated that excavation of DNAPL-impacted media is the preferred cleanup option. The ROD also stated that in-situ stabilization/solidification (S/S) could be used to treat the DNAPL at the HDA if it was more viable than excavation. USEPA decided to use the in-situ S/S method based on an evaluation performed by EA Engineering, Science, and Technology, Inc. (EA, 2009a).

EA retained Schnabel Engineering, LLC (Schnabel) to perform an S/S treatability study on the HDA DNAPL-contaminated soils, focusing on the compounds explicitly called out in the site Remedial Action Objectives (RAOs) for groundwater.

1.1 Remedial Action Objectives (RAOs)

The objectives of the remedial action are listed in Section 8 of the ROD. The excavation or in-situ S/S portion of the remedial action is necessary to meet or help meet the following RAOs:

a. Reduce human health risks from exposure, including ingestion, inhalation, and dermal contact, to site-related contaminants in the groundwater to acceptable levels.

b. Minimize the migration of contaminants from the unsaturated soils to the groundwater.

c. Reduce the ability of the creosote DNAPL present in the groundwater to migrate deeper into the aquifer system.

d. Minimize the migration of site-related groundwater contamination to Paradise Creek through the existing storm drain, or its gravel bed, in order to limit any potential environmental impacts.

The use of S/S prevents DNAPL from further migration by homogenizing the soil and dispersing the DNAPL throughout the resulting soil-cement mixture. Once redistributed across the soil-cement volume, there is often insufficient total mass of the DNAPL for it to persist as a separate phase. Even if DNAPL remains as a separate phase (e.g., very high contamination scenarios), the creation of a cemented soil mass with a very low hydraulic conductivity and its distribution of fine pores will lead to prohibitively high pore entry pressures for DNAPL flow, thus significantly limiting or preventing migration of a separate phase within or into the soil-cement mass. The S/S also greatly minimizes the potential for migration of dissolved groundwater contamination, contributing to the groundwater’s ability to meet the ROD-specified performance standards beyond the edge of the soil cover (i.e., the waste management area).

Project 09150034 / March 8, 2010 Page 2 Schnabel Engineering, LLC

1.2 Groundwater Cleanup Criteria

USEPA selected cleanup criteria for groundwater that are based on federal maximum contaminant levels (MCLs) or maximum contaminant level goals (MCLGs). Not all of the groundwater contaminants of concern have regulatory criteria, and the selected remedy does not set a cleanup criterion for all contaminants.

However, by meeting the regulatory criteria that do apply, it is assumed that the contaminants without cleanup criteria generally will be reduced as well. Groundwater cleanup criteria are presented in the following table:

Groundwater Cleanup Criteria for AWI Site (from Table 7 of the ROD)

Contaminant Criteria (mg/L) Requirement

Volatile Organic Compounds (VOCs) Benzene 0.005 40 CFR 141.61 Toluene 1.0 40 CFR 141.50

Ethylbenzene 0.7 40 CFR 141.50 Xylenes (total) 10 40 CFR 141.50

Semivolatile Organic Compounds (SVOCs) Pentachlorophenol 0.001 40 CFR 141.61

2,4,5-Trichlorophenol 0.05 40 CFR 141.61 Benzo(a)pyrene 0.0002 40 CFR 141.61

Metals Antimony 0.006 40 CFR 141.51 Cadmium 0.005 40 CFR 141.51 Chromium 0.1 40 CFR 141.51

Copper 1.3 40 CFR 141.51 Mercury 0.002 40 CFR 141.51 Arsenic 0.010 40 CFR 141.62

Lead 0.015 40 CFR 141.80(c) Dioxins/Furans

2,3,7,8-TCDD

(Dioxin) 0.00000003 40 CFR 141.61

1.3 Performance Standards for S/S

The S/S must be conducted in such a way as to prevent future downward migration of DNAPL and to limit the ability of contaminants to migrate in groundwater. With this accomplished, monitored natural attenuation (MNA) has the potential to further reduce the groundwater contamination beyond the waste management area to the above criteria. Immobilizing the contaminants and reducing permeability of the soil aid in achieving this objective. Note that since there is only a minimal area of groundwater contamination that must be addressed by MNA, a small reduction of migration potential can have significant beneficial effects.

Project 09150034 / March 8, 2010 Page 3 Schnabel Engineering, LLC

2.0 TREATABILITY STUDY SCOPE

2.1 Background and Context

An S/S treatability study of HDA soils was the focus of this Scope of Work (SOW). The approach to developing the mix designs was based on a combination of environmental and geotechnical testing to ensure suitable overall performance, and the ability to use geotechnical attributes for specifications development and quality assurance and quality control (QA/QC) purposes during remedial action. The environmental and geotechnical performances are linked; in essence, the stronger and less permeable a soil-cement specimen is, the less it will leach COCs.

The geotechnical attributes that are relevant for S/S media are the unconfined compressive strength (UCS) and hydraulic conductivity (k), and the attained values are influenced by the ability to inject the reagents/grout mixture as a slurry or grout (mix design) into the soil, then liquefy and homogenize the soil. The UCS of the soil-cement samples can be evaluated at any curing time, with the typical industry conventions usually being 3, 7 and/or 28 days. For this study, 7 and 28 days were adopted. The hydraulic conductivity is typically evaluated on the basis of 28-day cured samples. In terms of environmental performance, the totals concentrations are usually documented, and performance is evaluated on the corresponding leach characteristics of the 28-day cured specimens. For applications in the northeast US where exposure to acid rain is a concern, a synthetic precipitation leaching procedure (SPLP) solution is often used instead of the toxicity characteristic leaching procedure (TCLP) to assess the post-treatment leachate quality. In some cases, de-ionized (DI) water may also be used for leaching of soil-cement specimens if it more closely simulates in-situ conditions. The foregoing constitutes the typical suite for soil-cement mix design for environmental S/S applications, but other testing may be added. For the AWI study, semi-dynamic leaching tests (ASTM C1308) were added to the suite of tests mid-project because there was concern that given the wide range of COCs and their properties versus the composition of individual mixes, some mixes may immobilize some, but perhaps not all, COCs.

ASTM C1308 involves the use of monolithic cylinder samples submerged in a DI water bath. The test aims to measure the radial diffusion of COCs from a cylinder suspended into the surrounding water bath; the water bath is switched out at specific time intervals to measure the COC mass leached into solution over time. These conditions are more representative of the stabilized soil columns potentially created in the AWI HDA than the TCLP and SPLP tests, where testing methods specify the grinding of soils and solid media samples to pass a

9.5 mm sieve prior to extraction. The difference in testing conditions is not so much an issue for metals immobilization, which is often pH controlled, as it is for organics whose mobility may be controlled by sorption processes. For instance, by grinding samples, the SPLP test protocol directly exposes granular activated carbon (GAC) encapsulated within the cement stabilized media directly to the extracting solution, potentially releasing the sorbed compounds.

Because the SPLP-PCP results from the mix designs were much more elevated compared to the groundwater cleanup criteria, the decision to perform the more site-specific and realistic semi-dynamic leaching tests (ASTM C1308) turned out to be fortuitous, since this enabled the contaminant retardation coefficient of PCP in the actual soil-cement to be estimated.

2.2 Treatability Study Approach

Figure 1 shows a schematic of the treatability study approach and the data collection process. Since it is advantageous for treatability studies to be completed on contaminant source zone materials to establish the upper threshold for reagent dosing, EA and Schnabel personnel decided to identify and collect HDA site soils for the treatability study from an area that exhibited the greatest presence of visible DNAPL and was therefore presumed to be the most contaminated.

Project 09150034 / March 8, 2010 Page 4 Schnabel Engineering, LLC

The first activity was to collect contaminated soils as discrete samples (denoted in this study as S-series samples) and as soil boring cuttings to ~26 ft bgs. Recovered soils and the boring cuttings from the upper wood fill material (S-3; 0 to ~8 ft bgs) were visibly and heavily contaminated with DNAPL. The underlying sandy soils (S-1, S-2) and the boring cuttings to 26 ft bgs were much less visually contaminated, with only occasional stringers of DNAPL. To conservatively simulate the potential downdrag of contaminants from the upper wood fill into the underlying sandy soils during the S/S homogenization process, the wood fill and sandy soils (approximately 8 ft and 16 ft thick, respectively) were blended at a ratio of 1:2 to create laboratory composites, or L-series samples. In this way, the S-series samples served as background samples for the existing DNAPL-impacted site conditions, whereas the L-series samples served as the composite soils to be treated with various reagent combinations during the mix design evaluation process. The following environmental testing suite (totals and SPLP) was completed on each of the discrete field samples and the laboratory composites (S- and L-series, respectively):

BTEX (SW846-8260B)

SVOCs, Priority Pollutant List (PPL) (SW846-8270B) PPL Metals (USEPA Method 6000/7000 series) Dioxins/Furans, High Resolution (SW846-8290B)

The selection of reagents (R) and their doses shown in Figure 1 for the mix design process for the HDA soils at the AWI site was influenced by a number of considerations:

Knowledge of the site geology, groundwater conditions, and COCs to be treated Experience with other treatability studies at similar sites Review of other treatability studies Industry experience with in-situ S/S actually implemented at similar sites Reasonableness of cost

The review of site data plus industry experience suggested that mix designs that were successful at former wood treating and/or manufactured gas plant (MGP) sites would also be effective based on the known COCs concentrations at the HDA soils. As such, a total reagent dose on the order of 10 to 15% by dry weight was anticipated. The following key reagents were identified:

Portland-slag cement blend (to keep heat of hydration low) Bentonite (to attain a pumpable grout for in-situ mixing and achieve a low hydraulic conductivity) Granular activated carbon (GAC) (to restrict mobility of a broad class of organic compounds)

Proprietary reagents were avoided due to their anticipated high unit costs and potential limitations on competition and corresponding cost impacts during the bidding process. Mix designs were then prepared by blending known amounts of reagents prepared as a slurry with the laboratory composite (L) soil to make the six different mix designs (M-Series samples), as summarized in Figure 1 and Table 1. Figure 1 indicates that numerous soil-cement cylinders (Replicates A to H) from each mix design were fabricated to provide sufficient specimens to conduct the following:

1) 7-day and 28-day cured UCS strength tests in triplicate (replicates A to C; D to F; respectively)

2) Hydraulic conductivity testing on a 28-cured specimen (replicate G)

3) Totals and SPLP analyses for the suite of environmental analytes listed above for the S and L sample series.

4) Semi-dynamic leaching tests on a 50-day cured specimen (replicate H); see Section 7.

Project 09150034 / March 8, 2010 Page 5 Schnabel Engineering, LLC

The suffixes to the mix design numbers (appearing in summary tables) denote the replicate of a mix design, but have no significance beyond the replicate reference. To be conservative, the replicates having the two lowest UCS strengths at 28 days for each mix design were used for the environmental analytical testing suite (shown above). Hydraulic conductivity (ASTM D5084) were only performed on Mixes 1, 2 and 6 as they represented the minimum and maximum reagent doses (M-2; M-6) and the control without GAC (M-1).

Added to the treatability study mid-project were the semi-dynamic leaching tests (ASTM C1308) performed on Mixes 1, 2 and 6. Procedurally, additional 28-day cured specimens were held over for the ASTM C1308 testing and the semi-dynamic leaching tests were begun as soon as the SPLP results became available (i.e., 50 days). The SPLP tests showed PCP concentrations in excess of the ROD performance standard leaching from the mix designs.

The following sections describe in more detail the individual elements and the chronology of the treatability study SOW.

3.0 FIELD SAMPLE COLLECTION

The objective of this task was the collection of soil samples for the S/S treatability study. The sampling event was designed to collect some of the most contaminated soils and soil-like media from the HDA, and to use this media for the treatability study. Accordingly, two test borings were located by EA personnel on May 20, 2009 in the vicinity of Borings 3.8-06, -08 and -14 (EA 2009b), which were logged by Schnabel personnel as Borings 3.8-6A and 3.8-6B. Appendix A contains the complete geotechnical engineering data report of the collection of site soils to be used in the treatability study.

The two boreholes (Borings 3.8-6A and 3.8-6B) were drilled to collect site materials from two general strata:

(1) the upper existing fill material containing wood wastes from the ground surface to depths of approximately 8 ft bgs; and (2) the underlying site soils consisting of silty to clayey sands to approximately 25 ft bgs. From these strata, three discrete site samples or field controls (S-1 to S-3) were collected to determine the range of concentrations of the regulated COCs to which the mix design’s performance would ultimately be compared. Field controls S-1 and S-2 were silty to clayey sands collected from Boring 3.8-6A at approximately 11 and 17 ft bgs, respectively. Field control S-3 was from Boring 3.8-6B collected between 1 to 5 ft bgs; this was representative of the upper fill material that was visually saturated with DNAPL.

Photos 1 to 3 show images of the DNAPL that was encountered during split spoon sampling of site soils in Boring 3.8-6A. Photo 1 shows the DNAPL-impacted (black colored) site soils near the interface between the pond bottoms and the upper reaches of the sandy site soils. The relative ease at which the DNAPL spread (bled) in the open soil sample suggested it was fairly soluble and/or possessed a low interfacial tension.

Photos 2 and 3 show DNAPL lenses occurring deeper into the silty/clayey sands. Photo 3 is a close up of Photo 2 which, due to the vibrations associated with the drilling rig upon which the split spoon sat, was liquefying from the positive pore water pressures generated from the vibrations, liberating DNAPL within the sample to the surface (near the 7.5 inch mark). It should be noted that the RAO requires reducing the migration of the creosote DNAPL such as that shown in the photos.

The soil cuttings from both boreholes were used as the source material for the laboratory composite series (L-series) and for the mix designs (M-series). From Borehole 3.8-6A, four buckets of soil media were collected:

one bucket of upper fill material; and three buckets of sandy soils were collected from approximately 10 to 26 ft bgs. From Borehole 3.8-6B, the upper fill material was collected between 0 and 6 ft bgs, with an additional two buckets of sandy soils collected below the pond bottoms down to 20 ft bgs. These buckets were sealed and shipped to Schnabel’s Blacksburg, Virginia laboratory.

Project 09150034 / March 8, 2010 Page 6 Schnabel Engineering, LLC

4.0 PREPARATION OF LABORATORY COMPOSITE SOIL

For the in-situ S/S technology, mixing occurs both during the downward drilling and upward reaming process (thus homogenizing the entire vertical column of soil). As such, the soil that is ultimately stabilized using reagents can be significantly different from discrete field samples, both in terms of soil type and extent of contamination. Therefore, a representative laboratory composite soil (L-series) was developed (without reagents) that attempted to replicate the soil blending process through a vertically-averaged approach based on the anticipated maximum treatment depth (26 ft bgs).

In the laboratory, soil cuttings from the upper fill and underlying sands were individually blended until they were visually homogeneous. Then, the upper fill material was added to the sandy soil at a ratio of 1:2 and homogenized to reflect the average blending rate based on the site stratigraphy to 26 ft bgs (e.g., approximately 1-9 ft bgs of upper fill material and approximately 9-26 ft bgs of underlying sands). This laboratory composite soil was then sealed in 5-gallon buckets and stored until blending the actual mixes.

Totals analyses for the field samples (S-1, S-2, and S-3) and the laboratory composite samples (L-1, L-2 and L-3) were analyzed for BTEX, PPL SVOCs, PPL metals and Dioxins/Furans. Results of these analyses are presented for the 15 groundwater COCs listed in the ROD are presented in Table 2. For comparability to the upcoming mix design evaluations, the field and laboratory control samples were also analyzed by the SPLP method; these results are presented in Table 3.

Due to the blending of woody upper fill material media with the silty sands, the specific gravity of the laboratory composite soil had to be measured. While the specific gravity of typical soil solids is on the order of 2.65 to 2.8, the measured specific gravity of the composite soil was 2.57 (Appendix B). Additionally, since the mixes were completed over a series of days, the collection of the L-series samples (L-1 to L-3) from the storage buckets was staggered, occurring immediately prior to the creation of the S/S mixes. This was done to monitor potential changes to the VOC concentrations (e.g., volatilization) in the L-series samples during storage. The similarity in the VOC concentrations shown in Table 2 for the L-series suggests that VOC losses during the experimental program were negligible.

4.1 Soil-Cement Mix Design

For S/S to be successful, the mixing process must homogenize the soil with the reagents, which makes the wet mix design process and deployment the preferred option at the AWI site. Reagents often include pure and/or blended cements, which may include other pozzolans (cement kiln dust, fly ash, etc.) and specialty additives (bentonite, organophilic clay, GAC, etc.). The total dose and choice of reagent are a function of the site soils, COCs and their concentrations, pH behavior, flowability, and site remediation goals. Three factors are important for the success of the wet mix process: (1) proper reagent dosing rates, which requires knowledge of the in-situ density and moisture content of the soils; (2) a pumpable grout (which depends on water:cement ratio); and (3) liquefiable soil (the soil-reagent mix in-situ must be above the liquid limit of the soil).

Typically, a pure or blended cement is the primary reagent, and experience has shown that the water:cement (w:c) ratio can be up to 2 to achieve total reagent doses to soil between 5% and 20%.

For the AWI treatability study, six mixes providing a range of reagent doses were prepared based on the need for a pumpable grout and soil liquefaction. The cement used for the mix design work was provided by Lafarge, which has a terminal at Money Point on the Elizabeth River (approx. one mile south of the AWI site). To emphasize strength and to mitigate against heating, the cement used was a blend of 60% Type I/II Portland Cement (Ravena, NY) with 40% NewCem®, a form of slag cement (Sparrows Point, MD). To reduce hydraulic conductivity, promote metals sorption, facilitate cementation, and to prepare a slurry for the wet process, a commercially available bentonite powder was selected for use in the mix designs (Wyo-Ben Hydrogel; Billings, MT). Since many of the COCs are organic compounds, GAC (Calgon, Type WPH;

Project 09150034 / March 8, 2010 Page 7 Schnabel Engineering, LLC

Pittsburgh, PA) was used to increase the sorption and encapsulation of organic compounds in the stabilized mix. All reagents are locally available in the greater Portsmouth, Virginia, area.

The six mix designs are shown in Table 1, and the total reagent dose (dry weight basis) range reflects prior project experiences focused on economically immobilizing metals and organics in contaminated soils. The dry mix doses were limited to the 8 to 13% range, because cement doses in the 15 to 20% range, while sometimes necessary, often become cost prohibitive or not cost-effective compared to other remedial options.

High dry doses are difficult to effectively deliver in-situ because grout flowability decreases with decreasing water content. Since the bentonite slurry is the carrier fluid for the cement and GAC, bentonite slurries on the order of 4% to 6% were required to ensure sufficient pumpability and flowability of the corresponding dry mix doses. To determine the influence of GAC on hydrocarbon immobilization, doses between 0% and 0.1% were used; the upper bound was selected because GAC has a relatively high unit cost and 0.1 wt% GAC represents a significant volume of material due to its low density. Other key ratios relevant to mix design are shown in Table 1.

The mix designs were prepared by adding the reagents first to the bentonite slurry, then adding the slurry to a known quantity of the wet, composited soil. Once added, the blends were thoroughly mixed by hand with a spatula or automated mixer until visually homogeneous. The blended media was then placed in cylinders as a poured slurry (without compaction) for curing periods of 7 and 28 days, at which point the samples were extruded and tested for the various geotechnical and environmental parameters discussed in Section 2.2.

Three replicates of each mix design were prepared and strength tested for each curing period to ensure reproducibility of the results. As a conservative measure, the two weakest 28-day cured UCS specimens for each mix design were subject to totals and SPLP testing, as shown in Figure 1. Additional cured soil cement specimens of selected mix designs (M-1, -2, -6) were prepared for hydraulic conductivity (ASTM D5084) and the semi-dynamic leaching tests (ASTM C1308, performed as a later step in the treatability study), as they represented the minimum and maximum reagent doses (M-2; M-6) and the control mix without GAC (M-1).

5.0 MIX DESIGN (SOIL-CEMENT) RESULTS

The geotechnical results for the mix designs are contained in Appendix B and include UCS testing results (ASTM D1633) for all six mix designs at 7 and 28 days and hydraulic conductivity testing (ASTM D5084) for three 28-day cured mixes (M-1, M-2, and M-6).

Table 1 provides a summary of the mix design components and resulting data. The mix identification number appears in the first column. The next four columns (dry mix characteristics) show the total pozzolan dose by dry weight to the wet weight of the composite soil (~125 lb/ft3) and its breakdown by component (cement/slag, bentonite, GAC). Since the reagents were administered as a slurry to the wet, composite soil, the next three columns (slurry characteristics) are provided for convenience as operational parameters for potential full scale implementation. The next four columns (mix properties) show the average unconfined compressive strength (UCS) values for 7 and 28-day curing, respectively, followed by the reported values of the corresponding hydraulic conductivities for the three selected mix designs.

5.1 Unconfined Compressive Strength Testing

Table 1 summarizes the 7-day and 28-day UCS testing results conducted in triplicate for each curing period.

The average strength shown in Table 1 is based on the three replicates except when a significant outlier was recorded (as was the case with replicates 1-D and 2-D). In such cases, the outlier was dropped from the averaging process because it was approximately 60 psi less than the other two replicates, clearly indicating something was wrong with the replicate. After the 28-day replicates were broken, the sample breaks for the weaker two of the three replicates were subjected to environmental sampling (see Section 6.3). This sample selection was based on strength being positively correlated with contaminant immobilization, thus the selection of the weaker specimens would be the more conservative approach.

Project 09150034 / March 8, 2010 Page 8 Schnabel Engineering, LLC

The 28-day strength of all mix designs will be adequate for site construction and future site use. Assuming a factor of safety of 3 on the maximum load, the weakest mix design (approximately 90 psi) has the ability to support the equivalent of 35 ft of fill having a unit weight of 125 lb/ft3. This should exceed the future anticipated dead loads and construction vehicle live loads on the site following implementation of the complete AWI remedy.

It must be noted that once installed, the 28-day cured mixed media will likely require the use of backhoes and excavators if the soil-cement needs to be removed and/or graded.

5.2 Hydraulic Conductivity Testing

Hydraulic conductivity determinations of soil-cement matrices are difficult because the stiff soil fabric complicates the back pressure saturation process, which can take between 2 to 4 weeks. Once saturated, the samples were consolidated to a cell pressure commensurate with a position approximately 10 ft bgs, with the water table assumed to be at 5 ft bgs. For typical S/S applications, a maximum hydraulic conductivity (k) of the mix is specified on the order of 10-6 or 10-7 cm/s.

Table 1 shows the measured K-values for the three tested mix designs are very low, in the range of 10-6 to almost 10-10 cm/s. These hydraulic conductivity values will render an intact soil-cement mass as a diffusion-controlled system. The presence of GAC in the mix design appears to significantly lower the measured K-value, very likely due to its hydrophobic nature and its ability to clog the finer pores. Since the site soils underlying the HDA ponds are sandy in nature, the issue arises as to what the impact of a large, low permeable S/S-treated zone will be on the local groundwater regime and site structures.

Schnabel understands from EA that following S/S of the HDA, other AWI remedy components will be implemented, including the construction of a containment berm which will be placed directly over the in-situ stabilized soil. This berm will provide future containment of contaminated sediments dredged from the Elizabeth River. Management of both runon/runoff and groundwater is an important design aspect for the overall AWI remedy. Once the perimeter berm is installed over the stabilized area, accumulation of water can be controlled by careful grading of an interim soil cap over the completed S/S area so it drains to the north and runoff can be managed within the empty cell. Future placement of dredged sediments on the stabilized area will also be graded to drain to the north (towards the inside of the containment cell). Drainage of any infiltration on the inside of the containment cell will be addressed through engineering controls, such as the use of drainage pipes inside the cell. Once the cell is ultimately capped, the S/S area should be essentially isolated from precipitation/runon. It is expected that groundwater will ultimately seek its path of least resistance and go around the stabilized monolith.

6.0 ENVIRONMENTAL RESULTS

6.1 General Remarks

The treatability study focused on immobilizing the COCs that have numeric groundwater cleanup criteria (see Section 1.2). It must be acknowledged at the outset of this discussion that the environmental quality of a near neutral pH, dilute groundwater sample used to establish groundwater cleanup criteria in the ROD is significantly different than a TCLP, SPLP or even a DI water extraction from a soil-cement specimen. Soil-cement leachates do not resemble dilute groundwater in terms of turbidity, suspended solids, ionic strength, composition, pH, and pH-dependent behavior (solubility, precipitation, sorption) of solutes, which can be either exponentially increased or decreased relative to natural groundwater conditions. In addition, the TCLP/SPLP/DI water extractions are performed on sample particulates (i.e., the analytical method requires grinding of samples prior to extraction). Actual leaching conditions following stabilization will be from a monolithic mass. As such, these testing methods for treatability studies likely overestimate the actual

Project 09150034 / March 8, 2010 Page 9 Schnabel Engineering, LLC contaminant leaching that will occur in the field. Regardless, such direct comparisons are made for purposes performance of a particular reagent mix even though they may not be realistic.

In terms of chronology, it is also important to note that given the fast track nature of the treatability study, the environmental analytical results from the S- and L-series samples (Section 6.2) were not available for review prior to creating the mix designs. Recall that the mix design took a general industry approach to treating typical creosote and MGP-like COCs in the soils. Thus, when the total and SPLP results on the S- and L-series soil became available (while the mixes were curing), the data were compared and the measured PCP concentrations in the SPLP leachates were significantly higher than expected. Accordingly, concern immediately arose that the mixes would not strongly immobilize PCP and like compounds. This set in motion several supplemental analyses which included:

A re-verification of the SPLP results from the mix designs using a DI water leach (described in Section 6.3) The semi-dynamic leaching testing (Section 7.0) A DNAPL stability evaluation for the soil-cement samples (Section 8.0)

The purpose of these three sets of analyses was to resolve analytical discrepancies and to more completely characterize how the mixes would restrict the mobility of the DNAPL and dissolved phase migration of PCP, since 14 of 15 of the COCs in the ROD appeared to be already below ROD criteria and/or successfully treated. Sections 6.3, 7.0 and 8.0 outline how these issues were addressed during the execution of the treatability study.

In terms of presentation, the primary environmental data summary tables for the field and laboratory controls and the mix design SPLP/DI data (Tables 3 to 7) and totals (Tables 2, 8 and 9) data focus on the groundwater COCs in the ROD. Additional summaries of the environmental testing suite and the raw data are included in Appendix H. The leftmost columns in Tables 3 to 7 show the actual ROD groundwater standards1 and the adjacent reporting limits for each replicate series, followed by the measured COC values for each replicate.

SPLP concentrations that are both above the reporting limit and the ROD standards are yellow-highlighted for ease of reference. The highlighted COC concentrations are the focus of the discussion, treatment and immobilization strategy.

6.2 Field Controls and Laboratory Composites (S-1 to S-3; L-1 to L-3, Tables 2 and 3)

The baseline reference for the effectiveness of the stabilized soil mixes is the environmental data (leach, totals) associated with the field controls and laboratory composites. Compounds detected in the discrete field and lab prepared samples in excess of the regulatory values include benzene, PCP, benzo(a)pyrene and arsenic (see Table 3). The rest of the detected COCs listed in Table 3 were below their respective regulatory values. Only one replicate showed any trace of dioxin, the value (7.4x10-10 mg/L), well below the ROD value (3x10-8 mg/L).

1 Note that the ROD does not require that the SPLP test meet the groundwater cleanup criteria. However, comparing the groundwater criteria to the SPLP data provides a conservative evaluation that the S/S will be adequate for MNA to function as planned.

Project 09150034 / March 8, 2010 Page 10 Schnabel Engineering, LLC

The benzene, benzo(a)pyrene and arsenic concentrations appear to fluctuate near the ROD standard, whereas the PCP concentrations are well above the ROD values and approach the solubility limit of PCP (~14 mg/L).

Based on the SVOCs explicitly listed in the ROD, this suggested that the DNAPL contained in S-3 (upper fill media) was dominated by PCP, which also serves as the carrier medium for most of the contaminants.

To put these observations into a site context, the site owner has indicated that the HDA was developed in three separate cells, starting in the southwestern corner of the AWI property and proceeding eastward towards Burton’s Point Road. Wastes deposited in the two western-most cells were primarily wood debris and creosote. The eastern-most cell was filled with wastes containing large amounts of PCP. The test soils for this treatability study were intentionally taken from a location which during the 2008 site investigation were observed to have the most amount of DNAPL based on visual characterization. The area where the treatability test soil was collected is located from the last cell utilized for waste disposal (and consequently was anticipated to have the highest PCP levels), and represents a conservative “worst case for DNAPL” sample.

For the lab composite values (L-series), the measured concentrations, on average, generally exceeded those of the field controls (by a factor up to 10). The corresponding totals data for the S- and L-series soil samples are shown in Table 2. The PCP and metals concentrations show best the effects of soil blending to generate the laboratory composite samples—the values between the L replicates are generally more uniform than the field controls (S samples).

6.3 Mix Designs (M-1 to M-6) SPLP Data (Tables 4 to 7)

As indicated in an earlier section, the wood fill from the HDA was blended with the underlying sands at a ratio of 1 to 2 to conservatively simulate the soil (and thus contamination) down drag effects that occur during the in-situ mixing process. This approach is conservative and builds a factor of safety into the mix design process in that the COC concentration treated in the composite soil is well above the average COC condition in the field.

Tables 4 and 5 summarize the SPLP leaching results for the mix designs. While aggressive leaching conditions applied to cemented media often result in significant matrix interferences which prevent accurately determining COC concentrations to 0.001 mg/L or less, the reporting limits for SPLP testing results were often below the groundwater cleanup criteria for some COCs listed in the ROD. The main exception to this is benzo(a)pyrene which has reporting limits for individual S-, L-, or M-series replicates above its ROD standard by about 5x.

In only one case (mix design M-1) did benzene exceed the ROD criteria (M-1D); all other COC concentrations except for PCP were compliant with the ROD and/or were non-detect. The aqueous solubility limit of PCP is approximately 14 mg/L at 20ºC (Verschueren, 1983), and the leached concentrations of PCP approach this value for almost all of the mix designs tested.

6.4 PCP Analytical Data Issues

This section describes the analytical issues associated with the SPLP and totals data for PCP and how they were resolved.

The first indication that problems existed with the PCP analysis was evident when the total PCP concentrations from the laboratory composites (L-1 to L-3) which ranged from 400 to 460 mg/kg (Table 2) were compared to the mix designs (M-series) which were <7.6 mg/kg (Tables 8 and 9). This large discrepancy was not exhibited by the totals data for the other listed COCs. For instance, the total VOC, benzo(a)pyrene and metals concentrations between the lab composites and mix designs are similar (compare Tables 2, 8 and 9), except that the mix designs are approximately 10% less, due to reagent dosing. Likewise, Project 09150034 / March 8, 2010 Page 11 Schnabel Engineering, LLC the total PCP concentrations for the mix designs should generally reflect the average dilution rate (~10% ave.), assuming no significant evaporation of PCP. These reductions were not observed.