A Review of the Role of Plastics in Energy Recovery

Transcription

1 A Review of the Role of Plastics in Energy Recovery Prepared on behalf of: the Environment and Plastics Industry Council January 1997

2 This report was sponsored by the Environment and Plastics Industry Council (EPIC, formally the Environment and Plastics Institute of Canada and now a Council of the Canadian Plastics Industry Association). It was prepared on behalf of EPIC by Proctor & Redfern Limited, a consulting engineering and environmental management firm located in Toronto, Ontario. Although EPIC has endeavoured to provide accurate and reliable information to the best of its ability, it cannot be held liable for any loss or damages resulting from interpretation or application of this information. This information is intended as a guide for use at your discretion and risk. EPIC cannot guarantee favourable results and assumes no liability in connection with its use. The contents of this publication, in whole and in part, may not be reproduced or transmitted without proper accreditation to EPIC. EPIC is a Council of the Canadian Plastics Industry Association, 5925 Airport Road, Suite 500, Mississauga, Ontario L4V 1W1. Telephone: (905) Fax: (905) ISBN # TABLE OF CONTENTS 2

3 1.0 INTRODUCTION EMISSIONS STANDARDS AND GUIDELINES INVESTIGATIONS WURZBURG STUDY SELCHP STUDY PITTSFIELD STUDY EBARA STUDY UMEA STUDY UNIVERSITY OF FLORIDA TESTS NESTE STUDY SUMMARY OF RESULTS PROCESS PERFORMANCE HEAVY METALS AND PARTICULATE MATTER EMISSIONS DIOXIN AND FURAN (PCDDS AND PCDFS) EMISSIONS HYDROGEN CHLORIDE EMISSIONS CONTAMINANT CONCENTRATIONS IN SOLID RESIDUES CONCLUSIONS REFERENCES

4 ABSTRACT This report reviews the major scientific investigations that have been carried out world wide over the last ten years on the effect of plastics on the process and emission performance of energy from waste (EFW) facilities. Concentrations of contaminants in air emissions and other process residues measured during tests with added plastics are compared with concentrations during normal operations and with the stringent emission limits stipulated in municipal waste combustor standards and guidelines in Canada, the United States and the European Union. The emission limits in these standards and guidelines are based on the use of the best technology available (Maximum Achievable Control Technology (MACT) standards in North America and Best Available Technology (BAT) standards in Europe) and are amongst the most stringent in the world. The addition of plastics to feedstocks of EFW plants appears to improve combustion and waste burn out. The majority of studies reviewed found that the addition of plastics had no effect on stack emissions of particulate matter, heavy metals, and dioxins and furans (PCDD/Fs). While the addition of plastics did result in higher chlorine input and consequently higher concentrations of hydrogen chloride, in the raw gas, no correlation between chlorine input and PCDD/Fs could be deduced. Further, emissions of hydrogen chloride at facilities employing lime injection for HCl control were not discernibly higher than baseline. Similarly, results of the studies reviewed showed that the addition of plastics had no effect on the concentrations of heavy metals or trace organics in solid residues from combustion. The comparison with emission limits in municipal waste combustion standards and guidelines in Canada, the United States and Europe indicates that modern energy from waste facilities are able to meet these limits even when additional plastics are added to those normally found in municipal solid waste (MSW). This report will be of interest to municipalities, government waste management officials, private sector firms and concerned citizens in the examination of the relative environmental impacts of different recovery and disposal scenarios for particular material streams. EPIC expects that this report will be of value to the development of an integrated approach to waste management. 1.0 INTRODUCTION 4

5 The management of post consumer plastics has been the focus of considerable attention in waste management. Rapid developments in plastics technology have led to the use of plastics in innumerable applications ranging from candy wrappers to high strength engineered plastics for industrial use. The large number of different resins and resin composites found in plastic products, limit the extent to which these products can be handled by mechanical recycling programs which require the segregation of resins. On the other hand, because of their high calorific value, the recovery of energy from plastic wastes represents a potentially efficient management alternative, particularly as it would allow the management of residual plastics i as a single stream. Public concern with respect to emissions of toxic organics and heavy metals from waste combustion has limited the growth of the energy from waste (EFW) industry in North America and is manifest in the strict regulatory limits that have been imposed on waste combustion in all major jurisdictions in the world. Technology has developed to meet the stringent environmental performance required of these facilities and modern EFW plants are able to meet the most rigorous air emission standards. This report reviews the results of four full-scale and three pilot-scale investigations carried out over the last ten years to determine the effect of plastics on the process and emission performance of EFW plants. The objective of the review is to provide a sound scientific basis for determining the role of plastics in energy recovery programs. Concentrations of contaminants in air emissions and other process residues measured during tests with added plastics are compared with concentrations during normal operations and with the emission limits in standards and guidelines for municipal waste combustion in Canada, the United States and Europe. The emission limits used for comparison are amongst the most stringent in the world. i Waste plastics remaining after 3Rs. 5

6 2.0 EMISSIONS STANDARDS AND GUIDELINES Concentrations of contaminants in air emissions measured during tests with added plastics are compared with emission limits in standards and guidelines for municipal waste combustion in Canada (Ontario Guideline A-7, OMOEE (1995)), the United States (Maximum Achievable Control Technology (MACT) standards promulgated under the Clean Air Act, U.S.EPA (1995)) and Europe (German 17.BImSch V, (1990) and EC (1989)). Table 1 shows the emission limits for particulate matter, acid gases, heavy metals and dioxins and furans in the jurisdictions considered. Contaminant Table 1 Emission Limits for Municipal Waste Combustion Canada (Ontario Guideline A-7) 1 United States (MACT Standards) 2 Emission Limits Germany (17. BImSch V) 4 Particulate Matter (mg/m 3 ) (daily mean) Acid Gases HCl (mg/m 3 ) 27 or 95% reduction 25 ppmv or 95% reduction SO 2 (mg/m 3 ) ppmv or 80% Heavy Metals (ug/m 3 ) Cd: 14 Dioxins and Furans ng/m 3 (TEQ) Hg: 57 Pb: 142 reduction Cd: 20 Hg: 80 Pb: (half-hour mean) 10 (daily mean) 60 (half-hour mean) 50 (daily mean) 200 (half-hour mean) 0.14 ng 13 ng (total congeners) Cd+Tl: 50 Hg: 50 Sb+As+Pb+Cr+C o+cu+mn+ni+v+ Sn: 500 Europe (EC Directive) Cd + Hg Ni + As Pb+Cr+Cu+Mn: Notes: 1. Reference conditions for Ontario Guideline A-7 limits are: 11%O 2, 25 C, kpa, dry gas. 2. Reference conditions for U.S. MACT standards are: 7% O 2, 20 C, kpa, dry gas. The Ontario Guideline A-7 limits and U.S. MACT standards are approximately equivalent when converted to the same reference conditions. 3. The U.S. limit for dioxin is stated in the standards in terms of total congeners. The Background Information Document to the standard notes that the limit of 13 ng (total congeners) is roughly equivalent to ng/m 3 (TEQ). 4. Reference conditions for the German 17.BImSch V (Seventeenth Ordinance on the Implementation of the Federal Emission Control Act) and EC Directive limits are: 11% O 2, 0 C, kpa, dry gas

7 3.0 INVESTIGATIONS The paper reviews the following full- and pilot-scale investigations undertaken in the last 10 years to investigate the effect of plastics on the performance of energy from waste facilities. In addition, the review draws upon the results of other major studies that have been carried out under the Environment Canada National Incinerator Testing and Evaluation Program (NITEP) and the Waste Analysis, Sampling, Testing and Evaluation (WASTE) Program. Also, findings reported in a recent study (Rigo et al., 1995) investigating the relationship between chlorine in waste streams and dioxin emissions from numerous waste combustor stacks were considered. Full Scale Investigations 1. The Würzburg trials, carried out in 1993/1994. The results of the study are documented in summary reports published by the Association of Plastics Manufacturers of Europe (Mark, 1994, Mark et al., 1994 and Mark, 1995) and full technical reports obtained from APME. 2. The South East London Combined Heat and Power (SELCHP) trials carried out in February/March The results of this investigation are documented in a report by the British Plastics Federation (1995). 3. Trials carried out at the VICON Resource Recovery Facility in Pittsfield, Massachusetts between May and July The results of these trials are documented in a report by the Mid West Research Institute (MWRI, 1987). 4. Tests carried out at the Ebara Corporation incineration plant located in Fujisawa City in Japan on the combustion of non-recyclable waste plastic separated from municipal waste that was destined for landfilling in an internally circulating fluidized bed boiler. The results of this study are documented in Tsukamoto and Kurihara (1995). Pilot Scale Investigations 1. A recent study carried out at the pilot bubbling fluidized bed reactor at the Institute of Environmental Chemistry at the University of Umea to investigate the effect of varying the chlorine and copper content of the feed. The results of this study are documented in Marklund, et al. (1994) and re-analyzed in Rigo et al. (1995). 7

8 The results of stack emission measurements carried out at the University of Florida-Tacachale-Clean Combustion Technology Laboratory between 1988 and VOCs were measured when firing non-hazardous institutional waste with various levels of PVC. The results of the study are documented in Wagner and Green (1993). 2. Testing conducted at the Neste Chemicals facility in Kulloo, Finland during co-combustion of mixed plastics with coal in a bubbling fluidized bed steam boiler. The results of these tests are documented in Frankenhauser, et al. (1993) and Ruuskanen, et al. (1994). 3.1 Wurzburg Study This study is the most comprehensive investigation carried out to-date to examine the effect of increased plastic content in the feed stream of a well-operated, modern energy-from-waste facility. The main objective of the study was to determine the role of plastics for energy recovery in the management of unrecyclable plastic wastes. Tests were conducted at the municipal waste combustor in Würzburg, Germany in October 1993 and again in January The facility has two process lines each consisting of a 12.5 tonne/hr Martin grate, mass burn waterwall incinerator equipped with a cyclone and fabric filter baghouse for gas cleaning. All tests were carried out on process line number 2. Under normal operations, these units burn residual MSW and sewage sludge. In the 1993 trials, a total of 10 tests were carried out under the following three test conditions: Test condition A: normal operations, whereby residual MSW and sewage sludge, containing between 8.5% to 12% plastics was combusted (3 tests in duplicate); Test condition B: 7.5% added mixed plastics waste raising the total plastic content of the feed stream to between 16%-19.5% (1 test in duplicate); Test condition C: 15% added mixed plastics waste raising the total plastic content of the feed stream to approximately 23.5 to 27% (1 test in duplicate). In January, 1994, an additional 4 tests were carried out. These included 2 baseline tests (with no added plastics), one test with 7.5% added plastics and one test with 15% added plastics. The main difference between the tests carried out in October, 1993 and January, 1994 lies in the 8

9 scrubbing agent used. In October 1993, lime injection alone was used. In January 1994, an additive containing activated carbon and lime was used. The trials were sequenced so that tests with added mixed plastics were carried out in between tests with no added plastics as follows: A1, A2, B1, B2, A3, A4, C1, C2, A5, A6. Each test condition was run for a minimum 48 hours with testing carried out in the last six hours. Mixed plastic wastes consisted of packaging material collected under the German DSD system supplemented by polyvinyl chloride (PVC) wastes to bring the percentage of this polymer in total plastic wastes to the European average of 10%. The study included: sampling and analysis of the mixed plastic feed stream; continuous monitoring of process conditions (O 2, grate temperature, furnace exit temperature, boiler inlet and exit temperature, CO, CO 2, HCl, SO 2, lime addition rate and steam production rate); sampling of exhaust gases at three sampling locations: at the stack; at the entrance to the cyclone and at the entrance to the baghouse for heavy metals, acid gases, and dioxins and furans; sampling and analysis of contaminant concentrations in all solid residues (grate ash, boiler ash, cyclone residue and filter residue) for carbon content, halogens, PCDD/Fs, and heavy metals. Feed Characteristics The heavy metal content (sum of Cd, Be, Zn, Mn, Cu, Pb, Cr, Ni, Sn, As, Se, Sb, Hg) of the mixed plastic waste measured during the trials varied from 0.7 g/kg (plastic waste used in condition C) to 1.4 g/kg (plastic waste used in condition B). For comparison, the heavy metal content of residual MSW feed measured at the Hartford Connecticut facility during trials conducted under the National Incinerator and Testing Evaluation Program (NITEP) ranged between 0.39 g/kg and 9.76 g/kg (Environment Canada, 1990). The study on the effect of waste stream characteristics on MSW incineration carried out under the Waste Analysis, Sampling, Testing and Evaluation (WASTE) Program in Burnaby, British Columbia (A.J. Chandler et al., 1993) found the total concentration of the heavy metals listed above in the mixed MSW feed stream to be 2.67 g/kg. The concentration of the same metals in the plastics fraction of the waste, was found to be less than 0.11 g/kg. 9

10 The concentrations of PCDD/F measured in the mixed waste plastic used during trial B was 49.3 ng/kg (TEQ). The concentration measured in the mixed plastic waste during trial C was 5.1 ng/kg (TEQ). In comparison, values reported in literature for PCDD/F concentrations in mixed MSW range from ng/kg (TEQ) (Eduljee, et al., 1995). Feed PCDD/F analysis carried out at the Hartford Connecticut facility under the NITEP program yielded results that ranged between 1.6 ng/kg and 85.5 ng/kg (TEQ). The results of the feed characterization study carried out at the Wurzburg facility showed that the chlorine content of the mixed plastic waste was 12 g/kg during test condition B and 17 g/kg during test condition C. Chlorine content in typical municipal waste combustor feed is reported to range between 6 and 10 g/kg of feed. The addition of the mixed plastic wastes therefore increases the total chlorine content of the feed stream. The wide variations in contaminant feed concentrations reported by the different studies are typical of MSW and are a result of the heterogeneity of municipal solid waste which makes it difficult to obtain representative samples. Process Performance Incinerators are designed for a specific maximum heat release. Due to the higher calorific value of plastics compared to that of the residual waste being combusted at Wurzburg the mass feed rate was reduced during the trials with added plastic in order to remain within the thermal capacity of the unit. Steam production during all the trials remained fairly constant, with lower variability observed during the trials with added plastics, indicating smoother operation during these tests. Results of the process monitoring conducted during the trials showed that all parameters for combustion control remained within normal operating ranges. Lower CO was observed in the furnace during tests with added plastics which indicates improved combustion during these tests. Heavy Metals and Particulate Matter in Flue Gases To evaluate the impact of the addition of plastics on heavy metal emissions, concentrations of heavy metals were measured in the raw gases entering the cyclone, in the raw gases entering the baghouse and in the cleaned gases exiting the stack. The concentration of heavy metals in the raw gases is primarily a function of the composition of the feed, whereas, the heavy metal concentration in gases exiting the stack is heavily influenced by the efficiency of the air pollution control device (APCD). A comparison of the raw gas concentrations (before the cyclone) for test conditions B and C (7.5% and 15% added plastics, respectively) with those for the baseline tests (condition A) 10

11 therefore provides an indication of whether the addition of plastics increases the feed concentration of heavy metals. A comparison of the concentrations in stack gases, shows whether the removal efficiency of the air pollution control device is affected by the addition of plastics (through a change in process conditions or flue gas characteristics). Table 2 shows the heavy metal concentrations in raw gases during test conditions A, B and C. The emissions during the tests with added plastics are within the same range of concentrations found during the baseline tests, except for Cu which appears to be enriched in test condition C, Hg which is enriched under test condition B and Sb which is enriched under test condition C. Table 2 Wurzburg Trials: Heavy Metal and Particulate Concentrations in the Flue Gases before the Air Pollution Control Device under Test Conditions A, B, C Heavy Metal A B C (Baseline) (7.5% Added Plastic) (15% Added Plastic) (mg/m 3 ) (mg/m 3 ) (mg/m 3 ) Cd Hg Pb Cu Sb Cr Sn Particulate Notes: all concentrations corrected to 11% O 2 Table 3, which summarizes the stack emissions of heavy metals under different test conditions, shows that final emissions of these metals in the tests with added plastics are not higher than those in the baseline tests. A comparison of the stack emission concentrations in Table 3 with emission limits for waste combustion in Table 1 shows that emissions of all metals except Hg are well below limits in all the jurisdictions considered. Emissions of Hg are significantly higher than the limits in Table 1 for all test conditions, with the highest emission rate observed during the baseline tests. Since Hg emissions are a function of feed concentration, this indicates that the added plastics did not contribute significantly to the Hg emissions. It should also be noted that testing for heavy metal emissions was only carried out in the October 1993 trials at which time carbon injection (which is normally employed at this facility) was not used. Activated carbon injection is considered maximum achievable control technology (MACT) for Hg, which because of its volatility is not 11

12 captured to any large degree by particulate control devices such as baghouses. The emission limits for Hg shown in Table 1 are based on the use of activated carbon injection. Tables 2 and 3 also show the particulate matter concentrations in the flue gases under different test conditions. As can be seen, the addition of plastics did not result in an increase in particulate matter concentrations in either the raw gases or the stack gases. Further, stack emissions of particulate were well below limits for municipal waste combustion in all the jurisdictions considered. 12

13 Table 3 Wurzburg Trials: Stack Emissions of Heavy Metals & Particulate Contaminants Baseline Tests Tests with 7.5% Added Plastics Tests with 15% Added Plastics A1 A2 A3 A4 A5 A6 B1 B2 C1 C2 Heavy Metals (ug/m 3 ) Cd < < 0.6 < 0.6 < 0.3 < 0.4 < 0.3 < 0.3 < 0.5 < 0.5 Tl <18.6 < 12.4 < 15.4 < 15 < 8.1 < 8.8 < 8 < 8 < 12.1 < 11.9 Hg Pb 19.7 < 6.2 < 7.7 < 7.5 < 4.1 < 4.4 < 3.8 < < 6 Cu 4.1 < 2.5 < 3.1 < 3 < 1.6 < 1.8 < 1.5 < 1.6 < 2.4 < 2.4 Sb <0.5 < 6.2 < 7.7 < 7.5 < 4.1 < 4.4 < 3.8 < 4 < 6.1 < 6 Cr <1.9 < 1.2 < 1.6 < 1.5 < 0.8 < < 0.8 < 1.2 < 1.2 As <18.6 < 12.4 < 15.4 < 15 < 8.1 < 8.8 < 7.6 < 8 < 12.1 < 11.9 Co <1.9 < 1.2 < 1.6 < 1.5 < 0.8 < 0.9 < 0.8 < 0.8 < 1.2 < 1.2 Mn <9.3 < 6.2 < 7.7 < 7.5 < 4.1 < 4.4 < 3.8 < 4.3 < 6.1 < 6 Ni 8.7 < 1.2 < 1.6 < 1.5 < 0.8 < 0.9 < 1.9 < < 1.2 V <1.9 < 1.7 < 1.6 < 1.5 < < 0.8 < 0.8 < 1.2 < 1.2 Sn <18 < 12.4 < 15.4 < 15 < 8.1 < 8.8 < 7.6 < 8 < 12.1 < 11.9 Cd+Hg < < 66.3 < 64.1 < 371 < < < < < Cd+Tl <19.3 < 13.6 < 16 < 15.6 < 8.4 < 9.2 < 8.3 < 8.3 < 12.6 < 12.4 Ni+As <27.3 < 13.6 < 17 < 16.5 < 8.9 < 9.7 < 9.5 < 8.8 < 15.4 < 13.1 Pb+Cr+Cu+Mn <35 < 16.1 < 20.1 < 19.5 < 10.6 < 11.5 < 22.1 < 10.7 < 21.9 < 15.6 Sb+As+Pb+Cr+Co+Cu +Mn+Ni+V+Sn <84.6 < 51.2 < 63.4 < 61.5 < 33.3 < 36.4 < 44.6 < 33.1 < 57.9 < 49 Particulate (mg/m 3 ) <1 < 1 < 1 < 1 < 1 < 1 < 1 < 1 < 1 < 1 Notes: (1) For non-detects, detection limits in HNO 3 solution are reported based on the assumption that all metals in the sample analyzed are captured in the particulate phase and the HNO 3 acid solution. (2) All concentrations corrected to 11% O 2, 0 C, kpa, dry gas. 13

14 HCl Emissions As would be expected, the addition of plastics increased the concentration of chlorine in the feed. This resulted in correspondingly higher concentrations of HCl in the raw gases under conditions B and C as shown in Table 4. Due to the efficiency of the scrubbing system used however, stack emissions of HCl under conditions B and C were not higher than baseline (condition A) and the stack concentrations in all the October 1993 tests with added plastics were below the limits shown in Table 1. Table 4 Wurzburg Trials: HCl Concentrations in the Flue Gases Under Test Conditions A, B, C Test Before Cyclone 1 Stack Gases 1 mg/m 3 mg/m 3 A A C C A A B B A A Notes: 1. Concentrations shown are half-hour averages. PCDD/F Emissions Table 5 shows the concentrations of PCDD/Fs in flue gases before and after the APCD. PCDD/F emissions in the raw gases range between 2.75 ng/m 3 and 4.27 ng/m 3 for the baseline tests, between 2.89 ng/m 3 and 4.58 ng/m 3 for test condition B, and between 2.76 ng/m 3 and 3.09 ng/m 3 for test condition C. PCDD/F concentrations in the stack gases similarly show that emissions with added plastics lie in the same range of emissions found during baseline tests. Table 5 14

15 Wurzburg Trials: PCDD/F Concentrations in the Flue Gases Under Test Conditions A, B,C Test Before Cyclone Stack Gases ng/m 3 ng/m 3 October, A A C C A A B B A A January, A B C with lime injection only 2 lime injection and activated carbon injection All concentrations expressed under reference conditions of 0 C, 1013 kpa, 11% O 2. A recent study carried out to investigate the relationship between chlorine in the waste stream and dioxin emissions from waste combustor stacks (Rigo et al., 1995), re-analyzed the results of the Wurzburg trials. The study used cluster analysis to examine the effect of plastics spiking on the congener distribution of PCDD/Fs and principal component analysis (PCA) and ANOVA techniques to determine whether a relationship existed between uncontrolled emissions of HCl (a surrogate of chlorine in the waste feed) and emissions of PCDD/F emissions. The study reconfirmed the original conclusions of the study that the addition of plastics did not result in any observable change in PCDD/F concentrations within the system nor did it change the congener distribution of the PCDD/Fs at the different locations sampled (at the entrance to the cyclone, at the entrance to the baghouse and at the stack). A critical factor influencing stack emissions of PCDD/Fs is the scrubbing system used. This is seen in Table 5 where stack emission levels measured in the 1994 tests (when a proprietary scrubbing agent containing lime and activated carbon was used) are seen to be between one and two orders of magnitude lower than those measured in 1993 (when lime alone was used). Stack 15

16 emissions with activated carbon injection are well below the emission limits in waste combustion standards in Canada, the United States, and Germany (see Table 1). Contaminant Concentration in Solid Residues During the Wurzburg trials, all the solid process residues (grate ash, boiler ash, cyclone ash and baghouse dust) were analyzed for halogens, sulphur, organic content, heavy metals and PCDD/Fs. Table 6 shows the concentrations of chlorine, total organic carbon and PCCD/F in grate ash, boiler ash, cyclone ash and fabric filter ash. Chlorine concentrations in grate ash and boiler ash are elevated for test condition C relative to baseline tests. Chlorine concentrations in fabric filter fly ash are in the same range for all test conditions. The total organic carbon (TOC) content of grate ash and boiler ash TOC was found to be lower for condition C relative to TOC during baseline tests. This indicates improved combustion resulting from the addition of the high calorific value plastics and is consistent with the lower furnace CO concentrations measured during the tests with added plastics. TOC concentrations in boiler ash and cyclone ash were three to five times higher in the January 1994 trials compared to the October 1993 trials. It appears probable that this is due to the use of activated carbon in the 1994 trials. PCDD/F concentrations in all solid residues were in the same range for tests with and without added plastics. The majority of the concentrations measured were below the Ontario criteria for remediation of residential/parkland land use for a potable groundwater condition of 1 ug/kg TEQ (OMOEE, 1996). 16

17 Table 6 Wurzburg Trials: Concentrations of Cl, TOC and PCDD/Fs in Solid Residues Grate Ash Boiler Ash Cyclone Ash Fabric Filter Ash Chlorine g/kg October, 1993 A B C January, 1994 A B C TOC g/kg October, 1993 A < B C <0.1 January, 1994 A B C PCDD/F ug/kg October, 1993 A B C January, 1994 A B C The concentrations of heavy metals measured in solid residues are summarized in the Table 7. Concentrations in residues for test conditions B and C are within the range of concentrations obtained in the baseline tests, indicating that the addition of plastics had no discernible effect on the heavy metal concentrations of solid residues. 17

18 Table 7 Wurzburg Trials: Concentrations of Heavy Metals in Solid Residues Grate Ash Boiler Ash Cyclone Ash Fabric Filter Ash g/kg g/kg g/kg g/kg Cd A B C Hg A B C Sb A B C Pb A B C Cr A B C Cu A B C Sn A B C SELCHP Study This investigation was carried out at the South East London Combined Heat and Power (SELCHP) energy-from-waste facility in London, England in February/March The objective of the program was to compare stack emissions when firing mixed MSW spiked with 5% added plastics to those during normal operations The facility has two independent process trains each consisting of 29 tonne/hour Martin grate incinerators equipped with acid gas scrubbers (hydrated lime injection), activated carbon injection and fabric filter baghouse. All testing was carried out on train no

19 Air emission testing was conducted in duplicate under the following conditions: Test 1: Test 2: Test 3: Test 4: MSW (control test) 95% MSW and 4.6% mixed plastic 95% MSW and 4.6% mixed plastic MSW (control test) Addition of the mixed plastics waste is estimated to have brought the total plastic content of the feed stream up to 11.8% (based on a United Kingdom average of 7.2% plastics in mixed MSW). During the trials, atmospheric emissions, heat recovery (steam generated per tonne of MSW) and general plant performance were monitored. Testing was carried out at the stack only for particulate, hydrogen chloride, hydrogen fluoride, sulphur dioxide, nitrogen oxides, VOCs, heavy metals, carbon monoxide and PCDD/Fs. Process Performance As for the Wurzburg trials, it was necessary to reduce the throughput of the unit when burning additional plastics to compensate for the higher calorific value of the plastic waste. All other process parameters remained within normal operating ranges during the tests with added plastics. As expected, steam production per tonne of waste increased from tonnes in the baseline tests (tests 1 and 4) to tonnes during the tests with added plastics (tests 2 and 3). Lower variability in CO emissions was found during tests with added plastics indicating steadier combustion conditions. Air Emissions Table 8 summarizes the results of the stack testing program carried out at SELCHP with and without added plastics. A comparison with emission limits in Table 1 shows that emissions of particulate matter, acid gases, heavy metals and PCDD/Fs are consistently low for all four tests and well within emission limits in Canada, the United States and Europe. 19

20 Table 8 SELCHP Trials: Emission Testing Results Test 1 Test 2 Test 3 Test 4 mg/m 3 mg/m 3 mg/m 3 mg/m 3 (Baseline) (Baseline) Particulate HCl SO Cd <0.001 Hg As <0.001 <0.001 <0.001 <0.001 Cr Cu Mn < <0.001 Ni <0.001 Pb Sn <0.001 <0.001 <0.001 <0.001 PCDD/Fs (TEQ) ng/m 3 Notes: All concentrations at 273 C, kpa, 11% O Pittsfield Study This study attempted to identify the relationship of incinerator operating conditions, waste characteristics, and combustion variables to levels of PCDDs and PCDFs and other selected semi-volatile organic compounds. Tests were conducted at the VICON Resource Recovery Facility in Pittsfield, Massachusetts between May and July The facility consists of three modular, excess air 2-stage combustors each with a rated capacity of 120 tons per day. Tests were carried out on two of the three identical units, as this is the normal operating mode of the facility. The air pollution control equipment employed during the tests was an electrified gravel bed filter. During this investigation emissions were measured under a number of different test conditions of which the following are of interest to this review: Normal MSW (runs 9 and 14). A PVC-free waste consisting mainly of cardboard and wood wastes (runs 12 and 17). 20

21 PVC spiked waste (runs 18 and 25). The three selected paired tests were all undertaken at the same primary chamber exit set point temperature of 1800 C. For the normal MSW and PVC spiked conditions, testing was carried out at the boiler outlet and the stack. For the PVC free waste, testing was carried out at the furnace outlet and at the boiler outlet. All waste feeds were analyzed for chlorine. The chlorine content in samples taken during the six runs ranged from 0.03% for the PVC free waste to 1.24% for the PVC spiked MSW. As part of the study, a statistical analysis of the test results was carried out to determine whether there was a relationship between the PVC content of the feed and PCDD/F concentrations. The study concluded that differences in the PCDD/F concentrations under the three different feed regimes were lower than the standard error of the difference, making it not significantly different from zero. The study therefore concluded that the PVC content had no effect on concentrations of PCDD/Fs in flue gases. Rigo et al. re-analyzed the results of the Pittsfield study using cluster analysis, principal component analysis and ANOVA techniques. Their findings reconfirmed the original study findings that the test data collected during this study do not indicate a statistically significant relationship between PVC content and flue gas PCDD/F concentrations. 3.4 Ebara Study This study was carried out to determine whether stable energy recovery and adequate control of emissions could be achieved while burning non-recyclable waste plastic separated from municipal waste that was destined for landfilling. The tests were carried out at a unit located at the Ebara facility in Fujisawa City in Japan, that is normally used to combust industrial waste. The unit is an internally circulating fluidized bed boiler with a baghouse, acid gas scrubber and coke filter Tests were carried out with two waste streams: waste plastics segregated from residual MSW; and, plastic pellets of composition similar to the residual plastics. This investigation did not carry out a baseline test with mixed MSW or plastic free material for comparison. Stack emission testing for dioxins, furans, CO, NOx, SO 2, HCl, total hydrocarbons, particulate matter, and heavy metals was carried out. Stable combustion and steam production was demonstrated, with average CO levels of around 22 ppm. Stack emissions of PCDD/Fs of 0.1 ng/m 3 (TEQ) were obtained. Particulate matter emissions were below the detection limit of 1 mg/m 3. Heavy metals concentration in the particulate phase at the baghouse outlet were below detection limit. 21

22 3.5 Umea Study A number of combustion studies have been undertaken at the pilot bubbling fluidized bed reactor at the Institute of Environmental Chemistry at the University of Umea. The unit has a capacity of 2.2 lbs/hr (5kW)and is capable of simulating combustion conditions in full-scale municipal waste combustion units. Bed materials are removed from the gas stream in a cyclone. Flue gases are then cooled and drawn through an activated carbon filter. A recent study carried out at this facility (Marklund, et al., 1994) investigated the effect of varying the chlorine and copper content of the feed on emissions of PCDD/Fs. The chlorine content was varied between 0.1% and 2% through the addition of inorganic and organic chlorine (in the form of CaCl 2 and PVC, respectively) to the fuel pellets. Copper additions varied from 0.000% to 0.012%. Feed composition was closely controlled in these experiments. Sampling was carried out in triplicate (before the carbon filter) under nine test conditions. The results of these experiments are analyzed in Rigo et al., The original study concluded that there was no correlation between the chlorine in the feed and the PCDD/F concentrations up to a chlorine content of 1%. Increasing the chlorine content of the feed above 1% was found to increase the emissions of PCDDs/PCDFs. Rigo et al., 1995 cautions that this should be considered a tentative finding since the high variability observed in carbon monoxide emissions indicate that operating conditions varied between runs during individual test conditions. Variability in CO emissions was higher during the two high chlorine tests. It is therefore possible that the increase in PCDD/F emissions observed during these tests were the result of poorer combustion rather than higher chlorine. A number of other studies have found a statistically significant relationship between combustion conditions and emissions of PCDD/PCDFs. The analysis carried out by Rigo et al. also showed a non-linearity in the data which would indicate that a threshold level of PCDD/PCDFs may exist regardless of the amount of chlorine in the feed. 3.6 University of Florida Tests Tests were carried out at the pilot scale combustion unit at the Clean Combustion Technology Laboratory of the University of Florida to determine whether a correlation between the plastic content of incinerator feed and emissions of chlorinated organic compounds exists. Non hazardous institutional waste was used together with various levels of added PVC ranging from 0.7% to 4.7%. The unit is a pilot scale batch 500 lb/hr two stage incinerator. 22

23 In tests carried out at this facility, HCl and volatile organic compounds (VOCs) were measured at the stack. A total of 32 runs appear to have been undertaken. PVC was spiked in the form of either PVC resin containing 56.8% chlorine or PVC pipe containing 22.7% chlorine (the researchers assumed that the pipe contained 60% filler and 40% PVC) in 12 runs. The researchers report that valid VOC results were obtained for only 18 of the 32 runs. The reason for this high failure rate is not clear. Six of the runs for which valid VOC results were obtained had PVC resin added to the incinerator while non-hazardous waste was burning. The study examined the relationship between HCl emissions (as a surrogate for chlorine content of the feed) and emissions of dichlorobenzene using multivariate regression analysis. The study concluded that: stack gas concentrations of HCl increase with increasing PVC in the feed; and, that dichlorobenzene emissions are strongly related to the product of chlorobenzene and HCl. Based on these findings, the authors conclude that a reduction in the chlorinated plastics content of the waste stream would result in a reduction in other chlorinated aromatic hydrocarbon emissions such as PCDD/Fs (which were not measured). The statistical analysis carried out by the original researchers consisted of an exploratory series of over 460 single and multivariate regression analysis. The authors report that a number of correlations were found at the 95% confidence level. Rigo et al. (1995) points out that at the 95% confidence level it can be expected that at least 5% of the models would fit the data due to chance alone. Rigo et al. (1995) reanalyzed the results of this study. They found no significant linear relationship between HCl and VOC s. The product of HCl and chlorobenzene was found to produce a statistically significant prediction of dichlorobenzene concentration. However, Rigo et al. suggests that the design of the test program raises questions on the validity of the data. To determine input/output relationships, testing must be carried out under steady state conditions. In the trials at the University of Florida, tests with and without PVC were, in several instances, carried out on the same day. Under these circumstances it is not clear that process equilibrium was achieved prior to testing. A further premise of the University of Florida study is that VOCs are a suitable surrogate for PCDD/Fs. The relationship between VOCs and PCDD/Fs is not well established. 3.7 Neste Study 23

24 Testing was conducted at the 7 MW bubbling fluidized bed steam boiler at the Neste facility in Kulloo, Finland during co-combustion of mixed plastics with high sulphur coal. The boiler is rated at 7MW. Flue gases exiting the boiler are cleaned in an electrostatic precipitator. The objective of these tests was to determine the effects of sulphur-rich coal on the formation of chlorinated hydrocarbons like PCDD/PCDF in the co-combustion of mixed waste plastics with coal. Ten tests were carried out in the following sequence, whereby the proportion of plastics in the feed was gradually increased from 0% to 100% and then decreased again down to 0%: A 1-100% coal; B 1-10% plastics; C 1-36% plastics; D 1-51% plastics; E 1-100% plastics; E 2-100% plastics; D 2-51% plastics; B 2-10% plastics; F 2-10% plastics; and, A 2-100% coal. Two tests were run per day. Testing began 4 hours after the feed was changed to allow the process to stabilize. The waste plastic used during all tests except F 2 contained 60% polyethylene, 20% polypropylene, 15% polystyrene, and 5% polyvinyl chloride. The chlorine content of the mixed plastics was approximately 4%. In test F 2 a mixture containing 75% polystyrene and 25% PVC was used. Emission testing for hydrogen chloride, trace organics and heavy metals was conducted at the stack. A raw gas sample was only taken during test D 2 (~50% plastics). Oxygen, CO, SO 2, NOx and N 2 0 were monitored continuously. Chloride and heavy metal concentrations in fly ash from the electrostatic precipitator were also measured for each of the 10 feed regimes. The higher volatility of plastics relative to coal, lead to more combustion in the freeboard during tests with added plastics. High CO and soot formation were observed during test E 1 (100% plastics). In the subsequent test E 2, secondary air was increased. This reduced both the CO level as well as the residual carbon content of the fly ash. Emission test results for CO, HCl, heavy metals, and PCDD/Fs under the different fuel regimes are summarized in Table 9. Table 10 shows the chloride, trace organics and heavy metal concentrations measured in fly ash. The original study found no correlation between PCDD/F formation and the plastics and chlorine content in the feed. The study found correlations between HCl emissions and PCDD/F concentrations in the fly ash. The study also found a correlation between CO emissions and PCDD/F formation. The apparent increase in PCDD/F emissions in the 100% plastics tests (E 1 and E 2 ) is attributed to poor combustion (as evidenced by high CO) during these tests rather than the plastic content of the feed. Ruuskanen (1994) analyzed the results of the Neste study using principal component analysis. His analysis reconfirmed the original study findings. Rigo, et al. (1995) reanalyzed the results using cluster analysis, principal component analysis and analysis of variance techniques. They concluded that there is no change in either the type or quantity of PCDD/Fs in response to plastics spiking and that there was a limited response under the 100% plastics condition which is likely attributable to poor combustion. 24

25 A comparison of the emission results in Table 9 with the emission limits in Table 1 shows that all emissions are well within these limits. The concentrations of heavy metals and trace organics in fly ash are compared with the Ontario soil remediation criteria for residential/parkland land use for a potable groundwater condition in the recently adopted Guideline for Use at Contaminated Sites in Ontario (June,1996) in Table 10. The comparison shows that for the majority of contaminants, concentrations are within this criteria. Table 9 Neste Study: Stack Emission Results Under Different Test Conditions A 1 B 1 C 1 D 1 E 1 E 2 D 2 B 2 F 2 A 2 % plastic 0% 10% 36% 51% 100% 100% 51% 10% 10% 0% CO, mg/nm HCl, mg/nm Metals (ug/nm 3 ) Hg Pb Cu nd nd nd Ni nd nd nd 0.2 nd nd nd nd nd nd As nd nd nd nd nd nd nd nd nd nd Sb nd nd nd nd nd nd nd nd nd nd Cd nd nd nd nd nd nd nd nd nd nd Co nd nd nd nd nd nd nd nd nd nd Cr nd nd nd nd nd nd nd nd nd nd Mn nd nd nd nd nd nd nd nd nd nd Sn nd nd nd nd nd nd nd nd nd nd PCDD/F ng/nm 3 (Nordic TEQ) nd = not detected 25

26 Table 10 Neste Study: Concentrations of Contaminants in Fly Ash (ppm) A 1 B 1 C 1 D 1 E 1 E 2 D 2 B 2 A 2 Ontario Criteria for Soil Remediation Cl % Cd Cu (300) Hg Pb Sn Zn PCDD/Fs (TEQ) ng/g ng/g PCBs ng/g Total PAHs ug/g

27 4.0 SUMMARY OF RESULTS 4.1 Process Performance The investigations reviewed (Mark, 1994, BPF, 1995) indicate that an increase in the plastics content of the feed to an EFW facility results in improved combustion stability, manifest in fewer CO peaks and less variability in the rate of steam production; lower CO emissions as compared to baseline which is indicative of the better waste burn out achieved; and lower residual unburned carbon content in the grate ash, boiler ash and fly ash. These findings confirm the value of plastics in energy recovery programs. However, a significant reduction in the throughput of the units was observed at both the Wurzburg and SELCHP facilities due to the high heating value of plastics relative to municipal solid waste. This emphasizes the need to predict the changes in plastic levels that can be expected in MSW in the coming years for integration into the design of new energy recovery units. 4.2 Heavy Metals and Particulate Matter Emissions Of the seven studies reviewed in this paper, four (the Wurzburg study, the SELCHP study, the Ebara study and the Neste study) investigated the effect of plastics addition on emissions of heavy metals. All the studies found low emissions of heavy metals, with and without plastics addition. The results of the Wurzburg, SELCHP and Ebara studies also indicate that emissions of particulate matter from well-operated waste combustion facilities are very low and that they are not effected by the addition of plastics. 4.3 Dioxin and Furan (PCDDs and PCDFs) Emissions PCDD and PCDF emissions from waste combustion and the effect on these of plastics have been the subject of extensive study. The findings of each of the studies reviewed with respect to PCDD/PCDF emissions are as follows. The Wurzburg study in Germany found no correlation between plastics content of the feed and emissions of PCDD/Fs. These findings were reconfirmed by tests carried out at the SELCHP facility in London, UK; the VICON incinerator in Pittsfield, Massachusetts; and the Neste facility in Kuopio, Finland. These results are supported by the findings of Rigo et al. (1995) which reviewed the results of tests from 169 facilities including municipal waste combustors, biomedical waste combustors, hazardous waste incinerators, and cement kilns to determine whether a relationship exists between PCDD/F concentrations and uncontrolled gas phase concentrations of 27

28 HCl (as a surrogate for chlorine concentration in the input feed). Chlorine feed concentrations in the studies reviewed in that report ranged from less than 0.1 percent up to 80%. Rigo et al. (1995) concluded that no relationship exists between PCDD/F concentrations and the chlorine content of the feed to combustion units. The study carried out at the University of Florida concludes that a correlation exists between chlorinated hydrocarbons and HCl emissions (a surrogate for plastics content). The study did not measure PCDD/Fs and the paper reviewed does not provide sufficient data on the testing program design to allow a critical review of the validity of the data. The study carried out at the University of Umea did not find any correlation between PCDD/F concentrations and chlorine content up to a chlorine content of 1%. Above 1% a correlation was found. However, high variability in CO emissions was observed during the high chlorine tests suggesting that poor combustion rather than high chlorine may have been the cause of the elevated PCDD/F concentrations. Thus, both studies indicating the possibility of a correlation between PCDD/F emissions and plastics content of the waste feed have the potential for confounded results. 4.4 Hydrogen Chloride Emissions The results of the review show that the addition of plastics to the feed of combustion units increases the concentrations of HCl in the raw gas. However, there is no discernible increase in HCl emissions from modern energy from waste facilities employing lime injection, when plastics are added. 4.5 Contaminant Concentrations in Solid Residues Two studies reviewed in this paper measured concentrations of contaminants in solid residues during tests with added plastics: the Wurzburg study and the Neste study. The Neste study found a correlation between concentrations of PCDD/F in the ash from the ESP and HCl emissions. Extensive testing at the Wurzburg facility did not confirm the findings of the Neste study. The Wurzburg study results indicate the addition of plastics had no effect upon the contaminant concentrations in grate ash, boiler ash, cyclone ash or fabric filter ash. The concentrations of trace organics found in the ash in the majority of tests in both studies were below the Ontario soil remediation criteria for residential/parkland with potable groundwater use. 5.0 CONCLUSIONS The results of the review show that: 28