United States Environmental Protection Agency Office of Water (4303) EPA821-R-97-006 November 1997 Statistical Support Document for Proposed Pretreatment Standards for Existing and New Sources for the Industrial Laundries Point Source Category ------- ------- Statistical Support Document for Proposed Pretreatment Standards for Existing and New Sources for the Industrial Laundries Point Source Category (EPA-821-R-97-006) Prepared for: U.S. Environmental Protection Agency Office of Water, Engineering and Analysis Division (4303) 401 M Street SW Washington, DC 20460 Prepared by: Science Applications International Corporation Environmental and Health Sciences Group Health and Environment Studies and Systems Division 11251 Roger Bacon Drive Reston, VA 20190 j November 1997 ------- ------- ACKNOWLEDGMENTS AND DISCLAIMER This report has been reviewed and approved for publication by the Engineering and Analysis Division, Office of Science and Technology. This report was prepared with the support of Science Applications International Corporation (contract 68-C4-0046) under the direction and review of the EPA's Office of Science and Technology. Neither the United States Government nor any of its employees, contractors, subcontractors, or then* employees makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third party's use of, or the results of such use of, any information, apparatus, product, or process discussed in this report, or represents that its use by such third party would not infringe on privately owned rights. ------- ABSTRACT This document describes the sample design and development of survey weights for the Industrial Laundries questionnaires. It also provides the statistical analyses used in developing the proposed pretreatment standards. A list of the data used in calculating long-term averages, variability factors, and limitations is included. ------- Table of Contents 1. 2. Overview and Organization 3. 4. 5. Survey Design ; 2.1. Trade Association Population 2.1.1. Trade Association Screerier Questionnaire 2.1.2. Trade Association Detailed Questionnaire 2.2. Dun and Bradstreet Population 2.2.1. Dun and Bradstreet Detailed Questionnaire 2.2.2. Dun and Bradstreet Screener Questionnaire 2.3. Hotels, Hospitals, and Prisons Screener Questionnaire 2.4. Industrial Laundries Population - Final Detailed Questionnaire Design • Estimation Methodology for National Estimates 3.1. Detailed Questionnaire 3.1.1. Estimation from Complete Data 3.1.2. Estimation with Item-Level Non-Response 3.1.3. Estimation for Domains with Complete Response 3.1.4. Estimation for Domains with Item-Level Non-Response 3.2. National Estimates Analytical Data Collection Efforts and Definition of Options 4.1. Detailed Monitoring Questionnaire 4.2. EPA Wastewater Sampling Program 4.3. Definition of Proposed Options Description of Data Conventions 5.1. Data Review . 5.2. Data Types 5.3. Data Aggregation 5.3.1. Data Aggregation Across Multiple Grabs 5.3.2. Data Aggregation for Field Duplicates Statistical Methodology ' 6.1. Basic Overview of Adapted Delta-Lognormal Distribution 6.2. Modifications to the Adapted Delta-Lognormal Model Estimation under the Modified Delta-Lognormal Model 7.1. Facility-specific Estimates : 7.1.1. Estimation of Facility-specific Long-Term Averages 7.1.2. Estimation of Facility-specific Variability Factors 7.1.2.1 Estimation of Facility-specific 1-day Variability Factors 7.1.2.2 Estimation of Facility-specific 4-day Variability Factors Page 1-1 2-1 2-1 2-1 2-3 2-9 2-9 2-10 2-12 2-14 3-1 3-1 3-1 3-2 3-4 3-5 3-7 4-1 4-1 4-1 4-1 5-1 5-1 5-1 5-2 5-2 5-2 6-1 6-1 6-4 7-1 7-2 7-2 7-2 7-3 7-4 ------- 7.2. Pollutant-specific Estimates 7.2.1. Estimation of Pollutant-specific Long-Term Averages 7.2.2. Estimation of Pollutant-specific Variability Factors 7.2.2.1 Estimation of Pollutant-specific 1-day Variability Factors 7.2.2.2 Estimation of Pollutant-specific 4-day Variability Factors 8. Derivation of the Proposed Limitations 9. Raw Wastewater Concentration Comparisons 9.1 Comparison of Industrial Laundry Influent to Linen Supply Influent 9.2 Comparison of Linen Supply Influent to Denim Pre-wash Influent Appendices A. Listing of Daily Data A. 1 Listing of Daily Data for Dissolved Air Flotation A.2 Listing of Daily Data for Chemical Precipitation B. C. D. E. Listing of Summary Statistics for Regulated Pollutants B. 1 Listing of Summary Statistics for Regulated Pollutants for Dissolved Air Flotation B.2 Listing of Summary Statistics for Regulated Pollutants for Chemical Precipitation Listing of Facility-Level Long-Term Averages and Variability Factors C.I Listing of Facility-Level Long-Term Averages and Variability Factors for Dissolved Air Flotation C.2 Listing of Facility-Level Long-Term Averages and Variability Factors for Chemical Precipitation 7-7 7-7 7-7 7-7 7-7 8-1 9-1 9-1 9-4 A-l A-l A-10 B-l B-l B-5 C-l C-l C-4 Listing of Pollutant-Level Long-Term Averages, Variability Factors and Limitations D-l D.I Listing of Pollutant-Level Long-Term Averages, Variability Factors, and D-l Limitations for Dissolved Air Flotation D.2 Listing of Pollutant-Level Long-Term Averages, Variability Factors and D-5 Limitations for Chemical Precipitation Episode, Sample Point and Data Source Used in Raw Wastewater Concentration E-1 Comparisons E.I Industrial Laundry and Linen Comparisons E-l E.2 Linen and Denim Comparisons E-2 ------- CHAPTER 1 OVERVIEW OF ORGANIZATION AND CONTENTS OF DOCUMENT This document describes the statistical analyses performed for the Effluent Limitations Guidelines and Pretreatment Standards for the Industrial Laundries Point Source Category. These statistical analyses were used in estimating the number of industrial laundry facilities, estimating the number of industrial laundry facilities with particular characteristics of interest, developing the proposed effluent guideline standards and comparing influent pollutant concentrations of linen, industrial laundry, and denim prewash facilities. : This document is organized into nine chapters and five appendices, The following list summarizes the content of each chapter and appendix: Chapter 1: Overview of Organization and Contents of Document - Describes the organization of the document and summarizes the contents of each chapter and appendix. Chapter 2: Survey Design - Describes the development of the sample frame and selection of facilities to receive the detailed and screener questionnaires. Chapter 3: Estimation Methodology for National Estimates : - Describes the methodology used in calculating national estimates from the detailed questionnaire and provides some national estimates. ! Chapter 4: Analytical Data Collection Efforts and Definition of Options - Provides an overview of the analytical data collection efforts and defines the technology options. Chapter 5: Description of Data Conventions - Describes data conventions and how the data were treated, including aggregation and review. Chapter 6: Statistical Methodology - Describes the modified delta-lognormal distribution that was used to derive the proposed limitations. Chapter 7: Estimation under the Modified Delta-Lognormal Distribution - Describes the estimation of long-term averages and variability factors at the facility and pollutant levels. i ChapterS: Derivation of the Proposed Limitations • - Describes the derivation of the proposed limitations. Chapter 9: Raw Wastewater Concentration Comparisons ; - Describes the comparison of raw wastewater for facilities washing mostly linen items versus facilities washing mostly industrial items and a comparison between facilities washing mostly linen items versus facilities washing mostly denim pre-wash items. 1-1 ------- Appendix A: Listing of Daily Data - Provides a listing of the concentration data from each facility used to characterize the treatment in the options. Appendix B: Listing of Summary Statistics for Regulated Pollutants - Provides summary statistics for the data from each facility used to characterize the treatment in each option. Appendix C: Listing of Facility-Level Long-Term Averages and Variability Factors - Provides a summary of the facility-specific long-term averages and variability factors for the proposed option. Appendix D: Listing of Pollutant-Level Long-Term Averages, Variability Factors and Limitations - Provides the pollutant-level long-term averages, variability factors and the proposed limitations. Appendix E: Episode, Sample Point and Data Source Used in Industrial Laundry and Linen Comparisons - Provides the episode, sample point and data source used in the industrial laundry raw wastewater comparisons. 1-2 ------- CHAPTER 2 SURVEY DESIGN The industrial laundry industry screener and detailed questionnaires were sent to a random selection of facilities that were identified from two sources. These two sources of population information were trade association listings and information obtained from Dun & Bradstreet. The trade association listings were compiled from Uniform and Textile Service Association (UTSA, formerly IIL) customer and prospective customer lists, the Textile Rental Service Association (TRSA) mailing list, and the Occupational Safety and Health Administration's (OSHA) list of violations for industrial laundries. Industrial laundry facilities were identified in the Dun & Bradstreet listing by their reported Standard Industry Classification (SIC) codes. Facilities with primary SIC codes of 7218 (industrial laundering) or 7213 (linen supply servicing), and facilities with a secondary SIC code of 7218 were considered to be industrial laundries. i The original screener questionnaires were sent to all facilities in the trade association listing. Detailed questionnaires were sent to a random selection of facilities that reported generating wastewater from the trade association screener responses. After the frame was developed from the trade association listings, it was realized that the entire population of industrial laundries was not covered. Therefore, the Dun & Bradstreet information was used to supplement the trade association listings. Additional screener and detailed questionnaires were sent to a selection of facilities from the Dun & Bradstreet listing in order to capture industrial laundry facilities in the nation that were not originally identified in the trade association listing. : The two population listings initially contained duplicate facilities, due to an overlap between the trade association and Dun & Bradstreet listings. Extensive efforts were used to select only facilities from the Dun & Bradstreet listings that did not appear in the trade association lists. However, due to inconsistent recording of addresses and ownership status, some facilities were included in both sampling frames. After removing duplicate facilities from these two listings, the two populations are mutually exclusive. Different sample selection methods were used to randomly sample facilities within each of these two populations. Because the two populations are mutually exclusive, national estimates are generated within each population separately, and then combined to characterize the entire population of industrial launderers in the nation. The development of the sampling frames, the sample selection process, and the resulting survey weights is summarized below. The survey weights were developed independently for the screener and detailed questionnaires within each population (trade association and Dun & Bradstreet). 2.1 Trade Association Population \ 2.1.1 Trade Association Screener Questionnaire : The final mailing list from the trade association listings contained 1,751 industrial laundry facilities. Screener questionnaires were mailed to all 1,751 facilities. Therefore, a census was taken from the trade association listing with a frame size (N) and a sample size (n) equal to 1,751. 2-1 ------- Of the 1,751 screeners that were matted, 1,543 were returned. In addition, three facilities that were not on the original mailing list received copies of the screener from their parent company and completed and returned them to the EPA. Therefore, the frame (N) and sample size (n) are increased to 1,754, and the number of respondents is 1,546. However, 46 of the 1,546 respondents were duplicate facilities (i.e., 46 facilities were sent two screeners and returned both). Two copies of the questionnaire were mistakenly sent to each of these facilities due to inconsistent documentation of the addresses, facility names, and/or ownership status. Despite efforts to remove duplicate facilities before the mailing lists were established, differences in the recorded mailing addresses and physical locations and owner and operator names caused some facilities to be listed twice. After removing the duplications, the frame and sample size is 1,708 and the number of unique respondents is 1,500. There were 208 nonrespondents to the screener questionnaire. Among the 1,500 screener questionnaires that were returned, 1,127 of the facilities were identified as "in-scope". In-scope facilities are defined as facilities that generated laundry wastewater in 19931. An additional two facilities were later identified as in-scope facilities, which brought the total number of in-scope respondent facilities to 1,129. During the development of the detailed questionnaire frame from the list of in-scope screener respondents (as documented in Section 2.1.2), there were 15 facilities for which information regarding 1992 revenue, wastewater treatment type, or items laundered was not available. Subsequent telephone calls to these facilities were used to obtain this information. Prom the responses to these telephone calls, one facility that was originally identified as in-scope was found to be out-of-scope. Therefore, the number of in-scope respondents to the screener questionnaire was reduced by one to 1,128. The number of in-scope facilities among the 208 nonrespondents to the screener questionnaire is not known because scope was determined from the response to the screener question regarding wastewater generation. The EPA conducted an assessment to characterize the 208 nonrespondents to determine the likeliness that these facilities were in-scope. Characteristics assessed include mail delivery, business status, and laundry wastewater generation. Efforts to obtain more information about these facilities resulted in the identification of 86 of the 208 nonrespondents as out-of-scope. This was because 65 screener questionnaires were returned by the post office and new addresses were not available, implying that the facilities were out of business, and 21 facilities were excused by the EPA from completing the screener questionnaire because they did not generate laundry wastewater, were out of business, or were duplicates of other facilities. The remaining 122 nonrespondents to the screener questionnaire are possibly in-scope, but the status has not been verified as of this writing. Of these 122 facilities, 3 were excused by the EPA from completing the screener questionnaire but are likely to be in-scope and 119 were not returned by the post office, implying that the facility received the questionnaire, but no completed screener was returned from the facility. Among the 208 nonrespondents to the screener questionnaire from the trade association listing, five facilities also were sent a screener questionnaire from the Dun & Bradstreet listing (as documented in Section 2.2.2). This duplication was discovered after the screener questionnaire was mailed from the "In-scope" at this time includes denim prewash facilities, linen facilities, and all facilities generating laundry wastewater, regardless of production amount. 2-2 ------- Dun & Bradstreet listing. These five facilities were retained in the Dun & Bradstreet frame to maintain the probability structure of the Dun & Bradstreet sample. (The Dun & Bradstreet sample was a probability sample, whereas the trade association sample was a census.) Therefore, these five facilities were removed from the trade association frame, and accounted for in the Dun & Bradstreet population only. Of these five facilities, one was from the set of 86 nonrespohdents that are believed to be out-of- scope because the screener was returned by the post office and a new address was unavailable. The other four facilities were from the set of 122 nonrespondents that are possibly in-scope because the screeners were not returned by the post office. By removing these four facilities from the trade association population, there were 118 nonrespondents to the screener questionnaire that were possibly in-scope. Because it was not known if the nonrespondents were in-scope, and because auxiliary information was not available for these facilities, the EPA estimated the number of in-scope nonrespondents in the following way. The EPA assumed that the proportion of the 118 nonrespondents that were estimated to be in-scope is equivalent to the proportion of respondents that were identified as in-scope. There were 1,128 in-scope facilities among the 1,500 respondents, so it was estimated that 89 of the 118 nonrespondents also were in-scope. : Therefore, the estimated total number of in-scope facilities from the trade association population was 1,217 (i.e. the 1,128 in-scope respondents plus the estimated 89 in-scope nonrespondents). After the trade association list was established, the five largest industrial launderers in the nation were examined to identify facilities that may not have been included in the trade association list. There were 48 facilities identified as belonging to these five industrial launderers (Aratex, Cintas, Omni, Unifirst, and Unitog). Also, mailing addresses were identified for four additional facilities that were not originally included in the trade association list due to lack of address information. Abbreviated versions of the screeners were sent to these 52 facilities to obtain information regarding their operating practices and status. From this information, 29 facilities were identified as being in-scope and did not duplicate facilities originally in the trade association list. These 29 facilities were added to the trade association population. Therefore, the total number of in-scope industrial laundry facilities in the trade association population was 1,246, of which, information from the screener questionnaire ;was available from 1,128 in-scope respondents. 2.1.2 Trade Association Detailed Questionnaire The original trade association frame, from which a random sample of facilities was selected to receive the detailed questionnaire, was based upon the list of in-scope facilities that responded to the screener questionnaire. The original list of in-scope screener respondents contained 1,127 facilities (see Section 2.1.1). Only the in-scope respondent facilities to the screener were used as the sampling frame because information collected from the screener questionnaire responses was used to construct the detailed questionnaire sampling frame strata. The stratification scheme wa£ based on items laundered, 1992 revenues, and wastewater treatment processes, for a total of 48 strata (see Table 2-1). 2-3 ------- Table 2-1 Trade Association Detailed Questionnaire Strata A: B: C: D: 1: 2: 3: 4: I: H: HI: -- - "" Items Laundered ,-j > , i 1 I \V 5% or more printer towels (and possibly anything else) 5% or more shop towels (and possibly less than 5% anything else) 10% or more industrial garments (and possibly less less than 5% shop towels or anything else) Anything not covered by A, B, or C . „ '", "" 1992 Revenues Less than $1 million Greater than or equal to$1 million and less than $3 printer towels or then 5% printer towels or " ' * ilt i Oft***" <• f "" *' *• ', i"1 ; . ' J%J"',. -t^HH .5 million Greater than or equal to$3.5 million and less than $7 million Greater than or equal to$7 million i > » * _, i f , Wastewater .Treatment - • • -, , V., i 4. IH ! ^ a >. f 1 k S-W ,^i * I "^ - f' J rf * ' Biotreatment, air stripper, centrifuge, membrane filtration, pressure • filtration, and/or media filtration, and/or carbon adsorption (and possibly anything else) Dissolved air flotation, oil/water separation, and/or anything else) Anything not covered by I or n clarifier (and possibly At the time that the sampling frame was developed, stratification information was not available for 15 of the in-scope respondent facilities. Therefore, the sampling frame contained 1,112 facilities, divided into the 48 strata. These facilities and strata are summarized .in Column (a) of Table 2-2. From the sampling frame of 1,112 facilities, a sample of 214 facilities was randomly selected within the 48 strata. Among the 214 facilities that were randomly sampled, five were facilities that received a pre-test of the detailed questionnaire. Two of these pre-test facilities were replaced by redrawing from the sampling frame within the respective strata. One additional facility, which was not a pre-test facility, was identified during the selection process as being closed, so it was replaced by a facility within the same stratum that was not originally sampled. The other three pre-test facilities were in strata from which all of the facilities were sampled, so there were no alternative facilities to use as replacements. For these three facilities, responses from the pre-test were incorporated into the detailed questionnaire response database wherever possible. Therefore, the sample size remained at 214. All 214 of the sampled facilities for the detailed questionnaire were defined to be in-scope. 2-4 ------- Following the selection of the random sample of 214 facilities to receive the detailed questionnaire, 17 additional facilities were selected deliberately by the EPA to receive the detailed questionnaire based upon their wastewater treatment processes. Because these 17 facilities were not randomly selected, they could not be included as part of the random sample. But, because detailed questionnaires were sent to these facilities and responses were available to be used in the calculation of national estimates, they were included in the detailed questionnaire sample. To accommodate this situation, the 17 deliberately-sampled facilities were removed from the appropriate strata in the sampling frame, and were placed in a separate stratum from which it is assumed that all 17 facilities were selected. This adjusted sampling frame, with the additional stratum, is presented in Column (b) of Table 2-2. After the establishment of the sampling frame, the selection of the detailed questionnaire sample, and the mailing of the questionnaires, information was obtained regarding the stratification of the 15 in- . scope facilities that were excluded from the sampling frame due to lack of stratum information in the screener responses. One of these facilities was men found to be out-of-scope from the additional information obtained. Therefore, the sampling frame was increased by 14 to a total of 1,126. These additions to the appropriate strata in the sampling frame are reflected in Column (c) of Table 2-2. After the list of in-scope respondents was established for the detailed questionnaire sampling frame, two additional screener respondents were classified as in-scope. Because these two facilities were not included in the original sampling frame and, thus, were not available for selection to receive the .detailed questionnaire, the population was increased by two facilities. The population also was increased to account for the estimated number of in-scope facilities that did not respond to the screener and, thus, were not included in the original sampling frame for the detailed questionnaire. Of the 208 screener nonrespondents, 122 were identified as possibly being in-scope through follow-up telephone calls. Of these 122 facilities, 3 facilities that are likely to be in-scope were excused by the EPA from completing the screener questionnaire, and 119 screener questionnaires were not returned by the post office, implying that the facilities received the questionnaires, but completed screeners were not returned from the facilities. There were 1,128 in-scope facilities among the 1,500 screener respondents, so it is estimated that 92 of the 122 nonrespondents also are in-scope. This assumes that the proportion of the 122 nonrespondents that are estimated to be in-scope is equivalent to the proportion of respondents that were identified as in-scope. : The population was adjusted for these estimated 92 in-scope facilities plus the two facilities that were declared to be in-scope following the establishment of the original sampling frame. Because stratification information was not known for these facilities, the stratum frames were increased in proportion to the frame sizes. For example, the sample frame for stratum A-l-DI contained 31 facilities and the total frame size was 1,109 (excluding the 17 deliberately-sampled facilities). Therefore, three (94*31/1109) of the 94 in-scope facilities were apportioned to stratum A-l-m. These adjusted stratum frames, totaling 1,220 facilities, are presented in Column (d) of Table 2-2, and were calculated based on the stratum frames under Column (c). i It should be noted that the screener population (in Section 2.1.1) was adjusted by only 89 of the nonrespondents because four of the nonrespondents were duplicated in the Dun & Bradstreet screener sample. However, the detailed questionnaire frame is not affected by these four duplicates because the Dun & Bradstreet screener frame was developed after the Dun & Bradstreet detailed questionnaire frame (as documented in Section 2.2). 2-5 ------- Two of the possibly in-scope nonrespondents to the trade association screener questionnaire were also selected to receive a Dun & Bradstreet detailed questionnaire (i.e., the trade association list and the Dun & Bradstreet list duplicated these two facilities, as documented in Section 2.1). These two facilities were accounted for in the Dun & Bradstreet detailed questionnaire sample to retain the probability structure, because they did not respond to the trade association screener questionnaire. Thus, they had no stratification information for the trade association detailed questionnaire sample. Therefore, the estimated number of in-scope facilities from the screener nonrespondents was calculated from only 120 facilities, rather than 122. The revised estimate is 90 in-scope facilities, dictating an increase in the detailed questionnaire of 92 in-scope facilities, rather than the 94 facilities that were added to create Column (d). These adjusted stratum frame sizes, based on Column (c) and the additional 92 in-scope facilities, are presented under Column (e) of Table 2-2 and result in a total frame size of 1,218 facilities. After the trade association list and the subsequent detailed questionnaire sampling frame were established, the five largest industrial launderers in the nation were examined to identify facilities that may not have been included in the trade association list. There were 48 facilities identified as belonging to these five industrial launderers (Aratex, Cintas, Omni, Unifirst, and Unitog). Also, mailing addresses were identified for four additional facilities that were not originally included in the trade association list due to lack of address information. Abbreviated versions of the screeners were sent to these 52 facilities to obtain information regarding then* operating practices and status. From this information, 29 facilities were identified as being in-scope and did not duplicate facilities originally hi the trade association list. These 29 facilities were added to the adjusted sampling frame into the appropriate strata that were identified from the abbreviated screener responses. These final adjusted stratum frames are presented hi Column (f) of Table 2-2. The final adjusted frame size is 1,247 facilities, from which 231 facilities were sent detailed questionnaires. The final frame sizes for each stratum are listed under Column (f) of Table 2-2 and the sample sizes for each stratum are listed in Column (g). 2-6 ------- Table 2-2 Trade Association Detailed Questionnaire Sampling Frame Report >, } K~? >• ''• Stratum Items A A A A A A A A A A A A B B B B B B B B B B B B C C Revenue 1 1 1 2 2 2 3 3 3 4 4 4 1 1 1 2 2 2 3 3 3 4 4 4 1 1 Treatment I n m I n m i n m i n m i n m i n m i n m i n m i n Original Rrame (a) I 2 7 31 12 22 71 8 35 44 9 42 . 30 4 6 57 15 35 109 5 42 90 5 25 21 1 1 C l K J _^ i -1 - - - '""••/ Intermediate Frames ** ' „ *' -W' 2 7 31 12 22 69 8 33 42 9 40 28 4 6 56 15 35 108 5 41 90 5 25 21 1 1 <$_ 2 7 31 12 22 71 8 34 42 9 40 28 4 6 57 15 36 109 5 41 91 ; • 5 25 21 1 1 '(d) 2 ! 8 :34 13 '24 77 9 :37 46 ; 10 ! 43 30 4 1 7 62 16 :39 118 5 45 : 99 5 27 ; 23 ' 1 1 » - 2 8 34 13 24 77 9 37 45 10 43 30 4 7 62 16 39 118 5 44 99 5 27 23 1 1 final Frame" k . «3>:- (f) • 2 8 34 13 24 77 9 37 45 10 43 31 4 7 62 16 41 121 5 50 106 5 29 26 1 1 Sample Size (Oh) (g) 2 7 4 12 4 4 8 4 4 9 4 4 4 6 4 15 4 4 5 4 4 5 4 4 1 1 2-7 ------- Table 2-2 Trade Association Detailed Questionnaire Sampling Frame Report (Continued) SF Stratum Items C C C C C C C C C C D D D D D D D D D D D D Revenue 1 2 2 2 3 3 3 4 4 4 1 1 1 2 2 2 3 3 3 4 4 4 Treatment m I n m I n m i n m i n m i n m i n m i n m Deliberate-sample Total Original Frame Ca) 21 1 3 34 3 9 26 1 3 15 5 2 80 4 6 69 8 9 51 5 9 19 1112 V i i Intermediate Frames :. 0>> 21 1 3 34 3 9 26 1 3 14 .5 2 80 4 6 69 8 9 49 5 9 18 17 1112 (c) 21 1 3 36 3 10 26 1 3 14 5 2 81 4 6 71 8 9 50 5 9 18 17 1126 (d) 23 1 3 39 3 11 28 1 3 15 5 2 88 4 7 77 9 10 54 5 10 20 17 1220 . 23 1 3 39 3 11 28 1 3 15 5 2 88 4 7 77 9 10 54 5 10 20 17 1218 Final Frame /'<&>'.., .",/*-* '#-,* 23 1 3 39 3 11 28 1 3 16 5 2 88 4 7 79 9 10 55 5 10 21 17 1247 l, Sample - Size " \------- 2.2 Dun & Bradstreet Population The Dun & Bradstreet listing was used to increase the population of industrial laundry facilities to include facilities that were not captured by the trade association lists. Additional screener and detailed questionnaires were sent to a random sample of the facilities that were identified as industrial laundry facilities, but were not included in the trade association lists. From the Dun & Bradstreet listing, 24 facilities were selected to receive the additional detailed questionnaires. Following the selection of the sample for the additional detailed questionnaire, 200 of the facilities in the Dun & Bradstreet listing were chosen to receive additional screener questionnaires. Therefore, the development of the detailed questionnaire frame is discussed prior to the discussion of the additional screener questionnaire frame in this section. 2.2.1 Dun & Bradstreet Detailed Questionnaire : Industrial laundry facilities were identified in the Dun & Bradstreet listing as facilities with primary SIC codes of 7218 (industrial laundering) or 7213 (linen supply servicing), and facilities with a secondary SIC code of 7218. These three SIC code categories were used to define three strata for the sampling design: (1) primary SIC code 7218, (2) primary SIC code 7213, and (3) secondary SIC code 7218. The Dun & Bradstreet listing was compared with the facilities that responded to the trade association screener questionnaire to avoid duplication of the facilities. Duplicate facilities within the Dun & Bradstreet listing and between the Dun & Bradstreet listing and the trade association respondents were removed from the Dun & Bradstreet sampling frame. This resulted in a sampling frame of 2,249 facilities (714 in D&B Stratum 1; 1,372 in D&B Stratum 2; 163 in D&B Stratum 3). i Twenty-four in-scope facilities were selected by randomly sampling facilities from the sample frame and calling each facility to verify that it was in-scope (i.e., generated wastewater). If a selected facility was hot in-scope, another facility was randomly selected. This process was continued until 24 in-scope facilities were identified. To gain 24 in-scope facilities (12 from D&B Stratum 1; 7 from D&B Stratum 2; 5 from D&B Stratum 3), a total of 66 facilities were selected (36 from D&B Stratum 1; 19 from D&B Stratum 2; 11 from D&B Stratum 3). In order to develop survey weights, this selection process is treated as a stratified random sample of 66 facilities, of which 24 were found to be in-scope. During the sampling process, one of the 24 in-scope facilities was identified as a management facility, so it was replaced by a randomly selected facility from the same stratum, thus increasing the total number of facilities sampled to 67. But, because the replacement facility was randomly selected from the sampling frame, the survey weights (66 facilities selected, of which, 24 are in-scope) are not affected; that is, each facility had the same probability of being selected in the sample. During the development of the Dun & Bradstreet screener frame (as documented in Section 2.2.2), duplicate facilities were found between the Dun & Bradstreet listing and the respondents to the trade association screener questionnaire. The revised sampling frame, adjusted to remove these additional duplicates, is 1,977. However, this adjusted sampling frame also removed the 67 facilities that were selected for the detailed questionnaire, in addition to the duplicate facilities. Therefore, the 66 detailed questionnaire facilities (not including the one management facility that was replaced) should not be removed from the frame. The adjusted sampling frame, after removing only the additional duplicate facilities, is 2,043 facilities (631 in D&B Stratum 1; 1331 in D&B, Stratum 2; 81 in D&B Stratum 3). 2-9 ------- Duplicate facilities also were found between the Dun & Bradstreet listing and the nonrespondents to the trade association screener questionnaire. There were 73 duplicates found with the trade association nonrespondents that should be removed from the Dun & Bradstreet frame. However, two of these facilities had been previously selected to receive the Dun & Bradstreet detailed questionnaire. Therefore, the Dun & Bradstreet detailed questionnaire frame was reduced by only 71 facilities. The two duplicate facilities that were included in the Dun & Bradstreet detailed questionnaire sample were removed from the trade association detailed questionnaire frame (as documented in Section 2.1.2). The final sampling frame for the Dun & Bradstreet detailed questionnaire is 1,972 (605 in D&B Stratum 1; 1286 in D&B Stratum 2; 81 in D&B Stratum 3). The final Dun & Bradstreet detailed questionnaire sampling frame (N), sample sizes (n), and the number of sampled facilities that are in-scope (n') are summarized in Table 2-3. 2.2.2 Dun & Bradstreet Screener Questionnaire The sampling frame for the Dun & Bradstreet screener questionnaire was established after the Dun & Bradstreet detailed questionnaire was administered. During the development of this frame, duplicate facilities were found between the Dun & Bradstreet listing and the respondents to the trade association screener questionnaire. This resulted in a different sampling frame than was used for the Dun & Bradstreet detailed questionnaire. This sampling frame contained 1,977 facilities (595 in D&B Stratum 1; 1312 in D&B Stratum 2; 70 in D&B Stratum 3). This frame does not include the facilities that were selected for the Dun & Bradstreet detailed questionnaire. • These facilities should be included in the screener questionnaire frame because they are known to be in the population. Therefore, the frame was adjusted to include the 66 detailed questionnaire facilities (not including the one management facility that was replaced). The resulting frame contains 2,043 facilities (631 in D&B Stratum 1; 1331 in D&B Stratum 2; 81 in D&B Stratum 3). Duplicate facilities were also found between the Dun & Bradstreet sampling frame for the screener questionnaire and the nonrespondents to the trade association screener questionnaire. There were 60 duplicates found with the trade association nonrespondents that should be removed from the Dun & Bradstreet frame. However, five of these facilities also were selected to receive the Dun & Bradstreet screener questionnaire. Therefore, the Dun & Bradstreet screener questionnaire frame was reduced by only 55 facilities. The five duplicate facilities that were included in the Dun & Bradstreet screener questionnaire sample were removed from the trade association screener questionnaire frame (as documented in Section 2.1.1). The adjusted sampling frame for the Dun & Bradstreet screener questionnaire is 1,988 (613 in D&B Stratum 1; 1294 in D&B Stratum 2; 81 in D&B Stratum 3). From the Dun & Bradstreet screener questionnaire frame, 200 facilities were randomly selected. This sample was selected from the three SIC code strata and included 100 facilities from D&B Stratum 1, 60 facilities from D&B Stratum 2, and 40 facilities from D&B Stratum 3. From the 200 screener questionnaires that were mailed, responses were received from 133 facilities. Among these responses, 6 facilities were identified as duplicates because they received a previous screener questionnaire. Therefore, these 6 facilities were removed from the sampling frame and the sample. The final sampling frame for the Dun & Bradstreet screener questionnaire is 1,982 (608 in 2-10 ------- D&B Stratum 1; 1293 in D&B Stratum 2; 81 in D&B Stratum 3). The final Dun & Bradstreet screener questionnaire sampling frame (N), sample sizes (n), number of respondents, and number of in-scope respondents are summarized in Table 2-4. Among the 127 unique respondents, 11 facilities were identified as out of scope, because they were sold, out of business, or not an industrial laundry facility. Therefore, the estimated number of in-scope facilities within each stratum has been estimated. Table 2-3a summarizes the estimated in-scope sampling frame (N') and number of in-scope respondents (n'). : Table 2-3 Dun & Bradstreet Detailed Questionnaire Sampling Frame »"V - <- -„ *•":."•. * ^ ^_-= ^&*z -?, * L - •" .:A--* : * '~ " ",-!, -,.--"' f •*""" j* ^ - ,.«' - ^rJSjfccatum , rj / 1. Primary SIC 7218 2. Primary SIC 7213 3. Secondary SIC 7218 Total Sample Frame "" '-';(& * '": 605 1268 81 1972 "± * .A Sample Size : ' '-.&. \ 36 19 11 66 t 'i! js Ol 'I JSfumber of In-Scope Sampled ^acililies (H!> 12 7 5 24 Table 2-3a Dun & Bradstreet Detailed Questionnaire Sampling Frame for In-scope Facilities - '-* -- -„ ~ ^~",-«*~ . ., * . •i, * -K, J<-"^K -<• ^ K , „ ^ /"f - / _ ' • ; *- . tstratuin " , 1. Primary SIC 7218 2. Primary SIC 7213 3. Secondary SIC 7218 Total Sample Frames . ^ en ---i 202 474 37 i 713 1 Number of In-Scope Sampled Facilities (n') * 12 7 5 24 2-11 ------- Table 2-4 Dun & Bradstreet Screener Questionnaire Sampling Frame Stratum 1. Primary SIC 7218 2. Primary SIC 7213 3. Secondary SIC 7218 Total Sample Frame (N) 608 1293 81 1982 t- i fc t uV Sample Size - (n)" 95 59 40 194 Number of Respondent Facilities 58 41 28 127 i k "> t, Number of In-scope ; Respondent * Facilities 51 39 26 116 2.3 Hotels, Hospitals and Prisons Screener Questionnaire In response to comments from industrial laundry and linen supply trade associations, the EPA mailed out screener questionnaires to facilities such as hospitals, hotels, and prisons (HHP's). The trade associations indicated that the HHP's may generate revenues by accepting laundry from off-site, thereby reducing the profits of more traditional industrial laundries and linen supply facilities. The EPA mailed 100 screener questionnaires to HHP's. HHP's are not traditional industrial laundry facilities, but generate wastewater from laundering. To obtain the 100 facility addresses, the EPA randomly selected 25 facility addresses from each of four lists (see Table 2-5). The results of this survey effort cannot be used to estimate a national number of these types of facilities or national estimates of any characteristics of these types of facilities. The purpose of the sampling was to get a snapshot of the activities of nontraditional laundries to help determine whether these facilities should be considered within the scope of the regulation. 2-12 ------- Table2-5 HHP Screener Questionnaire Sampling Frame List Name Textile Rental Services Association of America (TRSA) Uniform and Textile Service Association (UTSA) Responses to Question 25(Q25)inthe Industrial Laundries Detailed Questionnaire (as of 11/18/94) National Association of Institutional Linen Management (NAILM) Members Total Addresses 2,416 208 312 1,504 Complete Addresses none (0) all (208) some some (1,057) i Incomplete Addresses All (2,416) none (0) some some (447) The estimated number of facilities in the mailing that were anticipated to be industrial laundries, prisons, health care facilities, hotels, and miscellaneous industrial facilities are displayed in Table 2-6. Table 2-6 Estimated number of HHP facilities in the mailing by Facility type , List ^Name TRSA UTSA Q25 NAILM Total -•" ""^L •4< V""^" l?acifity'J^Pe " * *- '" Prisons * " »% v w 3 0 3 0 6 ' Hef ttf>_ '^•" 8 14 7 23 52 t. *• Hotels - 6 1 7 1 15 Industrial Laundries 1 4 2 0 7: Misc. Industrial 7 6 6 1 20 "$ j, •* jjv/*'! , Total* J m , Mailing 25 25 25 25 100 2-13 ------- 2.4 Industrial Laundries Population The industrial laundries industry was characterized through the use of a screener questionnaire and was distributed a detailed questionnaire. The detailed questionnaire was distributed to a stratified simple random sampling of facilities from two mutually exclusive populations of industrial launderers identified through two sources; trade association listings and information obtained from Dun & Bradstreet. 2.4.1 Final Detailed Questionnaire Design Through the process of generating national estimates for the industrial laundries industry, an issue was identified with regard to the stratification of the trade association population listing. One basic motivation for designing a stratified sampling design is to reduce variability through the identification of homogeneous strata. That is, stratification is based on the grouping of like sampling elements. The industrial laundries industry trade association population listing was stratified by types of items laundered, revenue range, and type of wastewater treatment. Table 2-7 displays the trade association detailed questionnaire sampling frame previously described hi Section 2.1.2. Notice that four strata, C-l-I, C-l-E, C-2-I, and C-4-I contain only one facility each. Additionally, notice that within several strata all facilities within each stratum eligible to receive a detailed questionnaire were sent a detailed questionnaire. Many of these strata have small sample sizes. Due to the small sample sizes within these strata, it is difficult to assess the presence of homogeneity within strata or heterogeneity between strata. In addition, since sample unit nonresponse occurs (addressed later in this section) within these small strata, all strata in which all eligible facilities received detailed questionnaires were collapsed into one stratum. By collapsing all of these strata into a single stratum, sample unit nonresponse may be distributed over a larger number of facilities. This will eliminate any one facility or few facilities "over-representing" or "misrepresenting" the nonrespondents. As indicated, sample unit nonresponse occurred hi this sampling effort. Sample unit nonresponse is defined by a sampling unit, i.e., a facility, not returning a detailed questionnaire or not providing enough information hi the detailed questionnaire responses to adequately identify a facility as a respondent. If the response rate is considered to be random, then the sample of respondents is considered to be a simple random sample, and it is assumed that there are no differences between the set of respondents and nonrespondents. Due to the presence of sample unit nonresponse, an adjustment was made to the detailed questionnaire sampling design as described below. Stratum B-l-n contains only one respondent to the detailed questionnaire, as indicated in Table 2-7. Since there is only one respondent in stratum B-l-n, this stratum was collapsed with stratum A-l-II. Although the original stratum weights are not identical, the difference is rninimal. By collapsing these strata, the sample unit nonresponse adjustment will be distributed across the seven sites hi the collapsed stratum (W'h=2.14) versus retaining the original B-l-II stratum with an adjusted weight of 7.00. Therefore, all censused strata were collapsed into a single stratum and strata A-l-II and B-l-II were collapsed into a single stratum. Once these strata were collapsed, weights were adjusted for sample 2-14 ------- unit nonresponse in the following manner: n. where rh = the number of sampled units in stratum h responding to the detailed questionnaire and Nh and % are as defined in Table 2-7. Table 2-8 displays the final detailed questionnaire sample design for the combined trade association and Dun & Bradstreet frames with sample unit non-response adjustment. For each stratum, the number of facilities in the population, the number of facilities sampled, the number of respondent facilities, the original weight, and the adjusted weights are presented. , 2-15 ------- Table 2-7 Trade Association Detailed Questionnaire Sampling Frame -Stratum ' , - Items A A A A A A . A A A A A A B B B B B B B B B B B B C Revenue 1 1 1 2 , 2 2 3 3 3 4 4 4 1 1 1 2 2 2 3 3 3 4 4 4 1 Treatment I n m I n ffl I n m i n m i n m i n in i n m i n m i Final Frame 2 7 4 12 4 4 8 4 4 9 4 4 4 6 4 15 4 4 5 4 4 5 4 4 1 J Number of Respondents ' , >h> '^ 2 6 3 10 4 3 8 2 4 9 4 4 3 1 2 12 3 4 4 4 4 4 4 4 0 i i Stratum ' Weight J (WhT1 1.00 1.14 8.50 1.08 6.00 19.25 1.13 9.25 11.25 1.11 10.75 7.75 1.00 1.17 15.50 1.07 10.25 30.25 1.00 12.50 26.50 1.00 7.25 6.50 1.00 2-16 ------- Table 2-7 Trade Association Detailed Questionnaire Sampling Frame (Continued) ->. ~ ~ \ i * Stratum Items C C C C C C C C C C C D D D D D D D D D D D D Revenue 1 1 2 2 2 3 3 3 4 4 4 1 1 1 2 2 2 3 3 3 4 4 4 Treatment n m I n m i n m i n m i n m i n m i n m i n m Deliberate-sample Total Final Frame (¥*) i 23 1 3 39 3 11 28 1 3 16 -5 2 88 4 7 79 9 10 55 5 10 21 17 1247 Sample Size - ------- Table 2-8 Industrial Laundries Proposed Detailed Questionnaire Sampling Frame with Sample Unit Nonresponse Adjustment and Single PSU Adjustment Stratum '- Items Revenue Treatment Collapsed Census Stratum A,B A A A A A A A A A A B B B B B B B B C C C C 1 1 2 2 2 3 3 3 4 4 4 1 2 2 2 3 3 4 4 1 2 3 3 n m I n m i n m i n m m i n m n m n in m m n m , final •JFrame^ (N*) 62 15 34 13 24 77 9 37 45 10 43 31 62 16 41 121 50 106 29 26 23 39 11 28 Sample Size "(Dh) 62 13 4 12 4 4 8 4 4 9 4 4 4 15 4 4 4 4 4 4 4 4 4 4 Number of, Respondents- fey " 45 7 3 10 4 3 8 2 . 4 9 4 4 2 12 3 4 4 4 4 4 4 3 4 4 i Stratum •Weight (W 1.00 1.15 8.50 1.08 6.00 19.25 1.13 9.25 11.25 1.11 10.75 7.75 15.50 1.07 10.25 30.25 12.50 26.50 7.25 6.50 5.75 9.75 2.75 7.00 Adjusted Stratum Weight , <&*> : 1.38 2.14 11.33 1.30 6.00 25.67 1.13 18.50 11.25 1.11 10.75 7.75 31.00 1.33 13.67 30.25 12.50 26.50 7.25 6.50 5.75 13.00 2.75 7.00 2-18 ------- Table 2-8 Industrial Laundries Proposed Detailed Questionnaire Sampling Frame with Sample Unit Nonresponse Adjustment and Single PSU Adjustment (Continued) Stratum Items C D D D D D D D D Revenue 4 1 2 2 3 3 3 4 4 Treatment m m n m i n m n m D&B: P7213 D&B: P7218 D&B: S7218 Total Final JErame * 16 88 7 79 9 10 55 10 21 474 202 37 1960 Sample - Size &a 4 4 4 4 8 4 4 4 4 7 12 5 255 Number of Respondents 4 ~ \ &K 4 :4 :4 ,2 :7 '4 3 4 ,4 !6 i9 '2 208 Stratum Weight ' (w*) 4.00 22.00 1.75 19.75 1.13 2.50 13.75 2.50 5.25 67.68 16.81 7.36 Adjusted Stratum Weight, (WV 4.00 22.00 1.75 39.50 1.29 2.50 18.33 2.50 5.25 78.96 22.41 18.41 2-19 ------- ------- CHAPTER 3 ESTIMATION METHODOLOGY This section presents the general methodology and equations for calculating estimates from the Industrial Laundries detailed questionnaire sampling efforts. 3.1 Detailed Questionnaire A stratified random sample of 255 industrial laundry facilities (231 in Trade Association, 24 in Dun and Bradstreet) was selected from the 1,960 facilities in the population. Of the 255 facilities that received detailed questionnaires, 208 facilities responded. 3.1.1 Estimation from Complete Data Many characteristics of interest estimated from the detailed questionnaire responses were provided by every detailed questionnaire respondent (PSU respondent). Therefore, based on a stratified simple random sample with complete response, stratum weights are used to obtain mean estimates from a continuous response variable. The stratum weights are the proportion of available facilities in each stratum (Wh = Nh/N), where Nh is the total number of available facilities for the sample from stratum h and N is the total number of available facilities for the sample (N=1,960). The sampling fraction, which is used to estimate totals from a continuous response or the; total number of facilities with a given characteristic within each stratum, is the fraction of facilities within each stratum that are sampled (fj, = The stratum weights, Wh = Nh/N, are used to estimate means according to the following formula: N 'h . 7=1 N (3.1) where, N = total number of facilities (N= 1,960) Nh = total number of facilities in stratum h HJ, = number of facilities sampled in stratum h y^ = response from i"1 facility in stratum h The variance of the estimated mean is: n. N2 h (3.2) where 3-1 ------- The estimated total number of facilities with a given attribute, or the estimated total from a continuous response is: (3.3) where, Nh = total number of facilities in stratum h % = number of facilities sampled hi stratum h yu = response from 1th facility in stratum h The estimated number of facilities, within stratum h, with a given attribute assumes VM = 1 if the 1th facility has the given characteristic 0 if the 1th facility does not have the given characteristic. The variance of the estimated total is: (3.4) where 3.1.2 Estimation with Item-level Non-Response If responses are available from only .mh of the % sampled facilities, then the population can be considered to be divided into two domains: respondents and non-respondents. The estimated mean for the domain of respondents can be used as an estimate of the population mean, assuming that the non- respondent facilities operate at the mean of the responding facilities.2 The estimated mean for the respondents is: 1 Cochran, W. G., Sampling Techniques, 3rd ed., New York: John Wiley and Sons, Inc., 1977. 3-2 ------- Y = N n (3-5) where, Nh = total number of facilities in stratum h ; i^ = number of facilities sampled in stratum h i mh = number of respondent facilities to the characteristic of interest in stratum h Yhi = response from 1th facility in stratum h. ; The estimated variance is: i m n. (3.6) If responses are available from only m^ of the n,, sampled facilities, then the estimated total is: 7 = (3.7) where, N = total number of facilities (N=1960) : Nh = total number of facilities in stratum h % = number of facilities sampled in stratum h : mh = number of respondent facilities to the characteristic of interest in stratum h Vhi = response from i"1 facility in stratum h. ' This assumes that the proportion of facilities with the given attribute, or the average response, is the same in the set of non-respondents as in the set of respondents. ; The estimated variance is: 3-3 ------- (3.8) 3.1.3 Estimation for Domains with Complete Response If estimates are to be calculated for a specific subset (domain) of the data other than the strata used in the sample design, then the formulae must be adjusted. An example of a domain estimate would be the estimated number of facilities within ranges of daily water flow, where the ranges of daily water flow are the domains. If there is complete response within each domain, the estimated mean is similar to equation (3.1), except that the responses, y^, and the number of facilities sampled, are restricted to the j* domain. The estimated domain mean is: - v^ f Nh } r'-?bN" •"A . »'=1 N (3.9) where, N = total number of facilities (N=1960) Nh= total number of facilities hi stratum h HJ, «= number of sampled facilities hi stratum h By = number of sampled facilities in stratum h of domain j Vhjj = response from i* facility hi stratum h of domain j. The variance of the estimated domain mean is: rvrj) = -T _ 1 N2 h -.,>••-E n. (3.10) where N = total number of facilities (N=1960) Nh = total number of facilities hi stratum h nj, = number of facilities sampled hi stratum h y^ = response from i* facility hi stratum h of domain j. The estimated total number of facilities in a domain with a given attribute, or the estimated domain total from a continuous response is: 3-4 ------- (3.11) where, Nh = total number of facilities in stratum h % = number of facilities sampled in stratum h ny = number of facilities sampled in stratum h of domain j yuj = response from 1th facility in stratum h of domain j. The estimated number of facilities, within stratum h of domain j, with a given attribute assumes Vhij = 1 if the 1th facility has the given characteristic 0 if the i* facility does not have the given characteristic. The variance of the estimated total is: = E » (3,12) where 3.1.4 Estimation for Domains with Item-level Non Response When there are responses from only m^ of the n^ sampled facilities, each domain is divided into respondents and non-respondents. If, however, a facility does not provide a response to .the domain of interest, it is not possible to characterize that facility. Therefore, facilities are excluded from estimates based on a domain to which that facility did not respond. The estimated mean for a given domain is calculated from the set of respondents within that domain. This assumes that the mean of the non- respondents in each domain is equivalent to the mean of the respondents in each domain. The estimated domain mean when item-level non response exists is: ; • YJ = •"m. n. (3.13) 3-5 ------- where, Nh = number of facilities in stratum h of the sample frame % = number of sampled facilities in stratum h my = number of respondent facilities to characteristic of interest in stratum h of domain j Dy* = number of respondents to domain identification question(s) in domain j y^ SB response from 1th facility in stratum h of domain j. The variance of the estimated domain mean is: i .v_l_M. -— _y-v2 (3.14) where The estimated total is calculated from the estimated mean, as hi equation (3.13), except that the population size (N) is adjusted for the estimated population size of the jm domain (Nj). where, Nh = I AT ^M I %•£*,] (3.15) h \ n number of facilities in stratum h of the sample frame number of sampled facilities in stratum h number of respondent facilities to characteristic of interest in stratum h of domain j number of respondents to domain identification questions) in domain j response from 1th facility in stratum h of domain j. The estimated variance for the domain total assumes that the population size for domain j is known. The calculation is similar to equation (3.14). It is noted that this estimate may contain bias due to the assumption that the population size of the domain is known, when, in practice, it must be estimated from the known size of the sample and the fraction of respondents in domain j. 3-6 ------- V(YJ) = I — (3.16) 3.2 National Estimates Using the estimation methodology described above, national estimates were calculated to determine the estimated number of facilities by different domains of interest. National estimates are based on facility responses to the detailed questionnaire in 1993. The following tables present the estimated number of facilities in 1993 by revenue range, total production per year, items laundered, employee range, and annual flow. The number of respondents to the domain of interest, estimated total number of facilities, estimated standard error, and 95% confidence intervals are presented separately for facilities processing less than 100% linen (193 responding facilities), facilities processing at least one million pounds of total laundry per year and/or at least 255,000 pounds of printer and shop towels (172 responding facilities), and facilities processing less than one million pounds total laundry per year and less than 255,000 pounds of printer and shop towels. 3-7 ------- Table 3-1 Estimated Number of Facilities in 1993 by Revenue Range [ Revenue RJangeH Facilities Processing < 100% Linen < $1,000,000$1,000,000 - $3,499,999$3,500,000 - $6,999,999$7,000,000 - $10,499,999 Overall* 22 51 69 51 193 198 564 664 321 1,747 36.57 124.63 114.68 84.85 93.79 126 - 269 320 - 809 440 - 889 155 - 488 1,564-1,931 Facilities Processing at least 1 Million Ibs Total Laundry per Year and/or at least 255,000 Ibs Printer and Shop Towels <$1,000,000 $1,000,000 -$3,499,999 $3;500,000 -$6,999,999 $7,000,000 -$10,499,999 Overall* 11 44 66 51 , 172 118 508 659 321 1,606 24.56 122.16 • 114.67 84.85 95.63 69 - 166 268 - 747 434 - 884 155 - 488 1,418-1,793 Facilities Processing < 1 Million Ibs Total Laundry per Year and < 255,000 Ibs Printer and Shop Towels < $1,000,000$1,000,000 - $3,499,999$3,500,000 - $6,999,999 Overall* 11 7 3 21 80 56 5 141 32.99 31.19 2.18 39.79 . 15 - 145 0-117 1-10 63 - 219 * Overall estimates may not equate to the sum of the domains due to rounding 3-8 ------- Table 3-2 Estimated Number of Facilities in 1993 by Total Production Range Facilities Processing < 100% Linen < 1,000,000 1,000,000-1,999,999 2,000,000 - 2,999,999 3,000,000 - 3,999,999 4,000,000 - 4,999,999 5,000,000 - 5,999,999 6,000,000 - 6,999,999 7,000,000-7,999,999 . ^10,000,000 Overall* 24 25 13 22 18 15 17 24 35 193 167 264 211 231 254 144 116 116 245 1,747 42.70 87.17 92.42 ; 56.41 96.37 84.01 1 35.83 45.26 ; 84.75 93.79 83 - 250 93 - 435 30-392 120 - 342 65-443 0-309 45 - 186 27 - 204 79 - 412 1,564-1,931 Facilities Processing at least 1 Million Ibs Total Laundry per Year and/or at least 255,000 Ibs Printer and Shop Towels < 1,000,000 1,000,000-1,999,999 2,000,000 - 2,999,999 3,000,000-3,999,999 4,000,000 - 4,999,999 5,000,000 - 5,999,999 6,000,000 - 6,999,999 7,000,000 - 7,999,999 z 10,000,000 Overall* 3 25 13 22 18 15 17 24 35 172 25 264 211 231 254 144 116 116 245 1,606 i 21.92 87.17 , 92.42 ; 56.41 96.37 ; 84.01 35.83 45.26 84.75 : 95.63 0-68 93 - 435 30 - 392 120 - 342 65-443 0-309 45 - 186 27-204 79 - 412 1,418 - 1,793 Facilities Processing < 1 Million Ibs Total Laundry per Year and < 255,000 Ibs Printer and Shop Towels < 1,000,000 Overall* 21 21 141 141 39.79 ': 39.79 63 - 219 63 - 219 * Overall estimates may not equate to the sum of the domains due to rounding 3-9 ------- Table 3-3 Estimated Number of Facilities in 1993 by Items Laundered (Item -•,;-; s-si Estimated Facilities Processing < 100% Linen Industrial Garments Shop Towels, Wipers, etc. Printer Towels Floor Mats Mops, Dust Cloths, etc. Linen Supply Garments Linen Flatwork/Flat Dry Health Care Item Types Fender Covers Continuous Roll Towels Clean Room Garments Other Item Types Laundry Bags Family Laundry New Item Types Executive Wear Miscellaneous Rewash Item Types Filters Buffing Pads 165 141 71 179 162 110 129 78 75 98 9 2 3 6 9 5 2 5 2 1 1,462 1,332 480 1,654 1,529 942 1,364 649 687 928 28 31 28 84 74 43 14 39 7 6 103.85 107.60 57.81 99.62 95.24 134.01 109.05 126.66 117.18 128.59 12.50 29.75 25.17 44.21 38.22 23.89 12.51 26.10 5.52 5.48 1,258 - 1,666 1,121 - 1,543 367 - 594 1,459-1,850 1,342 - 1,716 679 - 1,205 1,150-1,577 400-897 458 - 917 676-1,180 4-53 ' 0-90 0-77 0-171 0-149 0-90 0-39 0-90 0-18 0-17 3-10 ------- Table 3-3 Estimated Number of Facilities in 1993 by Items Laundered (Continued) Item' ~£ V'r^/^l ltem -« ^X J « kV v,,f>-O^ . •> * „ " % 1 '<• ? r* c ^ -^ "s Number of ,> Respondents "% , *4< (s f'*" Estimated To|al < J/ V- t. X ^ Estimated "^ Standard Error f l - 95%£X ,- ^ "~ "'v\ ";s,!H /-V H X-" Facilities Processing at least 1 Million Ibs Total Laundry per Year and/or at least 255,000 Ibs Printer and Shop Towels Industrial Garments Shop Towels, Wipers, etc. Printer Towels Floor Mats Mops, Dust Cloths, etc. Linen Supply Garments Linen Flatwork/Flat Dry Health Care Item Types Fender Covers Continuous Roll Towels Clean Room Garments Other Item Types Laundry Bags Family Laundry New Item Types Executive Wear Miscellaneous Rewash Item Types Filters Buffing Pads 150 129 67 164 153 104 120 72 72 • 93. 5 2 3 4 9 3 2 5 1 1 1,380 1,270 464 1,564 1,472 925 1,283 607 675 903 22 31 28 77 74 15 14 39 6 6 105.61 107.48 ! 57.82 97.11 96.97 133.87 ; 106.73 124.71 'i 117.02 128.45 • 12.30 ' 29.75 25.17 43.89 38.22 ; 9.03 12.51 . ; 26.10 5.48 ; 5.48 1,173 - 1,587 1,060 - 1,481 351 - 577 1,374-1,755 1,282 - 1,662 663 - 1,188 1,073 - 1,492 362-851 445-904 651 - 1,155 0-46 0-90 0-77 0-163 0-149 0-33 0-39 0-90 0-17 0-17 3-11 ------- Table 3-3 Estimated Number of Facilities in 1993 by Items Laundered (Continued) 1 Item Number of ^ Respondents Estimated Total Estimated, . Standard Error 95%Cy£ ,- "^ '', " ^/r'"vy *', k '"HI"' Facilities Processing < 1 Million Ibs Total Laundry per Year and < 255,000 Ibs Printer and Shop Towels Industrial Garments Shop Towels, Wipers, etc. Printer Towels Floor Mats Mops, Dust Cloths, etc. Linen Supply Garments Linen Flatwork/Flat Dry Health Care Item Types Fender Covers Continuous Roll Towels Clean Room Garments Family Laundry Executive Wear Filters 15 12 4 15 9 6 9 6 3 5 4 2 2 1 82. 61 16 90 57 17 81 42 13 • 26 7 7 28 1 29.10 25.09 10.98 27.18 25.06 6.19 27.09 22.14 6.08 12.43 2.28 5.28 22.12 0.72 25 - 139 12-110 0-38 37 - 143 8-106 5-29 28 - 134 0-85 1-25 1-50 2-11 0-17 0-71 0-3 3-12 ------- Table 3-4 Estimated Number of Facilities in 1993 by Employee Range Employee Range ' - ' .*> \. NO: of t*- Respondents .. .. *.-:**•'- Estimated Total i^ ** f Estimated :•• Standard Error^ f* 95% CJ. ll. L *" i J> A V Facilities Processing < 100% Linen <10 10-29 30-64 65-99 100 - 199 *200 Overall* 3 33 47 52 49 9 193 39 316 491 583 296 23 1,747 ; 24.98 88.78 | 106.49 ; 128.26 85.88 8.02 93.79 0-88 142 - 490 282 - 700 331 - 834 127 - 464 7-39 1,564-1,931 Facilities Processing at least 1 Million Ibs Total Laundry per Year and/or at least 255,000 Ibs Printer and Shop Towels 10-29 ' 30-64 65-99 100 - 199 ;>200 Overall* 21 41 52 49 9 172 244 461 583 296 23 1,606 84.57 ; 104.27 128.26 85.88 8.02 95.63 78 - 409 257 - 665 331 - 834 127 - 464 7-39 1,418-1,793 Facilities Processing < 1 Million Ibs Total Laundry per Year and < 255,000 Ibs Printer and Shop Towels <10 10-29 30-64 Overall* 3 12 6 21 39 72 30 141 ; 24.98 ; 31.21 21.62 . 39.79 0-88 11 - 133 0-72 63 - 219 * Overall estimates may not equate to the sum of the domains due to rounding 3-13 ------- Table 3-5 Estimated Number of Facilities in 1993 by Annual Flow Ranges (Gallons per Year) 1 Annual Flow No. of Respondents Estimated Total Estimated s Standard Error » ^95%C.L;^3 *' i *>, \ *, . < , * "X 1 "'V Facilities Processing < 100% Linen < 1,000,000 1,000,000 - 4,999,999 5,000,000 - 9,999,999 10,000,000 - 19,999,999 20,000,000 - 29,999,999 *30,000,000 Overall* 4 35 41 59 26 28 . 193 31 319 471 502 244 181 1,747 22.54 97.70 108.57 106.99 95.88 81.34 93.79 0-75 128-511 258 - 684 292-711 56 - 432 21 - 340 1,564 - 1,931 Facilities Processing at least 1 Million Ibs Total Laundry per Year and/or at least 255,000 Ibs Printer and Shop Towels 1,000,000 - 4,999,999 5,000,000 - 9,999,999 10,000,000 - 19,999,999 20,000,000 - 29,999,999 ^30,000,000 Overall* 18 41 59 26 28 172 209 471 502 244 181 1,606 92.38 108.57 106.99 95.88 81.34 95.63 27- 390 258 - 684 292-711 56 - 432 21 - 340 1,418-1,793 Facilities Processing < 1 Million Ibs Total Laundry per Year and < 255,000 Ibs Printer and Shop Towels < 1,000,000 1,000,000-4,999,999 Overall* 4 17 21 31 111 141 22.54 35.35 39.79 0-75 41 - 180 63 - 219 * Overall estimates may not equate to the sum of the domains due to rounding 3-14 ------- CHAPTER 4 ANALYTICAL DATA COLLECTION EFFORTS AND DEFINITION OF OPTIONS Description of Data Sources The data used to calculate the proposed pretreatment standards for existing sources (PSES) and pretreatment standards for new sources (PSNS) limitations were collected from the following two sources: (1) the EPA wastewater sampling effort and (2) the self-monitoring data submitted by the facilities in response to the detailed monitoring questionnaire. The EPA wastewater sampling effort resulted in a database containing the results of intensive sampling efforts conducted between February 1993 and April 1997 at 8 facilities. The self-monitoring data were supplied by 37 sites in the 1995 Detailed Monitoring Questionnaire (DMQ). A listing of the data used to support PSES and PSNS standards development can be found in Appendices A.I and A.2. 4.1 EPA Wastewater Sampling ; The EPA wastewater sampling effort consisted of five 24-hour composite samples collected at each of the 8 facilities. For most EPA-sampled facilities, five analytical data values were available for each pollutant at each sampling point. Extensive documentation of the data quality reviews can be found in Chapter 9 of the Technical Development Document for Proposed Pretreatment Standards for Existing and New Sources for the Industrial Laundries Point Source Category (EPA Report No. EPA-821-R-97- 007, DCN L04197). 4.2 Detailed Monitoring Questionnaire The EPA requested industrial laundry (IL) facilities to submit wastewater monitoring data in the form of individual daily data points, henceforth referred to as DMQ data. These data were reviewed and, where sufficient requirements were met, included in the calculation of LTAs, VFs, and limitations. Because the EPA labs did not analyze these samples, data points in which inconsistencies existed among, the detection limit were excluded from calculations. , 4.3 Definition of Proposed Options During the site visit and field sampling phases of the rule development, and during follow-up to responses in the detailed questionnaire, three major technologies were identified for further evaluation for use in developing regulatory options. These major technologies are Chemical Emulsion Breaking (CEB), Dissolved Air Flotation (DAF), and Chemical Precipitation (CP). CEB is used primarily to remove oil and grease, as well as other related pollutants, from process wastewater streams. CEB is effective in treating wastewater streams having stable oil-in-water emulsions. CEB is also used for the treatment of heavy industrial wastewater, which consists of wastewater from the washing of heavily soiled items (e.g., shop towels) and wastewater from certain breaks (i.e., wash water, first rinse, etc.) in the washing cycles for other items that contain high 4-1 ; ------- concentrations of pollutants. The treatment consists of lowering the pH of the wastewater to break the emulsions, adding chemical flocculents, and skimming the surface of the water to remove the floating substances. Under this option, the heavy industrial wastewater is treated by CEB, combined with the untreated wastewater from the rest of the facility, and discharged. DAF is used to remove suspended solids, oil, and some dissolved pollutants from process wastewater. DAF treatment involves coagulating and agglomerating the solids and oil and grease and then floating the resulting flocculents to the surface using pressurized air injected into the unit. Then, the floating material is removed. Some DAF systems also have the means to remove material which settles to the bottom of the tank without shutting down for maintenance. CP is used to remove dissolved pollutants from process wastewater. Precipitation aids, such as lime, work by reacting with the ions (e.g., metals) and some anions to convert them into an insoluble form (e.g., metal hydroxides). The pH of the wastewater also affects how much pollutant mass is precipitated, as pollutants precipitate more efficiently at different pH ranges. Coagulation and flocculation aids are usually added to facilitate the formation of large agglomerated particles that settle more readily and can be removed from the bottom of the clarifiers. Along with these major technologies, regulatory options were developed utilizing stream splitting, a common practice at some facilities. Stream splitting provides a means of treating a portion of the total wastewater generated at industrial laundries. Stream splitting may be used to isolate and treat a stream with a higher pollutant load, while a stream with a lower load is either recycled and reused or discharged to the Publicly-Owned Treatment Works (POTW) without treatment. A divided trench and sump system is used to split process wastewater streams. EPA evaluated these technology options (along with splitting the streams for these options) and proposed PSES and PSNS pretreatment standards based on the CP technology option. The proposed technology options are discussed in greater detail in Chapter 9 of the Technical Development Document for Proposed Pretreatment Standards for Existing and New Sources for the Industrial Laundries Point Source Category (EPA Report No. EPA-821-R-97-007, DCN L04197) along with the justification for selection of this option. Pretreatment standards based on DAF are included here since the EPA is soliciting comment on a combined option where facilities currently using DAF would receive limitations developed using DAF data. All other facilities would receive limitations based on CP. EPA proposed pretreatment standards for the following pollutants: • Non Conventional - Silica Gel Treated - Hexane Extractable Material (SGT-HEM) • Metals - Copper, Lead, Zinc • Organics - Bis(2-Ethylhexyl) Phthalate, Ethylbenzene, Naphthalene, Tetrachloroethene, Toluene, M-Xylene, O&P-Xylene 4-2 ------- CHAPTERS DESCRIPTION OF DATA CONVENTIONS This section discusses the types of data in the IL analytical database and the hierarchy and procedures for aggregating multiple sampling observations within a sampling day. 5.1 Data Review ; The EPA wastewater sampling data hi the analytical database were thoroughly reviewed and validated by the EPA's Sample Control Center (further discussions of this data are at times referred to as the "SCC" data for this reason). During this review, the integrity of each sample was assessed to ensure that all specifications of the sampling protocol were met. The reviewers determined that some samples should be excluded from the analyses. Samples with flags of "EXCLUDE" or "DETECTED," which indicate a value was detected but the concentration value was not recorded, were excluded from the analyses. '. Also during the data review, several samples were qualified with a, greater than (>) sign, indicating the reported concentration value is considered a lower limit of the actual value. This is because the reported concentration was outside the range of the analytical method. When possible, these samples are diluted and reanalyzed. Otherwise these samples were handle4 as right-censored samples and excluded from all calculations. ! An engineering review of the database was also conducted and a few additional data values were excluded from the analyses for the reasons summarized in Chapter 9 of the Technical Development Document for Proposed Pretreatment Standards for Existing and New Sources for the Industrial Laundries Point Source Category (EPA Report No. EPA-821-R-97-007). One reason for such an exclusion would be if a pollutant was not detected in sufficient concentrations to evaluate treatment effectiveness. : 5.2 Data Types The IL analytical database (from the SCC and DMQ data) contains the following three different types of samples delineated by certain qualifiers in the database: : • Non-censored (NC): a measured value, i.e., a sample measured above the level at which the detection decision was made. • Non-detect (ND): samples for which analytical measurement ;did not yield a concentration above the sample-specific detection limit. • Right-censored (RC): samples qualified with a greater than (>) sign, signifying that the reported value is considered a lower limit of the actual concentration. All RC values were excluded from the analyses because these values could not be quantified with certainty. 5-1 ------- 5.3 Data Aggregation Data aggregation for the IL analytical data was performed at two levels. This section discusses the different levels and approaches for data aggregation, including multiple grab samples (one or more samples collected for a particular sampling point over time, assigned different sample numbers, and not physically composited) and field duplicates (one or more samples collected for a particular sampling point at approximately the same time, assigned different sample numbers, and flagged as duplicates for a single episode number). 5.3.1 Data Aggregation Across Multiple Grab Samples The first type of data aggregation performed was for multiple grab samples. Within the SCC database, SGT-HEM was reported as concentrations of multiple grab samples taken during one-day sampling periods. Since long-term averages (LTAs) and limitations were based on daily concentrations, multiple observations on a single day at the same sample point were averaged. When all of the samples in a set were NC, i.e., detected samples, the arithmetic average of the samples was straightforward. However, when one or more of the samples were censored, or ND, multiple grab samples were aggregated within each sampling day/sample point combination using the methods identified in Table 5-1. Table 5-1 Method for Averaging Multiple Grab Samples If observations are: AUNG A11ND NCandND 1. Max. NC > Max. Detection Limit 2. Max. NC s Max. Detection Limit J Lalbefof "average" NC ND NC ND , Value of "average** fet ' SNQ/n Maximum Detection Limit (SNC; +SND;)/n Max. Detection Limit n=number of grab samples per day. 5.3.2 Aggregation of Field Duplicates Another type of data aggregation for the IL SCC data was performed due to the identification of field duplicates in the database. The field duplicates are defined as one or more samples collected for a particular sampling point at approximately the same time, assigned different sample numbers, and flagged as duplicates for a single episode number/sampling point. Duplicates were collected for purposes of quality assurance/quality control. Table 5-2 presents the methods used to aggregate duplicates. Note that within the DMQ data no field duplicates were labeled, but for a few sample days, two concentrations were reported. Since there were only two concentrations reported within sample day, the aggregation method would be the same regardless of whether they were treated as grab samples or duplicate samples. Thus, these concentrations were classified as duplicate samples and were aggregated according to the methods outlined hi Table 5.2. 5-2 ------- Listings of summary statistics following aggregation of grabs and field duplicates are presented in Appendix B.I for all regulated pollutants in DAF (Option 3A), and in Appendix B.2 for all regulated pollutants in CP (Option 3B). Table 5-2 Method for Averaging Field Duplicate Samples * ff. *!»**,<• **,{*• „.*->, 1 I , , •> f *• * «*• V, J I' ' * ' If observations are: * * ^ * '- - BothNC BothND NCandND 1. NC > Detection Limit 2. NC <; Detection Limit ^ " ^*> ? ^ s^ -s * ^ Label of ^ . ^average" NC ND NC ND •v-~~ * ;r " .-.V " - ^i^ - ' ~ > * ^j-1--"^ ^ ^ _ ^" Value of^ayerage^ist ^ ^^. SNQ/2 Maximum Detection Limit (NC + ND)/2 Detection Limit NC = non-censored values ; ND = non-detected values If a sample had both multiple grabs and field duplicates, the multiple grabs were aggregated first. 5-3 ------- ------- CHAFFER 6 STATISTICAL METHODOLOGY - MODIFIED DELTA-LOGNORMAL MODEL 6.1 Basic Overview of Delta-lognormal Distribution The lognormal distribution is often appropriate for modeling effluent data. However, the presence of ND and very low concentration measurements in the IL effluent data led to the consideration of a modification to the lognormal distribution in modeling such data for several reasons. First, the lognormal model assumes that all concentration values are positively-valued. Second, the actual values of NDs are not known, though each ND has a concentration somewhere between zero and the reported detection limit. In this sense, ND measurements represent, in statistical terms, what are known as censored samples. '' In general, censored samples are measurements for which the exact value is not known but are bounded either by an upper or lower numerical limit. Non-detects qualify in this framework as left-censored samples, which have an upper bound at the detection limit and a lower bound at zero. To model NDs as left-censored samples under a strictly lognormal density model, it is necessary to assume that the exact (but unknown) values of these measurements follow the same lognormal distributional pattern as the rest of the detected measurements and that they are positively-valued (i.e., greater than zero). Therefore, two reasonably simple modifications to the lognormal density model have been used by the . EPA for several years. The first modification is known as the classical delta-lognormal model (Figure 6-1), first used in economic analysis to model income and revenue patterns (see Atchison and Brown, 1955). In this adaptation of the simple lognormal density, the model is expanded to include zero amounts. To do this, all positive (dollar) amounts are grouped together and fit to a lognormal density. Then all zero amounts are segregated into another group of measurements representing a discrete distributional "spike" at zero. The resulting mixed distribution, combining a continuous density portion with a discrete-valued spike, is known as the delta-lognormal distribution. The delta in the name refers to the percentage of the overall distribution contained in the spike at zero, that is, the percentage of zero amounts. : Figure 6-1 ; Delta-lognormal Model Non-Detects Detects 6-1 ------- Researchers at the EPA (see Kahn and Rubin, 1989) further adapted the classical delta-lognormal model ("adapted model") to account for ND measurements in the same fashion that zero measurements were handled in the original delta-lognormal. Instead of zero amounts and non-zero (positive) amounts, the data consisted of NDs and detects. Rather than assuming that NDs represented a spike of zero concentrations, these samples were allowed to have a single positive value, usually equal to the minimum level of the analytical method (Figure 6-2). Since each ND was assigned the same positive value, the distributional spike in this adapted model was located not at zero, but at the minimum level. This adaptation is appropriate since it is known that the NDs are some value greater than zero. This adapted model was used in developing limitations for the Organic Chemicals, Plastics, and Synthetic Fibers (OCPSF) and pesticides manufacturing rulemaking. Figure 6-2 Adapted Delta-lognormal Model Detects Non-Detects 0 S101S20 In the adapted delta-lognormal model, the delta again referred to those measurements contained in the discrete spike, this time representing the proportion of ND values observed within the data set. By using this approach, computation of estimates for the population mean and variance could be done easily by hand, and NDs were not assumed to follow the same distributional pattern as the detected measurements. The adapted delta-lognormal model can be expressed mathematically as follows: Pr (Uzu) = (1-6) 3> [(log(M) - ^)/a] if o< M < D 6 + (1 -8)$ [(logCD) - (i)/qj if u = D 6 + (1-6) $[(log(w) - n)/o] if u> D (6.1) where 6 represents the true proportion of NDs (or the probability that any randomly drawn measurement will be a ND), D equals the minimum level value of the discrete spike assigned to all NDs, <£(•) represents the standard normal cumulative distribution function, and p and o are the parameters of the lognormal density portion of the model. This model assumes that all non-detected values have a single detection limit D. It is also possible to represent the adapted delta-lognormal model in another mathematical form, one in which it is particularly easy to derive formulas for the expected value (i.e., LTA) and variance of the model. In this case, a random variable distributed according to the adapted delta-lognormal distribution can be represented as the stochastic combination of three other independent random variables. The first of these variables is an indicator variable, !„, equal to one when the measurement u is a ND and 6-2 ------- equal to zero when u is a detected value. The second variable, XD, represents the value of a ND measurement (discrete). In the adapted delta-lognormal, this variable is always a constant equal to the concentration value assigned to each ND (i.e., equal to D in the adapted delta-lognormal model). In general, however, XD need not be a constant, as will be seen below in the modified delta-lognormal model. The final random variable, Xc, represents the value of a detected measurement, and is distributed according to a lognormal distribution (continuous) with parameters /* and o. Using this formulation, a random variable from the adapted delta-lognormal model can be written as: U = IaXD +(1-/UKC ; (6.2) and the expected value of U is then derived by substituting the expected value of each quantity in the right-hand side of the equation. Because the variables !„, XD, and Xc are mutually independent, this leads to the expression E(U) = (l-6)exp(|i + 0.5 o2) (6.3) where again 8 is the probability that any random measurement will be ND and the exponentiated expression is the familiar mean of a lognormal distribution. In a similar fashion, the variance of the adapted delta-lognormal model can be established by squaring the expression for U above, taking expectations, and subtracting the square of E(U) to get: Var(U) = -&)Var(Xc) (6.4) Since, in the adapted delta-lognormal formulation, XD is a constant, this expression can be reduced to the following: \ Var(U) = (l-8)exp(2|i+CF2)[exp(a2)-(l-8)] + 8(1-8)£>[D -2exp(|i +O.S02)]. (6:5) In order to estimate the adapted delta-lognormal mean and variance from a set of observed sample measurements, it is necessary to derive sample estimates for the parameters 8, /*, and o. 6 is typically estimated by the observed proportion of NDs in the data set. \i and a are estimated using the log values of the detected samples where ^ is estimated using the arithmetic mean of the log detected measurements and a is estimated using the standard deviation of these same log values; NDs are not included in the calculations. Once the parameter estimates are obtained, they are used in the formulas above to derive the estimated adapted delta-lognormal mean and viariance. To calculate effluent limitations and/or standards, it is also necessary to estimate upper percentiles from the underlying data model. Using the delta-lognormal formulation above in equation (6.1), letting U0 represent the lOO*^ percentile of random variable U, and adopting the standard notation of zs for the 8th percentile of the standard normal distribution, an arbitrary delta-lognormal percentile can be expressed as the following: U« = exp(n +0" D exp(n+a i if if if 8+(l-S)((log(D)-n)/0) (6.6) 6-3 ------- The daily maximum limitations are established on the basis of an estimated upper 99th percentile from the underlying data model, so that 0.99 would be substituted for a in the above expression. To derive the daily VF for the 99th percentile based on the adapted delta-lognormal model, divide U-99 in the expression above by the previous formula for the LTA, namely U-99/E(U). 6.2 Motivations for Modifications to the Adapted Delta-Lognormal Model While the adapted delta-lognormal model has been used successfully for years by the EPA in a variety of settings, the model makes two key assumptions about the observed data that are not fully satisfied within the IL analytical database. First, the discrete spike portion of the adapted delta-lognormal model is a fixed, single-valued probability mass associated (typically) with all ND measurements. If all ND samples in the IL database had roughly the same reported detection limit, this assumption would be adequately satisfied. However, the detection limits reported are sample specific and, therefore, varied as a result of factors such as dilution. Because of this variation in detection limits, a single-valued discrete spike could not adequately represent the set of ND measurements observed in the IL database and a modification to the model was considered. In addition, the adapted delta-lognormal model sets all NC values below the detection to the minimum level of the analytical method. For example, if the minimum level for Toluene was .10 mg/1, then any NC samples reported below .10 mg/1 were set to .10 mg/1. There were a few instances in the IL analytical studies where a NC value was reported below the minimum level of the analytical method. 6.2.1 Modification of the Discrete Spike To appropriately modify the adapted delta-lognormal model for the observed IL database, a modification was made to the discrete, single-valued spike representing ND measurements. Because ND samples have varying detection limits, the spike of the delta-lognormal model has been replaced by a discrete distribution made up of multiple spikes. Each spike in this modification is associated with a distinct detection limit observed in the IL database. Thus, instead of assigning all NDs to a single, fixed value, as hi the adapted model, NDs can be associated with multiple values depending on how the detection limits vary (Figure 6-3). Figure 6-3 Modified Adapted Delta-lognormal Model Detects Non-Detects \ 0 5101520 6-4 ------- . In particular, because the detection limit associated with a ND sample is considered to be an upper bound on the true value, which could range conceivably from zero up to the detection limit, the modified delta-lognormal model used here assigns each ND sample to its reported detection limit. Once each ND has been associated with its reported detection limit, the discrete "delta" portion of the modified model is estimated in a way similar to the adapted delta-lognormal distribution, where multiple spikes are constructed and linked to the distinct detection Jimits observed in the data set. In the adapted model, the parameter 6 is estimated by computing the proportion of NDs. In the modified model, 8 again represents the proportion of NDs, but is divided into the sum of smaller fractions, 6;, each representing the proportion of NDs associated with a particular and distinct detection limit. This can be written as: ; v-\ * (6.7) If Dj equals the value of the i* smallest distinct detection limit in the data set, and the random variable X represents a randomly chosen ND sample, then the discrete distribution portion of the modified delta-lognormal model can be mathematically expressed as: (6.8) The mean and variance of this discrete distribution can be calculated using the following formulas: and Var(XD) = -?-£ £ i6/6/^/ " D¥- (6.9) It is important to recognize that, while replacing the single discrete spike in the adapted delta-lognormal distribution with a more general discrete distribution of multiple spikes increases the complexity of the model, the discrete portion with multiple spikes plays a role in limitations and standards development identically parallel to the single spike case and offers flexibility for handling multiple observed detection limits. 6-5 ------- ------- CHAPTER 7 ESTIMATION UNDER THE MODIFIED DELTA-LOGNORMAL MODEL Once the modifications to the adapted delta-lognormal distribution!are made, it is possible to fit a wide variety of observed effluent data sets to the modified model. Multiple detection limits for NDs can be handled. The same basic framework can be used even if there are no ND values or censored data. I Combining the discrete portion of the model with the continuous portion, the cumulative probability distribution of the modified delta-lognormal model can be expressed as follows, where Dn denotes the largest distinct detection limit observed among the NDs, and the first summation is taken over all those values, Dj, that are less than u. : [(log(w)-u)/a)] if u------- 7.1 Facility-Specific Estimates 7.1.1 Estimation of Facility-Specific LTAs For the purposes of estimating facility-specific LTAs (equal to the expected value in the equation (7.3)), the EPA chose to divide the IL data sets into two groups based on their size (number of samples) and the type of samples in the subset because the computations differ for each group. The groups were defined as follows: Group 1: Less than 2 NC samples or less than 4 total samples. Group 2: Two or more NC samples or 4 or more total samples. For Group 1, the LTAs were calculated as the arithmetic average of the samples, since the sample sizes for either the discrete portion or the continuous lognormal portion of the data were too small to allow distributional assumptions to be made. Specifically, Group 1 contained all data subsets with all NDs or only one detect. Sample-specific detection limits were substituted as the values associated with non- detectable samples. For Group 2, the LTAs were calculated using the procedures outlined in the preceding section using equation (7.3) and the Maximum Likelihood Estimates (MLEs) for \L and a. 7.1.2 Estimation of Facility-Specific VFs After determining estimated LTA values for each pollutant, facility, and option combination, the EPA developed 1-day variability factors (VF1) and/or 4-day variability factors (VF4) depending on the proposed frequency of monitoring, as outlined in Table 7-4.. Table 7-1 EPA Proposed Monitoring Frequencies ' ' - "- ^ -^- ,, ?,yX Pollutant Category Metals, Organics Classicals I ' „ "" - " I If *!" Frequency of Monitoring , • Monthly (VF1) Weekly (VF1, VF4) Similar to the calculations for the LTAs, the data were divided into the same two computation groups based on the number and type of samples in each data subset for purposes of estimating variability factor. These computation groups are defined as follows: Group 1: Less than 2 NC samples or less than 4 total samples. Upper percentiles and VFs could not be computed using the modified delta-lognormal methodology. Group 2: Two or more NC samples and 4 or more total samples. The estimates of the parameters for the modified delta-lognormal distribution of the data were calculated using maximum likelihood estimation in the log-domain. Upper percentiles and VFs were calculated using these estimated parameters. 7-2 ------- Several data subsets belong in Group 1, and therefore have missing 99th percentiles and VFs. 7.1.2.1 Estimation of Facility-Specific VF1 The VF1 are a function of the LTA, E(U), and the 99th percentile. An iterative approach was used in finding the 99th percentile of each data subset using the modified delta-lognormal methodology by first defining D0=0, 50=0, and Dk+1 = «> as boundary conditions, where D; equals the i* smallest detection limit, and 8; is the associated proportion of NDs at the r* detection limit. A cumulative distribution function, p, for each data subset was computed as a step function ranging from 0 to 1. The general form, for a given value c, is = E V. 0.99, was determined and labeled as PJ. If no such m existed, steps 3 and 4 were skipped and step 5 was computed instead. i 3. Computed p* = pj - 6j. . 4. If p* < 0.99, thenPj, = Dj? else if p* .>_ 0.99, then P99=exp (1-6) (7.6) 5. If no suchmexists, such thatpmj>. 0.99 (m=l,...k), then ! 0.99-6 (1-6) (7.7) The daily variability factor, VF1, was then calculated as P99 VF1 = E(U) (7.8) 7-3 ------- 7.1.2.2 Estimation of Facility-Specific VF4 Since the EPA is assuming for costing purposes that the Classical Pollutant, SGT-HEM, will be monitored weekly (approximately 4 times a month), the EPA calculated a VF for monthly averages based on the distribution of 4-day averages. In order to calculate the VF4, the assumption was made that the approximating distribution of U4, the sample mean for a random sample of 4 independent concentration values, is also derived from this modified delta-lognormal distribution, with the same mean as the distribution of the concentration values. The mean of this distribution of 4-day averages is (7.9) where (X^ denotes the mean of the discrete portion of the distribution of the average of four independent concentration values (i.e., when all observations are not detected), and (X4)c denotes the mean of the continuous lognormal portion of the distribution. First, it is assumed that the probability of detection (8) on each of the four days is independent of that on the other days, since these samples are not taken on consecutive days and are therefore not correlated such that 84 = S4. Also, since ~4)0 = E(XD) then (7.10) and since E(04) = E(U), then V-4 = log 7=1 d-64) -0.502,. (7.11) The expression for o?4 was derived from the following relationship: (7.12) Since and (7.13) then 7-4 ------- = 54 64(1 - (7.14) This further simplifies to 482 •64(l-64) 7=1 and furthermore, exp(024)-l = Then, from (7.10) above, exPGi4+0.5o24)=- (l-84)exp(2u4 k j=l (1-84) and letting -64) then, exp^+O.So2,) = -, since E(U.) =E(U) Furthermore, 1 + • *££w,-* 4 4 (1 (1 92 82(1 84)(^6Z> 6T» I' [w ' '' (1-84)J -8V -54)2 (7.15) *EEa,»/J>,-^ 4 4 S2n frh u V.* u J * I2 ^ A exp(u4+ . 04) (7.16) (7.17) (7.18) (7.19) 7-5 ------- Since Var(04) = Var(U)/4, then, by rearranging terras, 024 = log 4T12 4T12 /=! (7.20) Thus, estimates of /*4 and 04 were derived by using estimates of S^..^ (sample proportion of NDs at observed detection limits D^.-.Dk), p. (MLE of logged values), and o2 (MLE logvariance with sample bias adjustment) in the equations above. In finding the estimated 95th percentile of the average of four observations (four NDs, not all at the same detection limit), an average can be generated that is not necessarily equal to Dj, D2,..., or Dk. Consequently, more than k discrete points exist in the distribution of the 4-day averages. For example, the average of four NDs at k=2 detection limits are at the following discrete points with the associated probabilities: 1 2 3 4 5 (3£V (2Z>,+22>2)/4 V In general, when all four observations are not detected, and when k detection limits exist, the multinomial distribution can be used to determine associated probabilities, that is, Pr ^4 = 1=1 4! (7.21) The number of possible discrete points, k*, for k= 1 ,2,3 ,4, and 5 are given below: 1 2 3 4 5 1 5 15 35 70 7-6 ------- To find the estimated 95th percentile of the distribution of the average of four observations, the same basic steps (described in Section 7.1.2.1) as used for the 99th percentile of the distribution of daily observations were followed with the following changes: \ 1. Change Pj, to P9S, and 0.99 to 0.95. 2. Change Dm to Dm*, the weighted averages of the detection limits. 3. Change 8; to 8;*. 4. Change k to k*, the number of possible discrete points based on k detection limits. 5. Change the estimates of 8, p, and 0 to estimates of 84, jtt4, and a4, respectively. Then, the estimate of the 95th percentile 4-day mean VF is: P95 VF4 = since E(U) j) = E(U). (7.22) Appendices C.I and C.2 display LTAs, VF1, and VF4 by analyte and facility for DAF and CP, respectively. 7.2 Pollutant-Specific Estimates j 7.2.1 Estimation of Pollutant-Specific LTAs After estimating the facility-specific LTA for each pollutant and option, as described in section 7.1.1, pollutant-specific LTAs were calculated. Within each option, the pollutant specific LTAs were calculated as the median of the facility-specific LTAs for that pollutant. 7.2.2 Estimation of Pollutant-Specific VFs \ 7.2.2.1 Estimation of Pollutant-Specific VF1 After the facility-specific VF1 were estimated for each pollutant and option, as described in section 7.1.2.1, the pollutant-specific VF1 was calculated. The pollutant-specific daily VF was the median of the facility-specific daily VFs for that pollutant in the option. 7.2.2.2 Estimation of Pollutant-Specific VF4 After the facility-specific VF4 were estimated for each pollutant and option, as described in section 7.1.2.2, the pollutant-specific VF4 was calculated. The pollutant-specific VF4 was the median of the facility-specific VF4 for that pollutant in the option. 7-7 ------- ------- CHAPTERS DERIVATION OF THE PROPOSED STANDARDS The proposed daily maximum limitations for each pollutant were calculated as the product of the pollutant-specific LTA and the pollutant-specific daily VF. Similarly, the proposed 4-day limitation for SGT-HEM was calculated as the product of the pollutant-specific LTA and the pollutant-specific VF4. Limitations for CP are presented in Table 8-1. Table 8-1 Daily and 4-day Limitations for CP Option 3B Pollutant , , ' ^ ^ ?• i- t '- *--^ >. , _ ' '*.>/> „ 1 =• ' *•?,< f t a-- ..•. •> „-„ ^ ^, >t \, rf SGT-HEM Copper Lead Zinc O+P Xylene M-Xylene Ethylbenzene Naphthalene Bis (2-Ethylhexyl) Phthalate Tetrachloroethane Toluene Daily Limit ; ,(mg/L) ^ ; "" v^ 27.5 .24 .27 .61 .95 1.33 1.64 .23 .13 1.71 2.76 Monthly "A-Terage ^ (nfgflb)* ^ 12.7 NA NA NA NA NA NA NA NA NA NA *Based on an assumption for costing purposes of 4 days of sampling per month. Appendices D.I and D.2 present pollutant-specific LTAs, VF1 and VF4, and daily and 4-day limitations for DAF and CP, respectively. i 8-1 ------- ------- CHAPTER 9 ! RAW WASTEWATER CONCENTRATION COMPARISONS 9.1 Comparison of Industrial Laundry Influent and Linen Influent Statistical analyses were conducted to assess whether the mean influent concentrations of 98 pollutants of concern differed significantly by type of laundry facility. The EPA wanted to compare linen influent concentrations to industrial laundry influent concentrations. Unfortunately, none of the raw wastewater data available was for 100% linen items. Therefore, the EPA used facilities that were mostly linen (i.e. between 60% and 99% linen). The data from these facilities will be referred to as linen wastewater and the wastewater from facilities doing mostly industrial laundry items will be referred to as industrial laundry (IL). Appendix E.I lists the facility, sample point, and data source information used in this analysis. It was observed that not all of the 98 pollutants of concern had both IL and linen information. Also for several pollutants, only one linen facility was reported. Furthermore, only those facilities which reported at least 3 concentrations were included in the analyses. The following pollutants had sufficient data for analyses (at least 2 linen and IL facilities with at least 3 reported concentrations): BOD, Cadmium, COD, Chromium, Copper, Lead, Nickel, Silver, Total Recoverable Oil and Grease, Total Suspended Solids, Zinc, and pH. For each of the pollutants with sufficient data, a comparison was made between wastewater concentrations reported by facilities within IL and linen supply, respectively. Differences in pollutant wastewater concentrations within IL and linen facilities were observed for some pollutants. These pollutants were not considered further in the analysis to determine if IL wastewater concentrations differ significantly from linen wastewater concentrations. Thus, the list of pollutants for which comparisons would be made was reduced from 14 to 8. Table 9-1 displays results from the analysis of variance (ANOVA) which was used to compare the mean log concentration (log(conc)) between the linen facilities. Results from this analysis indicated that the mean log(conc) between linen facilities differed significantly for pollutants BOD, COD, Lead, Silver, and Nickel at a=0.01. • 9-1 ------- Table 9-1 ANOVA Results for Linen Comparisons Analyte . " ".^$, ^ 4» " ~" BOD COD TPH (as SGT-HEM) Total Recoverable Oil and Grease Total Suspended Solids pH Cadmium Chromium Copper Iron Lead Nickel Silver Zinc K *^ Number of J Facilities 3 2 2 3 3 4 4 4 4 2 4 4 3 4 Number,of ; Concentration Values 9 7 5 8 9 10 15 15 15 5 15 15 13 17 p-value "" ^ 0.0013 0.0068 0.2732 0.0202 0.1345 0.0713 0.2733 0.1567 0.4553 0.1776 0.0054 0.0017 0.0041 0.1070 Signn1<-apt;at >f" > f -a=0.dl f'X Yes Yes No No No No No No No No Yes Yes Yes No Table 9-2 displays results from the ANOVA analysis which was used to compare the mean log(conc) between the IL facilities. Results from this analysis indicated that the mean log(conc) between IL facilities differed significantly for pollutants pH, Nickel, and Silver at a=0.01. 9-2 ------- Table 9-2 ANOVA Results for EL Comparisons ABaayte^ • - *- *"*'* » * > ' ' ' * •»*••« 4 ,, M. '•* " 5 .« -.' > '"*'' , s ' - . ,. ~ V*' **?,-* .Jk>- ,! _ H „ _ >• „•>• BOD COD TPH (as SGT-HEM) Total Recoverable Oil and Grease Total Suspended Solids pH Cadmium Chromium Copper Iron Lead Nickel Silver Zinc f. V". ifcJSj«- C j^ rNumfter-'af . ,-• _ «_»!» W1 ' - ^Facilities • *. -^S^,1" u, » ^ , C' 6 6 5 2 6 6 6 6 6 6 6 6 6 6 rf 's i. ?* — Nppberof ; Coiicentration •^ ^ 'i )• M "Values •# «t 33 34 30 8 34 33 34 34 34 34 34 34 34 34 p-fahie" .' !*', •- . ^ *^ i \ i0.0252 b.1084 0.3625 0.4317 0.1543 0.0002 0.5284 0.0364 0.1385 0.1971 0.1945 0.0065 0.0001 0.7447 ** «* h v -jp- i Significant " '» at -«=OJi« j, j, k e N> Vf^fei, ^ ^ & "'-•^^ £ * •> ^ No No No No No Yes No No No No No Yes Yes No Pollutants in which the significance level exceeded a=0.01 for either linen facilities or IL facilities were excluded from further analysis. Thus, comparisons of wastewater concentrations between IL facilities and linen facilities were conducted for the following pollutants: Total Petroleum Hydrocarbon (TPH), Total Suspended Solids, Total Recoverable Oil and Grease, Cadmium, Chromium, Copper, Iron, and Zinc. Table 9-3 displays results from the t-test analysis which was used to compare the mean log(conc) of the linen daily wastewater concentrations to the mean log(conc) of the IL daily wastewater concentrations. Results from this analysis indicated that the mean log(conc) of linen wastewater differed significantly from the mean log(conc) of IL wastewater for pollutants Total Petroleum Hydrocarbon, Oil and Grease, Total Suspended Solids, Cadmium, Chromium, Copper, Iron, and Zinc at a=0.01. 9-3 ------- Table 9-3 Comparison of Mean Pollutant Log Concentrations in Linen Facilities vs. EL Facilities Analyte TPH (as SGT-HEM) Total Recoverable Oil and Grease Total Suspended Solids Cadmium Chromium Copper Iron Zinc TvpeofFacSiiy~ j »»* k jt "" „' s _ Industrial Linen Industrial Linen Industrial Linen Industrial Linen Industrial Linen Industrial • Linen Industrial Linen Industrial Linen Sample Size 30 5 8 8 34 9 34 15 34 15 . 34 15 34 5 34 17 Mean' ^ log(conc) 6.05 2.64 7.18 4.56 7.10 5.08 -2.66 -4.33 .-1.47 -3.19 0.85 -1.54 3.23 1.00 1.47 1.15 Mean Cone 425 14 1310 96 1206 161 .070 .013 .230 .041 2.32 0.21 25.2 2.71 4.16 0.32 p-value ' , f , ' " 0.0001 0.0012 < 0.000 0.0001 < 0.000 •< o.ooo < 0.000 < 0.000 SignifiKwt atla^O.ftL Yes Yes Yes Yes Yes Yes Yes Yes Table 9-3 illustrates that for each of the analytes listed, IL wastewater concentrations are significantly different from linen wastewater concentrations. Also, note that the IL mean wastewater concentration is consistently higher than the linen mean wastewater concentration. Although the linen facilities used were not 100% linen, EPA assumes that these results would hold if the proportion of linen items at these facilities were even greater. 9.2 Comparison of Linen Influent to Denim Pre-Wash Influent Statistical comparisons were conducted between denim prewash wastewater and linen wastewater to determine if pollutant concentrations in untreated denim prewash wastewater were similar to the pollutant concentrations in linen wastewater. Appendix E.2 lists the facility, sample point, and data source information used in this analysis. Prior to comparing pollutant concentrations from denim facilities to pollutant concentrations from linen facilities, it was first determined whether the pollutant 9-4 ------- concentrations differed significantly across facilities among the linen facilities. For each pollutant, ANOVA was used to compare the mean log(conc) in each facility that reported two or more concentrations among untreated linen facilities. : Results indicated that the mean log(conc) for each linen facility differed significantly at the a=0.01 significance level for pollutants BOD, COD, Nickel, and Lead. Thus, the only pollutants that were used in further analyses included: Cadmium, Chromium, Copper, Iron, Oil and Grease, Total Suspended Solids, and Zinc. Note that the concentrations reported for Total Petroleum Hydrocarbon did not differ significantly among linen facilities, but the denim prewash facility did not report wastewater concentrations for this pollutant. Influent concentrations were available for only one denim prewash facility. Because of this, EPA was unable to compare concentrations between denim prewash facilities to determine if there were significant differences between influent concentrations for denim prewash facilities. Therefore, the following results from the t-test analysis represent the comparison between linen facilities and the sampled denim facility. Table 9-4 displays results from the t-test analysis which was used to compare the mean log(conc) from the untreated linen facilities to the mean log(conc) from the untreated denim facility. Results indicated that the mean log(conc) from untreated linen wastewater differed significantly from the mean log(conc) from untreated denim wastewater for pollutants Cadmium, Chromium, and Copper (p < .01). There was no significant difference in the mean log(conc) from linen vs. denim wastewater for Oil and Grease, Total Suspended Solids, Iron, and Zinc. Table 9-4 Comparison of Mean Pollutant Log Concentrations in Linen Facilities vs. Untreated Denim Facilities ^ \*r Oil and Grease Total Suspended Solids Cadmium Chromium Copper Iron Zinc Type of lacfllfy^ : v >> i f ,t * ( *, „ - Linen Untreated Denim Linen Untreated Denim Linen Untreated Denim Linen Untreated Denim Linen Untreated Denim Linen Untreated Denim Linen Untreated Denim Sample ^Size 8 7 9 15 15 13 15 13 15 13 5 12 17 8 Mean * log(conc){. 4.56 : 2.96 5.08 6.15 ' -4.33 -5.68 : -3.19 : -4.47 : -1.54 ! -2.85 1.00 ! -0.69 -1.15 I -2.87 : "Mean/ cone ^ 95 19 161 470 0.013 0.003 0.04 0.01 0.21 0.06 2.71 0.50 0.32 0.06 p-value -v '«il* .018 .021 .0001 .0014 .001 .027 .114 Significant Jat ceMUH No No Yes Yes Yes No No 9-5 ------- Thus, it was observed that the pollutant log(conc) for the analytes Cadmium, Chromium, and Copper was significantly higher in untreated linen wastewater than in untreated denim prewash wastewater at a—0.01. Similar concentrations (0.01) were reported in untreated linen wastewater and untreated denim prewash wastewater for the analytes Oil and Grease, Total Suspended Solids, Iron, and Zinc. 9-6 ------- APPENDICES A.I and A.2 ------- ------- u c o u S§!o88 o o \o o o ooooo ooooooooooooo ooooo ooooo o u-> o o o r--ocoN.ooopjcQOococo «- o o o io t- co o o o o CM in o in CM «— \o po ^ vo o N. ^ O «~ CM o in o co ^ 10 in ^~ sf o ^ o *<~ N- in CM in ^o ^o r^ o CM <^ CM o* in CM CM T- i KJ r-o in CM *~ «~ CM »~ CM ^ O O- -O O f> CM «-«-«- o o o o «- «- o o o o o o o ooooo ooooo --, U U U U I u z z z z : ICJUOU CJCJUtJOUQOOOOOO OUUUU OOOOO :zzzz zzzzzzzzzz:zzz' zzzzz zzzzz LOJ : O MrofotOfOOfO *•" *O ! CMOCDOO»-O«—OO»-OO • ooooeaoooooo«-»- ; ... T- CM Q. ca «- CM KI sf in >->->->->• « ^ .£ 2 J-> C9 0) •*- «*- -I O N- S u § s u co o S Nf § 2 CO a> 0) 0) 4^ H- < Ill I U UI CA ! t; ra . (0 -^ ; u u CO 4-» ------- oooooo ooooooooooooooo o o o o o ooooooo -• •» ooooooooooooooo oooi _ _ _ I O O O CO LTt O* O O O O „. v _. -_ v- ~. — . - v., j o f^ OfOintnotcMocxjooco *o »o c\) o o *JT r^ o rxj ro «~ ^ »- O(O»-inm«o-*ooin»-ofM»- eoco»*»-«- focMc\jc\it'>o»- OOOOOOOO OOCM^OOOOOOv— OT— O^ OOO^^ OOOOOOO ^. OOCJCJOCJCJO UUUUUOUUUUUUUUU UUCJUU O O O O O O O ZZSZZZZZ ^ZZ^ZZIEZZZZZZZZ ZZZZZ ZZZZZZ %o«~ro. •»,-s, v^-«^ v^ x» to fo tn o to o* v- o o o o o o «- rtoNlin>o>o-d'>oec>c>c>r^rj >*>* ^nJro^FintnvJ t O O O O OOOOOO 7~ v-^ OOO^^ OOOOO O O & *J a o CA ~* 4^ ' C a *o camtamcacamco <<<<<<<<<<<<<<< <<<<< CD ca co CD CD ca CQ **- co a. w ^ c 5 T- U->r- WO a*i 'S. tn &» •*- «a «,£ *-» a 4-» _i w— 63 O ••- "-^ ill s s i°"B Q. U. m & O 4J O «C w 3: _i o g L. a> ro ro sf I o •M a: _>. ui II & < z u o: ui a. ct ui a. g ui fi. - 13 4-* U V) OT to IS 4-> 0) 0) ca 4-* ------- . . eo ro o> *- «- o CM «- O to -4- O O r— O O «* O -o o- r- « r- _ - - _..___ ,.r\jv-o*t^ in >o ** in ** oooooooo cM^OD^^-f^o^N-info _o <\j m ro *— *— *— c\j o c\j «— «— o ooooo ^-^-^-T—«-«—T—T— OCMC\J«—oooaroo oooooooo oooooooo oooooooo ooooooooo ooooooooo ooooooo ooooo ooooo oooooooo ooooooooo o o u u o u a z z z z z z z OOOOOOOCJCJ — — » ^ O* ^ O» O* ^ its ^sl^siS 5° ^ggg^^^ — X. D- 10 •r-<\jro-*in »-e\iro-*in mil iiiii c o —• 4^ £L_C -* -4- -4- -t -J- CO CO CO CO O3 (D O mmcacommcocQ <<<<<<<<< in c U. '5. HI lfm c 3 8 ^^ u E § tn -* O 5 °: O>J3 £Z v S§ 3 u z r- I I (0 O) 2 D> 3 : co : u o 4-* a ------- gooooo oooo oooooooooooo o o o o o oooo OOOOOOOCMOO**N- ro o o ro NO o NO NO CM o in ON ojj* CM sa* r1*- CM o* o* o CD ON co *~ co *~" co CD "si* ON *~~ ^r ON ^* in *^ o* ^~ c\i sf ^~ cs in ro T—S-i-cMocM 5-»j'CM»- ocM«*T-eONOr-srcMNom»* cM*-»*roro CMrocMrocM in oooooo oooo CM< • o o «— o o o o o o ooooo «- 1 OOOOOOOOOOOO OOOOO O CJ O O O NO ro ro «— NO . n. CD fa • CM ro-* in «- CM ro »* in ' ^C 4 ^ 4 4 ^ 4 ^C ^C ioOOO Q Q O O Q o < cu «<<< <<<£< cocommmcacococomoo CO c w ro U iC UJ ** «— >»• ro S 3 *-.s% < a o 111 m a _ •»- i- ** v> u •*- »*- _j o § I SJ II I -o VO § o sr I 11 I § u z N il co -• CJ 0. o Q. m a. a 4-> CO CJ s: ------- o o o o o o o ooooooooooooooo o in co N. NO -* o to ooooppooooooooo • o o o o *** NO t o o -4* in O O O O O O O OOMOOO^-r-OOr^OOO*- *— «- §000 in o o tniin in CM CM SN- o> m in o» -a- i o o «- CM o o o o o o o -., CO O O CO CO CO O CJ CO CO CO CO CJOCOCOCOCOOCOCOCO CJ O O CO CO O CO CO CO CO CO O U ZZZZZZZ ZZZZZZZZZZZZZZZ ZZZZZ ZZZZZ ZZ O) fo ro fo ro ^o FO ^o C O* O* O* O* O^ O* O1* — 2 tvji^o-tfcoo.'eo' dcj-*ec5^Ni^ Q.OJ CMCM»-CMOO«- OCMOOrOOOO»-«- »-»-tM(oiomNO«o>o«ocooo>«-CM -*,-a-o'o o" CO OOOOOi-r- OOOOOOOOOOOOO^-^- OOO«-»- I •M to — X Q. a 1° CO «- CM (O -* in «- CM §§§§§ §§ a g eommmoomm «<<<«<<<<<< 0) c_ D) NO MI NO s INN. V +•* M- < Ul Z UJ Ul LJ Cm •-> < z N U IM Q. s CL & C in 4J CU 3 -D Z8 S o o. s! —• L. O O) £_ O ------- ro *o CM ^- co *o N- *o oooininininininoooo o ro ro co CM r^r^xt-h-oooo o SI coo r^cM«-roc\i co 5 " o o o o o o o o o o o in in in CM o o «- o «- «- «- «- J •Jroe* rw S. o ?-in o o o o o o o o o «- r- «- «- t»- r^ vs o o oooooooo «-<\5cvi c\iro«»inro ooocaoocsoooocjeD '^'^''J^''^ °°°°°°°° T^OO OOv-OO OOOOOC3OOOC3C3C3CD C3OOC3C3 OOOOOOOO O O O CJ CJ I z z z z z : CJ U O 0 O 0 O Q O O O 0 O CJUOCJO OOOOOOOO £. •»- ro«*tn T-CM ro •* in fg 555 55555 3 o o o ooooo «- CM ro >* in 55555 o o o o a co co co co co ta c 0> "§ o ca «— Ul o vO • 4J <3 < o u 2 x"!! §"5 ° g. O TjJ """S™ S| w*3 tj'S § *i ^ C C3 3 E CM fe 1 ^ 55 S ^» s 5 O _J **s. i -» C3 -* o O CM o «- r- O* «C C. O Ul z Ul x a. s Ul ui Ul I Ul CD _1 I— 111 i 4-* (U O* 3 "8 ^ •S" z Q. co 52 co ro Q> C_ U tj-w "g CO >M O) CO I. < U O g en £ D> ------- § o N- CO o N. O T— CM O •.* O co o «— in «— T- «- O O O o» in co o in ro «— oj oo o m o* *o c? Is*" ^o t\i r*"- *~" >o o*o**oeo incvj-r-^a-fM T- -c— CM.«— T- OOOOO OOOOO OOOOO OOOOO OOOOO OOOOO o «- o o o o u o o u u o o o o 0 O O Q O .?o StScSSS (0 CO O> O> «- fM in o o CD «- «- a o co co m co m ro to to to to •§ e» O : s . <-. oj < to o I SI •M _J 'E S3 u s co g rvi S z c> 8 CM S CM R N. e c o £ en LU Ul CD C «t IU t CM U) ca UJ 3= _1 >- ^ I— LU I CM CO CO lit i CM CO m Ul o Ctf I— ai c ro D-D ±S o a. -o •o x <- o 4J O — x ro c —•0) (0 (0 4-> O) C CO t- <: o o ff o o> c. O ------- o o o o o o a o o o o ooooo ooooo «- o o o o o ooooooo I Mz-zzzzzzlzz zgzzM MMzzz zzllzz zizzzzi to 10 to to to to to to rp 50 to 50 K> O* Os O* O* O* O* O* O* w* O^ ^K o* u* ^^tOOs»-fM T- O CO »O •* f> N rviooocuo CM o o o o «- o T-tO->->->->- >->->->->£ O —' 4-* i I! CD CO CO CO CO CO CO %z & {5* **-§ T- O iS SJS? ss = 11 s JO 3 s s s £8 o» tn IU z Ul § g C «s : g QC s ^•g s •s0 •- a. o t-. u ** ° 'S >» O) C l—- m 03 ro -M TO ra CT) (0 D> ------- o c o u m CD cs c? c? cs *™ o *— o CM «- oo-^ooo ^-^-ro^oro N.N-OO*CO OOOOOO OOOOJO **^- N« CO O» O* ^^ CM o o o o *• «- «£<<<< I O 4-* o s s- tn cu r- CO 4J • -M CD < CO O I . ra ro o in -a- *nt- ta 2! C CO 4-» CD a u : S CO § ra co 0) • D) c —• o co CO -M O> C CO C- < CJ O c CO O) ------- §00 o o *- > o o o N- t in o in to ro to -a- r »•••• o o o o o o o S°°° gggggggggggg >*TOCMOCMN.cor^o.o — -— — — — — --— SSS8J g g £ £ O O O O oooooooooooooooo o o o o o cj CJ CJ CJ CJ CM CM O £ Rj O t RicMCMCMfVlNrarJ! * (3 CO a 5555 2 CO CO CO CO CO CO CO CO •g N. $. s CM «— 0 CM O .sB ^a« = i g s a o tu o_ o § s 0 N. S S CO o s vr s 7440508 743992 I a. a. o u —• o to £ co 4-1 ------- >, CO o ^» CO o c "— "g- ta c Q. '— U J= 4-* L. ••- 0 2 **- CM 4J 4-> • CO CO < O O X >»— •• S-?§~ Z CO O tit o 0- "O 0- H- C < o ra m c/> •>- CO 4-» £- tn (s ~"o c o 4-* CO ra CD ra ra E_ U 5 oooo oooooooooooooooo o *- o o ooooooooooo u co o ro o oooooooooooooooo «— «— o* «— rooroooooo%?*or*~ • c eot*»CM«- OOOOOOOOOOOOOCMCOO (O •» *- ro rom^«-«-«-«-.sX%XN^*X*^»^ — • 4*>* to ON ->->- ^> (0 OOOO 0) . — * 4-> n. c co co co co ro*o <<<< <<<<<<<<<<<<<<<< NrNTisTrsT cacacococococococQCQca co a. • o T3 ; O CM ^O N- CO w ^- co ro -- , N- 1 c CO 4-> 0) 3 T3 —• O CO o" - O L. 4-» O 01 >. O) -- —' OJ CO ------- o o o o ooooooo oooooooooooooooo oooo oooroo-a-o ocooooooooooo*- mintnr- c\j in ro o «— o «— _ ooo *— ro 10 ** N* «— o o ooo -T-o^cpincMcMinEjM^ •* N- »» "O o to in OOOO «-r-«-(M^-«-(M OOOOOOOOOOOOOOOO OOOO OOO U OUOCJ UCJUUOOU UUUUUCJUUOUUUUUUU O U U CJ O CJ O Z2SZ ZZZZZSEZ ZZZZZZZZZZZZZZZZ ZZZZ ZZZ SJSSS 888 1?^ ssso" 2S >. a a co o> f.~ c So — tn a. co co co co :««<< <<< 42° 5: -""S g o § -a- s (9 E o «» S S •a o NO vO s sr s §«£ ' UJ O Is! § 4-f 3 ' in as 4-* ------- O c co co eo co eo co co co co co co co «— N) -» in 1111 coco eo eo o> o in N- .2 g (8 s- * •* CM , O S eo XT 10 4-> ------- gfeSfe o o o o ooo intoN-cM o o CM o or^o-* ooo o in o -o -o «- -o o to eo o tn o •* o> >o uj o o o o o - > CO I . ro •» co in •-^ CM o «-^ SSS2 o o «- ooo o o o o -s. CJ O O O g zzzz , CJ O CJ O CJ O O CJ CJ O O CJ CJ CJ CJ O O O CJ CJ •z zzzz zzzz zzzz zzz zzzz ro o o o o «- T-f» >* in 5555 a o o o nil iiii CO 3 sr CO g o LJ Z Ul g 4-» 0) 3 "8 2° . D) C —* O c (D O> C_ O ------- APPENDICES B.I and B.2 ------- ------- x cu o CO 3 Z O O «- cj -o «- o 3 z o CM o C O CJ — 3 Z •o > cj 4-> 0) Z CO O CM >O sf o o e\i c o i xcucj o co 04 o o >o co £) z CM in CM co ro eo "eo «-^ o o o CM o C O CJ r- rO -* f- O O •i- 3 Z «-*-*-«- O «- ^""ctj OC30000 >cj ^ >o -*-»«- eo o z ro «- o CM o ro TJ > O 4J tt) Z to o c a> Q «- — 3 z o c en o o •»- 3 Z «— CO •a > o c\j •M CU Z T— CO a "O 111 (1) CD t- 4-» TJ < eo 3 z eo o -J o —• • D) 5. I— « CD UI 3= fi cj ro in ro o T— co = ^| * SCJ O •* r— z >o ro «- o o o o o c> ii 4^ 'c CO CJ ^-1 _ 3 z o o ro CD O ~* O M- >- z O x coco ro -J- o UlC04-> ui J3C03 O «— CM ~ " ^~ CO ^3 C9 O o; UJ a. a. o C ••- < a> "S Sro in tn t- o «M in ro «^ ro in •* "ro o o o o o o > >- = UI ui en c V 4-> >- E 4-> CO CM CO v^ >. " t- m E cu E 4-* 3 >. CO —• _c 4J CO a> c to sj- «- T3 eo o «— ^i- o cu • • • ME O O O •8 o vo o in w o «- to ••- in ^ in Q. sf -a- r- ui u u c? CJ CJ E CO CO O CO"O> - .a -M o ro«-oc\i«oro ocoo ...••• o o o o o o o c >o >tf" r^ co CM r*- Tjeo inro»-ro^>* a. UI T- o o o o o ro in in co tn in I CS C3 CS CO C3 o «— >O o v- in • co ro o «- o ro • ••- ro in so ^ in i o co o 4JC. OCJCJOOO C03 ECJUEEE O O OCOCOOOO CO w T3 > J2 «-• o o co a E 'a. ui o in -o o o ro !§ - cu O «™ >O C3 co ro o «- •r- ro in >o Q. ^* •* -m* UI 0> CO O 4-1 t- a o cj CO 3 E CJ CJ a o o co co T3 > <-• a UI cu to o *j t- CO 3 ------- u. 0 c 19 *-» I 3 — • __• O x o> o ta 3 z "1 •SSI 1 X O U OJ 3 Z El c a> o =! 4-* O IE C/N ^"N CM in O CM O CM O «- O CM O O O O O r- in O CM O «— O V~ O T-* o o o o o ON IN*. *^ ON *- O tN. N* 00 00 to NO to O o o o o §5 28 o o o o X Q> Q ^> .SSI > x aj cj in o co 3 z «— »* "3 o ^ ^> C O CJ CM CO •— 3 Z O CM "co o o "O > u tn NO •M 0) Z O s? o o ss§ ZI .SSI SI X O O CO 3 Z ZI •SSz E"ro > "O > CJ 4-» o NO in (M «- 0 ^ NO O O CM o o CO r— o-a- o o U* S-D ; 2 m • r- TO D. 4 • OJ Ui •' CXI C£ 1 scsl i §*J- « f? -H - * 1 4-f CO • CO t ! CO »*- o •*— (0 -J J C fl> O *» ° 5 z £ "S .> "* T3 3 S i r* "^^ *K -^* 1 O O ^^ ^^ JO "O OJ •* £ ^ 3 i 52 S O CO £3 || .2 ^ '5. u §0 z z to 0 •w S 3 O 3 — » 1— SS CO 1 1 U '5. UJ o CO U «-> c- CO 3 O O CO O NO CM O r- O CM «* O O O O T- NO O ON in T- O «- «- IO 0 O O 0 O CO r- O CO O O O O *- CM o o o o o tO NO O O CM *-0 T- CM 10 o o o o o CM CM CO T- CM in in co in sf -J — J —3 —3 — 1 O C3 C3 C3 C3 NO o ^" in *4" O «- ON tO -a- in NO NO in to •* NT m |N«. co SSigg CO CO £3 O O > C ^ J u a L > > 1 « ^ "e E < • j 3 C 0 CJ FCJ 3 Z fli _* ^ sTS j 3 j CO C O t XI 10 13 : ° "5 "<5 O > > s: ^ >^ ^ 5 • WTI > ja •(-> o , o w o 11 o z; '5. UI 5 iO LE Z O> 4J E 3 O 3 — ' H- Z CJ > 1 M '5. UJ O 2 if CO 3 O O c do i i i SO 3 .*°. S 00 C < S i 4 " in NO i 0 sf < d d NO O o in d d o o in in ^ ^ 11 , §5 u o u u CO CO J ?C 0 CJ CO 3 Z j Q) _> J E J J • : M C fl> *tJ CO * o > r z » J X, •« J • J2 4-» O (U C "U CO O Qi *EL UJ z z CO s§s O 3 — •» h- Z CO 4-» 'E 13 1 0} °EL m 0) CO O 4J &. CO 3 0 0 CO £° " d d o in O O O T— o o o O IN. in 000 d o o O IN. CO o o «- odd IO CM «— to in in _i — i — i iii ^NOO to in NO £ CJ CJ O CO CO ^ I < J; 1 | < 5 FC 0) CJ CO 3 Z J> Q) ^ : SE w i > 9 J J J c to c £ > > >v J : w "o > J3 4-» c ^ CO O (U w & '5. LU =1 O z z to 0) 4-» E 3 O D — • j— Z (0 •E 3 O TJ O U> a Ul a> CO O 4-* t. CO 3 o o CO CM CM «- si- d d O> N* 0 CM d d CO «- o si- d d CM CM «- Nf O O o o in in _j _j C9 C9 •z. E NO O o «- in NO sf-4- CJ CJ CJ CJ CO CO ------- o c CO c CD *J 3 O O. •a a> cu •• 4J TJ C •^ co «- So. ? • CU LU co a; : X J- C 3 «-• O CO L Q *• 5 Z O UJ CO ** t- 0. 0 >. 1 a. •• • 6 £ «S CO >. - *> m i CO I < t. I H- 0 D> c w J . X 0> O BJ D Z Z — * CO .£§§ s"co cISz "co > C CU CJ E — • CO •a > o *> CU Z co a u •. C CD CJ 9 CD 3 Z : o — • 1 E CO i > • 3 | t/> C CD 5 £.2.2 ?"O CD 1) E5" ^ X ^ 2 £ CO T3 > o tn a CD C •o co O 0) CO E 'S. Ill E Z Z O 3 3 h- Z CO 'E •8 o CO 'EL UJ cu CO U 4-> L. CO 3 0 0 CO O o in o o in o to O *"• -O CN CM CM •O 0 <00 -O in o o co co c : . CM in *~ *~ i j L < O CM ' t> ro < c )• u •; < ; co in : ** O CO CO : < JO [O «— in «- o in in CD CJ sO 0 O CJ CJ CJ (/) CO X (U Q 03 _3 Z "re £Si "ca X CU CJ CO 3 Z "co > CfOO CO T» > 0 4J CU Z CO O _] B- f I _> c a* cj r CD a z 3 — •* J E > J C J > 5 co c a> J -Q CO D ! 0:sl j r u ^ •* /^ 4J O 3 O CO O t CU C T3 CO O CU _CO E 'S. Ul lo z z CO 0 3 — 1— Z CO 'E 1 CO '5. UJ CD U +•* f- m 3 Q O CO 5 . .3 O " " O o o O -tf o o N- CM N- O O T- I1*- O CD CD O O ^O . »- ro CM in o O «— co 0 0 C> 0 (M TO tn CO o o 10 *- O O O i- ro ro ^o o in ^ o o ro so c O CD CD t> «- J j U s L ^ ^o co in ; O O T- CM o o o ^ i- * ( i i ! CM ro in CM o o ro <— o o o \d CM CM vo c> ro o o ro CM o o o ro' * -* O O CM ro in in so —i _i _i _j •^ ~> ^ ^» C9 C9 C9 C9 «— >o o in ro o «- ro ro in -NO in sr sf sr r-- §CJ CJ O u u E CO CO o x cu 1^ CO -ss "S g§ ^ "co > .£ 3 ca 3S co a j r cu -^ j > 5 j E J 3 ca c J ^ ,f£ » ,* a ' 55 O CO o •g CO "a. UJ § Z £i CO CO o g 5 . . 0 i z o 0 o ro «— v- z ^* >o co o CM in i O r- rO CM O O *- o in N-in z «- o >o o «- «^ o o -* : . • _ en «- <- >o i ^ o ro l^- i > , i < > CM h~ in O ^~ C3 %o ° Orr . cu ... E 0 0 -J- 1 . s£) CD CO o "^ . CO a> : 3 ro in in CO *J — 1 _J _l E C3 C3 C9 3 E E E O ^~ ^O O _co ro o «- ui cu 0 S- O CJ CJ 3 E CJ u O O C/> CO CO SSi = "co |§i CO x§0 ^* "(O > c cuo CO •o > o 4J CU Z CO O j c cu o 9 S3 Z £ E ro 1 > .J 3 J - CO C O J .Q CO 3 1 o "~ — •* ) TJ CO -I CD > ^ n c CO TJ > si u aj o co o 0 C "88 ca E '5. Ul §0 z z CO O 3 — I— Z CO ^> 4-t *c -s o CO 'EL Ul a> CO U •*-* L, CO 3 a o CO o in in s^- o «- 10 in CM T- O t- ON.CMJO o o o o ro -o -o o >o ro o ro o o o o CO «- N. >O o in CM r«- «- o o o ro o in co CM (O CM VO »- o o o ro -o o o >o to o ro o o o o o in ro r— «-^ 0 00 o o o o CM in in co « o C3 C3 *." >O *™* ^^ ro o r- o ro in >o *o •^ ^d" ^u* in a cj u o E U U E O CO CO O ------- JM SS 5 ro «r- C fl) O _ •~ 3 =s o rvj 1 "S a* *•• T3 -«•* .si g ,-S'EL £ **; J< Sfel o i§!S » I «*-. _* 1 •M C3 4 w c : **• « — 4-» C GJ >. | I in o a -M Jj -. 4-» W C O ^ 3 c ja tJ o> o o e> c ^3 ------- 8s c t/I c CO * = = 0 a- J -a O L — •» O 3 CO < O>M- a CM O Q. K • QS lu a CO C X O C «• — H- CO o > Z CO O J £O -M U "- >•» Q. •*-»—• •* ,M 2 : CO 1 CO CO C >« c 1 '' in 1 **- ' 1 0 ! 2 CO 1 x o o in CO ^ Z 0 ~a ' o £38' .8 "co " o X O CJ CO fs. co ^ z o w "CO O CD c o u so rs. co o o •o > cj *- • J w O Z O *» t/y o • f J — z 3 u C Q* CJ N. r- *• 5(0 3 Z OK) C o> — • • • • C E ffl 00 •" • c i L ; •* C J 1 5 JSgS oS f O •*- — * • • t J -0 eo o o jj E*" J J - ( J ! 0 •• w "0 > ^ t> ? € co Q °. *: ) 0 0 >. I o c is. in T3 CO C3 v- V) E O O '5. UJ S <=> CM i§ CO *> e S SI-M O 3 — • 1— Z CO 4-» i t C CD (S => E E o O N* CO w ro ^* UJ 0 CO U ra 3 cj E op wo CO go o 3 Z 21 'co |§1 > X O O CO 3 Z y - > c o cj a? ^ 5 •o > o Soz J s, C O CJ D TO 3 :z ™ 0> ^^ r E co 5 tf LJ L, en c o i ~5 to > _f 3 = *. « ^o > J2 -M Cl> CD CO O II 0> £ 'a. UJ I§ *,£§ O 3 — > 1— Z CO 4_» 'E 13 eu •g u 'EL UJ o CO U 4-* C— CO 3 0 0 O CM N- •O CM O o o o ^* O CM 10 ^* o o o o O >fr CM o o o 0 C3 0 ; o ** in ^4- *— O • • • 1 o o o . 1 I r 3 u ^- ^* *O ( ^o *•• CD — o o o : 5 ( ^ o o o • o o o o ^ in •* «- o o o o o o o Is. vO SC 1 1 1 O C3 C3 CM O Is. «- co ro ^5 ""J* *o O GF CJ O O CO X 0 0 « 3 = > .COO = 5 g§i E"co £SS =l •D > CJ •w O z J ** " £ §J- i =5 LJ E LJ sj = LJ 0 W C 0 J J3 CO 3 : |5 > ^ X 3 C ^ 4? 0 0 W 0 0) C T3 CO Ss '5. UJ lo z z CO •^ E 3 O 3 — • t— Z CO 4^ c ^ o CO '5. UJ o CO U t-> i- ro o CM «— o o o p o SCO *O o »— o p o sO O «- CM O CM - o o o : i . ' 1 ; • 1 O *- »- 1 o o o i i . « °°^ o o o 1 CM p CM o o o 0 «- 0 ^* fO ^1" E E E |s- CO «- o co o XT -» CO cj o a CJ E E CO O O xoo CO > coo = 3Z 3 x. o cj CO 3 Z CO > C 0 CJ "co > •o > cj •w O z j c o cj -. CO 3 Z 3 O —* f =5 5 J W C CD ;> "5 co i E> S c CO T) > f^ 4.1 (|) O tO Q 0 C •D CO CO E '5. UJ S •w S 3 I— Z "CO =3 o •g CO 'EL UJ o CO O •M C. o o CO °SoS o o o o o m f- in «- o o o 0000 CM CM CM »O «- M«-«-o « o o o o o o> eo •* CM in 00000 in «— >o CM co o o o o o o o o o o 8£ooS o o o o o co o vr CM o CM T- O O CM o o o o o o o o o «- o o o o o c3Soo^ 00000 o «- CM in r- fv. »o SI- r- -4- «•" «•" *^s *X "^ "^ ^ CD C3 CD CD CD OJ *O iV CO in ilill O O CJ O O o o co a o ------- - g S S« s <3T3 S —• S n 5 " 4J .™.S£- I <° t_-Q u. 2S£l g & CD O >? US O -M X O, 4-» —* SC S£ w « n *j 5 S a >. ** >• —• C/> CO CO c d> o «— •-Jj Z CM T3 > CJ •w en z CO O |SS I .£§ O r- O O O T- 000 O O O x a> cj «- «o o •M a> z. CO O c a> o CO D ^ 0) —• §§ gg o o M O M «f— • O gg o o o f c => •o > o 4J 0) Z CO O C 0> CJ ca 3 z 0) __, .5 Si TJ > O •MUZ CO O ro in f in o o o ooo > T3 > 1 4-* O I CO O to ro s- Q. -T Ul 0) ra u 4J t_ (O 3 00 I w TJ > J3 -M s '5. Ul Is °l| •M => CO E t/> 'a. UI 0) a o •M C- to 3 °cS OOO ooo ooo «- «- M M «» « ~l — 1 — 1 C3 CO C3 z: z: E CM 1^ CO «- to •* O ^ CO »--»>»• O CJ O SUE: 0 CO Q M T3 > J2 J-> tt> O CO Q . CD C «- T> CO PJ O O • a> s: o '5. UJ i Q : z u -g O Q. UJ is u •ML. CO 3 - 12 -a- to TJ > .3 4-> CU o co o rn in -SiS a. tu I . UJ (a u v i_ eo 3 a o N. 52 u u U) ------- a "8 u u •>-> Q. — >, O. 4-> — • < W g a* 4-> >. CO CO O E n LU Z -UJ X O , *J' C 0) Q — 3 Z X O CM O «- eo CJ 4-> 0> Z CO O 33i emu •— 3 z "I 4-> o -o CM <3 C => UJ z LU o C < . LU O N. T- .2 S5 Q. -3- CO LU O » O O o ^- —» ... > 10 in o o o •* u i C f •gg O V V> E o § 3 ^* m "* •8 o a. LU co «- S5 »»• co - - <0 3 U E E a o co Q Q m ID o 4J L. CO 3 O O CO o in ^- ro o •* CM o o o T3 > CJ O O «- O in T- o CM o o o o « "D > O ( J2 *-> o> in • o co o "O (O l^- «~ C3 CM O CU •'•.. WE T- O O O 'a. LU •w E 3 N. >O st -* O 3 —• «- t— z o o Q. «--^->0 CM >o l^ in «- eo co «* O O U O E E U E O O CO O ------- ------- APPENDICES C.I and C.2 ------- ------- u. o L. o H- to c_ 0 u CO X 4-* ^ .5 CO 'u CO •* "2 CO O CD X 0) SS uj E §T a! < 1— 1 f _J 0) o 1 *o CO •5 O> C i^- 4-» U) § 3 J2 (_ CO 0 "co F 1 CD 0> g .§• £_ ro c 'E CO CO 1 i 3= CO It CD •g u •w CO "o Q. r*. s u II o o £ UJ sc 1 1- CD CO to ^£ § m fyf i s Q | UJ O o: i— IU a. _, | n i CO o CO c .assicals CO o ro 0) CO u HI CO c CO ul o • >. • CO U. 1 > co to UJ o cy) 4-» 1— CO —1 UJ .Q CD ° a: -I *J3 ° -S o CO 'a. ui M- 4-* O •*— _^ o ••- i- £ CO -J- "" si- CM -O NO . ro CM .) o o : 0 0 r- in : CM r- i co r^ C eo in ** o ed CO t c I c > N . !• o o c II i T— ^J I C L I o p ! v- in «r— *— < j S" c 1 *- O ( 4 i : i in in i i 4 C L < SO O 4 SS 1 -* sr I « o o CJ O CO CO CO U. o • -J CO U. S I > F "• J f— * i -o 4-» 1— 3 CO CO J Itt 1 s 5 : •« 5 1. % % 5 » tf- r ^ • < E 4-> !•= 1 CO —I O Ul 3 £ U . [ Is J E 1> j X 3 C j | % 0 > z I N. > J D J D •§ J O X CO -» «^- o a. ~ in c "o-^ • ^* CO 1— u. -O O N. CO CO -J in sc co CM «- in *- CM «- CM r- r~ in in CM o o N- T- o* in sr in o ro >o xf i O O O O xf O in N. co eo ^ CM : ro in sr o ro *— (M CO in O O CM C o o o o o o i c >$• co r*. «— s3~ %o CM ro ro O o «- o o o o o o e « c I ^ r> ro in in *o T— ro o o o o o o < c £ 1 U co f»- \o CM t^ in : ro •* in >o »- ro c o o o o o o J c 4 1 4 C 0 0 «- 0 0 0 ~< 4 1 : I co in ro in in in < *""" i < c c 4- «- »»• «- in o vo : c^ ^f ro ro ^^ co •• >o ro ro tn NO in c in co xr f^ sr sc i a o a a e_> o a a o o co co CD u. a • eo u! 3 ' "* f j \ • Q 3 4-> H- CO CO SLU |> 3 O J : co o a .a i— j o co ^ 3 L J \ rt f r • < 3 4-> 1— E CO — J S 5 J J 1 CO C > .Q (D j °E L) j >^ ^ s c 0 | %0 • >* ? ^S a i D 0) •> ^ >> O ' CO I *f- E S- H- 4-» O *^- fl> >r- 1— U. ' O o in o in S3 O CM Q CM 0 0 O CO 0,*- 0 0 o o § S o o ss o o eo «- co in lie §' r^ 1"^ ro in in si- «— r*. o in o CM o. in ro N- CM «- «- ro c ' o o s. o in l>- co o : st Pw O «— »- »- o «- : o o o o • ro o *- co «- CM O O 0000 c > > c N o o ro o ro o in in c ro CM in ro : CM ro o i- o o o o < ^ » ro CM >o ro CM ro o *- o o o o _ . c ; < C ro CM CM CM "c 4 : -» >*in in i i ^ ^ t < 4 Rsr «- o - ro >o in i < §0 r« PJ Q CO CO CO 11- -i- CO U. i ?> a «- L i 5 4^ £ CO C/> Z UJ si i g J j : co o D J3 1— 3 > 1. j D j f r • < 3 4J H- o1 £" } J : ? coc ) ,Q {JJ : O O I 3E D i C 0 I *1 1* f CO > o ,> f J L 1 ' CO ! '5. c uj M- 4-* O "~ X CO 1— u- CM o in in si- CM «- CM CM >O ^ N« o ro CM CM r>! vo o o o o o o ro in «- sr sr CM 0 ^ ^ (^ *~ ^0 ro c> NO o o o o p o co r*- «- N- CO C> 000 ^o o o* f- CO l»- o o o o o o co in in «- tn si- o ro sr ------- r >. . cs co r— KJ u- o T- in II 'c CM t*- CM co •* in >. . in n> u. in • to «* to o o o o o CO H? • O CO O II 4J t— -» «- « to co o T- tJ UJ O tO o • • u o o . ea o > I— NO h- O CO fO to —i ru ^o T— tu CM CM in O-d-0 f~ fO ^ CM to sr o to o o to £Z .O 63 0 E «- CO N. «- o CM in § o (VI %^ CO UJ CD 0) 4-* >. o -o o to C CO 0) in o o : o «- to o «- •r- in in Q. UJ -s o o to «- o u U U I CO CO « a. o >. co I U O : cj u I CO CO o *^ a) ^ s: cj a. o o co >- a> ,^ s L U IV> *. CO ------- O h- •* . - !E GJ U. Ji i. I II X • 4-1 03 U. tr- O • § ^> in l»- m ~f ca u. o • i > •* in ro in N. O I O sf ( co OfO O in o> eo f ^> £' in • o I— 4J I— II to CO O 111 T3 O U c 2 .2 3 .81 •t- £t CL ••- tl (t .a -w o o to CM > —• o TJ 1 =1 C L. CO to o * «4- CM O sss o h- in ooo O Q> O D)-- 111 t a> 4-* < : >-• S i < a> >— o »* o 000 CO C O l>w CO J2 «J CD CO t— o o ... o o —• S3 o co «- o -» I 1 LU -j- o co o in o co CM t>j o ro M o o o in o <) o o CM O O • HI O I gs; dK3' CO C CM I o a> *"• E 01 « C J3 CO o 4^ >s •8 O T- • a. UJ O •** —• o u u (U >p- E O O a. o o co co o. 111 O *r- —' O U a> •>- u u a. o co co ^ o a> X I . to M . . _ •£ «*S SO«O 111 in NO o O «- _ IS X M- ** O *r- r u u i o o en co 01 T3 O T- VJ O CO to O T- "- ro in >o a. sr • • • o a. o X ai ------- Q. U 1 1 4^ f.1 £ jj 11 |S CM1! x ST §*"* > 4-» cu Cu Ei O ^ o ra m"^ ?? ™"* o c- « co O D> •- « 'i U. O 2 10 1 1 ^ CM S *? > «~ f ** o co u- in - u ? > CM c. §«— s j D 3 ~e § * °- SM S i; i "" "1 H C *! i CO C5 i2 *O i S* 1- 10 4 o c/> • : UJ 1 A e o t CO C "S, CO > 5 r> • < o c CD UJ CO C or • * — ' 33 j 3 O O • uj 1:0 o • 1 : _j § . t— U CO O - 9) 4^ CO 1 W OJ — J T3 CO ON- •< o « to •*- tp- \O " « CL -3- §UJ • 1* >- S °^ CJ 4-* Q. O CO O 1— U- 0 4-» J *. 3 I j 3 \ 3 5 j 3 .* ^ 3 L 3 > f f o1 5 £ U 1_ ^ f >_ 5 £ U u> X 0 CO _* g 11 F _ & D M a _j a> j _* a § CD U. *? > ** CO u. *? > • o 4-» 1— co tn UJ CO O O CO CO —1 UJ CO C A CD o o CO CO •§ CO '5. UJ >. *fr- 4^ O *r* 41 it- >— u. CM IN. in C\J *- NO *-*-*- IN- in in — — to ^ c • 4 Si! a 000 1 1 c. 4 -» O CM ! OOO C • • • J ooo : c f t ooto ? ^7 ^j PO O^ CO ^O ' M o in : 000 j c < L •* o in *~*^ °. ooo • « 000 ' ^O N" **it" , 1 . sO CJ N- co «— w ^^ C3 ^D ^- v ^r o o u z: £ u Q O CO J IN 3 - J 5 1 i J J i 3 _* ^ 3 • L s 1 !• i s' 3 X u i = 3 s u J : X. in a \^ u £ 1 >- 3 J to J 01 -1 o CD U. ? > "* CD U. ? > CO CO UJ CO O o co CO —1 UJ CO C J3 CD ° E CO CO o cu •g c/> 'S. UJ >. "S^ 1— U. N- o in *- co «— T— T— T— CM O O >O CM O CO CM NO in CM to *- CM »- in ** c. a 1 4- T- T- 0 t- 0 'I CM CO >O fO in- = CM o «— in NO 0.0*^00 a o o o o o i •j c S5-SS 1 ooooo : "c 0 > Nl V c V •j- o o o O- •O ca in o — . c CM O O IN- NO : O — — CM O ooooo < c N I* SO O CO NO — — CM 0 O O IO O O j , 5 < . " 1 in — — o CM . c eo NO in CM N- •N* OO N* — IO *^ ^— *Q v» ^f 1 ooooo s E z: s: cj O O O O CO &u| 1 > ' "* CD U. I ? > ', > : • o 3 4J H- CO CO : ui 1 > ~ Cf> O D J3 t— > o co J ^ 3 L 5 3 3 > f " • 5 IE s u .1 X 3 i J) J at W i § F 5 co [) o J 18 J S -8 >• o -• CO [B •»- : Q. S. Ul H. £• 01 ^ Q. U >• CD 1— u- •f o m ~s- IN- ST 10 CM *- CM t- »— O VT VT ON O O tr- IN. IO NC3 CM — 0 — 0 o in o in «- CM N- O C3 O «- C3 0000 NO ** IN- •------- o. o c_ o H- 0) L. O 4-* U ca tl 4-» 4-» !5 '£ CO 4J CO O > ll CJ D) O CO — > X £_ i Q > •*-» LU 0) Soro Si O t- _l 0 TJ — ' C O 3 > 2 j? i ••— >. E 4-> 3 "- CO — * CO 'o **• u. H- O en c 4-» ca —i £ f 4-> 'E =3 5 »V- & ^ CJ s CO 4-» D O CL. t^ 1 Z CJ Ul 1— < PHTHAL *-X _J >- X UJ rc _i >- 3= H- UJ i CM \*f W 5 E GJ Z a> s. ca c < CO u c CO CO I o CO 0) 4-» ca u CD 4^ >. 1 < CD U. , > N* CD U- Q • • > *Je CO CO UJ o co ~5 CO — J UJ CO C ja ca O CL> = CO % 0 •Si *l CD 'EL Ul >* O •'- a.'o >. CO t— u. ,_ S ! • • 0 t< 0 CO h CM in ^- o » • 1 00 C •i . Otr- '- «•* O - • • ( 00 C •v «• > c V C c o «- « 3S ' "~ ° £ O O 1 r i 0 5 £& ? 00 c S c J 1 < CM O C C •• I c 1 M >» C i C 1 c 4 5r-- c to t •* •* < 4 ~l t « O CJ s: cj a co ca "-' ?> -* j •*. r ca u. 5 0 r> 1 4J 1— 1> CO CO 3 Ul J _• 5 _» 3 3 O CO L !• S 5 i c • < J 4-> 1— CO — J LJ UJ S LJ U S a _i : co c LJ O O 1 E 3 S u ^ X 3 C ca ; z a 3> I O 5 a> •• TJ a o 3 CO u '5. -> I4J K I £ H-& O — 0) ^ Q. O >• CO 1— u. in «- 0 ^> ' i»i «^ co r>- -^ • • • s O CM II 4^ 'E rs in o o in in f>- T- o SS^ ? • . • o ooo -o o u 4-» c ca -OOO 4-> °. "^ ° 3 ooo — o a. S S o 17 O N3 CM U OOO Ul UJ _J J- X t •* «- >O E O CM CM II OOO E 2» a> 4J >. ca S CO «— o o o 'E ca en £. IO »* •* >* £_ O CD CO eo «- t<- «^ «— ro CD CO 0 -O 4-> •** co *a- >• ca c O O CO CO U. *? > •& £*: ^_ > CO CO Ul ca o o co J£ CO — J 111 II z «: J3 o CD "8 CO 'E. UJ ^ o ••- 0> •!— h— U- ,_ s co : ro '. u o u -o t o 1 o < c 4 . -o : CM 0 c r r v c 5 c o < Is- C to i o i < a H a c ro < ro = 0 i 1 2 C .j j ^- C 1 '] c 1 •* i i c ' c t 4. fe I ~ 1 u CJ CO ca u. i > i *4" j ; I1 &ul I Q • -> 1 > 5 ' ^ • "* , • O 4-> (— 1 ' CO CO t UJ J .» ' 3 , 3 , CO 0 -• ja i— - o co . i \i S I ' E ! 5! jss CO _J J Ul 1 J , : c : | : co c ; ' |> .« :' >« i i 0 CO - . z a , » • 5 ! co %-Q K . O 3 ! 3> < ) •> ' 0) a . TJ > , o !• 1 K : iu ^ S • >» o — Q. O >- CD H- U. >^- CO «— ] °; • 'ro' '. c T c CM CM 0 : in co ~o in i»o o ooo ... ooo c 4 . ooo - • • • ooo c e N r r ^ V r< T1 if* in h- •• c ooo * • • • c ooo L i s 3 >0>OI^ ooo c ... cs o o c 1 ! i < 4 < < t ( ir i ( t ro N> ^3- i ! i ( | 1 CM CO Is- J r- >* rO C O CO NO C x— sr ~a- •^ S E: i >» • >-» ca u. •~ o • § °> >. s • a 1 4-* 1— 1> CO CO 3 Ul D J _» a _» 3 O CO o 3 3 2 1 S • < 1 4-* 1— O CO — J 5 •" u E U J j CO C L .Q (U 3 °S? > 3 2 D _» ^ j 0 Z J 3 I X 5 7) O D (0 t) UJ .» X ^ a r >. c 'o.ti H- l^ ^_I CM «— -* 0 CM O oo o 0 « o CM CM O >!• 1^ fO NT § in cj ------- > (VI CM •— CO U. _. _ C O • * " a i > f>- in ^> o» o •* ro •o o CM CM 4J F roin co tn ro «- o ui T- in CO 4 o u in ro o g 1 I. ul ,7 O o NO CM O •* o o CO O O o I I . eu co v; ro & ••- O -*Q n. eo •* u 4J >• i o> c. o H L. O O) s o. UJ SCM o i X **- -u o ••- o ••- — a a u a— S Z CJ ^O QOW ------- APPENDICES D.I and D.2 ------- ------- N. ro o o u ^ o -J- I g 5 O • Q.E D) w ro~o.ro o — u. ••- ro • CM v^_, s o ro o C .C ••- T— NO C\J o -M _j jg CM ro co |Sj — ° f\i r-- in in ro N. - WO —•* < "-d* u. 2 O N^ o § to o in o II o ro o • «> CM T- »- I « 4J E II 4-» 'E ro ? o §c • as — u- ••- in xr in -J- CL> O • o o ro <- S §- O ••- - c c t— ro o • nj uj o «- u. •— Q. I 4-> . TJ I II 0) 4-> c < UJ 3: CO !i •g u o a. w •*-» co *o uj 0. o o s I o: Ul 8: S <1) 4^ >. | 8 O •. — to 4-> 8- >• E o «- o -"--- £%£ g g ^ j : o ro •* i>- •* CM >O (M in C 03 1 V "S >. c CD o - O - 111 Q. o c to ro eo 1 O -CO t> «O O. i .r- u_ .-. t> in o 1 *- r\i sr »- o- ro ooo c ro .g, W 4-> _1 73 UJ Q. 0) O E o -a o £ UJ Q. 0) O £ E U z'S. •a v " o ro I o u en c_> cj : cj cj en co 3 O U O O E CJ CJ to o co co ------- u. § H- H-> 1 |i _i ca i 1 £• •^ 1 ir « ca APPENDIX 1 Averages, v. ^ V _J !i cu _i 1 4J o D. O jf 4J M _1 S « i CU 1 4-* 4 S •• J J 4-» ~ Ul V •^ C CU v a ? § ». i is ? Q O 11 ^ 86J ( _*: 4 c?cu - Ttg , 5§ : -•s r .SJ2 c — * U e> o t» 4-* .5 i ' §1 - V* CO W Ul J <_ . •? • CO _J 4-* i ' Z ( 151 j f|; ° < J > l"lu! 5 i 4-» * f ^a-5* r 3 i 1^1 i Q-° x 3 ° E i 1 >* n (D O • D Q ••- U. •» i 4-» • x *— Q.> I ° c s c a • o o I JJ .r- t— J CO 4J CO | m«§- _t 1 . g<: 3 4-. — f± - CO 4-> _J s '"a- 1_ J) S E i z > u M 0 J CD J >* .^ 1 % i C CD g 1 g 1 CO -§ o 'EL Ul Source ^ c. o in ** 4 p T- CM f- CO N T- IO N- f CM «- T- 1 G < C O CO O U •a- CM i^. = T- O. S3- 1 in o to : c - o> co -a- >o to «- = in CM -* o in to • s± CO CM C\J -e- 03 f «— o o « • •* o N. e C3 O> tO os to eo o o o . c -* ru >o • < o < £ a • IBS O CO CO • • vx — » £ O11 — C _^T «^ *tf ^ J ° 5 j - — I S >. C C f 3 CO O • CO >O CC 1 O "- "- — O C I 4J • TJ • «• 1 *&>S ^ i u c/ 5 ^ - £ j s^ >». E **O U E c — • — N- a J O •*- -J % O^ U J •*- CO St » = - 4-» ca • o > < Q. X > , ° S a i i 3 CO O • CO ^" E O •»" U- **~ ^O ( i> ^- Q.> o K> ; •* O 3£ *" 2- ! 2 ' 5 .gog § i! i S£S;:s °i Ul Q. O O = 0 = u c 3 4 ;! jljsl fc : co '£ — J TI • i ° E « 2 i^ (D t 10 O TJ J 0 CM S z S" i 3 J 3> S ' 0 J j eu "c >. CJ O -• t, • CO 3 CJ = 0 CJ cc co co Option(4) i \ , 3 CD > 5 S - : c j ••— - 4-» < Q. O i » i •» i K «- ^ J C -i \ • r « ) UJ 3 J j a • •* 4-* 3 0) •* Ul •• 3 i. A J i j> ? >. 3 » o> t^ !D J U ^ •^ ID C f 4J — 2M ^ 2 J= • S CO 3 4J _l g O I 0 • 0 * E ro £ ^ N *" < Sc c • CD in *• •r- U. •*- O I g.>| - § < C, •M U •<— = -3 - ! O • CM C E : J 1 1 C C ! o • CD in < ..- u. •*- in = *j . TJ • a>o M ^ • I §c *o < .^ e .2 5 4-» CO TJ • f Q. O C3 H 1 c O < CO O 4 r!3Tj "! S- £ ° ^ c u M- a> O TJ O «- • §.2 Z D- C ^ < >» CD ^ 0 O 3 CJ O CJ « CO CO > 'C ~C! - .22 > 4-1 O : o * j xe > ca o — CS •"" 1 4-. 3, "8- • > j «- Z —CO 3 «|Q } 1 &g a o •>- 5 ^5 1 jj 3 CO 4J 2 mg > 3 3 j j . o - V- a w 4J 5 WB ^ ^ 3 1. a > E D J> _ > - ^ 0) a> k» D J D £ D ^s .g Lt_ 'F- >1 'i i .g LL. ••— Qg S| g *TJ _j d> § -2 z'5. UJ Source si o o 0 0 !S in 0 in o ^- in o in o «- St- vQ in «- to >* SI o to 0 0 CO O ro o sj- NS- 0 N- o to o o -- £ CJ a co u. "a! S 1 'o (0 I TJ 0 .c ^> TJ 0) "3. °3 S « Facility-le> c CO 1 0) JC CO CO I CO "3 o CO 0) *i e Indicates I i ------- LU o C- 0 H- tf> C O 4-» (D 4-» ic o °ZI CO 4-> U. CA >>S S s- T— CO O o '£ m CO O x > — APPEND Averages, 'ing Delta >» ?i f= D) _1 C 0) 0) -I < t 4-> C CD 4-» 3 •g t|_ O o? c »^ 4-» CA «^- _I i • • S: 4-> -, •,- c *J C 0) * UJ J CD • -J C i __* 1 o i u. : o i L =5 I- CD — > 3 •• o — * c CO O 4 » 0 J *; : CO — E •— c 4-> C o LU C 1 Q) i CU 1 T1 c! 4-» ; CO i 3 I — * 1 1> 1 0. C < •i ( £ < J *-x x. %3- 3 *-* f g J (^ E Q. t i 3 X i 1 z >J- "} J *"^ : «- sl C E O U "- a 4-> J Q. : u 1 I cS" E O x^ 3 i 5 1 . 30 4-> 20 » ^x LU ) 3 3 J 5 co j LU 3 _f ^ 3 L 0 J I 3) 3 1 X 5 3) > ,> D J l> j X ^ f C >•% — ' £ f-S z o> > * O • CO '5 • "5 O.> 0> 0 Z 4-1 CO **fc CD Z §c • CO • •- 12. •>- 4*J • T3 «§->! Sc O CO '•W CO T3 S- z '£ !j T3 Q. 0) O Z «f* OJ E CO z 'EL LU CU U 1 CO S ! o 4 • _ r< c r * CO ' c t L ; N- L CM N- « O f I • ' 8 = i < CO in u w- u d -> C O 4 CO ! v- C » 4 o : "< c t < •» CM 1 i 1 I i I I •1 C O C Z * o : v. " E O < U «*% j ^f s» C D 0 ? £ u* o : ° 3 n 3 VI JK f ? i( * O ^s J «- J ^ J O C 4* E ° c s i i i co- = o I> ^- X 3 c "V 4-» UJ I > j J 5 -M* -> w 3 LU _» _r a _ It J j 3> K ^ L> ^ J D j X .• j C 4-» ii« z en ^ c c •p- U. *p- 4-» . T3 Q-> CU O Z 4-> _>.'! 0 ^ CO" z §c • CO 4J . ^ Q-> CU o s: c c O CO 4-* f— ^3 Q.CO CU 0 Z '.M 5 "5 Q 1 G> 0 Z •sl E O z'5. LU CU § o CO §s» CM CM «- O O _ss «— ^ in «- o •* «a- CM " o o in in po to §£i §§s o o o »* C3 O O «— O CD O O «- CM m Cf z ^« z u u a v> - sc H- LU CM CO 5 CO 0 1 < o o 1 u 4-* c 01 D I .2 'E CO o> 1 u 0 o> CU 4-1 CD U o "co c 1 ^ 4-» C -C •— .2^^^ Q.Z O> O > < >. c c CU O • CO Q •— U. •— 1 4J i . -O sr Q-> cu o . z *•% 4-» J ^ J ^ E >> C , C CO O • CD ?r-:5 «- Q.> cu » *» CO ^5 • O < CO 4J — 1- .- CO 4J ^J *O LU Q. CU O : Z "S 73 -S I'E, LU CU 1 3 O i CO «o ro CM f~ «- CM 000 CM O CM co ** in o co ro T- O O ro CM ro o \o rj SlSSi ro o co o o o o o o CM CO -J- st O «- o o o «- CM ro o £ o ^ o o o z u u Q tn c/> u. > S ^ 4-* •p— U CD n C CO 1 o -C . 4-» £ T3 O |3. 4^ < 1— «J 1 1 >» +J U CO u. c CO 5 o JZ CO •g CO 3 o i CO 4-> |i i U) o 4-» CO U 1 =ft ------- - g s §ro o ••r in jOod °§- a <§•= «- O CO o in «- is *-• 4^ co \~* in •g 8> CM 17 o o • o •» in to o °"~u;!5 °J *} °, to1 "* o"> E *~ g n o >. c . c fQ O • (0 •* o.> o • 03 O^_.. 11..,- NO O NO ro r>J cj «_• >. E o CM i c ^/— o T- i **• O C£. t- o o O a 4-> u. "• • c. o> . 0§OJJ 2 ««j ** O W 4J W ^ o —^ a £?** E^S ca to a g ul 17 U £ CO —• E -^ 3 o co (M o 1^- rvj g • o < a > it— |— *^- • • • I -M -J *O (MOO UJ 3 _J o f H, ••g u . O l»» O O to o ro NO CM NO 14* • TJ 10 in r<- "c§->^ - CO 4-1 CO Tf LU Q. a> 0 E . O < CO 4J — I- — w 4^ —i -a mg. o ro o o S- o N^ ** in *?- o •» r>- .2 •* •a u o u u EC-JO . o II z a. LU > u u : u u I CO CO 3 U ------- s u g •£ f CM ro o OJ O -O ro o «- o -j- in o CM in CM T- «- S 0) • a> CO O • CO CM O •- U. -~ *f 1o > 4J O CM c O • CO - -M • " CL> i CVJ O •*" U- o ro ? >O o in "- ro o in <-£ >• E «-«- CO c —• •*- ^j* o ro O ••- _J % «*T CM CM '5 o • o o o § C CM > ro ro c co u JC "8 IU .J II ID ; 0 -• C CO CO CO >* O 0) CM 4-> C E • if- D> O — O HI §••- 4-» CO 1 > CO UJ CO 0) 0- > 0 T3 Q- CO «S • B) 4-> co c ra ro'x'3 2 "H 2 0> O CO *!« 0) c ra l-.£ E i E »^- fco to to ui _l < __ — * 0> 4> > > o 0 _J _J 1 t 4-> c to co 44 ti 3 3 — * ^ ^ 0 Q. O- i O o: i— UJ a. < | 0) >• 1 UJ CA fl> •g u ca ^ "o a. CO ca u i?g . Q «r- U. • O O « 4-» CO U, Q. J.22 CO 4J _l UJ O. O **- O c to ^^ •5 m IU CM C CO »9- o> in i co 1 ^ S •a o «- to S> E ro o __* 1 3 & •g CJ ollutant a. "co 4-> 1 O D) •M CO u £ _>. II 1 4J 4J.2 « *J UJ Q. C • o CO iJ uj a c • ca U_ (^ c O CO CO ^ Q) C <_co j -5 - o o S8S o o o o o o CM »- ro a E ra z a> ft CO s ca * Q> "U 4-* C CO 4-» _D O Ct_ CA CO 4_> t. O) 0 4^ CO U 0) £ "i >» c ca o Q •— Jg is'i. 0 .s 4-* »- CO 4-> UJ Q. C • 'CO U. *r-> , o to m '•£ 'c < CO 5:5 CO M- O o-a o ."' z '5. UJ CM in CM r— in ro 2.°° o o o SO 0 ------- :££•! ONOO c j: ~- NO r«- NO O ±* —I 3t CM NO O — c 3t in o ro O ro ro NO f i 5 2 •r- C * O O Q O •* S a o • ra sr >* KI o »r- u. »^ N- CM in " - Q.> g ¥-3 ex «- CM O >O & *§•>! - tn g p UJ *^f ' Ul O • CM in ON o s 8 §•§ • § ro o o ••- u. ••- co S 3 ^ *• o — _J •»- CO >- 4-» O & CO g Nl | >• >. C C CO co o • CD in T— •*- Q — u. — o si- TJ I 4J • TJ • • <1> xf Q-> cu ro«- E 0 = o O NO ro \o co NO s- D. tu ^. I Q> •g U 4-> i S W »> O 4-* C!) 1 cS o s. >. n c CD O • CO *r- Q.> O O. E • O Q (O « *J W "5 UJ OL d> • o ro 'cS- * in O N- O o co o sr sr m tn v- NO •* o CM o o ro o CM tn "S-8 §8 z'EL C CO CM CD < C C xf • O O CD v~ O V- t-— CM T- 0> *• CO TJ • ^— nt n Q) i"^ Z O £ . o - . o '^- I— ••- CM CO UJ Q. 0> O « O E o a. c CO O) z a. UJ in c c NO z *J •— t— •— ro u tn «-> »T3 • a> tu Q. o o eg 3 o a. c CO o> . o < § •*-»••- i— ••- u> -w _i -a til f>^ dJ o s ^s C C ^C CO O • CD CO I1^ H; S -frs C C O O CD g. • •g u - | • S3 ** °° 8 C C O» O* CD ^ ° 2.2 « ^ 5 o Li 01 o. tu •g CO u 0) Si g Ul a u u £ u u Q V) U> _ CD 0) u c ------- o z II •If §53: 4J — C = ••- 4-> O • -o o O in JO o •E a-= ilz CO — O .1- CO ^ 4-» U. CO (0 £. M- S- 000 o — c ro 10 CD u. E •-- o -^» o cu X JD I 4J "-> CO CO W C OJ fl) •.- _* O) >* 3 a> o> co > T3 CJ < c D) 3 4-> C W CO O O) Ul -J < — I 4-> *-• C i S II o CO g UJ >- >— Ul X C C CO O • CO at O £ ~ >s S •r- (0 : 4J O • Q. X O CO CO O • CO CO Q ._ U. ._ O C C T- • O O CO CM 4-> ••- ^- >•- T— en 4-> at T3 ui O. u o S CM tn g % I CO z tt) •g o CD +.§- C CD O O •— I c CD >- c CO O O •— L As-: .§« UJ O. O - UJ Q. O o co O CM >O CM O O I • • ro to it o z O O •£ ill! ;2 i 4-» • T3 in sr Q-> o S °' = I— ^~ 4-> a. Xi1'^ ^ o ^ H » —J ••— CO . Sfc >• 4-» iO • I ° I >- o «- o o N- t1- O O •o o ro CM §52 ' o o UJ I CM >v C ! ro o i CB «- *- I c 'r- Q .•- CM IT! O CO 4-» I— T3 O C? O UJ Q.OT co 4- 0) O TJ O CM «- I *£ CO Z '5. Ul o 0) o 4-1 CD O U 2 II •8 S 8 'c o D) cu . 0) 4-> UJ Q. o •- 73 0) o T- o CO O O •* >O O «- O »- §.2 Z D. UJ c CO •5 « a E O a u u E: o o Q en en ------- s 1 " 4 M . O- W » " £ c. t- a f o a ( W 09 1 n ui ; •*• — * c 4J CD « 0 > C 4j e> «*- _1 L Jii. t _j ^- 4.* ; "HiS ! 10 *^™ O 1 «.£ £ ! o o o « 4-* 1 *1| ! >. o 5 OJ 4J C E o — o o *p p^ Q> « f'J.'S ' — • o a> e> "3 51" t- CT>4-> !•• **•• fil §5 fe> u> - « UI —I < — • O > o> • V -J i 1 Z? o O Q. Ou 1 J i" ^- _* *E= ro in m -» C.C«^ Of\J^ - O4-*-J% OJ O» ^O N 5 '£ § • ooo : 50 > * N> _ - h i &g .§ m,-,o 1 s o ^ u- ^ in ^ ro j o *^ QL> o> CM CM CNJ ' JOE ^ 11 1 S £ • ! - ^ >. e T- o o S S-r;^ §3ic = XT •*- OJ =£ • • • 1 S 4Jo- OCM«- : J Q. X - | ° = ! S u * - >. e c ^_ : o ^~ QL> cu - N« in *O z O E (D j • -> c c to in -* ^ s • 5 o oj to»-cM e r *j — H- •— «- in to i ^ CO 4-* CO "O • • • * Ul g. g 000 s 1 5 *O^CO OvCOOv _} 4jra-p>^ r^toin CO 4-> _1 *p O -* CM •J III O (I) • • • • g O E O O O -> _• _. in O H— CD • X O T3 O «- «- CM en £ co S I'D. c ui I o CD o S 1 LJ 0 cu a> o •»« U L. >. 3 a to o — * O £ CJ O O • i Q.E ro J ° rf 3 ^ D >» C C 1 O E 2 1 o i *•* 4-< J C — * *^- 3 4-1 a • = a- | 1 1 xc c CO O "CD D Q «f- U. «r- i. J- S.> "S a c =c , 0 C C o • o o ra 7 CO 4-i CO T3 U Hi O Q» 3 0 E o J gp c c • O «^ CO M 4J »^ »- -^ D W 4-» — J "D -* UJ Q. 0) I ° = 8 « ^ H- CD c o -o CO O 3) E CO 5 z a. II Ul 1 to u a> mmf go o tf L. g CO to o o •e— O T— co «- in O CN1 «— o o -j- >* to sr in N- O in N. f*» *Z to ro o o to oo toQj CM CM CM CM N. CO to >o in ooo ~» 0 0 ^3- co in in in O C3 ^ «^ CM «- tO g o E CJ CJ O CO CO > 0) 1 o •s JJ* CO c CO CO CJ ------- APPENDICES E.I and E.2 ------- ------- cs fc "Si ** ^i^ co .c oT "O •o « 55 i& t^H ft+ a* ft- CO ; » I ' 1 , j ? tl^ & a 1 < i pa f^ T— < vo 1 ^ S 8 CO a ------- w I a> •a s CO Q> 15 0 a> I CO jb •'<*' § 0> 0, ? t 3 ? •o °* 4> JH Q eS « 1 I "w I S •s o> Q 1 1 & OQ 1 I < I 1 M T-t Y— 1 VD I 1 §5 5? VO O "H. 00 1 a -------