ichnicol Proposal to Arkansas Department of dilution Control and Ecology )rthe )N-SITE DESTRUCTION OF STORED VASTE ON TNE VERTAC SITE IN ACKSONVILLE, ARKANSAS

ugust13,1987

^ WESTINGHOUSE ELECTRIC CORPORATION Environmental Technology Division iZTECH INC. Westinghouse Power Systems Box 286 Electric Corporation Business Unit Madison Pennsylvania 15663-0286 E P Rahe Jr General Manager Environmental Technology Division

•August 13. 1987 ^ \ A ^

Mr. Paul Mean Director Arkansas Department of Pollution Control and Ecology 8001 National Drive P.O. Box 9583 Little Rock, AR 72209 Subject: Technical Proposal for Destruction of Stored Waste on the Vertac Site in Jacksonville, Arkansas

Dear Mr. Mean:

Vestinghouse Electric Corporation and HAZTECH, Inc. are pleased to submit the attached technical proposal for the on-site destruction of the stored waste on the Vertac site in Jacksonville, Arkansas. The proposal addresses three options for the handling and destruction of all materials stored above ground on the site. Vestinghouse will be the project manager for all work on site. The recommended option is to destroy the liquid or crystalline material using the Vestinghouse Pyroplasma process which is a thermal treatment technology. The system uses a plasma torch which produces Cemperatures between 5000°C and 15,000°C. The most important advantages of this process, relative to combustion processes, are very low emissions, very high destruction efficiencies, and low PIC (product of incomplete combustion) formation. In addition, this system is truly mobile. All of the process equipment is contained in one 48-foot trailer and can be sec up on site within one week.

Ve believe that the Pyroplasaa destruction process is very well suited for your specific waste material. It is estimated that two Pyroplasma units will be required and that work will be completed in approximately one and one half years.

The balance of the waste (solids material, sludges, and debris) will be incinerated using HAZTECH's transportable infrared incinerator, designed and manufactured by Shirco Infrared Systems, Inc. The system consists of a waste prepreparation system, feed metering system, infrared primary chamber, supplemental propane-fired secondary chamber, exhaust gas scrubber, data acquisition and control systems, and heating element power centers, all mounted on transportable trailers. All of these systems are described in detail in the attached proposal.

03107609 •3.1.11; As mentioned in my previous letter, the Pyroplasma system has been selected by the EPA for demonstration under the SITE (Superfund Innovative Technology Evaluation) program. This program provides for the on-site demonstration of innovative technologies where the EPA funds the sampling and analysis portion of a test and provides a written evaluation of the technology. Ve propose that Vestinghouse and the State of Arkansas jointly approach the EPA with the recommendation to perform the required trial burn for the Vertac site under the SITE program.

The second option provides for the destruction of only the 2,4,5-T waste using the Pyroplasma system, and destroying the 2,4,-D waste and all of the solids with the HAZTECH incinerator. This option will apply the unique destruction advantages of Pyroplasma to the waste stream containing the higher levels of 2,3,7,8-TCDD, and reduces the overall program cost. The third option is to destroy all of the material on site using the HAZTECH incinerator.

In all cases, Vestinghouse would be pleased to provide the services of our •— Corporate public relations personnel, at the State's request and at no (^J charge, to assist in any public hearings or meetings.

It should be noted that this letter is a technical proposal with a Ln budgetary estimate. Since we were unable to determine the exact 0 quantities and conditions of the waste to be destroyed, we have made some Q assumptions based on the best available information and have provided cost estimates on a per unit basis. Based on our assumptions, the maximum price quoted is a firm price. Our assumptions are stated. Any contract would be subject to mutually acceptable terms and conditions.

Ve regret that we were not notified of the original request for proposal; however, we sincerely appreciates the opportunity to provide a proposal for the remediation of this site. As you are aware, we also actively pursued this clean-up directly with Vertac. Ve look forward to more detailed discussions with you on this project. If you have any questions please contact Carrie Penman at 412-722-5709.

Sincerely,

^ \^^,^. E. P. Rahe, Jr.

Attachments

TVO-87-292

cc: Steven Cunningham, HAZTECH

03107610 Technical Proposal for the ON-SITE DESTRUCTION OF STORED WASTE ON THE VERTAC SITE IN JACKSONVILLE, ARKANSAS

August 13, 1987

Submitted to Arkansas Department of Pollution Control and Ecology 8001 National Drive Little Rock, Arkansas 72209

Westinghouse Electric Corporation HAZTECH INC. Environmental Technology Division Corporate Headquarters Madison, PA 15663-0286 Decatur.GA 30035-4013

03107611 TECHNICAL DESCRIPTION OF WASTE PROCESSING FOR VERTAC CHEMICAL CORPORATION SITE. JACKSONVILLE. ARKANSAS

PROPOSAL TO THE ARKANSAS DEPARTMENT OF POLLUTION CONTROL AND ECOLOGY

1.0 INTRODUCTION

1.1 SUMMARY OF PROPOSED APPROACH

m This proposal provides three options for the destruction of the waste material currently stored on the Vertac site in Jacksonville, Arkansas. A c0 broad base of equipment and experience in tfestinghouse and HAZTECH allows m us to offer options that enable ADPCE to respond to its particular ^ technical, regulatory, and political needs.

1.11 OPTION 1

Option 1 is the recommended option and provides for the destruction of the majority of the stored waste using the Vestinghouse Pyroplasma system. The liquid or crystalline material which can be made pumpable by heating to 150°F, would be destroyed in a plasma field with temperatures ranging from 5000°C to 15,000°C. This process is a pyrolytic process which means the waste is destroyed in the near absence of oxygen. The most important advantages of this process relative to combustion processes, are very low emissions, very high destruction efficiencies due to the high temperatures produced by the plasma torch, and low PIC (product of incomplete combustion) formation. Vestinghouse can set-up this system on site in approximately one week.

The balance of the waste stored on site in solid fora (solids material, sludges, and debris) will be incinerated using the HA2TECH transportable incinerator designed and manufactured by Shirco Infrared Systems, Inc. This system consists of a waste preparation, feed metering system,

-i- 03107612 infrared primary chamber, supplemental propane-fired secondary chamber, exhaust gas scrubber, data acquisition and control systems, and heating element power centers, all mounted on transportable trailers.

1.12 OPTION 2

The second option is to process only the 2,4,5-T waste material contained in approximately 3300 drums through the Pyroplasma system. This waste has much higher levels of 2,3,7,8-TCDD than the 2,4-D waste. Use of the Pyroplasma system on this waste would provide the unique advantages of the plasma torch, very high destruction efficiencies, low emissions, and low PIC formation, to the more highly contaminated waste stream and lower the '^t' overall waste destruction program costs. The balance of the waste, liquid -. and solid would be destroyed using HAZTECH's transportable incinerator. QQ in 1.13 OPTION 3 o 0

The third option is to destroy all of the waste material through the HA2TECH transportable incinerator.

1.2 SCHEDULE

Vestinghouse and HAZTECH understand that the proposed schedule for this project is as follows:

August 1987 Preliminary proposal assessment Presentation at public meeting in Jacksonville

September 1987 Conduct final contract negotiations

October 15, 1987 Contract placement

December 1987 Mobilization

January/February 1988 Demonstration testing of equipment

March/April 1988 Begin operations

-2- 03107613 Vestinghouse and HAZTECH will have all equipment available per the requested schedule. In addition, Vestinghouse and HAZTECH will make any required personnel available for public meetings/hearings, negotiations, permit applications, etc. to support the program and schedule.

-3- 03107614 1.3 DESCRIPTION OF THE PYROFLA.SMA SYSTEM

The Vestinghouse Plasma Systems mobile liquid waste destruction unit has been developed to destroy liquid organic waste materials by dissociating the organic contents into their atomic components. The Pyroplasma process is based upon the concept of pyrolyzing waste molecules using a thermal plasma field.

The heart of the destruction system is a plasma torch. The system will use 700 to 850 kW of electric power across a colinear electrode assembly to produce an electric arc in a medium of dry low pressure air. The intense energy causes the air stream to be ionized, producing a thermal _ plasma with temperatures in the 5-15,000°C range, waste liquids are (\J injected directly into the plasma where the waste molecules are broken 00 into their atomic states in an oxygen deficient atmosphere. The atoms in o recombine according to chemical kinetics to produce hydrogen, carbon o monoxide, nitrogen, hydrogen chloride, paniculate carbon, and small amounts of carbon dioxide, ethylene and acetylene. The product gas is scrubbed with caustic soda to neutralize and remove acid gas (HC1) and to remove particulate carbon. The remaining gas is drawn off by an induction fan and flared. Each system will process 2-3 gallons per minute or approximately one ton per hour.

r The entire system is process computer controlled. The computer updates temperature, pressure, flow, fluid reserve and other performance parameters, as well as provides continuous on-line monitoring of the process. The system is designed to fail safe even in the event of a total power failure. On board monitoring is installed in the control room to analyze the bulk gas constituents of the pre-flare gas.

The complete unit is contained in one 48-foot trailer. In the three gallon per minute unit, the front of the mobile unit is the laboratory with the computerized monitoring system. The plasma torch, the reaction chamber, and all other process equipment are located in center of the trailer. The power supply and associated electrical equipment are located in the rear of the trailer.

-4- 03107615 2.0 WASTE DESTRUCTION

2.1 WASTE DESCRIPTION

From discussions with the EPA On Scene Coordinator at the Vertac site, the quantity and description of waste to be destroyed or disposed of is as follows:

1. Waste 2,4,5-T Still Bottoms - 3,300 drums

2. Waste 2,4-D Still Bottoms - 25,574 drums plus 194 tanks containing approximately 700,000 gallons p,i CO 3. Contaminated trash and debris - 900 boxes - each containing 1.5 Lr\ cubic yards of material — 0

In Option 1, the Pyroplasma unit will destroy all of the still bottom material which can be made pumpable by heating to approximately 150°F. The trash debris will be destroyed by the HAZTECH transportable incinerator.

For cost estimating purposes, it was assumed that each drum contained 70 percent pumpable liquids and 30 percent solids. All drums or overpacks were assumed to contain a volume equivalent to 55 gallons. Therefore, 39 gallons of material per drum were estimated for Pyroplasma processing. The balance of the material has been estimated for processing in the HAZTECH transportable incinerator.

2.2 DESTRUCTION OF LIQUIDS

Prior to processing any still bottom waste material, the waste will have to be in a pumpable liquid form. A conceptual design has been completed for a solid waste/drum handling system which accomplishes liquification; it is summarized in Section 2.4 and detailed in Appendix A.

-5- 03107616 The current design of the mobile unit has a processing capacity of 2-3 gallons per minute. Two mobile units will be operating two shifts per day, five days per week. At this capacity, the liquid waste will require approximately one and one half years to destroy.

At this time, no provisions have been made for the disposition of the scrubber water or particulate carbon produced by the Pyroplasma unit or the ash produced by the HAZTECH incinerator. Disposition will depend on an EPA ruling on the delisting of materials produced in the destruction of an acutely hazardous material. It should be noted that the volume of scrubber water produced by the Pyroplasma unit is approximately one fifth the volume produced by an incinerator and the Pyroplasma unit does not produce an ash, only some paniculate carbon.

2.3 SOLIDS AND DEBRIS

A variety of solid waste materials and debris may be in the drums. The conceptual design of the drum handling system described in Section 2.4 will provide for the separation of these materials from the liquids. Large solids inside the drum can be removed and repacked or decontaminated with the drum. Disposition of these materials will be handled by HAZTECH in their scope of work proposal which is contained in Appendix B.

2.4 SOLID WASTE HANDLING SYSTEM

A conceptual design has been completed for a solid waste handling system and is detailed in Appendix A to this proposal. This system allows the drummed material to be handled in a manner that minimizes the possibility of environmental release or exposure of personnel. The function of this system will be to heat waste drums to liquify waste, screen out solids, and store liquids for subsequent plasma processing. This system will be used in Options 1 and 2 only. The system will be capable of handling 30 gallon drums, 55 gallon drums (metal and plastic), and 80-85 gallon overpack drums. Some minor modifications to the conceptual design will be made to acommodate the larger capacity overpack drums. The system will be

-6- capable of processing 5500 gallons of waste per day and storing 3000 gallons of liquid waste at 150°F. The solid waste handling system would be located in a concrete-floored, metal-roofed structure supplied by Vestinghouse.

The solid waste handling system is operated on two (2) eight hour shifts per day. Normal daily operation requires eight persons, (4 per shift) and a supervisor. This includes labor to transport drums to the waste handling area, process drums through the system (load and unload the hot room, drain drums, cut drum lids and remove solids), decontaminate the drums, and cut drums for salvage. The process rate for the system is 75 drums per shift. Site services required to operate the system are 120 v, 230 v, and 460 v electrical power, a fork truck, and a drum C<| decontamination area. 00 in 0 0

-7- 03107618 3.0 PERMITTING

Vestinghouse proposes submitting an application for the Dioxin Certificate for the use of the Pyroplasma technology based on data from destruction tests on PCB's. These tests were performed using a smaller capacity (1 gallon per minute) unit in Kingston, Ontario, Canada and were monitored by the USEPA. These data demonstrate the ability of this technology to achieve superior destruction efficiencies for PCB compounds, which are much more difficult to destroy than the Vertac wastes. (These data are also being reviewed by USEPA Region II and New York State Department of Environmental Conservation (NYSDEC) in connection with dioxin-contaminated sludges for the Love Canal remediation project.) Approval of the Vertac site applications based on these data is key to our effort to mobilize by 0 December 31, 1987. Copies of these data are provided in Appendix C to this proposal. * - in 0 Another key factor which will determine the timing of this project will be 0 the timely resolution of the issue of scrubber water disposal. It is essential that EPA approve requests to downgrade the classification of the scrubber water resulting from the processing .of the Vertac waste. In addition, approval will have to be obtained for discharging this water to the Jacksonville sewage treatment system or a suitable alternative identified and committed. Water treatment for removal of metals, etc. will be provided on-site by HAZTECH prior to discharge to any public water treatment system.

-8- 03107619 4.0 BUDGETARY COST ESTIMATE

4.1 OPTIONS 1 - VESTINGHOUSE PYROPLASMA UNIT WITH HAZTECH INCINERATOR

The price of processing all of the pumpable liquid waste through the Pyroplasma system will range from $10 to $13 per gallon.

The price of processing all of the remaining solids through the incinerator will range from $450 to $600 per ton. A minimum of 3000 tons of waste are required for this price. If a minimum of 8000 tons are to be destroyed, the price will range from $200 to $450 per ton.

The price of the solid waste handling facility will be approximately $860,000.

4.2 OPTION 2 - VESTINGHOUSE PYROPLASMA UNIT WITH HAZTECH INCINERATOR

The price of processing the 2,4,5-T waste through the Pyroplasna system will range from $10 to $13 per gallon.

The price of processing all of the 2.4-D waste plus the remaining solids through the incinerator will range from $200 to $450 per ton.

The price of the solid waste handling facility will be approximately $860,000.

4.3 OPTION 3 - HAZTECH INCINERATOR

The price of processing all of the stored waste on site will range from $200 to $450 per ton.

No solid waste handling facility will be required.

-9- 03107620 5.0 VESTINGHOUSE QUALIFICATIONS AND EXPERIENCE

5.1 CORPORATE EXPERIENCE

The Westinghouse Electric Corporation, headquartered in Pittsburgh, Pennsylvania is one of the largest, most diversified companies in the world. Westinghouse, as the the twenty"eighth largest company in terms of sales and operating revenues in the United States, is an extremely strong and financially stable organization. In 1986, sales reached $10.7 billion; Corporate assets new exceed $8.4 billion.

Vestinghouse manages, operates, and maintains more than 100 manufacturing facilities producing over 8,000 products such as xircaloy-clad uranium ^ fuel for ouclear power plants, a full range of electrical generation CO transmission and distribution equipment, people mover systems, and LQ industrial robots. Managing the waste streams that are byproducts of 0 these manufacturing processes is part of the Corporation's daily activities.

Vestinghouse has expanded the Corporate capabilities and has made a major commitment of its corporate resources to develop services and advanced technologies related to hazardous waste management. This commitment is exemplified by the formation of the Environmental Technology Division (ETD) with a corporate charter to provide services and technology to manage the safe disposal of industrial and hazardous wastes. The relationship of the Environmental Technology Division to the overall Vestinghouse Corporate structure is shown in Figure 1.

The objective of the division is to provide state-of-the-art technology with single source project management to solve hazardous and toxic waste problems. Vestinghouse ETD has developed a number of advanced thermal destruction technologies for on-site treatment and destruction. These technologies respond to the major goals set forth in The Superfund Amendments and Reauthorizatlon Act of 1986. Specifically, a faster pace of cleanup nationwide with emphasis on achieving remedies that permanently and significantly reduce the nobility, toxicity, or volumes of waste.

-10- 03107621 005855 The capabilties of the Environmental Technology Division go well beyond just providing technology. ETD's comprehensive services include project management, system development, site remediation, risk assessments, and regulations permitting.

5.2 PERMITTING EXPERIENCE

Vestinghouse ETO has extensive experience and expertise in all phases of the permitting process which include the following: o Defining the total range of permitting and regulatory requirements for new or developing facilities and technologies «=d- m o Defining the total scope of permitting activities for existing 00 facilities LU 0 o Reviewing and assessing operations in existing facilities for compliance with regulatory requirements, including operating permits o Recommending/implementing changes in design or operations for permitting and regulatory compliance

Vestinghouse ETD has experience with many permitting programs in the areas of RCRA, TSCA, CERCLA FERC, Dioxin Certification, Clean Water Act, Clean Air Act, and Corps of Engineer permits. Much of this experience includes the actual preparation of permit applications and subsequent review and negotiations with regulatory authorities.

The Vestinghouse ETD experience with state and local permitting requirements is equally comprehensive. Vestinghouse has developed, filed applications for, and received permits in a variety of areas including: state air quality permits, state solid waste permits, state clean water act NPDES permits, state floodway permits, local construction permits, local highway access permits, and local sewage discharge permits.

-12- 03107623 In particular, Westinghouse ETD has experience in the RCRA RD&D permitting program having submitted an application in September 1986 which is currently undergoing technical review by EPA Region III. In the permit application, Wescinghouse is seeking approval for its three innovative thermal treatment technologies, including Pyroplasma, and to test a variety of hazardous waste materials. The objectives of the Westinghosue permitting activities are technology and product development, not commercial waste processing.

Vestinghouse ETD plans to conduct the RO&D experiments at its Waltz Mill Site in Madison, Pennsylvania approximately 35 miles southeast of Pittsburgh, Pennsylvania. Westinghouse will demonstrate technologies on hazardous waste samples submitted and characterized by potential customers. Sufficient quantities of waste will be used to adequately perform the experiments and collect the required information.

5.3 PROJECT MANAGEMENT EXPERIENCE

Westinghouse has considerable experience in plant maintenance and as the lead contractor/project manager from our many years in the field of reliable power generation. Our capabilities are backed by 100 years of lead contractor experience in the United States and over 40 years of international experience on numerous power generation projects, ranging from small industrial fossil power plants to the largest operating nuclear plants in the world.

Westinghouse Project Management ensures: o A tested organizational structure to implement project plans o Experienced, qualified personnel o Integrated project schedule identifying and scheduling events, milestones and interfaces, and reporting project programs o Encouragement of customer involvement

•13- 03107624 This system reflects the considerable experience gained by tfestinghouse project management assignments relative to energy related programs.

The objective of the Vestinghouse Environmental Technology Division (ETD) is to provide state-of-the-art technology with single source project management to solve hazardous and toxic waste problems. A sample of the Vestinghouse project management experience in the area of hazardous waste management is included in Table 1-1.

-i4- 03107625 TABLE 1 PROJECT MANAGEMENT EXPERIENCE

PCS Sludge Pond PCS Excavation and Decontamination RI/FS and Cleanup- RI/FS and Incineration and Remediation Tennessee Nuclear Remediation Bloomington, IN Decommissioning Sharon,PA Service, Inc. Columbia, SC Trafford.PA Jonesboro, TN

Client Weitinghouie flectrk W«(llnghouc Wtillnghouc Itectrk ATO)*! Ordnance Wettlnghouie Nuclear Corporation, Corporate Environmental Affalrf Corporation, Sharon Company Fuels Division Law Dtpiilment Plant

P»ilodo» 19H-2000(«) 4/K S/M •/•6-12/li 10/1) •2/14 •/•4 • 12/14 Peitormance

Coy Confidential •• S1.000.000 •• S100.000

Contiact Mi Sfphtnr Mr.Paul Jack Mr. Ronald Hair Dr. Stephen Prewatt Mr. Edward l.Slech OKidil Wtidiiniki (4U)64M««1 (412)10-1137 (61S|751-46BI (803)776-2610 (4U)642-SM1

levelOt Plinr (ont(«(t0f Prim* contcctoi Prim* (ontiattof Prime contractor Prlm« contractor Rt»pontibim»

Scope of ll«in«dl«tion o( wry ••mowd 10.000 cubic "l/TS and cmtdlatlon •t S«l«tttv cmoval of PerfMmedRVrSand l«ig« PCB lilt oioupt by (ftolPCB plant lit* wh«r« PC1 12.500 cubic (••t of then removed 1.000.000 Work (hcrmil ptocciting (oniamlnated m*l«rlal contaminafd »oH. contammafd dudg*. gallont of calcium under U.S. court conttnl fiorn plant tit* and rubbi*. ttcl scrap, and fluorlde sludge. d«(i«« adjaccnl properly wat«r wr dealt with. UtHirda unique tpin owninlind (ontrol plan resulting In cott iivingi exceeding S 110.000.

005837 TABLE 1 PROJECT MANAGEMENT EXPERIENCE (Continued)

Technical Study to Technical Study of Develop Waste Isolation Alternatives for Maxey Flats Shrivcr's Corner Alternatives for the Closing Savannah Pilot Plant Project Disposal of Decommissioning Gettysburg. PA River Plant 643-6 Study (WIPP) Depleted Uranium Burial Ground Waste Aiken. SC EglinAFB

Client Wedinghoute 001 Albuquerque Ah fore* Armament 1.1. duPont de Nemourt Kentucky Department tnvironmental Allalri Operation* Office Laboratory AFB •nd Company (or Environmental F'lotection, Oiwitionof Waite Management

rerlod of io«( ii/if 1«71-f>reient 10/14-1/K s«s-««s 7/B3-12/B) Cerfoimarue

Con •• SSO.OOO,000(1«IS) $182.000 S41.000 S100.000

Contract Mr. Paul Jack Mr. 1. G. Hoyl Oi.M Patrick Mi. Jame< Cook Mr. Ooyle MlHi Orticlal (412)6421112 (SOS)146•4)35 «04)IB2-4446 (101)725-1161 (S02)S64-6n«

level of Prim* contractor M«n*g«m«nt •nd Crime contfdof Advlioryftkfofce Prim* contractor Htlponiibility operating contractor

Scope of Conduced several Demonstrated' AtteiteddltpOtal Developed cloture Evaluated and Work remedial action* lor We(tlnghou(e

005858 TABLE I PROJECT MANAGEMENT EXPERIENCE (Continued)

Evaluation of Alternate Design Development of an Disposal Methods of Improved Disposal Trench in Maxey Flats Disposal Site West Valley Radioactivety Fractured Geologic Media Demonstration Project Contaminated Soils, for a Humid Environment Mixed Waste Disposal Cheswick.PA

Client DepJilment of Environment*) Kentucky Department (or DOE Idaho Operation* Office Westinghouse Electro- Protection, Kentucky Niturt) (nviionmenlil Protection, Mechanical Oiviilon Rttouicd and fnviionmcniil Division of Waif Minigement Piofction Cibintt

Period of 1911 Present 1911Present 6/8610/86 Performjnce

Cod S400.000 Avenge UOO.OOO/year

Contract Mr OoylcMilll Mr. Doyle Mills Mr. 0 P. L«fndr« HughWoodion Oltjcial (S02)S64i7(i (502)5646716 (201)S26-1«SO (412)9615012

l«y«lof Prim* Contractor Operating Contractor Sit Manager Prlm« Contractor Ropontibility

Scop* of Developed the deiign of an Based on the tll/FS •watuatlon, Decontamlnatlngi and Packaged uranium improved dilpoul trench sited In a Work constructed an Improved decommissioning former contaminated wastes using fractured geologic media and In a disposal trench. nuclear fuel reprocessing site concrete as a stabiliser. Sludge humid environment. and vitrifying 600.000 gallons of contaminated with uranium and high-level waste for transport oil also stabilized with cement to a federal repository. as an additive.

ItHH^L t

005859 Appendix A

SOLID WASTE HANDLING SYSTEM CONCEPTUAL DESIGN

03107629 1.0 INTRODUCTION

This report describes the conceptual design of a Solid Waste Handling System, to be used in conjunction with two (2) Westinghouse Plasma System Mobile Units, for processing dioxin-contaminated chemical wastes at a Vertac Chemical Corporation facility in Jacksonville, Arkansas.

In general, the Solid Waste Handling System will heat waste drums to liquify waste, screen out solids, and store liquids for subsequent plasma processing.

The report discusses the functional requirements developed for the Solid Waste Handling System, the conceptual design description, as well as cost and schedule information. Included in the attachments are a conceptual drawing of the Solid Waste Handling System (Attachment 1), and a List of Assumptions ^-i. (Attachment 2). CO in 2.0 FUNCTIONAL REQUIREMENTS °

Listed below are 'the functional requirements for the Solid Waste Handling System. These requirements were developed using limited information from Vertac reports and from Westinghouse Thermal Programs. These requirements should be updated as the project progresses. Additional information concerning the waste drum material properties is required before detailed design is initiated. Discussions will be held with ADPCE and EPA regarding approval of the Solid Waste Handling System.

2.1 The Solid Waste Handling System should minimize cost. This equipment will be operated by HAZTECH.

2.2 The Solid Waste Handling System shall be capable of handling 30 gallon drums, 55 gallon drums (metal and plastic), and 80-85 gallon overpack drums. Minor modifications will be made to acommodate the larger overpacks.

-1- 03107630 2.3 The Solid Waste Handling System shall be capable of processing 5500 gallons of waste per day.

2.4 The Solid Waste Handling System shall be capable of storing 3000 gallons of liquid waste at 150*F.

2.5 The Solid Waste Handling System shall handle the two (2) types of waste material listed below:

A. F-020 Waste - 2,4,5-T Still Bottoms. The composition of the chemical constituent is:

1 percent aethanol 8 percent toluene 3 percent Di and Trichlorobenzene 18 percent Dichloro Dimethoxybenzene 56 percent 2,4,5-Trichloro Anisole 7 percent Sodium salt of 2,4,5-Trichlorophenol 7 percent Sodium salt of 2,4,5-T 2,3,7,8-TCDD, 15-50 ppn (average 18 ppm)

B. F-023 Waste - 2,4-D Still Bottoms. The composition of the chemical constituent is:

25 percent 2,4-Dichlorophenozyacetic Acid 20 percent Bis 2,4-Dichlorophenoxyacetic Acid 40 percent 2,6-Dichlorophenoxyacetic Acid 10 percent Orthochlorophenozyacetic Acid 5 percent 2,4,6-Trichlorophenoxyacetic Acid

The flash point of these materials is >140*F. These wastes are solid at room temperature; a thick, heavy liquid at 80*F; and, are similar to motor oil in viscosity/flow properties at 150*F.

03107631 -2- Note: Flash points of individual solvents listed are as low as 40*F. Also, melting point of individual acids listed as high as 300*F.

2.6 The Solid Waste Handling System's interface with the Pyroplasma trailer is a storage tank from vhich the plasma system can pump the liquid waste. The Solid Vaste Handling System shall provide liquid waste at 150*F to the plasma system. The liquid shall be screened to at least 100 mesh.

2.7 Drum decontamination and solid waste decontamination will be performed by HA2TECH.

2.8 Drums shall be cut in half lengthwise for subsequent material salvage. Equipment for drum cutting shall be provided by tfestinghouse or HA2TECH. ,_(- 00 2.9 The wastes included in drums are wastes listed in Requirement 2.5 as well LQ as a small amount of contaminated soil, equipment, and clothing. 0

2.10 Materials of construction for components contacting the waste material shall be similar to those used for the plasma system. This includes 304 SST components, Viton seals, and commercially available equipment.

2.11 Ventilation control is required when heating waste material. The exhaust gases shall not be vented directly into the atmosphere.

2.12 Personnel exposure to waste material and heated areas shall be minimized.

3.0 CONCEPTUAL DESIGN DESCRIPTION

The Vestinghouse proposed design for a Solid Waste Handling System is illustrated on drawing 1743E64 (Attachment 1}. This system is intended for use in conjunction with two (2) Vestinghouse Plasma Systems Mobile Units for processing dioxin-contaminated chemical wastes. These wastes are stored in drums at a Vertac Chemical Corporation Facility in Jacksonville, Arkansas.

-3- 03107632 In general, the Solid Vaste Handling System will heat waste drums to liquify wastes, screen out solids, and store liquids for subsequent plasma processing. Empty drums will be processed HA2TECH.

The proposed Solid Vaste Handling System is economical, flexible, and maintains a high degree of operator safety. Vestinghouse plans to design, construct, and test the Solid waste Handling System. After successful checkout of the system, it will be turned over to HAZTECH for operation.

The system process is a batch-type operation. Vaste drums are prepared and loaded into a heated area. This area will hold approximately 150 55-gallon drums. Drums are heated for 12 hours to 150'F. At this point, drums can be removed from the heated area. The liquid waste is then pumped from the waste '^l' drum, through a screening tank and into heated storage tanks. The storage tanks supply the plasma processing units as required. Emptied drums are QQ processed by cutting drum lids and removing any large solid wastes remaining in iTk the drum. Drums are then transferred to a drum decontamination area. ^ 0 The Solid Vaste Handling System is. located in a concrete-floored, metal-roofed structure supplied by Vestinghous«. The primary components of this system are discussed below.

The hot room is a 30 foot by 37-1/2 foot concrete block structure. The walls, / floor, and ceiling are insulated to minimize heat loss. The room is heated by the post-flare stack gas from the Pyroplasma units and supplemented with three (3) 100 kV electric resistance-type heaters. The auxiliary heaters are required primarily for start-up and maintaining the necessary hot room temperature during off-shifts. These heaters were chosen because they can be purchased for a relatively low cost and for safety reasons. Room temperature is maintained at 150*F to 175T.

Vsste drums are supplied and removed from the hot room through drum sized vail penetrations. This will minimize heat loss during drum changeout. An overhead gantry crane is provided to transport the waste drums inside the hot room. This eliminates the need for manned access into the hot room during normal operations. Crane controls are located outside the hot room and operators have visual access through windows in the hot room walls.

-4- 03107633 The hot room is ventilated to prevent any gas vapor build-up inside the room. Liquid vapors are condensed out of the vent stack.

A small hole (1-2 inch diameter) is placed in the top of each waste drum before heating to prevent any pressure increases inside the drums during heat-up. The small hole also minimizes evaporation from the waste drums. After heating, drums are removed from the hot room and transported to a drum emptying station. At this station, liquid waste is pumped from the drums using a portable, electric drum pump. The liquid is pumped to a screening tank where solids are filtered out. The tank has a series of screens for screening. These screens are manually removable and can be cleaned at the decontamination station. Upon passing through the screening tank liquids are pumped into one of three storage tanks. The tanks are located inside the hot room to minimize heat loss to the atmosphere. All controls for the storage system are located in ^ outside the hoc room. Three 1000 gallon storage tanks provide an eight-hour oo supply of liquid waste for the Pyroplasma units. ^ 0 As drums are emptied, they are transferred to a drum lid cutoff station, where ^ lids are cut using an electric drum cutter. Large solids inside the drum can be removed and repacked or decontaminated with the drum-. If small amounts of liquid remain inside the drums, the drum can be mechanically lifted and drained into the screening tank using the drum handlers.

Decontaminated drums are longitudinally cut in half at the drum splitting station. The cutting can be accomplished by any one of three methods: ozyacetylene torch cutting, portable abrasive saw cutting, or handsaw cutting, depending on specific job requirements.

The solid waste handling system is operated on two (2) eight hour shifts. Normal daily operation requires eight (8) laborers (4 per shift) and a supervisor. This includes labor to transport drums to the waste handling area, process drums through the system (load and unload the hot room, drain drums, cut drums lids and remove solids), decontaminate the drums, and cut drums for salvage. Process rate for the solid waste handling system is 75 drums per shift.

-5- Vertac site services required to operate the system, in addition to labor, are 120 v, 230 v, and 460 v electrical power, a fork truck, and a drum decontamination area.

4.0 COST ESTIMATE

Total estimated cost for this system is $860,000. The estimate assumes that all work is done to Quality Level 4 requirements.

Equipment estimates assure all materials of construction are standard, commercial-type, materials (e.g., carbon steel, stainless steel, standard motors and pumps). No costs are included for environmental monitoring equipment or qualification test monitoring equipment. ^0 •^ 5.0 SCHEDULE 00 in The estimated schedule to design, build, and test the Solid Waste Handling ° System is approximately six months.

-6- 03107635 ATTACHMENT 1

SOLID WASTE HANDLING FACILITY DRAWING

03107636 -7- ATTACHMENT 2

LIST OF ASSUMPTIONS

-9- 03107637 LIST OF ASSUMPTIONS

1. Heating time for one drum of waste Is 12-15 hours.

2. All waste material properties are as listed in Functional Requirements.

3. No explosion proof components are required.

4. Standard components and materials are applicable to this design.

5. A ventilation system is required for area where drums are heated. The vent gases cannot exhaust directly into the atmosphere. 0^ 6. Cost estimate does not include permitting costs or related activities ^ CQ specific to the Solid Vaste Handling System. m 0 7. Drums are first heated and poured as a liquid. Drums cannot be emptied Q with waste in solid form.

-10- 03107638 VERTAC PROPOSAL

Proposal No. 1-708-701

Submitted by: HAZTECH, Inc. 5350 Snapfinger Woods Drive Decatur, Georgia 30035-4027

August ^, 1987

Submitted to: Environmental Technology Division Westinghouse Electric Corp. Waltz Mill Site Madison, PA 15663-0286

-1- 03107639 VERTAC The Westinghouse Corporation and HAZTECH, Inc. will be working together to remove and dispose of hazardous material containing 2, 4-D and 2, 4, 5-T from the Vertac site in Jacksonville, Arkansas. Both companies will use on-site waste destruction units to remove the hazardous constituents in an efficient and cost-effective manner. Materials will be crystalized or liquid phase will be treated using the Westinghouse Pyroplasma Process (WPP). Solid material, sludges, and debris will be destroyed using HAZTECH's Trans- portable Incinerator (HTI). The following sections describe the scope of work and preliminary budget estimates which will be performed by HAZTECH for this effort. Attachment A described HAZTECH's transportable incinera- tor and Attachment B list pertinent corporate work experience.

1.0 SCOPE OF WORK ^ 00 1.1 Site Preparation i^ Prior to commencement of work, site preparation will be required to set up equipment, establish work areas and access, and perform 0 basic operational services and functions. Mobilization of equip- ment, materials and personnel will originate from both the Atlanta, Georgia, and Toledo, Ohio, operation offices. Major equipment will consist of a trackhoe, backhoe, -forklift, a decon- tamination (decon) office trailer, office and supply trailers, a portable water treatment system, and the HTI and WPP units. 1.2 Pro-Construction Areas Concrete pads will be constructed for both the HTI and WPP units. An adequate area will be required to accommodate the waste destruction unit, water treatment system, stock pile areas, and control units. Water, gas and electricity will be required to maintain the units during operation. 1.3 Material Handling Material handling encompasses all hazardous waste material which is to be transported from the storage areas to the waste destruc- tion units. HAZTECH will supply the necessary equipment, material and personnel to transport and stage the drummed waste material, storage tank material, and debris to the waste destruc- tion units. A forklift and trackhoe with drum grappler will physically handle the drums from the storage area to a trailer for transportation to the staging/bulking area. Liquids and crystalized solids (those solids which can be melted) will be staged at the Westinghouse Solid Waste Handling Facility for processing in the WPP unit.

-3- 03107640 ATTACHMENT A

PROCESS ENGINEERING DESCRIPTION

The wastes described in the previous section will be destroyed in HAZTECH's transportable infrared incinerator, designed and manu- factured by Shirco Infrared Systems, Inc. of Dallas, Texas. The system consists of a waste preparation system, feed metering system, infrared primary chamber, supplemental propane-fired secondary chamber, exhaust gas scrubber, data acquisition and control systems, and heating element power centers, all mounted on transportable trailers. A brief description of the infrared incineration process is presented below to familiarize the reader with the general concept of the technology. Specific details of the HAZTECH incinerator design and operation are presented in the following sub-sections. (M General Process Description m CD Figure 4-1 presents a generalized schematic of the HAZTECH tech- ^\ nology. Waste material is first processed in waste preparation equipment designed to reduce particle sizes to dimensions that can be handled by the incinerator. After leaving the waste pre- ° paration equipment, the feed is weighed. Waste material is then fed to a hopper mounted over the furnace conveyor belt. A feed chute on the hopper distributes the material across the width of the conveyor belt. The feed hopper screw speed is used in con- junction with the conveyor belt speed to control the feed rate and bed depth. The incinerator conveyor, a tightly woven wire belt, moves the waste material through the insulated heating modules (primary unit) where it is brought to combustion temperature by infrared heating elements. Rotating rakes gently stir the material to ensure adequate mixing and complete burnout. When the material (ash) reaches the discharge end of the furnace, it is cooled with a water spray. The material is then discharged by means of screw conveyors to an ash hopper. Combustion air is supplied to the primary unit through a series of overfire air ports (not shown in Figure 4-1) located at various points along the length of the chamber, and flows coun- tercurrent to the conveyed waste. Exhaust gases exit the primary chamber near the feed module to a secondary chamber (afterburner), where propane-fired burners are used to ignite any organics present in the exhaust stream, and burn them at a predetermined set-point temperature. Secondary air is supplied to the afterburner to ensure adequate excess oxy- gen levels for complete combustion. Exhaust gases from the secondary chamber then pass through a scrubber type pollution control system (PCS) to the exhaust stack.

-5- 03107641 Weigh and Feed Hoppers Waste will be conveyed into the weigh hopper from the crusher until a previously prescribed weight is reached. At that time feed to the weigh hopper will be stopped and waste will be con- veyed from the weigh hopper to the feed hopper via a belt con- veyor. The feed hopper has a live bottom consisting of six 9 in. screws which will feed the waste in consistent depths across the width of the incinerator belt. Primary Chamber The Primary Chamber is comprised of six powered, one feed, and one discharge module constructed of mild carbon steel. These modules are bolted together and mounted on a skid which has a removable "goose neck" and transportation "dolly" attached for towing to each designated site. Each module is insulated with a 1 in. thick layer of ceramic tr\ fiber blanket and 3 mil stainless steel vapor barrier next to the LH steel shell with additional "Z Block" fiber insulation added as GO the innermost temperature barrier. The interior steel surface of ^ each module is sprayed with stilastic before the insulation is installed to further protect the shell from corrosive volatiles — which might penetrate the insulation at process temperatures. 0 The exterior shell is primed and painted with high temperature resistant paint to provide a durable long-term protective sur- face. The powered modules are fired by transversely mounted silicon carbide resistance heating elements which are insulated from the steel shell with ceramic sleeves. Electrical connections to the heating elements are made by attaching braided steel straps to their aluminized ends with spring tensioned C-clamps. These electrical connections are protected by ventilated wireways. The feed material enters the Primary Chamber through the top of the Feed Module for processing. It drops onto a metal alloy belt which transports the material through the chamber. The belt is supported in the chamber by high temperature alloy top and bottom rollers and shafts which penetrate the chamber shell and are sup- ported by externally mounted bearings. The belt is pulled through the chamber by a chain driven roller system mounted in the *feed module. Chain driven rotating rakes are mounted in the powered modules to stir the feed material periodically and increase process effi- ciency. The rotating rakes are rollers with an array of high temperature alloy "fingers" attached which slowly stoke the material layer on the belt as it moves through the fired zones in the chamber.

-7- 03107642 - . «ti -J

ft ^ ^ u i

B.j I il I

4 __r

03107643 emergency. Under normal operating conditions it would never be used while processing hazardous wastes. The Emergency Bypass Stack is a vertically mounted, rectangular cross-section carbon steel shell which is insulated in the same manner as the Primary and Secondary Chambers. It is sealed at the top with counter-weighted doors which would be opened during an emergency. Scrubber The normal flow of exhaust gases from the Secondary Chamber will be through the base of the Emergency Bypass Stack where the waste gases are split into two separate streams prior to entering the scrubber section. Both streams exit the Emergency Bypass Stack into Inconel tubes where the hot waste gases are cooled with quench water sprays prior to entering the dual fiberglass rein- forced plastic (FRP) Venturis. in Water injected into the venturi throats atomizes and increases LT\ particulate precipitation as the gases enter the front section of oo the FRP packed tower scrubber. The particulates entrapped in ^ water droplets drain into an open blowdown holding area in the bottom of this section. The particulate-free waste gases con- ° tinue into the downstream section of the scrubber where an alka- ° line liquid is injected to neutralize acid vapor in the stream. The neutralized and cleaned gas stream exits the scrubber in a single duct leading to the Induced Draft Blower. Exhaust System The scrubbed gases are drawn from the scrubber by the Induced Draft Blower which propels them up the FRP Exhaust Stack. The exhaust stack is mounted on a pad as a free standing unit with sampling ladder, platform, and ports attached. Control Van The System Control Van is a specially designed unit built to house the controls required to start, run, and shut down all subsystems. The Control Cabinet is located in the rear of the unit and contains all system alarms, annunciators, recorders, Hand-Off-Auto (HOA) switches, process controllers, and process indicators. The Stack Gas Analyzer and solid state belt drive controller are located in the middle section and the Motor Control Center (MCC) is mounted in the front. All electrical power leads and control wiring exit the van through conduit mounted in the flooring.

-9-

03107644 The following HAZTECH jobs were either completed or are currently in-progress for Westinghouse Electric Corp. Information regarding HAZTECH's quality of performance on these jobs may be obtained from the Westinghouse contact listed below. Project #/ Description of Work/ Contact Name/ Location Cost to Client_____ Phone t_____ 2323-87-0755 PCB decontamination J.L. Kaylor Latrobe, PA $525,000 (412) 963-4948 2322-87-0668 Decontamination/ Jan Chizzonite Baltimore, MD Miscellaneous $57,806 2322-87-0666 Trial burn/ Oakridge, TN Miscellaneous $47,400 2322-87-0646 Sampling Jan Chizzonite Jacksonville, PL $2,325 2323-87-0625 PCB sampling and Jan Chizzonite Baltimore, MD analysis $1,900

•ll- 03107645 Specifically, HAZTECH can perform the following services to control such situations and minimize health and environmental risk: o Drum overpacking and removal o Transfer of liquids from deteriorated drums and tanks o Use of fire blankets, fire-suppressant foam o Use of neutralizing, fixation agents o Filtering of pressure-relief gases Because prompt response is essential to proper control of poten- tial releases, the necessary equipment and supplies to mitigate such hazards is stocked at all HAZTECH response centers. Extent of Contamination Surveys HAZTECH engineers and scientists have experience in identifying the nature and extent of contamination and have conducted assessments at more than 300 sites ranging from sampling of con- taminated media to full remedial investigation/feasibility stu- Ln dies. 00 in Equipped with electromagnetometers, resistivity meters, hand Q augers, flame and photoionization detectors, explosimeters, ^ Draeger tubes, and air, soil, waste, and water sampling instru- ments and containers, HAZTECH can evaluate most types of surface and subsurface contaminant conditions. The experience of staff geologists and hydrogeologists in the tracking and modeling of groundwater contaminant dispersion in numerous Region IV projects enhances HAZTECH's ability to effectively contain and/or extract contamination in aquifers serving as water supply. Waste Sampling and Analysis / HAZTECH's sampling capabilities discussed above are regularly applied in the characterization and profiling of wastes as required under the provisions of the Resource Conservation and Recovery Act for proper transportation and disposal. HAZTECH technicians are trained in methodologies for specific and com- posite waste sample collection, documentation, and transpor- tation. To assure reliable analytical data, HAZTECH sends all field samples to reputable, licensed analytical laboratories with experience in hazardous waste analyses and rapid turnaround. When required for the efficient cleanup of a site, HAZTECH will use a mobile laboratory trailer. This trailer will be supplied with all essential equipment and qualified field chemistry staff. Substance Handling and Removal HAZTECH's recovery equipment and experienced equipment operators speed the removal of hazardous materials for transportation and

-13-

03107646 ties nationwide. In addition, HAZTECH has working agreements with various permitted hazardous waste handling, treatment, and/or disposal companies, with capabilities that include sophisticated wastewater treatment, chemical fixation, and PCB destruction. HAZTECH Response Managers, foremen, and technical support staff are experienced in identifying the most cost-effective transpor- tation and disposal alternatives when required. It is the full- time responsibility of at least one Atlanta HAZTECH staff member to maintain up-to-date information on the pricing and availabi- lity of all transportation, treatment and disposal facilities throughout the United States. This information is made readily available to the HAZTECH Response Manager as is assistance in procuring and arranging transportation and disposal. Project Descriptions This subsection presents an overview of all of HAZTECH's projects for the EPA and selected projects for other clients. 00 EPA Projects iT\ 0 HAZTECH has received 111 delivery orders under the current ERCS Q Zone II contract. It characterizes each delivery order as to type of work, media (air, land, surface water, and ground water) and regulatory authority actually governing by the released hazardous substance and the level of subcontracting. Other Projects HAZTECH has performed over 400 projects for private industry and governmental clients other than EPA. The following project sum- maries exemplifying HAZTECH's capabilities in incineration and large tank removal and cleanup jobs.

-15- 03107647 Summary No. 04 Project No. _____173-84-118 Response Mgr. Graham Foreman: G. Garrett

IMMEDIATE REMOVAL AT SWAINSBORO PRINT WORKS

Location: Swainsboro, Georgia Client: USE PA Project Dates: Fall 1984 Total Cost: $350,000

HAZTECH was responsible for sampling and characterization of approximately 350 drums left in an abandoned warehouse at Swainsboro Print Works. Large tanks containing acids, cyanides, epoxys, and caustic soda were also discovered* Drains and trenches running the length of the warehouse were cleaned of all sludges. The materials were solidified and bulked for shipment. Approximately 65 drums of flammable materials were overpacked, and liquids from the tanks were tested and prepared for disposal.

Reference: Mr. Ed Hatcher, USEPA Region IV (404) 347-3931

-17- 03107648 Summary No. 07 Project No. 321-84-162______Response Mgr. Morrow______Foremen: Berquist, Cox, Lee

IMMEDIATE REMOVAL AT SMITH FARM SITE

Location: Shepardsville, Kentucky Client: USEPA Project Dates: Summer 1984 Total Cost: $900,000

With a crew of 25, HAZTECH excavated 5,000 drums buried and scat- 0 sO tered over a 40-acre area at Smith Farm* Mountainous terrain and QQ the extreme heat at the time of cleanup made the task difficult. Ln 0 The drums were staged, sampled and characterized, ultimately 0 requiring transportation and disposal of five different waste streams. The cleanup included liquid compatibility testing and mixing, overpacking of flammable solids, acid neutralization, handling of PCB liquids and solids, contaminated soil excavation, and preparation of the site for future EPA remedial action.

Reference: Mr. Fred B. Stroud III, USEPA Region IV (404) 347-3931

-19- 03107649 Summary No. 020 Project No. 321-86^447 Project Mgr. Neville Kingham

SANFORD PLATING

Client: USEPA

Location: Sanford, NC

Project Dates: Summer 1986

Total Cost: $176,000

HAZTECH was responsible for sampling, characterizing and the cleanup of waste material of a defunct metal plating operation. v0 00 After reviewing analytical results that revealed the presence of .p. cyanide-bearing acidic waste, it was determined the material ^ 0 would require stabilizing before being removed. In-situ treat- ment included pH adjustment and cyanide destruction. The resulting aqueous waste was taken to a waste water treatment plant. The sludge was chemically fixed and taken to a hazardous waste landfill. Because of the dangerous, unstable nature of the cyanide waste, all residents located within one-quarter mile of the site were evacuated during waste treatment and destruction.

Reference: Ned Jessup

(404) 347-3931

-21- 03107650 Summary No. 029 Project No. 322-85-0076 Project Mgr. Graham Supervisors: Johansen, Riffe, Lee ____

TURNKEY SITE CLEANUP AT DRUM BURIAL SITE

Location: Largo, Florida Client: General Electric Neutron Devices Project Dates: Ongoing Total Cost: $120,000 CM HAZTECH, with geotechnical support from S&ME, Inc., provided \Q 00 General Electric with turnkey site assessment and cleanup ser- in vices at an abandoned disposal site near Largo, Florida. 0 0 HAZTECH technicians conducted a magnetometer survey of the suspected burial areas, then installed ground-water monitoring wells to determine if ground-water contamination existed. While ground-water samples were being analyzed, HAZTECH excavated buried drums and other containers located by the initial survey, bulked their contents, and removed the wastes for disposal. This project was accomplished below budget and within the critically short time frame established by General Electric due to the intensive news media coverage given the site.

Reference: Mr. Fred Hoyt, General Electric Company (813) 541-8943

-23- 03107651 Summary No. 040 Project No. 321-86-0400 Project Mgr. Williams Supervisors: Morrow, Gaffga

PCB CONTAINMENT AND CLEANUP

Client: PEI & Associates/USEPA Project Date: Summer, Fall, 1986 Total Cost: $1,300,000

HAZTECH was requested by PEI & Associates to respond under the ^ ERCS Contract to a PCB contaminated site in Detroit's inner city. v£> Contamination on site ranged from 750,000 PPM to 150 PPM throughout a three block area. Immediate containment and 0 0 security was provided while extensive sampling was performed. Office trailers, decon units and mobile laboratories provided work areas for over 25 personnel on site. After sampling was complete, HA2TECH began cleanup operations, which included exca- vation of backyards and lots, decontamination of roads and «• alleys, staging and securing 60,000 cubic yards of waste and water treatment of decontamination solutions.

Additionally, HAZTECH was asked to supply specially designed grappling equipment to allow proper decontamination of scrap metal on site. Packaging of PCB capacitors and installation of surface runoff sumps were also completed by HAZTECH personnel. This site has been listed as one of the worst sites of PCB con- tamination in Michigan.

Reference: Mr. Thomas Wey, PEI & Associates (513) 782-4700

-25- 03107652 Summary No. 052______Project No. 173-83-005 Response Mgr. Cunningham Foreman: Morrow

CAUSTIC CLEANUP

Location: Miami, Florida Client: Hooker Chemical Company Project Date: Summer 1983 Total Cost: $270,000

Hooker Chemical's Miami site housed thousands of pallets of bagged caustic* Many of these bags were in a deteriorated con- dition and there were many resulting spills throughout the ware- house* All ruptured bags and spilled product were drummed with the rest of the bags and repalletized. Both drums and pallets were transported to a new warehouse.

••2.1- 03107653 Summary No. 066 Project No. 173-83-008 Response Mgr. J. Graham

IMMEDIATE REMOVAL AT PINEY WOODS SITE

rocation: Atlanta, Georgia Client: USEPA Project Dates: December 1983 Total Cost: $35,000

In December 1983, HAZTECH was contracted to sample, characterize in and remove approximately 320 drums illegally dumped in southwest \0 00 Atlanta. During the investigation phase, it was discovered that .p, several of the drums were labeled "Dioxin". A mobile laboratory ^ 0 specially designed for instant analysis was brought to the site to determine if the drummed waste did contain dioxin. The drums were staged and overpacked, and composite samples were taken for screening. After verification that no dioxins were present, the waste streams were compatabilized, solidified and removed to a secure hazardous waste disposal facility.

References: Ed Hatcher, USEPA Region IV (404) 881-3931

-29- Summary No. 071 Project No. 321-84-212 Response Mgr. T. Morrow

IMMEDIATE REMOVAL/WESTERN CAROLINA SMELTING

Location: Asheville, North Carolina

Client: USEPA

Project Dates: October 1984 to December 1984

Total Cost: $83,000

HA2TECH was contracted to provide sampling and analysis for v0 approximately 100 drums and for the decontamination of a building vo on site. The cleanup included acid and base neutralization and the cleanup and disposal of drummed material and contaminated 0 0 soil* The presence of concentrated cyanide and acid solutions made working conditions particularly hazardous* All materials were bulked and shipped to a hazardous waste landfill after acids were neutralized and solidified.

Reference: Carol Walsh, USEPA (404) 881-2930

•3i- 03107655 Prepared for

U.S. ENVIRONMENTAL PROTECTION AGtN^Y Hazardous Waace Engineering Research Laboratory Cincinnaci, OH 4522b

Concracc No. 6b-03-3243 Work Aasignoenc No. 6

Technical Project Monitor Dr. C. C. Lee

r- \D 00 LT\ STACK TESTING OF THE 0 MOBILE PLASMA ARC UNIT 0

Final He pore

October 1986

Prepared by

Mark Gollanda Edward Peduco Joanna Hall Howard Schiff

ALLIANCE TECHNOLOGIES CORPORATION (Formerly GCA Technology Division, Inc.) 213 Burlingcon Road Bedford, Hascachuaecea 01730

03107656 NOTICE

This Find Report was fumi«hed Co ch« Environaencal Proeeeeion Agency by che Alliance Technologies Corporation (formerly GCA TechnoloRy Uivition, Inc.), Bedford, H«s««chu«ece« 01730, in fulfillment of Contract No. 68-03-3243, Work Assignment No. 6. The opinions, findings, and conclusions expressed are Chose of che authors and not necessarily chose of Che Environmental Protection Agencv or Che cooperating agencies. Mention of company or produce names is not co be considered as an endorsement by the Environmental Protection Agency. 00 v0 00 in o o

XL

03107657 FOREWORD

The Environmental Protection Agency was created because of increasing public •nd governmental concern about the danger* of pollution Co the health and welfare of the American people. Noxioua air, foul water, and •polled land are tragic testimony to the deterioration of our natural environment. The complexity of the environment and the interplay between if components require a concentrated and integrated attack on the problem*.

Research aud development ia the first necessary step in problem solution; it involves defining the problem, measuring if impact, and searching for solutions. The Hazardous Waste Engineering Research Laboratory develop* new (J\ and improved technology and systems to prevent, treat, and manage hazardous ^Q waste pollutant discharges. This publication is one of the produces of Chat — research. 00 in This document presents information which can be used to assess the 0 feasibility of destroying hazardous waste using a mobile plasma pyrolysis o unii.. Trial burns involving RCRA and TSCA regulated compounds were conducted during which time all environmental release points were sampled and actual release rates quantified.

Thomas VL. Hauler, Director Hazardous Waste Engineering Regional Laboratory ABSTRACT

The mobile plasma arc system developed by Pyrolysis Systems, Incorporated (PS I) underwent •n extensive trial burn program in King*con, Ontario, Canada. The objective* of this program were to evaluate the performance of che •ycea and co establish if destruction and removal efficiency (OK£) capabilities while pyrolying both RCRA aad TSCA regulated hazardous wa«c« feeds. Ihe emissions vere saapled and analyzed for: carbon tecrachloride (CCl^), hydrogen chloride (HC1), polychlorinaced biphenyls (PCBs), polychlorinated dibenzo-p-dioxins (PCDO), polychlorinaced dibenxofurans (PCDF), and parciculace matter. Of notable concern were polynuclear aromatics (PNAa) contained in che scrubber waCer discharge. Samples of the scrubber water o generated during the ays tea operations were also analysed for CCl^, HC1, p_ PCBs, and PCDD/PCDF.

During the CCl4 trial burns, the pyrolysis system aec the established Ln RCRA requireaent with a DR£ of CCl^ of greater than 99.99 percent. Ihe 0 CCl^ emissions averaged 24.98 x 10~^ kg/hr with an average input of 0 63.0 kg CCl^/hr. HC1 eaissions averaged 0.35 kg/hr. N0^ and CO eaissions were 0*35 and 0*14 kg/hr, respectively. CCl^ discharged through che scrubber water averaged only 6.21 x 10~6 kg/hr.

Results of the PCB trial burns indicate that che pyrolysis system destroyed che PCBs at a level of greater chan or equal Co 99.9999 percenc DRJE. PCB scack emissions during che three test runs ranged froa Mot Detected (ND) co 0.11 x 10~6 kg/hr with a system ORE of greater than 99.9999 percent. PCOD and PCDF emissions were in ranges of ND Co U.U28 x lu~6 kg/hr (PCDO) and 0.082 x 10"6 to 0.304 x 10"6 kg/hr (PCDF). HC1 and parciculace maecer emissions averaged 0.0039 and 0.028 kg/hr, respeccively. NO, and CU emissions averaged 0.476 and 0.053 kg/hr, respeccively* PCDUs were not detected in the scrubber water discharge. PCOFs were detected in very small concentrations in only che first test run. PCBs discharged in che scrubber water ranged from NO to 93.1 x 10**6 kg/hr. PNA concentrations in the ppb range were detected in both Che spent scrubber water and the stack gas. CONTENTS

Notice ...... ii Foreword .•.•••••••,.•.••••.••••...••.... iii Abscracc ...... iv Figures ...... vii Table* ...... viii

1. InCroduccion ...... I 2. Summary, Oiacuaaion of Ra«ulCa, «nd Concluaiona ...... J Project Suoaary ...... 3 ,— Carbon Tecrachloride Trial Burn ...... 6 p_ PCB Trial Burn ...... II Concluaioaa ...... 2^ 3. Facilicv Deacripeion ...... ;>0 ln Proceaa DaacripCion ...... JU 0 4. Sampling Location* ...... 34 0 Wacce Feed ...... J4 Reactor Aah ...... 34 Scrubber Water ...... 34 Poscflare Produce Gaa ...... 34 Pre flare Produce Ga« ...... J7 5. Samoling Procedure! ...... 41 Waste Feed ...... 41 Reaccor Hearth Aah ...... 41 Scrubber Water ...... 42 Poscflare Stack Gaa ...... 42 6. Analytical Hechoda ...... 57 Waace Feed and Scrubber Water ...... 57 Poac flare Stack Gaa ...... 3d

03107660 FIGURES

Number Page

3-1 Facility are* cop view ...... 31

3-2 Process schematic of the PSI plaaoa pyrolyic unic ...... j2

.4-1 Plasma pyrolysis system flow diagram ...... 35

4-2 Waace feed aaapling location ...... JO C\J 1"~ 6-3 Poacflare eaiaaion aaapling locaciona ...... JBOO m 6-4 Samolinx point locaciona ...... J9 —

5-1 Continuoua nonicoring aaapling schematic ...... 44 -'

5-2 Volatile organic aaopling Crain achematic ...... 46

5-3 Incexraced xas aaaplinK train ...... 5u

5-4 Modified Method 5 train ...... 52

5-5 Modified Method 5 aaaole point locaciona ...... ^3

5-6 Gaseous HCl sampling craia ...... 55

6-1 Modified Method 5 train organic analyaia flow scheme ...... 59

vn

03107661 CONTENTS (continued)

7. Quality Assurance/Quality Control ...... & 1. Introduce ion ...... ol Project Description ...... ol Project Organization and Responsibilities ...... o- Preciaion, Accuracy, Completeness, Representativeness and Comparability ...... 82 Sampling Procedures ...... o^ Samole Custody ...... 71 Calibration Procedures and Frequency...... 71 Analytical Procedure* ...... 72 Data Reduction, Validation and Beporting...... 72 Internal Quality Control Checks ...... 72 Performance and System Audits ...... 72 Preventive Maintenance ...... 72 Assessment of Precision, Accuracy, and Completeness . • . 72 Corrective Action ...... '. iso^3

References ...... ^" 7u 00 in o o

VI

03107662 TABLES

Number Page

2-1 Stage II, Teat Series I Sampling Parameters ...... 4

2-2 Stage III, Test Series 3 Sampling Parameters ...... 5

2-3 CCl^ Test Burn Schedule ...... 7

2-4 Stage II CCl^ and HCl Emissions ...... 6 ^ r- 2-5 Combustion Parameters • CCl4 Trial Burns ...... 9 QQ in 2-6 PCB Test Burn Schedule ...... 12

2-7 Waste Feed Composition and Feed Race ...... 14 °

2-8 Scrubber Wacer - Volatile Mass Emission Races ...... 15

2-9 Scrubber Water - Semivolatile Mass Emission Races ...... lb

2-10 Scrubber Water - Chlorinated Species Mass Eaission Kaces .... ly

2-11 Poscflare Stack Gas Combustion Data - PCb Trial Burns ...... 2U

2-12 Poscflare Stack G«s Part icu lace and HCI Emissions Uaca ..... 22

2-13 Poscflare Stack Gas - Semivolatile Mass Eaission Kates ..... 2J

2-14 Postflare Stack Gas - Chlorinated Seat volatile Organic Compound Mass Emission Race ...... 24

2-15 Plasma Pyrolysis System Oft£ for PCAs in a Liquid Waste Feed ... 211

5-1 Pro flare and Poscflare bias ion Parameters Measurements ..... 43

5-2 CEM Sampling Parameters and Methodology ...... 45

7-1 Summary of CEM Precision, Accuracy, and Completeness ...... t>J

7-2 Analysis of a Laboratory Control Sample for Cnlorides ...... 04

viii •

03107663 TABLES (continued)

Number

7-3 Analysis of a Matrix Spike Into Scrubber Wacer for Volatile Organics ...... OJ

7-^ Analysis of a Surrogate Spike Into Scrubber Water for Volatile Organic* ...... 60

7-5 Analyst* of Senivolatile Organic Matrix Soiked Compound* (Percent Recoveries) ...... &7

7-6 Analyia of Seat volatile Organic Surrogate Spiked Compounds in Sample Aliquocs ...... oa

7-7 Particulate Hatter Emissiona Correction ...... 70 in ("- CD in o 0

IX

03107664 SECTION I

INTRODUCTION

The U.S. Environmental Protection Agency's Hazardous Waste tegineeriag Research Laboratory (HUEKL), Cincinnati, Ohio, and the New York Dcace Department of Environmental Conservation (NYSDEC) established a Cooperative Agreement in 1982. The Cooperative Agreement called for the construction ana testing of a •obile pilot-scale plasma arc system for the high efficiency destruction of hazardous waste. The capacity of the system ia noainally designed Co be four kilograms (8.8 pounds) per minute, and to fit, with ./-> ancillary equipment, in a 45 foot trailer. The concept of the plasma arc ia chat ic uses very high intensity energy with temperatures approaching lU,OUU*i; co break bonds of hazardous waste chemical molecules down Co the atomic a^ state. The recombination of these acorns results in molecules such aa LT\ hydrogen, carbon monoxide, carbon dioxide and hydrochloric acid. The 0 off-gases from the plasma system are subsequently scrubbed to remove ,->. hydrochloric acid and flared Co remove combustibles.

In general,- the approach taken for this EPA/NYSDEt Cooperative Agreement involves four phases of activity. Implementation of each phase waa subject to the EPA/NYSDEC approval of the preceding phase results. The first two phases were performed in Canada with the cooperation of Canadian federal, Provincial and local authorities; the third and fourth phases will be performed in the state of New York. The total Cooperative Program is configured as followH:

Phase I: • Construction and shakedown of the mobile plasma arc system by the vendor, Pyrolysis System*, Inc. (rttlJ.

Phase II: Performance testing of the plasma system at the Kingston, Ontario test sice (GCA involvement}.

Phaxe III: Transportation, installation and verification of system performance ac a sice in New York State.

Phase IV: Demonstration tests as designated by NYSDEC for permitting purposes at a New York State hazardous waste sice*

The vendor, Pyrolysis Systems, Inc., completed the construction and shakedown phase (Phase I) at the Kingston, Ontario test sice by December 1984. The unit was then suitable for the initiation of the Phase II performance Cests. GCA's involvement began with Che initiation of Phase II sampling activities in February 1985. Several postponements occurred prior Co chia time which further delayed preparatory activities and eventual oobilization or the test crew and equipment. Mobilization and the initiation of Phase II activities were finally accomplished in February 1945.

Phase II consisted of several stages of performance testing. Stage I included equipment operation and shakedown which was performed by PSI prior Co GCA's arrival onsice. Stage II and Stage III were further broken down into several test series which were designed to assess system performance using different waste feeds. Stage II testing utilized carbon cecraehloride (CCl^) as the waste feed component during three 60 minute test. burns (Test Series 1). CCl^ destruction and HCl removal were the primary goals during these performance tests. Stage II testing concluded in February 1985.

Stage III testing utilized an Aakarel waste feea blend to prove cne system's performance capabilities for PCB destruction. Testing under Stage III Test Series 2 was not conducted by GCA. The system was proven acceptable for further endurance fating during the three 60 minute cesca ^ conducted by IMET, a local test company, during the period December l9o5 to February 1986. Stage III, Test Series 3 was initiated and completed duringr"' February 1986 by GCA. These three endurance tests were originally scheduled^) to be 6 hour test burns with the system's waate feed consisting of the AakattfM blend. The first attempt ended prematurely due to a system shutdown. When o the first endurance burn attempt ended after only 115 minutes, the target C^M for the remaining tests was shortened to 4 hours. Only two full term endurance tests were completed during this test series.

Measurement activities during Stages II and III were conducted in order to obtain information OB the effectiveness of the plasma arc, scrubber, and flare systems in destroying or removing certain compounds of interest present in the waste feed. These activities also served to assess the reliability of system components and the stability of destruction performance. SECTION 2

SUMMARY, DISCUSSION OF RESULTS, ANU CONCLUSIONS

PROJECT SUMMARY

The measurement activities were conducted in accordance with che Quality Assurance ProjecC Plan (QAPP, Reference I) which was prepared ana submitted under EPA Contract No. 68-02-3698, Task No. Oil. Any deviations froa this QAPP have been addressed in Section 7 of this report. —.

The primary purpose of Stage II was co deaonstrace destruction r"' capabilicies of waste materials containing regulated chain chlorinated ^ compounds. One test series was conducted and involved three 60 •iouce cests. LP> The only waste feed component was CCl^ contained in an MEK, aechanol, and o water blend. The primary purpose waa Co demonstrate proper nCl removal Q through the scrubber process and the destruction of one of the harder Co destroy compounds (CCl^).

The Stage II sampling and analytical parameters are shown in Table 2-1. The various samples were collected and analyzed onsice. Spent scrubber wacer was scored in one cubic meter Canks pending completion of the.onsice analyses and a demonstration of compliance with the Ontario Ministry ot che t-nvironnenc (MOE) effluent guidelines.

/ The primary purpose of Stage III was Co demonstrate the destruction capabilities of che system using a waste feed concaining more complex chlorinated aromatic compounds, namely Askarel. The Aakarel was comprised of a mixture of three Aroclors and crichlorobenzone wich che balance of the waste feed being MEK and mechanol.

Stage III Test Series 2 consisted of three 60 minute test burns which were monitored by IMET. Tescing was initiated in December 1^05 and was completed by February 1986. Aa GCA had no involvement in these I hour PCS burns, no data from these burns are presented in chis report. i*CA was mobilized and onsice again in February 1986 co begin SCage III Tesc Series 3 which consisced of three 6-hour endurance test burns. Uue to cne possibility of system malfunctions and lengthy delays, che sampling period was shortened co 4 hours. Testing commenced on February 12 and ended February 22, 1966. che first test lasted 115 minutes (I hr 55 ain) before shutting down due to a torch power malfunction. The second test on February 20 lasted for Cbc full 4 hours, as did che third and final burn on February 22. The parameter* measured during Sca»;e III Tesc Series 3 are shown in Table 2-2.

03107667 TABLE 2-1. STAGE II, TEST SERIES I SAMPLING PARAMETERS

Sampling poinc

Pose flare produce K«a 03, CD, CO;, HCl, NO,, CCl^, flue {«« velocity and cuperacure

Spent •crubber veer CCl4 0^ r- Wasce f»ed S«reple and arcnive co R««ccor ash Saaple if «vil«bl« and archive 0

03107668 TABLE 2-2. STAGE III. TEST SERIES 3 SAMPLING PA&AtCKRS

Sampling poinc H«««ur*acQC p«r«a«C«n*

Po«c£l«r« product ga« Oz, CO;, CO, HO,( HC1, p«rticul«c« B«Cter •«aivol«cil««, VOC», PCODfl/PCOFf, PCB«, TCB« flu* g«« velocicy, e«apcracure, •oi«Cur* 0 00 Sp«nc •crubb«r vc«r •«ai-VOC«, VOCt, TCB«, POf, QO PCODt/PCOFa ^

(Reactor ••h) if «vil«bl« 0 0 W««C« £««d PCDD«/PCDF«, PCB«, TCB«

*VOC« • vol«til« orffaie coBpounda PCDDs • polychlorinafd dib«nzo-p-dioxin« PCDFa • polychlorinaeed dib«nzofur«na PCBa • polychlorinaeed biphenylc TCBs • Coeal chlorinated bcnzenei

03107669 The data are reported in essercially two pares. The fine represents data obcained during Stage. II, Tec Series I during which time carbon tecrschloride was the selected principal organic hazardous conscicuenc (POHC). These data include scrubber waste feed, water, and stack gas parameters obcained during che costing. The second data group represents information obtained under Stage III Test Series 3 during whicii an Askarel waste blend was the selected wasce feed.

CARBON TETRACHLORIDE TRIAL BURN

Sampling for CCl4 emissions was conducted during Stage II, Test Series I to determine the overall CCl4 destruction and removal efficiency (ORE) of the system. CCl4 was selected as che principal organic baxardous constituent (POHC) because its very low beat of coobustion suggests it is a difficult material to be ChenBally destroyed, based on ERA'S current rank ing guidance (Reference 2); it is readily available, and it is relatively inexpensive. Consequently, trial bums are frequently conducted using CL'l^ as the POHC. Testing was initiated on February 16, l9o5, after 4 days of plasma arc system preparation and test equipaent sec up. The test achedufe^ was as shown in Table 2-3. The results froa the three completed 60 •inute0 test runs are shown in Tables 2-4 and 2-5• CO in During the three tests, waste feed ratea of 2.82, 2.26 and 2.o3 litex^ per •inute, respectively, were introduced to the system. This corresponda^co •ass feed rates of 64.2, 60.6, and 64.2 kilograms CCl^ per hour (kg/hr), respectively. Scrubber water flow races during these tests were, respectively 33, 30, and 32 liters per minute (L/nin). Stack gas flow rates during these tests were 38.13, 29.69, and 29.81 dry standard cubic meters per minute (m^min):

Waste Feed - CCIA/MEK/MEOH

The waste feed blend of CCl4 and methyl ethyl ketone was introduced at rates averaging 2.82, 2.26, and 2.83 L/min. These correspond co CCl4 mass feed races of 64.2, 60.6, and 64.2 kg/hr. These figures were used in calculating the destruction and removal efficiency (DRS) of the system shown in Table 2-4.

Scrubber Water

Scrubber water samples wre taken and analyzed for CCl4 concentrations which were then combined with scrubber water flow races Co yield CCl4 discharged co the sewer. The concentrations of CCl4 found in the scrubber water were 1.27, 5.47, and 3.26 ppb (ug/L), respectively, for the three l-hour tests. The mass discharge rates are presented in Table 2-4.

Postflare Stack Gas

During the C.Cl^ waste feed burns, samples of the postflare stack gac were obtained and analyzed for HCl and CC^. In addition Co Chese parameters, the acack gaa flow race, cemperature, and bulk gas constituents (0^, CO, OU^ and NO,,) were monitored on a coocinuous basis.

03107670 TABLE 2-3. CCl^ TEST BURN SCHfcDL'LE

Dace (1985) Occurrence Test Duration (ain) Cooacnca

2/16 lac burn 15 Syacea aal function

2/18 2nd burn 60 lac complaca Caac run

2/23 3rd burn 26 iiyacea

03107671 TABLE 2-4. STAGE II CCl^ AND HCl EMISSIONS

^«* ««• i •

Run I Run ; Hii'i I Av-.r.'*

•J«ce. l,9o5 :/18 2/26 ;/:'>

Test duration, ain. 60 60 ^U 60

-«s;e Fttd Pa-ame;

-ai;e F««d CoaooiiCLon CC1<., «aa< : 35 &0 35 36 •(SK/M«OH S5 60 65 64 Specific gravity k»/L 1.08 1.12 1.08 1.0?

•-«*c* F««d Flow Rac«, L/min 2.82 2.26 2.83 2.64 R|/ain 3.05 2.53 3.06 2.88

C::4 F««d Race. •((/•in 1.07 1.01 1.07 1.05 kl/hr 6&.2 60.6 64.: 63.0 ^ Scrubber U«c«r ?»r»m«t

Scack S«( •«riB«cer»

Av«r«(« Flow Rate. •-/•in* 38.13 29.69 29.81 32.54 ft3 /•in- 1.346.3 1,0^8.3 1,05:.7 l,l'.9.1

Avcrif '*iBp«r«;urt "C 908 821 ,642 807 •T 1.666 1.510 1,:77 1.484

HCl Cone.. «»/«3* b 138 2<.7 193 "Cl Sai«iion«, k(/hr H/A 0.:5 0.4- 0.35 Ib/hr N/A 0.35 0.97 0.76

CC1<, Cone.. ppb' e e C:!,. Sai«iea«, k|/hr 29.27xU•s 22.79xl0'6 22.89x10"* 24.»6xl3"6 Ib/hr 6A.39X10"0 50.14x10"* 50.36x10"* 54.96x10"*

Syscm C;l^ DR£, : >»».9» >99.99 >99.99 >99.99

•Ory «tand«rd eondieioni •• defined by 20'C •nd 760 — H|. ''HCl •—plini conducted «c pr«fl«ra loeazioa. S—plia( «u*p«nded due Co carbon plu(gin( of train (not analyfd). 'Remit* are bacd on •»ci«ac«d d«e«eeioe liric of 2 ppb. Actual d«c«e:ion limi.c w in ch« ran{« of 2-15 ppb and waa noc quancif.ad.

03107672 TABLE 2-5. COMBUSTION PARAMETEKS - CCl^ TKLAL BURNS*

Tesc run I 2 3 Average •

Dace, 1985 2/18 2/2o 2/26

Stack Gas a^/min* 38.13 29.09 29.61 32.54 Flowarate fc^nin* 1,346.3 1,048.2 1,052.7 l,l«»9.l

Scack Gas Temperature, "C 908 821 692 607 •F 1.666 1,510 1,277 1.484 <^- N0^ Concencracion, ppo (v/v) 106 92 bl 93 00 Emission Race, kg/hr 0.46 0.31 0.28 0.35 00 Ib/hr 1.02 0.69 U.02 0.7U m CO Concencracion, PDO (v/v) 48 57 ai 62 0 Emission Race, kx/hr 0.13 0.12 0.17 0.14 0 Ib/hr 0.28 0.26 0.37 0.30

0^, percenc 12.7 14. h 15. 1 14.1 CO^, percenc 6.0 5.7 4.9 5.5

*^rv standard conditions as defined by 20*C and 7b0 ma Hg.

03107673 HCl— Durine Run I, hydrogen chloride was sampled at the preflare location («ee Figure 5-6) buC was aborted afcer 20 minutes due Co plugging of Ch« •ample line bv che high carbon and moisture concent of che gas scream. Thi» •ample was invalidated. No further Costing was acceapced ac this locacxon after several flarebacks occurred, causing safety concerns ac this sampling location. The sample train utilized in obtaining the nCl samples is described in Section 4.0. The concentrations of HCl found in the stack gas were 137.7 and 247.2 mg/a^ for Runs 2 and 3, respectively. The results of the HCl testing are further summarized in Table 2-4.

Carbon Tetrachloride— CCl4 samples were obtained from the stack gas using an integrated bag sampling technique whereby a volume of stack gas was drawn into a Tedlar bag at a constant race over a period of time The gaseous samples were cnen subjected to onsice analyses by gas chromatography equipped with.an electron capture detector (GC/ECD). The concentrations of CCl^ present in the stack gas during Runs I, 2 and 3 were all below the detection lioic of tne analytical insCruaenc (less than 2 ppb). For the purposes of this report, ^*pd the establishment of a reportable ORE for the system while fired with a ~_ CCl^-eonCaining waste feed, CCl4 emission races of 29.27 x LU"6, 22.79 x l0~6, and 24.98 x l0~6 kg/hr were used in the calculations. The c0 ORE is calculated using only the stack gas emission rate and does not include CCl^ discharged in the scrubber water. Therefore, the established ORE'S foZ> Runs I, 2, and 3 are all greater than 99.99 percent. Table 2-4 contains BO summary of the stack gas data including the syscem's UK&.

O;, CO^, CO and NO^— In addition to HCl and CCl^. sampling and analysis, combustion parameters were measured in the poscflare stack gas via GCA's continuous emission monitor system (CEHS). Saaples were extracted on a continuous basis bv means of an in-staek probe, filter, and heated sample line and passed through a gas conditioning syscea and valving system to the calibrated analyzers. The gas streams were analyzed for 0^, CO^, CU, and N0^ concentrations. Resultant data (millivolt output) were input aireccly to the Fluke Data Acquisition Svscem and strip chare recorder. The Fluke output waa formatted into test report form and yielded results in ppm, percent, and pounds per hour, as necessary. In addition, stack gas velocity was recorded continuously via a pitoc tube/pressure transducer hookup to the Fluke computer system. Temperature was monitored similarly using a thermocouple/Fluke hookup. All test combustion data are summarized in Table 2-5•

Aa is the case with most combustion sources, NO, and CU are good indicators of combustion Ceaperatures and efficiency. During startup and shutdown operations, the flare is quite unstable due to the nonsceady reactor produce gas supply. This is due largely to the face chac, during startup, the reactor requires a minimum time period to reach equilibrium temperature. Because of the instability upon startup, the system was brought up to operating temperatures on a waste feed containing nonchlorioaced compounds, such as methyl echyl Itecone (MEK). Once online and up to temperature, the chlorinated waste was then introduced. There was usually a slight instability

10

03107674 in che syscem as che new wasce encered che plasma •• may be evidenced by changes in che poacflare stack gas ceaperacures ana coocencraciona ot U^» CO;, and N0^. Generally, once system ceaperacures •caoilized, CO concentracions were relatively constant at levels of less than U. 17 kg/hr. Different types of waste, and even the same waste with slitfiicly varying compositions, seemed to affect varying system responses. As such, although the system could be controlled to operate within a certain range, che repeatability of tests using different waste feeda remains a question. This is due, in part, to the chlorine composition, density, and solida content found in different types of liquid waste feeda.

PCS TRIAL BURN

GCA's involvement during Stage III began with Test Series 3 wherein sampling cook place during the conduce of three endurance • PCH trial burns. The waste feed during these burna was comprised of a blend of three Aroclors, crichlorobenzene, methyl ethyl kecone, and aethanol. Aakarel (Aroclor/triehlorobenzene blend) comprised approximately 25 percent of the waste feed by weight. Teat Series 3 was included in the program Co teac Che vD plasma pvrolysis system over a period of 4 hours while a waste of this cype was introduced. Originally, three 6-hour runs were slated for this test 00 series. However, the target run times were reduced eo four hours in a joint 00 decision by PSI and NYSDCC personnel in an effort Co conserve waste feed and in to complete the sample runs in a timely aanner. This shortened run time was Q also deemed sufficient for allowing adequate detection limits for tne required —. analytical parameters.

During the-first trial burn (GCA Run 3-1, February 12, 1986), sampling ended after 115 minutes (I hr 55 ain.) due co torch power prooleoa within cne pvrolysis system. The aecond and third burns (GCA Runs 3-2 and 3-3, February 20 and 22, 1986, respectively) were each completed after 24U ainuces (4 hours) of sample time. Uuring Run 3-2, sampling was interrupted for approximately 30 minutes due to an offsice power grid loss causing « system upset. The system was brought back on line with the MtK/MEUM waste feed and sampling was resumed 10 ainuces after the switch back to the fCb waste teed blend was made. During all operation of the pyrolysis system, no PCB containing wasce was introduced prior to the system first being stabilized on the MEK/MEOH feed. Additionally, DO sampling was conducted while Che system was solely on the MEK/MEOH feed, in transition to the PCb waste feed, nor during a system upset period. The test schedule for Stage III Test Series 3 was as shown in Table 2-6. The resultant data from these test runs are presented in this section.

During the three tost runs, the PCB waste blend waa introduced to the svscea at rates of 2.10, 2.33, and 2.20 kilogram* per minute (kg/nin), resoectively. The average PCB content of this feed (total, aono through decachlorinaced biphenyls) was 12.6 percent by weight. The total mass rCB input was 0.26, 0.29, and 0.28 kg/min. , respectively, for the three runs. Scrubber water flow races averaged 36.5, 33.0, and 32.5 liters per minute (L/min). Stack gas flow rates were 45.43, 3b.4l, and 35.81 dry standard cubic meters per minute ^m'/rain), for Runs 3-1, 3-2, and 3-3, respectively.

II

03107675 TABLE 2-6. PC8 TEST BURN SCHEDULE

Occurrence T««c duracion (ainJ Cc ienc« l«C burn 115 Run 3-1 Abbreviacec du« Co •yceo ia«l- fuaccion.

2ad burn 240 kun 3-2 incerrupcec for about l/2jir- coopl«c«d 00 3rd burn 240 Run 3-3, •litfKt fan problciu duri^ port ehaoge CLUJ no infrrupcion-^ coapl«c«a.

12

03107676 u^ace Feed - Aakarel/MEK/MEOH

The PCB waste feed blend was introduced at «n average race of 2.21 kg/min wich a PCB "rsa input of 0.28 kg/min or 16.7 kg/hr. This maca input includes mono through decachlorinated biphenyls. Integrated sample* were obtained during each test run from che velving assembly just prior co the feed ring ot che reactor vessel. Ac this point, the waste feed blend was well mixed and representative of chat fed inco che plasma reactor. The aaaples were analysed for eoCal PCBa, ehlorobenzenes, polychlorinated dibenzo-p-dioxina (PCUDs), and pclychlorinated dibenzofurans (PCDFs). Daca on waste feed composition and feed race are presenced in Table 2-7.

Scrubber Water

Scrubber water samples were collecced during eacn test run and analyzed for volatile and seaivolacile compounds including rCbs and CCUUs/rCUFs. During Runs 3-1, 3-2, and 3-3, scrubber water flow race was 36.5, 33.0, and 32.5 L/min, respectively. The scrubber water samples were analysed by (iCA's laboratory and che resultant data are summarized in the following paragrapha.

Volacilea— c0 The volatile compounds found in Che scrubber water were principally 00 benzene, toluene, chlorobenzene, and styrene. Their mass emission races, LO based on their aqueous concentrations (ug/L) and che scrubber discharge race o are shown in Table 2-8. In addicion Co chose compounds listed above, two (—> others were detected albeit at concent rat ions beneath Che dececcion limit of che instrument. These were crans 1,2-dichloroechene in Kun 3-2 and chloroform in Run 3-3. Chloroechane and 2-butanone were also found in Run 3-3 in measurable quantities. The scrubber water sample from Kun J-l was lost due to freezing and breakage of Che VOA vials.

Senivolatiles— Semivolatile components of the scrubber water discharge were sampled by means of compositing samples in a large container during che course of each test run. Samples were split in order Chat a large aliquot could be sent Co Zenon Environmental, Inc. Co conduct the PCB, chlorobenzene, and fCUO/PCUF analyses. These data are presenced and discussed separacely. The remaining aliquocs were transported co GCA's laboratory Co conduce further semivolacila analyses. Generally, che compounds detected and quantified are siscer compounds Co naphthalene and pyrene. They are presented in Table 2-9 wixn their associated concentrations and mass discharge rates. The concentrations presented in this cable are combined aqueous and carbon analyzed oacrxces, aa the samples were two-phased. Generally, che carbon layer had higher concent rat ion a of seaivolacile compounds than the aqueous'phase. In most canes, 'the carbon separated from the aqueous solution, forming a Cop layer with a light, meringue-type consistency. In other samples, tne carbon remained in suspension or gradually settled out over a period of time. This inconsistency in carbon layer formation may be due to varying consistency ot the scrubber water in which the density of the carbon is greater than Chat of the aqueous solution phase. Additional analytical data are presented xn subsequent sections.

13 TABLE 2-7. WASTE PEED COMPOSITION AND FEED RATE

PCB content* Waate feed PCB oasa Dace Run No. percent flow race (kg/ain) inpuc (kg/aiaJ

2/12/86 3-1 14.3'' 2.10 0.30

2/20/86 3-2 U.S6 2.33 0.29

2/22/86 3-3 12.3^ 2.20 0.28 0^ 00 *Total PC3« (oono-decachlorinaced biphenyla). CO in ^FroiB Zenon'a analyses. 0 o

14

03107678 TABLE 2-8. SCRUBBER WTEH - VOLATILE MASS EMISSION RATES

Concen- Scrubber Compound oasa tration veer flow eaisaion race Oace Run f Compound (ug/L) race (L/nin) (10~6 kg/hr)

2/20/86 3-2 1,2-dichloroethene 2* 33.0 4 Benzene 840 1,663 Toluene 41 81 Chlorobenzene 83 164 Seyrene 86 17C

2/22/86 3-3 Chloroechane 23 32.S 45 0 Chloroform 3* 6 Benzene 770 1.502 0^ Toluene 29 57 CD Chlorobenzene 50 98 in 2-bucanone 12 23 o Seyrene 68 133 o

•Component present beneath detection liaita. Reauica ahould be conaidered seaiquantitacive.

15 TABLE 2-9. SCRUBBER WATER - SEMIVOLATILE MASS EMISSION RATES

Scrubber Coocen- WCT flow Compound aas* tracion race «ai««ioQ race Date Run f Compound (U

2/12/86 3-1 Napthalene 21,QUO 36.5 45,9^0 Ace naphthalene 72,000 157.6BU Acenaphchene 240 526 Fluorene I, SOU 3.2H5 Phenanchrene 35.UOO 76,b50 Anthracene 160 35U Fluoranchene 21,000 45,990 P/rene 2l,UOO 45,^0 Benxo (A) Anthracene 1,200 2,b26^- Chry—n« 1.900 ^.H'lQN Benxo (B) Fluoranchene 1.800 3,942^ Benxo (K) Fluoranthene 1,000 2,1^ . Benzo (A) Pyrene 3.200 7.uuaLn Iodeno(l23-CD) Pyr«ne 3,600 7,Bb40 Beazo(CHI) Piryleae 7.700 16,8630 2-H«chylaapCh«lea« 4U ttU

2/22/86 3-2 Naphchalcoe 11,000 33.0 21,7b0 Ace naphthalene 54.000 106,920 Acenaphcheoe 340 673 Fluorene 1.400 2,772 Phenanchrene 28.000 55,440 Anehracene 1,300 2,^74 Fluoranchene 20.000 39.6UU Pyrene 16,000 31,660 Benzo (A) Anthracene 1.000 1,980 Chryene 1,600 3.160 Benxo (B) Fluorantheoe 1.300 2,574 Benxo (K) Fluoranchene 860 1,70.1 Beoxo (A) Pyrene 2,700 5,346 Indeno (123-CD) Pfrene 2,600 5,148 Benxo (Oil) Perylene 5,100. 10,098 2-Methyl naphthalene 300 59<»

(continued)

16 TABLE 2-9 (continued)

Scrubber Concen- veer flow Compound oasa tration race •airioa race Dace Run * Compound (ug/L) (L/ittia)* (l0-<» tg/hr)

2/22/86 3-3 N«phch«l«ne 8.90U 32.5 17,355 Acenaphehal«n« 39.UUU 76,030 Acenaphchene 69 135 Fluoreoe B1U 1,56U PheQaachreoe 17,000 33,15U Anchraceoe 730 1,424 Fluor«nth«ne 12,000 23,400 Pyretic 12,000 23.4UU B«nxo(A)Anchr«ceo« 690 l,3<»6 (M Chryeae 850 l,b5tf Benzo( B)Fluor«ach«ne 960 1.872 0^ B«nzo(K)Fluoraachcne 440 «5« 00 Benzo(A)Pyr«ne 1.800 3.510 m Indeno(123-CD) Pyrene 1,700 3,315 0 B«nzo(GHI)?«rylea* 5,400 10.53U 2-««chyln«phehal«ae 100 195 0

•Serublxr wc«r flowraf obC«ia«d froo PSI via NYSDEC.

17

03107681 PC3s, Chlorobenzenes, PCUDs/PCDFs-- Solic scrubber wacer samples were analyzed by Zenon bnvironoencal, Inc. for PCB, PCDD, and PCUF content. In addicion, Run 3-1 scruober wacer samples were analyzed for chlorobenzenes, chlorophenols, and benzo (a) pyrene. The resultant daca from these analyses are given in Taole 2-LU. The concent rac i.ons given are combined aqueous and carbon phase concencracion* of each compound. As can be seen from che daca in Table 2-10, fCUus were noc dececced in che scrubber wacer in any of the runs. PCDFs w«re dececced in only Che first run and aono through deeachlorinaced biphenyla in Che laac twu runs. It should be noted chat aono aod dichlorinacad biphenyla represenc approximately 89 and 81 percent of the Cocal PCB mass in Huns 3-2 and 3-3, respectiveIv.

Poscflare Stack Gas

Stack gas samples were collected during each run utilizing a variety of sampling trains and methods to obtain Che required parameters. The stack gas constituents sampled for included 0^, CO^, CO, MO^, parciculate matter, HCl, volaciles. semivolacilas,- PCBs, and PCDDs/PCOFs. Also included were ^ •neasuremencs of gas temperature, velocity, and •oiature. Aa seated earlier, 0s Run 3-1 was limited to 115 minutes of sampling time due Co a malfunction of 03 Che power supply Co che torch. During Che Chree test runs, some problems •^»f. arose with che sampling equipment due to che very cold weather causing sample lines and pumps co freeze. Extremely high temperatures wichin che stack ^ created numerous problems especially during Run 3-2 when gas temperatures 0 approached ll50*C (2100*F). Rune 3-2 and 3-3 were sampled Co completion, however, for coca I run times of 240 •inutes (4 hours) each. The case runs and resultant data are summarized and discussed in che following subsections.

Combustion Daca— During Che Chree operational periods in which sampling runs 3-1, 3-2, and 3-3 were conducted, Che postflare scack gas was monitored cor U^, CU^, Co, and N0^ using GCA's continuous emission monitoring syscem (CtMS). These analyzers are frequently used in determining combustion efficiency for diagnostic purposes as well as for determining overall Cu ana/or blUx emission races for regulatory purposes. The emission races are calculated using scack gas flow races and che analysers' responses in concencracioo (pDBi-Dollutanc). The daca are summarized and presented in Table 2-11 in conjunction with scack gas flow races obtained during the semivolacile sampling via Modified Method 5 (MH5).

HCl— Emissions of hydrochloric acid were sampled at the poscflare scack Co determine scack gas concentrations as well aa che HCl mass emission races. Concentracions in the gas scream were quite low during all three runs averaging only 1.68 mg/m3 for an average emission race of 64.1 ag/nin. or 0.0084 Ib/hr. The data sumaary is presenced in Table 2-12 wich Che paniculate emission daca.

18 TABLE 2-10. SCRUBBER WATER - CHLORINATED SPECIES MASS bMIaSION KATES*

Scrubber Cone en- wacer flow Coapouad aass Cracion race «o&aaxon race Dace Run » Compound (Ug/L) tl./nxn)11 U0-<» kg/ttr)

2/12/86 3-1 Di-PenCachlorophenola NU 36.5 NA Benzo(A) pyrene 329 720.1 Tetra-OcCaehlorinaCed dibenzo dioxina NO NA Tecra-Oceach lorinaced dibenzo furans O.U0072 U.OUlb Diehlorobftnzene U.29 U.b4 t;1' Trichlorobenzene 0.20 0.44 ^ Mono-D«cachlorinaced 00 biphenyla NU NA m 0 2/20/86 3-2 Mono-decachlorinaced biphenyla 47.0 33.0 93.1 0 Tecra-Octachlorlnaced dibenzodioxina ND NA •Tetra-Occachlorinaced dibenzo furana ND NA

2/22/86 3-3 Mono-decachlorinaced biphenyla 10.0 32.5 19.5 Tecra-Oceachlorinaced ^ dibenzo dioxina ND NA Tetra-Octachlorinaced dibenzo furana NO NA

• Analytical result* froa Zenon Environmental, Inc.

b Scrubber wacer flow race obcained froa PS I vie NYSUEC.

19

03107683 TABLE 2-11. PUSTPLARb STACK (SA.S COMBUSTION DATA - I'Ul TRIAL BURNS

Run number

Parameter 3-1 3-2 3-3 Average

Dace 2/12/86 2/20/8b 2/22/86

Tesc Duration, Bin. 115 240 240

Stack Ga« Temperature, *C 576 907 671 765 m 1.664 •F 1,070 1,599 1,444 0^ CO Stack Gaa Velocity, •/••c 17.8 20.8 19.5 19.4 ft/ain 3,511 4,090 3,843 3,814 in 0 Stack Gac Flow Race, B^/rin* 45.&3 .16.41 J5.ai 39.22 0 ft^/ain* 1,604.0 1,285.6 1,264.4 1,384.7

Oxygen, percent 15.8 14.U 15..1 15.U

Carbon Dioxide, percent 3.8 5.1 4.3 4.^

Carbon Monoxide, ppo 18 20 20 19 kg/hr 0.057 0.051 0.050 U.053 Ib/hr 0.126 0.112 U.IIO 0.116

Oxidex of Nitrogen, ppa 96 115 l0d l0b kg/hr 0.502 U.4B2 U.445 U.^76 Ib/hr 1.104 1.060 0.979 1.048

'Dry standard conditions defined as 20'C and 760 mm HR.

20

03107684 particulace Maccer— Prior to analyzing the MM5 filters and probe rinses for aeroivolatile compounds, che particulace catch was weighed and uaed in calculating particulace emissions from the stack. The results of the three Celt runs show an average particulace c .-me enc ration of U.OU5 grainu per dry standard cubic foot (gr/dscf) with an average emission rate of 463.2 mg/min or U.061 Ib/hr. Run 3-1 results were almost twice as high ac chose from Run 3-2 or 3-3. During run 3-1, the stack gas tempera cure was auch lower and the stack gas flow race was higher than the two subsequent runs. The system problems which led to a shortening of the test period nay also have caused the increased grain loading (i.e., higher carbon concentrations in the reactor gaa and poscflare stack gas). The data from the three test periods are compiled in Table 2-12 along with HCl emission data.

Volatile Organic Compounds— The postflare stack gas was sampled for volatile organic compounds (VOCs) using a Volatile Organic Sampling Train (VOST). The results of the sampling are not available because the holding times of the samples and the upper ^0 temperature limits for storage were exceeded. The results of toe analyses Q^ would be deemed erroneous because it is unclear what the breakdown components 03 would be after the samples wore allowed to became warm. Extrapolating frum i.n^ the data obtained from the scrubber water analyses, the moat prevalent conscicuenes in the preflare produce gas could be primarily benzene, chlorobenzene, toluene (methyl benzene) and styrene (ethenylbenzene) with ^ boiling points ranging from 80 to 110*C. Ic is escinac«d chat in the poscflare stack x«s, only the higher boiling compounds would be present (i.e., styrene). However, as the PCB concentrations in the stack gas were not detectable, it day be Chat the more volatile species would not have withstood the 900'C stack temperatures and thus, not have been detected.

Semivolatile Organic Compounds— Sampling for senivolacile organics Cook place during each test period using a Modified Method 5 (MM5) sampling train with an XAU sorbenc module in place. Coincident with this sampling was another similar train used for Che collection of semivolacile organics solely for analysis for rCBs, PCUUs, ana PCDFs.

The sereivolacile samples were analyzed by GCA's laboratory utilizing GC/MS. Aa with the scrubber watar samples, the principal coapon«ncs found were naphthalene and its sister compounds. The various cone ancrat ions and resulcanc emission races are shown on Table 2-13. The sampling and analytical methods are described in Sections S and 6, respectively.

Chlorinated Species - VCBs, PCUUs, and PCUFs— A sampling train similar to the one used for the collection of nonehlorinaced semivolacile organic compounds was used co collecc chlorinacau samples Co be analyzed for polychlorinaced biphenyls (PCHs), polychlorinaced dibenzo-p-dtoxins (PCDUs), and polychlorinaced dibenzofurana (PCDFs). These samples were delivered Co Zenon Environmental, Inc. for subsequent analysis following each test run. The resultant data from the three test runs are presented in T

21

03107685 TABLE 2-12. POSTFLARE STACK GAS PARTICULATE ANU HCl EMISSIONS DATA

Run nunber

Parsnfeter 3-1 3-2 3-3 Average

Dace 2/12/86 2/20/86 2/22/86

Test duration, nin. 115 240 240

Stack fi«» now Rice, •-/•in* 45.43 36.41 35.81 39.22 ft^Bin* 1,604.0 1,285.6 1,264.4 1,384.7 r-- Stack Gas Temper*cure, *C 576 907 871 785 0^ •T 1,070 1.664 1,599 1,444 00 in Paniculate Matter 0 Concentration, gr/dsef 0.00692 0.00332 0.00479 0.00500 0 •g/B3 15.84 7.60 10.96 11.47

Eaission Rate, ag/ain 720.0 276.6 J9J.O 4o3.2 ttg/hr 0.043 0.017 0.024 0.028

HC1 Concencracion, ing/n-^ 1.07 2.68 1.29 1.68

Emission Rate, mg/rain0 48.3 97.8 46.3 64.1 kg/h^ 0.0029 0.0059 U.0028 0.0039 Ib/hr0 0.0063 0.0129 0.0061 U.0084

•Ory standard conditions a* defined by 20'C and 760 — Hg. ^WX Concent rat ions and Emission Races during Run 3-1 are the results of I test during the test run. Data fron Runs 3-2 and 3-3 are averages of 3 tests during each test run.

22

03107686 TAB1.E 2-13. POST-FLARE STACK CAS - SRMIVOLATILE MASS EMISSION RATES1'

Stack gas Concentration flow rate Compound Ha•a Enigaion Rate Date Run 1 Compound (ug/«3) (^/nin) (Ib/hr) (mg/m in)

2/12/86 3-1 Naphthalene 46.40 45.43 27.88xl0''5 2.11 Acenaphthalene 37.96 22.81 1.72 Phenanthrene 50.62 30.42 2.30 Fluoranthene 21.09 12.67 0.96 Pyrene 7.59 4.56 0.34 2-Hethylnaphthalene 6.33 3.80 0.29 2-Hethyphenol 8.59 4.56 . 0.34

2/20/86 3-2 Naphthelene 244.42 36.41 117.71 8.90 Acenaphthalene 8.65 4.17 0.32 Phenanthrene 63.92 30.79 2.33 2-HeChylnaphthalene 30.08 14.49 1.10 Dibenxofuran 28.20 13.58 1.03

2/21/86 3-3 Naphthalene 8.92 35.81 4.22 0.32 Acenaphthalene 1.55 0.73 0.06 Phenanthrene 8.14 3.92 0.30 2-Hethylnaphthalene 1.36 0.64 0.05 Dibenxofuran 1.74 0.83 0.06

• Doea not include PCB, PCDD/PCDF iaaa eriaaion ratea - aee Table 2-14.

005898 TABLE 2-14. POST-FLARE STACK GAS - CHLORINATED SEMIVOLAT1LE ORGANIC COMPOUND MASS EMISSION RATE"

Stack gaa Compound masa emission rate C

2/12/86 3-1 Dichlorophenol 88.9 44.69 0.004 0.24 Trichlorophenol 164.1 0.007 0.44 Te C rac h1orophe no 1 74.8 0.003 0.20 Pantachlorophenol 244.0 0.011 0.65 Dichlorobenxene 495.0 0.022 1.32 Trichlorobenxene 385.9 0.017 1.03 Tetrachlorobenzene 233.9 0.010 0.63 Pentachlorobenxene 424.8 0.019 1.14 Dichlorinated biphenyl 39.0 0.002 0.10 TrichlorinaCed biphenyl . 2.7 0.0001 0.01 Mono-dec•chlorinated biphenyl 41.7 0.002 0.11 Hexachlorinated dibenxodioxin 1.4 6.3xl0~5 0.004 Heptachlorinated dibenxodioxin 2.0 8.9x10-5 0.005 Octachlorinated dibenxodioxin 0.6 2.7xl0~5 0.002 Hexa-octachlurinated dibenxodioxin0 4.0 l7.9xl0~5 0.011 Tetr«chlorin«ted dibenxofuran 25.7 ll4.9xl0-5 0.069 Pent•chlorinated dibenxofuran 26.0 ll6.2xl0~5 0.070 Hexachlorinated dibenxofuran 21.8 97.4xl0~5 0.058 Heptachlorinated dibenxofuran 9.6 42.9X10"3 0.026 Octachlorinated dibenxofuran 6.5 29.0xl0~5 0.017 Tetra-octachlorinaced dibenxofuran 89.6 400.4xl0~5 0.240 .————a 05899.. • .. ...— --—— (continued) TABLE 2-14 (continued)

Stack ga« Compound •aas eniaaion rate Concentration flow rate Date Run 9 Co-pound (ng/a3) (.^•in) (•g/«in) (10-6 kg/hr)

2/20/86 3-2 Hono-decachlocinated -• biphenyl ND 36.46 NA NA Pentachlorinated dibensodioxin 0.2 0.7xl0"5 0.0004 Hexachlorinated dibenxodioxin 2.1 7.7xl0-5 0.005 Heptachlorinated dibenxodioxin 4.8 l7.5xl0~5 0.011 Octachlorinated dibenxodioxin 5.6 20.4xl0~5 0.012 Penta-octachlorincted dibenxodioxin 12.6 45.9xl0'5 0.028 Tetrachlorinated dibenxofuran 12.0 43.8xl0"5 0.026 PenCachlorinated dibenxofuran 18.1 66.0xl0~5 0.040 Hexachlorinated dibenxofuran 26.1 95.2xl0'5 0.057 HepCachlorinaCed dibenxofuran 39.5 l44.0xl0~5 0.086 Oceach lorinated dibenxofuran 43.1 l57.lxl0~5 0.094 Tetra-octachlorinated dibenxofuran 138.8 506.1x10"^ 0.304

(continued)

005900 TABLE 2-14 (continued)

Stack gaa Conpound raaa erission rate

• 2/22/86 3-3 Mono-decachi orinated biphenyl ND 34.95 HA NA Penta-occachlorinated dibenxodioxin1' HO HA NA Tetrachlorinated dibenxofuran 4. 5 l5.7xl0~5 0.009 Pentachlorinated dibenxofuran 7. 6 26.6xl0-5 0.016 Hexachlorinated dibenxofuran 7.2 25.2xl0~5 0.015 Hepcachlorinated dibenxofuran 11. 1 38.8xl0-5 0.023 Occachlorinated dibenxofuran 8. 8 30.8xl0~5 0.018 Tetra-octachlorinated dibenxofuran 39. 1 l36.7xl0"5 . 0.082

"Analytical data from report received fror Zenon Environiental, Inc. via Dra. Hugh Dibba (EPS) and Thoraa Barton (PSI).

^etrachlorinated dibenxodioxin not detected in all three runa.

005901 plasma pvrolysis system when firing PCB-con fin ing liquid wees was also calculated for each run and is presented in Table 2-IS. Sampling and analytical methods are described in Seccions 5 and t>, recpeccively.

In calculacing the ORE for PCBa during Runs 3-2, and 3-.), an eacioace of the maximum possible PCS emission race had co be used for Chese runs because the sample analyses yielded results below che instrument dececcion liaics. This estimate uses Che sum of Che minimum detection liaics for mono ehrough de each lor inaced biphenyls. Ic is expected chat the accual PCU emission race is far below ehac calculated using the ainiaum detection liaics. The emission race values are therefore preceded by a "less than" symbol to signify that the value given is a maximum possible eaissioa rate. Therefore, the calculated DREs from Runs 3-2 and 3-3 aay not be representative of actual conditions buc serve Co provide an absolute minimum destruction/removal efficiency value. The calculations used in this determination are as follows:

PCB Congener (N 0 Cl-l Cl-2 Cl-3 Cl-4 Cl-5 Cl-6 Cl-7 Cl-8 Cl-9 Cl-10 (^ Detection limit* 2 2 2 2 2 2 6 4 4 4 m (eocal ng) 0 0 Sum of Detection Limits (Cl-l • Cl-10 PCB) 28 ng

Therefore,

Run 3-2

28 ng 36.46 a3 60 a in KR I.I x 10-8 kg/hr 3 5.54 « •in hr 1 X l012 ng

28 ng 34.95 a3 60 ain kg - I.I x l0~8 kg/br 5.14 a3 •in hr I x l0l2 ng

Where: 36.46 m^/ain • Volumetric flow rate of stack gas during iun 3-2. 34.95 B^min • Volumetric flow rate of atack xas during Run 3-3. 5.54 a3 - Volume sampled by MM5 train during Run 3-2. 5.14 •3 • Volume sampled by MM5 train during Run 3-3.

•Detection limit data obtained from Zenon Environmental, Inc.

27

03107691 TABLE 2-15. PLASMA PYROLYSIS SYSTEM ORE FOR PCBa* IN A LIQUID WSTE PEED

Run duration Uaate feed rate PCB content PCB raaa PCB •«•• Syter ORE Date Run f (win) (kg/«in) (t weight) input (fcg/hr) out (kg/hr)6 percent

2/12/86 3-1 115 2.10 14.3 18.018 O.llxlO'6 >99.9999

2/20/86 3-2 240 2.33 12.5 17.475 99.9999

2/22/86 3-3 240 2.20 12.8 16.896 ^.IxlO'^ >99.9999

^ 'Total PCR« «• cono (I) through dec* (10) polychlorinated biphenyl*. PC8 (•••• out doe« not include PCS •••f discharged through •crubber water. Only atack eaiiaaiona are used in the calculation*. CConcentraciona of PCBa were below the inatru«ent detection lixita according to Zenon*a analyaea. In order Co eatabliah a rininui DRE, the au» of their detection limita for Cl-l - Cl-10 waa uaed to obtain a arxiruw poaaible emiaaion rate.

005905 CONCLUSIONS

Based on che cest results and che operational experience associated with this test program, several conclusions can be drawn. First and foremost, the technology should be created as a promising emerging cecnnology which should be further demonstrated during subsequent trial burn programs.

The notable conclusions which are drawn from the test program are summarized below. These conclusions are focused on the demonstration of an acceptable destruction and removal efficiency as delineated in Che KC8A and TSCA regulations.

• Results from the carbon cecrachloride test burns indicate Chat che system is capable of destroying a "difficult Co destroy" RCKA regulated waste. The OREs from each of che three test burn* exceeded Che minimum RCRA requirement of >^.9^ percenc destruction removal efficiency, '^t- 0 • HCl emission rates conformed to the allowable limits of <4 kg/br and 0^ >99 percenc removal efficiency based on total inlec chlurme concent. in • Concentrations of CCl^ in che scrubber effluenc ranged from ^ 1.27-5.47 UK/L. Effluent levels met the criteria for discharge to 0 Che sewage treatment plane.

• Resulcs from che PCS Cesc burns indicate chat the system i* capaole of destroying a PCB liquid wasce blend consistent wxth cne TSCA requirement of ^99.9999 percent ORE.

• HCl emission races were again consistent with Che requirement ot >99 percent removal efficiency and ^4 kg/hr emission race based on the chlorine input.

• High coneencrat ions of polynuclear aromatic hydrocarbon compounds were detected in the two-phased scrubber effluent* The predomi.nant species were naphthalene, acenaphchalene, phenanchrene, pyreoe, and fluoranchene. Levels were in the range of l2,UUO-72,UUU-J g/L. Correspond ins levels in the flue gas discharge were less than 245UR/B3.

• No appreciable levels of dioxin or furan compounds (a* total tecra through octa) were detected in the scrubber water. In all cases, levels were either nondeeectable or significantly less than I ng/L. Corresponding levels in the flue gas were in the range of 39 - 139 nR/m3 for the total tetra-octaehlorinated dibenzofuran compounds and MD - 12.6 ng/«3 for the Cetra-ocCachloruiafd dibenzo-p-dioxin compounds.

29 SECTION 3

FACILITY DESCRIPTION

The mobile plaama pyrolyaia syacea, operated by Pyrolysis.Systems, Inc., was tested while located on Che grounds of Cbe Royal Military College in Kingston, Ontario, Canada* The •ajor portion of the ay tea waa contained within a 45 foot, apecially adapted trailer, capable of being cranaporced from •ice to aite. Ancillary equipment, aucb aa the power transformer, waate feed blending area, and limited waate atorage facilities were located in a aecu^, contained area within adjacent Building 62. The flare and poaeflare Stacks were located on the opposite aide of the trailer from the building in a f«^rly open area (Figure 3-1). u\ in PROCESS DESCRIPTION 0 0 The PSI plaaaa pyrolyia process ia baaed on the concept of reducing (pyrolysing) waste •oleeulea Co the atomic atate uaing a cheraal plaaaa field. A co-linear electrode aaa—bly ia uaed to produce toe electric arc. Dried, low preaaure air ia uaed aa the medium through which the electric current passes. Air •oleeulea are subsequently ionised forming the plasma field. Upon return to the ground state, the ionised moleculea emit ultraviolet radiation.

Hazardous waste fixtures are injected into the field end interact with the plaaaa field. This interaction results in a reducing mechanism in wnich the molecules are atomised. Upon cooling, ampler •olecules such aa hydrogen, carbon dioxide, carbon monoxide, hydrogen chloride and other ainor matrix compounds such aa acetylene and ethene are formed.

PSI operated the plaaaa system and the online analytical equipment. The online system generated composition data associated with the produce gaa (prior to flaring operacioua). PSI waa alao responsible for providing and preparing synthetic waste feed blends for subsequent tasting. These test blends were identified in the PSI Quality Aaauranee teat plan.

Figure 3-2 ahowa a block diagran of the plaaaa pyrolyais unit which ia the focus of this prograr. The system consists of a liquid waste feed ayscea, pyrolyis reactor, caustic scrubber, flare and online analytical equipment.

Gaseous effluents from the reactor pass through a caustic venturi type scrubber where acid gas removal ia effected. Scrubber water discharga is on the order of 32 liters per minute. Subsequently, the product gas is flared Co

30

•

03107694 BULDINQ 62

BUILDING ENGINEERING OPFRAT10NS AREA SHOP v£) (BIdo. 02) 0 PARKING 0» n. LOT m o PLASMA PYROLYSIS o SYSTEM TRALER

I PREFLARE I PIPE

FLARE STACK AND PLATFORM •o'}

SAMPLING TRAILER

PARKING AREA

BUILDING

H.gur« 3-1. Facility *ra« top view.

31

03107695 • FLOWCNAGRAM

PROCESS«MWtR —— nusHNGnu0 |(VTGASESAonwc

^ CAUSTIC SaUIION n r^ EMERGENCY CWBON n TEK g ^ 1...... NXJCEOOWIMN1^ ^ SCRUlODLK/ n SFfWw ORy . 4«» -rr^- GAS010MNOQWH- ts* I1' i MASsstifciMrruNn EKCTMC.. ^1 WkSrEfEtO oc/ocr KEACIOR LABORMORY ICCTHR- SAMFII ANALYSIS EQUnCNI L pm R GASQAOMMOQWH WAIERSIPWAIW ? (St^HM) l| /^^r——^ ^n i—l^———an.mwh- XRVMER OXXtC)IMftll R <* I' "•-.. =^L4 ) 01SCMARGE IOHRMN

0 CJ ——————— ^ 0 ^3 OS

Figure 3-2. Procesa •cheutic QfQhS ^OpPasia pyrolyals unit. (from Pyrulyula Systuniu, Inc.) complete the cleanup cvcl«. The poat flare gea then encera the acack and i« diacharzed «t che approximate heighc of 8 a«c«r« «bov« grade.

The noqin«l creaciaenc c«p«cicy of Che •yc«n i» 4 kg/ain of vc« feed or approximately 200 L/hr. Produce ga* production race* «c chi« op«r«cing level are on the order of 5-6 n'/ain prior Co flaring operacions.

For the purpoaes of Chia Ceac proxraa, • flare concainoenc cb«aber and • •cack were included Co facilicace ceacing. After coabuacion, Che flue g«c flow race uaa on the order of 36 ••/rin ac acandard condiciona.

GO 0 0 in o o

33

03107697 SECTION 4

SAMPLING LOCATIONS

The locations for collecting che various sample types •re shown in Figure 4-1. The location* remained unchanged from Che original Quality Assurance Project Plan wich one exception. Ac the preflare produce gas sampling location, difficulty wee encountered in obtaining representative aamples due to very high cerbon loading end entrained moisture. Additionally, during •everal system upsets, hydrogen flarebacka occurred creating a sefefy hazard at that location and cauaed sample probea Co be blown out of the PQnc. No further Coating took place at Chat location. 0^ WASTE PEED . IH 0 The waace feed was sampled downstream of the blending and pumping Q operaciona through a valve assembly just prior to entering the reactor feed ring (Figure 4-2). The feed line was under pressure thus enabling an integrated waste feed sample Co be obtained during each test run.

REACTOR ASH

Reactor ash was sampled from the interior of the reactor. Becauae cue entire torch assembly, cooling water jacket feed ring, and graphite core had to be removed to accomplish this, aah samples were cakan only when available.

SCRUBBER WATER

Scrubber water samples were obtained at che discharge point of Che drain hoae aa shown previously in Figure 4-1.

POSTFLARE PRODUCT GAS

The stack ia constructed of 1/4 inch acainlesa sceel place rolled to an I.D. of 16 inches. The flare eoncaiasenc vessel ia similarly constructed, but with a 48 inch to a 16 inch I.D. caper, beginning approximately 21 inches from the baae. The flare head, constructed of a 4 inch stainless steel elbow, protrudes inco the concaionenc veasel where Che reactor produce cases ere ignited. The flare ia self-sustaining and could be ignited either remotely with an igniter fixed in place, or •anually, by placing the ignitor into position until the flare waa lie and then withdrawing it. It waa found that the IsCter was the more reliable as the ignicor waa not continually subjected to the vibration and extreme heat present at che flare head.

34

03107698 POST PLANE PHOOUCT I OAS (4AJ

(4B)

0 c^ in o o

PLASMA AMC HEACTOR

WASTE FCEO HCACTOH HEARTH ASH (2) (1)

WASTE FEED SUPPLY SCRUBBER WATER (3) DISCHARGE

Figure 4-1. Plasma pyrolyaia system flow diagram (abbr»vi*fd to only show aaapling locations).

35

03107699 TO WASTE FROM WASTE FEED FEED RING PUMP (POS. PRESSUR^

VALVE ASSEMBLY

1/4" LINE

TO SAMPLE CONTAINER

Figure 4-2. Waste feed saBpling location. 00591 1 The flare «c«ck waa decigned wich the flare concainaeac veaael to allow for che rapid expamoo of Ch« coabuacad r«accor produce gaa. Th« v««l i« ooe° 4t che ^••e ^ch •n opening area of 1.17 •qu«r« aeeera (12*56 fc~). Thi* opening, in conjunction wich the rapidly heaead gaaea, allow for « ficne and «cack gaa buoyancy which aliainacea the Q«ed for induced draft or forced drafc f«°« In •Pic* of the lack of • fan, once che flare ia lie and eaaparacuras h«ve gcabilized coaewhac, the ga« flow racaa through Cha icack ,re alao relacivaly •c«bl«. Th«r« !• che presence of Ceaperacura apikaa •c eiae* which if uaually aCCribueed Co r«-«ntraio«d carbon •c Ch« fl«r« h««d or eb«nxin8 bydrox«n eoncenC of Kh« reactor produce x««.

Po«tflar« produce ««• •«•?!•• wr« takan froa ewo loc«cion« down«cr*«a of the flare. Th« flow diaxr— in Figure 4-1 •how ebcr locacionc in r«l«cion co 6h« r««c of eh« proc«c«. Figure 4-3 illuicraf cb« po«cfl«r« •Cack confixuracion •nd •—pling port location* wich cb«ir •••ociacftd ••aauriunc*. It •hould R« noc«d that, although ch* a«apling location for Cha MH5 traiaa •aciafiad Cha 8/2 critaria for laminar flow, Cha aaall diaaacar of Cba •cack (16 inchai), in conjunction wich Cha nuabar and Cypaa of aaapling probaa raauirad for Cha progr—, ia noc eooduciva to obtaining aecuraf flow CM •aaauramanea. Tha high taaparaturaa found in Cba •tack oacaaaieatad Cba uaa .— of wacar coolad probaf for Cha MM5 traina. Two HH5 eraina wara raquirad by 0^ cha Adainiaeraeor to ba run •iauLtanaoualy in ordar to provida aaparaca pCB/Dioxin and •amivolacila aaaplaa. Tha blockaga eauaad by chaaa prob«a LP> alona approaehaa 22 parcane. Furtbar flow diacurbancac within Cba atack could ^ hava baan eauaad by Cha ocbar in-ataek probaa aituacad 72 inchaa upaera— of 0 cha wacar coolad probaa. Howavar, in •pica of Cba blockaga and poaaiola flow disCurbancaa, cha valocicia* and flow raca* •aaaurad by boch eraina during all chraa Ca«c runa.wara in cloaa agraaaanc (wichin 5 parcenc). Tha •f'spling poinca for all eraina ara illuacraead in Figura 4-4.

Tha HCl and VOST aaapla probaa wara locacad 3.7d diaaaeara downacraaa of cha flare concaicnane vacaal. During Cha CCl^ burna, Che CEM probe and filear wara eolocacad wich eha HCl probe, aa Chara wre no VOST runa requirad during Seage II, TaaC Sariaa 1. The CCl^ ineagracad bag aaapling aycaaa were aae up on Cha aaapling placfom and aaapling waa eonduecad froa one of che upper porea.

During Taac Sariaa 3, Che two HH5 eraina, aa well aa eha CEH probe and filCar hocbox, wara aec up ac che upper porta, 8.28 diaoecera dovnacraaa of Che flara eoaeainBaoe vaaaal* Tha CEM proba waa bane ac a 45 degree angia in ordar Co aliainaca incarfaranea wich eha (015 aaapling probe in port A. The CEM probe cio reaained on eha a—e horixoncal plan aa Che Htt5 noxslaa.

PREFLARE PRODUCT GAS

Aa wncionad aarliar, aaapling afforca ae Chia locacion were abortad due Co unsafe aaapling eondieiona and very high carbon loading and ancrainad (ooiacure. The acainleaa aceal preflare pipe axica che Crailar approxiaacaly eight (8) faac off che ground, cakac a downward band, and ehen runa along Cba ground approximacaly ewency (20) feec before Caking a 90* upward bend Co fora

37

03107701 ys"

MMS SAMPLING PORTS

CEM SAMPLING so- PORT

m VOST/HQ SAMPLING PORT ^~ o^ rTACX 10.* 1« MOCS m o NEAREST UPSTKEAM OISTURaANCE: o «^» OtAMETCBS

NEAREST DOWNSTREAM DISTURBANCE: 4JMOIAMCTCM

-Merrs a/2 CWTERIA

T2"

FLARE

Figur* A-3. PoBtflare eaifion ««aplin( locaciorr.

38

03107702 Project Suanury

STACK TESTING OF THE MOBILE PLASMA ARC UNIT

October 1986

Prepared by

Mark Gollands Edward Peduco '=1 Joanna Hall •c— Howard Schiff (^ in o o ALLIANCE TECHNOLOGIES CORPORATION (FORMERLY GCA TECHNOLOGY DIVISION, INC.) Bedford, Ma«cachu««cf 01730

Contract No. 68-03-3243

Technical Projact Monitor

Dr. C. C. Lae

U.S. ENVIRONMENTAL PROTECTION AWNCY Hazardous Waste Engineering Research Laboratory Cincinnati, OH 45226

03107703 ABSTRACT

The Plasma Arc System developed by Pyrolysis Sycems, Incorporated trs>i^

nderwenc an extensive crial burn program in Kingston, Oncario, Canada. The

lasma arc reactor with ancillary equipment was designed as a mooile unit for che highly efficient destruction of liquid hazardous waste. Produce gases

froa the pyrolysis reaction are scrubbed and then flared for che final destruction of any remaining hazardous constituents.

The objectives of this program were to evaluate Che performance of the system and to establish its destruction and removal efficiency (ORE) while pyrolyzing both RCRA and TSCA regulated hazardous wastes. In February 1965, the system was operated over a two week period while introducing a liquid in waste containing CCl,* This effort was essentially repeated in 0^ in February 1986 when a liquid waste containing PCHs was introduced co the Q system. The emissions were sampled and analyzed for: carbon cecrachloride 0

(CCl,), hydrogen chloride (HCl), polychlorinaced bipheayls (PCSs), polychlorinated dibenio-p-d toxins (PCOD), polychlorinaced dibenzofurans

(PCDF), and pareiculace —ccer. Of notable concern were polynuclear aroaatxcs

(PNAs), oxides of nitrogen (NO ) and carbon monoxide (CO) emissions.

Samples of the scrubber water generated were also analyzed for CCl., UCl,

PCBs, PCDD/PCDF, and PNAs. The system was evaluated during two separate efforts.

During the CCl^ trial burns, the pyrolysis system met the established

RCRA requirement with a ORE for CCl4 of greater than 99.99 percent. The

CCl, emissions averaged 24.98 x 10 kg/hr with an average input of

63.0 kg CCl,/hr. HCl emissions averaged 0.35 kg/hr. NO and CO emissions were 0.35 and 0.14 kg/hr, respectively. CCl, discharged through the scrubber water averaged 6.21 x l0~ kg/hr.

03107704 Reaulcs of the PCB trial burns indicate that the pyrolyis •ycea

•croved the PCBs to « level of greater than or equal Co 99.9999 percent

-HE. PCB stack eniaaiona during the three te«c run* ranged fro® Not Detected —6 (NO) co 0.11 x 10 kg/hr with a ay tern ORE of greater than 99.9999

ptt-cenc. PCDO and PCDF •tack eaiaaiona were in the range of NO to

0.028 x 10'6 kg/hr (PCDO) and 0.082 x l0"6 to 0.304 x l0'6 kg/hr

(PCDF). HCl and particulace •atter —iaaiona averaged 0.0039 and O.U28 kg/hr, refpectively. NO and CO euiaaiona averaged 0.476 and 0.053 kg/hr, recpectively. PCUD« were not detected in the •crubber water diacharge; howevec, PCDFs were detected in very aaall concentration* in only the first test run. PCBs diacharged in the •crubber water ranged froa NU to

93.1 x 1l0~0 "6 kg/hr and PNA concentration! in the ppb range were detected in u^ 0 both the •tack gaa and the •crubber water. 0

03107705 ^TKOOUCTIW

The U.S. Environmental Protection Agency's Hazardous Waste Engineering

—,e,rch Laboratory (WERL), Cincinnati, Ohio, and the New York Stace

Department of Environmental Conservation (NYSDEC) established • Cooperative

Agreement in 1982. The Agreement called for the construction end tenting ot a

•obile pilot-scale plasma arc system for the high efficiency destruction of hazardous waste. The capacity of the system is nominally designed Co be four kilograms (8.8 pounds) per minute, and to fit, wich ancillary equipment, in a

45 foot trailer. The concept of the plasma arc is Chat it Uses high intensity c- energy with temperatures approaching 10,000'C to break bonds of hazardous 0\ waste chemical molecules down to the atomic state. The recombination of these ^0 0 atoms produces molecules such as hydrogen, carbon monoxide, carbon dioxide and hydrochloric acid. The off-gases from the plasma system are subsequently scrubbed to remove hydrochloric acid and flared to remove combustibles* A multi-stage sampling program was designed to characterize and quantify emissions from the plasma are unit in addition to establishing its destruction and removal efficiency capabilities.

In general, the approach taken for this EPA/NYSDtC Cooperative Agreement involves four phases of activity. The first two phases were performed in

Canada with the cooperation of Canadian Federal, Provincial and local authorities; the third and fourth phases will be performed in the State Of

New York.. The total Cooperative Program is as follows:

Phase I: Construction and shakedown of the mobile plasma arc system

by the vendor, Pyrolysis Systems, Inc. (PS I).

03107706 Phase II: Performance testing o£ the plasma lyacem ac Che Kingston,

Ontario cesc •ice (t»CA involvement/.

Phase III: Transportation, installation and verification of system

performance at « sice in New York Stace.

Phase IV: Demon•CraCion tests as designated by NYSDEC for permitting

purposes at a New York State hazardous waste sice.

CO Project Objectives <(— (^ in 0 The primary objectives of the program were Co demonstrate the destruction _ capabilities of the system on waste materials containing chlorinated compounds. The first aeries of teats in Phase II used carbon tetracbloride as

the chlorinated feed coBponent and as the target compound in the analyses.

"This series was essentially a preliminary test run which could prove the / performance of the system on a "difficult Co destroy" compound with a very low

heat of combustion. Upon the successful demonstration of compliance with

state, provincial, and Federal requirements, the next Cesc series involved

introducing an Aakarel blend waste feed containing 12-14 percent

polychlorinated biphenyls (PCT) by weight.

Phase II measurement activities were conducted in accordance with the

Quality Assurance Project Plan. The purpose was Co obtain information on the

effectiveness of the plasma are, scrubber, and flare systems in destroying or

romovine the compounds of interest in the waste feed and Co assess the reliability of avstem components and the stability of destruction performance. The parameters measured in each of the two test series are shown in Tables I and 2.

TABLE \. CARBON TETRACHLOR1DE SAMPLING PARAMETEKS

Sampling point Measurement parameter

Postflare product gas (>2, CU. CO^, UCl, N0^, CCl^, flue gas velocity and temperature

Spent •crubber water CCl^

Waste feed Sample and archive

Reactor ash Sample if available and archive

TABLE 2. PCB SAMPLING PARAMETERS

Sampling point Measurement parameter a*

Postflare produce gas O;, CO^, CO. N0^ HCl, paniculate matter semivolatiles, VUCs, PCDUs/rCUFs, PCUs, TCBs flue gaa velocity, temperature, •oiacure

Spent scrubber water semi-VOCs, WCs, TCBa, PCBs, PCDD./PCDFs

(Reactor ashJ if available

Waste feed •PCDDa/PCDFs, PCTa", TCBa

^voc* • volatile orRanic compounds PCODs • polvchlorinated dibeneo-p-d toxins PCUFs • polychlorinated dibenzofurans PCBs • polvchlorinaced biphenyls TCBs • tocal chlorinated benzenes

03107708 Teat Facility

All Phase II sampling was conducCed on Che grounds of Cbe ttoyal

Military College in Kingston, Ontario, Canada. The mobile plasaa pyrolysis system, as shown in Figure I, was operated by Pyrolysis Systems, Inc. (PSI).

The plasaa reactor and ancillary equipment were housed in a 45 foot, specially adapted trailer, capable of being transported froa sice to site. PS»I was responsible for providing and preparing Che synthetic waste feed blends for che subsequent testing. Table 3 outlines the waste feed paraaeters froa the o (N two test series. in 0 CARBON TETRACHLORIOE TRIAL BURN 0

Sampling for CCl, emissions was conducted during Stage II, Test

Series I, to determine the overall CCl, destruction and removal efficiency

(ORE) of the system. CCl, was selected as the principal organic hazardous constituent (POHC) because its very low heat of combustion suggests ic is a difficult material to be thermally destroyed, based on EPA's currrent ranking guidance (Ref. I); it is readily available, and it is relatively inexpensive.

Testing was initiated on February 16, 1985, after four days of plasaa arc system preparation and test equipment set up. Results for the carbon tecrachloride and hydrogen chloride gas test runs are presented in Table 4.

The waste feed blend of CCl^, methyl ethyl kecone and methanol was

introduced at rstes averaging 2.82, 2.26, and 2.83 liters per minute (L/min).

These correspond to CCl, mass feed rates of 64.2, 60.6, and 64.2 kilograms

per hour (kK/hr). These figures were used in calculating the destruction ana

removal efficiency (ORE) of the system shown in Table 4.

6

03107709 FLOW DIAGRAM

OT GASES TO »tWE

EMERGENCY CARBON FUtER

SrnilMIR WUER DISCHARGE IOUWN

Figure 1. Process schematic of the PSI plasma pyrolysia unit (from Pyrolysis Systems, Inc.) 005921 TABLE 3. WASTE FEED PARAMETERS

Parameters ' Kun I Kun 2 Run J Average carbon tecrachloride burns-2/85

CCU, aas« X 35 40 35 36 Feed flow rate, L/min 2.82 2.26 2.83 2.64 Specific Gravity, kg/L 1.08 1.12 I.Ob 1.09 CCU feed rate, mg/min 1.07 1.01 1.07 1.05 kg/hr 64.2 60.6 64.2 63.0 pCB Burnt - 2/86

PCB, mass Z 14.3 12.5 12.a 13.2 Feed rate, ltd/rein 2.10 2.33 2.20 2.21 PCS feed rate, kg/min 0.30 0.29 U.28 0.29 ItR/hr 18.0 17.4 16.8 17.4

Scrubber water samples were taken and analyzed for CCl, concentrations which were then combined with scrubber water flow rates Co yield the CCl,

•ass discharged to the sewer. The concentrations of CCl, detected in Cne scrubber water were 1.27, 5.47, and 3*26 ppb (ug/L), respectively, for the three l-hour tests. The mass discharge races are presented in Table 4.

During the CCl, waste feed bums, samples of the post flare stack gas were obtained and analyzed for HCl and CCl,. In addition to these parameters, the stack gas flow race, temperature, and bulk gas constituents

(0 , 00, 0). and NO ) were monitored on a continuous basis.

During Run I, hydrogen chloride gas (HCl) was sampled at the pro flare location but was aborted after 20 minutes due to plugging of the sample line by the high carbon and moisture content of the gas stream. This sample was invalidated. No further testing was attempted at this location after several » flarebacks occurred, causing safety concerns at this sampling location. The TABLE 4. CCl^ TRIAL BURNS - CCl^ AND HCl EMISSIONS

• & *^

Run 1 Run 2 Run 3 Average

Dace, L9B5 2/18 2/26 2/26

Tesc duration, min. 60 60 60 60

Uasce Feed Parameter*

•Ja«ce Feed Coapoiicion CCl^, •••• t 35 40 35 36 MEK/MtOH 65 60 65 64 Specific gravity k»/'L 1.08 1.12 1.08 1.09

Uasc* F««d Flow !Lac«, L/nin 2.82 2.26 2.83 2.64 kg/min 3.05 2.53 3.06 2.88

CCl(, Fed Race, kg/rin 1.07 1.01 1.07 1.05 kg/hr 64.2 60.6 64.2 63.0 K^

Scrubbier Uacer Parameter* CM 0^ Discharge Flow Uce, in L/ni.n 33 30 32 32 CCl^ Conc«ncracion, 0 ppb tug/L) 1.27 5.47 3.26 3.33 0 CCl^ Oitcharge Kac«, -6 lig/hr 2.51x10 9.85x10"* 6.26x10"* 6.21x10"* Ib/hr 5.54xl0"» 21.71x10"* 13.80x10"* 13.68x10"*

Scack Ga* Parameter*

Average Flow Raf, •^/•in* 38.13 29.69 29.81 32.54 ft3 /•ina 1,346.3 1,048.3 1,052.7 1,149.1

Av«r«(e T—peracure *C 908 821 692 807 •F 1,666 1,310 1,277 1.484

HCl Cone.. •i/a3* b 138 247 193 HCl Eairioo*, k(/hr H/A 0.25 0.44 0.35 Ib/hr II/A 0.55 0.97 0.76

CCl4 Cone., ppb' e e CCl4 Cai««ion«> kg/hr 29.27xl0~* 22.79K10"6 22.89x10"* 24.9»xl0~* Ib/hr &A.3»xlO"* 50.14x10"* 50.36x10"* 54.96x10"*

Sye«« CClfc ORE, Z >99.99 >»9.99 >99.99 >99.99

'Dry icandard condicion* a* defined by 20*C and 760 — H(. ''HCl •aaplin^ conducted aC preflar* location. S—pliat luapendcd du« Co carbon plugging of train (noc analyfd). 'Result* are based on escraced deceecion licic of 2 ppb. Aecual d«C«ccion limic wa« in Ch« range of 2-15 ppb and was noc quantified.

03107712 concentrations of HCl found in the stack gas were 137.7 and 247.2 milligram per cubic mecer (ma/m ) for Runs 2 and 3, respectively.

The concencracions of CCl, present in Che stack ^as during Runs I, 2 and 3 were below the detection liaic of the electron capture gas chromatograph

(GC/ECD) analytical instrument (leas than 2 ppb). For the purposes of this report, and the establishment of a reportable ORE for the system while firing a CCl,-concaining waste feed, CCl emission rates of 29*27 x 10" ,

22.79 x 10" , and 24.98 x 10 kg/hr were used in the calculations. The

DRE is calculated using only the stack gas eaission rate and do«s not include

CCl discharged in the scrubber water. Therefore, the established URE's for ^ - CM Runs 1, 2, and 3 are all greater than 99.99 percent, asauaing a detection 0's in lireic of less than 15 ppb (the next highest calibration standard above the _ blank). °

The test combustion data are summarized in Table 5. As is the case with most combustion sources, NO and CO are good indicators of combustion

temperatures and efficiency. During startup and shutdown operations, Che

flare is quite unstable due to the non-steady reactor produce gas supply.

This is due largely Co the face chat, during startup, the reactor requires a minimum time period to reach equilibrium temperature. Because of the

instability upon startup, the system was brought up to operating temperatures

on a waste feed containing nonchlorinated compounds, such as methyl ethyl

ketone (MEK). Once online and up to temperature, the chlorinated waste was

then introduced. There was usually a slight instability in the system aa coe

new waste entered the plasma as was evidenced by changes in the postflare

stack gas temperatures and concentrations of 0., CO., and bU .

(•enerally, once system temperatures stabilized, CU concentrations were

relatively constant at a level of less than 0.17 kg/hr. 10

03107713 TABLE 5. COMBUSTION PARAMETERS - CCl^ TRIAL BURNS

Tesc run 123 Average

Dace, 1985 2/18 2/26 2/26

Stack Cas Flow Rate, a^/aiin* 38.13 29.69 29.81 32.54 ft^ain* 1.346.3 1,048.2 1,052.7 1.149.1

Scack Gas Temperature, •C 908 821 692 BU7 •F 1,666 1,510 1,277 1.4B4

MO^ Concentration, ppo (v/v) 106 92 »l 9J Eniia«ion Rate, kg/hr 0.46 0.31 0.28 0.35 Ib/hr 1.02 (J.69 U.62 U.78

CO Concentration, ppo (v/v) 48 57 81 62 in Emission Rate, kg/br 0.13 0.12 0.17 U. 14 Ib/hr 0.28 0.26 0.37 0.30 c\! C^ 0^, percent 12.7 14.4 15.1 14.1 m C02. percent 6.0 5.7 4.9 5.5 0 —————————————————————————————^————————————-____——————— 0 *Drv standard conditions as defined by 20*C and 760 no Kg.

PCB TRIAL BURN

In February 1985, three endurance PCB trial burns were conducted. The waste feed during these burns was comprised of a blend of three Aroclors, trichlorobenzene, methyl ethyl ketone, aod •ethanol. Askarel

(Aroclor/trichlorobensene blend) comprised approximately 25 percent of the waste feed by weight. This test series was included in che program to test the plasma pyrolysis systea over a period of four hours while a waste ot this type was introduced.

II Wasce Feed - Askarel/MEK/MEOH

The PCB wasce feed blend was introduced •c •n average race of 2.21 kg/min

with a PCB mass input of 0.28 kg/min. or 16.7 kg/hr* This mass input induces

mono through decachloriaated biphenyls. Integrated samples were obtained

during each test run froa the valving assembly just prior to the feed ring of

the reactor vessel. At this point, the waste feed blend was well •ixed and

representative of that fed into the plasma reactor. The samples were analysed

for total PCBs, c h lo rob enx ones, polychlorinated dibenzo-p-dtoxins (PCDOs), and

polychlorinaced dibenzofurans (PCUFs). \0 CM ON Scrubber Water Ln —————————— . . 0 0

Scrubber water samples were collected during each test run and analyzed

for volatile and seaivolatile compounds including PCBs and PCUOs/rCUFs.

During Runs 3-1, 3-2, and 3-3, scrubber water flow race was 36.5, 33.0, and

32.5 L/min, respectively.

The volatile compounds found in the scrubber water wer« principally

benzene, toluene, chlorobensene, and styrene. Chloroechane and 2-butanone

were also found in Run 3-3 in measurable quantities.

Generally, the seaivolatile compounds detected and quantified in cne

scrubber water are sister compounds to naphthalene and pyrene. The samples

were two-phased, and the carbon layer typically had higher concentrations of

seaivolatile compounds than the aqueous phase. In most cases, the carbon

12

0310*7715 separated from the aqueous solution, forming • Cop layer with a light, meringue-type con«istency. In other samples, the carbon remained in susoension or gradually •ettled out over a period of time. Thia inconsistency in carbon layer formation may be due to varying consistency of the •crubber water in which the density of the carbon is greater than Chat of the aqueous solution phase.

Split scrubber water samples were analyzed by a second laboratory for

PCB, PCDD, and PCDF concent. In addition. Bun 3-1 scrubber water saples were analyzed for chlorobenzenes, chlorophenols, and benzo (a) pyrene. PCUDs were not detected in the scrubber water in any of the run*. fCUFs wre detected in r— CM only the first run and mono through decachlorinated biphenyls in the last two ^ runs. Mono and dichlorinated biphenyls represented approximately o9 and 0 81 percent of the total PCS mass in Runs 3-2 and 3-3, respectively. 0

Postflare Stack Gas

Stack gas samples were collected during each run utilizing a variety of sampling trains and methods to obtain the required parameters. The stack gaa constituents sampled for included 0-, CO., CO, NO , parciculate matter,

HCl, volaciles, semivolatiles, PC&s, and PCDDs/PCDFs. Alao included were measurements of gas temperature, velocity, and moisture.

During the three operational perioda in which sampling runs 3-1, 3-2, and

3-3 were conducted, the postflare stack gas was monitored for U_, CU», Cu,

and NU using GCA's continuous emission monitoring system (CEHi>).

13

03107716 Emissions of hydrochloric acid were sampled aC Che postflare sCack Co determine stack gas concentrations a* well •« Che HC1 mass emission races.

Concentrations in thf gas scream were quite low during all three runs averaging only 1.68 mg/in for an average emission race of 64.1 •g/ain. or

0.0084 Ib/hr.

Paniculate emissions from che •tack. show an average parciculaCe concenCracion of 0.005 grains per dry acandard cubic foot (gr/dacfJ wicb an average emission race of 463.2 ag/min or 0.061 Ib/hr. Run 3-1 resulcs were almost cwice as high aa Chose from Run 3-2 or 3-3. During Run 3-1, che scack gas temperature was much lower and che scack gas flow race waa higher than Cbe two subsequenc runs. System problems which led Co a shorcening of Run 3-1 Bay also have caused che increased grain loading (i.e., higher carbon concentrations in Che reacCor gas and poscflare scack gas).

Sampling for senivolacile organica cook place during each cesc period using a Modified Method 5 (MM5) saapling train wich an XAU sorbenc •odule in place. The seaivolaeile samples were analysed by GC/HS. As with the scrubber wacer samples, Che principal componencs found were naphthalene and ics sister comoounds.

A sampling Crain similar Co che one uaed for che colleccioo of nonchlorinaCed seaivolaeile organic co-pounds was uaed co collecC samples Co be analyzed for polychlorinaCed biphenyls (PCBs), polychlorioaCed dibenzo-p-dioxins (PCDDs), and polychlorinaCed dibenzofurans (PCUFsJ. A destruction and removal efficiency (ORE) for Che plasma pyrolysis aystem when

firing PCB-containing liquid wastes was calculaCed for each run and ia

presented in Table 6.

14

03107717 TABLE 6. PLASMA PYROLYSIS SYSTEM DRE FOR PCB«" IN A LIQUID WASTE FEED

Run duration Wacte feed rate PCB content PCB —•• PCB •a«B Syaten DRC Date Run t (•in) (kg/rin) (X weight) input (kg/hr) out (kg/hr)11 percent

2/12/86 3-1 115 2.10 14.3 18.018 O.llxlO-6 >99.9999

2/20/86 3-2 240 2.33 12.5 17.475 cl.lxlO-80 >99.9999

2/22/86 3-3 240 2.20 12.8 16.896 ^.IxlO^ ->99.9999

•Total PCBa a« Mono (1) through Deca (10) polychlorinated biphenyla.

"HCB maaa out doe* not include PCB •••• diacharged through acrubber water. Only atack eririona are uaed in the calculationa.

CConcentrationa of PCBa were below the inatruxent detection lirita according to Zenon'a analyaea. In order to eatabliah a •ini«ui DRE, the •ur of their detection liiita for Cl-1 - Cl-10 w used Co obtain a laaxiimini poaaible e—iaaion rate.

005929 In calculating the DRE for PCBs during Runs 3-2, •nd 3-3, an estimate of die maximum possible PCB emission race had to be u«ed tor chese rune because che sample analyses yielded results below the instrument detection limits.

CONCLUSIONS

Based on the test results •nd the operational experience associated with this test program, it can be concluded that the technology •how* promise as an emernine technology which should be further demonstrated. The data contained herein are useful for engineering research purposes and support the conclusion o m that the technology shows promise for future trial burn programs. 0^ The further conclusions are focused on the demonstration of an acceptable Ln 0 destruction and removal efficiency as delineated in the RCKA and TSCA 0 regulations.

• Results from the carbon Cetrachloride test burn* indicate Chat the system is capable of destroying "difficult to destroy" compounds. The DREs from each of the three test burn* exceeded the •inimum KCItA requirement o£ <99.99Z destruction removal efficiency.

• HCl emission rates conformed to the allowable limits of ^4 kg/hr and <992 removal efficiency based on inlet total chlorine content.

• Concentration of CCl^ in the scrubber effluent ranged from 1.27-5.47 ug/L. Effluent levels met the criteria for discharge to the sewage treatment plane.

• Results from the PCS test burns indicate Chat the system is capable of destroying a PCB liquid waste blend consistent with the TSLA requirement of >99.99992 ORE.

• HCl emission rates were again consistent with the requirement of ^ Icg/hr and ^992 removal efficiency based on total chlorine input.

16 • High concent rations of polynuclear aromatic hydrocarbon compound! were detected in Che two phased •crubber effluenc. The predommanc species were naphthalene, acenaphchalene, phenanchrene, pyrene, and fluoranthene. Level* were in Che range of l2,UUU-72,UUU ug/L. Corresponding level* in Che flue gas discharge were lea* cban 245 UR/a3.

• No appreciable levels of dioxin or furan compounds (as cocal Cecra through occa) were deCecced in Che •crubber water. In all cases, levels were etcher noodeCectable or significantly less than I ng/L. Corresponding level* in the flue ga* were in the range* of 39.1-139 nR/«3 for the total coopounds and N0-12.6 ng/«3 for the Cetra-occachlorinated dibenio-p-dioxin.

REFERENCE

1. "Presention of a Method for the Selection of POHCs in Accordance with the ^ ri RCRA Interim Final Rule, Incinerator Standard*," January 23, 1981, Office 0\ in of Solid Waste. 0 0

17

03107720 Figure 4-A. Sampling point locations.

39 the flare. The ixnicor was originally a penaanenciv aounced SRTK source, buc ho since been •leered •o chat it hac Co be aanually put inco place and removed during che ignition sequence. The pro flare •ample pore* were located 90* Co each ocher in the 6 inch stainless •Ceel pipe approximately ten ieec uoscre«o of the flare. The heavy, aoiat caroon present in the preflare product gas caused plugxinx of the aaapling equipment. It was theorized that «c cimea, some of the carbon buildup would break loose which would then hit the flare, possibly causing it to blow out. A 'T' connection was installed, replJicim the original flare tip. The new 'T' acted as a settling point for sooe of the carbon instead of letting all of it pasa through the flare. Thr can be seen in Fixure 4-3.

m m o^ in o o

40 SECTION 5

SAMPLING PKOCEDUKCS

The procedures for obtaining samples of the waste feed, reactor ash, scrubber water, and postflare stack gas are described in this section. Tne methods used in sampling remained essentially unchanged from those described in the Quality Assurance Project Plan submitted fur this program. Any deviations from the described methods are called out in Section 7. All containers used in collecting and/or scoring these samples were prepared as stated in the QAPP. A summary of the sampling aetnods used follows. '^1' WASTE FEED ^ C^ Liquid wastes entering the plasma arc reactor enter by way of a machined \f\ waste feed ring. The waste feed is pumped from the drums in which it is Q blended, through the calibrated rotamecer, through the stator pump, and into the waste feed ring. Just prior to the feed ring is a 'T* connection and valve assembly which allows a sample to be taken under positive pressure. The valve allows fine adjustment of the liquid flow so that an integrated saaple can be a taken over the duration of rhe test run. Waste feed sampling coomenced after the system was switched over to the desired waste feed (i.e., CCl^ or PCB, depending on the burn schedule). This ensured sampling only the target waste feed and not the flushing solvent blend {i.e., MEK/MEUH). If a system upset occurred during the sampling run, necessitating switching off the target waste feed, the sample valve was closed and not reopened until the target waste feed was switched back on and it was felt that sufficient volume had passed through to diminish any effects of dilution by the flushing solvent blend. Ac the end of each test run, the waste feed samples were returned to a waste feed drum after obtaining the required aliquocs of the liquid which were then transported to Zenon Environmental, Inc. for subsequent analyses. GCA also archived waste feed samples from each test run.

REACTOR HEARTH ASH

Ash samples were not taken and analyzed by GCA. Some samples of carbon found deposited on the snow around the stack area were collected for NYbUbC to be analyzed by their laboratory. Also, samples of carbon were taken from inside the preflare gas pipe and relinquished to NYSUtC for analysis oy their laboratory. The carbon samples were collected in wide mouth jars or 40 oL VQA vials, as appropriate. They were scooped into the containers using hexane rinsed stainless steel spatulas.

41 SCRUBBER WATER

The scrubber water samples were composited in • 10 liter j«r •very 30 minutes during each ceac run. During each grab sampling episode, two 40 mL VOA vials samples were taken for subsequent VOC analyses. A composite scrubber water sample for each run was collected for seaivolatile analysis by the GCA laboratory. The 10 liter sanple jars were transported to Zenon for analyses for PCBs and PCDDs/PCDFs. In all compositing efforts, the scrubber water was mixed thoroughly to ensure a representative carbon/aqueous solution mix. During some sapling periods, the amount of carbon present in Cbe scrubber water was noticeably less than at other times.

POSTFLARE STACK GAS

Table 5-1 summarizes the sampling methodologies utilized in this program to characterize emissions from the pyrolysis system while pyrolyzing two types of chlorinated water feeds and ftaring the product gas*

Bulk Cases

Continuous Monitoring— 0^ A continuous monitoring system was in operation during the two test LP» series of the project to monitor concent rat ions of CO, 0;, CO;, and N0^ Q in the flue gas. In addition, continuous sensors for measuring postfl arenas flow rates were included during the CCl^ burns. An attempt Co monitor flow rates and temperatures at the preflare location was aborted due to carbon plugging and the hydrogen flarebacks. The monitoring system was comprised of a gas conditioning system, for measuring CO, 0;, CO; and N0^ and a data acquisition system as shown in Figure 5-1.

The gas conditioning system consisted of a glass fiber filtration unit mounted on the probe to remove particulates and a condenaace crap for primary moisture removal from the flue gas. The final seep'in moisture removal is achieved by an inline permeation drier. Sample gas exiting the permeation drier is then ready for analysis. Gas analyses will be performed using the instruments described in Table 5-2*

Carbon monoxide concentrations were measured using a Horiba Model FIR 2000 NDIR Analyser in the operating range of 0 Co U. 1 percent full scale. Calibrations preceded and succeeded each test by injecting the appropriate xero and span gases.

Oxygen concentrations wore measured using a MSA Model 8U2 0; Analyzer in the operating range 0 to 25 percent full scale. The analyzer was calibrated before and after each test with a zero gas of ultrapure nitrogen and calibration span gaaes of the appropriate concent rations.

Carbon dioxide was measured using a Horiba FIR 2000 NOI& CO; Analyzer in the operating range 0 to 25 percent full scale. This analyzer was calibrated before and after each test with the applicable zero and span gases.

42

03107724 TABLE 5-1. PREFLARE AND POSTFLARE EMISSION PAHAMETLKS M£ASUk£M£MS

Parameter Co I leecion —thod

HCl Impingers (TACB Method)*

Volacile Ornanica1' Integrated Tedlar Bag (onsite analyes for CCl^)

Volatile Organic Sampling Train [(VOiiT) (offsite analyses)] • Semivolatilea Modified Method 5 (MM5)

PCBs Modified Method 5

PCDD/PCDF Modified Method 5

Particulate aatter Modified Method 5

•Texas Air Control Board Method.

^Baa saaplinx/oasite GC used for all l-hour burns; VUST used during endurance burns. ' •

43

03107725 Figure 5-1. Continuoua nonicoring sampling schematic.

44

03107726 TABLE 5-2. CEM SAMPLING PARAMETERS AND METHODOLOGY

Available oeaaureaenc ranges ParameCer Iircruinenc aodel (dececcion) (up Co)

CO Horiba PIR 2000 (NUIR) 5,000 ppm 00 02 MSA Model 802 (Paramagnetic) Z5JI m cn^ Horiba PIR 2000 (MDIR) 252 0^ in ^x TECO Model IOA (Chenii luminescence ) 10,000 ppa 0 0 Velocity Rosemonc InacruoenCi 5 inch W.C. P

45

03107727 Oxides of nitrogen (NO,) were measured using « TECU Model IUAR Chemi luminescent Nt\ Analyzer in the range of 0 Co 1000 ppm. Calibration of the analyzer was accomplished using nitrogen zero gas and span gases of the appropriate concentrations.

Continuous reoni.coritie of the flue gas was performed according to the following sequence:

1. Arrived onsice, inspected condition of equipoent.

2. Set up and leak checked conditioning system through manifold.

3. Connected all four analyzers to the •anifold and data acquisition syscem.

4. Performed initial calibration of all monitors with zero, aid and high span certified gases* Hade any accessary adjuscaenca on the uonitors.

5. Monitored CO, O;, CO^ and NO, throughout the flue gaa testing as making sure to mark the scrip charts noting the beginning and end^ot the test runs. 0 in 6. At the end of each run, recalibrated the monitors and noted all -^ values on the appropriate data sheet to determine moaicor drift*

7. Monitoring data were reduced and presented as average concentrations and, for CO and NO,, hourly emission races* • The continuous monitoring system inspection, installation and operation was performed in accordance with the applicable iastruoenc manuals*

Flue Gas Molecular Weight by Integrated Orsat— The flue gas molecular weight is required by TSCA and &CRA to be caleulacad from data representing each sampling point in the stack* Because the CEMS was situated for single-point monitoring, ic was necessary to calculate the molecular weight from the average of the data froo the four 1-hour integrated bag samples taken during each run. During Run 3-1, ie was only possible to obtain two l-hour samples due to the shortening of the teat run. A lung sampling system wea used to collect the integrated stack gas sample in a Tedlar bag* This system was leak-checked before and after each sampling run to ensure no leakage occurred during the run.

Trace Gases

The trace gas samples were collected from the poscflare stack and were analyzed for HCl, s«mivolatiles, and, more specifically from the latter group, PCBs and PCDD/PCDF. Carbon tecrachloride and polychlorinated bipbenyls were introduced into the plasma arc aa separate waste matrices during the two test •t4«e«. The capability of the plasma arc system to either destroy or remove the target compounds were measured by the methods outlined in this subsection.

46 Volatile Organic Sampling Train (VOST)— The VOST was used Co collect VOCs present in the produce flue gas during ehe lone Cerm endurance run*. VOC refer* Co Chose organic compound! wicb boiling points less than 150'C. The method utilizes Ten«x «nd Tenax/Charcoal cartridges each of which is preceded by a condensing module Co adsorb Che VUCs.

The train consisted of a glass-lined probe with a glass wool plug to remove particulace, followed by an asseably of condensers and organic resin Craps as illustrated in Figure 5-2. The first condenser cooled Che gas screaa and condensed the water vapor present. The flue gas and condensed aoisture then passed through a cartridge containing 1*5 gr—s of Tenax resin (60-80 oesh). cbe condensate was collected in Che first iapinger which was continually purged by the gas sere—. The second condenser and trap containing Tenax/charcoal served as a backup for low volume breakthrough compounds. Following the second Tenax trap is a silica gel drying Cube for residual uoisture removal. The sampling train was operated at a flow rate of 0.3 liters per minute and the total collection volume did not exceed 2U standard liters.

Saple temperature was monitored at the outlet of the saaple probe and ^ the inlet to the Tenax cartridge using cheraocouples. The gas teaperature C^ through ehe probe was maincaioed above 150'C to prevent the premature lT\ condensation of the volatile components. The temperature of the gas through Q the resin cartridges was maintained at less than 20*C. ^

Extensive sorbent preparation and quality aasurance procedures were instituted to ensure ehe integrity of these samples. All components of the system coming into contact with the samples were rinsed with UI water and dried in an oven at 150'C for a period of 2 hours prior to use. The Tenax adsorbent and glass wool packing were precleaned as described in the Quality Assurance Project Plan.

The presanpling preparations for the VOST included:

e Washing the train and saaple containers using the following sequence: soap and water, pre-excracced ODI water rinse, and •ethanol rinse.

• Preparing VOA vials for coodensate recovery which were not solvent rinsed but were heated in an oven at HO* C for 2 hours.

• Setting up a field biased blank sampling train which was disassembled end recovered in the same maooer as the actual VOST following each run.

• Collecting method blanks of DI water, Tenax, end charcoal.

47 HEATED 3-WAV SAMPLE VALVE PROBE

ICE WATER CONDENSER

TENAK CARTRIOGE

CONDENSING IMPINGE R U^Ai H^O

005941 Figure 5-2. Volatile organic sampling train schematic. The recovery activities for Che VOST Included:

• Removing VUST from stack •nd transporting Co nearby recovery area.

• Sealing the •orbeac cartridges with teflon cape and placing Cheni in cheir original glass culture tube* with glass wool Co abaorb shocn.

• Measuring Che volume of Che condensace iapinger with a precleaned graduated cylinder.

• Tranaferring Cbe aeaaured condensate voluae Co 40 mL VGA vials and diluting to voluae with DI water to decrease headapace and Cbe poaaibility of revolatiliiation of Cbe coapounda.

• Further reducing reactivity by •Coring all aaaples at 4*C.

The •aaplea which were collected during each VOST run eonaiaeed of a Tenax cartridge, a Tenax/charcoal cartridge, and the produce «aa condenaata (captured by the mini iapinger), All •ample* were labeled according to their (-\j •erica number, run number, sampling train type, aaapling, component, and ^ parameter to be analyzed for. (^ in Leak cbecka on the a—pling train were performed before and after.each •aopling run. No portion of the traina were disassembled and reaaaeabled o during a teat run. The ample train leak checks are documented on the field c-' teat data aheet for each respective run.

Integrated Bax peopling for VOCa— For the purpoae of ooaite analyaea for the target volatile compound* during l-hour CCl4 bum*, an integrated Tedlar bag collection oechod waa utilised to collect the aaaplea. A ayringe waa then uaed to draw a aaaple froo the bag and inject onto the GC/ECD column.

The bag •«nplo« were collected on an hourly baaia and analyzed for VOCa onaite by GC/ECD. One bag per day waa filled with prepurified N3 for uae aa a field-biaaed blank.

The gaa bag aanpling aaaeably eaployed ia ahown in Figure S-J. It eonaiaced of a cleaned, evacuated Tedlar bag placed inaide a rigid container Chat waa evacuated at a known rate during the aaapling period. Prior to sampling, Tedlar bags were purged with prepurified nitrogen and evacuated. After the sample was drawn, a quick-disconnect valve atop the container aealed the sample in the bag for direct enalyaia.

Leak checks on the integrated bag sampling train were performed before and after each sampling run* The sample train leak checks and leakage race (if applicable) are documented on the field test data sheet for each respective run.

Modified Method 5 (MM5)— Modified Method 5 sampling traina were uaed for the collection of particulates, aeaivolaciles, polychlorinaced biphenyla, and polyeblonnaced

49 TO PUMP

005945 Figure 5-3. Integrated gas sainptlnR train. riibensso-p-dioxins/polychlorinaced dibenzofurans (PCUD/rcUF). A schematic ot the MM5 sampling train is shown in Figure 5-4.

Additionally, a field-biased blank was sec up at the sice for eacn parameter (i.e., one semivolacile field blank, one f<-b field blanm during each cesc day. 1C was sec up and recovered exactly che same way as che accual sample buC without having had sample gas flow through che system. The field-biased blanks were Created analytically che same way as actual samples and che results will provided appropriate blank corrections.

The sample train consisted of quartz glass-lined heat-traced water-cooled probe with an ineonel button hook nozzle and attached thermocouple and pxcoc Cube. The probe was maintained at a Cemperacure of 250T •»• 25*F. After the probe, the gas passed through a heated glass fiber filter TKeeve Angel 9J4 AH filter paper). Downstream of the heated filter, the sample gas passed through a water-cooled •odule, then through a sorbent module containing approximately 25s. of XAD-2 resin. The XAO module, which was kept at a temperature below 20*C, is followed by a series of four impingers. The first iapinger, acted a« «^- a condensate reservoir connect to Che outlet of the XAO module, and was _.. modified with a short stem so that the sample gas did not bubble through cue collected condensate. The first and third impingers were empty, the second contained 100 •L of DI water, and the fourth contained a known weight of Ln silica gel. All connections within the train were glass or Teflon and no 0 sealant greases were used* The impingers were followed by a pump, dry gas o meter, and a calibrated orifice —cer.

Readings of flue gas parameters were recorded at every sampling poxnc during the sampling traverse. la the event that steady operation was ni.;, maintained, or there were atypical fluctuations in monitored gas parameters (CO, 02), the testing was stopped until these conditions were stabilized. Steady operation of the pyrolysis unit was the responsibility of Pyrolysis Systems, Inc., personnel, but the flue gas parameters and composition was monitored by GCA. Any changes were noted and relayed Co PSI personnel so thai appropriate action could be taken.

Sampling was conducted while traversing the MM5 train across the two diameters shown in Figure S-5. The stack satisfied the tf and 2 criteria for a minimum number of sampling points and a twelve point test was selected for these runs.

Prior to sampling, all gas-contacting components of the train were washed with alconox and water, thoroughly rinsed with DI water, and oven driea ac IIO'C for at least I hour. Immediately prior to use, the components were rinsed with hexane. Ac each test point within the stack, a'll necessary train parameters were measured. All field data sheets are included in the appendix.

Leak checks on the MM5 sampling train were performed before and after each sampling run. The sample train leak checks and leakage rate (if applicable) were documented on the field test data sheet for each respective run.

51 REVERSE-TYPE PI TOT TUBE

Figure 5-4. Modified Method 5 train.

52 Distance Point from Wall (in.)

1 0.70 2 2.34 3 4.74 6 11.26 5 13.66 6 15.30

Figure 5-5. Modified Method 5 sample point locations.

53

03107735 Following completion of ««ch test run* the MM5 trains were transported co che oniice trailer for recovery. Recovery procedures differed depending on whecher the 4-hour MH5 samples were Co be analyzed for PCUD/PCDF and PCas or semivolatiles. The PCDD/PCDF end PCB trein was analyzed •olely for choee parameters end ooc for perciculate emissions. The recovery procedure! outlined in Che Quality Aaaurance Project Plan were followed.

Becauae Che particulace catch froa the Mtt5 eraias incended for semivolacile GC/HS analyse* were Co be firac used co determine particulace emission races, • slightly aore involved recovery procedure was required. The recovery technique was conducted on chose 4-hour W5 runs not incended for PCDD/PCDF analyses. The procedure followed during this recovery was aa outlined in che Quality Assurance Project Plan.

Afcer Che eocal particulace catch was determined, the acecone rinse residue waa redissolved in hexane and che organic analyses proceeded as described in Cbe analytical aeceion of Cbis report.

Filters for all modified Method 5 trains were tare-weighed prior to uae~ for the determination of parciculaee loading. These fileers were then platfda in seeled class containers for shipment Co Che CesC sice. Q-,

Sorbene resins uaed in MH5 sasipling Grains require extensive preparation and quality aaaurance •easures prior Co use in Cbe field Co eliainace the possibility of saaple bias due Co sorbenc contaainacion. The XA&-2 resin u«ed in ehis saaple Grain waa obtained froa Supeico, Inc. Tbe resin was precleaned by soxhiec extraction using the sequence outliaed in the IESL-BIP Procedures Manual: Level I Environaencal Aasesaaenc aa described in the Quality Aasurance Project Plan. The XAD-2 resin uaed in the HH5 train for PCDD/PCuF collection was soxhiec excraeced ovemighe in coluene in addition Co Che sequence outlined in che IERL-BTP Level I assessment procedure.

Gaseous HC1— Sampling waa conducted at Che poscflare saapling location Co quantify HC1 emissions during each scage of this Cesc program. The •echod uaed in collecting the gaseoua HC1 was the Texas Air control Board (TACBJ •echod with deionised water in the first two iapingers aa the absorbing solutions.

The saapling train is shown in Figure 5-6. It consists of a probe liner, an iapinger train with DI H^O and silica |el, a puoip, a dry gas aecer, and a •aooAecer. Chloride was decenxined onsice eoloricetrically during Che CCl4 burns. After the 4 hour PCB enduranca burns, the saaples were transported back to GCA for analysis.

Leak checks on the HC1 saapling train were conducted before and afcer each saapling run. The sample train leak checks and leakage race (if aoplicable) are docuoenced on the field test data sheet for each respective run.

54

03107736 SAMPLE PROBE

ABSORBING IMPINQER8

PUMP

FiRuru 5-6. Caseous HC1 sampling tr.iin. 005948 Moisture (H;0)— Hoiature in the poacflare acack gaa was determined by irpinger wight gain from the beginning of the test run Co che end. In Chi« mechod, initial. •nd final weight* of che impingera are obtained by weighing on a balance accurate to 0.1 gram.

Temperatures and Flow Rate*— During che 1 hour CCl^ ceac runa, a preaaure cranaducer and thermocouple were uaed to continuously monitor acack gas velocity and temperature* ac che poatflare location. Hoaicoriag the pr«fl«r« location waa attempted during the firat CCI^ bum but waa aborted following aeveral hydrogen flarebacka in eb« preflare pipe aa well aa experiencing heavy carbon plugging of Che pitot Cube.

Flow races and tea per at urea during che <* hour PCB endurance Ceac runa were aeasured during the HH5 sampling when velocity and temperature reading a were recorded at each aaapling point. The aenivolacile sampling train measurements were uaed in calculating che pollutant emission ratea during che run. The flow rates calculated from these measurements were in cloae 0^ agreement (<5 percenC difference) with chose flow rataa aeaaured with Cha «^- PCDD/PCDF sampling probe borrowed from che Oncario Ministry of Che Envi.rooae^c. in o o

56

03107738 SECTION 6

ANALYTICAL METHODS

The following section delineates the analytical prococols which were used to analyze samples in the field as well as the GCA laboratory. As discussed in earlier sections, the following process streams were sampled:

• Waste Feed

• Reactor Hearth Ash 0 in • Spent Scrubber Water 0^ in • Poscflare Produce Gas 0 WASTE FEED AND SCRUBBER WATER 0

Volatile Organic Compounds • Field samples for VGA analysis were collected in duplicate. Waste feed samples were archived. Analyses of aqueous samples were conducted using conventional purge and crap GC/MS procedures in accordance with EPA Method 624. Surrogate spikes consisting of dy-coluene, d^-l,2-dichloroethane, and bromofluorobenzene, were added to all samples a« described in Section 9 of the Quality Assurance Project Plan. Analyses were conducted using a Finnigan OWA computerized gas chromacograph/mass spectrometer. Priority pollutants were identified by computer match of retention times and spectra of standards determined on that day.

The identification of additional compounds detected by the GC-MS system were determined using a computer search of the httS/EPA/Nly library Co provide tentative spectral matches. These, as well as unknown spectra, were reviewed for major peaks and fragmentation paCCerns. Further identifications were made by the operator and verified by comparison of available reference spectra (NBS/EPA/NIH library) to the background corrected component spectra. Component concentrations were calculated relative to the closest eluting internal standards. All values represent approximations due r.a inherent variabilities in component response factors in the absence of refereace materials.

57

03107739 The following steps were taken co dececc conciminacion introduced in sample handling or analysis:

1. Analysis of field biased blanks'-Deionized water caken co the field and carried through che scorage and analysis procedures were analyzed.

2. Daily analysis of method blanks—A sample of deionized waccr was analyzed. This dececcs concaminacion introduced by the purge gaa or the Cubing in front of Che crap.

3. Afcer analysis of a high (>200 ppb) level sample, a blank of deionized water was analyzed co ensure chac concaminacion of subsequent samples by carryover had ooc occurred.

Semivolatile Organic Compounds

Aqueous samples collecced for semivolacile organic compound decenoinatioA- were analyzed in accordance wich EPA Hechod 625. iiurrogace compounds were i^ added eo che samples and chen serially excracced wich mechylene chloride ac a pH greacer than II and again ac pH less ehan 2, using a separacory funnel. The nechylene chloride excracc is dried and subsequently concentrated Co a Lr\ volume of I Co 2 oL. The concentrated excracc is Chen analyzed by gas 0 chromacographv/mass spectromecry. Paniculate oaccer was separated from 0 aqueous samples via filtration and subsequencly excracced wich a soxhiec apparatus using aechylene chloride. The extract was then subjected to concentration and GC/MS analysis as described for aqueous aliquocs.

POSTFLARE STACK GAS

Modified Method 5 (for Semivolacile Organics)— Method 5 Grain samples were analyzed by GCA for semivolacile organic compounds. Each Grain, as noted earlier, produced four types of samples: 1) parciculaee collected on a filter; 2) probe rinses; ^) XAU suroenc samples; and 4) impinger/condensates. Figure 6-1 shows the analytical flow scheme.

The recovered filters and probe rinses were prepared tor gravimetric analysis. Once particulate weights were recorded, catches and probe rinses were combined wich the appropriace XAD sample and excracced. Three samples were extracted composited and analyzed from each MM5 Train:

• impinger waters and eondensace,

• probe rinse (front half) and filter,

• back half rinses and XAD adsorbenc.

The latter two samples were surrogate-spiked with de-nicrobenzene, 2-fluorobiphenyl, d^-terphenyl, d5phenol, 2-(:luorophenol, and 2,3,6-cribromophenol, and then soxhiet-extracced for a period of 24-hours in mechylene chloride.

5o

03107740 IMPINGBR WATERS ACETONE .FILTER(S) HEXCANE XAD SORBENT & CONDENSATE PROBE RINSE | TRAIN RINSES MODULES 1 1 1 DRY, HEIGH WEIGH SURROGATE SPIKE \ \ t t ' <

EXTRACT WITH METHYLENE CHLORIDE (24 hn)

DRY AND CONCENTRATE TO 1.0 ML

GC/MS ANALYSIS IN TOTAL ION MODE

Figure 6-1. Mudlfle'I Method 5 i.riln organic airlyaiB flow aches

005952 09

0 0 U1 \D ~ •(•»3«»rt po» .r3»n JO •fAfuv '[•oiiato JO) •poq3»H) 7"^t Poq3»M m q3aoJ 3*« lowaoad »q3 03 »u}pJ033» u J»xXiruyo3nv 009104301 • 9ofn p»2X-[»u» paw p»33»ll09 9J»A canpf*^ •qaoq 2J»j • u-i «.t»q

••.i39»d« p»ATJ9p-^rJqf[ »q3 q3tA tJ33»d* 309uodao3 JO oo»iJ»dB03 rnaw Aq p«30»BMtddn« »3»a T».233»ds •••N JO Aai.iq'n SflK »U3 ?o q3Jf PJVAJOJ • Ifofn p»iJi3Q»pi •JM •punodoos i»ao'i3'ippv •»pJ»pu»3c 03 ••q33»« T.l33»d» pn» •nn) uoi3a»3»J Xorn »p»a ••A (d3H) •llJOJd 30UJH3 VOX T303 »q3 01 •Ifd JO UO'l3»3l?13a»PI •"•Id 33»fo.l^ a3D«Jn««V ^3f[»n(l 9q3 20 g-^ »iq»l ni pa3»fl 919 •ooi3ipuo9 8ui3«J»do 3oauru3»u-i ?W/3S •apoB 001 l»303 aq3 01 p»33»llo3 aJ»A •i399d« 3uaaoduo3 iiv •tp^»pu»3» 1»u3a3a't aq3 pa« (5M/3*)) XJ3»ooJ339d« •cvai/Aqd«JHo3*inoJq3 •«X omnios ^x»i-i}ae3 iturn 'caiufXJo a'['i3flOAia»« JOJ p»ua»3« C*A 33«.i3x» pauiqaoa »qJ,

•T" O'l 03 sr\y939dd9 q«iu»G •uJ»pn^ •TA p»3«J3oa3uo3 pir 33U3X* ^»3IiJ/(JVX ^3 Ml1" paaiooo? »3»J'in» unipo* J«AO pxi^p u»q3 a^fi f3;:;3x» »iu, 'apiJo-iqa »u»lAq3«"u q3iA •«ai3 ••Jq3 pa33«J3XB aJ»A «»idu»« »3»«u9puao pu» J»3»A J»»niauii SECTION 7

QUALITY ASSURANCE/QUALITY CONTROL

INTRODUCTION

Quality Assurance/Quality Control (QA/QC) protocol* followed in this program were based upon routine sampling •nd analytical practical and the revised Quality Aaaurancc Project Plan (Q4PP) entitled Performance Testing of the Pyrolysis Systems, Inc. Plaaaa fyrolysis Unit, submitted for chit program on February 4, 1985. This Quality A««uranc« •ection will detail areas where ^f changes in laboratory and/or field procedure! were made. In order Co LO facilitate review of pertinent QC data. Chic •ectioo will roughly follow the ^ outline of the QAPP.

PROJECT DESCRIPTION 0 0 Stage II Te»C3

Preflare produce gas wax not sampled for €014, HC1, or •onitored for velocity and temperature. The high carbon and •oiature concent of the gas •cream were not conducive to aucb measurements. A pitot Cube/pressure transducer and thermocouple were mounted in the •aaple port but were removed because hydrogen flarebacks created a hazard at the preflare sampling location* The hydrogen flarebacks were essentially ignition of the preflare product gas from the flarehead back to the 1*0 fan within the trailer, resulting in a •inor explosion inside the fan. No further testing was attempted at this location. However, the ports wre periodically used for obtaining carbon samples, between runs*

The availability of reactor ash after test runs was limited due to the lengthiness of the cool down period and the difficulty associated in obtaining the •ample. This necessitated the reaoval of plasaa torch, the waste feed ring, coolant flange, and graphite core*

Scxge III Tests

Two test series (2 and 3) were scheduled for the Stage III Level of testing involving the destruction of PCB containing liquid waste feed. In the effort to complete Stage II Test Series 1 and Stage III Test Series 2 in February 1985. fundine was depleted due to the length of cine the crew was required to be onsite. These delays were caused by equipment problems within the pyrolysis system, ones which, largely, could not have been foreseen prior

61

03107743 to actual tesL attempts on Che different waste feeds. Stage II toeing was completed after several attempt*. Stage III Test Series 2 underwent several attempts in February and March, 1985 at completing a single teat run but ch« efforts were unsuccessful after repeated pyrolysis system equipment proolems. Testing was suspended until Che equipment problems could be corrected. It was decided chat a local test company would conduce the Stage III Test Series 2 testing on an "on call" basis. This testing was completed in January l9bo.

GCA was concracced co conduce the Stage III Test Series 3 fCB endurance cescs in 1986, which were Co be a minimum of 6 hours in duration. The first burn attempt lasted only 115 minutes. It was decided ac Chat point by NY;»Ufc,(. and PSI that shorter test burns would be necessary because it was questionable as Co whether the system would stay online for the duration of a full e-nour test. Rather than risk another aborted run, the sampling tiae was shortened to 240 minutes (4 hours). Two successful test burns wer« then conducted, finishing Stage III Test Series 3 and GCA's involvement in Phase II activities. Stage IV testing, as described in the yAff, was cancelled. in PROJECT ORGANIZATION AND RESPONSIBILITIES tpi

During the course of this program, three major changes were made xn G(-A'« project and quality assurance organization. Mr. Paul Exner replaced Dr. Paul Fennelly as Program Manager. Ms. Joan Schlosscein replaced 0 Ms. Andrea Cutter as Analytical QC Coordinator, and Mr. Howard Schiff replacedO Mr. Richard Graziano as Measurements QC Coordinator*

PRECISION, ACCURACY, COMPLETENESS, REPRESENTATIVENESS AND COMPARABILITY

During the 4-hour PCB endurance tests, the continuous emission monitors and orsat gas analyses were the only measurements made onsice. Precision estimates could not be made on these measurements with any consistency becauae there were so few data points. Only two complete test runs were conducted and these for only four hours. Accuracy estimates of the CWs were not made due to no provision being; made for audit gas cylinders under this program. A summary of measurement precision, accuracy, and completeness goals and accomplishments is shown in Table 7-1. The hourly orsat data for kuna 3-2 ana 3-3 did yield 0^ —an concentrations of 14.0 and 15*3 percent, respectively, and COq mean concentrations of 5.1 and 4.3 percent, respectively. Standard deviations for hourly 0; readings were 1.04 and 0.15 percent for the two runs. CO^ concentrations yielded standard deviations of U.64 and U.38 for the two runs. The true precision of the measurements cannot be esciaated due to there being only four hourly measurements made per run and no sioultaneuus measurements made.

Since insufficient sample volume was available for replicate aliquots in most cases, analytical precision could not be determined. Analytical accuracy was determined through the analysis of EPA Environmental Monitoring and Support Laboratory (EMSL) Quality Control Samples and the analysis of matrix and surrogate spiked sample aliquots. Results of these analyses broken down by parameter are presented in Tables 7-2 through 7-b and are discussed below.

62

03107744 TABLE 7-1. SUMMARY OF OEM PRECISION, ACCURACY, AND COMPLETENESS

Preciaion Accuracy Completeness Measurement (parameter) Goal (X) Actual* Goal(X) Actual11 Goal (I) Actual (X^

Oxygen <0.5 (v/v) NA IS ND 95 99 (MSA 802)

Carbon Dioxide <0.5 (v/v) NA 15 ND 95 99 (Horiba PIR 2000)

Carbon Monoxide <2.5 (v/v) NA IS ND 95 99 (Horiba PIR 2000)

Oxidea of Nitrogen 10 RSD NA IS ND 95 99 (TECO 10A)

Oxygen 0.6 (v/v) NA 3.0 ND 95 100 (EPA Method 3)

Carbon Dioxide 0.6 (v/v) NA 3.0 ND 95 100 (EPA Method 3)

'Preciaion eatimatea could not he determined from only two complete teat runa (four data pointa). "No audit gaa cylindera were provided for chia program. Original acope of program changed greatly - only two complete teat runa. 'Six (ainiitea of CEM data were loat during Run 3-3 due Co purging of condenaate from the aample line.

005956 TABLE 7-2. ANALYSIS OF A LABORATORY CONTROL SAMPLE PUR CHLORIDES

Expecced Recovered Percent QC ««mple (og/L) (og/L) recovery

VP-882-1 8.52 9.18 108 r- m 0^ in o o

64

03107746 TABLE 7-3. ANALYSIS OF A MATRIX SPIKE INTU SCRUBBER WATER FOR VOLATILE OKGANICS

Percent QA objective Compound Concentric ion (ug/L) recovery (Percent RecoveryJ

Exoected Recovered

I, l-Dichloroethene 50 52 104 bO-l8U 00 in Trichloroethene 50 43 86 60-140 0^ Ch lorobenzene 50 45 90 6U-14U m o Toluene 50 44 88 60-14U o

Benzene 50 49 9U 6U-15U

65

03107747 TAaLE 7-4. ANALYSIS OF A SURROGATE SPIKE INTO SCRUBBER WATt.R FOR VOLATILE ORGAN I CS

Compound ?erc«nc Recovery QA Objective Run 2 Run 3 Average (Percent Recovery.)

Toluene-OS 96 9V 9d none

Bromofluorobeniene 121 luy 11^ l,2-UitfhloroeCh«ne-D4 9t» 116 lub m o o

66

03107748 TABLE 7-^. ANALYSIS UF St. Ml VOLATILE ORGANIC MATRIX SPIJCEU COMPOUNDS (PERCENT KECUVEKIES)

Compound Filt race Water Filter XAU UA Objectives l,2,4--Trichlorol>enzeiie 91 91 04 2U-11U

Acenaphchene 100 96 8^ 3U-140

2,4-OiniCrocoluene 114 11U do 2U-9U fyrene 128 1U9 71 3U-1JXJ

N-nitrosodi-n-propyi;mine IU3 9J 76 <*U-UU

1,4-Uichlorobenzene 92 99 73 2U-11U 0 v0 Pencachloropheaol 74 8U 51 IU-110 0^ Phenol 34 75 36 2U-90 m • 0 2-Chlorophenol 66 80 46 20-110 0

4-Chloro-3-Bethylphenol 68 75 4U 2U-110

4-Nitrophenol 29 9u 56 IU-12U

67

03107749 TABLE 7-6. ANALYSIS OF SEMIVOLATILE ORGANIC SUlUOWAlB SPIKED COMPOUNDS IN SAMPLE ALIUUOTS

Sfple ID P«rc«nt R«>covry UlcrobenzBaa Z-Fluoro- 1'•rphcayl Pb«aol——3t-tluoro- •2.i,^ KT- -°5 blph«nyl I'5 tituaol Broao- phaaol

Scrubber Wat«r 81 99 86 57 25 83 Method Blank •!- Scrubber U«c«r 78 98 79 49 21 96 ^ MuLrIz Spike 0^

Run 1 Scrubber 71 104 63 81 31 78 ^ Vacer 0 kun 1 Scrubber 59 104 58 80 50 .6 ° W«c«r

Kua 3 Scrubb«r 64 102 50 49 50 81

XAA/Fllcar Blank 133 157 104 158 41 85

XAC Matrix Splk« 130 154 93 142 69 116 kua 1 M-5 Blank . 128 147 95 148 68 85

Run 1 M-5 129 168 82 99 30 164

Run 2 M-5 Blank 138 154 97 162 40 87 kua 2 M-5 137 169 68 0" 10* 147

Run 3 M-5 Blank 102 130 99 125 13 103

Run 3 M-5 131 147 81 7* 4« 96

QA Objectives 41-120 44-119 33-128 10-110 10-130 aoae

68

03107750 Complftfneaa, defined as the percentage of •II measurements whose results •re judged velid, waa escimaced Co have achieved che goal of ^5 percent. Uherever possible, reference methoda and •Candard sampling procedure* were uaed a« scaced in the QA Plan Co ensure comparability with ocher representative measuremeaf aade by GCA or another organization.

SAMPLING PROCEDURES

Senivolatile. PCDD/PCDF Train*

Several quality control problems occurred during Che sampling prograa conducted in February L986. Theae problems eooaiaced of aonisokiaecic sampling, pitoe cube and sampling train leak checks, configuration of the PCDD/PCDF train probe tip, and the eziaeence of blockage within cbe •tack due Co the simultaneou* sampling require—ac« of the program.

Simultaneoua sampling of a —all area (i.e., a 16-inch diameter •cacx) with two 3-inch di«a«cer water-cooled a—pling probea, would lead to flow blockade at Cbe sampling plane* This blockage causes a greater velocity of C\J the flue gas paaaing the pitot Cube* and •ample nozzle than would be measured ^Q uaing —Her diaaecar probea or a aingle probe. The effect of the blockage ^ on the velocity readinx* by each individual train could not be ascertained. However, Cbe effect on each train would be equal aa ebe velocities were in close aRreeaent. For exaaple, during Run 3-3, ebe velocity and flow rate ° measured by the semivolatile train was 3,843 ft/ain and 1,264 fc^/ain, 0 respectively. The sane iMasureaencs taken by the PCDD/PCUF train were 3,797 ft/ain and 1,234 fc"/axn, respectively. The concerns regarding the possibility of flow blockage were presented to ERA and New York State technical personnel prior to the co—enceaent of Stage III testing. They concurred chat although flow blockage aay exist within the stack, the operational and Ciae constraints of the prograa were such chat silltaneous sampling for semivolatile and chlorinated organic* (PCDD/PCUF) was necessary.

The pilot tube bead of the PCDO/PCOF probe was positioned approximately 2 inches behind the Quartz sanple nozzle due to the length of toe integral quartz probe liner supplied with the probe. The aethod calls tor close alignaent of the pitot tips and sampling nozzle. No other configuration of the probe waa poasible without extenaive Modifications of the probe and saaplinx hoe box. This •isaligcent uaually would have an effect on the isokinetie saapling aa the velocity at the pitoc head would not necessarily be the saae as the velocity at the sample nozzle. However, the velocity profile of the stack waa fairly flat across each line of traverse with very little variation between the points. Therefore, the effect of Cbe •isaligaoenc on isokinetie saapling waa not aa great as if the velocity profile waa skewed or erratic.

All sampling train and pieoc tube leak check* were conducted prior to, and on the coaplecion of each run. The leak rate liaic for the saapling train of ^0.02 cfm was ret for all runs conducted. However, the leak check on the PCDD/PCDF sample train following Run 3-2 was conducted from the front-half of the filter, becauae the quartz nozzle was thermally the BOSC weakened at two

69 or three points approximately 3 inches back froa Che 9U" bend. Affr cooling •nd •n attempted leak check, this portion broke off completely «od fell Co ch« ground. The nozzle was oot rioted iaco the •«aplc container becau«« of ch« possible contamination of the sample froa carbon deposits on Che ground. The probe section was recovered. The train leak check vac then conducted from Che front of the filcer holder. It is assumed chac the breakage occurred because of: I) burnthrough of the liner occurred within Che stack due to the extremely high temperatures, and 2) upon removing the probe from cne stack, the nozzle cracked through froa thermal shock resulting from the very cold ambient temperatures. The amount of paniculate •atter loac in the nozzle was detenoined to be negligible due to the very low grain loading within the flue ties.

The pose-run PCDD/PCDF train picot cube leak check for Run 3-2 failed Co meet the leak rate reauireaencs on ebe impact •ide of the pitoc. The high •tack ceapersture caused a bumchrough in the probe, including the quartz liner and picot tube*. 1C is assumed Chat this burncbrough occurred during the final half-hour of the run when temperatures were highest. Theoretical^ Che test run should be voided. However, ebe velocity head aeasured tty bocn^ trains were in fairly eloae agreement (0.36 in. H^O for cne JS—ivolatile train and 0.37 in. H^O for Cbe PCDD/PCDF train during Run 3-2). If it is u assumed that the bumchrough occurred in the final •inucss of the test run, Lr> the effect on overall velocity measurements would be minimal. 0 0 There was a problem with nonisokinetic sampling of the flue gaa during Test Runs 3-1 and 3-2 (semivolatile train) and Run* 3-2 and 3-3 (PCDD/PCOF train). A test run is valid and no correction or invalidation is required when the isokinetic ratio is between 0.90 and 1.10. Due to the variabilicy of the waste feed and resultant product gas constituents, the stack temperatures after the flare varied widely* The operators apparently were unable Co keep up with the temperature chances when setting their Homographs. The nonisokinetic* of the involved sampling run* affect* only the collection of parciculace mattar end does not affect the collection of compounds in the vapor state. Vapor state material is only affected by concentration scracificaeion across the s—plinx plane which is alleviated by multipoint sampling across that plane. Parciculace concentrations and emission races can be corrected for a nooisokioetic condition by multiplying these values by the isokinetie ratio. This results in a value that would have been obtained, bad the sample been obtained isokioetieally. This correction is given in Table 7-7.

TABLE 7-7. PARTICULATE HATTER EMISSIONS CORRECTION

Run 3-1 Run 3-2 Run 3-3

Ib/hr concencracioo 0.0952 0.0366 O.U520

Isokioecic ratio 0.8190 1.112 I.U93

Ib/hr, corrected 0.0780 0.0407 0.0568

70

03107752 AB can be seen by the above eorrectiona, the change in emiaaion rafa is negligible and doe* not affect Che data appreciably.

VOST. HCl. Fixed Gaaea

No problema were experienced when a—pling for volatile organiea (VOST), HCl, fixed gases (EPA H3>, or fixed gaaea (CEHa), During Run 3-J, however, approximately 6 minutea of CEM data waa loaC when condenaate built up in cbe sample liae and required purging.

• Chloridea

Quality control procedurea for ehloridea included Cbe analyaia of laboratory control aaaplea and field blanka*

Preeiaion waa ooc •eaaured. Accuracy, •eaaured aa the percent recovery of laboratory control a—plea waa eaciaated at 108 percent. Theae reaulta are presented in Table 7-2. No precision or accuracy goala wre eacabliabed for chloride analyaia* ^ v0 Volatile and Senivolatile Organica /-^

Quality control procedurea for volatile and semivolatile organics decerainationa included the analyaia of •atrix and surrogate spiked a—pie aliauoca. Results of cheae analyses, uaed aa an indication of accuracy, are ° presented in Tables 7-3 through 7-6. QA objectivea are presented alongaide experiaencal values for coaparison. Completeness objectives of 95Z were achieved.

Deviations Froa QA Plan

In an effort to cut progr— coaca, precision of analytical measurements through the analyaia of replicate sample aliquota was not perforoed. Recoveries of aurrogate apiked compounds did not •eec QA objectives in all caaes.

SAMPLE CUSTODY

Sample Chain-of-Custody waa maintained throughout the progr—. Theae procedurea are described in Section 5 of the Quality Aaaurance Project Plan prepared for this progr—. It should be noted that the poac-flare stack gas volatile organic aaaplea were allowed to became warm during shipaent. In addition, the holding tiaea for these samples were exceeded. These a—plea were not analysed aa the results could have been deemed erroneoua and (insupportable.

CALIBRATION PROCEDURES AND FREQUENCY

Calibration procedurea deaeribed in Section 6 of the QA Plan were followed during this sampling program.

71

03107753 ANALYTICAL PROCEDURES

Analytical procedural ai—arixed ia T«bl« 7-2 of the QA Plan were followed during this program. Several con—ones on the QA Plan were oade ia Che February 21, 1986, EPA meao froa Charles Porfert, Deputy Quality Assurance Officer, regarding enelyticel procedure! (coomeacs I, 2, •od 3)* Coaaenc I requested infonsecion on QA objective* for precision, accuracy, and core? Ie ce ness. The analytical QC •eehoda uaed are described in Section 9 of Che QA Plan; percenC recovery objective* are given in Table 3-1 of Section 3; and detection liaits are not reQuired. Cooaenc 2 regarded certain EPA analytical •etboda which abould be uaed for analyaia of aediaent s—plea. Cooaenc 3 requested the uae of EPA Method 314B for bardaeaa. However, GCA did not have the reagent* neeeaaarr Co perfora thia procedure and, ia order not to delay the program. Method 314A waa uaed.

DATA REDUCTION, VALIDATION, AND REPORTING

Data reduction, validation, and reporting procedure* described in Section 8 of the QA Plan were followed during this progran. Ln \0 INTERNAL QUALITY CONTROL CHECKS CT\ in Internal QC procedure* described in Section 9 of the QA Plan were Q followed during thia progr—, with the exception of duplicate •ample analy—J which were not perforaed. •

PERFORMANCE AND SYSTEM AUDITS

Analytical perforaance was audited through the uae of internal quality control check* described ia Section 9 of the QA Plan. In addition, the laboratory participated ia EPA Water Pollution (WP) Study 016 and water aupply (WS) Study 018 during the period of perfozaaaee of thia work* The results of these analyses, for corpouods pertinent to this test progr—, can be found in Appendix B. No syscea audit was conducted during the period of perforaance of this work.

PREVENTIVE MAINTENANCE

Preventive Maintenance procedures described in Section 11 of the QA Plan were followed during this progrea.

ASSESSMENT OF PRECISION, ACCURACY AND COMPLETENESS

Analytical preeiaion waa not reported. Accuracy assessments were based on the results of analyses of EPA Standard Reference Materials and of aacrix spiked samples and reported ia ceras of percent recovery which waa calculated as shown below:

Percent Recovery - 100 /»«""^d Value\ \ True Value /

72

03107754 The following forcula w u«ed Co ••cinate compleeeoef:

c - loo(^) where: C • Pereeoc coapleceoc**,

V • Nuiaber of •e««ur«aenc« judged valid, •nd

T • Toeal ouab«r of ••••ur«oenc«.

CORRECTIVE ACTION

There were ao Corrective Action Requ««c fonai iaicieeed ia regard Co chic proxr—.

73

03107755 REFERENCES

1. Quality Aaaurance Project Plan, "Performance Testing of th« Pyrolyia Sycenr, Inc. Plaaoa Pyrolyia Uhit," Final V«r»ion, February 1985, GCA/T«chnolo(y Oivia ion*

2. "PreaanCaeion of a Method for Cba Selaction of PUHCa in Accordance wich the RCRA InCeria Final Rule, Incineration Standards," January 23, l^bl, Office of Solid Waste.

74 "" — ^SSyCAt KC^OMT PATA

F WOMT NO.

«. TiTkt AMO»U«TlTk< k. »••»&•—-a —6*«i««'rioN eeoc Stack Testing of the Mobilei Plasma Are Unit

».AU+-em»» Mark Go Hands]roanna Hall Howard Schiff Edward Peduco r M6a*A« IklklMV i.6———————— CCA TECHNOLOGY DIVISION, Iwe. 213 Burlington Road Bedford, Massachusetts 01730

<2 •rO—OMl-a AOl-CT CAMC AMO AOC>«1U «3 T»»t of —»ODT ACO ••••00 eevfis •«. f»o

r »y^k(M(MTA«T MOTU ^0 0^ o A trial bura prograi iJrrolxing a plaaaa pyrolysia aye— was eonduetad at th« loyal Milieary Collf. Kingacoo, Ontario* The plaau pyrolyais tail was d«r«lop*d by Pyrolyia 3yc——» Inc* undar coaeract to th« Raw York Stata D«parta«nfc of Ka^riroa—ntal Cona«rratioa (RTSDEC). BVESL-Ci uader a joint agra—nc fuad«d-th« p«rforxaaee Taluacion phaaa.

Trial buraa v«r« coaduetad whila firing carbon tatraehlorida and poly ebloriaafd biph«ayla «rhieh are XCXA wad TSCA regulated eoopouadc reapee- Civaly* Xaaulta froB the trial burna indieata that the ayatex ia acceptable for treating tbaae two cOBpouod elaaaes ia reference to the KCSA (^ 99.991 OBZ) and TSCA (^ 99.99991 DBZ) requireaenta*

«

»». •IT •OttOe A-O OOCUMCNT AMAkVia l ——C«tfTO«U k.lOIMTI»llR«'0»€klMO(BTrM« C. Ce»*TI FwM.-GfMr

fl'ftO O* »«if

M ••CWTT CkAS» iflmmr, iS-Wci—

IP* P— ON»1 (Bw»- 4>m — ——TK

03107757