GEI Consultants, Inc. › work › 01 › 567890.pdf · 567890 : GEI Consultants, Inc. SOURCE CONTROL : V . SUPPLEMENTAL TREATABILITY TESTING O'CONNOR COMPANY SUPERFUND SITE AUGUSTA,

f \

SDMS DOCID 567890

GEI Consul tan ts , Inc .

SOURCE CONTROL V

SUPPLEMENTAL TREATABILITY TESTING

O'CONNOR COMPANY SUPERFUND SITE

AUGUSTA, MAINE

Submitted by

GEI Consultants, Inc.

Jeffrfey A. Klaiber, P.E.

Senior Project Manager

PRIVILEGED AND CONFIDENTIAL 53 Regional Drive Concord, New Hampshire (603) 224-7979

Project 94170 Revision 0

March 24, 1995

SC Supplemental Treatability Testing Revision 0

March 24, 1995

EXECUTIVE SUMMARY

Supplemental treatability testing was conducted by CF Systems (CF) for Central Maine Power Company (CMP) to provide information for preparation of a response to a solvent extraction Request for Quotation (RFQ), which is due to CMP April 10, 1995. Resources Conservation Company (RCC) voluntarily undertook supplemental treatability testing as part of its response to the solvent extraction RFQ. RCC's results will be available upon their publication in RCC's response to the RFQ. The CF Systems' data herein will be used in a competitive bidding process, therefore, CMP considers these data to be confidential and requests that the U.S. Environmental Protection Agency (EPA) not release them to the public until final selection of the Solvent Extraction Vendor.

Previous pilot-scale treatability testing by CF showed poor reproducibility of testing results. Polychlorinated biphenyls (PCB) concentrations in fill were reduced from between 320 parts per million (ppm) and 345 ppm to between 6 ppm and 32 ppm. PCB concentrations in clay were reduced from between 83 ppm and 185 ppm to between 5.4 ppm and 33 ppm. Poor physical mixing was identified as the primary likely cause for the poor results. Additional concerns were also raised that the treated solids which had been flocculated during dewatering would not be suitable for backfill. Based on pilot-scale results, it was estimated that approximately 15 percent of soil could fail to meet the target cleanup goal of 10 ppm.

The objectives of CF's supplemental treatability testing were to determine if CF's modified mixing configuration improves solvent extraction treatment effectiveness, to determine the effect of initial moisture content on treatment effectiveness, to determine the geotechnical properties of the treated soil, to determine the effectiveness of post-treatment dewatering options, and to generate sufficient data to develop a cost estimate of full-scale treatment costs.

CF's supplemental treatability testing demonstrated improved treatment effectiveness in most samples by reducing PCB concentrations from approximately 600 ppm to less than 6 ppm in fill and from approximately 300 ppm to less than 7 ppm in a feedstock consisting of two parts clay to one part fill (clay/fill). In one fill sample, PCB concentrations in the treated soil were only reduced to 18 ppm. Improved solvent extraction treatment effectiveness appears to be the result of two process changes: a new mixer blade and additional flushes of clean propane at the end of each extraction stage.

Air-drying prior to treatment appeared to enhance treatment effectiveness particularly in the clay/fill feedstock. The air-dried clay/fill feedstock was treated to PCB concentrations less than 2 ppm. This contrasts to the results for the same feedstock being extracted without altering its as-receiVed moisture content (treated soil had 7 ppm PCBs) and being extracted as a slurry (treated soil had 4.3 ppm PCBs.)

PRIVILEGED AND CONFIDENTIAL

i *


March 24, 1995

GF developed a dewatering process for the slurried treated soil that included use of two flocculants. GEI Consultants, Inc. (GEI) recommended reversing the dosing order of the flocculants which reduced the amount of time needed to produce a filter cake. The physical properties of uncontaminated clay, extracted clay, and the filter cakes produced by CF's dewatering process and their process as modified by GEI were compared.

There was no significant change in the suitability of the treated clay as backfill when CF's solvent extraction treatment process was used. The moisture content of the soil after vadium press dewatering was significantly higher than the optimum moisture content for compaction and additional dewatering/drying will he required. Alternately, it is recommended that CF consider using pneumatic means to remove the soil from the extraction vessel without the use of water (dry method).

Two mechanical problems were identified during supplemental treatability testing and need to be addressed in design of the full-scale treatment unit: (1) excessive high rate of mixer wear and (2) clogging of frit filters, fill nozzles and drain ports. The information developed during this study will be used by CF to provide a quotation to CMP for hill-scale treatment costs.



March 24, 1995

TABLE OF CONTENTS

EXECUTIVE SUMMARY TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF APPENDICES

PAGE NO.

1. INTRODUCTION 1

1.1 Purpose and Objectives 1

1.1.1 Confidential Information 1

1.1.2 Purpose 1

1.1.3 Objectives 2

1.2 Summary of Previous Treatability Testing 4

1.2.1 Laboratory-ScaleTesting 4

1.2.2 Bench-Scale Testing 5

1.2.3 Pilot-Scale Testing 6

2. DESCRIPTION OF SUPPLEMENTAL TREATABILITY TESTING 9

2.1 Sample Collection 9

2.2 CP's Pilot-Scale Treatability Testing 9

2.3 RCC's Bench-Scale Treatability Testing 10

3. DESCRIPTION OF GEOTECHNICAL TESTING 11

4. SITE VISIT 13

4.1 Objectives 13

4.2 Feedstock Preparation 13

4.3 Mobile Demonstration Unit Operation 14

4.4 Preliminary Treatability Test Results 16

4.5 Dewatering Operation 16

4.6 Summary of Observations 17

5. TREATABILITY TESTING RESULTS 18

6. GEOTECHNICAL TESTING RESULTS 20

TABLES APPENDICES


March 24, 1995

LIST OF TABLES

1 -2 -3 -4 -

CFSystems Feedstock Preparation CF Systems Preliminary Treatability Test Results Summary of Geotechnical Testing Moisture Content Data for Contaminated Soil Test Runs

LIST OF FIGURES

1 - Sample Locations

LIST OF APPENDICES

A -B -C -

CF Systems, Inc. Summary Report Geotechnical Laboratory Test Results by James Valentine Memorandum, A. P. Davis, Jr. to Charles R. Nickerson, Dated February 16, 1995

SC Supplemental treatability Testing Revision 0

March 24, 1995 Page 1

1. INTRODUCTION

1.1 Purpose and Objectives

The results of supplemental treatability testing for the Source Control (SC) component of the O'Connor Company Superfund Site (Site) Remedy established in the 1989 Record of Decision (ROD) and as modified by the 1994 Explanation of Significant Differences (ESD) are presented in this report. This report was prepared by GET Consultants, Inc. (GEI) for Central Maine Power Company (CMP) of Augusta, Maine, Work was performed in accordance with the Scope of Work for Supplemental Treatability Testing dated October 7, 1994 (Scope of Work) and the information provided in a December 21, 1994 GEI letter entitled "Response to U.S. Environmental Protection Agency and Maine Department of Environmental Protection Comments Supplemental Treatability Testing".

1.1.1 Confidential Information

Supplemental treatability testing Was undertaken to provide information for preparation of solvent extraction vendor proposals for Remedial Action, Since these data will be used in a competitive bidding process, CMP considers these data to be confidential and requests that the U.S. Environmental Protection Agency (EPA) not release them to the public until final selection of the Solvent Extraction Vendor for this project.

1.1.2 Purpose

Two out of eleven solvent extraction Vendors solicited by CMP have been prequalified to provide quotations for solvent extraction treatment of soil and sediment during the Remedial Action. The two vendors, Resources Conservation Company (RCC) and CF Systems (CF), have been asked to provide proposals for the Remedial Action. CF has previously performed bench-scale and pilot-scale treatability testing at the Site. RCC has previously performed laboratory-scale and bench-scale treatability testing on soil and sediment from the Site.

Based on the previous treatability testing, it was estimated that most Of the soil With initial polychlorinated biphenyls (PCB) concentrations greater than 10 parts per million (ppm) would not achieve the 1989 ROD PCB target cleanup goal of 1 ppm and that most of the soil With initial concentrations of carcinogenic polycyclic aromatic hydrocarbons (cPAHs) greater than 1 ppm would not achieve the 1989 ROD cPAH target cleanup goal of 1 ppm.




Pursuant to an ESD finalized by the EPA on July 11, 1994, die SC remedy for the Site was revised after all the previous treatability testing had been completed. The ESD changed the target cleanup goal of 1 ppm for PCB and cPAHs in soil and sediment within an on-site "Designated Area" to a target cleanup goal of 10 ppm. Soils with concentrations of PCBs and cPAHs between 1 and 10 ppm will be placed in the Designated Area without treatment. Based on pilot-scale testing, it was estimated that approximately 15 percent of soil could fail to meet the target cleanup goal of 10 ppm PCBs.

The purpose of conducting the supplemental treatability testing is to allow the prequalified vendors to develop information they need to prepare proposals for the solvent extraction component of the SC remedy. In particular, the information developed will be used by the vendors to determine their respective full-scale treatment costs based on the target cleanup goals.

1.1.3 Objectives

Objectives of the proposed treatability testing programs for each vendor are described below. The objectives for a specific vendor are dependent upon their solvent extraction process and the results of their previous treatability testing. Some of the stated objectives are needed to enable CMP and the Supervising Contractor, GEI, to assess the vendor's solvent extraction technology capabilities.

CF Supplemental Treatability Testing

The results of CF's pilot-scale testing indicated physical mixing of soil and solvent to be the most significant process parameter affecting treatment effectiveness. Further studies were recommended in the "Source Control Summary Report Pilot-Scale Treatability Study" Revision 1 dated November 1, 1993 (SC Report) to establish the appropriate mixing configuration for fUll-scale implementation of the solvent extraction process. Results of the pilot-scale testing also indicated that the treated soil and sediment were not dewatered adequately for the purpose of backfilling the material on-site. Further studies were recommended in the SC Report to examine whether the use of water in the treatment process could be reduced or eliminated or, alternatively, whether dewatering techniques could be improved.

CF performed their treatability testing on pilot-scale batches with their Mobile Demonstration Unit (MDU) (note: with mixer modifications) previously used at the Site. The objectives of CF's supplemental treatability testing were as follows:


0



Determine if CF's modified mixing configuration improves solvent extraction treatment effectiveness and enables CF's solvent extraction treatment technology to achieve the target cleanup goals of 10 ppm for PCBs and cPAHs in a greater percentage of soil.

• Determine the effect of initial feedstock moisture content on treatment efficiency.

• Determine the geotechnical properties of the treated soil.

• Generate sufficient data to develop a detailed estimate of full-scale treatment costs.

RCC Supplemental Treatability Testing

RCC's treatability testing was performed at bench-scale in their laboratory and pilot test facility. RCC's Toxic Substances Control Act (TSCA) permit does not allow them to conduct pilot-scale treatability testing at the facility. RCC has stated that they can prepare a proposal based on the testing proposed under this Work Plan and their previous treatability testing results fftim the Site. The objectives of RCC's supplemental treatability testing were as follows:

• Verify ability of RCC's B.E.S.T.® process to remove PCBs and cPAHs from the Site soils.

• Record observations and data to predict full-scale performance of the B.E.S.T.® process.

• Take samples during simulation of the treatment train and conduct sufficient analyses to determine the removal efficiency for PCBs.

• Generate sufficient data to develop a detailed estimate of full-scale treatment costs.

RCC voluntarily undertook this work as part of their response to CMP's Request for Quotation (RFQ). They will only release their results as part of their submittal to the RFQ due to CMP April 10, 1995. Consequently the treatability test results are not provided in this report but will be submitted to EPA after CMP's receipt of RCC's response to the RFQ.




Summary of Previous Treatability Testing

1.2.1 Laboratory-Scale Testing

RCC: Feasibility Study

During the Feasibility Study, RCC conducted laboratory-scale testing on soil from the Site. Three different soils were tested: an overburden clayey silt; an overburden fill material which was a heterogeneous mixture of silt, clay, and sand; and a surficial soil which was a Hollis series fine sandy loam. Treatability testing results were:

The treated silty clay and sandy loam samples passed the EP Toxicity test for metals while the treated fill sample exceeded the EP Toxicity criterion for lead. Initial lead concentrations were 85 ppm, 150 ppm, and 1,900 ppm, respectively.

The results of the solvent extraction treatability study showed that the PCB concentrations in the Site soils were reduced to 2.6 ppm to 19 ppm after three extractions. The majority of the extractable PCBs was removed in the first three or four extraction stages. PCB concentrations were reduced by over three orders of magnitude in two soil types and by over two orders of magnitude in the other soil type after three extractions. The residual PCBs appeared to be essentially non extractable.




1,2.2 Bench-Scale Testing

Bench-Scale testing was performed by both CF and RGC during Pre-Design studies.

RCC: Pre-Design

Upon receipt, RCC screened the soil and sediment samples through standard Tyler sieves to remove any materials greater than one-quarter inch in size and to further homogenize the samples. As a result of homogenization, PCB concentrations detected in the feed samples were significantly higher than those detected in grab samples from initial characterization. The change in PCB concentrations in bench-scale treatability samples after homogenization is likely the result of in situ heterogeneities in PCB distributions. PCB concentrations are anticipated to be even more variable in full-scale treatment batches.

Five of the six samples were treated to below the former target cleanup goal of 1 ppm.

CF: Pre-Design

Upon receipt, CF manually mixed the soil and sediment to homogenize the samples. PCB concentrations in the homogenized feed samples were higher or lower than those detected in grab samples from initial characterization. The change in PCB concentrations in bench-scale treatability samples after homogenization is likely the result of in situ heterogeneities in PCB distributions. These results reflect the heterogeneity of PCB concentrations in relatively small laboratory-scale sample volumes and further indicate complications with PCB concentration variability during full-scale implementation.

Four of the six samples were treated to levels below the former target cleanup goal of 1 ppm after five extractions. The fill sample containing a feed PCB concentration of 535 ppm and the sediment sample did not achieve the cleanup goal. The data indicated that CF's liquified propane extraction system will not achieve title PCB target cleanup goal for soil containing initial PCB concentrations greater than 500 ppm. Note that initial PCB concentrations in fill or clay between 100 ppm and 500 ppm were not tested by RCC in the previous laboratory program performed during the Feasibility Study. Poor reproducibility of results was exhibited in the "duplicate" clay samples with initial PCB concentrations of 92 ppm and 113 ppm. Achieving the former target cleanup goal (1 ppm) was anticipated to be more difficult in pilot-scale and full-scale implementation.

In general, PCB concentrations significantly decreased with each additional extraction stage, Up to five extractions. PCB concentrations were reduced to below the fonner




target cleanup goal of 1 ppm after one extraction in the feedstocks containing less than 10 ppm PCBs. The duplicate clay samples containing 113 ppm and 92 ppm PCBs achieved the former target cleanup goal after two and six extractions, respectively. The data indicated that PCB concentrations could be reduced to less than 1 ppm within two to five extractions in bench-scale samples if the feed concentration of PCBs were less than 100 ppm. Three to six extractions would be needed to approach the former target cleanup goal of 1 ppm in soil with initial PCB concentrations greater than 100 ppm. More than six extraction stages did not improve the effectiveness of the technology in the bench-scale samples.

The concentration of cPAHs in the sediment feed sample and three clay feed samples were below detection limits and, therefore, cPAH extraction efficiencies could not be determined. Concentrations of cPAHs were reduced from 13.5 ppm to below detection limits in one fill sample after two extractions. However, in the other fill sample, cPAH concentrations were only reduced to 59.6 ppm after six extractions from an initial concentration of 139 ppm. These data indicate that CF's solvent extraction process may have difficulty in reducing cPAH levels in soil containing high initial cPAH concentrations. The CF solvent extraction process also seems to be less efficient in treating sediments with high moisture and organic content.

Total lead concentrations increased in all treated samples. CF's solvent extraction process does not reduce lead contamination. This change in concentration of lead appears to be primarily due to a decrease in the sample's total weight due to loss of extracted organic material and a small loss of solids during the solvent extraction process. No Toxicity Characteristic Leaching Procedure (TCLP) analyses or lead stabilization tests were performed during the CF bench-scale testing.

1.2.3 Pilot-Scale Testing

Pilot-scale treatability testing was performed by CF as part of Pre-Design Studies during the summer and fall of 1992 to examine whether solvent extraction technology could achieve the target cleanup goals of 1 ppm PCBs and cPAHs. The materials on which this technology was tested were:

• fill containing initial PCB concentrations between 320 ppm and 345 ppm and initial cPAH concentrations of approximately 45 ppm;

• clay with initial PCB concentrations less than 5 ppm;

• Clay with initial PCB concentrations between 93 and 97 ppm;




clay with initial PCB concentrations between 83 and 185 ppm; and

sediments with initial PCB concentrations of approximately 45 ppm.

Pilot-scale treatability testing results were:

Concentrations of cPAHs were reduced from an average initial concentration .of approximately 45 ppm to between 7 ppm and 17 ppm. Based on pilot-scale treatability testing results, it was estimated that only soil with initial PCB concentrations less than 10 ppm and Sediments would reliably achieve the PCB target cleanup goal of 1 ppm. Therefore, it was estimated that 12,000 cubic yards (cy) out of 31,500 cy of soil and sediment would fail to meet the PCB and cPAH target cleanup goal of 1 ppm stipulated in the 1989 ROD.

As a result of the inability to obtain the 1989 ROD target cleanup goals, the treatability testing scope of work was expanded to examine whether key process changes would improve performance of CF's solvent extraction technology. Of the process changes examined, physical mixing of soil With the propane solvent seemed to be the most significant process parameter affecting treatment efficiency.

Poor physical mixing was attributed to the type of mixing equipment used and the potential that isolated pockets of poorly-mixed soil may have adhered to the walls of the extraction vessel. CF believed that during pilot-scale operation these "dead zones" were forced out of the reaction vessel at the end of each test by flushing with water. However, in the full-scale unit, any "dead zones" within the extraction would become filled with solids and allowed to remain so between treatment batches. Only treated solids that are readily drained from the extractor would be removed at the end of each treatment batch. CF anticipated that these treated soils would be better mixed and, therefore, have lower concentrations of PCBs. Additional laboratory-scale studies were recommended in the SC Report to establish the appropriate mixing configuration for full-scale.




As described in the SC Report, water was used to displace the propane after the final pilot-scale extraction stage and expel the treated solids from the extraction vessel. This process resulted in a slurry of water and treated solids at the completion of solvent extraction. This slurry requires dewatering prior to backfilling the treated soil on-site.

During pilot-scale, a flocculant was added to the slurry, and the free Water was Her-antpd Belt filter pressing was then used to dewater the flocculated solids. However, this dewatering method was not sufficient to reduce the water content to a level such that the treated soil would be readily placed and compacted as on-site fill. Based on visual examination and the results of the preliminary laboratory testing, the physical properties of the flocculated clay were expected to be similar to those of an organic soil.

Of particular concern were the expected low strength and high compressibility of the treated material. The material may also exhibit large secondary compression characteristics (i.e., long-term post remediation settlement). This treated material would be difficult to handle and backfill during full-scale remediation, would create excessive bulking compared to the material's current in situ volume, and would likely possess insufficient strength to support itself, resulting in mud-waves, unstable slopes and large post-remediation differential settlement of the ground surface.




2. DESCRIPTION OF SUPPLEMENTAL TREATABILITY TESTING

2.1 Sample Collection

On December 22, 1992, eight 55-gallon drums of contaminated (on-site) soils were excavated from a test pit located in the northeastern portion of Transformer Work Area I, Previous data indicated that this was an area of high concentrations of PCBs (greater than 100 ppm PCBs) (Figure 1). This material was used by CF and RCC to conduct supplemental treatability testing. Four 55-gallon drums each of fill and silty clay were obtained for use for supplemental treatability testing. The fill was sieved to remove materials greater than 3/8-inch in diameter. No confirmation analyses were performed at the time of excavation- The eight drums of soil were covered with plastic sheeting and staged on wooden pallets within the equipment decontamination corridor. The drums were transported to the Solvent Extraction Vendors in accordance with the requirements of the TSCA, (40 CFR 761), the U.S. Department of Transportation (DOT), and in accordance with CF's and RCC's respective TSCA permits.

Clay was also collected from an uncontaminated portion of the Site and shipped directly to CF for use in evaluating its dewatering procedure.

2.2 CF's Pilot-Scale Treatability Testing

CF's summary report is provided in Appendix A. As described in the Scope of Work, CF's work was performed in general accordance with the EPA-approved SC Pilot-Scale Treatability Study Pre-Design Work Plan and Project Operations Plan (Revision 1) dated June 19,1992. The work was performed at Hazen Research, Inc. (Hazen) in Golden, Colorado.

CF performed five treatability tests on two feed types, fill and clay. CF initial analytical screening of the soil indicated that only low concentrations (less than 2 ppm) of PCBs were present in the clay. In order to have feed that contained greater than 100 ppm PCBs, CF mixed two parts clay (approximately 2 ppm) with one part fill (approximately 400 ppm) to represent the clay feed (clay/fill feedstock).

The first three treatment batches were run under the following conditions:

» Batch No. 1

Feed was run with "as received" moisture.

• Batch No. 2

Feed was air-dried.



- March 24, 1995 Page 10

• Batch No. 3

Feed was slurried with water.

• Treatment batches Nos. 4 and 5 were replicate batches of the treatment batch that showed the best treatment effectiveness.

Treatment batches Nos. 4 and 5 were run on air-dried batches of feed similar to run No. 2. Process parameters are described in CF's summary report. As indicated in the Scope of Work, duplicate samples by EPA's oversight contractor were not collected;

The treated material referred to as raffinate was recovered from the extraction vessel by two different methods: a dry method and a slurry/flocculate/filter method. The dry raffinate recovery method indicates that the treated solids were physically removed from the extraction vessel without the addition of water. The slurry/floceulate/filter method indicates removal of the treated solids from the extraction vessel with water producing a slurry, The slurry was, in most cases, treated with flocculants and dewatered with a vacuum filter press. Two fill samples were dewatered without the use of flocculants.

2.3 RCC's Bench-Scale Treatability Testing

RCC planned to perform one bench-scale treatability test on clay containing between 50 and 500 ppm PCBs. As mentioned previously, results of this testing will be submitted to CMP as part of RCC's response to the RFQ due to CMP on April 10, 1995. As described in the Scope of Work, RCC's treatability testing was to be performed in general accordance with the EPA-approved SC Pre-Design Work Plan (Revision 1) dated August 7, 1991.

The proposed bench scale treatability test was designed to provide the following data:

• Solvent/feed ratio (volume basis) required to extract the organic contaminants (PCBs) from the raw waste;

• Mass balances (oil, water, solids, PCBs);

• Volume and mass determinations;

• Chemical treatment parameters required to reach the desired treatment level; and

• Data required to determine equipment configurations, treatment rate and operating costs.

As indicated in the Scope of Work, duplicate samples by EPA's oversight contractor were not collected.




3. DESCRIPTION OF GEOTECHNICAL TESTING

CF subcontracted a filter press vendor, Parkson Corporation (Parkson) of Fort Lauderdale, Florida, to determine the appropriate dewatering procedure for clay at the Site. CF shipped uncontaminated extracted clay to Parkson for evaluation. As described in CF's summary report, Parkson determined that flocculation in series with Percol 712, an anionic polymer, and Percol 767, a cationic polymer, was necessary to procure a filter cake.

During a subsequent site visit (see Section 4), GEI recommended reversing the dosing order of the flocculants, which reduced the time necessary to press the material into a filter cake.

To evaluate CF's selected dewatering process and GEI's modification, the following samples were submitted for geotechnical testing:

Sample No. 1 As received uncontaminated clay

Sample No. 2 Uncontamhiated extracted clay

Sample No. 3 Filter Cake-1 CF's flocculation method using Percol 712 followed by Percol 767

Sample No. 4 Filter Cake-2 Modified flocculation method using Percol 767 followed by 712

Sample No. 1 was collected to compare the physical properties of the clay prior to solvent extraction. Sample No. 2 was collected directly from the extraction vessel after the propane had been allowed to evaporate from the clay. Sampling Nos. 3 and 4 were collected after the extracted clay was removed from the extraction vessel with Water. The resultant slurry was flocculated, and dewatered by filtering with a vacuum press,

The samples were Submitted to James Valentine, Inc. of Niwot, Colorado for the following geotechnical testing:

• Standard Proctor compactor Test by ASTM 698.

• Grain size with hydrometer, and

• Atterberg limits.




The geotechnical results from the tests of uncontaminated clay, as described above, were also compared to water contents for test runs of contaminated fill and clay/fill. Water content data (weight of water divided by dry weight of the sample) was obtained by converting CF's percent moisture data (weight of water divided by total weight of the sample) by the following equation:

Percent Moisture

Moisture Content = —- - x 100

(100-Percent Moisture)

It is important that soil handling procedures prior to and after treatment influenced the water contents of the contaminated material. As described in Section 2, contaminated material was extracted at three moisture contents: the material's in situ moisture content, air-dried, and as a slurry. In addition, the treated solids were removed from the extraction vessels by two methods: dry (without use of water) and as a slurry. The slurry Was then treated with fiocculants and deWatered with a vacuum filter press. Two treated fill samples were removed as a slurry and dewatered without the use of fiocculants.




4. SITE VISIT

4.1 Objectives

On January 9 to 13, 1995, Dr. Benjamin Su of GEI visited Hazen to observe CF's supplemental treatability testing. The objectives of this site visit included:

• Observing the pilot-scale treatability test runs on CF's MDU;

• Observing the soil dewatering operation; and

• Examining the mixer impeller (i.e., modified mixer) for wear.

4.2 Feedstock Preparation

During GEI's observations of treatability testing, GEI also observed feedstock preparation. Contaminated fill and clay soil were received at Hazen in 55-gallon plastic drums, two for each type of soil. Uncontaminated clay soil was received in 5-gallon plastic pails. The clay was hand picked for debris separation and delumped with a garden spade on a tarpaulin-lined concrete floor. The pre-sereened clay was then placed into 20-pound bags and replaced into the Original shipping drums. The fill was pre-screened through a SWECO with a 1/4-inch opening. Screen rejects were delumped with a garden spade on a taipaulin-lined concrete floor. Prepared fill was then placed into 20-pound bags and stored in original shipping drums until testing.

Prior to solvent extraction, the soil was air-dried. Air drying was also needed to prevent clumping so that the clay could be sieved. Sieving was necessary to reduce the particle sizes of the clay samples to below 1/4 inch as required for feed to the MDU system.




Table 1 presents the soil moisture 1 measured by Hazen during feedstock preparation. Percent moisture represents the weight of water divided by the total weight of the sample. The drying time varied widely from 6 hours to 96 hours depending upon the ambient air temperature and other weather conditions. On January 6th, the drying of the 140 to 160 pounds of clay/fill feedstock was accomplished with a propane hot air blower. Percent moisture of this feedstock was reduced from 13.8 percent to 6.8 percent after 60 hours of drying. On January 12th, a sunny, windy day with a high temperature of 65°F, the total drying time for 125 pounds of the uncontaminated clay was about 6 hours. Percent moisture for this feedstock was reduced from 19.5 percent to 10.8 percent.

4.3 Mobile Demonstration Unit Operation

Daily operation of the MDU was conducted by the following personnel from Hazen:

• Steve Will - Project Engineer • Thomas Pinnow - Chief Operator • Brad Leyshock - Assistant Operator

Daily operational directives were received from Mr. John Maekiewiez of CF. An additional technician from Hazen was responsible for preparing the daily feedstocks.

Daily operation included the following sequence of steps:

1. Feedstock was prepared by drying and breaking lumps of soil to sizes less than 1/4 inch. The feedstock was usually prepared at least one day before use.

1 In order to be consistent with CF's Summary Report, the term "percent moisture" as measured by an analytical laboratory is presented throughout the text of this report, except in Section 6 and Tables 3 and 4, where "moisture content" is more appropriately used. Percent moisture refers to weight of water divided by total weight of the sample. This contrasts with the geotechnical term, moisture content, which is weight of water divided by the jky weight of the sample. Percent moisture can be converted to moisture content by the following equation:

Moisture Content- P"cm< * 100

100-Percent Moisture




2. The solvent was preheated.

3. The extractor was charged with 100 lbs. of prepared feedstock.

4. The extractor was filled with 150 lbs. of preheated propane.

5. The soil/propane charge was mixed for 20 minutes.

6. The soil was allowed to settle for 2 minutes.

7. The propane extract was drained.

8. The above steps were repeated until three extraction stages were completed.

9. The extractor was then flushed with 100 lbs. fresh propane and the propane was allowed to drain from the extraction.

10. The propane flush was repeated.

11. The residual propane was scavenged by activating the main compressor or displacing the propane with water injected from the bottom of the extractor.

12. The treated soil slurry was dewatered by vacuum to form a filter cake of about 1-inch thickness.

A daily test run started by heating up the hot oil tank to a temperature at 250°F to 3009F. The hot oil was circulated to the extractor vessel to maintain a desired solvent extraction temperature at 120°F. The oil heating was slow, and it usually took about 2 hours to reach the desired temperature. During this period, operators would clean the extractor drain lines, fill valves, and frit filters to prepare the MDU system for operation. OnCe the extractor vessel reached the desired operation temperature, 100 lbs. of prepared feedstock was charged manually into the extractor from a flanged top opening. Then 150 lbs. of preheated solvent was pumped into the extractor to start the first stage of extraction sequence including, mix (20 min.), settle (2 min.), and bottom drain (25 to 55 min.). The above sequence of operation from fill to drain was repeated to complete three stages of extraction as planned, After the last draining, the treated soil was flushed twice with 100 lbs. of fresh propane and the residual propane was scavenged from the extractor. A water displacement prior to propane scavenging was attempted if the extractor bottom frits were not plugged, A three-stage extraction treatability test was completed in 5 to 6 hours, discounting the 2 hours oil heating.




Excessive propane draining time was experienced during GEI's site visit due to frequent plugging of the frit filters installed at the bottom of the extractor. An improvement of these frit filters will be needed to reduce the maintenance and to shorten the cycle time during full-scale operation.

On January 12, 1995, the extractor vessel was opened to examine mixer wear. The new mixer assembly is a single, three-blade, air foil type of impeller designed by I lightning The extractor walls were fitted with baffles to aid mixing. After six treatability test runs the mixer already showed signs of severe abrasion loss. The blade edges Were rounded off, contrasting sharply with the 90-degree angle of a new blade. The blade was dismounted and weighted. A loss of 7.3 percent from its original weight of 713.8 grams had occurred during the six test runs. Such an abrasion loss is excessive, and an improved construction material is strongly recommended for full-scale operation.

4.4 Preliminary Treatability Test Results

Table 2 presents the preliminary test runs obtained from CF during GEI's site visit. All PCB analyses were conducted by an outside laboratory based on a 24-hour quick turnaround method utilizing sonication as opposed to Soxhlet extraction. Based on these preliminary data, PCB target cleanup goals of 10 ppm were obtained for all feedstocks using three stages of extraction. Lower PCB concentrations were obtained for feedstock soils preconditioned by air drying. For the fill, initial PCB concentrations of 230 to 290 ppm were reduced from 1.7 ppm to a value below the method detection limit. Fill processed as a slurry was treated to 2.3 ppm PCBs. Air-dried fill was treated to below the method detection limit. Initial PCB concentrations in the clay/fill feedstock of 170 to 230 ppm were reduced to 10 ppm in the feedstock processed as-received and to not detected in the air-dried feedstock.

4.5 Dewatering Operation

A vacuum slurry dewatering system was constructed by Hazen's operators to perform tests of dewatering procedures. The system included a 4-foot by 4-foot by 6-inch filtration bed lined with a fabric filter cloth supported by a bottom grid structure. Vacuum was drawn from the bottom of the filtration bed, and the filtrate Was collected in a knockout tank. Vacuum filtration of the solvent extracted soils without polymer conditioning was observed to be slow; it took about 5 hours to dewater one batch of soil slurry at a relatively high vacuum of about 20 inches of mercury (Hg). On January 12, 1995 a vacuum dewatering experiment for the treated clay (uncontaminated) was conducted to evaluate effectiveness of polymer conditioning by two polymers, Percol 712 and Percol 767, as recommended by Parkson. A smaller filtration bed, 2foot diameter, was constructed. The procedures specified by CF were followed by Hazen's operator to conduct this dewatering experimentation. The filtration run took more than 18 hours to complete. GEI recommended that a second test run be conducted using the fiocculants in




reverse order. The recommended modification reduced the dewatering time to 1 hour. The second test run was not observed by GEI.

4.6 Summary of Observations

During GEI's site visit, a total of six treatability test runs including the contaminated fill, Clay/fill, and the uncontaminated clay were observed. These test runs were conducted in accordance with the Scope of Work. The following summarizes GEI's observation of the MDU performance:

• The treatability test runs were apparently successful in removal of PCBs from the treated fill and clay/fill feedstocks to below the target cleanup goal of 10 ppm PCBs. Improved PCB removal appeared to be achieved when the feedstock was air-dried relative to treatment at as-received moisture.

• Vacuum dewatering of the propane extracted soils without the polymer flocculation was unsuccessful. Vacuum dewatering of the treated soils with the aid of polymer flocculation, Percol 712 and 767, as recommended by Parkson, was also disappointing. It took more than 18 hours to complete the first polymer aided dewatering operation on January 12,1995. A subsequent dewatering operation with the polymers added in reverse order as recommended by GEI was reported by CF to be complete in one hour.

• The extractor vessel was opened for inspection of the mixer assembly at the end of the last test run on January 12, 1995. This new mixer is a single three-blade, air foil type of impeller designed by Lightnin. After six batches of soil treatment, the mixer blades were rounded off at the edges and measured a significant weight loss of 7.3 percent from the original weight of 713.8 grams. Based on the above observation, an improved construction material for the mixer will be needed for full-scale operation.

• The treated soils appeared to be granular, fluid, fine, and uniform in sizes. The absence of lump agglomeration in the treated soils could be effects of improved mixing and air-dried feedstock preparation. Inspection of the inside of the extractor on January 12, 1995 showed no large areas of extracted soils smeared against the sides of the extractor (i.e., no dead zones).




5. TREATABILITY TESTING RESULTS

Supplemental treatability testing demonstrated improved treatment effectiveness in most samples by reducing PCB concentrations from approximately 600 ppm to less than 6 ppm in fill and from approximately 300 ppm to less than 7 ppm in clay/fill (Table 2). PCB concentrations in test 5 of the fill were only reduced to 18 ppm. In addition, a duplicate analysis of Test 3 fill showed poor precision (6.1 ppm compared to 12 ppm). Note that the clay/fill feedstock was actually two parts clay to one part fill and that the PCB contamination was associated primarily with the fill material. Therefore, these data may not accurately reflect treatment effectiveness in highly contaminated clay. However, these results are significantly lower than past pilot-scale treatability testing results for either fill or clay.

Aroclor 1242 was the predominant PCB Aroclor detected in the treated soil. Three feedstock samples also contained Aroclor 1260. This contrasts with on-site pilot-scale treatability results which reported predominantly Aroclor 1260. The PCB chromatograms were reviewed and the Aroclor identifications appear to be correct. Reviewing all previous soil data, both Aroclors 1242 and 1260 have been delected in the soil at the Site. As indicated on Figure 1 the sampling location for supplemental treatability testing was different from the sampling locations for the on-site pilot-scale treatability study. In addition, soil samples from other pre-design studies collected near this location indicated a predominance of Aroclor 1242.

Air drying appeared to enhance treatment effectiveness for the clay/fill feedstock and had little, if any, effect on the fill feedstock. Slurrying the clay/fill feedstock produced results in between air drying and extracting the feedstock without modifying the as-received percent moisture, The positive effect of air drying is likely due to two factors: removal of the water barrier surrounding the soil particles enabling better contact with the solvent and better suspension and, hence mixing of the particles in the solvent. Visual observations indicated that drying reduced the amount of feedstock that agglomerated or smeared onto the walls of the extractor. Extracted soil that had been air-dried prior to treatment appeared more uniform in grain size and did not appear to have clumped into balls or clods.

Improved treatment effectiveness appears to result from two process changes: a new mixer and additional flushes of clean propane. The new mixer is a single three blade airfoil, axial flow impeller Which provides high volumetric axial mixing at about 1,200 revolutions per minute (rpm). The previous mixer used during on-site pilot-scale testing produced radial mixing at about 200 rpm. In addition, to a new mixing apparatus, two flushes of propane were performed at the end of each extraction stage. During on-site pilot-scale testing, no flushes of clean propane were performed because water was used to displace the "dirty" propane containing the extracted PCBs.




The additional propane flushes insure a complete removal of the propane extract from the treated soil and may have contributed to the improved treatment effectiveness. However, this procedure will also increase treatment time.

Two major mechanical problems will need to be resolved during design of die full-scale treatment unit;

• high rate of mixer wear, and

• clogging of frit fillers, fill nozzles, and drain ports.




6. GEOTECHNICAL TESTING RESULTS

As part of the supplemental treatability study, a series of geotechnical laboratory tests were conducted on a sample of clean clay soil obtained from a depth of 7 feet in test pit TP-601 (see Appendix C). The laboratory tests were performed to assess the potential changes in soil characteristics that might occur during solvent extraction treatment. TTie testing was performed on soil subsamples obtained during the Stages of processing, as described below, in the CF MDU. The results of the geotechnical laboratory testing are presented in Appendix B and summarized in Table 3.

Descriptions of the samples tested and discussion of the testing results are presented below:

• Sample 1 is a Specimen of the clay as received from TP-601. This test series was conducted to ensure that the clay sample shipped to Hazen Research in Golden, Colorado (Hazen) for testing in the MDU was unchanged during shipment. Based on comparison of sample 1 testing results and the results of testing previously performed on other samples obtained at a depth of 7 feet from TP-601 (presented in Appendix C), it was concluded that the soil tested in the MDU was representative of the clay stratum of interest.

• Sample 2 was taken from the MDU immediately after treatment to represent the clay post-treatment but before being slurried. The post-treatment classification properties were essentially unchanged from the pre-treatment classification properties. However, the maximum dry density, as determined by ASTM D698, was noticeably higher and the optimum moisture content lower than the other test results presented in Table 3. It may be that CF's solvent extraction process induces changes in the clay particle structure or mechanically causes particle breakage thereby permitting denser packing of particles and a resulting higher maximum dry density and lower optimum moisture content.

• Samples 3 and 4 were taken from the vacuum press filter cake following dewatering from a slurry treated with a flocculant. The flocCulant consisted of Percol 712 (anionic) and Percol 767 (cationic). For sample 3, Percol 712 was added first, followed by the addition of Percol 767. For sample 4 the order of flocculant addition was reversed (i.e., Percol 767 followed by Percol 712). The slurry was prepared at a ratio of 1,8 parts water to 1.0 parts soil based on weight.




The results of classification and compaction testing on samples 3 and 4, shown in Table 3, were similar to those obtained for the other samples tested.

It was observed, however, that the Atterberg limits (liquid and plastic limits) test results for samples 3 and 4 were higher than other reported values. For sample 3 the liquid limit test result of 51 classified the soil as a clay with high plasticity (CH). Samples 1, 2, and 4 were classified as clay with low plasticity (CL). Atterberg limit data were plotted and are shown in Appendix B. The observed increase in the Atterberg limits may have been due to constituents of the water used to slurry the treated clay from the MDU and/or the addition of the flocculants. However, these observed changes in properties are not expected to impact handling or backfill requirements at the site.

The results of compaction testing on samples 3 and 4 were similar to the test results for the clay soil prior to its treatment in CF's MDU (Sample 1). The decrease in maximum density as compared to sample 2 is likely related to the addition of the flocculants and the samples' corresponding increase in plasticity.

In summary, test results indicated that these particular flocculants, at the ratio applied, will not significantly change the compaction characteristics of the clay soil. However, it was noted that the moisture content of the filter cakes greatly exceeds the optimum moisture content for compaction, and additional measures for drying of the clay soil would be required.

The Solvent Extraction Vendor will be required to deliver the treated soil to the Remediation Contractor suitable for backfill. Backfill requirements for treated soil require that the soil be placed at a moisture content between the optimal moisture content and the optimal moisture minus 5 percentage points. As shown on Table 4, reproduced from CF's Summary Report, the Water content of the treated and filter-pressed fill ranged from 15.6 percent to 50.8 percent and from 50.4 percent to 65.5 percent for the treated and filter-pressed clay. CF's dewatering process did not produce a filter cake suitable for backfill. If CF is selected as the Solvent Extraction Vendor, drying or more effective filter pressing by CF will be needed to meet the design criteria for the Remedial Action.

The treated soil removed dry from the MDU during this pilot-scale test series (Sample 2 on Table 3) had a moisture content below the specified range for backfill requirements. If this "dry" treatment method is selected, the addition of water will likely be necessary to increase the moisture content to the specified range. In addition, the extracted dry soil had a powdery characteristic that could cause particulate emission problems during pneumatic transport. If CF uses these procedures during the Remedial Action, water must be added to the dry soil to reduce dust formation prior to placement in temporary storage for analytical testing.




Based on our evaluation of the testing data, the following conclusions are offered:

• There was no significant change in the clay soil compaction characteristics due to CF's solvent extraction treatment process.

0 The addition of flocculants slightly increased the plasticity of the soil. However the observed change did not impact soil handling or backfill requirements.

• The vacuum press dewatering process was inadequate. The moisture content of the soil after vacuum press dewatering was significantly higher than the optimum moisture content for compaction and additional dewatering/drying would be required.

• Alternately, it is recommended that CF consider using pneumatic means to remove the soil from the extraction vessel without the use of water (dry method). If pneumatic means are used to remove dry soil from the production solvent extraction units, care must be taken to control particulate air emissions. In addition, water will likely need to be added to the soil to achieve specified moisture contents.


TABLE 1- CF SYSTEMS FEEDSTOCK PREPARATION

Supplemental Treatability Testing O'Connor Company Superfund Site Augusta, Maine

Test Soil Type Percent Moisture'11 Approximate Drying Time

Before After hours'21 Drying Drying

572-1-P01 Fill 12/27/94 13.8 13.8 none

572-1-P02 Fill 12/29/94 13.8 5.3 24

572-1-P03 Fill 1/5/95 13.8 25.9 none

572-1-P04 Fill 1/9/95 13.8 6.9 60

572-1-P05 Fill 1/9/95 13.8 6.8 60

572-2-P01 Clay/Fill, 2:1 1/3/95 25.2 25.2 none

572-2-P02 Clay/Fill, 2 1 1/4/95 25.2 11.9 96

572-2-P03 Clay/Fill, 2 1 1/6/95 25.2 31.3 none

57202-PO4 Clay/Fill, 2 1 1/10/95 25.2 9.0 84

572-2-P05 Clay/Fill, 2 1 1/10/95 25.2 9.3 84

572-2-P06 Clay 1/11/95 28:5 12.6 48

572-3-P01 Clean Clay 1/12/95 19.5 10.8

Notes:

1. Percent moisture is the weight of water divided: by the total weight of the sample.

PRIVILEGED AND CONFIDENTIAL Project 94170

GEI Consultants, Inc. March 24, 1995

TABLE 2- CF SYSTEMS PRELIMINARY TREATABILITY TEST RESULTS Supplemental Treatability Testing O'Connor Company Superfund Site Augusta, Maine

PCB Results (ppm) Test No. Soil Type Condition Date

Screening Full Analyses(4>

Fead Treated Feed Treated

Fill As Received 12/27/94 230 1.7 640 5.7

Fill Air Dried 12/29/94 290 ND(2) 640 4.1

Fill Slurry 1/5/95 280 2.3 650 6:1/12®

J3)Fill Air Dried 1/9/95 590 5.9

Fill Air Dried 1/9/95 600 18

Clay/Fill, 2:1 As Received 1/3/95 170 10 280 7.0

Clay/Fill, 2:1 Air Dried 1/4/95 230 ND 390 1.2

Clay/Fill, 2:1 Slurry 1/6/95 150 330 4.3

Clay/Fill, 2:1 Air Dried 1/10/95 260 0.9

Clay/Fill, 2:1 Air Dried 1/10/95 42 0.9

NOTES:

1. PCB analyses were performed with a 24-hour quick turnaround method using sonication as opposed to Soxhlet extraction requiredby the Scope of Work. Results were recorded during GEI Consultants, Inc's. Site visit.

2. ND -not detected.

3. GEI did not observe testing.

4. Polychlorinated biphenyl (PCB) results from full analyses using a Soxhlet extraction procedure Were obtained from CF Systems Summary Report.

5. Duplicate analyses.


Project 94170 GEI Consultants, Inc. March 24, 1995

TABLE 3 - SUMMARY OF GEOTECHNICAL TESTING Supplemental Treatability Testing O'Connor Company Superfund Site Augusta, Maine

Classification Properties'21 ASTM D698(2)

Sample NoJDescription Maximum Dr

Optimum Moisture Contents

Water

Content*31

uses Gs LL'3» PL'31 P|(3) % Gravel

% Sand Fines

(pcf)

1. As received'11 _(4)CL 2.8 38 20 18 2.6 97.4 102 22,2

2. Uncontaminated extracted clay CL 2.77 38 19 19 1.5 98.5 108.2 18.3 9.6(5)

3. Filter cake-1 CH 2.79 51 28 23 1.4 98.6 103.9 19,9 50.4'51

4. Filter cake-2 CL 2.76 44 21 23 1.2 98.8 104.4 20.3 55.2'51

Notes:

1. Material Source: Test pit TP601 at depth of 7 feet (see Appendix C).

2. Tests by Hazen Research Inc., Golden Colorado (see Appendix B).

3. Liquid Limit (LL), Plastic Limit (PL), Plasticity Index (PI), Optimum MoistureContent and water content are expressed as weight of water divided by weight of dry solids.

4. Data not obtained.

5. Data obtained from CF Systems.


Project 94170 GEI Consultants, Inc. March 24,1995

TABLE 4 - MOISTURE CONTENT DATA FOR CONTAMINATED SOIL TEST RUNS

Supplemental Treatability Testing O'Connor Company Superfund Site Augusta, Maine

Moisture Raffinate Moisture Content11' Adjustment Prior Recovery

to Solvent Extraction

Method Pre-Treatment Post-Treatment

Fill Test 1 None Dry 16.0(2> 15.7(3)

Test 2 Air-dried Dry 5.6 5.4

Test 3 Slurried Slurry/Filter 35.0 23.2

Test 4 Air-dried Slurry/Filter 7.4 15.6

Test 5 Air-dried Slurry/Flocculate/ 7.3 50.8 Filter

Clay/Fill Test 1 None Dry 33.7 31,1

Test 2 Air-dried Dry 13.5 12.4

Test 3 Slurried Slurry/Flocculate/ 45.6 63.4 Filter



Notes:

1. Moisture content refers to the weight of water divided by the dry weight of the sample ̂

2. Data were obtained by converting CF Systems' data for percent moisture in Tables 2 and 3 of their Summary Report (see Appendix A) by the following equation:

,, . _ Percent MoistureMoisture Content = : x 100

(100-Percent Moisture)

3. Moisture content data was obtained from Table 4 of CF System's Summary Report provided in Appendix A.


Project 94170 GEI Consultants, Inc. March 24,1995

9+170 P1 3/20/95 EHV/PGL

LEGEND

AREA ok;'V ; &

BBS®®

WETLAND BOUNDARY AS MAPPED BY SMART ASSOCIATES ENVIRONMENTAL CONSULTANTS, INC.

EXTENT OF DESIGNATED AREA

LEDGE

THE

41T, ste MK,/

jh TEST PIT SAMPLE LOCATION FOR 1992 ^ PILOT-SCALE TREATABILITY STUDY

W TEST PIT SAMPLE LOCATION FOR 1994 i ^ SUPPLEMENTAL TREATABILITY TESTING

UPLAND i TP201 yft- /-VMARSH / / />|i\iv;v 7*n

\ ir-/ LOW \ H / -

/-T/^K \V'!r. Gwki

"Z AU Y TE202-A? ''7ZZY\yk> >" 4rr / /K' iazzzznn

•\Q VZ

UPPER -LAGpON

\ 1 fcSr"A' -1'" -"• ./pp tif

' . • » • '

' W-''/

-TO AUGUSTA

I ' » Y_

'\\V-.. /«/

/ROUTE 17 S •~T-

. is

TT TO TOGUS

V

100 200

NOTES

1. EXISTING GROUND SURFACE CONTOURS SITE SURVEY PERFORMED BY CENTRAL

AND ABOVEGROUND FEATURES BASED ON MAINE POWER COMPANY (CMP), 1994.

SCALE, FEET

Central Maine Power Company Supplemental Treatability Testing SAMPLE

2. SOURCE OF ADDITIONAL SITE FEATURES INCLUDING TWAs, SCRAP AREA, LOW AREA, Augusta, Maine O'Connor Co. Superfund Site LOCATIONSAugusta, MaineUPLAND MARSH, AND UPPER AND LOWER LAGOONS FROM EC JORDAN CO.

FEASIBILITY STUDY (FS), FIGURE 1-3, DATED JUNE 1989. 0 GEI Consultants, Inc. Project 94170 March 1995 Fig. 1

APPENDIX A

CF Systems, Inc. Summary Report

3 CF SYSTEMS 5^ A MORRISON KNUDSEN COMPANY

PILOT-SCALE

SOLVENT EXTRACTION

OF PCB-CONTAMINATED SOILS FROM THE

O'CONNOR SUPERFUND SITE

REPORT PREPARED FOR:

CENTRAL MAINE POWER

PREPARED BY:

CF SYSTEMS

BOSTON TECHNOLOGY CENTER

March, 1995

1

Report to Central Maine Power Solvent Extraction of O'Conner Site PCB Contaminated Soils

Rev. 0 - Page 2

EXECUTIVE SUMMARY

CF Systems was contracted by Central Maine Power (CMP) to continue a pilot-scale demonstration of its solvent extraction technology for the separation of PCB-laden oil from two soils from the O'Connor Superfund Site in Augusta, ME. The treatment objective was to demonstrate that consistent extraction efficiency could be attained such that treated solids would contain less than 10 ppm PCBs and 10 ppm cPAHs. A second objective of this study was to demonstrate that the dewatered treated solids obtained at the completion of an extraction would meet certain physical properties that would allow the soil to be backfilled. The demonstration was carried out at Hazen Research in Golden, CO utilizing CF Systems' Mobile Demonstration Unit (MDU).

The pilot-scale study, devised by CMP, CF Systems, and GEI Consultants was a continuation of pilot-scale work performed on the same materials by CF Systems at the O'Connor Site in Augusta, ME in 1993. It called for a total of 10 tests to be completed on two separate feedstocks (clay and fill). The fill sample contained about 600 ppm PCBs and the Clay Mixture sample Contained about 300 ppm PCBs. Aroclor 1242 was identified in both samples, with some evidence of Aroclor 1260 also present. Neither of the feedstocks contained cPAHs significantly above the treatment target. An additional test was completed on an uncontaminated Maine clay so that nonhazardous dewatered solids could be generated for geotechnical testing.

Extraction of PCBs from both soils was readily accomplished. The treatment objective of 10 ppm was consistently attained. The variability in the results observed in the initial pilot work was not repeated during this study. The treatment objective was attained at each weight percent moisture (the major variable of this study) evaluated.

The dewatering portion of the study also went well. The fill, the clay, and the clay mixture were easily dewatered with the aid of an anionic/catiOnic floccUlent combination. Dewatered solids were produced that contained less than 40 wt% moisture (based on the total weight of the sample: i.e. [mass of Water]/[total mass of sample]*100). Samples were provided to GEI for geotechnical testing.

MDU0315.RP0 March, 1995 CF SYSTEMS A Morrison Knudsen Corporation

2


Rev. 0 - Page 3

INTRODUCTION

During December, 1994 through January, 1995 CF Systems conducted a series of pilot-scale tests in order to demonstrate that liquefied-gas solvent extraction can effectively treat PCB-contaminated soils from the O'Connor Superfund Site in Augusta, ME. This pilot program was a continuation of work which was performed in 1993. During that initial work, operational difficulties were encountered that produced highly variable results. The purpose of this phase of the program was to confirm that the variability encountered during the first phase was due to operational difficulties, and that these operational difficulties were corrected.

Questions were also raised during the initial work concerning the physical qualities of filter cake produced by the CF Systems process. This filter Cake must meet certain strength requirements so that it can be backfilled. While quality of the filter cake was not a stated objective during the initial phase, concerns Were voiced. During this phase, two methods for handling the treated solids were demonstrated.

The work plan for this phase of pilot-scale testing called for testing two of the soil types found at the O'Connor site: fill and clay. These were the same sources of contaminated soil as were used for the initial work. One hundred pound batches of feed were subjected to solvent extraction using liquified propane solvent.

Reproducibility of the process was evaluated by subjecting each feedstock to multiple runs holding processing conditions (solvent-to-feed ratio, residence time, and number of stages) constant. In addition, a series of tests were incorporated into the program to determine the effect of feed weight percent moisture on the PCB content of the treated residue. During the initial work, feed moisture was held constant at the soil's excavated weight percent moisture.

The work was performed at Hazen Research in Golden, CO under CF Systems' direction using the same equipment used during the O'Connor site demonstration. Modifications Of the extractor were made to address problems encountered during the initial work. These modifications included a conceptual change for the mixer (from a lower speed turbine to a high speed axial flow impeller) and drain port modification.

The balance of this report contains a detailed explanation of CF Systems' liquefied gas solvent extraction process, a summary of the pilot-scale testing program, a report of the analytical results obtained, and a discussion of those results.

CF SYSTEMSMDU0315.RP0 March, 1995 A Morrison Knudsen Corporation

3


Rev. 0 - Page 4

PROCESS DESCRIPTION

The treatment process description described in the report summarizing the 1993 work is repeated here. The process has not conceptually changed, but certain operations have been modified to improve the process. Those modifications are described in this section.

The process can be divided into three primary operations: feed preparation, extraction, and solvent recovery. The pilot-scale solvent extraction process flow diagram is shown in Figure 1.

3.1 FEED PREPARATION

Feed preparation for this solvent extraction process entails size reduction and moisture control. Feedstocks are typically crushed and/or screened to obtain a -1/4" particle size. Once the proper size is obtained, feed moisture is adjusted to a predetermined level which is optimum for the particular soil type (discussed in more detail in the Discussion section). The purpose of size reduction and moisture control is to expose a large surface area for mass transfer and to create an environment so that efficient mixing can occur.

3.2 EXTRACTION

The extraction section of the plant is comprised of the following major pieces of equipment: (See Figure 1)

1.) Vessel D-2: Extraction Vessel. Externally-jacketed, agitated vessel in which contact between feedstock and extraction solvent is accomplished.

2.) Vessel D-5: Solvent Pre-Heat Vessel. Externally-jacketed vessel where extraction solvent is heated to desired temperature and stored for eventual use in D-2.

3.) Vessel D-3: Extract Surge Vessel. Receiving vessel for extract (solvent plus extracted organics) from D-2.

4.) Pump P-5: Water Pump. High-pressure pump used to introduce water at the end Of the final extraction stage in order to displace residual extract.

MDU0315.RP0 March, 1995 CF SYSTEMS A Morrison Kriudsen Corporation

2 u a o

osr

NONO

> 2

i/) ̂n3 *T) Q CO *o CAi Ha; m§• 23 LO

c "1 2-f —. :H>

u ^ c:

OQ 3 -a

3n> o-

o o CDC/l 3C/l

Oa* fT O«T> 3 o3c/i

O3

£

1•-»

&

2

O*a

Report to Central Maine Power Solvent Extraction of O'Cornier Site PCB Contaminated Soils

1 Rev. 0 - Page 6

The extraction process is a series of fill-mix-settle-drain-refill cycles, where each cycle is defined as an extraction stage. The number of extraction stages, mixing time, and solvent to feed ratio are process parameters which are determined primarily by the degree of product recovery required and the physical behavior of the feedstock under process conditions.

In Order to perform an extraction test, a 100 pound quantity of a particular feedstock is manually loaded into D-2, Once this is complete, the vessel is sealed and pressurized with solvent vapor. The purpose of initial pressurization with solvent vapor is to prevent any instantaneous freezing which could occur if liquid solvent were introduced directly into a vessel at atmospheric pressure.

Vessels D-2 and D-5 are jacketed and heated with the integral hot oil system to the prescribed temperature and pressure. The pressure in each vessel is controlled by an independent control loop. Once at temperature and pressure, liquid solvent is typically introduced into D-2 through two fill ports in the bottom of the vessel. The flow of propane is manually controlled by opening and closing solenoid valves based on readings obtained from a MicroMotion mass flowmeter and/or level indicators on the extractor. The liquid solvent is transported from D-5 to D-2 via pressure differential. Once the proper quantity of solvent is transferred, the mixer mounted on D-2 is started by the operator. After a pre-determined amount Of time, the mixer is turned off (manually) and the solids are allowed time to settle (typically 5-10 minutes). Since liquified propane has a low density and low viscosity, settling is readily accomplished.

Once settling is complete, the liquid contents of the vessel (extract) is drained from D-2 to the extract surge vessel (D-3) via pressure differential. Draining is accomplished by allowing the extract to flow through the settled bed of solids and out the bottom of the extractor. As the extract flows from the extractor, it will first pass through sintered metal frits designed to remove the courser material, then will pass through finer cartridge filters (F-l) before emptying into D-3. Once the draining operation is complete, the drain port is closed. Multiple mixing stages are completed by adding fresh, recycled solvent into D-2 by opening the fill solenoid valve. Again, solvent is introduced into the bottom of this vessel. When the desired amount of solvent has been charged, the vessel is isolated, the mixer turned on, and the next extraction stage begins. (There are also ports above the settled bed of solids Where filling and draining can occur.)

After the final mixing stage and drain is completed, the treated solids can be recovered in one of two ways. Using the first method, fresh solvent is brought into the extractor through an upper side port to flush the bed of solids of any residual extract. During the flush, a quantity of liquefied propane is brought into the extractor above the bed of solids



- Rev. 0 - Page 7

and is allowed to flow through the bed in a plug-flow fashion, essentially pushing out the residual extracted organics. Typically, 100-150 pounds of fresh solvent is used for this operation. Once the final flush and drain have been completed, the extractor is depreSsurized by opening the extractor to the secondary compressor (C-2). When all the propane vapor has been recovered, the extractor is opened by lowering the bottom flange and the treated solids are recovered. Using this method, the treated solids are recovered with virtually the same weight percent moisture as the feed.

In the second method (called "water displacement"), water is used to float the residual extract out the top of the extractor. After the final mix and drain have occurred, the process of removing the residual extract contained in D-2 begins. First, the flow path used for draining solvent is slightly modified. Rather than proceeding directly from D-2 to D-3, the solvent is routed from one of the extractor side ports through a small vessel equipped with a water sensitive level probe. Once it passes through this vessel it goes on to D-3 as with other drain operations. When vapor is obtained at the drain port being Used for this operation, the water pump (P-5) is turned on and water begins to flow into D-2 through the same bottom ports used for solvent loading. Since liquified propane is lighter than water, the water flow forces any remaining liquid solvent out through the drain port, through the vessel with the probe and on to D-3. When a level of water begins to build in the small vessel the water sensitive probe signals the control panel and the water pump shuts off. This action signifies that all the liquid solvent in D-2 has been removed and what now remains are treated solids, water and solvent vapor. The solvent vapor is slowly removed by opening D-2 to the secondary compressor. Once depressurized, the bottom valve of the extraction vessel is opened whereupon the water/solids mixture flows into a product holding tank. The slurry is later dewatered to obtain the treated solids.

3.3 SOLVENT RECOVERY

Solvent recovery is comprised of the following major pieces of equipment: (See Figure 1)

1.) Vessel T-l: Distillation Tower. Single-stage column with reboiler which separates solvent from dissolved oils.

2.) Vessel D-9: Product Storage Vessel. concentrated PCB oil.

Final holding tank for the

3.) C-l / C-2: Primary and Secondary Compressors. Primary utilized to take solvent vapor from top of T-l,

MDU0315.RP0 March, 1995 CF SYSTEMS A Morrison Kriudsen Corporation

Report to Central Maine Power Solvent Extraction of O 'Conner Site PCB Contaminated Soils

Rev. 0 - Page 8

compress it and eventually recycle solvent to extraction section. Secondary compressor used to scavenge residual solvent vapor from extraction vessel and product storage vessel.

4.) Vessel D-8: Solvent Surge Vessel. Receiving vessel for recycled solvent from T-l via C-l.

The solvent recovery section of the plant is where the PCB-laden oil is separated from the solvent and the solvent is subsequently recycled for reuse in the extraction end of the plant. As described above, extract (solvent containing the dissolved organics) is drained into the extract surge vessel (D-3). D-3 serves as the supply vessel for the distillation tower (T-l). Solvent containing extracted organics is fed, via pressure differential, from the bottom of D-3 into T-l through a Control valve.

The distillation tower consists of a horizontal reservoir containing a tube bundle which acts as a reboiler and a vertical section where the vaporized solvent flows out to the compressor. Vaporization of the solvent is accomplished using hot, compressed solvent vapor flowing through the tube side of the heat exchanger bundle while the solvent/organic contaminant mixture boils along the outside surface of the tubes. The solvent, having a very low boiling point relative to the dissolved organics, is vaporized and flows from the top of the tower to the inlet of the main compressor (C-l).

The main compressor compresses the solvent vapor to a higher pressure. This compressed gas flows on the inside of the tube bundle where it cools and partially condenses. Full condensation is realized immediately downstream by means of a water-cooled condenser (HX-2). Condensed, distilled solvent then flows to the solvent surge vessel (D-8) where it is stored until makeup solvent is required in the extraction end of the plant. When solvent is required for extraction, it is pumped from D-8 into the bottom of D-5, being heated along the way as necessary.

The recovered oil is periodically drained from the bottom of T-l into the product storage vessel (D-9). The top of D-9 is connected to the inlet of the secondary compressor, allowing removal and recycling of residual solvent from the concentrated organic mixture. The oil is then drained from the bottom of D-9 into a product drum.



Rev. 0 - Page 9

3.4 ADDITIONAL EQUIPMENT

Additional pieces of equipment which are illustrated in Figure 1 represent parts of either the on-board heating system or the pressure relief safety system. The heat required to perform extractions at elevated temperatures is provided by a hot Oil system. This consists of an insulated reservoir (which houses the oil and immersion heaters) and a gear pump for circulating hot oil to the external jackets of D-5 and D-2. As previously stated, pressure of each vessel is controlled by regulating the flow of hot oil into each jacket.

There are six single loop controllers (PID control) that control the major unit operations of the process. Operator control is required to start and stop the mixer, and to initiate and stop the flow of solvent and extract by manipulating various solenoid and hand valves.

All pressurized vessels are connected to a common relief header by means of individual pressure relief valves. The header in turn flows to a knockout drum (D-ll) which is piped to an activated carbon bed. A rupture disc is piped in parallel with the carbon bed to protect it in event of over-pressurization. The carbon bed and rupture disc bypass then vent to an atmospheric stack and flare.

3.5 PROCESS MODIFICATIONS SINCE 1993 WORK

There have been two main changes in the operation of the CFS system since the demonstration in Augusta, Maine in 1993. The first change addressed a suspected mixing problem. The initial configuration of the mixer was a low speed turbine agitator located low in the vessel (actually in the charge of feed). During the initial testing an axial flow prop was also tried, but We were not able to change the speed of the mixer shaft or to position the mixing prop at an appropriate height with respect to the bed of solids. Since those tests, the system has been reconfigured to allow high speed (up to 1200 rpm) and the use of the axial flow impeller at a variety of heights in the vessel. By increasing the speed and changing the impeller configuration, CF Systems has been able to substantially increase the turbulence inside the extractor and so greatly reduce the possibility of "dead zones" where the feed remains static. By exposing all the feed to a high velocity flow of solvent, much higher degrees of extraction can be achieved.

The second change allows us to drain the extractor from the bottom of the vessel, through the charge of solids. This removes more of the extract from the vessel and so lowers residual organics. This change is not as significant as the mixer configuration, but it does lower the ultimate levels of contamination that can be reached, especially in the absence of the water displacement step.

MDU0315.RP0 March, 1995 CF SVSTEMS A Morrison Knudsen Corporation

4


. Rev. 0 - Page 10

PILOT-SCALE TESTING PROGRAM

The work plan for this phase of the pilot study outlined a total of ten tests to be performed on two soils. In addition, an eleventh test was to be performed on an uncontaminated clay so that dewatering could be demonstrated. An uncontaminated clay was used so that physical testing could be performed on a nonhazardous material. This soil was obtained outside the boundaries of the O'Connor SUperfund Site, thus was representative of a Maine clay.

The fill and clay samples were obtained from suspected "hot spots" within the O'Connor site. Grab samples Were taken at the O'Connor site to verify PCB concentration. The fill sample was screened and sampled. Due to time constraints, the clay sample was left as large lumps; samples were obtained by scraping small amounts of material from the surface. While PCBs were found on the fill material, only trace levels ( < 10 ppm) were found on the clay. Since the clay was in very large agglomerates (>6") and the samples taken from a previously determined hot spot, it was thought higher PCB concentrations would be found once these lumps were broken up. The samples were shipped to Hazen Research for the test program.

'J" Once at Hazen Research, the clay samples were crushed, screened, sampled, and sent out for 24-hour PCB analysis by a modified Method 8080 (Quick Turnaround PCB, or

t QTPCB). The results of these analyses are shown in Table 1. The expectation of Table x QTPCB Analyses of Samples a higher PCB concentration in the clay After Preparation at Hazen sample after crushing and homogenization Research Was not realized. To obtain a clay sample with a significant PCB concentration for

Sample Aroclor 1242 (ppm)the extraction experiments, two parts clay were mixed with one part fill. This Fill 390

mixture Was used as the feed for the Clay 400

Mixture tests.

1.4 1.2

MDU.0315.RP0 March, 1995 CF SYSTEMS A Morrison Knudsen Corporation


. Rev. 0 t Page 11

Five tests were made on each of the two contaminated soils. Process conditions were held constant for all tests:

Solvent Liquefied propane Solvent to feed ratio 1.1:1 - 1.5:1 (weight basis) Number of mixing stages 3 Residence time (per stage) 20 minutes Operating temperature 110°F - 130°F Mixing apparatus 10" Hydrofoil-type impeller

(A-310) Mixer speed 1200 rpm

The first three tests with each material evaluated the effect of weight percent moisture on extraction. For these tests, the soils were processed at the as-received weight percent moisture, air-dried, and slurried. To demonstrate reproducibility of the process, tests four and five with each material were run at the weight percent moisture that appeared to result in the best extraction efficiency as determined by preliminary QTPCB analyses (See Appendix B). The actual process conditions for each test are detailed in Tables 2 and 3.

Two options for the recovery of the treated solids were also evaluated. In Tests 1 and 2 of each series (as-received moisture and air dried), the solids were recovered in a dry form by opening the extractor at the bottom flange and manually digging the solids out. In the remaining tests, a water displacement technique produced a slurry of treated solids in water. This stream was later dewatered to obtain the treated solids.

5 RESULTS

5.1 ANALYTICAL RESULTS

The analytical results for this test program are shown in Tables 2 and 3 along with the process conditions. These results were provided by Quanterra Laboratories in Pittsburgh, PA. GC/MS was used to detect the PCBs; a Soxhlet extraction Was used in the preparation of the samples for analysis. Also shown are the results of QTPCB analyses (using a sonication technique for initial extraction and a simplified extract cleanup technique that results in a 1 ppm detection limit). The feed and treated solids were reported to contain Aroclor 1242. Traces of Aroclor 1260 Were tentatively identified in some of the samples, but were below reporting and quantification limits. All other Aroclors were below detection. cPAHs for the first three tests are also shown in Tables



Rev. 0 - Page 12

2 and 3. Since significant quantities (> 50 ppm) were not detected in either feed or treated solids, cPAHs were not analyzed for in the last two tests on each soil.

In an effort to attain representative samples, all samples were collected by compositing a minimum of four small grab samples from various points and depths in the sample container.

As shown in these tables, the treatment objective of 10 ppm residual PCBs in the treated solids was attained with each of the feedstocks. In the case of the Fill material, percent moisture did not seem to have an effect on the extraction efficiency. All runs, with the exception of the anomalous result found for Test 5, resulted in very close results. In the case of the Clay Mixture sample, the presence of water appears to hinder extraction. Those results strongly suggest that an air-dried feed will respond more favorably to CF Systems' solvent extraction process.

5.2 MASS BALANCE

Mass balance closures for all tests were excellent. In each test, 100 pounds of soil was loaded into the extractor. Over 95 pounds of treated product was recovered in each test. The PCB-laden oil was allowed to accumulate in the extract product tank until all ten tests had been run. At the completion of the final test, 33 pounds of oil was recovered.

5.3 DEWATERING TREATED SOLIDS TESTS

In addition to the ten tests performed to determine extraction efficiency, two tests were made to explore handling of the treated solids after extraction. One of these tests, was performed using the original clay material (unmixed with fill) to ascertain if the clay could be processed by itself. Another run was made on an uncontaminated clay taken from a location near the O'Connor Site so that a PCB-free dewatered cake could be produced for geotechnical testing. The weight percent moistures of each of the dewatered cakes is shown in Table 4.



Rev. 0 - Page 13

Table 2

FILL

Test 1 Test 2 Test 3 Test 4 Test 5

Operating Parameters

Feed Weight (#) 100 100 100 100 100

Solvent Weight (#, per 145 140 110 140 130

stage)

10"-A310 10"-A310 10"-A310 10"-A310 10"-A310Mixing Device

# Mixing Stages

Mixing Time (min per 20 20 20 20 20 stage)

Moisture Adjustment None Dried Slurried Dried Dried

Feed Moisture 13.8 5.3 25.9 6.9 6.8

Temperature (°F) 111-121 103-125 90-118 103-124 117-127

Pressure (psig) 258-274 243-319 212-293 245-305 276-303

Analytical Results

PGBs in Feed (ppm) 640 640 650 590 600 230'

6.12PCBs in Treated Solids 5.7 4.1 5.9 18 (ppm) .29, .41, ND1 12.3

1.7'

cPAHs in Feed (ppm) 7.7 25.4 4.5 NP NP

cPAHs in Treated Solids 1.6 2.9 NP NP NP (ppm)

NP Not Performed ND Not Detected 1 Quick Turnaround PCB Analysis 2 Sample received at Laboratory technically out of holding time 3 Analysis run in duplicate

CF SYSTEMSMDU0315.RP0 March, 1995 A Morrison Knutlsen Corporation


Rev. 0 - Page 14

Table 3

CLAY MIXTURE

CLAY:FILL::1:2

Test 1 Test 2 Test 3 Test 4 Test 5

Operating Parameters

Feed Weight (#) 100 100 100 100 100

Solvent Weight (#, per 145 140 115 140 125

10"-A310 10"-A310 10"-A3l0 10"-A310 10"-A310Mixing Device

# Mixing Stages 3

Mixing Time (min per 20 20 20 20 20

stage)

Moisture Adjustment None Dried Slurried Dried Dried

Feed Moisture 25.2 11.9 31,3 8.9 9.3

Temperature (°F) 101-122 107-119 118-124 105-118 112-123

Pressure (psig) 217-286 276-296 274-317 273-313 292-305

Analytical Results

PCBs in Feed (ppm) 280 390 330 260 42

170' 230, 290' 150'

PCBs in Treated Solids 7.0 1.2 4.3 0.9 0.9 (ppm) 10' ND'

cPAHs in Feed (ppm) 1.7 3.0 2.2 NP NP

cPAHs in Treated Solids 1.2 8.0 NP NP NP

NP Not Performed

ND Not Detected

1 Quick Turnaround PCB Analysis


6


Rev. 0 - Page 15

DISCUSSION

6.1 EXTRACTION

Evaluating the ten extraction tests which were conducted as a group reveals several interesting points. The treatment objective of 10 ppm PCBs was attained at moderate operating conditions (3 stages at 20 minute mixing cycles, a maximum of a 1.5:1 solvent to feed ratio) for all but one run. The analytical result for the run where the treatment objective was not met (Fill Test 5) appears anomalous in th

Documents

GEI Consultants, Inc. › work › 01 › 567890.pdf · 567890 : GEI Consultants, Inc. SOURCE CONTROL : V . SUPPLEMENTAL TREATABILITY TESTING O'CONNOR COMPANY SUPERFUND SITE AUGUSTA,