September 1987 PCS Sediment Decontamination Processes—Selection for Test and Evaluation Project Summary by Ben H. Carpenter Research Triangle Institute Research Triangle Park, NC 27709 Contract No. 68-02-3992 RTI Project No. 471U-3065-65 Project Officers: Donald L. Wilson, T. David Ferguson Hazardous Waste Environmental Research Laboratory Cincinnati, OH 45268 Hazardous Waste Engineering Research Laboratory Office of Research and Development U.S. Environmental Protection Agency Cincinnati, OH 45268 ------- PROJECT SUMMARY PCB SEDIMENT DECONTAMINATION PROCESSES - SELECTION FOR TEST AND EVALUATION by Ben H. Carpenter Research Triangle Institute Research Triangle Park, NC 27709 Contract No.: 68-02-3992 RTI Project No.: 471U-306S-65 Project Officers: Donald L. Wilson and T. David Ferguson Hazardous Waste Engineering Research Laboratory Cincinnati. OH 45268 Hazardous Waste Engineering Research Laboratory Office of Research and Development U.S. Environmental Protection Agency Cincinnati. OH 45268 September 1987 ------- PROJECT SUMMARY PCS SEDIMENT DECONTAMINATION PROCESSES - SELECTION FOR TEST AND EVALUATION Ben H. Carpenter ABSTRACT Eight alternative treatments for PCB-contaminated sediments have been assessed as candidates for immediate thorough test and evaluation. The proc- esses are: Basic Extraction Sludge Treatment (B.E.S.T), UV/Ozone or Hydrogen/ Ultrasonics Technology, Bio-Clean Naturally-Adapted Microbe. Potassium Poly- ethylene Glycolate (KPEG). Low Energy Extraction, MODAR Supercritical Water Oxidation, Critical Fluid Systems (CFS) Propane Extraction, and Battelle In Situ Vitrification. The processes were evaluated using five criteria: the probability of cleaning sediments to 2 ppm or less; the availability of a teat system; the test and evaluation effort required; the time required for future availability of a commercial treatment process; and the probable cost of treatment using t process. These criteria were addressed by engineering analysis of avail- e data and site visits to developers' facilities. The processes were ranked comparatively as to the overall desirability of thorough test and evaluation using all five criteria collectively. Two rating methods were applied: a multiplicative model using a Desirability Function and a linear model, d-SSYS, using weighted utility functions. Both methods converted the process characteristics to ratings on a scale from 0 to 1 (worst to best). The Desirability approach normalized the characteristic using the difference between acceptable and borderline values; d-SSYS normalized the characteristic using the difference between the maximum and minimum values. In calculating the overall score, the factors were weighted equally in the Desirability Function. Probable cost of treatment and test and evaluation effort were assigned weights 4 to 5 times those of the other three character- istics in the d-SSYS ranking. These Independent approaches gave final overall desirability scores as follows: ------- Desirability d-SSYS Process score score Kale Extraction Sludge Treatment, 0.623 0.8127 Resources Conservation Conpany UV/Ozone or Hydrogen/Ultrasonics Treatment. 0.621 0.8010 Ozonic Technology, Inc. Naturally-Adapted Microbes Process, 0.617 0.7583 Bio-Clean. Inc. Potassium Polyethylene Glycolate (KPEG), 0.615 0.7434 Galson Research, Corp. Low Energy Extraction. New York University 0.614 0.4529 Supercritical Water Oxidation. NODAR. Inc. 0.600 0.4738 Propane Extraction Process. 0.590 0.6214 Critical Fluid Systems In Situ Vitrification. Battelle 0.460 0.2299 While all the processes except In Situ Vitrification appear to merit further development for this application, those three with the highest compar- ative ratings are recommended for immediate EPA-supported thorough test and evaluation. These are the Basic Extraction Sludge Treatment, UV/Ozone or Hydrogen/Ultrasonics Technology, and Bio-Clean Naturally-Adapted Microbe proc- esses . This recommendation does not mean that the other processes merit no fur- Ber T and E. The Potassium Polyethylene Glycolate (Galson), Low Energy Extraction (New York University), MODAR Supercritical Water Oxidation, and Critical Fluid Systems Propane Extraction processes rank very close to the top three. Thus these seven processes merit consideration for testing at least through the preliminary phases to confirm their performances. This project summary was developed by EPA's Hazardous Waste Engineering Research Laboratory, Cincinnati. Ohio, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). INTRODUCTION The PCB-contaminated sediment problems in New Bedford. Massachusetts (EPA Region I) in New York State (EPA Region II) and In Waukegan. Illinois (EPA Region V) are reported to be the worst in the nation In terms of PCB concen- tration and the total quantity of PCBs present. In addition, there are numerous industrial lagoons with large quantities of PCB contaminated sedi- ments. PCB contamination poses threats to both drinking water and the fishing istry. ------- The only available proven treatment technology Is dredging and expensive J neration. Land disposal of the sediments untreated has legal restric- B. The Environmental Protection Agency (EPA) has initiated a three-phase research program to identify, validate, and demonstrate effective and economi- cal chemical/biological processes for the removal/destruction of PCBs in sedi- ments. The Phase 1 study screened 64 emerging process technologies and selected eleven for evaluation: Potassium Polyethylene Glycolate with Dimethyl Sulfoxide (Galson Research Corporation); O.H.N. Methanol Extraction; Advanced Electric Reactor (J.M Huber Corporation); Acurex Solvent Wash (Electric Power Research Institute); Bio-Clean Naturally-Adapted Microbe (Bio- Clean. Inc.); Battelle In Situ Vitrification; Light Activated Reduction of Chemicals ((LARC). Atlantic Research Corporation); MODAR Supercritical Water Oxidation: Sollex Solvent Extraction; Sybron Bl-Chem 1006B; and Composting (as studied by the Atlantic Research Corporation) (Carpenter. 1987). The evalu- ation showed the first eight of these to have potential for reduction of PCB concentrations to the desired background levels or less, with minimal environ- mental impacts and low to moderate coat. All of the eight except the Advanced Electric Reactor required further development and testing. Phase 2 study was undertaken to establish suitable factors for fur- assessment of candidate processes, to identify additional data needs, and provide a basis for selection of three processes for a defensible, thorough technical assessment, including laboratory experiemnts and field evaluation. The study involved consultations with treatment process developers, technical assessment of the processes, and the selection of the three highest ranking processes for immediate teat and evaluation. SCREENING OF CANDIDATE PROCESSES The seven candidate processes that required further development and test- Ing were screened at the start of this Phase 2 (Validation) study for avail- ability of a continuing developer and a treatment system for use in test and evaluation of the process. The results of this screening are given in Table 1. Three processes were eliminated from further consideration. The Solvent Wash process is not available for assessment because its sponsor, the Electric Power Research Institute, is seeking a firm to undertake the further needed development of the process before it is ready for further consideration. The ------- Process TABLE 1. INITIAL SCREE '• OF TREATMENT PROCESSES Contact Continuing Developer Test Systeem) Available KPEG OHM Extraction EPRI Solvent Hash8 Battelle Vitrification Bio-Clean. Inc. NOOAR Supercritical Mater Dr. Robert L. Peterson Yes Galson Research Corporation 6601 Klrkvllle Road Bast Syracuse, NY 13057 Sue Maaon No OH Materials 16406 U.S. Rt. 224 E. Plndlay, OH 45839-0551 (419) 423-3526 Ms. Mary McLearn No Electric Power Research Institute 3412 HlllvleN Avenue Palo Alto, CA 94304 Craig TlMeraan Battelle Pacific Northwest Laboratory Yes P. 0. Box 999 Rlchland, HA 99352 Dr. Lance B. Croable, Yes Director of Labs 201 H. Burnsvllle Prkwy., Suite 130G Burnsvllle. MN 65337 (612) 890-1118 Ralph A. Morgan Yes Modar, Inc. 3200 Wllcrest, Suite 220 Houston, TX 77042 (713) 785-5615 Yes No No Yes Yes Yes (Continued) ------- Process TABLE 1 Continued) Contact Continuing Developer Test Syste*(s) Available LARC Basic Extraction Sludge Treatment CP Systeas Propane Extraction Ultrasonics/UV Technology Low Energy Extraction Process George Anspach Atlantic Research Corporation 5390 Cherokee Avenue Alexandria, VA 22312 Mark lose Resources Conservation Co. 3101 N.E. Northup Hay Bellevlew, HA 08004 (208) 828-2376 Thomas J. Cody, Jr. CP Systems Corporation 25 Acorn Park Cambridge MA 02140 (817) 402-1631 Edward A. Pedzy Ozonic Technology, Inc. 90 Herbert Avenue P. 0. Box 320 Closter, NJ 07624 (201) 767-1225 Halter Brenner/Barry Rugg New York University Dept. of Applied Science 26-36 Stuyvesant Street New York. NY 10003 (212) 508-2471 No No Yes Yes Yes Yes Yes Yes Yes Planned aThis process was identified as the Acurex process in the Phase 1 study. ------- developer of the OHM Extraction process has chosen not to invest In this proc- Is. The developer of the LARC process has not identified sufficient Markets d the process is not available from the*. Meanwhile, four technologies not assessed in the Phase 1 study have become available: the Basic Extraction Sludge Treatment (B.E.S.T.) process (Resources Conservation Company); the Critical Fluids Systems Propane Extrac- tion process; the (JV/Ozone or Hydrogen/Ultrasonics process (Ozonic Technology. Inc.); and the Low Energy Extraction process (New York University). These have all been assessed as candidates for thorough test and evaluation. The UV/Ozone or Hydrogen/Ultrasonics Technology provides continuity for the radiant-energy approach previously represented by the LARC process. The other three processes provide improved approaches to the extraction technology. The results of Initial screening of these processes are also given in Table 1. DESCRIPTIONS OP TREATMENT PROCESSES The Basic Extraction Sludge Treatment process pretreats the feed (sedi- ment, sludge, oily contaminants, water) with an alkaline composition, then admixes with triethylamine (TEA) while cooling below the critical solution temperature (U.S. Patents 3899410. 3925201. 4002562. 40S6466). A single Huld phase is formed from which the solid matter is separated. The liquid is then heated to above the critical solution temperature to form an amlne- rich phase and a water-rich phase, after which the water phase is decanted. The amine phase contains all of the oil contaminants. It is processed to recover the oil and contaminants, and the TEA is recycled for the processing of additional material. The UV/Ozone or Hydrogen/Ultrasonics Technology Involves three factors, all of which have been shown to be effective. The UV/Ozone technology has been demonstrated for destroying PCBs in industrial waste waters (Arisman et al.. 1981). PCBs have been extracted from soils using Tween 80 surfactant (Scholz and Mllanowski. 1984). They have been removed from metallic surfaces using surfactants and ultrasonics (Smith and Sltabkhan. 1986). UV/Hydrogen has been shown to destroy PCBs in non-aqueous solvents (Kitchens et al., 1979, 1984). Ozonics Technology. Inc. Is equipped to apply all three factors in a comparative evaluation of ozone vs hydrogen. The Bio-Clean process utilizes selected naturally adapted microbes to destroy PCBs under aerobic conditions. Contaminated sediment is charged as a ------- water slurry to a digester. The charge is adjusted to optlaua pH and heated ^a extract the PCBs. Surfactants May be added to promote extraction. After Brcraction, the slurry is cooled, neutralized, and inoculated. The PCBs are degraded in from 48 to 72 hours depending on the process conditions and the •icrobes employed (Bio-Clean. Inc.. 1985) (Crombie, 1986, 1987) (Unteraan, 1985). The Potassium Polyethylene Glycolate process degrades PCBs by nucleophlllc substitution. An equal volume of contaainated sediment and reagent are blend- ed in a reactor, and heated to remove excess water and promote the reaction. The Galson version of the process promotes extraction of PCBs with dimethyl sulfoxide (Peterson, 1986). The treated sediment is settled, the reagent removed by decantation, and the solids are washed with water (Research Demon- stration Permit Application, 1987). The Low Energy Extraction process Involves separation of water from the sediments, a solvent leaching with a hydrophylic solvent (e.g. acetone) usually carried out in countercurrent stages, and transfer of the leached organic contaminants to a hydrophobia solvent (e.g. kerosene) in which it Is concentrated for final destructive treatment. The final treatment is a «rate process. Residual contaminants in the water stream are adsorbed onto ntaminated sediments. The system recycles all solvents and returns only decontaminated sediments and water to the environment (Brenner et al., 1986). The MODAR Supercritical Water Oxidation process feeds a water slurry of contaminated sediments together with liquid oxygen to a pipe reactor where, at 400-600 *C and 22-25 MPa, the contaminants dissolve and react rapidly with the oxygen. The reactor effluent is cooled by heat exchange with fresh feed. Pressure let down and separation of sediments, liquid, and gases is carried out in multiple stages to minimize erosion of valves and optimize equilibria (Staszak et al.. 1987). The Critical Fluid System Propane Extraction process uses propane at ambi- ent temperature and 1.8 MPa (200 lb/in2) to extract PCBs along with other oily organlcs from a water slurry of the sediment. The batch extraction is repeat- ed as required to achieve specified reductions in contaminants. The treated slurry is discharged after separation from the liquid propane which contains the dissolved contaminants. The propane solution is fed to a separator where the solvent is removed by vaporization and recycled. The contaminants are off for final treatment in a separate process. ------- The Battelle In Situ Vitrification process was developed to stabilize fdionucllde-contamlnated soils by melting into a durable glass and crystal- ne fora (Buelt et al.. 1987). Submerged sediments are dredged and relocated for treatment. Four electrodes are Inserted into the sediments in a square array. A path for electric current is made by placing a mixture of graphite and glass frit between the electrodes. Dissipation of power through the starter materials creates temperatures high enough to melt a layer of sedi- ments, which establishes a conductive path. The molten zone grows downward through the contaminated soil. At the high temperatures created (> 1700 *C) organic materials pyrolyze, diffuse to the surface, and combust. Off-gases are collected, monitored, and treated (Tlmmerman, 1986). DEVELOPMENT OF EVALUATION CRITERIA The following criteria -were used to select treatment processes for thorough test and evaluation: 1. The likelihood that the process will acceptably clean up the PCB- contaminated sediments; 2. The probable cost of the application of the treatment after perform- ance is proven: 3. The relative level of T and E effort to be supported by the Environmental Protection Agency; 4. The availability of a processing system to test; and 5. The likely future commercial availability of the process. The standard selected for acceptable cleanup is a PCB concentration in treated sediments (or soils) of 2 ppm or less. EPA has considered PCB requirements preparatory to their promulgation as an amendment to the PCB regulations (40 CFR, Part 781, April 2, 1987, pp. 10688-10710). The proposed standards require cleanup as follows. 1. Nonrestricted area — <, 10 ppm plus 10" cap of soil £ 1 ppm. 2. Restricted area — 25 ppm. 3. Outdoor electrical substation — 25 ppm, or 50 if a warning sign is maintained. These standards would apply to PCB spills occurring after promulgation. occurring earlier, and spills which are apt to result in spread of ------- PCS's into other media (groundwater, surface waters, grazing landa. and Ketable gardens) are to be decontaminated to requirements established at the cretlon of the EPA regional officer. These levels (10 ppm. 25 ppm, and SO ppm) are to be attained by removing all contaminated soil exceeding these levels. The removed soil Is subject to disposal regulations: cleanup to <2 ppm. For this reason, permits issued for alternative destruction processes generally will require that all treated materials and by-product waste streams must have PCB concentrations of less than 2 ug/g resolvable chromatographic peak (2 ppm). If this condition is not met, the effluents containing 2 ppm or greater must be disposed as if they contained the PCB concentration of the original Influent material. If the PCB feed material being treated by the process is over 50 ppm PCB, then the resulting effluents must be incinerated unless an analysis is conducted and indicates that the PCB concentration is below 2 ppm per PCB peak. In accordance with these policy and treatments requirements, the cleanup target for alternative treatments has been set at <2 ppm PCB. The probable cost of treatment Is presented as the cost per cubic meter of sediment treated, based on a system sufficiently large to process 380,000 m3 « Hudson River sediments in 2.5 years. By focusing on a specific site and e of cleanup task, each process could be assessed using available data from sampling and analyses to characterize the feed materials to the processes, and comparative cost estimates for a specific application could be obtained. The sediments from the Hudson River present a variety of sediment types for test- ing PCB-treatment processes. The probable cost of treatment was obtained from the developers for those processes sufficiently supported by commercial firms, or was estimated using as major cost elements capital costs, and operation and maintenance costs. Treatment process requirements determined capital, energy, and maintenance costs. Labor rates, overhead, contingency, profit, and health and safety were costed using standard unit values for all the processes. Since no full-scale systems exist for the processes under assessment. capital costs were estimated by designing a full-scale system, utilizing the data available as a basis. Equipment costs were then obtained as budget esti- mates from manufacturers or developers, or estimated using the method of exponents: ------- Cf - Cj (Qi/Q!)n where: Cj - coat for ith capacity (size); Qj - itn capacity; n - empirical constant; Cj - cost of reference capacity; and Ql - reference capacity. Values for the reference capital cost and the exponent, n, were obtained in part from the literature and In part from equipment manufacturers. Labor hours were estimated based on an automated industrial chemical proc- essing plant (Peters and Tlmmerhaus, 1980): e2 - «i (Q2/Ql)n where: ej - Operator hours per day and processing step of reference case: 62 • Operator hours per day and processing step of case 2; Qj - Process capacity of reference case; Q2 - Process capacity of case 2; and n • Empirical constant. The values used in this evaluation are: ej - 18 h/d x step; Q! - 9.07 mt/d n - 0.22 The number of foremen and chemists are taken to be 15 percent of the num- ber of operators. In addition to these workers, there is one site manager. The hourly wages are assumed to be: Operators: 15 $/hr Foreman: 18 $/hr Chemist: 25 $/hr Manager: 60 $/hr Maintenance was estimated at 10 to 15 percent of the capital cost, depend- ing on the number of unit operations Involved and engineering practices for the operation. The allowance for safety equipment was generally $0.30/»3 of sediment treated ($114.000 for the cleanup of 380,000 m3 of sediment). 10 ------- The treatment system cost estimate was capitalized (recovered) over the K years of operation taken as the base period. Some developers provided atment costs that were correspondingly lower for subsequent applications. The test and evaluation effort required was estimated based on a compari- son of available process data with the requirements for thorough test and evaluation. The following information must be provided to qualify a process for a permit to test: 1. Waste characteristics; 2. Process engineering description; 3. Sampling and monitoring plan; 4. Accident and spill prevention and countermeasure; and 5. Demonstration test plan. For these assessments, Hudson River sediments were selected as the character- ized wastes. Hudson River sediment samples have been classified according to their con- tent of clay. silt. muck, muck and wood chips, sand, sand and wood chips, «se sand, and coarse sand and wood chips (Tofflemire and Qulnn, 1979). ments have been shown to range from clay to cobbles, with the largest mass fraction being in the sand sizes. The highest PCB concentration was in the muck with wood chips class, which typically had over 30 percent silt and clay, high volatile solids and some small but visible wood chips. The size lowest in PCB was medium sized sand or gravel without wood chips. The coarse fraction (>0.42 mm) of the sediments typically contained wood chips, sawdust, shale chips, cinders, and coal fragments. The fine size frac- tions contained some fragments of the above, plus sand (containing quartz and feldspar), silt, clay, and organic material. Process engineering descriptions were developed for each process assessed. These varied in completeness because the processes varied in stage of develop- ment from conceptual to field tested. While unit operations were identified and described for all processes, the descriptions were based only on per- formance requirements. Detailed equipment specifications have not been made. except where necessary to obtain cost estimates (e.g., high pressure compres- _fk| and slurry pumps). 11 ------- The descriptions included process flow diagrams and identified all product K waste streams. Additional process information included summaries of bench ts, pilot tests, and field tests, if available. Sampling and monitoring plans were then developed, baaed on the scale of process tests required, the purposes of the tests, and the extent of data needed to characterize the process performance and scaleup the system to full- scale. Some of the processes, when the developers' prior experience justifies it, can be scaled-up from bench-scale teats. Thus, the size of system indi- cated for test and evaluation (T and E) is the size the developer feels can be scaled-up with confidence. For the needed testa, the extent of sampling and analyses was indicated. Methods of analysis were specified. Prom the infor- mation developed. T and E coats were estimated. For most of the processes assessed, test systems are available from the developer. Most developers would need financial support of the testing time and effort. The availability of a suitable test system and any conditions/ restrictions on its use were considered for each process. Accident and spill prevention and counter-measures needs have been identi- fied. Part of the estimated treatment cost is allocated to these factors. The processes vary as to the strength and extent of their sponsorship. ^He developers are commercial firms in the waate treatment business with resources committed to further commercialization of their process. Some developers have a need for financial support to achieve commercialization. In all cases, the short-term (2.5 years) effort projected in this evaluation, and the uncertainties of further markets make the construction of a full-scale treatment system contingent upon completion of the T and E (with attendant EPA approval of the process) followed by a contract for the cleanup work itself. Under these necessary conditions, all the processes assessed would be commer- cially available. The estimated time required to make them available varies from process to process, however, and this has been taken into consideration in this evaluation. PROCESS EVALUATION All the processes assessed have merit. In selecting among them, a ranking system has been employed for comparative simultaneous evaluation of all five criteria characteristics (Harrington, 1965). For each process, the 12 ------- desirability of immediate thorough test and evaluation was expressed by a r ability value. Dj: where: djj = the rating of process J for criteria i for 1 from l to 5, 0.0 < d <. 1.0. This function is a multiplicative decision model . although it »ay be regarded as a linear model if it is utilized in its logarithmic form. The methodology does not in its original form provide for weighting of the factors involved. Instead, it applies equal weights to all factors. Each factor may be weighted, however, by the application of an appropriate exponent. This is shown as follows: 0 - (d Xld uj, modified valj a2j 3J "• nj where: xj - the weight of factor i The logarithmic form is: (3) U + X2 lo* d2J * X3 lo« d3J * ' ' ' * xn n 0 < d < 1 The value found for each characteristic, y. was transformed to a value of d according to the following judgements: Value of d 1.0-0.99 Represents the ultimate level of the characteristic y. Improvement beyond this point would have no appreciable value . 0.99-0.80 Acceptable and excellent. Unusually good performance. 0.80-0.63 Acceptable and good. 0.63-0.40 Acceptable. Some Improvement is desirable. 0.40-0.30 Borderline acceptability. 0.30-0.01 Unacceptable. This one characteristic could lead to rejection of the process. 13 ------- The scale of d so developed is a dimensionless scale to which any charac- Krlatic nay be transformed so that it »ay be Interpreted In terms of its sirability for the intended application. In this evaluation, the most cost- effective final process was sought that could be available in the shortest reasonable time. A characteristic assessed on a numerical scale was transformed to the scale of "d" by the basic equation: dl . e-e0.77941[(-yi + yn)/(ylh - yu) In this equation: yj is a value of a treatment process characteristic 1; is the acceptable valuable of yl; and is the borderline value of y<. (4) Table 2 shows the acceptable and borderline values of yj for each charac- teristic rated. TABLE 2. ACCEPTABLE AND BORDERLINE VALUES FOR PROCESS CHARACTERISTICS Character i s 1 1 c Probability of cleaning to £ 2 ppm Probable cost of treatment. $/m3 T and E effort, S/1000 Test system availability, rating Time to provide commercial system, months Acceptable Value* 0.9 100 300 0.9 IS Borderline Value b 0.3 300 900 0.3 36 ad « 0.63 for these values. bd * 0.37 for these values. The probability of cleaning to £ 2 ppm was set at 0.9 If such performance had been demonstrated with soils of any type, 0.8 if such performance was projected from test data, and 0.3 if no data were available. The use of 0.9 and 0.8 distinguishes slightly between processes reaching the goal of < 2 ppm on initial tests and those for which reaching this goal could be projected from initial teats. The probable cost of treatment was considered acceptable if $100/n3, and borderline if S300/m3. T and E effort was considered acceptable up to $300,000. Values above $900,000 were considered borderline, but could be justified if the process had 14 ------- potential for lowering the cost of treatment, or rated extremely well on other fepired characteristics. Test system availability was rated 0.9 for an available company-provided system with experienced operating staff and resources to commercialize the process; 0.8 if further government purchases were required to provide a system; and 0.3 if a suitable test system were not available. Time to provide a commercial system sized to effect cleanup of 152,000 m3 of sediment per year was considered acceptable at 18 months, but borderline if 36 months were required. Using the values of the characteristics shown in Table 2, Equation 4 was applied to calculate the individual ratings shown for each process in Table 3. This table also shows the overall desirability rating of the process, calcu- lated using Equation 1. All the processes show acceptable "D" values. The Basic Extraction Sludge Treatment, UV/Ozone or Hydrogen/Ultrasonics, and Bio-Clean Naturally-Adapted Microbe processes show the highest values, and are recommended for immediate test and evaluation. This recommendation does not mean that the other processes merit no fur- ^fe* T and E. The Potassium Polyethylene Glycolate (Galson). Low Energy Extraction (New York University). NOOAR Supercritical Water Oxidation, and Critical Fluid Systems Propane Extraction processes rank very close to the top three. Thus these seven processes merit consideration for testing at least through the preliminary phases to confirm their performances. The Potassium Polyethylene Glycolate process rates lower primarily because of the high esti- mated cost of treatment. The Low Energy Extraction process has a low rating because of the rela- tively higher cost of development that may be necessary and the length of time to commercialization. Uncertainty about the possible commercial sponsorship led to the lower rating for availability of a test unit. The $827,000 Test & Evaluation cost includes the cost of a pilot unit, however. For the other processes, this cost was not Included because it was contributed by the developer or had already been purchased by the government. The estimated cost of treatment using this process is lower, however, than any other process. If this estimated cost could be attained, the development cost for the process adds only $2.17/n3 of sediment treated: 15 ------- T*ft£ 3. MMU. OSIMBIUTY OF MQIATE T AND E OF THE HOT OMIMTC PBggg Probability of clean- Ing tD<2ppe d ratlfig FT09HDM GDft OT . »i_3 Grenenc, J/B* d rating T and E ei'ori $1000 d rating Availability of a fyetee for a tart jtura purdne by govern, raoulrad futerapunJuee by govern, not required d rating Lflerty futura avall- etnlTty or tre prooeei OMammmm BUIUB d rating OMrall detlrablllty, 0 erllett futura evedl. latBt future avail. KFGB. feleon 0.9 0.63 160-191 0.54* 216 0.66 0.9 0.63 19.5 0.62 0.615 0.615 0.615 fcdar Supercritical Hater O.I 0.59 86-136 0.62 483 0.56 0.9 0.63 21.5 0.59 0.60 0.60 0.60 Bio-Clem 0.8 0.63 156 0.57 166 0.68 0.9 0.63 19 0.62 0.617 0.617 0.617 W/tam- HydrogW Ultraetnla Tadnology 0.8 0.59 90-120 0.63 151 0.69 0.9 0.63 21-24 0.59-0.55 0.625 0.616 0.621 CfS Extraction 0.8 0.59 153-264 0.50 123 0.69 0.9 0.63 25 0.54 0.59 0.59 0.59 u»*»w 8.E.S.T. Extraction 0.8 0.9 0.59 0.63 133 50-57 0.59 0.68 149 ITO-eZT* 0.69 0.64 0.8 0.9 0.63 0.59 19 25 0.62 0.54 0.623 0.614 0.623 0.614 0.623 0.614 InSItu Vltrlflcsti 0.9 0.63 443483 0.16 400 0.59 0.8 0.59 19-24 0.62-0.55 0.46 0.45 0.46 of $170,000' 1f by epmearlng fin. A «t of $280,000 *» «a 1n the ^k«1m to allot fcr tfe uorealnty. 16 ------- $827.000 + 380.000 »3 - $2.17. •Mis would have a small impact on the final treatment cost. If added to the estimated $76-$83/m3. From this point of view, the major obstacle to a higher rating for this project is the 25-month development time required, and the lack of a firmly committed commercial sponsor for the test system. The NOOAR process has a high T and E cost, but a potential application to a broad range of contaminants besides PCBs. The In Situ Vitrification process has the highest estimated cost of treat- ment, which derives in significant proportion from the cost of electricity and consumable electrodes used in the treatment. As previously mentioned, the advantages of in situ treatment could not be bad in the treatment of submerged sediments. The fact that a solid mass is the product presents a problem in disposal. The process appears best suited for in-situ fixation of radioactive wastes. APPLICATION OF ALTERNATIVE PROCESS SELECTION METHODOLOGY While this project was under way, an alternative process selection method- ology became available at HWERL. The methodology is available as a computer •Bgram entitled "d-SSVS. A Computer Model for the Evaluation of Competing Uncertainties," (Klee. 1987). This method was applied In addition to the Desirability Function approach. The D-SSYS calculates weights for each evaluation factor using values of weight ratios assigned by the user. Weight ratios were assigned to emphasize the Importance of the ranges of treatment cost and test-and-evaluatlon cost. The range of ratings for probability of cleaning to <2 ppm PCBs is only 0.1 (Table 3), Indicating that all the processes might reasonably be expected to meet the requirement. The availability of a test system was not considered as Important as the total test and evaluation cost. The time required to make a commercial process available showed a range of only six months, and was judged of lesser importance than the two major costs assessed. All ratios among the five factors that resulted from these assignments are shown below as a matrix. For example, the ratio (test system avallablllty)/(T and E cost) is shown as the intersection of Row 4 and Column 3 as 0.2. 17 ------- Clean to 2 ppm Cost T and E Cost Test system availability Early com. availability Clean to 2 ppm 1 5 S 1 1.25 Cost 0.2 1 1 0.2 0.25 T & E Cost 0.2 1 1 0.2 0.25 Test System Availability 1 5 5 1 1.25 Early Commercial Availability 0.8 4 4 0.8 1 Prom these ratios and the following tabular algorithm, the factor weights (W) were generated. Factors Clean to 2 ppm T and E Cost Future commercial proc. Test system availability Cost Ratios 0.2000 4. 1. 0. 000 25 2000 0.20 1.00 0.25 0.20 5.000 Weights. 0.0755 0.3774 0.0943 0.0755 0.3774 2.65 The procedure for weight generation is as follows: Construct an Intermediate weighting scale (the w-column) by the follow- ing procedure. Opposite that last factor enter a "1". The remaining numbers in this column are formed by the product of its predecessor and Ratio value opposite it in a sort of zigzag route up the column. For example, the first w-value, 0.20 is the product of the second w-value (1.00) and the first Ratio-value (0.2000). • Total the w-values. This total is 2.65. Construct a column of stan- dardized weights by dividing each element of the w-column by this total to obtain the W-column. The elements in the W-column will, perforce. total one. The program then scales the factor scores to obtain a linear utility func- tion: y1 « b0 + bj (factor score) 0 < y1 < 1. (5) In applying the scaling procedure to the two factors "Probability of Cleaning to £ 2 ppm" and "Availability of a Test System," It is noted that these are positive factors (the higher the factor score, the better the proc- "s_L and the 'y1 values are obtained by: 18 ------- score}j - •inimua score}4 maximum score£j - minimum scorejj (6) The other three factors are noted to be negative (the higher the factor score, the worse the process) and the y1 values are obtained by: maximum scorej.| - acorejj maximum scorejj - minimum score ij (7) The yjj values for the five factors by which each of the eight processes were assessed are given In Table 4. TABLE 4. SCALED RATINGS OF EIGHT TREATMENT PROCESSES Clean 2 KPEG (Gal son) MODAR Bio-Clean UV/OZ or H2/Ultrasonlcs § Propane Extraction .S.T. Energy Extraction In Situ Vitrification to pp» 1 0 0 0 0 0 1 1 Probable Tr. Cost 0.60 0.87 0.67 0.89 0.45 0.77 1 0 T & E Cost 0.75 0.043 0.89 0.93 1 0.93 0 0.26 Test System Availability 1 1 1 1 1 1 0 0 Early Commercial Availability 0.92 0.58 1 0.42 0 1 0 0.58 Depending on the users degree of risk that he is willing to accept. d-SSYS fits a utility function to the y1 values via the following function: utility « y'f . (8) The exponent f is evaluated by presenting the user with a structured lottery. d-SSYS requires s comparison between two simple lotteries for each factor rated. Lottery 1 SOX chance of most undesirable rating. SOX chance of most desirable rating. 19 ------- Lottery 2 « | X value of the ratine for certain. I Using probable treatment cost as an example, RTI selected for Lottery 1: SOX chance of a treatment cost of $313/m3 SOX change of a treatment cost of $80/m3 and an X value equal to the mathematical expectation of Lottery 1 for Lottery 2: (0.5 X $313) + (O.S X 80) « $196.50/m3. The value of $196.50/m3 on the y' scale is $313 - $196.6 y« - « 0.5 $313 - $80 The utility of Lottery 2 is easily determined, since it is equal to the utility of Lottery 1: (0.5)(utility of $313/m3) + (0.5)(utility of $80/m3) - (O.S x 0.0) + (0.5 x 1) - 0.5. ^^m Equation 8: f - (In utility)/ln y' (9) f - (In 0.5)/ln 0.5 • 1 Note that if Lottery 2 had been set at a lower cost for certain, f would have been greater than 1 and the function would have been a risk-taking one, in that one would be willing to pay more for Lottery 1 in the hope of gaining a treatment cost of $80/m3. The remaining utilities for each factor are then calculated using Equation 8 (Klee, 1987, p. 23). Using the ratings scaled by Equation 8, the program computes an overall deterministic score for each treatment process as the sum of the scaled factor ratings times the scaled factor weights. Using the factor ratings of Table 44 (which equal the utility when f - 1) and the weights cited above, the follow- ing deterministic scores were obtained for the treatment processes (Table 5). 20 ------- TABLE 5. DETERMINISTIC SCORES FOR TREATMENT PROCESSES Proce88 Score Basic Extraction Sludge Treatment 0 8127 UV/Ozone or Hydrogen/Ultrasonics 0 8010 Bio-Clean Naturally-Adapted Microbe 0 7583 Potassium Polyethylene Glycolate. Galaon 0 7434 Critical Fluid Systems (CFS) Propane Extraction 0 6214 MOOAR Supercritical Water Oxidation 0*4738 Low Energy Extraction. New York University 0 In Situ Vitrification. Batteile ' The highest scores were attained by the Basic Extraction Sludge Treatment, UV/Ozone or Hydrogen/Ultrasonics Technology, and Bio-Clean Naturally-Adapted Microbe processes, the same processes that ranked highest using the Desir- ability Function ranking methodology. These are recommended for Immediate test and evaluation. In the application of this ranking, probable treatment cost and test and evaluation cost were assigned weights 4 to 6 times those of the other three gctors. This Increased emphasis on the costs involved did not change the top Pree processes. With different weights assigned, it would be possible to obtain a different ranking of the processes. CONCLUSIONS Eight emerging treatment processes for decontamination of PCB-contaminated sediments have been evaluated as candidates for thorough test and evaluation (T and E) using a test system judged of sufficient size by the developer to provide performance, cost, and sealeup data for a large commercial plant. The processes assessed include: Basic Extraction Sludge Treatment (B.E.S.T); Bio- Clean Naturally-Adapted Microbe; Critical Fluid Systems Propane Extraction; Potassium Polyethylene Glycolate. Galson; Low Energy Extraction. New York University; MODAR Supercritical Water Oxidation; UV/Oxone or Hydrogen/Ultra- sonics Technology; and Batteile In Situ Vitrification. The processes were evaluated using as criteria: • The probability of cleaning sediments to <2 ppm PCBs; • The probable cost of treatment: 21 ------- • The relative level of Test and Evaluation effort to be supported by EPA; • The availability of a processing system to test; and • The likely future commercial availability of the process. While all the processes except perhaps In Situ Vitrification merit further development for treatment of sediments, comparative simultaneous evaluation of their ratings on a scale of 0 to 1 gave the following results: Relative Desirability of Thorough Test and Evaluation Desirability d-SSYS Process score score Basic Extraction Sludge Treatment, 0.623 0.8127 Resources Conservation Company UV/Ozone or Hydrogen/Ultrasonics Treatment. 0.621 0.8010 Ozonic Technology, Inc. Naturally-Adapted Microbes Process, 0.617 0.7583 Bio-Clean. Inc. Polyethylene Glycolate (KPEG), 0.615 0.7434 Research, Corp. Extraction. New York University 0.614 0.4529 Supercritical Water Oxidation, MODAR, Inc. 0.600 0.4738 Propane Extraction Process, 0.590 0.6214 Critical Fluid Systems In Situ Vitrification, Battelle 0.460 0.2299 The Basic Extraction Sludge Treatment Process (Resources Conservation Co.), UV/Ozone or Hydrogen/Ultrasonics Technology, and Bio-Clean Naturally- Adapted Microbe processes have the highest ratings, and are recommended for immediate thorough test and evaluation. This evaluation was confirmed using the d-SSYS Computer Model for the Evaluation of Competing Alternatives (Klee. 1987). The Potassium Polyethylene Glycolate (Galson). Low Energy Extraction (New York University), MODAR Supercritical Water Oxidation, and Critical Fluid Systems Propane Extraction processes ranked very close to the top three. These processes have potential for treatment of a broad range of hazardous contaminants and are recommended for at least those preliminary phases of ^ough test and evaluation which confirm performance and establish process Peters for pilot-scale tests. References cited are identified fully in the full report. 22 ------- Ben H. Carpenter is with the Research Triangle Institute, Research Triangle Park. NC 27709. Donald L. Wilson and T. David Ferguson are the EPA Project Officers (see below). The complete report, entitled "PCB Sediment Decontamination Processes - Selection for Test and Evaluation." (Order No. PB , Cost , subject to change), will be available only from: National Technical Information Service 5285 Port Royal Road Springfield. VA 22161 Telephone: (703) 487-4650 The EPA Project Officer can be contacted at: Hazardous Waste Engineering Research Laboratory U.S. Environmental Protection Agency Cincinnati. OH 45268 23 ------- |