Regional Air Pollution Study (raps) Preliminary Emission Inven-tory

EPA450/3-74-030 January 1974 A REGIONAL AIR POLLUTION STUDY (RAPS) PRELIMINARY EMISSION INVENTORY U.S. ENVIRONMENTAL PROTECTION AGENCY Office of Air and Water Programs Office of Air Quality Planning and Standards Research Triangle Park, North Carolina 27711 ------- EPA-450/3-74-030 A REGIONAL AIR POLLUTION STUDY (RAPS) PRELIMINARY EMISSION INVENTORY by Fred E. Littman, Sylvan Rubin, Konrad T . Semrau, Walter F . Dabberdt Stanford Research Institute Menlo Park, California 94025 Contract No. 68-02-1026 EPA Project Officer: Charles Masser Prepared for ENVIRONMENTAL PROTECTION AGENCY Office of Air and Water Programs Office of Air Quality Planning and Standards Research Triangle Park, N. C. 27711 January 1974 ------- This report is issued by the Environmental Protection Agency to report technical data of interest to a limited number of readers. Copies are available free of charge to Federal employees, current contractors and grantees, and nonprofit organizations - as supplies permit - from the Air Pollution Technical Information Center, Environmental Protection Agency, Research Triangle Park, North Carolina 27711, or from the National Technical Information Service, 5285 Port Royal Road, Springfield, Virginia 22151. This report was furnished to the Environmental Protection Agency by Stanford Research Institute, Menlo Park, California 94025, in fulfillment of Contract No. 68-02-1026. The contents of this report are reproduced herein as received from Stanford Research Institute. The opinions, findings, and conclusions expressed are those of the author and not necessarily those of the Environmental Protection Agency. Mention of company or product names is not to be considered as an endorsement by the Environmental Protection Agency. Publication No. EPA-450/3-74-030 11 ------- CONTENTS LIST OF ILLUSTRATIONS vii LIST OF TABLES ix ACKNOWLEDGMENTS xi I INTRODUCTION 1 A. Saint Louis Regional Air Pollution Study (RAPS) .... 1 B. The Emission Inventory 2 C. Organization of the Report 3 II SUMMARY 5 A. Basic Concepts 5 B. Results of Specific Tasks 8 1. Task A 8 2. Task B 9 3. Task C 12 4. Task D 13 5. Task E 14 C. Recommendations 18 III TASK A: DEFINITION OF POTENTIAL USERS AND USES 21 A. General 21 B. Summary 21 C. RAPS Requirements 22 1. Atmospheric Structure Models 24 2. Transformation Processes Models 25 3. Mobile Sources 27 4. Other Studies 29 D. State and Local Control Agencies 30 1. City of Saint Louis 30 2. Saint Louis County 32 ------- Ill TASK A: DEFINITION OF POTENTIAL USERS AND USES (Continued) 3. Missouri Air Conservation Commission 4. Illinois Environmental Protection Agency . . . . 5. Federal Highway Department DOT E. The Planning Agencies 1. East-West Gateway Coordinating Council 2. Industrial-Waste Control Council F. Research Programs 1. Metromex 2. Community Health and Environmental Surveillance System (CHESS) 3. University Research 4. Plant Damage G. Conclusions IV TASK B: EMISSION INVENTORY CONTENT , A. General Principles 1. Introduction , 2. The Emission Inventory System B. Precision of Emission Estimates C. Inventory Resolution 1. Temporal Resolution 2. Spatial Resolution D. Pollutants , 1. General Discussion , 2. Nonreactive Gases . 3. Reactive Gases . 4. Particulate Matter , 5. Other Pollutants , 6. Heat and Water Vapor Releases E. Source Categories . 1. Classification Scheme 2. Stationary Sources . 3- Mobile Sources F. Emission or Data Conversion Factors G. Physical Aspects of Emission Sources 32 33 34 35 35 35 36 36 36 37 38 39 41 41 41 43 46 48 48 48 50 50 50 51 53 55 56 56 56 58 66 67 68 iv ------- IV TASK B: EMISSION INVENTORY CONTENT (Continued) H. Units of Measurement 68 1. Specification of the RAPS Emission Inventory 69 1. General 69 2. Point Sources 70 3. Point Combustion Sources . 75 4. Point Noncombustion Sources 77 J. Stationary Area Sources 77 1. General 77 2. Area Combustion Sources 77 3. Area Noncombustion Sources 79 K. Mobile Source Procedures 79 L. Natural Background Emissions 80 M. Particulates 80 N. Scheduling and Scale of Effort 81 V TASK C: EMISSION INVENTORY FILE SYSTEM 83 A. General 83 B. System Environment 84 C. Inventory Operation 87 D. Data Structure and Formats 93 1. Source Categories 93 2. Data Item Formats 94 3. Other Data 97 4. Data Element Formats 97 5. Summary Table 100 E. Format Extensions for Batch Data Entry 102 F. Implementation 104 G. Specifications for Computer Programs 105 1. Data Management System 105 2. Input and Output Subroutines for Emission Model Programs . 106 3. Verification Program 106 ------- VI TASK D: SURVEY AND EVALUATION OF EXISTING EMISSION INVENTORY DATA 107 A. Introduction 107 B. Existing Emission Inventories for the Metropolitan Saint Louis Interstate Air Quality Control Region . . . 108 1. IPP Inventory—1968 108 2. IBM Inventory—1970 109 3. DAQED Emission Inventory—1971 109 4. NATO Emission Inventory—1971 109 5. NEDS Inventory—1973 113 C. Traffic and Transportation Inventories 127 1. Streets and Highways 127 2. Railways and Vessels 129 D. Summary 133 VII TASK E: EMISSIONS MODELING 135 A. Introduction 135 B. Modeling 137 1. Review of Existing Models 137 2. Specification of Emissions 146 3. Emissions Model Verification--Mobile Sources . . . 153 4. Resolution of Source Location 158 C. Consideration of Natural Emissions 170 D. Summary of Conclusions and Recommendations . 172 REFERENCES 175 APPENDICES A SAMPLE EMISSION INVENTORY PRINTOUTS A-l B SUMMARY OF EMISSION MODEL REPORTS B-l vi ------- ILLUSTRATIONS 1 Metropolitan Saint Louis Interstate Air Quality Control Region 31 2 Emission Inventory System . 44 3 Schematic of Inventory System 92 4 Sample Point Source Item 95 5 Sample Area Source Item 96 6 Emission Inventory System Concept 115 7 Number of Point Sources and Emissions 120 8 Particulate Emissions for Saint Louis Air Quality Control Region 122 9 SOg Emissions for the Saint Louis Air Quality Control Region 123 10 NOX Emissions for the Saint Louis Air Quality Control Region 124 11 Hydrocarbon Emissions for the Saint Louis Air Quality Control Region 125 12 CO Emissions for the Saint Louis Air Quality Control Region 126 13 1969 Traffic Map of Saint Louis Metropolitan Area Interstate and Freeway System 130 14 Schematic Illustration of Classification Matrix for Emission Models 136 15 Performance Curves for Cooling Towers Giving the Exit Temperature as a Function of the Ambient Wet Bulb Temperatures for Various Exit Relative Humidities 154 16 Performance Curves for Cooling Towers Giving the Moisture Discharge as a Function of the Ambient Wet Bulb Temperature for Various Exit Relative Humidites 155 17 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height is at the Surface and Atmospheric Conditions Are Stable 162 vii ------- 18 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is Five Meters and Atmospheric Conditions Are Stable 163 19 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is 30 Meters and Atmospheric Conditions Are Stable 164 20 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is 100 Meters and Atmospheric Conditions Are Stable 165 21 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is at the Surface and Atmospheric Conditions Are Slightly Unstable 166 22 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is Five Meters and Atmospheric Conditions Are Slightly Unstable 167 23 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is 30 Meters and Atmospheric Conditions Are Slightly Unstable 168 24 Variation of Integrated Normalized Error with Longitudinal Distance from the Source When Source Height Is 100 Meters and Atmospheric Conditions Are Slightly Unstable 169 viii ------- TABLES 1 Emission Inventory Requirements 23 2 Emission Inventory Input Requirements for Atmospheric Structure (WATPIC) Model 25 3 Emission Inventory Input Requirements for the Transformation Processes Model 26 4 Emission Inventory Input Requirements for Vehicular Sources Model 27 5 Emission Inventory Input Requirements for Mobile Sources (Airports) Model 28 6 Emission Inventory Input Requirements of the Air Pollution Control Office, City of Saint Louis 32 7 Emission Inventory Input Requirements of the Air Pollution Control Division, Saint Louis County Health Department 33 8 Emission Inventory Input Requirements of the Illinois Environmental Protection Agency 34 9 Emission Inventory Input Requirements for CHESS Studies ... 37 10 Classification of Sources for Emission Inventory 57 11 Major Noncombustion Sources of Pollutants 64 12 Pollutant Sources by Class 72 13 Tentative Workload for the Monitoring of SO and NO Sources 73 ? x 14 Development Schedule 82 15 Sample of DAQED Inventory SO Emission Summary 110 16 Sample of DAQED Inventory Current and Trend Data, Ambient Concentrations ... Ill 17 Comparison of Emission Inventories for the Metropolitan Saint Louis Interstate Air Quality Region . . . 118 18 Number of Point Sources 119 IX ------- 19 Number of Sources Monitored 121 20 Breakdown of Area Sources . 128 21 Classification of Possible Emission Inventories 131 22 1967 Emissions from Rail Operations Within 100 Miles of Saint Louis 132 23 Summary of Content of Emission Models Reviewed 138 24 Emissions Models Input/Output Specifications 139 25 Ratio of Cold-Start Emissions to Hot-Start Emissions .... 150 26 Cold-Start Emissions from Light-Duty Vehicles 151 27 Natural Pollutant Emissions in the Saint Louis Area 172 ------- ACKNOWLEDGMENTS This study was carried out by a project team led by Dr. F. Littman of the Operations Evaluation Department of the Engineering Systems Divi- sion at Stanford Research Institute. The project supervisor was Mr. R. Rodden, Assistant Director of that department. R.T.H. Collis, Director of the Atmospheric Sciences Laboratory of the Electronics and Radio Sci- ences Division, acted as project advisor and assisted in the preparation of the final report. rlhe principal authors were as follows: Section III (Task A) Section IV (Task B) Section V (Task C) Section VI (Task D) Section VII (Task E) Dr. F. E. Littman K. T. Semrau, Chemical Engineer, Physical Science Division Dr. S. Rubin, Information Science Laboratory, Information Science and Engineering Division Dr. F. E, Littman Dr. W. F. Dabberdt, Atmospheric Sciences Laboratory Other contributors were as follows: P. A. Buder R. C. Robbins P. J. Mai-tin •\. E. Moon G. 1. Thompson XI ------- INTRODUCTION This report presents an operational plan for providing emissions data for the Saint Louis Regional Air Pollution Study (RAPS). It also describes existing emission inventories Tor the Saint Louis area and reviews in detail emission models that have in the past been used to provide emissions data. A. Saint Louis Regional Air Pollution Study (RAPS) The Saint Louis Regional Air Pollution Study is the most ambitious study of its type ever attempted. Focused principally upon the verifi- cation and development of air quality models, it is also designed to de- velop a better understanding of atmospheric transformation processes and to provide a basis for studying various air pollution control strategies. Material advancements in the technology and methodology of air quality monitoring and other aspects of air pollution control, including partic- ularly the improvement of emissions inventory procedures, are expected also. Considerable attention has already been paid to measuring meteoro- logical conditions and air quality in the test area, and a network of 25 automated data collection stations will be installed, stations which will continuously telemeter their readings to the data center. Data of this quality will be valuable to RAPS, particularly for diffusion model- ing on an extended scale. For ihe lull reali^atJon ol such purposes, however, especially lor the modeling and prediction of air quality, it is essential to know the emissions from the air-polluting sources. Without effective assessment ot sunh emissions to an appropriate degree of resolution in time and space, the experiments and studies of the RAPS program will be severely handicapped no matter how well other measurements are made or other fac- tors assessed. To collect data on critical emissions for the multitude of different sources in an area the size of Saint Louis is a very tall order. It can only be met by an effort of considerable magnitude and cost for exceeding any previous undertakings of a similar nature. Indeed, it must be recognized that the requirement demands and warrants the appli- cation of detailed data collection procedures considerably different from ------- those appropriate to a less rigorous determination of high resolution data and that many of the estimation or modeling techniques used hitherto cannot be used. B. The Emission Inventory To determine how to meet RAPS' needs for emission data, the problem was considered under five tasks. These logically follow each other se- quentially, and they were addressed, at least initially, in that order. As the study proceeded, however, the interrelationship of the problems addressed under each task became ever more apparent, and, although the separate submission of the task reports in draft form served the useful purpose of checking and correcting progress, it was agreed that the re- sults of the project were most useful when integrated as they are in this final report. The tasks and their objectives were • Task A. To ascertain the uses and users of emission data in the RAPS program (and elsewhere). • Task B. To specify the content and nature of the emission inventory for RAPS. • Task C. To specify the data formats and a data handling system needed for the inventory. • Task D. To locate and assess existing sources of data suitable for incorporation in the RAPS inventory. • Task E. To review the emission models available for developing emission data for studies of the type to be carried out in RAPS. As noted, the full value of these tasks is only realized if they are allowed to interact, and a failure to recognize this until late in the project led to difficulties in formulating and presenting our conclu- sions. In the final form, summarized below, the findings of the studies made under Tasks D and E have determined, to a certain degree, the con- tent of Task B, which in turn reflects certain constraints and opportu- nities of the data handling system described in Task C. ------- C. Organization of the Report The first part of this report, the Summary, is an overview of the project that summarizes the results. The second and largest part includes sections on the five separate tasks, each section reporting the appropri- ate detailed results. Appendix A is an assembly of data in printout form that pertains to Task D, and Appendix B is a lengthy summary of emission model reports. ------- I I SUMMARY A. Basic Concepts The requirements of the RAPS program are wide ranging, diverse, and, in many cases, stipulate degrees of spatial and temporal resolution that have not hitherto been achieved. For the multiple purposes of the RAPS program, it is thus more appropriate to consider data on pollutant emis- sions in terms of a system of data resources than in terms of a single all-serving emission inventory. Alternatively, one can look upon an emission inventory for RAPS as being a multifaceted data source that comprises a number of elements, which are capable of providing needed information as required. In either case, we are led to a system in which varying criteria are applied to determining the nature and extent of the inventory content. A major distinction can be drawn between the requirements of RAPS for high resolution data (in space and time), for such purposes as the verification and development of air quality models, and the more general needs of the program, such as, for example, investigations of economic, social, or political processes, or the investigation of control or abate- ment strategies not immediately served by the air quality modeling studies. This distinction leads immediately to a concept in which a special high resolution inventory (for pollutants of significance) is provided to the air quality modeling studies and other investigations (such as transpor- tation processes or improved inventory methodology) where such high res- olution data is appropriate. Those pollutants are S00, NO , CO, and ^ !X hydrocarbons (HCs). For other pollutants, the high costs and difficul- ties of providing such high resolution emission data on a routine basis are not justified. For such other pollutants, we recommend that the National Emissions Data System (NEDS) Inventory be used. When special information on emissions is required—for example, for extra precise high resolution data for limited 'maximum effort" periods of model validation or for intensive transformation process investiga- tions—we recommend that each such emission data collected be treated as a special case. This will avoid placing excessive requirements on the basic emission inventories and save cost and effort. ------- It is convenient, therefore, in speaking of the RAPS emission inven- tory, to consider it primarily as the specially developed, high resolution data source concerned initially with SC> , NO , CO, and HC. This inventory, A A as noted, must be flexible and capable of serving many purposes. To ensure this, it is necessary to retain the basic or prime data from which it is built in its compilations in order to make revisions or extensions readily possible. Data should not be lumped in a form that applies to a single, or narrowly perceived, requirement. The most satisfactory approach, wherever possible, is to provide files of emissions data for each pollutant in terms of direct statements of weight of pollutant emitted by each source or class of sources as a function of point or area location for every hour. Although this is a substantial undertaking, as described below, in most cases it can be ac- complished to a useful level of precision within the RAPS program. This is mainly because the bulk of the emissions is a result of a limited num- ber of sources. In a program of the scope of RAPS, it is possible to treat such sources individually and to use measured data providing direct or closely related information on emissions as the basis for hourly emis- sion estimates. The frequency and quality of the measurements will be proportional to the magnitude of each source. This approach differs from that commonly used in developing emission data for purposes similar to the RAPS main objectives, which has perforce relied heavily upon emis- sion modeling techniques to interpolate from gross annual data to provide information with high nominal temporal resolution. In the proposed RAPS inventory, such modeling techniques would only be applied to estimating hourly emissions from minor point sources or from area sources, although in certain of these cases it is recommended that the required precision demands the development of improved modeling methods for providing hourly data. In the case of mobile sources, however, where considerable progress has been made in assessing emissions with high spatial and temporal reso- lution, the current modeling approach is suitable for the RAPS inventory; the records of mobile sources to be incorporated in the inventory have a special form. The term "emission modeling" sometimes leads to confusion, and it is necessary to clarify its significance. Strictly speaking, any compu- tational manipulation of raw data can be described as modeling, for ex- ample, the combination of pollutant concentrations and flow rates to pro- vide emission estimates. More often, modeling applies to the derivation or estimations of emissions from limited prime data, such as the estima- tion of emissions from residential space heating on an area basis (from a consideration of population density and daily temperature). In partic- ular, the apportionment of emissions from any source or class of sources ------- on, say, an hourly basis, from data such as gross annual tonnage of fuel burnt, is commonly performed by modeling using expressions (which can be more or less arbitrary) to describe the patterns of variability that oc- cur by season, by day of the week, or by hour of the day. In a more comprehensive sense, various emission models have been developed for specific purposes. These may consist of series of sub- elements integrated into a single system for deriving required emissions data. Such models are generally intended to provide emissions data in a specified form suitable for a particular use, deriving their estimates from relatively limited prime data (usually that relating to gross annual fuel consumption). A typical example is the Argonne Hourly SC>2 Model. This model, described in detail in Section VII, provides hourly estimates of SOg and thermal emissions from power plants, industrial point sources, and one-square-mile areas of residential, commercial, and institutional sources. It derives such estimates from a range of input parameters based largely upon fuel consumption statistics as well as upon other factors such as temperature and time. In considering such comprehensive emission models, it is important to emphasize their limitations for the purposes of RAPS. Adequate though they may be for the uses for which they were developed, they reflect the expediency and economy that was appropriate to such use. They seek to provide the best emissions estimates possible from available (i.e., mini- mal) data. They are not the best approach to developing emission inven- tories if better data can be specially acquired--as it can in the case of RAPS. Another aspect of emission models is the way that the specific output requirements determine the manner of developing the emission esti- mates from the prime data. This generally results in a lack of the flex- ibility needed to meet the multiple and varied demands of RAPS. The RAPS Emission Inventory should make emissions information available in such a manner that any desired selection of data may be extracted and presented in any desired form. This requirement is met by providing emission data in the form noted above, i.e., as direct statements of weight of pollutant emitted by each source, or class of sources as a function of location for every hour. In this approach, the emission modeling techniques, where used, are limited to the production of a common database, i.e., the gen- eral purpose high resolution emission inventory. The task of extracting emissions data for any purpose and of presenting it in any appropriate form then becomes a separate function. It is recommended, therefore, that the RAPS inventory be developed upon the basis of specially collected data, rather than upon current emission models (with the exception of mobile source data). Further, it is recommended that the inventory provides flexibility and universality ------- by being a simple compilation of all required data in the lowest common denominator form, i.e.. hourly emissions by weight of each pollutant (SO , NO , HC, and CO). £ X For other pollutants, it is recommended that the National Emissions Data System (NEDS) be used. B. Results of Specific Tasks 1. Task A To determine the potential users and uses of the emission in- ventory for RAPS, a survey was made. Users range from the research groups directly involved in the RAPS program to city, county, and other planning organizations and air pollution control agencies in the area. Research programs other than RAPS were also considered. The principal uses of emissions data will be in the verification and development of transport and diffusion models and in studies of chemical transformation processes, the main interest of the RAPS program. Other uses will be in the analysis of economic, social, and political impact of air pollution and its control and abatement and in the development of such control and abatement strat- egies (particularly episode control). The RAPS emissions data could also be useful to programs such as the Community Health and Environmental Sur- veillance System (CHESS) or research studies such as METROMEX (a joint investigation of meteorological effects of air pollution) or other locally based investigations requiring detailed knowledge of pollutant emissions. Two factors emerge from this survey. First, the requirements for these purposes are multiple and diverse, covering a wide variety of pollutants and very different degrees of temporal and spatial resolution. Second, the needs of the air quality modelers are the most demanding, and they dominate the requirements for high resolution data in space and time. The needs for emissions data in the RAPS program can, thus, best be met routinely by • Relying on NEDS for data on the broad range of pollutants for which high resolution information is not essential. • Developing a special RAPS Emission Inventory capable of providing the high resolution data needed by the modelers and of serving other purposes. For special purposes, such as particular experiments to inves- tigate chemical transformation processes or studies of the dispersion of 8 ------- specific pollutants for which the above mentioned inventories are inade- quate, appropriate special data collections should be made to determine emissions on an experimental basis. (Such data may be collected, pro- cessed, and stored within the KAPS Emission Inventory data handling sys- tem, but are not an integral part of this inventory.) The specification of various air-quality model input require- ments that act as common denominators thus determines what, is required of the RAPS Emission Inventory. Specifically, the requirements are • Pollutants: SO , CO, NO , hydrocarbons (by types), and 2 x particulates. • Resolution: Temporal, hourly, for each hour; spatial, 0.01 km point sources, 1 km grid squares [area sources using Universal Transverse Mercator (UTM) coordinates in each case]. • Area covered: The Saint Louis Air Quality Control Ken ion (AQC'K). • Period i. overed: Continuous hour by hour throughout the years of full RAPS operations. • Units: Emissions by metric weight; distance in kilo- meters, in UTM coordinates. • Output iormat; Variable, but in l-OKTRAX compatible form. • Other information: Data on sources (e.g., stack height, exit temperature, velocity) where appropriate; data for mobile sources, traffic flow data, and aircraft movements as required for tise in appropriate emission models. Jn this task (specification of the content of the RAPS Emission Inventory) an extensive analysis was made of the problems of developing emission data with the necessary spatial and temporal resolution and ac- curacy to meet the requirements of RAPS. After an initial review of the basic form the inventory system should take to meet the multiple purposes of RAPS, we discussed each aspect of such a system in detail. Because of the need for precision in estimating the emission and the requirements for spatial and temporal resolution, a quantitative ------- basis can only be provided by analyses made for specific air quality or dispersion models (as discussed in Section VII). The emission inventory can potentially provide the most complete data input to such models, depending upon the cost and effort applied, although even in the most careful direct assessments of emissions at individual sites, uncertain- ties of ±10 percent must be expected. We described the pollutants of interest, distinguishing between the reactive gases, nonreactive gases, and particulates. The role of the emission inventory of such pollutants is discussed in the context of RAPS. While data can fairly readily be obtained for the key gaseous pollutants, particulate emissions are ex- tremely difficult to monitor. Existing data on such emibsions are often inaccurate and give no information on the size distribution of the parti- cles, which is so important in the modeling context. In the circumstances, it is proposed that particulates not be included in the high resolution ] n\ < MI t orv on .in hourly basis, and that I he needs •>[' RAPS for parliculate data for dispersion model studies or research on transformation processes be dealt with on a special experimental or project basis—including, spe- cifically, the use of tracer techniques. The study team described and analyzed sources categorized ac- cording to identity, function, and pollutant emitted. Each category is discussed in detail, necessary for the data collection procedures to be recommended later. The importance of emission and data conversion factors is then noted. Such factors are critical to any inventory and especially to an inventory of the type needed for RAPS. The economics of the problem, however, are such that, while it would be desirable to test every size- able source in the inventory on an individual basis, wherever reasonable confidence can be placed in standard emission factors, these should be used in practice. Finally, the physical characteristics of the sources needed for the inventory are noted, and recommendations are made regard- ing the metric units to be used. The second part of Task B was to specify the RAPS Emission In- ventory system outlined below. While NEDS is proposed as a source of information on annual or long-term emissions from a broad range of pollutants, it is recommended that a special, high-resolution RAPS Emission Inventory be developed to provide hourly data emissions of SOg, CO, NOX, and HC in the form of direct statements of weight of pollutant emitted in kilograms each hour from each point or line source or area element for the Saint Louis AQCR. Locational data will be provided in UTM coordinates with a resolution of at least 0.1 km for point sources. Mobile sources will be specified in terms of traffic links (with a resolution of approximately 0.2 km) for major routes and in areas for secondary routes. Both for stationary and 10 ------- mobile sources, the area elements will reflect either the spatial resolu- tion of the basic data from which the emissions are estimated (e.g., hous- ing developments) or one kilometer grid squares, whichever is the smaller. Information on the physical characteristics of point sources and the prime data used in compiling the inventory would also be provided. It is not considered practicable to develop comparable data for participates. For initial planning purposes we propose that only point sources emitting over 100 tons per year of each pollutant be considered individu- ally. The estimation oi emissions from such sources will be accomplished in classes according to the magnitude of each source. In general, we rec- ommend that emission data be based upon direct monitoring of emissions for the largest sources. For other major sources, direc! monitoring of hourly fuel consumption, operating, or process data should be used as the basis for estimating emissions. For smaller point sources, we propose that emissions be estimated on the basis of generalized patterns of tem- poral variation of emission as a function of source type applied to the best available data on fuel consumption, operating, or process data (per shift, daily, or weekly). This approach approximates current practice in which annual totals of such data are used as a basis for modeling. For area sources, we recommend the use of modifications and adaptions of already available emission models. Similarly, for a first approach to assessing mobile source emissions, we recommend trie use of available models. Improved mobile source models should be employed, as they become available, to provide higher quality data. A simple new model should be developed to estimate natural emissions of nonmethane hydrocarbons on summer days. in all cases, ue propose thai, special testing efforts be carried out to ensure that the emission and data conversion factors used to derive the emission estimates are as reliable as possible, and have regard for the high temporal resolution required. Standard emission fac- tors should only be employed if a review of the source on a case-by-case basis indicates that such a factor may be used with confidence. The concluding portion of Task B is a description of methodology and detailed procedures lor developing the inventory in the form proposed. Point sources using the NEDS dam are classified according to size to faciii iale (he description ol how emission data would be derived in four bio.id grounh accoichng 1 <,> the ija^ic records used to estimate emissions— Group I, continuous emissions monitoring records; Group II, continuous records of (ueL consumption, operating, or process data; Croup III, short term, periodic records of fuel consumption, operating, or process data (per shift, daily, or weekly); Group IV, long-term (semiannual or annual) records or estimates of iuel consumption, operating, or process data. Within Groups II and 111, a further distinction would be made between sources for which separate conversion or emission lac tors are established LI ------- and those for which representative factors will be established. Data from Groups I, II, and III will be collected on an ongoing basis. In principle, it would obviously be desirable to deal with all sources under Group I. In practice, sources should be treated in the highest ranking group possible within limits of cost and difficulty. Some 95 percent of the S02 emissions and 70 percent of the NO emissions can be derived from direct data (Groups I, II, and III) on the basis of routine data inputs from 101 sources with a high confidence based upon the testing of 65 individual sources. Similarly, in the case of CO sources, over 90 percent of the emissions can be estimated on an hourly basis by the treatment of direct data from only six sources. On the other hand, it would be necessary to obtain data from and test some ten or more sources to achieve hourly estimates from direct data for only 50 percent of the HC emissions. The proposed approach, however, is in- tended to provide considerable flexibility, so that optimum results can be achieved within budgetary and other limitations. Thus, the number of sources to be treated on an individual basis and the numbers of tests to be made should be decided upon the basis of an initial survey on a case- by-case basis. In the case of area sources, the shortcomings of existing emis- sion models require that special efforts be made to improve and verify these, for, although such sources contribute only a small proportion of the total burden, their local effect is significant. Similarly, in the case of mobile sources, while a simple form of existing models will suf- fice initially (with improved traffic data inputs), the products of more sophisticated models should be used as they become available. Finally, it is noted that a method for estimating emissions of nonmethane hydro- carbons from natural sources on summer days is needed. Task B includes a tentative schedule aimed at providing initial data on SO2 and NOX by the first quarter of the second year of full activ- ities, complete data by the third quarter of that year; effective initial data on CO and HC would become available by the end of the second year, and complete data would be available by the third quarter of the second year. 3. Task C Task C is the specification of a system for handling, storing, and processing the relevant data. Specification is based upon certain assumptions regarding the nature and capability of the data management system that will be available as part of the RAPS overall data handling 12 ------- system. These assumptions either indicate facilities that are normally provided by such a system or could be readily provided by appropriate adaptation or additional programming of a conventional type. In order to specify the data formats and structures, it is necessary to determine how the inventory files will be used as a system for source data preparation and entry, for processing data through emis- sion models, and finally, for production of the data files for the vari- ous user activities. The system is designed to minimize manual transfor- mations of the raw data to the form in which it is entered into the computer. Input data are thus based upon the sources, and all pertinent data (e.g., pollutants, fuels, consumption) are entered in direct form as reported. Rather than using integer codes such as the SAROAD manual, we propose the use of direct, mnemonic identifiers. We also specify the use of a hierarchical, list type structure to enter such data and similar necessary information. This leads to data structures that comprise ex- tendable data items within an appropriate data reference system allowing maximum flexibility both to accommodate disparate types of data and to allow ready extension to cope with new or additional data forms that might need to be added after the establishment of basic lists or thesauri. The master inventory file will consist of data items, each of which describes a source or a set of identical sources. Each item will include source identification data, source operating parameters, the set of primary measurements from which emissions will be computed, and the set of computed hourly emission values for each pertinent pollutant. This inventory file is desU.iu.vi as the core of the whole inventory sys- tem. It will be supported and maintained bj a gt-neral-purpose data management system and by such special-purpose programs as may be required for verification, emission computations, and production of data files properly formatted for use a.s input to air quality models. 4, Task JD Task D, i"he review of the existing emission data sources in the Saint L-,;uis area, has been disappointing. Despite the fact that Saint Louis ha:- beta the site of a number of advanced air quality studies and has active and well-organi'/ed local air pollution control agencies, the available emissions inventorv data is quite inadequate for the purposes of RAPS. Although the emissions data that have been collected in the area place it in the forefront of urban areas in the availability of such information and the existing data fairly well meets (or met) the require- ments for which it was collected, it falls far short of the new and highly demanding requirements of RAPS. Spec!fica11v, the data do not provide for hourlv resolution. ------- The five major inventories considered are the IPP (1968), IBM (1970), DAQED (1971), NATO (1971), and NEDS (1973) inventories. Of these, the first four are now essentially of historical interest only. The NEDS inventory contains the best current estimates and constitutes a valuable source of general data for the broad requirements of RAPS. For the prin- cipal purposes of RAPS (model verification and development and study of atmospheric transformation processes), NEDS will serve as a basis for developing the high-resolution RAPS Emission Inventory necessary to meet the requirements of the modelers and others. Thus, NEDS provides good information on point sources that is of use in planning and implementing the RAPS inventory. In addition, it provides basic inputs on area and mobile sources for the advanced pur- poses of that inventory. 5. Task E In this task, 28 existing emission models relevant to the ob- jectives of RAPS were reviewed in detail. In \ppendix B, complete infor- mal i ------- gross, and much reliance is placed upon extensive interpolation procedures. Since these modeling procedures are clearly of limited precision for use in the context of RAPS requirements and since, in the absence of extensive verification, confidence in their use must be guarded, it was not possible to recommend the use of such models unreservedly for the RAPS program. Certain models appear to have potential utility for the purposes of RAPS depending upon the degree to which interpolation procedures will be needed (as distinct from the preferable procedure of deriving emissions directly from time resolved related data). These models are noted below. In addition to the review of existing models, a brief analysis is also made of the nature of and need for verification procedures. In particular, a procedure is set out for evaluating the validity of mobile source models by combining specially collected experimental data on traf- fic volume with basic statistical data. Another point addressed in detail is the degree of resolution required for the specification of location of point sources. The inter- play of source strength, stack height, lateral and longitudinal receptor position, and atmospheric stability makes the definition of locational precision complicated and complex. Accordingly, a new methodology is developed to enable the varying interrelated factors to be considered for any given circumstances, enabling the effects of positional error to be assessed for any downwind distance or distance interval, and pro- viding a recommended spatial resolution for each source specification based upon stack height and emission rate. Finally, the role of natural mechanisms in adding significantly to the man-made pollutant burden of the Saint Louis region is considered. The specific conclusions that result from the studies carried out under Task E are listed here. i • No existing models adequately meet all the requirements of RAPS. • To provide emissions data with the necessary high resolu- tion in space and time, it is recommended that - Direct information of emissions or of factors deter- mining emissions be acquired to the extent possible. Where this is not possible (e.g., for small point sources and area sources) the most suitable of the existing models should be used in adapted and improved form. 15 ------- • Specifically, for stationary sources Direct data from emissions monitoring should be used wherever available (probably only for the largest sources, such as power stations). Information on hourly fuel consumption rates, process operations, and so forth should be obtained and used to derive emissions for other large sources. - Modeling procedures for small point sources and area sources, based upon the models noted below, should be used, but, with refinement of the input data wherever possible. - Recommended models are the Argonne model (for hourly SC>2 from distributed residential, commercial, and in- stitutional sources; and hourly SC>2 and heat from major point sources) and the Systems Applications, Inc. (SAI) model1'3 (for hourly NOV and HC from stationary point .A. and area sources). For automotive mobile sources, we recommend that an average route speed model be used with a link or line source geometry supplemented with measured data of traf- fic flow from fixed sensors on high volume freeways and on selected arterials. - The recommended models are the SAI model1'3 (for hourly CO, NC> and HC with modified inputs derived from Stanford j\. Research Institute's model4 for spatial and temporal dis- tribution of vehicle number and speed on a link basis for primary traffic and area basis for secondary traffic). - For other mobile sources, the recommended models are the Geomet model (for diurnal emissions from river vessels and railroads) and the Northern Research and Engineering model as revised by Geomet (in preparation, for aircraft emissions). For both stationary and mobile sources, the Ontario De- partment of the Environment 1971 Model for Toronto should be referred to for overall guidance and planning method- ology. • For microscale studies under the RAPS program, the basic inventory will need to be supplemented with special emis- sion inputs. These should be obtained on an experimental or project basis (e.g., the acquisition of detailed data 16 ------- from small stationary sources or for mobile sources, con- tinuous and detailed monitoring of vehicle behavior in a given area with the application of a multimodal emission model to forecast emissions). In view of the limitations of existing mobile emission source models and the impossibility of resorting to direct information, it is recommended that special steps be taken to improve and verify such mobile emission models. Spe- cifically, - An experimental evaluation should be made of mobile emissions models to refine the specification of process input parameters and the accuracy of postulated input- output relationships. Parameterization of inputs for mobile emissions (statis- tical database plus continuous measurement of vehicle volume, speed, and mix on key links to generate a dy- namic link-by-link inventory) should be verified on a selective basis via aerial photography and/or side- looking aerial radar. - Parameterization of outputs from mobile sources should be evaluated in various models to verify representation of time-averaged emissions on a per-link basis taking into account types of link and the impacts of deteriora- tion, nix, road grade, drag, and so forth. Of the pollutant emissions (CO, NOV, S00, H0S, nonmethane x ^ £ HC, and particulates) from natural phenomena in the Saint Louis area, nonmethane hydrocarbons may contribute a sig- nificant background during summer daylight hours and should be included in the RAPS emission inventory. Suitable proce- dures to estimate these should be developed early in the RAPS program. In the context of the RAPS program, the accuracy of point source locations is best considered in terms of the effect of errors throughout the study area. Accordingly, we pro- pose that a measure of error in source location be defined as the integral value of the error over both the lateral (crosswind) and longitudinal (downwind) directions. 17 ------- C. Recommendations The following recommendations provide an outline of an operational plan to derive the RAPS Emission Inventory. • An initial survey should be made of point sources emitting over 100 tons per year of S00, NO , CO, or HC. The survey should be £ X based upon the information of the NEDS inventory, and as many as possible of the sources should be inspected. An initial assessment, based upon the nature and quality of the records available, should be made on how each source should be treated by group (according to the data collection procedure). • With regard for budget and the results of the survey, detailed plans should be made as to the number of sources to be treated in each group, especially the number of individual source tests to be made. • Meanwhile, the proposed data handling system should be developed and coordinated with the specific data management system actually procured for the RAPS project. • As soon as possible, effort should be applied to the development and refinement of emission modeling procedures to meet the needs of the RAPS inventory. Specifically, improved techniques of handling shorter term input data for both point source and area sources should be applied to the Argonne model and the SAI model. For mobile sources, similar improvements should be made on the basis of the SAI model with modified inputs derived from the SRI model. A model must be developed to estimate summer day emissions of nonmethane hydrocarbons from natural sources. • Following the detailed plan developed from the initial survey, an intensive program of source testing should be undertaken to develop a reliable basis of emission factors or conversion fac- tors to relate the routinely collected input data to the emis- sions of specified pollutant on an hourly basis. • Arrangements must be made for the routine periodic collection of input data from the sources to be treated on that basis or additional data for modeling purposes not acquired in the source of the initial survey. Arrangements should also be made to in- stall any necessary monitoring equipment provided under the RAPS program for point sources or measuring traffic flow on selected high volume freeways or arterials. 18 ------- Finally, the system should be operated on a routine basis follow- ing the sequence noted in Task !; to provide data first on S()( and \O and later <», CO and 11C. x In support of (he inventory development effort, tin1 follouin<; inputs should he sought from otlur elements ot HAPS ,is e.irlv as poss i h i e; - Tests •)[ specific air qualit\ or d i spi >rs i on modi-Is should be wad e to ascertain '-ens i t i \r i t \ ti> the pri-c i s i on ruid reso- lution 01 emissions data, in (he Saint Louis ati',. (usit,^ \i.i)h 'ia t a a ;: a ba s j s ) . - The \'icinitv oi identilied so that consid- eration can b( 5'iven to t i'ea , i :ii; tlu>ni on an individual basis even though they do noi em 11 ovc.'r 100 tons of pollutants prr veai . ------- Ill TASK A: DEFINITION OF POTENTIAL USERS AND USES A. General An emission inventory is a fundamental part of any air pollution study. The Regional Air Pollution Study (RAPS) in Saint Louis is ex- pected to be unique in scope and depth. Consequently, the emission inventory contemplated for this study needs to be far more detailed and accurate than those currently available. The actual scope, content, and format of the RAPS inventory is de- termined by its proposed use, and, in order to ensure that the inventory will correspond as closely as possible to the demands placed on it by future users, a survey was made to determine the characteristics desired by potential users both within EPA and elsewhere. Primary consideration was given to RAPS requirements. To complete the picture, the needs of other potential users were then considered. B. Summary Although an emission inventory is a key information base for all air pollution studies and control strategies, it is of direct use to RAPS investigators concerned with the relationship between ambient concentrations of pollutants and their source, that is, the modelers. Other groups directly concerned with emission inventories are (1) the regulatory agencies charged with the achievement and maintenance of the National Air Standards (this group has the responsibility and authority to collect and store emission data) and (2) other workers in the field concerned with transportation problems, health effects, plant damage, and the like (these groups are primarily concerned with ambient concen- trations of pollutants of interest). The requirements of these users are quite different. The modelers are concerned primarily with those pollutants for which air quality standards have been promulgated, that is, sulfur dioxide, carbon monox- ide, hydrocarbons, nitrogen dioxide, and suspended particulates. The information they require is very detailed (accurate location of the sources, detailed information on emissions as a function of time, and precise estimates of quantities emitted are all essential). The period 21 ------- for which such high resolution data is needed will vary considerably. In many cases, model verifications will be limited to relatively short periods, of, say, 48 to 72 hours, selected for representative meteoro- logical conditions. Since such periods may not be readily determinable until a later time, when the best combination of circumstances is deter- mined, post facto, it is necessary to collect appropriate emissions data on a continuing basis, parallel with the even more comprehensive contin- uous measurements of air quality and meteorological factors. The requirements of the people working with health effects or plant damage, on the other hand, are quite different. The range of pollutants of interest is very broad, including not only compounds of known toxic effects such as mercury or cadmium but also an almost infinite variety of compounds suspected of having an effect on plants or people. For these purposes, detailed time resolution of emission is seldom required (except in cases of accidents or unusual meteorological events) since most effects are cumulative over long periods of time. Therefore, long- time historical records on many substances are of main importance in de- termining causative relationships of air pollution to health or vegeta- tion effects. State and local regulatory agencies are in a sense caught between these two requirements. They have to maintain inventories adequate for the enforcement of legal standards, and they would like to have records of emissions that are potentially hazardous. As RAPS may eventually include a number of investigations that cur- rently are not well defined, it is essential to include in the inventory design the flexibility required for the addition of other pollutants. For the moment, however, the primary users of the RAPS emission inventory are the modelers, and it is their requirements that have to be met. In the following section the details of these requirements are delineated. Table 1 summarizes the requirements. C. RAPS Requirements The Saint Louis emission inventory is of prime concern to the RAPS research groups engaged in the development, evaluation, and refinement of air pollution models. Several types of models are currently under consideration or active study as part of one RAPS study plan;7 in this plan, the four basic model types are emission, atmospheric structure, transformation, and removal process models. Also being considered are existing models that encompass one or more of the basic models, such as some models that treat the generation and dispersion of pollutants from certain transportation sources. 22 ------- g. fr s t> 3 a 8 S 3 « M X X X X X K X £ S Hazardou Material 1 C ^ ydr bon ydr rbo Gro XX XX tu oc Agency aint Louis is County State APC Air Contro n r f L i gul Cit Sai Ill Mis Com 23 ------- Upon examination of the four basic models, it can be seen that only the emission model is self-contained. The others exhibit various degrees of interdependence. For example, a model of boundary layer structure will need to account for the meteorological effects that result from emissions of, say, heat and water vapor. Similarly, transformation models must consider both emissions and atmospheric structure models, while removal process models consider all three models. Thus, in assess- ing the requirements of the various specific models, we need to recognize that, while the primary emphasis of the particular model may be one thing, it is quite likely and indeed probable that many of its input require- ments derive from the structure of the other model components. Up to now little effort has been devoted to the development of com- prehensive or nonspecialized emission models. In fact, the development of a framework for a RAPS emission model program is an objective of this study. In the absence of an established emission model that the emis- sion inventory would serve, we have surveyed the various model require- ments for emission inputs so as to specify criteria and objectives for such an inventory. These models and their requirements are discussed below. 1. Atmospheric Structure Models Atmospheric structure models describing transport and diffu- sion within the planetary layer on a regional scale (<30 km) are being developed at the Model Development Branch, Meteorology Laboratory, NERC, Research Triangle Park, North Carolina. The specific requirements re- lated to emissions are shown in Table 2. The model is currently under development, and, hence, these requirements reflect our best judgment based on available information. Three additional models used at the Meteorology Laboratory of NERC are • PTMAX, an interactive program that performs an analysis of the maximum, short-term concentration from a point source as a function of stability and wind speed. • PTDIS, an interactive program that computes short-term concentrations downwind from a point source at distances specified by the user. • PTMTP, an interactive program that computes, at multiple receptors, short-term concentrations resulting from mul- tiple point sources. 24 ------- Table 2 EMISSION INVENTORY INPUT REQUIREMENTS FOR ATMOSPHERIC STRUCTURE (WATPIC) MODEL Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence SO , CO, particulates, heat, water vapor £ 1 hour Point sources on ITM grid (.10.«L km) Area sources on variable1 grid from 1 km" Tape Grams/second, watts Stack height (meters), exit velocity (meters/second), temperature (°K), stack diameter (meters) ' Diurnal, daily, seasonal Depends on source of data (stack sampling, questionnaires) We believe that the emission inventory input requirements of these models are met by the specifications in Table 2. However, when the manuals for these models become available, we will review those require- ments. 2. Transformation Processes Models .vhiny materials emitted into the atmosphere undergo not only physical dispersion but chemical changes as well. We need to know the whole life history of a pollutant, from emission through chemical and physical transformation to eventual removal, before accurately predict- ing the relationship of emission to air quality. At present, the fate of six major pollutants is under investi- gation by modelers at Research Triangle Park and outside contractors under the guidance of the Chemistry and Physics Laboratory. The pollutants are 25 ------- sulfur oxides, nitrogen oxides, hydrocarbons, oxidant, carbon monoxide, and particulates. The photochemical transformations have received the most attention so far. The emission inventory inputs into this model are listed in Table 3. Table 3 EMISSION INVENTORY INPUT REQUIREMENTS FOR THE TRANSFORMATION PROCESSES* MODEL Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence S0 NO v,, iiu , GO, CO, HC, particulates (NO by com- & X O A pounds; HC by categories, e.g., paraffins, ole- fins, aromatics) 1 hour Point sources on UTM grid. Line sources (end points) in UTM coordinates. Area sources on 2 variable-size grid from 1 km . Eulerian models utilize a fixed rectilinear grid, while La- grangian models may require a flexible grid aligned along a given wind trajectory. Tape Grams, meters, seconds See Table 1 Diurnal, daily, seasonal Depends on source data Principal candidate-model EPA contractors: 1. Systems Applications, Inc. Beverly Hills, California 90202 2. General Research Corporation Santa Barbara, California 93105 3. Pacific Environmental Services, Inc. Santa Monica, California 90403 4. Systems, Science & Software, Inc. La Jolla, California 92037 26 ------- 3. Mobile Sources a. Surface Vehicles Emissions from mobile sources such as automobiles, trucks, railroads, and vessels constitute an important segment of urban pollution. They account for virtually all of the carbon monoxide, the bulk of hydro- carbons, and a significant share of nitrogen oxides. Since emissions from these sources are a function of the speed, number, and type of vehicles and the traffic patterns for given highway segments, they are best represented as line sources for heavily traveled streets. Lightly traveled areas can be considered area sources. Emission inventory inputs into transportation-related models will require items indicated in Table 4. Emission requirements of models that treat vehicular sources have been limited to the require- ments of mesoscale transportation modes. Microscale models—those treat- ing near-source effects—may have more stringent emission requirements both in quantity, types, and precision of the data. Consequently, these Table 4 EMISSION INVENTORY INPUT REQUIREMENTS FOR VEHICULAR SOURCES MODEL Content of inventory Time resolution Spatial resolution Preferred medium Uni ts Source descriptors Temporal variations Level of confidence CO, hydrocarbons (by types such as olefins, paraffins, aromatics), NO , particulates x 1 hour Street segments. Endpoints (in UTM coordi- nates), width (meters), center strip width (meters). Type of roadway at grade, elevated, o cut section, street canyon. 1 km (minimum) blocks for area sources. Tape Grams/second, meter Mean speed (meter/second), traffic volume by mix Diurnal, daily, seasonal Depends on source data (sampling, traffic counts, simulated traffic patterns, and the like) 27 ------- types of highly specialized requirements are not within the scope of the basic inventory and will need to be obtained as a special task in conjunc- tion with the particular microscale model experiments and evaluations. b. Aircraft and Airports Models dealing with pollutant emissions from planes on the ground level and above airports are also being developed. These are essentially microscale models dealing with the movement of individual aircraft during landings, taxiing, and takeoff. The degree of detail of emission data required for such a model (Table 5) is probably outside the scope of a regional inventory. This work is coordinated by the Model Applications Branch of the Meteorology Laboratory, NERC, Research Triangle Park. Table 5 EMISSION INVENTORY INPUT REQUIREMENTS FOR MOBILE SOURCES (AIRPORTS) MODEL Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence CO, hydrocarbons, NO , particulates X Every flight: taxiing time, takeoff and landing Each runway, taxiway, gate, direction of takeoff Tape Grams/second, meter Type of plane c. Railroads and Vessels The contribution of railroads and vessels to the air pol- lution burden of the Saint Louis area is under investigation by the Trans- portation Center, Department of Transportation, Cambridge, Massachusetts. 28 ------- The Center is conducting a study that will include a sur- vey, an experimental program to obtain emission data and some dispersion modeling studies. In the course of this work, the Center will be both a user of and a contributor to the RAPS emission inventory. The main pollutants of concern are SO and particulates. 4. Other Studies a. Particulate Emissions Visibility loss caused by suspended fine particles is one of the most obvious manifestations of air pollution. Particles also may be involved in gas phase reactions and provide one of the mechanisms for the removal of nitrogen and sulfur oxides. Thus, the formation and re- moval of fine particles is part of our understanding of atmospheric re- actions and will ultimately have to be accounted for in models describing such reactions. Particles also play an important role in health studies, since those fine enough to remain suspended for indefinite periods (2 to 3 microns) are breathed in and can deposit in the lungs. Thus, there is a particular interest in this fraction of particulate emissions. Studies are under way under the auspices of the Chemistry and Physics Laboratory, NEHC-KTP, in the Saint Louis area to measure par- ticle size distribution and their chemical analysis. Data on the emission of particles are needed, including data on size distribution of major sources. At the very least,, da+a on the coarse (greater than 3pJ and fine (less than 3^) fractions should be obtained. b. Various Analyses In addition to the modeling studies dealing with the phys- ical characteristics of air pollution, such as emissions, air movement, or ambient concentrations of pollutants, the RAPS data are expected to be used for a number of studies arid analyses dealing with the social, economic, and political impacts of air pollution and its control. Most of these studies will be based on long-term averages, and they do not require the fine resolution the RAPS inventory will provide. Still, they will profit from the increased accuracy of the new inventory. 29 ------- The development of an inventory of the scope and detail of the Saint Louis study is in itself no mean achievement. On the basis of the experience gained in this study, guidelines for improved emission inventory procedures will no doubt be developed. D. State and Local Control Agencies The Metropolitan Saint Louis Interstate Air Quality Control Region comprises two states—Missouri and Illinois—and 12 counties. Five counties are in Missouri (Saint Louis City, Saint Louis, Saint Charles, Jefferson, and Franklin); seven are in Illinois (Bond, Clinton, Madison, Randolph, Saint Clair, and Washington). The Missouri counties are under the jurisdiction of the EPA Region VII office at Kansas City, Missouri. The Illinois counties are the responsibility of the EPA Region V office at Chicago, Illinois. Saint Louis City and Saint Louis County have sepa- rate air pollution control offices; the other Missouri counties are ad- ministered directly by the Missouri Air Conservation Commission at Jef- ferson City. The seven Illinois counties are under the jurisdiction of the Illinois Environmental Protection Agency. The area is shown in Fig- ure 1. The crosshatched areas are those covered more completely by mon- itoring stations. The ultimate purpose of the state and local control agencies is to achieve and maintain the air quality that corresponds to at least the national primary and secondary air standards, although states may estab- lish more stringent standards. Thus, control agencies are concerned not only with pollution episodes (emergency conditions), but also with long- term pollution control programs aimed at a gradual reduction of air pol- lution levels and with community growth planning as it relates to added pollution. Ultimately, they are responsible for the effect of pollution on the health of the community, its vegetation, and perhaps even weather modifications. 1. City of Saint Louis The Air Pollution Control Office, Department of Public Safety, City of Saint Louis, has a source evaluation section in charge of the emission inventory; it has identified some 16,000 point sources, but reliable data are available primarily for the large sources. The sec- tion's primary mission is enforcement rather than data collection. The emission inventory is in the form of reports gathered by inspectors. The data were not computerized. See Table 6. 30 ------- I I / \ PIKE L. 0 10 20 30 40 50 SCALE — km EMISSION RATE — tons/day • 10 - 100 of any single ± > 100 pollutant SA-1365-33 FIGURE 1 METROPOLITAN SAINT LOUIS INTERSTATE AIR QUALITY CONTROL REGION 31 ------- Table 6 EMISSION INVENTORY INPUT REQUIREMENTS OF THE AIR POLLUTION CONTROL OFFICE, CITY OF SAINT LOUIS Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence SO (HC, CO), NO , particulates ^ X Yearly average output Point sources, UTM coordinates Reports Engineering Stack height, exit velocity, temperature Inspectors' reports 2. Saint Louis County The Air Pollution Control Division, Saint Louis County Health Department, like its Saint Louis City counterpart, is primarily an enforce- ment agency. It has identified the major pollution sources (about five that are emitting more than 100 tons per day, about 20 with emissions in excess of 25 tons per day). Some modeling of sulfur dioxide concentra- tions has been done, and some CO modeling is contemplated. The inventory (Table 7) is based on inspectors' reports. The control of additional pollution resulting from the growth of the area through proper land use is under consideration. 3. Missouri Air Conservation Commission, Jefferson City, Missouri This agency has responsibility for the air quality of the state of Missouri. It operates through county or local air pollution control boards where such agencies exist, and it operates directly in other areas. Its function is to promulgate and enforce air pollution standards for the state of Missouri. 32 ------- Table 7 EMISSION INVENTORY INPUT REQUIREMENTS OF THE AIR POLLUTION CONTROL DIVISION, SAINT LOUIS COUNTY HEALTH DEPARTMENT Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence SO . NO , CO, EC, particulates 2 x Yearly averages Point sources on UTM coordinates Printout or tape Engineering Stack height, exit velocity, temperature None Major sources have been updated and some stack sampling is performed every six months At present, the commission maintains data on particulates, SO , id NOX, total hydrocarbons, and CO. The data are based on an original sur- vey carried out in 1968 (for particulates and 862 only), which has been updated in 1970 by IBM to include the other three pollutants. It was further updated in 1972 by Radian, Inc., and is now in National Emissions Data System (NEDS) format. Emission input requirements are similar to those shown in Table 7 for Saint Louis County. 4. Illinois Environmental Protection Agency, Division of Air Pollution Control, Springfield, Illinois The Illinois side of the Saint Louis Air Quality Control Region, which is heavily industrialized, is under the jurisdiction of the Illinois Environmental Protection Agency through its regional office (State Region IV, Collinsville) and the central office at Springfield. The Surveillance Section is in charge of the emission inventory, which is continually re- vised and enlarged. Currently it contains information on about 2,500 sources, but it is expected to reach 14,000 sources. The inventory is in the form of a computer printout, and it is used constantly as a guide 33 ------- for issuance of permits (surveillance), as a source of information to other agencies and interested parties, and as an aid in modeling. The modeling is contracted to Argonne National Laboratories, the State Im- plementation Plan is based on the Argonne work. Emission inputs are shown in Table 8. Table 8 EMISSION INVENTORY INPUT REQUIREMENTS OF THE ILLINOIS ENVIRONMENTAL PROTECTION AGENCY, AIR POLLUTION CONTROL DIVISION Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence SO , CO, NO , HC, particulates £ X Quarterly, seasonal Daily during air pollution episodes Point sources: UTM coordinates ±0.1 km Area sources by counties Printout, tape Engineering Stack height, exit velocity, temperature 5. Federal Highway Department, Department of Transportation, 400 - 7th Street, N.W., Washington, D.C. The Federal Highway Department, in cooperation with the Air Data Office of EPA, will conduct a study designed to provide information on driving patterns on highways in the Saint Louis area. This informa- tion, in combination with emission data based on different driving modes, will permit an accurate assignment of the contribution of automotive traffic to air pollution in Saint Louis. The emission inputs necessary for this study will have to be developed from dynamometer runs under simulated driving conditions for the pollutants of interest (CO, hydro- carbons, NO ). x 34 ------- E. The Planning Agencies 1. East-West Gateway Coordinating Council The East-West Gateway Coordinating Council (Laclede Gas Build- ing, Saint Louis, Missouri) is a coordinating agency created to facili- tate the land use and transportation planning for the Saint Louis metro- politan area. II helps to coordinate the planning efforts of the Missouri State Highway Commission, the Illinois Department of Transportation, and various county and city planning dep;i tments. It has also received sup- port from the U.S. Department of Housing and Urban Development and the Federal Urban Mass Transportation Administration. The council has produced a comprehensive Transportation Study Report analyzing present problems and suggesting long- and short-range solutions to transportation needs based on socioeconomic projections, analyses of present traffic patterns, and estimates of future travel demands. The council has generated an "Average Daily Traffic Volume Summary" for the Saint Louis metropolitan area (1970), which is updated yearly and constitutes a valuable source of information on mobile sources. The council, thus, is both a source and a user of data on mo- bile sources. Additional traffic data are available from • The Illinois Department of Transportation, 930 St. Clair Avenue, Fairview Heights, Illinois (Mr. Robert Kronst, District Engineer), • The Missouri State Highway Department, State Highway Building, Jefferson City, Missouri (Mr. Carl Klam, Highway Engineer). • The Saint Louis County Department of Highways, 120 N. Gay Street, Clayton, Missouri (Mr. Richard Daykin, Director). • The Saint Louis City Traffic Department, City Hall, Saint Louis, Missouri (Mr. Jim Bauman). 2. Industrial Waste Control Council The Industrial Waste Control Council is an association of pol- lution control officers from the major industrial firms in the Saint Louis area. As the problems they face are to some extent common to all of them, they share technical information and develop public relations. Their members are probably in the best position to develop emission 35 ------- inventories for their respective companies, inventories which they need to satisfy legal requirements. The data thus developed are generally confidential and available only to organizations empowered to receive them. F. Research Programs 1. Metromex Metromex is a joint program involving groups from Argonne National Laboratory, Battelle Pacific Northwest Laboratories, University of Chicago, Illinois State Water Survey, Stanford Research Institute, and the University of Wyoming. It is basically a federation of individual scientists with diverse sources of funding, but common or at least compli- mentary research interests in the inadvertent weather modifications pro- duced by large urban complexes through the interaction of their emission of pollutants and their contribution to the energy balance of the area. The study is centered in the Saint Louis area. Pollutants of interest to Metromex are primarily suspended particulate matter. Thus, major sources of particles, their physical properties and chemical analysis, and their size distribution are all parameters of interest. 2. Community Health and Environmental Surveillance System The Community Health and Environmental Surveillance System (CHESS) is a national program that relates community health to changing environmental quality. It consists of a series of standardized epidemi- ologic studies designed to measure simultaneously environmental quality and sensitive health indicators in sets of communities representing ex- posure gradients for common air pollutants. The program is conducted by EPA in cooperation with local public health agencies, universities, and private research institutes. The purpose of the CHESS program is • To evaluate existing environmental standards « To obtain health intelligence for new standards • To document health benefits of air pollution control. One of the areas under study is Saint Louis. As mentioned earlier, CHESS, like other health effect studies, requires primarily ambient concentration data rather than emission data. At present, it obtains such data from its own network of air monitoring 36 ------- stations sited within each CHESS community. Eventually, RAPS data may be used. Since historical data on ambient concentrations are generally not available, estimates of past pollutant concentrations have to be based on past emission data from which estimated ambient concentrations are derived via meteorological dispersion models. Thus, CHESS has some need for emission data. The CHESS program is administered by the Human Studies Laboratory at Research Triangle Park. The Bioenvironmental Mea- surements Branch is responsible for obtaining exposure data for the use of epidemiologists. Compounds of interest to these studies are shown in Table 9. Table 9 EMISSION INVENTORY INPUT REQUIREMENTS FOR CHESS STUDIES Content of inventory Time resolution Spatial resolution Preferred medium Units Source descriptors Temporal variations Level of confidence SO, NO particulates, especially sulfates, nitrates,and organics (size fractionated, if possible), hydrocarbons, Oo, CO, trace metals, e.g., Cd, Zn, Pb, Hg, Cr, V, Ni, Cu, Mn, as- bestos, benzo-a-pyrene, PCB, pollens For acute studies: 1 to 4 hours For chronic studies: cumulative exposures School districts Printout Gram/cent imeter/second Seasonal 3. University Research Individual researchers at several of the universities in the study area are engaged in work requiring knowledge of pollutant levels. For example, at Washington University, Dr. Richard Gardener is conducting 37 ------- work in the area of thermal pollution— heat islands '—and constructing an aerosol submodel in conjunction with other agencies. Dr. Christopher Hill is studying the total (mining, processing, use, and disposal) envi- ronmental impact of certain industries. At the Center for the Biology of Natural Systems, Dr. Charles Lee is establishing pollution burdens in rain and soil samples and in human tissues. At Saint Louis University, Dr. Slavin is studying the relation- ship of air pollution to health problems. There seem to be strong cor- relations between airborne sulfates and asthma attacks, between S0? and emphysema, and between carbon monoxide and heart attacks. His department is collecting a daily pollen count and could make it available to the RAPS inventory. At Southern Illinois University, Dr. Al Kahn is carrying out studies of the carbon monoxide and hemoglobin reactions. As mentioned earlier in this report, these activities have only a peripheral requirement for emission data at this point. However, to a certain extent, the format of the inventory should be designed with the potential to accommodate such data if they should become necessary. 4. Plant Damage There are a number of compounds in the atmosphere that cause vegetation damage either by reducing yield and growth or by affecting the quality of the crop. Well over 90 percent of the losses are caused by the following pollutants: • Ozone • Peroxyacetyl nitrate (PAN) • Oxides of nitrogen • Sulfur dioxide • Fluorides Out of 65 standard metropolitan statistical areas, the Saint Louis metro- politan area is rated by Benedict8 16th with respect to potential damage due to hydrocarbons, 24th with respect to oxides of nitrogen, and 18th with respect to sulfur dioxide on a scale that takes into consideration total emissions, meteorological factors, area, and stagnation periods. Thus, plant damage constitutes a significant aspect of air pollution in Saint Louis, something that should be taken into account in setting up the emission inventory. Data on oxidants—ozone, PAN, and 38 ------- oxides of nitrogen—and sulfur dioxide would normally be covered by the inventory. It may be advisable to make provisions for a fluoride inven- tory as well. The main known sources of fluorides are aluminum, phosphate, and ceramic plants. G. Conclusions The emission requirements data for RAPS are such that inventories taken for other purposes—such as the Implementation Planning Program (IPP) or air pollution control agencies—are inadequate mainly in the degree of detail required for modeling studies. Since the verification of atmospheric dispersion models is an early and major goal of RAPS, the requirements of the modelers constitute the most important criterion of the adequacy of the inventory. The most important requirements for modeling are • Pollutants - From stationary sources: SO . CO, particulates, NO . 2 x - From mobile sources: CO, NO , particulates, hydrocarbons X (by types). • Resolution Temporal: 1 hour Spatial: Point sources—0.01 km. Area sources—variable size grid from 2 1 to 16 (km) . • Units. Depends on model, but always metric. • Output programs. Depends on model. Inventory should provide FORTRAN interface. Output to include source descriptors such as location (UTM coordinates), stack height, exit temperature, and velocity. For traffic models, description to include de- scriptors of roadway. • Area. The area to be covered coincides with the boundaries of the Metropolitan Saint Louis Interstate Air Quality Region. Alternatively, it could be resolved to a smaller, more central area covering only the City of Saint Louis and the Counties of Saint Charles, Jefferson, Saint Louis, Monroe, Madison, and Saint Clair (Figure 1). 39 ------- • Time. The period will vary considerably. In some cases, model verifications will be limited to relatively short periods (48 to 72 hours). The requirements of other study groups, such as the health study workers, are more extensive, but generally less intensive. They require the system to store and retrieve information on many compounds, but the demands for spatial and particularly temporal resolution are much less critical. Special programs, such as Metromex, have specific requirements for certain categories of pollutants, which may or may not be satisfied by a general emission inventory. 40 ------- IV TASK B: EMISSION INVENTORY CONTENT A. General Principles 1. Introduction As is shown in Section III, there are numerous agencies and groups--both inside and outside the RAPS program—that have actual or potential needs for emission inventory data from the Saint Louis Air Quality Control Region (AQCR). Wide differences exist in the specific data requirements of these agencies, particularly with respect to the time resolution of the estimated emissions. However, the most demanding requirements are set by the need to verify experimentally various multi- source atmospheric dispersion (air quality) models. In particular, the emission inventories used for this purpose must provide a high degree of time resolution, giving emission estimates at one-hour intervals. A high degree of spatial resolution is also required. Other possible users of emission inventories surveyed under Task A have generally far less demanding requirements. In particular, they do not commonly need the refinements of time resolution required for short-interval dispersion modeling. If they use the emission data for air quality modeling at all, it would probably be to estimate relatively long-term, average pollutant concentrations. Others may use inventory data only for surveillance and for determining the priorities for regu- latory programs. In any case, the high-resolution emission inventory designed for use in dispersion model verification will be more than ade- quate for any of the other uses. Attempts have been made previously to verify experimentally predictions of short-term air quality made from dispersion models. How- ever, the source emission data fed into the dispersion models have been derived primarily from long-term average emissions (e.g., annual emis- sions), at least for stationary sources. Furthermore, most of the annual emission data have been derived by applying emission factors to survey information on fuel consumption and materials flow, rather than by making direct measurements. Estimates of hourly emmissions have been derived or extrapolated from the annual rates by more or less elaborate calculation procedures (see Section VII). The resulting compilation of extrapolation procedures, algorithms, and correlations for simulation of short-term 41 ------- emissions may be described, for lack of a better term, as an emission model. Unfortunately, the errors and uncertainties, both in the basic data and in the assumptions required for their extrapolation, are so great (or so indefinite) that estimates of hourly emissions derived from gross annual emissions cannot be relied upon for the critical tests of short-term air quality predictions to be carried out in RAPS. At best, the predictions of pollutant concentrations from an air quality model can be no more precise than the source emission data used as input to the model. In addition to errors in emission data, other sources of error in the predictions, besides the inherent limitations of the dispersion model itself, include the description and measurement of the atmospheric structure, the measurement of the actual pollutant con- centrations, and the representativeness of the pollutant concentrations actually measured. Assuming that sufficient care is exercised, we can estimate source emissions more accurately than any of the other input data to the air quality model. The level of effort needed to ensure the accuracy of the emission estimates rises rapidly as the required accuracy is increased, particularly as the time resolution is extended to shorter and shorter periods. Nevertheless, the emission estimates, being the least equivocal of the inputs to the air quality model, should be made accurately enough to ensure that they do not become the limiting factor in the testing of the model. There is no practical way by which the emissions from all sources can actually be measured during the particular hours in which the test of an air quality model is carried out. Consequently, it is necessary to employ simulation techniques to estimate the hourly emissions from most sources, a process that also can be described as emission modeling. How- ever, such emission modeling should involve as little extrapolation oi data as possible. In particular, the emission model for a particular source should, where possible, contain an independent variable to be evaluated from data collected in real time at the source. Such data can, as in an extreme case, indicate whether the source was actually in operation during the period of time under consideration. In the general case, the data should indicate the value of some variable having a functional relationship to the emission. Most directly, in sources having continuous monitoring systems, the variable may be the concentration of the specified pollutant in the waste gas. In other cases, it may be a quantity such as the rate of firing of fuel. At all important sources, the relationship between the emission and the chosen independent variable should be established from actual measurements. Such measurements essentially constitute a calibra- tion of the emission model. It will, of course, be necessary to redeter- mine the relationship between the independent variable and the emission (recalibrate the model) whenever there are significant changes in the mode of operation of the source. 42 ------- It is feasible to carry out an emission inventory of the fore- going type for the Saint Louis AQCR and to store hourly values of the chosen independent variables and of the corresponding emission estimates for an appropriate period (for two or three years if desired). However, the process is necessarily difficult, time consuming, and expensive. It should therefore be carried out only for the minimum number of pollutants necessary for verification of atmospheric dispersion models; as discussed in Task A, these are considered to be SO , NO , HC, and CO. The inventory, x x which should be adequate for use in verification of mesoscale atmospheric dispersion models, may be designated as the RAPS Emission Inventory. It should be appreciated that the RAPS Emission Inventory will be essentially separate and distinct from the existing National Emissions Data System (NEDS) Inventory, which was compiled under far less demanding criteria. Some information from the NEDS Inventory will be useful in the RAPS Inventory; for example, the NEDS Inventory locates and identifies stationary point emission sources and indicates their relative magnitudes. Some of the data in the NEDS Inventory relative to area sources and some small point sources may be, if carefully evaluated, adequate for inclusion in the RAPS Inventory. In addition, NEDS will provide the needed data on emissions for purposes that do not require high resolution data of the principle pollutants covered by the RAPS Emission Inventory. For some anticipated microscale atmospheric dispersion studies, very high degrees of temporal and spatial resolution of emissions may be required for specific, small areas. Again, for particular studies, higher resolution data may be required for a pollutant and included in the RAPS Emission Inventory. The precise requirements for emissions data in such cases will vary considerably, and cannot be anticipated at this time. Specification and execution of these special emission data collections are basically research tasks that will be best handled by the research project staff carrying on the dispersion studies. 2. The Emission Inventory System The following discussion is intended to provide an understanding of the general process by which information is generated and flows through the proposed RAPS Emission Inventory system and is intended to present definitions of terms (see Figure 2). It should be appreciated that, whereas the eventual emission inventory is organized primarily according to pollutants, the execution of the inventory is necessarily organized primarily according to sources (the manner in which the original data must be obtained). The process 43 ------- DATA SURVEYS INTERMITTENT EMISSION MEASUREMENTS CONTINUOUS EMISSION MONITORING PROCESSED EXPERIMENTAL AND ANALYTICAL DATA ON EMISSIONS EMISSION SOURCE OPERATING VARIABLES BASIC SOURCE DATA FILE CONTINUOUS EMISSION MONITORING EMISSION MODEL \ EMISSION SOURCE OPERATING VARIABLES 1 EMISSION ESTIMATES EMISSION INVENTORY Output FIGURE 2 EMISSION INVENTORY SYSTEM SA-2579-42 44 ------- by which the RAPS inventory is to be carried out is essentially conven- tional; however, the special emphasis on and treatment of certain steps is determined by the high degree of temporal and spatial resolution required to meet the specific requirements of RAPS such as verification of atmospheric dispersion models. Original information from emission sources is to be compiled from data surveys and, as far as possible, directly from intermittent measurements or continuous monitoring of the sources themselves. Sub- sidiary data on the operating variables and parameters of the sources must be collected so that correlations can be made between the variables and parameters and the actual measured emissions. The source variables are to be quantities that are--or can be--routinely measured as part of the source operation. Hence, once a correlation is established, the measurement of the variable can be substituted for the more difficult operation of measuring the actual pollutant emission. For a relatively small number of the largest point sources, it is anticipated that continuous analyzers will be installed to monitor the concentration of key pollutants (SOg and NOX). For the remaining point sources as far as practicable, reliance must be placed on intermittent manual measurements of emissions correlated with continuously measured process variables as described above, rather than upon the use of exist- ing emission factors, although the latter can be used in certain cases. For the smaller point sources, and for area and line sources (stationary and mobile), individual emission measurements will not be made, but more accurate emission factors than are currently available for typical sources must be derived from a sampling program. Processing procedures for the raw data will depend on the nature of the information. At least initially, computation of the emissions from raw experimental data will probably be most conveniently carried out manually. Once the processed data is suitably organized, it is to be stored in the Basic Source Data File, which then is the source of infor- mation required for creation of the Emission Model for each source cate- gory of sources. The emission model in this context is essentially a collection of correlations or algorithms that permit calculation of estimated emis- sions from readily or routinely measured operating variables and parameters of the various emission sources (as discussed above). The relationships must be worked out from the basic source data; they may range from simple to complex, depending on the nature of the emission sources. 45 ------- To generate specific emission estimates, it will be necessary to supply to the emission model particular values of the independent emission source variables pertinent to the desired time interval. These input data must be collected from the sources on a continuing basis over whatever time period is required. The Emission Estimates thus generated provide the input to the Emission Inventory. The Emission Inventory will consist of the compiled and stored estimates of the hourly emissions of the specified pollutants on an hourly basis for the time period (two or more years) of interest to the RAPS program. The emission estimates corresponding to any chosen sequence of one-hour intervals will then be withdrawn from the Emission Inventory to form input for the air quality models and other purposes. B. Precision of Emission Estimates Since the requirements for the emission inventory are to be set by the needs of dispersion modeling, it is necessary to specify, within some limits, what degree of agreement between predicted and measured pollutant concentrations will be considered to constitute a verification of a dis- persion model. While the work of Hilst9 discussed below approaches this issue, no such specification of model precision has been made. Modelers have thus far simply striven to obtain the best results that they could get. In tests of models made to date, it has been common to encounter deviations of ±50 to 100 percent or more between predicted short-term pollutant concentrations and the measured values. Better results are reported for prediction of long-term average concentrations. An ideal multisource atmospheric dispersion model would permit accurate prediction of the concentration of a specified pollutant at any desired point in the geographical region of interest over as short a time interval as might be of practical interest. In practice, it is necessary to compromise the requirements for the model with respect to both temporal and spatial resolution. In addition to the inherent limitations of a model itself, the precision of measurement of emissions and meteorologi- cal variables and the practicability and precision of measuring atmospheric pollutants in the field set practical lower limits on the time intervals and the receptor areas that can be considered. The predicted pollutant concentration is assumed to represent the average value within one square kilometer, which is the smallest element of the standard geographical coordinate grid. The pollutant concentra- tion measured by a single monitor will bear a relation to the average concentration that will be determined by the distribution of the pollutant 46 ------- over the specified area. The distribution of the pollutant will depend on the geographical and topographical characteristics of the area. The distribution will therefore vary not only from one grid area to another, but also within any one grid area under different circumstances and at different times. Other experimental studies in the RAPS program are expected to address the problem of the spatial variation of pollution concentrations within the receptor area. However, where it is necessary to employ a single monitor for one square kilometer area, the relationship of the measured concentra- tion to the true average concentration for the area will necessarily be statistical and subject to a greater or lesser degree of uncertainty. For such studies, or the development of microscale submodels designed to couple the regional scale models to specific localities or points where measurements are made or air pollution effects are significant, special efforts must be made to collect detailed emission data on an experimental basis. The routinely collected d;i t a of the RAPS Emission Inventory should be designed to serve the broader scale needs of the modelers on a con- tinuing basis. It will be important, however, to make a special check in the vicinity of each permanent monitoring station (say within a radius of one kilometer) to ensure that any point or other source that could significantly effect the representativeness of readings made at the monitoring station, are taken into account. For example, it may be found necessary to treat an office building heater system efflux as a specially considered point source, rather than as part of a nonspecified area source. Alternatively, if such consideration cannot be made in connection with the emission inventory system, steps should be taken by those responsible for air quality moni- toring to develop weighting or calibration factors for the station in question, or even to set its readings aside for certain purposes under particular wind conditions As stated above, it is essential that the precision of the pollutant emission data not limit the precision of the dispersion model. The dif- ficulty of attaining highly precise measurements or estimations of pollut- ant emissions increases rapidly as time intervals are shortened. Hence, it is necessary to make compromises between precision and the cost and difficulty of measurement and estimation. The question of the significance of errors in specifying the strength or location of sources has been addressed in Section VII, especially in terms of the precision needed in defining position. The sensitivity of air quality models to such errors is a complicated question and specific to 47 ------- each model and each area. The work of Hi 1st" in a case study of the TRC Model in Connecticut is indicative of what can and should be done for the major models to be tested in the RAPS program. Depending upon the resolution in space and time of the predictions made by an air quality model, Hilst's findings in New England are encouraging because the cancelling effect of random uncertainties minimizes the errors due to inaccurate point source data. Most serious errors can result from systematic inaccuracies, such as those introduced by incorrectly esti- mating area sources, and the smaller the proportion of the total inven- tory reported in this way the better. It is recommended that as early as possible, specific studies (based initially on NEDS data) of the type reported by Hilst be performed to provide information on the effects of varying the degree of precision of emissions data in Saint Louis, espe- cially in terms of hourly resolution. In the meantime, it is highly desirable to develop the most detailed and accurate emissions database economically feasible in Saint Louis, both to ensure the best support of the air quality model studies and to provide the base data for re- search in emission modeling methodology. C. Inventory Resolution 1. Temporal Resolution As is indicated above, the difficulties both of formulating an adequate dispersion model and of testing it increase rapidly as the time interval is shortened. Ideally, a model should be capable of pre- dicting the atmospheric concentration of a pollutant over the shortest interval for which an air quality standard is specified. In fact, it is probably impractical to specify an interval shorter than one hour. One limitation is that it is practically impossible to specify significant values for the pollutant emissions from most sources for periods of less than one hour. Indeed, it is difficult even to make precise estimates of emissions for periods as short as one hour, at least without making direct measurements in the particular time interval of concern. In view of these considerations, the minimum time interval to be considered may be set at one hour. This degree of time resolution has been accepted by most modelers. 2. Spatial Resolution Stationary sources fall into two categories: point sources and area sources. The distinction between the two is generally arbitrary; point sources are those having emissions of a specified pollutant in excess of some selected minimum value, and they are treated individually. Smaller 48 ------- sources within a selected region are aggregated, and their total emission is assumed to arise uniformly from the area. Choice of the minimum emission rate for specification of a point source obviously depends on the nature of the pollutant and the ratio of the emission from the particular source to the total emission from all sources. x Specification of the location of a source, whether a point or an area, is to be made by reference to UTM coordinates- The precision required in specifying the position of a source is also partly related to the position of the receptor. Some dispersion modelers have asked that the position of point sources be determined within +10 meters. From the standpoint of a nearby receptor, the physical size of the stack or vent may be appreciable; large stacks may be 10 meters or more in diameter, and some building vents may be still larger in their principal dimension. An objectivi method for determining the spatial resolution required is presented in Section VII. The method enables one to estimate the allowable error Ay, in the cross wind position of a source (for any specified stack height, source strength, wind velocity, and stability conditions), as it affects pollutant concentrations at given downwind distances. "Allowable error" is defined arbitrarily by comparison with a reference source strength— e.g., the 100 tons/year commonly used as the lowest value for point sources, and a reference positional error—e.g., 100 m, the resolution of most current inventories. Although such a high degree of precision in specifying location may only be necessary in a limited number of applications, it is recom- mended that wherever possible the location of point sources be reported with a resolution of ±10 meters. Although existing data (in NEDS) is only specified to ±100 meters, additional surveys of significant sources will have to be made in any case, and it would therefore be readily possible to obtain the further locational data at that time. As noted above, verification of microscale dispersion models will generally be special exercises. They will require finer spatial resolution and more detailed specification for local sources than will be required for tests or use of the mesoscale models. This finer degree of resolution can be incorporated into the general emission inventory for the entire area covered by the major dispersion models. However, the addi- tional time and cost required may not be justified, since the increased resolution will probably never be required except in relatively few small areas being treated in the microscale dispersion studies. It is more practical to make special, fine-resolution inventories for the particular limited areas of the microscale studies. 49 ------- D. Pollutants 1. General Discussion Initially, inventories must be carried out for those pollutants that are to be used in model verification. In principle, it would be desirable to carry out the inventories simultaneously for as many of the pollutants as possible, particularly where surveys of the same sources are involved. But in practice, it may not be entirely feasible to do this, because the basic data may simply not be available for all of the pollu- tants at the same time. Acquisition of adequate data on emissions of nitrogen oxides and hydrocarbons will probably take much longer than will the gathering of equivalent information for sulfur oxides. 2. Nonreactive Gases Although none of the major pollutant gases that may be used in tests of dispersion models are actually nonreactive, both sulfur dioxide and carbon monoxide are relatively unreactive in relation to the time of dispersion that is of primary interest to dispersion model verification. Carbon monoxide is the more inert of the two. However, sulfur dioxide has the advantage that its reaction products (sulfuric acid mist and sulfates) can be traced by chemical analysis, whereas the identity of carbon monoxide is lost after reaction. Carbon monoxide is best suited to testing models of dispersion from automotive sources, since these are by far the major sources of this pollutant. It will be necessary, nevertheless, to inventory stationary sources, since their contribution to the ambient concentration of carbon monoxide may be appreciable in relation to the model testing. It has been found that large quantities of carbon monoxide are produced in nature by the oxidation of methane in the atmosphere. Carbon monoxide is removed from the atmosphere by microorganisms in the soil. The background concentration of carbon monoxide in the atmosphere is determined by the balance of the natural removal processes and of the natural and anthropogenic formation of the compound. It is unlikely that either natural formation or natural removal is important on the time scale of concern to the dispersion studies and within urban areas, where high concentrations of carbon monoxide may be generated. Task E presents a more detailed discussion of natural emissions- Sulfur dioxide is best suited to use in testing of dispersion models for pollutants from stationary sources, since relatively little is 50 ------- emitted from mobile sources. It is more reactive than carbon monoxide and more subject in the atmosphere to the action of various removal processes. The oxidation of the sulfur dioxide to sulfur trioxide re- sults in conversion to an aerosol (sulfuric acid mist), and subsequent or coincident reaction of the sulfuric acid with ammonia or other basic materials in the atmosphere results in formation of solid aerosols. If the time interval of interest to the dispersion study is no more than a few hours, these reactions will probably not be of great consequence. Fortunately, it is possible to sample for both sulfur dioxide and the sulfate aerosols and, thus, to deduce the probable history and dispersion pattern of the total sulfur originally being discharged to the atmosphere as sulfur dioxide. (About two to three percent of the sulfur emitted from combustion processes is usually already in the form of sulfur tri- oxide . ) Sulfur dioxide and its reaction products may be removed, or scavenged, from the atmosphere by a variety of mechanisms. The sulfur dioxide may be absorbed by vegetation or reactive solid surfaces or be adsorbed by a variety of solid surfaces. It may also be absorbed by rain drops. The sulfuric acid or sulfate aerosols may be rained out or washed out or may be mechanically deposited by inertial, diffusional, or gravita- tional mechanisms. In time, the aerosols will also coagulate, increasing the particle size and accentuating particle deposition by the inertial and gravitational mechanisms. The coagulation mechanism will be strongly affected by the total concentration of particulate matter present in the atmosphere In general, the influence of the various removal or scavenging processes on sulfur dioxide should be much greater than the corresponding influence on carbon monoxide. Allowances must nevertheless be made for the depletion of both pollutants by the various removal processes, but determining these adjustments is, of course, a task separate from the emission inventory. Estimating the depletion of the emitted pollutants is, in fact, an inherent part of the formulation and development of the dispersion models. It will call for an extensive supporting experimental program. 3. Reactive Gases The most important transformation process involving major air pollutants is the photochemical reaction of the nitrogen oxides with hydro- carbons to form PAN and ozone. The nitrogen oxides are contributed in roughly comparable amounts by both stationary and mobile sources, which 51 ------- also contribute the hydrocarbons. Consequently, it is necessary to devote comparable attention to an inventory of emissions of both of these types of pollutants from both classes of sources. As they are present in waste gases at elevated temperatures, the nitrogen oxides are primarily in the form of nitric oxide (NO). Partial conversion of the NO to nitrogen dioxide (NO2) takes place primarily after the waste gases are cooled and mixed with ambient air. There appears to be no compelling reason why the distribution of NO and NOg in the hot waste gases need be determined or specified as part of the emission inven- tory. The subsequent distribution of the two species in the atmosphere is, of course, of critical concern in the transformation models, but it will have to be determined from separate experimental and analytical studies. Developing an inventory for nitrogen oxides presents special problems. Most of the nitrogen oxides are formed by fixation of atmo- spheric nitrogen at elevated temperatures in the combustion processes, and the quantity formed is dependent on the temperature reached, the residence time at high temperature, the size and configuration of furnace or other combustion devices, the amount of excess air present, and any other fac- tors that may influence the time and temperature history of the combustion gases. The nature of the fuel also influences the formation of nitrogen oxides, primarily through those fuel properties that affect the time and temperature relationships--flame temperature and flame emissivity. Any nitrogen compounds in the fuel may also be partly oxidized. Hence, the quantity of nitrogen oxides emitted from a given piece of combustion equipment cannot be estimated from first principles by any currently available methods. Strictly, it must be determined experimentally for each device and each operating condition. Fortunately, the range of variation in nitrogen oxides formation is not so great that reasonably representative values cannot be obtained by testing specific types of devices over appropriate ranges of operating conditions. Nevertheless, extensive source sampling will be required to obtain adequate information for a precise emissions inventory. Inventory of the hydrocarbon gases and vapors presents a poten- tially complicated case because of the very large number of different organic compounds that may be emitted and that may also vary widely in photochemical reactivity. It is completely impractical even to attempt to inventory emissions compound by compound, except in instances where, for example, a relatively pure, known compound is lost by vaporization. Probably the most that can be done in a practical way is to identify the organic compounds by major groups or classes (e.g., paraffins, olefins, 52 ------- diolefins, aromatics, alcohols, ketones, and so on). For the most part, even this will require a major source-sampling program. Identification and quantitative determination of the classes of compounds, alone, will call for a large analytical effort. Hence, it is essential to keep the scale of the sampling and analysis to a minimum consistent with attain- ment of the program objectives. 4. Particulate Matter The development of dispersion models for particulate matter, as well as efforts at verification, are well behind comparable developments for dispersion models for gases. The dispersion of particulate matter is more complex than that of gases, since the dynamics of the particles are strongly affected by the particle size, as are the removal processes that result in a decay of the particulate cloud. In addition, the nonspecific nature of most particulate matter makes it difficult, if not impossible, to identify the source of the material, trace its path from source to receptor, and allow for losses resulting from removal processes. In using simple dispersion models, it has been common for workers to assume that all dust emitted from emission sources remains airborne and contributes to the resulting atmospheric concentration of suspended par- ticulate matter at the receptor. Such an assumption is highly unrealistic, since relatively coarse particles will, in fact, tend to settle out. Rel- atively fine particles (about 10 microns in diameter and smaller) are frequently assumed to behave more or less like a gas; this is a fairly safe assumption, but one still subject to somewhat indeterminate errors. Making a realistic test of models concerned with particulate formation or dispersion would require the following information: • The quantity of particulate material emitted from each source and the distribution of effective particle sizes of the particulates. It should be noted that the effec- tive particle size distribution, which may reflect particle flocculation, is not necessarily the same as the distribution that may be indicated by the measuring techniques commonly employed. The particle size should, of course, be expressed in terms of the Stokes diameter (i.e., the diameter of the equivalent spherical particle of the same density that would have the same terminal settling velocity as the particle in question). 53 ------- • The extent to which particles of each size are de- posited by removal processes during passage over the terrain between the source and the receptor. No essential quantitative information is available on these phenomena. The technical problem is similar to that of determining the efficiency of dust col- lection equipment in the collection of particles of various sizes. However, the studies of particle deposition from the open atmosphere are necessarily less controllable than experiments with dust collec- tors, and work in even the latter field is not well advanced from the standpoint of basic science and technology. Determination of the mass emissions and particle-size distri- butions of particulates emitted from even the major sources alone would require a major experimental program. The experimental methods and procedures for determination of the effective particle size of the mate- rials as emitted are still poorly developed. In addition, the study of particle deposition from the atmosphere calls for a far greater experi- mental effort. If a comprehensive effort is to be made at confirmation of particulate dispersion models using particulates normally discharged from various sources, a lengthy and costly preliminary experimental and theoretical investigation will be required. Although information on particulates is noted as a requirement by a number of users identified in Task A, there is no evidence that the implications of including such data have been seriously considered. It is therefore open to question whether an attempt to verify particulate dispersion models should be made using the actual particu- lates emitted from various sources. An alternative procedure using particulate tracers may be usable with less preliminary work and less expense and at an earlier date. Use of particulate as well as gaseous tracers has already been proposed as a part of the RAPS program for tracing wind movements and plume dispersion. Conversely, it is possible to use synthetic particu- lates prepared from tracer materials (e.g., dyestuffs) to test predictions of particulate concentrations at various receptor areas, and also to get a relatively direct measurement of the depletion of the particulate cloud in its movement from source to receptor. The synthetic particulate matter can be produced within reasonable ranges of desired particle size, and it can be discriminated from other particulate matter in the atmosphere by its chemical or physical characteristics. The most convenient materials of this kind are those that can be measured by colorimetric or fluoro- metric techniques. 54 ------- If the synthetic particulates can be used for the testing of the dispersion models, there will probably be no need to develop a par- ticulate emission inventory that has a high degree of resolution with respect to time intervals and particle-size distributions. This will result in major savings of both cost and time. For the other purposes of RAPS, e.g., research on transformation processes, there may be a need for detailed information on particulate con- centrations with high temporal and spatial resolution. Because of the difficulties noted above, however, it is suggested that research studies for these purposes be planned to avoid, as far as possible, dependence upon general source emissions data. Better results can be achieved by special monitoring of only the critical sources in a carefully selected area. The alternative is an extremely costly and difficult inventory development program. In summary, it is proposed that data on particulates should not be included in the high-resolution RAPS inventory emission, since the means are not currently at hand either to acquire the full range of data that may in principle be required or to employ the information effectively if it were obtained. 5. Other Pollutants In the RAPS area, only a few agencies have indicated an interest in obtaining inventories of pollutants other than the major ones discussed above, and none of the agencies that did were concerned with short-term dispersion modeling. Local and state regulatory agencies were interested in emission inventories for designated hazardous air pollutants, including mercury, beryllium, cadmium, lead, and asbestos. They called only for quarterly or yearly average emission rates or, at most, for daily average emissions during air pollution episodes. The CHESS program includes consideration of a wider number of trace materials as pollutants including cadmium, zinc, lead, mercury, chromium, vanadium, nickel, copper, manganese, asbestos, benzo-a-pyrene, and polychlorinated biphenyls. Although CHESS relies mainly on air con- centration measurements obtained from its monitoring network, it will probably require estimates of relatively long-term average concentrations in the years before the air monitoring network was established. Thus, it may be necessary to prepare estimates of emission rates for these earlier periods, so that the atmospheric concentrations at those times can be calculated from use of dispersion models. 55 ------- In general, there is little point in attempting to carry out emission inventories for pollutants other than the major ones discussed above until such time as the precise requirements are specified by the potential users. Variations in the required precision and the time and space resolution can greatly change the costs and time required for the inventory. 6. Heat and Water Vapor Releases Although not pollutants, sensible heat and water vapor releases are of interest to meteorology and dispersion modeling because of their effects on atmospheric stability. Studies are now being made of the effects of the urban "heat island. The sensible heat and water vapor content in waste gases can generally be calculated readily from data that will be collected in the course of the emission inventories for sulfur oxides, nitrogen oxides, and particulate matter. The heat transferred to the atmosphere by convection and radiation from the surfaces of buildings, paved areas, and the earth must, of course, be estimated by separate pro- cedures not related to the emissions inventory. The temperatures and volumetric flow rates of the waste gases will be part of the information collected during the inventories. Hence, it will be possible to calculate the sensible heat content of the gases above the ambient temperature level. For combustion processes, it will generally be possible to calculate the water vapor emission from the quantity and the hydrogen content of the fuel burned. For some other fuel-burning operations, such as direct-fired drying, the water vapor present in the waste gas will greatly exceed that formed by burning of the hydrogen in the fuel. However, the water vapor content of the gas may be obtained during the sampling for various pollutants. Sometimes data on the process may permit calculation of the water content of the gas. E. Source Categories 1. Classification Scheme A general classification scheme for emission sources is pre- sented in Table 10. It is broad enough to encompass all emissions of pollutants from all sources. The source categories (stationary and mobile) and subcategories (area, line, and point) are convenient from the standpoint of dispersion modeling as well as from inventory orga- nization. The division of source processes into combustion and 56 ------- o in 0 CJ 0 CQ CU H X! o en 0 O fi 3 o CQ ra c o >H p ra P CQ 0 r-> o o r. be 3 0 O p CQ CO U O) CD O o CQ 0 C •J c CO 0 < 0 CJ 3 0 CQ P C •H o OH to CD y 0 CO CO 0 ource :atego: w 2 3 C in 3 C d 0 H -p 01 & 0 u c 0 0", 3 XI E O O q o H J-J in 3 0 u c o 01 XI u o c K D X! U ource rocess •D C T3 iH CO in 0 0) 3 in <-H CH 0 O rH H bC > ^ Xl CQ CO 01 in CD rH CJ in o H 0 H .q M in o: ,c 0 oi 0 rH 3 CD > P 0 P-H co x; > c CM CJ CJ pi 0 H 0 H "C rH ca E be c ,C IH q -T in ? ft CH C 01 r-i 0 --H H a PCD 4Hcoi- 0 p o c: o o tn >> H P SH M 3 IH t4 P ro ra -O O T3 C H C — ' H 0 bo O -H bC cfl C C --» ^ q in p. H H c 0 -H H JH i—l rH O O E0cadracara CJ CQ>>HbC^3rH l in i/i i —1 r. !H rH H P p Ul 01 rH S-H ffl CO O rH CO P 0 CD ft O rH-P QJ -Hp,njOPH QJ rH 13 i— < 0CQ3^CQ MO E P H rn 3 ra C §ffi en ra &- £ H d CD E O M « CQ en p C o 0 CQ ca in C 0 CJ OP -H _Q ra u C JH rH H rt ro 3 c he CJ CJ CO rH 0 -H be o rH P f-l C 13 rH O rH 53 &( CP •a «-* c en ra X! > 0 01 > 01 rH -H P 0 H 0 •O ra P co (H P > d cj ca r-" 0 ca H tnHrHCJ.M'a HH 0 OOO >10'D-IJ £ 0 H 0T! W cO^UE'OO rHCQ^OQ^ ra ca O &H in P c p 3 a rH rH < CJ H H CJ CD in ft P in c T3 0 c en c M CO P 0 i- H ft 0 to S ft q 01 T3 o CO O0 to C U > Xi > CD ra in H P r-f O ra P ra _C ra c o ca rH to o ra 0 > M 3 co H in XX -ofi-H o-H>i0o ca ca E O a, CQ 01 rH 03 tn o t. ^ in ao o ft f- B ca - /-• in 0 > > in 0 0 ra PC P T3 cj c H ra cj C H O rH rH -H ra ^ M [H cj ra X in bcoco o H boo 0 rHtflhCTJP rH £ in o ^ O ^ o w ra ra O ft •o ra C 01 O to ra P H am 0 q in -o rC co 00 to c u tn > X! > O CO P t, H P o d T; ra P ra -C -H o> q u ra •-< w «H c ra o> oi3ra-no ^ H ;•< cj cj a in xx -OrHrH OH>.a)ea> 0OOO t>i0tJfe'^''-)r~'ri"O'* ra co E O CU co in p c ra P 3 -H O ft 57 -------An error occurred while trying to OCR this image. ------- Industrial power and steam plants will usually display load patterns substantially different from those of the public utility plants serving the general area. Where the manufacturing plants are operated on a 24-hour basis, both the electrical power and the steam required for process use should in general represent fairly steady loads. On the other hand, the power and steam plants will be essentially if not completely shut down when the manufacturing plant is out of operation. The power and steam required for heating and air conditioning the plant will superimpose a fluctuating load on that resulting from process re- quirements. The nature of the load cycle in an industrial power plant will also be affected if the plant is tied into the public utility power grid. In such cases the plant will usually send power into the public system when power demand is low in the manufacturing plant and draw power from the public system when demand is high. Depending on the type of manufacturing, electrical loads may exhibit cyclic behavior, sometimes over relatively short intervals (e.g., electrical power loads in a steel rolling mill). Batch chemical processes may have a cyclic steam demand. To obtain a significant emission inventory from such sources, particularly an inventory from which it is possible to simulate one-hour emissions, it is necessary to acquire detailed knowledge of the operating cycle of each installation as well as of the characteris- tics of the installation itself, the fuel or fuels used, and so on. A sampling program to acquire information for derivation of representative emission factors will have to be carried on. Although emissions from such sources are commonly esti- mated by emission modeling techniques (as described in Section VII) using information on load and operating patterns discussed above, we recommend that, wherever possible, emission data from the point sources responsible for the major proportion of emissions be acquired by direct methods using information on actual emissions, loads or fuel consumption. Details of this approach are given below. Incinerators constitute another point combustion source. The sulfur contents of common wastes are low, so that incinerators are only rarely significant sources of sulfur dioxide. The variability of the composition of wastes will, however, make it much more difficult to correlate the measured emissions with operating variables than with fuel-burning combustion sources. 59 ------- The design of most existing incinerators has not been favorable to efficient combustion, so that emissions of carbon monoxide and hydrocarbons tend to be much higher than in fuel-burning operations. This problem is frequently aggravated by improper operating conditions. The formation of nitrogen oxides can vary widely depending on incinerator operating temperatures as they are affected by the amount of excess air used in combustion. Since the variations between different incinerators may be extreme, emissions are relatively unpredictable. Hence, any large units (municipal incinerators and possibly some industrial incinerators) should be tested individually, i.e., on the basis of operating data, such as daily tonnage of throughput. Smaller industrial and commercial inciner- ators may be treated by modeling with only representative units being sampled, since their emissions will be small enough to be estimated with adequate precision from such data. Noncombustion Sources—Noncombustion point sources of pollutants are practically exclusively industrial in nature. They com- prise such a diverse collection of manufacturing operations that no generalizations can be offered concerning either their emissions of pollutants or the factors affecting them. Most noncombustion point sources will emit one or more of the pollutants of concern to the dis- persion model studies, as is shown in the following discussion under each pollutant. The principal point noncombustion sources of SO,, are found in petroleum refineries, nonferrous smelters, sulfuric acid plants, and-- possibly—integrated steel mills. In petroleum refineries, the process sources are the Claus sulfur plants, the regenerators of catalytic cracking units, and flares burning waste sour refinery gases. Refinery process heaters or boiler furnaces burning sulfur-containing fuels (including sour refinery gases) fall in the category of point combustion sources discussed above. The flare represents only a combustion source, but it is one of such special- ized nature that it is best classified with the process sources. Because of its irregular operation, the flare presents special problems of emis- sion estimation, which must be worked out in reference to process measure- ments available and with consideration of the actual level of S02 emmissions for which the source is responsible. Claus sulfur plants are likely to be fitted with continuous monitors for measurement of S02 in the tail gases. In the event that they are not, manual measurements of S02 emissions must be made and correlated with an available measured process variable, such 60 ------- so that the source location is rather diffuse. Control of such emissions has been made difficult by the small sizes of many of the individual leaks. However, vaporization losses from storage tanks has been a large and spe- cific source of hydrocarbon emissions. Losses from evaporation occur not only in refineries, but also in subsequent handling and storage of gasoline and other hydrocarbons liquids; estimates of such emissions should be made wherever possible. Other point sources of emissions of hydrocarbons and hydro- carbon derivatives will include large surface coating operations, such as the spray painting of automobiles at assembly plants. Data will have to be developed on the actual consumption of coatings and solvent in these operations and on the rate of consumption. Even the operators may not currently have reliable data on anything but the gross long-term con- sumption. Chemical analyses should be made to classify the hydrocar- bons and hydrocarbon derivatives emitted according the major groups (paraffins, olefins, diolefins, aromatics, alcohols, ketones, and alde- hydes). Where possible, the actual material vaporized, not the liquid stocks, should be sampled. To summarize, Table 11 shows the major noncombustion sources and the basis of the principal pollutants emitted. b. Area Sources Combustion Sources—The stationary area sources are com- posed of: • Residences. • Commercial establishments. • Institutional establishments (e.g., schools, hospitals, public buildings). • Some small industrial (manufacturing) estab- lishments. Most of the pollutants arise from the combustion of fuels used for cook- ing, space and water heating, and steam generation. Some may come from combustion of wastes in small incinerators. Industrial establishments in this category are composed mostly of small manufacturing operations that employ fuels primarily for space heating or that burn small quanti- ties of waste. 63 ------- Table 11 MAJOR NONCOMBUSTION SOURCES OF POLLUTANTS Pollutants Sources S° NO x Particulate matter Hydrocarbons and derivatives Contact sulfuric acid plants Claus sulfur plants Coke oven gas production Copper, lead, and zinc smelters Sulfonation of organic compounds Nitric acid plants Nitration of organic compounds EOF steel furnaces Secondary lead smelters Grey iron foundries Flour mills Cement plants Coke quenchers Organic chemicals manufacture Paint and varnish manufacture Coating of fabrics with rubber or plastics Venting of storage tanks for petroleum products and organic chemicals The rate of fuel consumption by the area combustion sources exhibits pronounced diurnal and seasonal variations as well as variations related to ambient temperatures. Fuel consumption peaks appear over periods of one to three hours- The consumption patterns of residences differ from those of commercial, institutional, and small industrial operations, which may operate at night and over weekends as well as during the day. The different types of sources may also use different kinds of fuel. Individual residences and small apartments tend to use natural gas, whereas large apartment houses and commercial, institutional, and industrial operations are relatively more likely to use oil or coal. 64 ------- The amounts and nature of the emissions from fuel combus- tion are related not only to the type of fuel burned but also to the type and the size of the combustion equipment. If waste incineration is used, the practices employed in incineration may vary widely. The only feasible way to develop an emission inventory for the area combustion sources is to apply emission factors characteristic of the combustion equipment to the quantities of fuels consumed. Unfor- tunately, there will seldom be data that will give directly the quantities of various fuels consumed by particular types of equipment and by specific types of users. The flow of natural gas to various sections of a city may be measured at metering stations. However, the measurements will indicate only the temporal pattern of the total flow to all users. Data on sales of oil and coal indicate only the average rates of consumption over rela- tively extended periods of time. For significant area sources, it will be necessary to deter- mine, by survey and experiment, the amounts and the patterns of fuel usage by typical components of area sources. For instance, the consumption of fuel by a typical apartment for cooking, water heating, and space heating could be measured and fitted to a model incorporating diurnal and seasonal influences and the effects of ambient temperature. Pollutant emission factors should be determined experimentally for the combustion devices unless they are already available. Similar determinations should be made for other representative source units: individual homes, offices or office buildings, retail stores, and such. Noncombustion Sources--The most important noncombustion area sources are those that emit organic vapors and may make appreciable contributions to the total emission of hydrocarbons. Prime examples include vapors from (1) gasoline storage and handling by distributors and service stations, (2) dry cleaners, (3) painting, (4) degreasing, and (5) application of asphalt to roofing and pavement. The total quantity of vapors released can be estimated fairly readily by applying appropriate emission factors to the quantity of gasoline handled, the amount of cleaning fluids consumed, the amount of paint applied, and so on. However, attainment of adequate spatial and temporal resolution will require a fairly extensive survey of the sources and a detailed study of operations at a representative sample of estab- lishments or activities of each type. The emissions of gasoline vapors from automobile service stations will vary with the day of the week and the hour of the day, as 65 ------- the volume of business fluctuates. The emissions of cleaning fluid vapors from dry-cleaning establishments are probably relatively constant during actual working hours. Solvent vapor emissions from the painting of products in small manufacturing establishments and from solvent de- greasing operations should have a similar temporal pattern. However, the emissions of solvent vapors from the painting of buildings (interior and exterior) are irregular with respect to quantity, time, and the loca- tion of the activity. Nonmethane hydrocarbons are also emitted in sig- nificant quantities in summer daylight hours. (See Section VII.) Specific estimates for the Saint Louis area are needed before that con- tribution can be properly evaluated. Again, for the major contributing sources of this type we recommend the direct approach to determining emissions data described later. 3. Mobile Sources The determination of emissions from mobile sources presents special difficulties. The problem has received considerable attention in connection with air quality modeling specifically related to the nature and variability of traffic sources. In studies of this type (as described in detail in Section VII) a considerable degree of sophistica- tion has been achieved in some cases. In these however, it has been con- venient and appropriate to consider the relationships between the behavior of the sources and the ultimate air quality conditions in models which integrate and combine the various parameters involved in ways specific to the purpose for which the model is designed. In many respects, current modeling capability in this area is inadequate, and its improvement will be an important aspect of the RAPS program. The basic RAPS Emission Inventory may properly be expected to provide basic data for the use of such modeling development, and thus data on traffic flow and traffic links will be part of the data collected and stored. For the primary needs of the principle users of RAPS, however, the inventory must provide data on the emissions of the key pollutants from mobile sources, which contribute to overall input of emission data to air quality modeling or transformation process studies. Accordingly, we recommend that, until more refined inputs are available as a result of progress in mobile source emissions modeling studies, the basic traffic data be used with simple models based upon average speed to develop the initial informa- tion on emissions of the key pollutants. 66 ------- F. Emission or Data Conversion Factors The derivation of emission estimates from continuously monitored operating variables is a critical step that will determine the validity of the high resolution inventory data. Emission factors as currently used in developing emission inventories are generalized factors that are most valid when applied to a large number of sources or processes espe- cially over extended time periods. They have limited applicability for use with single sources, certainly in the estimation of hourly emissions, and cannot therefore be used indiscriminately in deriving an inventory for RAPS in the manner proposed. The whole point of this inventory is to have high precision in terms of space and time, and for this reason, the best possible conversion factors must be used to develop the emissions data from the monitored variables. In general, this means that specific factors must be developed on an individual basis for each source for which the critical relationships are not already known with a high degree of confidence. (Even in the latter cases, some testing and verification would be desirable.) On the other hand, such specific source testing is a costly and time-consuming activity to be performed only when necessary. It is recommended, therefore, that consideration be given to using pub- lished, established emission factors for each source,* where the type of equipment or activity is sufficiently typical that the published emission factors can be identified as being truly representative of the case in question. It should be noted that, in general, the published factors have the most credibility for the major combustion sources (particularly in the case of SOg emissions). It is important to recognize, however, that in the case of the RAPS inventory, the role of minor sources is of great importance in local effect, i.e., at near-surface levels, with the spatial resolution required for RAPS. Lower confidence factors are assigned to emission factors for such lesser sources (quality factors B and lower in the categorization of Reference 11). Considerable caution must therefore be exercised in treating such lesser sources. This is especially the case for area sources, which, although minor contributors to the total burden, are important on the local scale. Special care is necssary to improve the accuracy of emission factors from such sources for use in refined modeling techniques. * For example, Reference 11. 67 ------- G. Physical Aspects of Emission Sources For purposes of dispersion modeling, it is essential to know not only the quantity of a pollutant emitted to the atmosphere but also the manner in which it is emitted. It is necessary to know the effective height from which the emitted pollutant can be assumed to start dispers- ing. Therefore, there must be sufficient information from which it will be possible to estimate the effective height of the plume of gas. The effective height of the plume is the sum of the physical height of the stack and of the plume rise, which is the distance that the plume ascends under the influence of its initial momentum and its buoyancy. For small source units, such as those that constitute area sources, the water gases are usually discharged effectively at ground level. Even where the gases are discharged a short distance above the roof of a build- ing, they will generally be carried quickly down to ground level. Hence, it is commonly assumed in dispersion modeling that the waste gases are discharged at ground level and that neither their initial temperature nor velocity influences subsequent dispersion. For point sources, and particularly for the larger ones, the emission inventory data must be supplemented by data on the gas flow rate, tempera- ture, pressure, and specific gravity, and on the physical height and diameter of the stack. Calculations of the plume rise require knowledge of the gas exit velocity and the stack inside diameter at the plane of discharge, and of the temperature and specific gravity of the gas. Cal- culation of the specific gravity of the gas requires knowledge of the composition of the gas stream including the amount of water vapor. In addition to the data noted above, the inventory should include information on the layout of the source stack with respect to other structures. As a rule of thumb, a stack should be at least 2.5 times the height of the nearest structure; otherwise, the proximity of the other structure; may lead to "downwash" of the gas from the stack and a consequent rapid descent of the plume to the ground. Short stacks pro- jecting from the roof of a building are particularly subject to this kind of problem, and the role of the stack in securing dilution and dispersion of the gas is largely lost. H. Units of Measurement Many agencies continue to use English engineering units, and it is virtually certain that many of the data coming into the inventory will continue to be expressed in engineering units despite a growing preference 68 ------- for the metric system. It is undesirable to convert such data into metric units manually before storage in the inventory because of the large amount of time it would take plus the time required for checking for conversion errors. It is recommended that the data be entered into computer storage in the same units in which they are originally received. The computer should then be programmed to convert the data to metric units when they are re- called and printed out. This procedure will diminish the amount of manual processing that the data must undergo and still retain flexibility. I. Specification of the RAPS Emission Inventory 1. General While NEDS as a source of information on annual or long-term emissions from a broad range of pollutants, it is recommended that a spe- cial high resolution RAPS Emission Inventory be developed to provide hourly data emissions of SC^, CO, NOX, and HC in the form of direct state- ments of weight of pollutant emitted in kilograms each hour from each point or line source or area element, for the Saint Louis (AQCR). Loca- tional data will be provided in UTM coordinates with a resolution of at least 0,1 kilometers for point sources. Mobile sources will be specified in terms of traffic links (with a resolution of approximately 0.02 kilo- meters) for major routes and of areas for secondary routes. Both for stationary and mobile sources, the area elements will reflect either the spatial resolution of the basic data from which the emissions are estimated (e.g., housing developments) or one kilometer grid squares, whichever is the smaller. Information on the physical characteristics of point sources and the prime data used in compiling the inventory would also be provided. Although it is not considered practicable to develop comparable data for particulates, it is recommended that additional data on particulate emissions be collected to supplement the NEDS inventory, specifically data to distinguish between coarse and fine particulate emissions from the major sources. For initial planning purposes we propose that only point sources emitting over 100 tons per year of each pollutant be considered individ- ually. The estimation of emissions from such sources will be accomplished in classes according to the magnitude of each source as specified in detail below. We recommend that emissions data be based upon direct monitoring of emissions for the largest sources. For other major sources, 69 ------- direct monitoring of hourly fuel consumption operating or process data should be used as the basis for estimating emissions. For smaller point sources, we propose that emissions be estimated on the basis of general- ized patterns of temporal variation of emission as a function of source type and using the best available data on fuel consumption, operating or process data (i.e., per shift, daily, weekly). In the limit, this ap- proach approximates current practice in which annual totals of such data are used as a basis for modeling. For area sources, we recommend the use of modifications and adaptions of already available emission models. Similarly, for a first approach to assessing mobile course emissions, we recommend the use of available models. Improved mobile source models should be employed as they become available to provide higher quality data. A simple new model should be developed to estimate natural emis- sions of nonmethane hydrocarbons on summer days. In all cases, we pro- pose that special testing efforts be carried out to ensure that the emission and conversion factors used to derive the emission estimates are as reliable as possible and have regard for the high temporal reso- lution required. Standard emission factors should be employed only if a review of the source on a case-by-case basis indicates that such a factor may be used with confidence in the particular case. 2. Point Sources a . Classification by Size It is convenient to classify point sources by size accord- ing to the amount of pollutant emitted annually. • Class I, over 100,000 tons per year. • Class II, over 10,000 but less than 100,000 tons per year. • Class III, over 5,000 but less than 10,000 tons per year. • Class IV, over 1,000 but less than 5,000 tons per year. • Class V, over 100 but less than 1,000 tons per year. • Class VI, over 1 but less than 100 tons per year. The numbers of sources comprising these classes for SO , CO, NO , and 2i x HC and their contribution to the total burden in each case is shown in 70 ------- Table 12 based upon NEDS data that is discussed in greater detail in Section VI. As indicated b\ these classifications, a relatively few sources account for the bulk of the emissions. The annual emissions, obtained from the NEDS inventory, although adequate for initial source classification, are not entirely representative of the probable contribution of the various sources to air quality at specific time periods. For example, units supplying steam for space heating will be operated only in periods of cold weather, and the actual emissions at that time will be higher than indicated by the annual average. Furthermore, the emission rate will follow a diurnal cycle even within a single day. The foregoing considerations are important because reports from atmospheric dispersion studies have indicated that ground-level pol- lutant concentrations are disproportionately affected by local low-level pollutant emissions, even though the latter are individually small and contribute—individually or even collectively—only a minor percentage of the total pollutant emission from all sources. Thus, the smaller emis- sion sources cannot be ignored even as part of the general RAPS inventory that is intended to supply emission estimates for use in mesoscale atmo- spheric dispersion models. They must be appropriately evaluated even though the effort required is more than proportionate to their relative contribution to the total emission. Within these reservations, the sta- tistics noted in Table 13 provide a useful basis for assessing the gen- eral magnitude of the problems involved in developing a program of point source emission estimates and describing the approach recommended. b. Data Acquisition Groups We propose that point source emission data be derived in four broad groups according to the basic records used to estimate emis- sions. Group I. Continuous emissions monitoring records Group II. Continuous records of fuel consumption, operating, or process data Group III. Short term, periodic records of fuel con- sumption, operating, or process data (per shift, day, or week) 71 ------- Ol 0- 01 O r- O o — o ^ 72 ------- w w CJ o w O LO Cn O o z 5 o H .a co O fa K O > I—I H W CD p CO I-H 3 E CJ •ogram -p CD H tH O p CD CJ H CD rH •H ft •a E a cs hH CO CO p CO T3 01 01 CD CJ o rf-i ft c bo b/) C S -H •H ^ P O CO P ^ -H CD C ft C O E co O CO t^ ^p in C71 CT1 rf LO m o co in 01 CO r-H CJ in ^ ^H o • CM O -H 01 01 /^ O 01 ^-v CD P m CM p m o 01 ^ in ^ r< CD >> CD 3 +J 01 r-H +J m 0 P CD P tO — 1 0! -P r-l 01 CO CD CO CO CD > 3 P E 3 P hH O -H T3 •HO) X H CD 01 > hH O > hH 01 -H ft rl H ft CO -C £ ft 73 £ •H C CO ft C CO O hH CO < hH CO d 0 •H CO P P ft CO E -a 3 01 01 S 01 0 CD CJ CJ o 1 — ' ?H CD ft 3 O £ be h< bo c! CD C -H hH p -H h- hH IPO hH p CO P £-1 ^4 H ft C CD C 3 s: ft c 0 CO 0 E o o m m CO in co i — i , — i 1 — i CO p Q H CO •c W) C •rH -P 01 CD "H o 73 ------- Group IV. Long-term (semiannual or annual) records or estimates of fuel consumption, operating, or process data. Within Groups II and III, a further distinction should be made between sources for which separate conversion or emission factors are established and those for which representative factors will be established. Data from Groups I, II, and III will be collected on an ongoing basis. In principle, it would be desirable to deal with all sources under Group I. In practice, sources should be treated in the highest ranking group possible while maintaining regard tor the costs and difficulties of so doing. A tentative work load is indicated in Table 13 for S00 and NO sources to show how the scope of effort can be Z> X analyzed and assessed. This shows that, by collecting data routinely from some 101 sources (of which 65 have been individually tested), ap- proximately 95 percent of the hourly S02 emissions and 70 percent of the hourly NO emissions can be estimated on a direct basis. Similarly, over 90 percent of the emissions of CO sources can be estimated on an hourly basis by the treatment of direct data from only six sources. Still, it would be necessary to obtain data from and test some 10 or more sources to achieve hourly estimates from direct data for only 50 percent of the HC emissions. It is recommended that the method of grouping sources for this purpose be determined after a survey has been made of all the sources over 100 tons per year. This survey will produce a better understanding of the data available and the difficulty of collecting useful direct data on fuel consumption or operating or process data from which hourly emis- sions may be derived. Such a survey should preferably be made in site visits to obtain more details of the source and operations than are avail- able from the NEDS review. It should ascertain the availability of direct monitoring data on omissions or fuel consumption and of operating or pro- ce<-s data that can provide direct information on hourly emissions. Alter— n..iivt'ly it should reveal the nature of short-term data from which hourly emission estimates can be derived. The survey should pay particular at- tention to the form of the available records, and the optimum frequency and method of collecting them. The survey will provide a basis for determining how much special source testing is necessary to provide reliable conversion or emission factors for individual sources and possible categories of source. This survey would also reveal the needs and opportunities for installing appropriate monitoring devices at selected sources, an approach that could usefully upgrade the inputs from certain sources and reduce data collec- tion effort. 74 ------- It will be appreciated that the methodology proposed allows considerable flexibility in the manner in which sources are treated. In this way, optimum benefit may be derived from the available RAPS inventory budget on a case-by-case basis. In any limitation of effort for budgetary reasons, it would be preferable to reduce the numbers of sources treated individually, rather than to restrict the source testing program, since operating patterns may be more confidently determinable at the present time than emissions or conversion factors. 3, Point Combustion Sources Combustion point sources burning oil and coal are responsible for most of the SOo emissions in the region, and are also major sources of NOV. Those burning natural gas produce NO , but do not produce SOr,. -• x ^ Neither type of point source is a major source of CO or HC, particularly when compared with automotive sources, but their emissions of these pol- lutants are significant and must be considered in the context of RAPS. The procedure for collecting emission data within the groups described in Table 13 is described here. For each source, appropriate reference data, such as type of unit, nominal ratings, control equipment, and so forth, together with physical characteristics of the emissions point (i.e., stack height, diameter at the discharge, location of stack in UTM coordinates) must be compiled. In addition, critical fuel consumption, operating, or process data must be acquired in routine collections. These constitute data as a function of time, ascertained in the initial survey to be basic for the determination of emissions. Typical of such data are • Type of fuel (e.g., coal, oil, natural gas). • Characteristics of fuel (e.g., sulfur and ash contents, higher heating value). • Fuel-firing rate (e.g., weight or volume rate, thermal rate if available). • For boiler furnaces, the steam generation rate at specified temperature and pressure. • For power boiler furnaces, the steam generation rate and electric power output. Then, the following procedures should be applied. 75 ------- Group I; Directly Monitored Emission Sources—Most of the sources in Class I (over 10,000 tons per year of pollutants) are expected to be equipped with continuous monitoring devices for the measurement of SO and NO and possibly with a Pitot ^ x element to measure flue gas flow continuously. Exit flue gas temperatures are likely to be recorded continuously at such large sources. The calibration of the permanently installed recorders will have to be verified experimentally unless data on such verification are available at the source. Similarly, flow measurements have to be obtained or verified. If no per- manently installed Pitot element is on hand, such flow measure- ments will have to be correlated with a measured furnace oper- ating parameter or variable. It may be advantageous to under- take the determination of other pollutants at the time of calibration. For example, from flue gas analysis, measurements of CO and hydrocarbons should be made, if the source is a sig- nificant (more than 100 tons per year) emitter of these pollu- tants. Group II; Directly Monitored Fuel Consumption, Process, or Operating Data—Typically these are sources emitting less than 10,000 tons per year of SO2 or NOx but more than 5,000 tons per year and any larger sources not equipped with continuous moni- toring devices. These units are not expected to have continu- ous monitoring systems for measurements of pollutant concentra- tions, but they should have good records of fuel consumption or other operating data with hourly resolution. In Group HA, conversion factors are derived linking the continuously moni- tored parameter with the emission rates are derived from manual sampling at each source. This is carried out initially in the program and thereafter repeated intermittently to ensure the continuing validity of the results. In Group IIB, standard emission factors may be assigned to categories of similar sources after they are determined by testing representative stacks. Group III: Short-Term, Measured Fuel Consumption, Processing, or Operating Data Available—In this group, in which both Class IV (1,000 to 5,000 tons per year) and Class V (100 to 1,000 tons per year) sources will be treated, records will be avail- able on a shift, daily, or weekly basis for fuel consumption, process, or operating data. (Direct hourly data must be inter- polated.) Again two possibilities are open. In Group IIIA, each plant is considered individually, and tests are made to establish both the time dependence and the amount of pollutant emissions that can be deduced from the recorded data. Appro- priate conversion or emission factors are thus assigned to each 76 ------- source tested. In Group IIIB, factors are established for ap- propriate categories of similar sources on the basis of tests of representative samples. Group IV: Long-Term Fuel Consumption, Processing, or Operating Data—For sources over 100 tons per year, for which suitable specific data are not available (or which must be treated in this fashion for budgetary reasons), we recommend the use of modeling procedures of the type described in Section VII. A suggested approach for S02 is to employ the Argonne3 model, and for NOX and HC the SAI model,1'3 but with adaptation and modi- fication to ensure that the best possible input data on fuel con- sumption are used and that conversion factors are verified by testing. 4. Point Noncombustion Sources Because of the extreme diversity in point noncombustion sources, no general and detailed procedure for correlating emissions with source operating variables can be oifered; appropriate procedures will have to be worked out, case by case, for each type of source, following the gen- eral principles outlined in Table 13 and in Section IV-E-2. For the purpose of organizing the inventory, such sources should also be considered according to the most important pollutant that they emit. In this grouping, particulate emission^ will not be used as a basis even though they may be the principal pollutant emitted by some sources. J. Stationary Area Sources i \| \ 1. General j Stationary area sources contribute minor proportions of the total burden of SO0, NO , CO, or HC as described here. Their importance i—i X may be significant, however, in the context of RAPS, since they may con- tribute significantly to the concentrations of the key pollutants at the near-surface level on a local basis. 2. Area Combustion Sources The NEDS inventory data indicate broadly that the emission sources in this category are primarily those using natural gas or dis- tillate oils. Consequently, the estimated SO^ emissions are minor (only 77 ------- about 1 percent of the estimated total). The relative contribution of NOX from these sources is greater, but still a minor part of the total. The NEDS inventory indicates that 28 percent of the total NOV emitted x comes from area sources, but these include vehicles that are known to be, in the aggregate, a large contributor. The contributions ol CO and hydrocarbons from the stationary area combustion sources are likely to be minimal. As discussed, the only practical way to estimate short-term emission rates from the area sources is the use of simulation techniques such as are discussed in Section VII. The two most advanced models to date are the Argonne model2 for S00 and the SAI model1'3 for CO and NO . x They are alone in their class in that they have received significant field verification. As discussed in Section VII, the models have limited precision for the estimation of area combustion source emissions for the RAPS inventory. Accordingly, it is suggested that to improve emissions estimates by such modeling techniques, more accurate and precise survey data on fuel consumptions as well as more accurate emission factors be obtained. Both will require that new experimental studies be made at typical sources. Data on fuel consumptions in specific, small areas of cities are not readily available. Commonly, continuous metering of natural gas is carried on only at a few points on lines supplying major sections of the city. The flows metered at these points give no indication of the breakdown of usage by type of source, and the temporal variations in flow reflect only the summation of the demand patterns of different users. The consumption of gas by individual customers can, of course, be obtained from meter records. However, these will, at best, give only the monthly totals of gas consumed; they will not show the different uses by a given customer. Data on fuel oil consumption by residential and commercial users are largely restricted to data on fuel oil sales. Even users will generally have no more than a qualitative idea of the actual rates of consumption over short periods. To obtain a high degree of both temporal and spatial resolu- tion in the estimates of emissions from area sources, it will be neces- sary to conduct extensive surveys of the patterns and amounts of fuel usage by representative sources. Sufficient sources must be surveyed to constitute statistically valid samples. It will probably be necessary to install meters at these sources that will record both the rates of fuel consumption at various times and the integrated total. 78 ------- Residential and commercial installations, in particular, will show marked diurnal variations in their fuel consumption patterns as well as variations in usage between week days and weekends or holidays. There will also be seasonal variations in the magnitude and distribution of fuel consumption. For example, the consumption of fuel for water heating may be relatively constant throughout the year, but the consumption for space heating will be highly seasonal. Because space heating will be the most highly variable as well as the largest consumer of fuel, the rate of fuel, consumption for this purpose should be correlated with ambient temperature. Better emission factors must be obtained for the typical fuel- burning units used in residences and commercial establishments. Some re- quired data may be derived from other studies currently in progress, but others will probably have to be made specifically to meet the needs of the RAPS inventory. Once emissions are determined for the representative sources, the emissions from specific areas can be estimated from the number and sizes of residences and commercial establishments in the areas. 3. Area Noncombustion Sources The only area noncombustion emission sources that appear to be significant are those emitting hydrocarbons and hydrocarbon derivatives (gasoline and solvents). The nature of the sources and their emissions was already discussed. The release of solvent vapors from the painting of buildings probably cannot be incorporated in a practical way into models for pre- diction of hourly emission rates; the emissions are too irregular in quantity, time, and location. Probably also, the solvent emissions from this source are too small to be of concern in mesoscale dispersion studies. However, a painting operation might be of concern to a microscale disper- sion study being conducted nearby. If so, making allowances for the emis- sion is best handled as a project task of the dispersion study. The area hydrocarbon sources of consistent importance will con- sist primarily of the smaller gasoline service stations and dry cleaning establishments. It will be necessary to survey the typical operating cycles of such establishments to establish the pattern of their emissions. 79 ------- K. Mobile Source Procedures For automotive mobile sources, we recommend that an average route speed model with a link or line source geometry supplemented with mea- sured data of traffic flow (via fixed sensors on high volume freeways and selected arterials) be used. The SAI (Systems Applications, Inc.) model1'3 for hourly CO, NO . X and HC emissions is recommended. It can have modified inputs derived from Stanford Research Institute's model4 for spatial and temporal dis- tribution of vehicle number and speed on a link basis for primary traffic and area basis for secondary traffic. As a basis for this approach, data should be assembled on traffic volumn and the location of the primary traffic links and of the areas of secondary traffic. Additionally, ar- rangements should be made to install and collect data from fixed sensors on selected routes, as noted above. As soon as improved methodology is developed for refining the inputs to such models, it should be used to update the inventory, and, in due course, the simple average speed approach could be replaced by more com- prehensive modeling techniques as they become available. In any case, prime data relating to route links and such would be available from the RAPS Emission Inventory records for use in such model development or other purposes. For other mobile sources the following models should be used to the extent possible: • Geomet model5 for diurnal emissions from river vessels and railroads. • Northern Research and Engineering Model,6 as revised by Geomet (in preparation), for aircraft emissions. L. Natural Background Emissions Techniques of estimating the natural emissions of nonmethane hydro- carbons in summer daytime should be developed so that such emissions may be added to the inventory in time for studies in which hydrocarbon emis- sions are considered. 80 ------- M. Particulates As noted, it is impracticable to develop a high resolution inventory of particulate emissions on a routine basis. For modeling studies it would appear that such emissions must be treated on a project basis with special collections of data (possibly limited in space and time) as re- quired. In addition, it is suggested that the tracer technique be em- ployed to investigate the dispersion of participates under various con- ditions. N. Scheduling and Scale of Effort The development of a complete RAPS Emission Inventory is scheduled to take up to three years. It would be desirable, however, to aim for an earlier provision of emissions data, especially for the first critical tests of air quality and dispersion models. These will use SOg (emphasizing stationary sources). Accordingly, the development of the inventory of these pol- lutants should be given priority, particularly in the source testing and modeling phases. (Since the majority of point sources of SC>2 also emit NOX, the collection of data on this pollutant can conveniently be carried on in parallel.) A possible schedule is shown in Table 14. The aim is to provide initial data on SO0 and NC) by the first quarter of the second year of Z> X full activities and complete data by the end of the third quarter of the second year. For CO and HC, effective initial data should become avail- able by the end of the second year and complete data by the third quarter of that year. 81 ------- Table 14 DEVELOPMENT SCHEDULE Inventory Operations Years Stationary sources Survey of point sources and assimilation of NEDS data Development and adaptation of modeling procedures Source sampling Organization of routine data collection Routine data collection Mobile sources Survey of mobile sources and routes and assimilation of existing data Development and adaptation of modeling procedure Organization of routine data collection Routine data collection Data handling system and system operation Development of programs for use with data management system selected Operation of system Completed acquisition of data SO2 and NOx Initial Full-seale CO and HC Initla 1 Full-scale Additional effort if interface facilities of basic system need major extension. 82 ------- V. TASK C: EMISSION INVENTORY FILE SYSTEM A. General This section concerns the specifications for data formats in the emission inventory. Because this inventory is intended primarily for air quality modeling in the RAPS program, the data elements were chosen to provide input data and supplementary documentation for that purpose. The data structures specified below are extensible data items in a data reference system that provides the facilities to manage such data types. Data systems of this kind have been developed and are available. We describe in this report the main functional features of the inventory system, and the computer system interfaces needed to provide data to modeling programs. Although most of the expected categories of data are used to demonstrate the data structures and formats, this chapter is not intended to be a complete listing of the content of the emission inventory. Because the data structures are specified to be extensible, any data not explicitly shown can be attached to the structure. Section B describes the computer system environment in which we are assuming this inventory will be implemented. Section C discusses the main operational aspects of the inventory. Both of these discussions provide the basis for the design of the data structures described in detail in Section D. Section E suggests a possible alternative format for implementing these data structures in the absence of a supporting data reference system. Section F lists the steps required for implementation of the inventory system, and Section G is a tabulation of the principal specifications of the computer programs needed. 83 ------- B. System Environment In this section, we discuss the key features of the system and software environment in which the emission inventory will be implemented. A data management system will be needed for the operation of the RAPS Emission Inventory, although it may be possible to perform the inventory operations with systems not having exactly the features described here,, Data management software is usually the responsibility of the central computing group of an organization, however, because this system must meet the needs of the Air Data Branch for the inventory functions, specifications for the system are given in this report. Either a general-purpose or a custom-designed data management software system must be obtained or constructed that will run on the computer facility at EPA-RTCC. This data management system should provide the basic functions of file definition; addition, revision, and deletion of data items; retrieval of selected items using selection criteria on data content; hierarchic sorting by content; and output of specified data components of selected items in specified formats to construct files for use in other programs. The system should provide for defining hierarchically structured data items, with components consisting either of single data records or lists of records, subordinate to other components of the data item. We assume that the system will provide means for assigning names to each of the components of a data item. The system should provide functional records of any desired length to contain character strings. The system should provide for data entry from punch cards, tapes, and terminals (if it functions in an on-line or interactive mode). It should also provide for merging of preliminary files from disk. Data should be readable as ASCII character strings. Conversion of declared numeric data to numeric internal representation may be provided, but is not necessary. The system should not pad card input with blanks and should provide for continuation of a data record to successive cards until the record is terminated by an end-of-record character. For the purpose of this emission inventory, all operations can be carried out in batch operating mode. However, the RAPS Data Manager has indicated that the RAPS program will have other data base requirements, and that compatibility among these different data bases would be opera- tionally and economically beneficial. One level of compatibility that can be obtained is to use the same data management software to implement all of the RAPS data bases. Some of the data base applications may require on-line use of the data system for retrieval and perhaps even for data entry. Therefore, it would be desirable for the data manage- ment system to have the potential capability of being used in an on-line mode, perhaps by means of additional software options. All of the presently existing air quality models are written in Fortran. The primary purpose of the emission inventory file is to provide input data to these models. Therefore, a system that has an interface for Fortran programs would be advantageous, because the input data files could be constructed by means of Fortran programs that could 84 ------- access the inventory files through the system interface and write data in the formats and data representations required by the models. Since the inventory file formats include data embedded in character strings, using various separator and operator characters, these records must be decomposed by the Fortran interface programs, the numerical values extracted as character strings, and converted by the interface into proper internal number representation for arithmetic operations. This translation of the internal formats will have to be added to any interface routines that a data management system may already have. The RAPS Data Manager has stated that EPA-RTCC will obtain the Univac DMS-1100 data management system when their Univac 1100 computer is installed. DMS-1100 is described in a catalog as a host language system: it is based on Cobol and can be utilized only by Cobol programs that issue calls to the data management system for file operations. DMS-1100 is not in itself a complete data management system. It must be implemented by writing Cobol programs to perform the functions of defining and labeling data structures, entering data, sorting and indexing, and formatting output data. There is no Fortran interface to DMS-1100, so additional system programming work would be needed to provide a facility for generating input files to models if the inventory is based on DMS-1100. (DMS-2000 is also being acquired.) At this time we do not have sufficiently detailed information about the facilities and file structures provided by DMS-1100 to be able to make any positive recommendation concerning its applicability to the RAPS Emission Inventory. It is estimated in Section IV on Task B that the total number of point sources that should be included in the RAPS inventory will probably not exceed 1000. The number of area sources will depend on how the areas are defined. If each area is only 1 square kilometer, then there may be several thousand area sources to be described. However, the format described here allows for any size and shape of area source and also for including sets of equivalent area sources in one item, so the number of inventory items needed to describe the area sources can be reduced to a few hundred if this feature is used. The line sources that need to be described will be mostly major traffic routes, and they should be relatively few in number. For the purpose of specifying the capacity requirements of a data management system for the RAPS inventory, we can probably put an upper bound of about 2000 data items per file. For some point sources, the data elements may be fairly detailed, but most source items will describe only one or two products or pollutants, and only a few typical patterns of time dependence can be predicted by emission models, so that most of the data items will contain from 200 to 500 characters for a total of about one million characters. The longest elements in the file will be the hourly data on consumption and emission for about 50 point sources, each of which will require up to 75,000 characters to describe hourly consumption to 1% precision for a year. Then, if hourly emission values are also stored for an average of three pollutants for each such source, the total size of each of these point source items will be about 300,000 characters, or 15 million characters for 50 sources. 85 ------- The master inventory files will be easily accomodated on a single large disk pack, such as the IBM 3330, or the Univac Fastrand drum, which holds over 100 million characters per drum. Therefore, the storage of emission values in the inventory file will not significantly affect the computer facility requirements. The accumulation of historical inventory data, over a period of years, will most conveniently be accomplished by making a separate file for each calendar year with the possible exception of the first one, which will have to average several prior years experience. If this system is eventually extended to maintain emission inventories for other regions, each Air Quality Control Region (AQCR) should be handled as a separate set of data files. 86 ------- C. Inventory Operations To define the data formats for a system file, it is necessary to have a view of how the file will be used as a part of an overall system operation. In the case of this emission inventory, there are three major operational activities: source data preparation and entry, computation of emission values of sources and production of data file inputs to air quality model programs. A principle we have followed in designing these data formats is that source data should be organized and filed with a minimum of manual transformations from the form in which it is gathered to the form in which it is entered into the computer. Thus we have chosen to organize the file with sources as the primary items, and we are making provision for entering the names of fuels, pollutants, and other materials in readable standard nomenclature. Consumption data should be entered in the units in which it is normally described to avoid manual computations and errors. The emission modelling programs can compute emission values in standard system units. The SAROAD parameter coding manual of EPA has assigned integer code numbers to many hundreds of compounds of interest in air pollution. However, only a relatively few pollutants will be included in the high resolution emission inventory for air quality model verification. Furthermore, the SAROAD codes do not include materials other than air pollutants, such as consumables that will be included in the emission inventory as a basis for computing emissions. Finally, writing 'S02' instead of '42401' to describe sulfur dioxide is more brief, more readable, requires no catalog look-up by an inspector or data clerk, and is far less error prone. We suggest, in this report, that compounds and other materials be labeled mnemonically and concisely. Of course, it is essential that the labels be absolutely consistent. Since the number of materials is expected to be relatively small, a standard thesaurus of material identifiers should be easily maintained. A considerable amount of data is needed to identify and describe each source in addition to the set of values for emission or consumption of a variety of possible pollutants or materials and a specification of the time dependence of those values. The number of different kinds of sources and different materials indicates that a hierarchic list type of data structure is most applicable to this purpose. In connection with model sensitivity studies or studies of control strategies, it will be desirable to be able to include the effects of simulated sources, or real sources operating under simulated conditions or under actual but very unusual conditions. All such sources can be entered into the Master Inventory in the standard way, but identified by an appropriate category name so that they can be selectively included with or substituted for real sources in data files for model runs. It is expected that most of the data will be entered in a batch mode via punched cards or using key-to-tape machines. In either case, 87 ------- the hierarchic structure of the data items must be represented in the character stream by additional delimiter characters. An extended format for this purpose is described in Section E on extensions for batch data entry; however, any existing system will have its own format for batch entry. This format is sufficiently complicated that it is recommended that the data be written by a skilled data clerk, who can transcribe original data from an inspector's notes or from data collection forms. This clerk should also be qualified to proof read data entries both for syntactic correctness and for data sense. Data preparation forms can be used to aid in the structuring of the data and to make it easier to use the correct syntax and punctuation characters. However, it is not possible to depend only on well-designed data preparation forms for data quality, because the content as well as the form of the data must be verified. A data management system generally will check its own data entry syntax, and should be used for this purpose. However, the internal formats of some of the data elements, such as consumption data, will be transparent to the system and will need to be tested by a separate emission inventory verification program. The same program should also verify whether material names are included in the material name thesaurus. This thesaurus therefore should be kept in the form of a computer reference file available to the data entry program as well as in paper reference copies in the data preparation rooms. For example, the substitution of the digit-10' for the letter-'O1 in S02 by a typist can be detected this way. The same kind of verification should be applied to all other data descriptors for which there is only a small number of correct choices. The verification program should make output listings of all items in which it detects errors in a form in which the erroneous data component or syntax is identified. Although the verification program could include facilities for user-entered corrections, it will be more practical to use the editing facilities of the data management system for this purpose, and this is recommended as the preferred approach. Data should be entered first into a preliminary data system file, so that they can more easily be further verified by printing out the entire content of a small data set in an expanded easily readable format for proof reading and further correction prior to adding these items to the main inventory file. Another reason for this preliminary stage is that most data management systems perform a significant amount of behind-the-scenes processing on new entries into a directly accessible file, such cS ".riding address tables, hash coding computations, making inverted lit_» GJ. key data elements. The small preliminary file, on the other hand, can be a simple sequential file with none of this extra processing until after the data manager has determined that it is accurate enough to merge it into the main file. The retrieval facility of the data management system will be used for the selection of groups of sources for which emission values can be computed by the same tuodel or are selected for production of a particular data file for model validation or for editing and updating of selected items. The retrieval of source items should be accomplished on the basis of any appropriate data in the item, such as source category, 88 ------- pollutant, location, or status of previous computation of emissions. The retrieval facility can also be used to produce special files organized in terms of any user-specified parameters, such as a pollutant, sources using fuel at greater than some rate, location. This can provide for a variety of special applications without requiring additional file programming. A sorting capability in the data management system should be provided to order and index these files by their content. In very few cases will source data include measured values of emission. For most actual sources, the emission must be calculated from consumption or other source data. Formats are provided in this inventory for both emission data and consumption data, using the same data syntax and structure for both. For those sources for which there are no direct emission data, emission must be calculated using emission factors or emission models applied to consumption or activity data. The emission factors have been compiled by EPA into a table as a function of the Source Classification Codes (SCC) for a wide variety of industrial processes and consumables. After a set of new data source items has been entered, the next logical operation is to calculate the emission values for those sources that have no pollutant data entered. The results of these computations can be appended to each source item as the content of a pollutant component using the same time interval description for each source as was used for consumption if only the emission factor is applied. If a specific emission model is used, the time interval description also must be generated by the model program. The emission inventory software system will require emission model programs that can access the consumption data element for each item, refer to the SCC code for that material, look up the appropriate emission factor in an auxiliary Emission Factor File maintained as a direct access file on the system, and compute the emission values for each of a prespecified set of pollutants. If the programs have access to the data system for file v,riling as well as reading, they can then update each item by adding a pollutant list component with an entry for each pollutant to be calculated. One of these will be a "no model" program that will only multiply consumption data by emission factors to obtain emission values. These emission models will generally also require input of ambient atmospheric data, which can be obtained from meteorolr.ylc,:J dr^a files based on the RAPS data network. These files are not considered in this report, because they are not part of the Emission Inventory system. In this manner, the master emission inventory file will serve both as a file of input source data and also as the repository for derived emission values with no duplication of source information. For the principal objective of the RAPS program the final function is that of production of input data files for air quality models. In general, the required emissior d?.a will be that for a specified pollutant for a specified time Interval or series of time intervals and 89 ------- also for specified positional coordinates. Many data management systems have search and retrieval features for selecting data items that satisfy specified data constraints or specifications. Some preliminary selections may be made using the system facilities, such as selection of a particular type of source, certain UTM coordinate ranges, and the like. However, the time dependence data are coded in the emission data element format in a special way, and a general system feature will not be able to decode it. If the general data management system includes the capability for calling procedural language subroutines, such a subroutine can be written to decode the time and emission data in the emission data element and can be linked to the data system so that it will automatically be invoked whenever the system is asked to search for emission data within some particular time range. If that capability is not available, then the data file production program must have a subroutine included within it to perform the time interval selection function. After the items containing the desired data have been identified, it is necessary to abstract the emission data for the particular pollutant and write an input data file for the model program. The input file must be written in exactly the format expected by the model; the values must be in the units defined for that model program. Since the model programs are written in Fortran, the data file must also be written with the appropriate integer or floating point representation for each quantity. The generation of these input data files will require interface programs designed to perform this task. These programs (or maybe one general-purpose file-building program) must read the emission data elements that are character-string data, decode them, compute new time sequences if the model requires them, transform the string representation of the emission value to internal numeric form, read the name of the unit: used and multiply the values by the. proper conversion factor, and write the file for the model. Most of the air quality models operate on the basis of allocation of emission values to a fixed geographical grid system. The data interface programs will also have the function of translating the geographical coordinates of point, area, and line sources in the inventory j.-.tr the appropriate grid units for the model. The dat-L file derived from the RAPS Emission Inventory can be combined with similar data files derived from other source inventories, such as NEDS, to obtain the most economical use of existing systems. It is assumed that programs to generate input data files for atmospheric models already exist in the NEDS system. If such programs do not exist, then it may be more economical to transfer all the NEDS data relating to the RAPS area into this RAPS inventory system. The only sources that may be duplicated iu both systems should be those for which time-resolved dat-a -in be obtained. Since there are expected to be not more than one or two hundred such sources, it will 90 ------- not be efficient to provide a program for computer translation of NEDS to RAPS data. It will be cheaper to manually copy these source data, especially since much of the data for each source will be new. If, eventually, the use of the RAPS inventory is extended to cover much more data from other areas, then the possibility of computer transfer of data between the two systems should be re-examined. Figure 3 shows a schematic of the inventory system. 91 ------- INPUT AND OUTPUT PROGRAMS DATA FILES REFERENCE FILES ORIGINAL \y SOURCE DATA / ERROR / LISTING N^ \ ERROR / MESSAGES >? RETRIEVAL \ SPECIFICATIONS/" EDITING \ CORRECTIONS/ SELECTED / DATA LISTINGS^ MODEL \ PARAMETERS / DATA ENTRY 1 r DATA MANAGEMENT SYSTEM PROGRAMS' MERGE, EDIT, SELECT, DELETE SORT, OUTPUT EMISSION MODELS: "NO MODEL" "£" .. "-7" ^ ^ ^^ * ^ PRELIM. cni IRPF DATA 1 MASTER EMISSION INVENTORY SELECTED SUBFILES ( , i THESAURUS OF IDENTIFIERS EMISSION FACTORS AND SCC CODES DATA FORMAT SPECIFICATIONS i DATA r FILE GENERATORS INPUT DATA FILES FOR AIR QUALITY MODELS SA-2579-8 FIGURE 3 SCHEMATIC OF INVENTORY SYSTEM ------- D. Data Structures and Formats 1. Source Categories The three principal categories of emission sources are point, line, and area. Because they require sufficiently different identification parameters, they should be filed by using different item data structures. They can be stored in separate computer files or in the same file in different item categories. Point sources can be identified by the use of a group of identification parameters selected from those given in the manual for the NEDS inventory. Although this is a RAPS emission inventory, some of the NEDS identifiers are included primarily to facilitate data interchange between the two systems. The primary identifiers should be: * Assigned source item number (sequence number assigned by inventory system); * Source category (e.g., power station, foundry, refinery); * Source name (name of company, organization, or generic name if it has no specific name); * Plant No., Point No.; * Street address (or nearest street identification); * City, County, State, AQCR No. The UTM coordinates are not included in the source identifiers of point sources, because each stack of a multistack source will have its UTM coordinate position in the data item. If some stacks at a plant have different emission values or time sequences than others, more than one data item will be needed to inventory that plant. The Point No. is intended to identify such different sources within a plant. The NEDS Source Category Code (SCC) is included as part of the identification of each consumed or manufactured product, rather than as a single-valued source identifier, because a source may utilize more than one product. The code includes both the material and process identification. Location, source category, and sequence number identify area sources. Predetermined area sizes or shapes need not be allocated. Any closed simple area bounded by straight lines can be uniquely specified by the coordinates of the intersections of these lines taken in order. Thus the areas can be selected on the basis of convenience and appropriate internal contents without size or shape restrictions. Area sources can overlap; for example, an area identified for automotive emissions may include or overlap an apartment house area. The recommended identifiers for area sources are: * Item number; * Source category (e.g., automobiles, residential, dump); * Area number (assigned sequence number); * List of UTM coordinates of corners of boundary (starting with most westerly or southwest corner and going clockwise); * City, County, State, AQCR No.; 93 ------- * English language paragraph for general descriptive purposes. This should always include a street or locality description to supplement the UTM coordinate specification. Identification of line sources will be similar to that for area sources except for the significance of the UTM coordinate list. The sequence of the UTM coordinates of the end points of a connected sequence of straight lines approximating the position of the source will locate each line source. 2. Data Item Formats The items of the file will consist of point sources, area sources, or line sources. Many sources will have fuel-burning data or process rate data instead of direct emission data. For some plants, the materials being produced may be significant sources of air pollutants, and these products can be included in the process category. Or, if data are lacking on either, an activity may have to be identified to which an average emission model can be applied (for example, automobile traffic). An activity might be included for some point sources (like a freight terminal), if the fuel or emission rates cannot be measured directly. Thus, three different components of the source data that can be used in combination for some sources are: * Emission data * Consumption and process data * Activities. In the case of consumables, the data element would describe rates of consumption. For activities, the possibilities are so varied that no single standard format for descriptive data is likely to be usable. One example of an activity, automobile traffic, is discussed in the line source category. Except where direct emission data are available for a source, the emission can be calculated from the consumption or activity data by application of emission factors and emission models, and the result entered into that source item in the inventory file as calculated emission data. a, roiut Sources Any point source may emit one or more of several pollutants of interest. If data capacity is provided in each item for all possible pollutants, the file structure will be inefficiently used (mostly empty). A more appropriate structure would use list elements implemented such that each pollutant or consumable actually included in the item list is associated with a pointer to an associated data element, each of which is a logical record of variable length. This will permit efficient use of computer storage and efficient retrieval and selection of data for construction of model input files. 94 ------- In some data reference systems, the system automatically maintains the pointers on the basis of the data linkages specified during data entry. If the system does not provide internal pointers to associated data elements, then explicit addresses (such as record number) must be written in the file, and the user must maintain the necessary address directory. Figure 4 illustrates a sample point source item. SOURCE IDENTIFIERS ITEM NO., CATEG., NAME PLANT NO., POINT NO., ADDRESS, CITY, CO., STATE AQCR NO. STACK 1 POINTERS CONSUMPTION COAL SCC CODE POINTER OIL SCC CODE POINTER 1 UTM COORD. HEIGHT ------- b. Area Sources Because the format of an area or a line source will have to provide for an indefinite number of UTM coordinates, these are placed in a separate data element. The area and line sources are defined in terms of a single principal pollutant; thus only that pollutant is identified, instead of a list of them as in a point source item. The emission of that pollutant will usually have to be computed from consumption data by means of standard emission factors, so the consumption data should also be included in each area source description in the inventory. Figure 5 shows a sample area source item. H1/V1, H2/V SOURCE IDENTIFIERS ITEM NO., AREA NO. CATEGORY ETC. H1/V1, H2/V 0 2' - * UNIT NAME \| UNIT FACTOR UTM COORD. 2 POINTERS I DESCR. COMMENT LOCATION APPROX AREA EMISSION DATA \| METHOD POLLUTANT NAME POINTER CONF. LEV. •— J STACK HT. FLOW TEMP. SA-2579-10 FIGURE 5 SAMPLE AREA SOURCE ITEM This example shows how two separate areas can be included in one area source item. An area source item can describe a set of non-contiguous areas to avoid duplication of the complete description of each of a set of area sources having identical emission characteristics. The number in the UTM Coordinate component specifies the number of different areas described in the item. For each area, there will be a separate data element containing its coordinates. The Stack data in an area source item refers to an equivalent stack for purposes of dispersion modeling. The stack area is assumed to be the whole source area; volumetric flow rate is estimated from fuel or other consumption data. The. consumption data component is not shown in this St-.'^p1-" it is strii^tut^d the same as the consumption component shown in the _-.t a^jice sample. Only a ,dry slight change in format is needed to use a single area source item to describe the emission of more than one pollutant, as in point sources. Therefore, although we are here suggesting a single pollutant per area source item, that choice can be made by the emission inventory manager at the time that the areas are defined. ------- c. Line Sources An important type of line source is automobile traffic. The format of line source items will differ from area sources primarily in the significance of the UTM coordinate list that will be the endpoints of straight-line segments approximating the road route. The category name "Road" can identify this type of source. The activity name can be "Automobile", and additional elements should be appended to the item under the names: Traffic (average vehicle flow, average speed, vehicle mix) * Way (type, dimensions). 3. Other Data File data describing the composition of fuels or other consumed materials may be useful in the inventory, if they are variable for different sources. A data element for composition can be attached to any entry in the consumption list by a second pointer. The pointer identification is: Pointer 1 to the consumption data element, Pointer 2 to the composition data element (if any). The composition element can be a paired list of impurity element symbol or compound formula and weight fraction for as many constituents as are signifacant. The element or formula name can be terminated by a blank, and each pair can be separated by a slash. A sample is shown below. S .02 / As .005 Another category of data that may be needed in some source items is certain special emission factors, such as those for nitrogen oxides, for sources where there are actual measurements of the factors. The measured emission factors can be added to the data structure by a third pointer under the appropriate consumable. The emission factor data element will have to identify each pollutant calculated by each factor. The data can be entered in the same format as that for composition, above, except that the identifiers will be pollutants and the numerical values will be emission factors. Other source data may be added to any item by use of a labeled pointer to an extension record. Any number of such extension records may be attached to any item, but no additional storage space need be allocated to items that do not include such data. 4. Data Element Formats Most of the data elements will be numeric or text constants and will require very little formal formatting. Real numbers must be written in standard notations such as decimal or exponential (ie. 94.27, 1.03 E-4). Text elements can be written free-form. It is often convenient to have one printing character such as "#" or "$" defined as a text element terminator. A continuation character can also be defined 97 ------- to enable the entry of multiline text elements. On some systems, the linefeed key is most effective for this purpose, because it initiates a new line at the terminal. The emission and consumption data elements, on the other hand, can have a large variation in amount of data and should allow for different levels of time resolution. These therefore require format rules to provide exact identification of descriptors and data. Since the data elements for consumption data may deal with a large variety of consumable products, a single unit of measure (e.g., grams/second) may not always be appropriate. Therefore, the first component of the data element is the name of the unit of measure and the second is the conversion factor to convert that unit to the standard inventory system unit. For those sources for which a series of time-sequence measurements are available, it will be useful to describe the emission or consumption by those measurements. On the other hand, there will be many sources, especially area sources, for which only approximate estimates can be made, usually on the basis of some repetitive patterns over time. The format rules described below can be used to describe either a consecutive time sequence or repetitive time patterns or any convenient combinations of both. This component of the data element may be very long (up to 75,000 characters), but it will still be one logical record. Each emission or consumption data element also includes a component for Method, for Confidence Level, and an optional component for any further comments that may be considered useful to add for documentation of the data. Method will generally be described by means of a concise standard descriptor or abbreviation of the name of the method used to obtain the data, For example, "EF'1 can identify emission values calculated from consumption data by application of the standard Emission Factor table. Confidence level should be an estimate of the probable error of the values given, based on a knowledge of the errors of the method of measurement, or the reliability of the source data. The emission element format is the following: Data Element = Unit name + Unit factor + Data Part + Method + Conf . Level 4 Comment(optional) Da;. n""t = Time P^t'Data Part or Value Ti.;.f- ?- - -- . ic...- Code:Time Range or (Time Part, Time Part, ...) Time K.c., ------- Operator characters are: : means a 'range/data1 sequence follows; / means a datum or a data set follows; , Is a separator between components of a simple data set; ; is a separator between data sets of a multiple data set; - is a connector between first and last time identifiers (Inclusive) of a time range; ( ) enclose a multiple group of time pattern descriptions. Some examples are: M: 1/1.4,2/1.7,3/1.9, 12/1.3; is a set of values for each of the 12 months. H: 1/9.7,2/10.3, 24/8.2; is a set of 24 hourly values, for some "average" day. (M:ll-3/D:7-l/H:21-7, M:6-7/D:l/H:22-6)/1.4; specifies a repeating pattern of 8 pm to 7 am on Saturdays and Sundays during November through March, and 9 pm to 6 am on Sundays during June and July, with an emission value of 1.4 units during each of these intervals. This looks complicated, but the notation is self-descriptive once the user becomes familiar with the meaning of the punctuation symbols. It satisfies two important criteria; it can concisely describe a large variety of possible time-dependent data, and it can be read by machine. There appears to be no way to design a preprinted, "fill in the blanks", form for data to be entered in this format. The reason is that the format is designed to provide a concise method of describing any sequence or repetitive pattern of time-dependence using several different time units, so that the syntax must be dynamic as in a language, rather than a static syntax that can be preconstructed . For each time interval code, the total period covered, and the notation is shown in Table 9. Table 9 Time Interval Codes and Ranges Time Interval Code Annual Monthly Weekly Daily Hourly M: W: D: H: Period range in calendar years to which the data applies (e.g., 69-73) 1 year 1-12 values 1 year 1-53 values 1 week 1-7 values; 4=WED, 5=THU, 1 day 1-24 values 1=SUN, 2=MON, 3=TUE, 6=FRI, 7=SAT All the time units refer to inclusive intervals (i.e., M:l-2 means January and February). Counted hours, not clock time, is used (i.e., H:7-8 means from 6 am to 8 am). Zero (0) is avoided because it leads to confusion in data entry and because it is not common usage. Since the 99 ------- week is defined to be Sunday through Saturday, a year will generally begin and end with a partial week, so that there will always be 53 identifiable weeks in every year. The emission data should correspond to the average emission during the specified interval, for a current example of the period. Multi-nested intervals can be used where sufficient data exist and the emission changes considerably over the larger time intervals. For example, hourly emission data for an average day during each of several months may be useful, if the emission rates change considerably through the year. The large choice of time intervals permits the entry of detailed time-dependence of emission if the information is available and less detailed averages when that is all that is known. If data are available only for portions of a data set, only those elements of the set need be identified and entered in the inventory because the format includes a numeric identifier for every time interval for which data are entered. Since M: and W: interval sets both cover a period of one year, they are mutually exclusive. The D: and H: intervals can be used with any one of the first two in a multiple set, and D: and H: can also be combined. By the use of these formats, each emission data element can contain as few as one value, averaged over a typical year, or a value for every hour of a year. Although it uses a concise notation, this format is sufficiently redundant to permit recognition of many entry errors. Manual proofreading will detect many common typing errors, and the use of the computer for checking the intern':! ; insistency of data and the syntactic correctness of the data elements will also be feasible. 5. Summar}- Listing The following listing summarizes the form and arrangement of identifiers in a source item. Generally, most of the entries will be data relative to higher level identifiers and an identifier for lower level data. They will only be entered into an item if they represent actual information so that only a small fraction of the entries in this table would occur in any r'r.gle inventory source item. Many of the entries in •'" - table (e.,\,t names of pollutants) represent only a few examples <_ £ u: ireitea set of possibilities and are included to display their arrangement and nrtational style. 100 ------- Point Identifiers NO. Item number. For file maintenance. Category Type of source Name Ordinary name of facility Street address City, County, State, AQCR No. Pollutants S02 Chemical formula or abbreviated name Unit Emission rate data Method C.L. Comment HC Unit Emission rate data Method C.L. Consumption Coal SCC code Unit Consumption rate data Method C.L. Activities Trucks Count/day Helicopters Count/day Comment Stacks Count UTM coord. Height Exit area Exit velocity Temp. Area Identifiers NO. Category Area Number Assigned for map reference City, County, State, AQCR No. UTM Coord Area count Coord list Description of area Pollutant Heat Unit Emission rate data Method C.L. Consumption Fuel Gas SCC code Unit Consumption rate data Method C.L. Activity Autos Flow Speed Veh. mix Population Persons Housing units Hous. mix Line (See Area—elements are essentially identical) 101 ------- E. Format Extensions for Batch Data Entry To prepare the data items off-line for batch entry, it is necessary to provide additional format conventions to indicate the logical boundaries of data components and the hierarchic level structure of each data item. If an existing data management system is obtained, it will have its own formats for this purpose. If a data management system is expected to be used to maintain the RAPS Emission Inventory, but the scheduling of data processing requires that data be prepared for machine entry before the entry formats for that system are available, then this format can be used to enter the data onto machine-readable media, and the form,. -inverted later to those required by the system. The special-pi^"" ~ .inversion program will be a temporary program use; ,'.: ' once t> , ; :',-" data previously entered. After that, a;. I new data 'i prepared directly in the system format. This extended format can a^so be used for the purpose of making sequential files for temporary use without the use or support of a hierarchic data management system. For this lain. •. > purpose, it is also useful to include a code to show whether a given J.ata it en is a current one in the file or has been replaced by an updated version, because the actual deletion of items and compaction of the i il< ^ill require processing by a file maintenance program. The characters chosen as delimiters for this extended format have not been used for the internal data formatting. These data structures, together with all format defining ,-naracters, can be implemented within the conventional 64-character ASCI! subset thai is available on most keyboard data entry devices and pij.ut.ers: [n Left bracket, followed by an integer, indicates opening of a hierarchic data level, n = data level n] Close data level n, and all inner levels. < > Delimiters of datu element nanes # Pound sign -• >-j.td. c lament separ^i.-n . Each - ' •-'at-'' itt-i:i •;> . , i t-c, n •> ith [1. • i id,, ate the status of the item, tb in;. ;et I cat> ' t-. - linked. [1 means mat the item is a current one jn the file; ' I;.LV,, i Vat it has been superseded. A partial example of a data j i t-m > . •»» che niannei of u^e of these format controls: [1 [2 # 234f/ MFG?; UNIVERSAL HORSE-COi l.ARS# 999 WHINNY ST., ST. LOUIS, M0.# 70# 2\| \|2 -..POLLUTANTS: # S02 I i/' G/S # M:l/1.4, 2/1.7, 3/1.9 12/1.3 J\| # HC [3/^ KG/HR « A:69-72/19.5# 2] [2 # 1 C3# 321.0,963. 1?> ':','=' 0.83/* '3V vji,,- H The denotes this as a paii.^ : (not no. -u .: unless all types of sources are included on one fi.it). ": lie denotar.;.o£i :J is the 102 ------- abbreviated name for Source Identifiers, a level 2 data structure. The denotation is the name of the next level 2 structure, but it contains two level 3 data sets. The terms S02 and EC are not names of data structures, they are data, but a level 3 data structure depends from each of them. is the name of another level 2 data component, containing the parameters of each stack. 103 ------- F. Implementation The next steps in the implementation of this emission inventory are the following: * Specify the set of point sources, area sources, and line sources in the St. Louis metropolitan area that will be used as the initial basis for the RAPS emission inventory. * Assemble the data applicable to these sources from other inventories and from other data sources. * If a general-purpose data management svstem is not available, then write a system of programs for fil, <.*.-finition, data entry, and updating of the emi.ss: Specify data t-ntr> : '«:.,. - rr :- 3 based joint described in this < . \|-;v. , •'. the capa'i.i: Any conflict.-, must ;• • ';ol\-id r>v ci.ai.- u1. or the choice of p' • c (,<.-)> : - ;i ra^l-. i • * Enter a small set of :at-: .-.» verify \ • \ structure and the furiat-, r» the datu In,,. ui t lie formats ) i, f t he system. <^ f riuat structures ; e o e f- s a r y. '-s of the * Enter and proof-,:; ; , * Write and tei.r tr t - Computing emibsu r. - Accessing data '.t^ of time, omisfs. • .. - Const rue ti.ig .1 ^..i. i n- mi i . f i um uo: he ousi -. -T SOUl- * Produce input data files i\,r models u: '.i procedures or separate cGrn;>iier progrj..ic. -1-•" vlata; ^ t - . , i i:d ranges (.at ions; p •. • . models. - tther data system I (14 ------- G. Specifications for Computer Programs 1. Data Management System These systems usually consist of a group of interconnected programs. This specification concerns only the external features related to the operation of the RAPS Emission Inventory. * Data bases of 20 million character size, containing up to 2000 items, each including 10 to 100 data elements, in up to 4 hierarchic levels. * Batch data entry from punch cards, magnetic tapes, or disk files. * Capability for defining hierarchic data structures, including list components. * Symbolic naming of data components. * Unrestricted lengths of lists and records. * Retrieval of items by content or partial content of any data component, including logical combinations of retrieval specifications. * Editing and updating individual items or groups of items. * Interface for external programs to access data items and data components in the files. * (Optional) Capability of calling special subroutines within the data system for operations requiring recognition of specialized formats within a data component. * (Optional) Construction of data files in specified formats for use by external programs. * (Optional) Sorting and indexing selected subfiles by content. 2. Input and Output Subroutines for Emission Model Programs These are intended to be standardized subroutines, callable by any emission model program to provide access to the emission inventory file. The input subroutine must: - Access source item via file address; - Read consumption data component from source item; - Extract SCC code, unit, time ranges, and consumption values; - Read emission factor for SCC code from Emission Factor File; Convert values to the units required by the model. The output subroutine must: - Generate calculated emission data element for the pollutant and for the time ranges defined by the model, in the format required by the inventory; - Append the emission element to the same source item from which the consumption data were read. 105 ------- 3. Verification Program This can operate in three different ways. It can be part of a data entry program; it can be a separate program to correct preliminary data files prior to merging into the Master Inventory; or it can operate within the data management system on data sets retrieved by the system. It should perform these tasks: * Verification of syntax of formatted data components for logically correct usage of operator and delimiter characters. * Comparison of material names with those in a thesaurus file to verify that every name is one of an allowed set. * Printing listings of erroneous items with bad elements marked. 4. Data File Generator Program This capability may be included in a complete data management system. It should do the following operations as a minimum, although more extensive formatting capabilities may be useful. * Accept user or file input of specifications for the exact sequence of data values and data representations required by an external program. * Take its input from a subset of the emission inventory file selected by the data system and extract data from data components (the input subroutine for the emission model can be used for this purpose). * Convert coordinate locations to the grid identification code used by the model. * Interpolate emission values, if necessary, to the time intervals required by the model. * Write the output file on the disk, magnetic tape, or punch cards, as specified. 106 ------- VI TASK D: SURVEY AND EVALUATION OF EXISTING EMISSION INVENTORY DATA A. Introduction Air pollution studies have been conducted in the Saint Louis area for many years, and several emission inventories have been developed. To avoid any unnecessary duplication of effort, it was important to review the existing emission inventories, to compare their content with the information needed for RAPS, and, if they appeared to contain useful information, to transfer this information to the RAPS inventory. The City of Saint Louis has been involved in air pollution control since its first antismoke ordinance was passed in 1893. Smoke and fumes were, of course, the first targets of pollution control efforts; even so, 50 or 60 years passed before effective pollution control measures were applied to the largest pollution sources of the Saint Louis area, the power plants and the steel mills. In 1964, an "interstate Air Pollution Study, Saint Louis-East Saint Louis Metropolitan Areas" was undertaken by the U.S. Public Health Service. Questionnaires were sent out to de- termine fuel use and combustible waste disposal practices in the area as well as manufacturing activities. A revised emission inventory, still based on 1963 data, was published in December 1966 as Phase II of the interstate study. After the Metropolitan Saint Louis Interstate Air Quality Control Region had been established, the first comprehensive inventory was taken in 1968, to serve as a basis for the Implementation Planning Program (IPP) This inventory included the jurisdictions of Jefferson, Saint Charles, and Saint Louis Counties and Saint Louis City in Missouri; and Madison, Monroe, and Saint Clair Counties in Illinois. Franklin County was added in 1969. Four additional Illinois counties, Bond, Clinton, Randolph, and Washington, ,vere added to the air quality control region in 1971. Only sulfur dioxide (SO?) ond particulates were included in the initial IPP inventory. The more recent inventories contain estimates of carbon monoxide (CO), hydrocarbons, and oxides of nitrogen (NO ) as well. The X following inventories are described in some detail in Section II: • IPP Emission Inventory-1968 • IBM Emission Inventory-1970 • DAQED Emission Inventory-1971 107 ------- • NATO Emission Inventory-1971 • NEDS Emission Inventory-1973 In addition, the following traffic and transportation inventories are discussed : • Streets and highways • Railways and vessels The emission inventories in current use by the Missouri and Illinois regulatory agencies were recently (Summer 1973) acquired and transferred to the NEDS files under another contract; therefore, they are not con- sidered separately. Contact was also established with research groups, such as the Metromex consortium, active in the Saint Louis area. However, these groups do not have any emission inventory data. T'or were any data avail- able from industrial sources, such as the Waste font]ol Council. The data available from the various air pollution control organiza- tions deal with pollution arising from industrial, commercial, residential and traffic activities. There appear to be no direct emission data on such transient sources as agricultural spraying or fugitive dust. B. Existing Emission Inventories lor the Metropolitan Saint Louis Interstate Air Quality Control Region 1. IPP Inventory-1968 The Implementation Planning Program (II'P) emission inventory covering sulfur dioxide and parti dilate matter in the six counties and the City of Saint Louis originally making up the Air Quality Control Region was assembled bv the EPA Region VII office in 1968. This inven- tory VDP used by Argo, National Laboratory as a basis for diffusion modeling is. the .u>\ <;lopinent of the IPP for Illinois and Missouri. The 1968 inventory lists yearly emissions of SO and particulates ^j from 116 major spot sources and 316 area sources, on the basis of ques- tionnaires and letters of intent filled out by local agencies. Typical printout pages are shown in Figures A-l and A-2 in Appendix A. * The pioneering interstate study by U.S. P,li,o. it> mentioned above on page 107. 108 ------- 2. IBM Inventory-1970 In 1970 the International Business Machine Company (IBM) compiled an emission inventory for the Office of Air Programs. This inventory included only three Missouri counties (Jefferson, Saint Charles, and Saint Louis) and Saint Louis City; therefore, it is not directly compar- able with other inventories in the area. Though it covers a smaller area, the IBM inventory is consider- ably more detailed than the IPP inventory. It contains data on 105 point sources, giving their location; stack parameters; fuel consumption data; and calculated S00, particulate, CO, hydrocarbon, and NO emissions in ^ x tons per year (Figure A-3, Appendix A). From these data, emission inventories were calculated for each of the political jurisdictions. In the IBM inventory, area sources were treated by counties. Each county was divided into source areas, varying in size from 4 to 700 square kilometers. For each area, emissions of particulates, S02, CO, hydrocarbons, and oxides of nitrogen were calculated for domestic, commercial, and industrial sources, as well as for incinerators and motor vehicles. A summary of the data is shown in Figure A-4, Appendix A. 3. DAQED Emission Inventory-1971 The former Division of Air Quality and Emission Data (DAQED) acquired an emission inventory for the Saint Louis area by using the rapid survey technique described by Ozolins and Smith.13 This technique is based on information available in most urban areas; it does not entail extensive surveys or sampling procedures. The results are, however, considerably more accurate than those derived by gross approximations based on published production or consumption data. The inventory included SO?, particulate matter, CO, hydrocarbons, and oxides of nitrogen. The data are given in metric units. Summary information on population, area, some background information, and air quality data, IF shown in Tables 15 and 16. 4. NATO Emission Inventory-1971 The North Atlantic Treaty Organization (NATO) emission inventory for the Metropolitan Saint Louis Interstate Air Quality Region was pre- pared in 1969 to serve as a comparison and model for the development of emission inventories in Frankfurt, Germany, and Ankara, Turkey. The so- called NATO Emission Inventory covered the following pollutants: 109 ------- CJ r; 1 -f C O rH o d o 0 C J^ -fj ^ c -r- d d Cfi •u t. O CJ d TH cj ,— r-H rH 5" 'X 3 ™ 0 cr. ix x rH H 5 " hJ "C c . d +j CO rn a C- ^" IT 0 /. rH rH £— ' :X < en Ci - o ' PL, O1 S5 -C P~j hH T-J 3 >. P ^ "S cd & ~ O E O O g § S -. 1-4 X, W OJ (0 03 r-H G G < r-H td o ra — a o -H rH < S. £ OX o w w x DH W p; <; w '"- p N-- r. 5 O -r-' S£ CO >-; HH J"' 1— .11 1 rH K £; 2 " O ~ C/] GO • I-H 4-1 P1- .0 'J~l i- i — i C 0 £ M_. H 'P CO W O BH i i C£ CJ 5- CJ •rJ C •H ~Z CO r-H 'rH in rH £ d J- Cj f-^ O _r, H 10 . !— • O .C •-> i_j ^ M CJ G O CJ , — 1 d o !S a ^ 0 fcx r- ' rH \_x rH d I 1 O t — J n o H d E • ™ rH £ £ "C C d 4- " <~t „ ^ C X J_ o ~ £ "3 X r- G 73 O o c > r-H j tn ^ c C i — i c d o — /: C o t-* -t-j 3 C Ji 0 ^ £ O > d I-H ^ C7 c; -T- ^ W W ^ m r^i 00 t^ CD , 1 ^ O I-H a a 0 g "C C o r= b£ (H d • ,£? X 0 CJ rl ^ •C 3 0 O X JH d i— * 'X rH C r" c -.- "j- 3 X 73 j- ___ ^ ~ Cj C " d •M ^ J] c 3 rH w-i a ^H x o ^ O 0 ^ a -^. C-H C C c -r- 0 0 -H P: X 3 X.! •• £ CO 0 y, o O I — i b£ CO C CO H hH rH S 3 W T: W H < O hH C y o "H '^ 5 - r-* i^ K 0 & s c 0 0 CO +J o ^, O CT: CO ^—. x O P- c/: "T, P^ 4-> ri C O d 1— 1 rH w {JO CO QJ s s & o o o rH U PH CO 00 i-J W E-u r-H CJ r-H -H CO r- 3 3 C 'r- C r-H d 3 J- O E cl 3 C CT n C 4- CJ O £ X. 7 ^ d C X o £1 O 3 --• — _3 r- C 'J / X ->-; O 0 C rH C- O rH c a G r-, 0 i~i -i-J -H G 3 rH G i — ( r-H rH •rH -r-l C. X M U • -M t^ ^d d i -r- E— C C o c c c C CC 0 0 0 0 o o LO -o y rH r-\| rH E d £ •H O 4-> CJ G "U •rH x -a OJ G 7v CJ -4_i +J •H E o ;_ C J T-- CO r-H O -rH C cx J o •^ o c r-l O ^ ~ c ~z cX c H ^, 3 +^ CJ CO SH G d •H ^ _p; X d tx ^. o •^, a n r; .^ ^ ^ _- r-; Q _ ^ r-^ - r- LT — -C J ^j O ,_^ c O ^, • — _ •H CO r^ Q_ O o C; "~ O O o J^. o CO w w 8 cc OH o ^ -t- o o — d r- X r-i CJ C w d £ 3 r-« O G -- C rH £ d c "" r— __ r^ C d J -r- O C -^ O .b r C O 7 ^ o x r" r-l d rH r- 1— O O rC O 4-J -i-1 O x ~ • ^ cv '- C71 w ^ ^ C O C 4-J H G C O t£ O O LO ~ 10 [-S. LO <£ CO ex CO 1 — 1 o w CO £ C5 o c c O r; 4-J c ^, - ^ r- L C ~ r- J J- ;-_ r^ G r-i JXH J. H O I-H O .-»; C ^L, — O1 w ^ r-i O < ~ '. _ 0 « ri CO ™ <• CQ & rH OH C- < c c o X c s ^ •^ 0 r , t-f H OS g CO i ^, 1-1 I— d C 3 • •" ~ c d 0 C _r; -«-H ^^ -- d c ~ 0 ' C - ^ tp ^ i£ E C O d O "C 3 O /. x; c 5 ? ~ ~~ ~ ~ ~ ,_ " — X X G '/EC co C O O "" - - — r- ^ " " .T/, ^ -C — t-j r-i -c " j=. c d O 3 - K 3 pG X ~ _ , O '•~ C C CO ^ 'rH Q. -r- X X rH OCX 4-- E 0 , r- p 3 " 3 H ^ -r-1 -r- Jj in rH P3 w_< X3 C- C E— O 0 C r-H C CT — r- C o C 0 _, ^5 o CD C d ^ a d rH a ,_j c o - --' r-i 0 - ^ O d C d X C pi O r: ,-- C r- o c t-' 3 H- d C d -r- O t: r = ^ w ^ : , r^ C C 'J c. ~ r_ 0 r; C- X r -f /- .-• 3 rH ™ / ~ d — ' rC / '/ "1 ~ 0 -^ E '•— d O "C r- O • d -^ !/ "J C -^ rH d r-l -— G -i— 3 'l-H J", P 1 F-H 1 — 1 ~ ~ c o - l ( ^ o o C Y rH T-J d C ,£ C C- p. p, . „ X 0 i -^ — — 5 T- 7; c / S _0 O C OH' OH LO 3 "U OH c a CX CO -^ o. - OH 0 J= • !H 0 > CC -^ o £J X - H C IB 0 b£ 0 0 OS O ID s ai -^ o •rH bjC « +J rH ,3 M (H SH SH . C 0 Ul C3 O O J-J O 0 •rH rH P O. CO > ~: S QJ 00 r-l LO d +-> d) ^H -M O 0 ^H £ f-H O 0 3 y — . s 1— H r^ r-^ 1 i ' [^ C y i — i " CT LJ q y 's ^ o r--. — r. - 110 ------- aMaax ii ^ i ii 00 I 1 CT) CM CT: ^^ CO O = CD n C CM LO l-i b£ ' E- x ^r CT; CD c CM x t= rt CO 1-1 N X X CO 1- O _ M _ — rl CO K < " " O ! t- _; i- x CT: ~H r "CO CT -r u; ^ CD ^r — n x M cr. r- r-i ss H * C 5; CD (^ :o 10 10 :^ — — cr. t; Xcoxcxi ro'O CDcn 10— o C " " 2S ^ CO X C T CD r- rH ~ co C £2 i O t E- W O C ^ O 0 E- C H S i> W G cs Q < 5 - G E"1 (i. C5 O G w y J W CO ft CS 2 E- c/i C ^ < r^ SI C W c ca Pi D 0 e^ H g G "^ > '"" > "^ >' "^ . 's', 1ft ^2 ±-> •< ? -— v -M r-l ^ O X O 3 X rH CM O — ' 3 x x c tn C -H H -H H i-4 Z3 "j 4-* p c o q c o •rl -t-J >, O C3 -H -H -4-1 < c^ c C t^ X c < f. ^ <^ ?* PJ t>-. ^-, ^ iJ J rJ 0 -H W Cfl . +j +J 4_) +j -H W DO CO r/) cn CQ W t—t 1-1 ,_C i— i 1 — i p~ O xl i—i cr' i— .3: r— cr. X r: S •^ 0 rw. T— "C 5_, C -; • C rH ™: c3 C fn ~ P rH 1 I 1 K • - ™ — ^ Cj rH — ~^ F-l rJ C rC C J~ I1 C 1 •- c C 0 •g ~ = ~ ," ? ~ "" X 1 0 + - ^ cr. CJ G -c o p rH O 4-1 CJ C U £ O 3 i-J r— 1 O C w Q H ~ +- C 4-1 -H CO t-i P O D r-J ^ * 111 ------- • Sulfur dioxide • Participates • Carbon monoxide • Hydrocarbons • Oxides of nitrogen Essentially, this inventory v.as an updating from the IPP and DAQKD emis- sion inventories of all area and major point source emissions. In June, 1971, the NATO inventory was further updated by using nev. emission factors. The NATO inventory is the most \ersatile of the inventories discussed so far. The data are grouped in the following 23 sections: (1) Summary if Air Pollutant l-missi ons, Study Area. (2) Summary of Air Pollutant Emissions, Saint Louis (ity, Missouri. (3) Summary of Air Pollutant Emissions, Saint Louis County, Missouri. (4) Summary of Air Pollutant Emissions, Jefferson County, Missouri. (5) Summary of Air Pollutant Emissions, Saint Charles County, Missouri. (6) Summary of Air Pollutant Emissions, Eranklin County, Missouri. (7) Summary of Air Pollutant Emissions, Madison County, Illinois. (8) Summary of Air Pollutant Emissions, Monroe County, Illinois, (9) Summary of Air Pollutant Emissions, Saint Clair County, Illinois. (10) Summary of Air Pollutants from Point Sources. (11) Summary of Air Pollutants from Area Sources. (12) Combustion of Fuels, Stationary Sources. (13) Transportation Sources. (14) Process Losses. (15) Emissions from Evaporative Losses. (16) Emissions from Solid \\astc Disposal. 112 ------- (17) Point Source Emissions. (18) Area Source Km i s s i on s (272 grids). (19) Annual Fuel Consumption (by counties). (20) Housing Units (bv .jurisdictions). (21) Population, Housing Units (by grids). •~> (22) Emission Densities (tons km"). (23) Flights per year (various airports). Samples from the inventory are shown in Figures A-.i through A-10, Appendix A. 5. NEDS Inventory-1973 The inventories discussed so far show the gradual evolvement of the techniques for collecting and processing inventory data. The earliest inventory, the IPP inventory, contained information on only two pollutants, sulfur dioxide and particulates, while subsequent inventories include also data on carbon monoxide, hydrocarbons, and oxides of nitro- gen. The data collection methods have also become More roll able. \- or the IPP imentory, information (Mi area source emissions was obtaimd from gross estimates on fuel delivery and consumption for the entire area under study. The use of the rapid survey technique for the DAQED inventory, which breaks down the information into geographic areas of the community and makes use of formulas using degree-days (a unit repre- senting the deviation from the mean ambient temperature—G5"F—for one clay), the number and size of dwellings, and other refinements, provided considerably more accurate estimates of emissions from area sources. Similar techniques were used for all subsequent inventories. A comparison of the presentations of data shows an increasing- degree of sophistication and usefulness indicated by the various ways in which the data are broken down and presented, for example, in the NATO inventory. The actual emissions of pollutants shown in the tabulations have been continuously updated to reflect current practices as new sources came into being or to reflect the installation and effect of control equipment. For this reason, the inventories currently maintained by the local and state regulatory agencies, such as the Saint Louis City or County Air Pollution Control Office, are the only ones which can be used for the purposes of RAPS. The data in these inventories, some of which 113 ------- are in the form of computer printouts (others are on Inspectors' Report Sheets), have recently (Summer 1973) been acquired and entered into the National Emission Data Systems (NEDS) inventory. The characteristics of this inventory system, presented in APTD-1135, "A Guide for Compiling a Comprehensive Emission Inventory,"14 are discussed below. The NEDS was created by the EPA Office of Air Programs to pro- vide a uniform format for source inventory data on a national scale. It consists of a number of separate files suitable for computerized process- ing and has provisions for storing data for both point and area systems as well as separate files containing various input/output, maintenance, and editing programs. The concept is shown in Figure 6. The system distinguishes between point sources (emitting at least 100 tons per year) and area sources for the primary pollutants: particulates, S02, CO, hydrocarbons, and NO . The decision to designate sources emitting more than 100 tons per year of primary pollutants as point sources is based on the estimate that a 100-tons-per-year source would produce, under average meteorological conditions, an ambient con- centration of about 5 to 8 micrograms of pollutant per cubic meter. This is about 10 percent of the secondary annual federal air quality standard for SO or particulates. (Page 10-2 of APTD-1135.14) £t In addition, about 70 compounds or categories of industrial sources are designated as point sources regardless of quantities of emissions. A point source is defined, therefore, as any emitting point or plant or facility whose summation of emitting points totals 100 tons or more per year of any one of the five primary pollutants: CO, SO , NO , ^ x particulates, or hydrocarbons; or as any of the pollutant sources listed below, regardless of quantity of emissions. • Chemical process industries: adipic acid, ammonia, ammonium nitrate, carbon black, charcoal, chlorine, detergent and soap, explosives (TNT and nitrocellulose), hydrofluoric acid, nitric acid, paint and varnish manufacturing, phosphoric acid, phthalic anhydride, plastics manufacturing, printing ink manufacturing, sodium carbonate, sulfuric acid, synthetic fibers, synthetic rubber, terephthalic acid. • Food and agricultural industries: alfalfa dehydrating, ammonium nitrate, coffee roasting, cotton ginning, feed and grain, fermentation processes, fertilizers, fish meal proces- sing, meat smoke houses, starch manufacturing, sugar cane processing. 114 ------- EMISSION INVENTORY DATA SYSTEM POINT SOURCE FILE EMISSION CALCULATION PROGRAMS AREA SOURCE FILE INPUT, OUTPUT PROGRAMS HAZARDOUS POLLUTANTS SOURCE FILE AREAWIDE INVENTORY PROGRAMS EMISSION FACTOR FILE IPP MODELING CONVERSION PROGRAM GEOGRAPHICAL ID FILE AREA SOURCE GRIDDING PROGRAM CONTROL EQUIPMENT ID FILE TREND PROJECTION ANALYSIS PROGRAMS IPP PROCESS ID FILE POPULATION DATA FILE SOURCE: "Guide for Compiling a Comprehensive Emission Inventory," Environmental Protection Agency (1972). SA-2579-11 FIGURE 6 EMISSION INVENTORY SYSTEM CONCEPT 115 ------- • Metallurgical industries: aluminum ore reduction, copper smelters, ferroalloy production, iron and steel mills, lead smelters, metallurgical coke manufacturing, zinc (primary metals industries) ; aluminum operations, brass and bronze smelting, ferroalloys, gray iron foundries, lead smelting, magnesium smelting, steel foundries, zinc processes (secondary metal industries). • Mineral products industries: asphalt roofing, asphaltic concrete batching, bricks and related clay refractories, calcium carbide, castable refractories, cement, ceramic and clay processes, clay and fly ash sinlering, coal clean- ing, concrete batching, fiberglass manufacturing, frit manu- facturing, glass manufacturing, gypsum manufacturing, lime manufacturing, mineral wool manufacturing, paperboard manu- facturing, perlite manufacturing, phosphate rock preparation, rock, gravel, and sand quarrying and processing. • All petroleum refining and petrochemical operations. • All wood processing operations. • Petroleum storage (storage tanks and bulk terminals). • Miscellaneous: fossil fuel steam electric power plants, municipal or equivalent incinerators, open burning dumps. » Hazardous pollutant sources. Another category of point sources, regardless of quantities of emissions, are pollutants designated as hazardous. A list of these is still in preparation. The location of all point sources is described using Universal Transverse Mercator (UTM) coordinates. This system, designed by the U.S. Army provides projections of square grid zones with convenient measur- ing units. The system has the advantage of continuity and uniformity for the country, and it is rapidly becoming the accepted system of base coordinates for a body of technical information. The NED System contains other files beside the source inventory. These are • The Population Data File, which contains such information as number of household units in an area, number of manu- facturing employees, number of retail and wholesale establish- ments. This information is used to calculate emissions 116 ------- where no direct data are available, as for example, emissions resulting from solid wastes, fuel consumption for space heating, and others. • The Emission Factor File, which contains conversion factors that provide an estimate of the production of a given pol- lutant by a process or activity. For example, it has been estimated that two pounds of solvent per person per year are emitted from dry cleaning plants in moderate climates. In conjunction with appropriate population data contained in the Population Data File, the total amount of hydrocarbons per unit area can thus be calculated. The most complete collec- tion of factors has been published by EPA in the Compilation of Air Pollutant Emission Factors (AP-42). • The Control Equipment File, which contains information on types and efficiencies of various control devices. As indicated in Figure 6, the NED System contains a number of programs; these include not only "housekeeping" programs, such as those for input and output, editing, and emission calculation, programs per- mitting outputs by area, source classification, and pollutant, as well as conversion programs selecting data for implementation planning pro- grams and trend analysis. The system is designed to provide data on annual emissions of pollutants; estimates of shorter emission rates—weekly, daily, or hourly-- can be calculated automatically by applying appropriately resolved pro- duction throughput, or other control data to the base emission figure. Such an approach can provide the time resolution necessary for RAPS where no actual hourly measurements are available. Typical printouts are shown in Figures A-ll and A-12 in Appendix A. Table 17 shows a comparison of the various emission inventories. Only limited conclusions can be drawn from a comparison of these data. The data in the IBM inventory, covering a much more limited area, are not comparable at all; nor do the other inventories cover exactly the same territory. The data shown for the NEDS inventory have been adjusted to include only the counties included in the two NATO inventories; thus, these three inventories are directly comparable. Good agreement is shown for sulfur oxides (if a correction of about 100,000 tons per year, corresponding to the SO emissions of Franklin County, is added to the IPP and DAQED figures). The figures for the total emission of partic- ulates and carbon monoxide are much higher in the NEDS inventory than in the older inventories; the emission values for hydrocarbons and nitrogen oxides again show quite good agreement (the contribution from Franklin County is fairly small). 117 ------- cd H * CD O W g W K PJ M O W H E- § M p •B, O PH O O co ns o o o K 0) rH cd '•^ -P £H O cd H CU d \ C 01 •H C 01 O to -p •H * — £ W c cu 01 fa£ CU O 73 r. -H -P X -H ° i tr. 0 C •c -5 >. t-( E Cd CJ o c -o 0 'H I* 0 a c CJ 0 i O 01 •H O +- -P Cd rH fl 01 3 CU 4H "G rH -H 3 X W O K>> rH o -p c cu CD 3 C TH hH g r* C CJ O CD •H H 01 01 -H g pq * cd cu ^ «jj Cd CU !H -p 01 cd cd cd cu Q P3 >H O -P CU c e cu cd > X c ~ C c X CM en i — i X co CM to ,O 0 rH r-H •P ^— v rH 3 E • ^ 0 O 0 tn 3 -P TJ IH •—' d cu cu r-H 01 I/I -P rH CU CU C • H -H rl 'H =H CJ -H C cd 'H 01 CU C O E ho C rH Cd 0 01 O -rl SH 'H rH -P CU cd c/i -p OH CJ OJ -P OH 0 P CU rH rH G1 rH -P C 3 -a o cu CJ T3 3 C rH •H O rH C c cd +-> IH O PH C X CO en rH OH 1 1 QJ r— 1 fQ C3 C3 p- E 0 O •g '^ cd -p cd T3 cu CJ ^^ ^ -o 3 CU O -P 01 cd T3 -p p. C 3 • H 0 .— - P. 00 to 0< Cn OH rH rH i— ' >> •£' -PC >. C 3 -P T3 3 C ^H C CU 0 CJ -H T3 3 "D CJ cd C O 3 C rH Cd CJ rH C -H CJ CJ O I-H >> CU C 01 r^ -P -P O -H •H C C C! SH •a cd 'M 3 c +-> cd h cd o O O S PH CO O S C o J^ CJ) L , CQ M c X t-- - o X •""" o oo CM CO OJ CO o r^ to .. ^ ^^T- CO ^ p-. c en O) ^ to X 0 en T CM to to ^f. cu' r^ 3 01 •c •H p. cd K -p c 3 0 O CU -P 0 0 e SH c -a •H C CU rH O >> -D ^J B -P 3 £.1 C i~-l cd -a 3 cj S« S O C PH Cd CJ -H c •P X to ^ en tO rH 1 — 1 Q W O1 Q C C 1-H ^ to co - x m i — i en CM CM c c m co to - r- x X X rH CM C C CO CM r- r- ~ i> in r-H tO \r x ~ ~ rH -— < O C rH f- cn en K . m T x x rH i — ' 0 0 to co rH CD CO TF ^ ^ i- r- 0! rH O .p cd cd cj -P -p cd cd cd VH T3 'O C OH OH O Oj OH -rl -H i-r 01 0) TJ "U -H CU CU E •P -P CD td cd •o -as a P. o P ^ ^ i i i i i i i i en en to to en en rH rH 0 0 H I-H H CM < 1 < 1 Z Z , , en T~ ^ x en CM CM X CD CM X M C to T . CO o X .. CO en i^ CO - t^ rr m 0 i — i t^ m ^ cd H-J cd •a OH OH 1 — 1 •Q CD -p cd TJ a D i i F 1 CM t> cn CO Q W C 3 O O 01 •H J ^j C •H cd co >, c 0 CJ 01 cu rH SH S O -P C •H a CO K- -P C 6 CJ c 0 01 0)	J -p c 3 0 o c 1—1 y c cd PH - -p O 01 •H 3 0 i-4 . -p W 01 .c -p C o g to >> cu > 0 TJ cu -p TJ a 3 01 •H CO c PJ ^ 0) rH cd cu cu • H rH rH Cd cu c o TJ CU 01 _Q CU rH CU ^ -P 3 •P fc co [^ en rH c •H CO Q H iz; o •p c •H -p 5. cu cu ^ cd -P cd TJ CU (H H 118 ------- The inventories are too close in time to allow any clear trends to show up. The differences are probably better viewed as examples of changes produced by better inventory techniques. Even so, closer examina- tion of NEDS data shows examples of values that suggest the desirability of rechecking the data; for example, a CO emission of 1,340,000 tons from a single blast furnace or evaporation of 14,000 tons from a single tank suggests a check. The former amounts to almost 35 percent of the total estimated emission of carbon monoxide for the air quality control region; if correct, it will certainly affect the view of carbon monoxide as a ~L. pollutant whose origin is primarily related to traffic sources. As the NEDS inventory represents the current understanding of emissions in the Saint Louis area, it could and should be used for plan- ning purposes. For example, it can serve as a basis for the determination of the number of point sources that should be monitored for the RAPS emission inventory, since only the most important point sources can be monitored. A review of the point sources listed in the NEDS inventory shows the breakdown of point sources by size in Table 18 and Figure 7. Table 18 NUMBER OF POINT SOURCES Pollutant Particulates SO X NO X HC CO Emission Category (tons/year) >10* 271 345 352 269 93 1 - - >102t 99 184 111 99 31 >io3t 28 67 24 23 11 >104t 6 13 7 1 6 t Data from NEDS printout of 19 December 1973. some sources not in earlier printouts. Data from NEDS printout of 28 August 1973. Includes Both of these figures are erroneous (Personal Communication, Illinois Environmental Protection Agency, December 1973). 119 ------- H "2 co A Z o „ ~ o Z A n CO CM 10 o /\ co «„ ^ ------- This information is even more relevant if it is arranged to indicate what percentage of total emissions can be evaluated by monitoring a given num- ber of sources. By referring to Figures 8 through 12, we see that we can monitor, say, 90 percent of all emissions from point sources greater than 1 ton/year by observing about 40 sources for SO2, about 5 for CO, 50 for particulates, and 100 for hydrocarbons; oxides of nitrogen, however, would require the monitoring of 330 sources to achieve the 90 percent level. Alternatively, we could decide to use an emission of 100 tons per year as a cutoff point. In that case, 184 sources for SO , 31 sources for CO, 99 sources for particulates and hydrocarbons, and 111 sources for NO should be monitored. Table 19 summarizes. As mentioned above (page 114) the 100 ton-per-year break point is consistent with a 10 percent sensitivity for modeling or measuring con- centrations near the levels of the national secondary air quality standards. While this is probably equally suitable for the modeling verification studies of RAPS, the sensitivity of such multisource models to this break point should be investigated by specific studies as recommended above on pages 48-49. Table 19 NUMBER OF SOURCES MONITORED Pollutant so2 Particulates CO Hydrocarbons NO X Type of Inventory 90 Percent of Emissions from Point Sources 40 50 5 99 330 100 Tons per Year Inventory 184 99 31 99 111 The above data can also be used to evaluate the relative im- portance, and thus the appropriate treatment, of area sources. For example, if we assume that all sources emitting more than 100 tons per year of any pollutant will be monitored directly, the remain- ing point sources (that is, those emitting between 1 and 100 tons per year) 121 ------- § h- < Z 1- 3£ cc LU LL 0- O CO LL W °£ £ 0 g<" £ uj ^- Q- 0 CL DC ^ LU CO 0 tr [2 "J o CD Q; i ° Z co C r- CO O O r- O r- CO — o q 8 CO o> CM •- r- S28 in i- •- in in tf CO CM CM O) O in en in CO O DC °£ o: Mp 8^0 < — 0 o - -I Q. , < A i j O -I «? 1- < o. t CO «- CO CD ^r en co CD CO co co co •- CM ^ CO >: i- H 0 II ~ ^* A A CO CO Q-' CL n R ^_ CO i OURCES CO <£ LU OC < < CO 8 o s 8 o Q_ 8i D Z o O o o z g o LU CC o cc O O a DC CO O CO tr O LL CO 2 O LU O I- cc < a. cc D a 8 o en 8 S30dnOS INIOd dO lN30d3d 122 ------- 100 H 2 LU O or LLI Q_ _f < 10,000 TON/YEARS 70 60 50 L-< 100,000 TON/YEARS 40 30 10 0 TOTAL AQCR ALL POINT SOURCES P.S. > 100 T/YR P.S. > 1000 T/YR P.S. > 5000 T/YR P.S. > 10,000 T/YR P.S. > 100,000 T/YR TOTAL AREA SOURCES NUMBER SOURCES 358 184 67 26 13 4 TONS/YEAR 1 ,233,805 1 ,220,897 1,182,909 1,144,906 1,060,480 990,500 608,000 12,908 PERCENT OF POINT SOURCES 100.0 96.9 93.8 86.8 81.1 49.8 PERCENT OF TOTAL 100.0 98.9 95.9 92.8 85.9 80.3 49.3 1.1 0 50 100 SOURCE: NEDS Inventory (1973). 150 200 250 NUMBER OF POINT SOURCES 300 350 400 SA-2579-14 FIGURE 9 S02 EMISSIONS FOR THE SAINT LOUIS AIR QUALITY CONTROL REGION 123 ------- 100 90 — 20 10 TOTAL AQCR ALL POINT SOURCES P.S. > 100 T/YR P.S. > 1000 T/YR P.S. > 10,000 T/YR AREA SOURCES NUMBER SOURCES 480 111 24 7 TONS/YEAR 432,790 311,577 231,745 205,770 162,100 121,213 PERCENT OF POINT SOURCES 100.0 74.3 66.0 52.0 PERCENT OF TOTAL 100.0 72.0 53.5 47.5 37.5 28.0 100 200 300 400 500 NUMBER OF POINT SOURCES 600 700 SA-2579-15 FIGURE 10 N0x EMISSIONS FOR THE SAINT LOUIS AIR QUALITY CONTROL REGION 124 ------- 100 o I- LL O 111 o cr ut Q_ > 1 TON/YEAR 30 20 10 TOTAL AQCR ALL POINT SOURCES P.S. > 100 TONS/YR P.S. > 1000 TONS/YR P.S. > 10,000 TONS/YR ALL AREA SOURCES NUMBER SOURCES 470 99 23 1 TONS/YEAR 294,908 78,295 71,051 45,960 14,100 216,613 PERCENT OF POINT SOURCES 100.0 90.7 58.7 18.0 PERCENT OF TOTAL 100.0 26.5 24.1 15.6 4.8 73.4 > 10,000 TONS/YEAR 100 200 300 400 500 NUMBER OF POINT SOURCES 600 700 SA-2579-16 FIGURE 11 HYDROCARBON EMISSIONS FOR THE SAINT LOUIS AIR QUALITY CONTROL REGION 125 ------- 100 2 LL. O HI O (t LU Q. 90 < 100 TONS/YEAR < 1000 TONS/YEAR < 10,000 TONS/YEAR OF POINT SOURCES 80 I" < 100,000 TONS/YEAR 70 60 50 40 30 20 10 0 1000 TONS/YEAR 100 TONS/YEAR OF ALL SOURCES < 10,000 TONS/YEAR < 100,000 TONS/YEAR TOTAL AQCR ALL POINT SOURCES P.S. > 100 T/YR P.S. > 1000 T/YR P.S. > 10,000 T/YR P.S. > 100,000 T/YR TOTAL AREA SOURCES NUMBER SOURCES 216 31 11 6 1 TONS/YEAR 3,852,627 1,684,792 1 ,680,920 1,674,300 1,661,300 1 ,340,000 2,167,835 PERCENT OF POINT SOURCES 100.0 99.8 99.4 98.6 79.5 I PERCENT OF TOTAL 100.0 43.7 43.6 43.5 43.1 34.8 56.3 0 50 100 150 200 250 300 350 NUMBER OF POINT SOURCES SA-2579-17 FIGURE 12 CO EMISSIONS FOR THE SAINT LOUIS AIR QUALITY CONTROL REGION 126 ------- will be assigned to area sources. For purposes other than those of the emission inventory itself, it may be desirable to monitor directly a few point sources emitting less than 100 tons per year. Thus, a source lo- cated near an air quality monitoring station may be monitored directly to aid in evaluating the measurements from the station, and would, there- fore, be treated as a point source rather than included as an area source. In addition to these minor point sources, the area sources also contain emissions of totally diffuse origin (mainly traffic, but also domestic heating). The hourly distribution patterns of these two types of sources are different. It thus becomes necessary to establish the distribution patterns of minor point sources if they amount to a significant fraction of the total area emissions. Table 20 shows the breakdown of minor point sources in relation to diffuse area sources. The data indicate, for example, that minor point sources are an important component of area SO emissions (75 percent) and small fractions of CO, hydrocarbon, and particulate emissions. On the other hand, only in the case of NO do minor point sources constitute an X appreciable fraction (18.4 percent) of the total emissions of the Saint Louis area. Thus, a realistic decision can be reached in the framework of actual emissions in the Saint Louis area. C. Traffic and Transportation Inventories 1. Streets and Highways A comprehensive study of surface transportation in the Saint Louis area was initiated in 1965 by the East-West Gateway Coordinating Council in cooperation with the State Highway Commission of Missouri, the Illinois Department of Transportation, and related federal and local agencies. The purpose of the study was to develop a balanced, multimodal transportation plan for the period extending to 1990 based on forecasts of socioeconomic and land use data. As part of this program, traffic volume data were gathered and appropriate traffic maps were prepared. Average daily traffic counts were obtained for all major streets and highways in the City of Saint Louis, Saint Louis County, and portions of Saint Charles, Franklin, and Jefferson Counties in Missouri and Madison, Monroe, and Saint Clair Counties in Illinois. The following functional definitions were used for classifica- tion purposes : 127 ------- Table 20 BREAKDOWN OF AREA SOURCES SO 2 Remaining point sources > 1 ton /year Area sources Total CO Point sources > 1 ton/year Area sources Total NO X Point sources > 1 ton/year Area sources Total HC Point sources > 1 ton/year Area sources Total Particulates Point source > 1 ton/year Area sources > 1 ton/year Total Number 174 185 369 371 460 Emissions (tons/year) 37,988 12,908 50,876 3,872 2,167,835 2,171,707 79,832 121 ,213 201,245 7,244 216,613 223,857 10,150 104,161 114,311 Percent of Point Sources 3.1% 0.2% 25.6% 9.2% 4.0% Percent of All Sources 3.1% 1.1 4.1% 0.1% 56.3 56.4% 18.4% 28.0 46.5% 2.5% 73.4 75.9% 2.9% 29.3 32.2% Remaining Point Sources as Percent of Area Sources 75 % 0.2% 39.6% 3.3% 9.0% 128 ------- • Freeways and expressways are high-volume, high-speed high- ways designed to carry about 12,000 vehicles per lane per day at speeds of 35 to 50 miles per hour. All cross traffic is carried on separate levels, and there are no traffic signals. • Principal arterials carry large volumes of traffic between different parts of the urban system and connect to the free- way system. Typically, they carry 6,000 vehicles per lane per day at average speeds of 25 to 35 miles per hour. Minor arterials carry traffic between the principal arterial net- work; they alsc serve as local streets. They frequently carry up to 6,000 vehicles per lane per day at speeds of 20 to 30 miles per hour. • Local streets provide access to all areas. They vary in design capacity. Data were gathered for all freeways and major and minor arterials. Street names were alphabetized, and segments were designated by intersec- tions (not by UTM coordinates). The resulting data are shown graphically on a traffic map in Figure 13, and they are updated yearly by the Missouri and Illinois traffic authorities. 2. Railways and Vessels The U.S. Department of Transportation is planning three trans- portation studies in the Saint Louis area, as a part of the Saint Louis Air Pollution Studies. The goal of these studies is to ascertain the air pollution attributable to railways an/3 vessels in addition to quantifying the emissions from specific transportation systems and facilities. The studies are being coordinated with EPA's Regional Air Pollution Study. The first of these studies deals with waterborne vessels, which constitute a portion of the mobile pollution sources of the Saint Louis area. A comprehensive listing of sources, such as tugboats, tankers, recreational vessels, and so forth, will be organized by engine horsepower. The frequency of usage will be determined, and the pollutant emission cal- culated using appropriate emission factors. If the data appear to warrant additional investigation, increasingly accurate inventories are planned according to the scheme shown in Table 21. A similar study is planned for rail operations' in the Saint Louis area. Rail activity in the region has been studied for many years. The Transportation Systems C'enter, DoT, recently estimated annual surface ------- -«/5^t<'if'" fe"l .,,„ »,, ™^?° • "....„, ,-«,,.-/ , ""-;•<' -%^'/ ' ' no v-,.-' ' "F,,,««-:"-p^--'4,j0 ' 15, X CH,^, _ V^yITM -^^r;^ MISSOURI STATE HIGHWAY DEPARTMENT OFFICE OF PLANNING U S DEPARTMENT OF TRANSPORTATIO FEDERAL HIGHWAY ADMINISTRATION BUREAU OF PUBLIC ROADS 1969 TRAFFIC MAP ST. LOUIS METROPOLITAN AREA INTERSTATE AND FREEWAY SYSTEM ( i "ttiMt0" """"** SA-2579-18 FIGURE 13 1969 TRAFFIC MAP OF SAINT LOUIS METROPOLITAN AREA INTERSTATE AND FREEWAY SYSTEM 130 ------- Table 21 CLASSIFICATION OF POSSIBLE EMISSION INVENTORIES , Type Inventory Based on published data Rapid survey Comprehensive Extensive field survey Source Description Exhaust averages from literature Exhaust averages calculated from opera- tors duty cycles Exhaust and evaporative averages from tests Individual exhaust and evaporative tests Position Resolution Region Terminals Mainlines Fine grid Fine grid Time Resolution Yearly average Weekly/daily average Weekly average and possible daily maximums Weekly averages and possible hourly maximums Source: "Air Pollution Guidelines," NATO Committee on Challenges of Modern Societies (1971). 131 ------- Table 22 1967 EMISSIONS FROM RAIL OPERATIONS WITHIN 100 MILES OF SAINT LOUIS Pollutant Particulates S°2 CO HC NO X Aldehydes Organic acids Total 103 Ib 97 253 273 195 292 16 27 1,153 Source: J. C. Sturm, Railroads and Air Pollution: A Perspective,' Report No. FRA-RT-73-33, U.S. D.O.T. Federal Railroad Admin- istration, Washington, D.C. (May 1973). freight transportation exhaust emissions within 100 miles of Saint Louis. These estimated rail exhaust emissions for 1967 are shown in Table 22. The rail source emission inventory will describe • The quantity of emissions by source type • The location of the sources The time pattern. As in the case of the vessel inventory, a series of inventory methodologies will be used (similar to the ones shown in Table 21) start- ing with a review of published data on rail operations and standard 132 ------- emission factors. Subsequent inventories with increasing refinements in location, emission, and time factors are planned for those pollutants whose emissions appear to be of sufficient magnitude to warrant further investigation. A separate study of emissions from Saint Louis railway terminals is also planned. This study will concern itself not only with emissions from locomotives, but also with emissions from stationary sources at terminals, such as power plants, fuel storage, incinerators, and with operational factors, such as freight handling by trucks or cargo spillage. D. Summary The requirements of accuracy and spatial and temporal resolution that the RAPS emission inventory should exhibit are not met by any of the existing inventories. The first four inventories—the IPP, IBM, DAQED, and NATO emission inventories—are essentially of only historical interest. The NEDS inventory contains the best current estimates, and it con- stitutes a useful data base for those emission sources that, for prac- tical reasons, cannot be individually measured or closely simulated by emission modeling. It can serve as a base for the determination of the scope of the task, since it details the source size and the descriptions of location, normal operating schedule, source classification code number, and stack parameters. It certainly could meet the requirement for long-term data on a wide range of pollutants. For model verification purposes for a limited range of pollutants, the content of the NEDS inventory is, however, inadequate since it only gives annual emission data and does not provide sufficiently detailed information for the determination of hourly emissions of suf- ficient precision even with the application of appropriate models. Data on traffic and transportation are in part contained in the area source emission data of NEDS. For freeways and major arteries, which will be treated as line sources, these data will have to be supplemented by information on the traffic mix, average speed, type of roadway, and other factors. 133 ------- VII TASK E: EMISSIONS MODELING A. Introduction The specification of the rate and distribution of emissions is basic to any air quality simulation model. However, emissions per se have rarely been measured in modeling studies conducted prior to RAPS; usually, emissions are parameterized on the basis of available emissions-related data. The so-called emissions model is thus a methodology that relates available data to the required mass flux density of pollutant emissions. Depending on the type of the emissions source and the requirements of the various air quality simulation models, emission models are con- veniently classified according to the mobility and configuration and type of the source. Figure 14 is a three-dimensional matrix illustrating some 64 possible combinations of emission features that describe various emission model types. These are only broad classifications, and the number of real possibilities is considerably greater. Referring to the discussion of Task B, for example, we note that there are at least four major categories of gaseous pollutants to be considered: carbon monoxide (CO), oxides of nitrogen (NOX). hydrocarbons (HC), and sulphur oxides (SOX) . Also, particulates may need to be classed as large (> 2(am) or small (f 2\|^m), while mobile surface emissions can be subdivided similarly as gasoline and diesel powered roadway vehicles, trains, on-ground aircraft, and so forth. The saving factors are; that not all combinations exist in the real world, some are of no major concern, and others can be grouped and treated in a single approach. Other constraints also limit the task of specifying the input require- ments and the structure of emission models likely to be required in the RAPS. Many such constraining factors have already been discussed in other task reports. Emission model requirements are determined by the specific requirements of simulation models to be evaluated in RAPS. For example, the minimum emission-time resolution for most model applications is specified as one hour. (We do consider, later in this section, sub- hourly specifications for vehicular emissions.) Spatial resolution has also been addressed in Task B (Section IV), but an objective methodology for evaluating the required geographical resolution of major point sources is presented in this task report. 135 ------- < CO \ O H CC U u. o u LU u CC D LU 5 D O CC < 01 Zi POINT ° LO CO > u "~ S02 FROM POWER PLANT CO D O LU CO FLY ASH FROM POWER PLANT HI H _J O 1- cc Q_ SENSIBLE HEAT FLUX FROM STACK \- in X WATER VAPOR FROM COOLING TOWER CC LU h- g Q O CO to oc O or I— < o CO CO U LL O z. o I- cc h- co _i _i u LU I o CO LU CC D O 136 ------- B. Modeling 1. Review of Existing Models An extensive and in-depth review of the available literature on emissions modeling resulted in the identification of 28 papers and reports deemed significant in the context of this RAPS study. Appendix B summarizes these models by listing (1) the author's abstract, (2) our summary of the significant features of the models, and (3) the form and availability of documentation. Table 23 lists the models reviewed and summarizes the models according to emissions type, source mobility and configuration, and the degree of experimental verification. The last aspect is particularly important to RAPS where appropriate short-term specification of emissions is essential to the overall evaluation of air quality simulation model performance. In this regard, it is sig- nificant that only three of the 28 models reviewed have undergone in- dependent experimental evaluation of the emission estimates. While Table 23 provides an overview of the emission models, Table 24 lists details of the input requirements and outputs of selected emission models. Generally, these models are the more comprehensive ones reviewed; as such, they reflect the state of the art and outline the facility and process parameters needed to simulate emissions from the smaller sources (discussed under Task B). In Table 24, the input requirements of the various models surveyed are subdivided into model inputs and model parameters, two categories with subtle distinctions. The attempt was to list as model parameters those data that are essentially internal to the emission model in that they are usually specified only initially. Of course, they may need to be revised periodically as updated or more representative values become available. Model inputs on the other hand encompass operational inputs required to actually run the model and generate emission rates. The classification system should not be viewed as rigid or formal, but rather as a convenience. The discussion in Section IV points out many of the inadequacies of emission modeling techniques for use in a research program focused on the development and evaluation of air quality simulation models. In the past, emission data have been derived primarily from long-term average emission estimates--especially for the stationary sources. In turn, most of the long-term data have been derived not from direct measurement, but, by applying emission factors to survey information (e.g., fuel consumption). Hourly emission estimates (and finer spatial resolution) have then usually been derived from simulation techniques surveyed in this review. The resulting aggregation of emission factors and correlation and algorithms 137 ------- a w £ > w C£ V! o J} 6 « M CM C/] 3 i 13 T3 [H H O H H I W O S <£ S w UJ -o Q ! 1 « D ^5 Ci S 3 C 0 ^ Emissions Type 1 c w CQ c 0 H H Li T3 t-t Hi e 0) S D- X < j; c x; o o S c 2 xT CC SJ Li C 2 c o S 3 O c 0) rt O o +J a r~i Ll S 3 3 X XX X X ' 1 , ,. ., 1 X, X XXXXXXXX XXXX XX X XXX X X X X X XXXX X XX X XXX X X X XX XXX XX XXXX X 1 XX XX 3J C 0) H > 1 o c O be -P bo c c c r^ C C (LOJCO -t->H OJ -3T inrH Li C WWW flj H H p-( i-i M D, rH Li fn H Li 0 O Li-pppoLi C 03 ^^ O- --1 -s'H-i «i; ui IB m -, u 0 SH X)j3 r^ca LI he ccra ccd-n u cuo a a S 0 s c-j-pr-t-Hi-i^C-M ^^3 =j=j=^p- ccoc^cccioiaiaaj >, >, z: H M C """ C "3 -C -C Ll H H S ce -fl^i cs-M c. :jHHLieoco6=p-' LI LI # CC Li Lid) C< ------- CJ J -i C M 3 C. i C C d -H -I ,C < .n S < < < Q. 0. O ;£ !-• S E < < ;£ m O QJ J3 CL, 139 ------- 5J C. l > c J ,-t •p c 0> Jl JJ C ^ M i a >, n S-, 0 n o 3 i) Q O (tf ^ =- t, M O 0 o 1-1 a C H 0 3 o. a -HOG -" ai c a 3 6 Ji > 0 M iH O U g. 1 '" I - 140 ------- -CO* be a* H 3 S w C 3 3 T3 1) (Li 3 rH o -a ta (X w C 0> O E 3 0 d •P o n o a o in O c CS be 01 c 3 1 tr 0 U J2 ?• 0> 1 o ,4 « 3 H Jl h ~a c t i •° 0 in E 141 ------- C 3 6 g .fl ------- G C O £ C a; Q, -• o .0 « m E 0) ,/• <-, O D C OJ O C O rn a; a -o a 3 -o C CJ Q O J -O h ^ h BJ 3 i» C DSUEOS03 H P 0) -I-1 £ J g i-l O J HJ ~ 143 ------- 01 01 "tj g CH 11 •D £ 01 C 01 u bj 1/1 w c. +J o 0) ^ 1> 1 £ o o E Ll 3 T3 -n c 01 n -H Q. C be +j 5 L. ^ (5 ra 01 -i ii C t-> o •>-> H TO K "" E ° £ C T3 '•H CO !H -Q LI O O D E Q. >, O -i 01 en ~- H n •o^££jaHa * 3 < < [/} H c-1 M 0 H c § c >•> O P-H T3 3 Oi 0 U] X CJ M 3 'c O TO a. 01 g -. 3 01 C 0 > CX 0* 01 L, U £ 3 a •Q u 3 C +•» c U '-' u: "H ^ 1) L, L, j be oJ; T3 0 T3 i! O H C U rH OJ CO Ll 01 -a tn 01 L, o L, 01 O Li Li ^ 0) tc ^: DC cj^^x uga >. S -M o E r a — ' >i O p-t C 01 O -1 'M C H 0; ^ >H-QO >vL. c -H 01 O Jd H 5) 3) ^ !H (1) (D cuag cue g J Cn t>, J O t- H a> e C C C 0 C 0 tn in D H H a c Li Li C r2 rH 01 3 3 01 1 1 i Oi O ^ jtf p2 n c 3 H H U M r-t H -C ca ca > HE c _ u 0 0 (S* ^ ^ Ll "~ -. g 0 § IE U IE - S 0) 41 U U Oi 0> L. L, 0) 01 01 D O> D HH 0 —01 f EC w ce w os 03 ^ CO ^ H P 1) 01 P a o >. u ^ di c o £ a 1 c 0 o 3 •a 01 c u ra L! r-n 3 O O C - O C •- 1 1 o u c c Ul E 0) 01 § 0) 3 O" d Ul 0 O .3 TO O 0) 0) X L. 01 Oi L. CC > (D H 01 •a C 0) O Li Q, 3 - 2 1 C 0> hti •o 01 rj -> 01 M " £• 2 " D p > li > •t-1 11 H GO n p, -M aj ,= o U 3 fl -" -O 0 O O 3 H r- 1) C C p-t E -t-1 (I) L- 0 (C - 1-1 G E 0 5 •3 B S C Oi TJ a LI c ac P 01 01 ^ S a o so O (U 0) u (fi a. •a e m 01 0 c T-> 2 01 a c •° u ^ u C Q o a S "^ S ^ 1-, Li O Li >, < "t-t M Q; -a Lt H -OK -J L. C ^ E ^ " C 1 ? L, C Jl . LI o n a w H f>- O C c •o c * B 0) QJ CC £ 01 " ^ r~ O r- Z- D. 51 e a ^ cr >. 1) n P O S r£ 41 a 3 0 o 1> a ^ Li 0 a PI 3 01 •D D 3 0 bl f\|H O Li < 3 O a 0) E C H Opcrat •H S E n. cr 0) •o OJ 1 c 0 u 01 3 Li O 01 L, 0) C Oi 3 03 to Q. 0) B C O tfi £ 0) t« O C £ L, a 3 T3 Btu re con sum c g Q O C -H C3 £ U a 0) D, 0> U ^ 01 >, CJ c 0 i a CJ QJ C -H D U 01 H 01 > ^ 4H ~0 0 1 Types Number _^ a r< w p ^ ^ 01 i- p-< CO u -c H C £ 3 0) 0 01 O •o c QJ bC C 0> •Jl a 3 o J3 Oi a Oi E C H Li 01 P. O 144 ------- for the simulation of high resolution spatial and temporal emissions is, thus, defined as the emission model. Because of the uncertainties and potential for error in these methods, an emission inventory system for RAPS that relies heavily on measurement techniques has been proposed. However, two areas have been identified where the measurement approach is not feasible and where the use of simulation techniques is recommended. As discussed in Section IV, small stationary sources fall into this cate- gory. And it will be shown here that the magnitude of the mobile emission (i.e., roadway emission) network requires the use of emission simulation techniques together with some special measurements. In spite of the limitations of many of the emission models, we have sought to identify those models that are appropriate to the treatment of small stationary sources and mobile sources. We have also included models that have heretofore been used to simulate emissions from the larger point sources; without the special, direct measurements recommended in Section IV, these models would provide the best (though admittedly coarse) emission estimates for these sources. Having thus placed in perspective the role of emission models in RAPS, we discuss below a cross section of models and the needs they appear to serve. • The Argonne Model of Roberts8* for the simulation of hourly SO2 emissions from distributed residential, commercial, and institutional sources, and for hourly SC>2 and thermal emissions from major point sources. • The emissions model developed at Systems Applications, Inc. (SAI), by Roberts1;3 for hourly stationary point and area source emissions of NOX and hydrocarbons. This model is also recommended for the simulation of vehicular NOX, CO, and hydrocarbon emissions subject to one reservation. The SAI aggregates vehicular emissions into two square mile area sources—a feature that may be too gross for direct application to RAPS. Therefore, we suggest that the SAI model be modified to provide link-by-link emissions for the major links in the manner of the Stanford Research Institute (SRI) model of Ludwig4 while retaining the spatial averaging (on the smaller scale of one square kilometer) method for the secondary links. * Model reports are reviewed in Appendix B. 145 ------- • The Geomet model6 for diurnal emissions from river vessels and railroads. • The Platt's model for Northern Research and Engineering,6* for aircraft emissions. Several categories of emissions are notable by their absence from the above listing. To the best of our knowledge, there are no models available that treat heat emissions from vehicular or distributed area sources although the inputs to such models already seem to be re- quired in the SAI, SRI, Geomet, and Argonne models. However, appropriate thermal emission factors are required. Another shortcoming of the available models is the treatment of particulate emissions. While four of the 28 models reviewed treat particulates, none provides output on size dis- tribution. Furthermore, very little has been done to develop a regionally applicable model of man-induced water vapor emissions. To be sure, the knowledge and expertise are available to develop such a model, but a rigorous and comprehensive methodology has not been compiled. Perhaps the closest facsimile to what is required is the work at the University of Alaska by Benson20 for Fairbanks, Alaska, where water is considered a primary atmospheric pollutant in winter. Thus, we recommend the use of an aqueous emissions model, similar to Benson's, but expanded as dis- cussed later. Lastly, the emissions modeling work of the Ontario Department of the Environment for Toronto is perhaps the singularly most comprehensive, available methodology; it is not, however, generally available. We suggest that official requests be submitted to the Ontario Department of the Environment since many of their methods and experiences might be of benefit in the development of the RAPS emissions inventory. 2. Specification of Emissions The basic requirement of the RAPS emissions inventory program is to provide hour-by-hour emission rates for the various pollutants and heat and water vapor. The pollutant emissions are of primary importance * This model is currently being modified by Geomet under Contract 68-02- 0665 with the Meteorology Laboratory, Environmental Protection Agency, Research Triangle Park, North Carolina. 146 ------- (they specify the basic emission inventory requirements), while heat and water vapor can be determined from data acquired in the course of listing the pollutants (see Task B). For stationary sources,, there is a three-stage sequence in which emissions should be specified in RAPS. (1) Direct measurement of emissions. (2) Measurement of inputs to the pollutant-generating process with subsequent parameterization of the input-output (i.e., emissions) relationship. (3) Parameterization of both the inputs and input- output relationship. An example of the first would be the use of in-stack monitors at major power-generating facilities. The second method would employ measurement of process parameters directly related to emissions (e.g. fuel consumption rate), while in the third case these parameters might be determined from a statistical base or as a function of a known indirect indicator (e.g., degree-hours). Prior to RAPS, regional studies employed the third method almost exclusively. In RAPS, however, greater emphasis must be placed on the first and second methods. a. Stationary Sources Direct monitoring (input or output) of all stationary sources is not practicable in view of financial constraints, nor is it required; hence, there is a need for emission models. The available inventories for Saint Louis have been reviewed in discussions of Tasks B and D to identify the number and size of the major point sources of SOX and NOX (CO and HC are emitted primarily by distributed sources). A major point source may be considered as one that receives individual treatment in an air quality dispersion model, not one that is aggregated into an area source. About 13 point sources (e.g., power-generating plants) have been identified as "principal" major emitters with emission rates far in excess of "secondary" major point sources. As a first step, we have proposed (Task B) routine emissions monitoring for all principal sources and routine input monitoring for secondary point sources. Routine input or emissions monitoring is not required for the remaining individual stationary point and area sources. The best available emissions models will serve adequately. 147 ------- The types of input or emissions measurements recommended above for various stationary sources are those necessary to provide the accurate and current emissions inventory required as input to the oper- ation or evaluation of mesoscale air quality simulation models. Where the intent is to investigate microscale phenomena,, the basic or routine inventory will need to be supplemented with special emission inputs. Depending on model receptor locations (or the location of air quality monitoring stations), it may be necessary to provide higher resolution emissions data for some of the lesser sources. Thus, for example, minor point sources near a particular receptor may either need to be monitored continuously or, at least extensively, during selected periods. More specific requirements cannot be set until actual monitoring sites are determined or specific research studies defined. This is especially true for emissions from mobile sources; further considerations along these lines are presented later in this section. b. Mobile Sources For routine use, the mobile emission inventory should be estimated for link or line source geometry using an average route speed model formulation such as that used in APRAC-1..31 Links are straight segments of major arterial streets and freeways that serve as the trans- portation network for the area. In most urban complexes, this kind of network has about a one-half to one-mile span in suburban areas and a one-block span in downtown areas. The link geometry is recommended because it is conveniently related to the parameters needed to estimate emission, links can conveniently be aggregated into area sources (but not the reverse), and control strategies to be investigated later will deal with modifications of the locations or conditions on the links (not the areas). The mobile emission inventory is thus envisaged to comprise three distinct elements: (1) the basic roadway network, (2) traffic characteristics, and (3) an emissions model. The first element is a listing of the characteristic features of each link in the roadway net- work, such as • Link location (geographical location of the endpoints). • Facility type (e.g., suburban freeway, downtown arterial). • Roadway characteristics (e.g., number of lanes, grade, elevation, foundation). 148 ------- Traffic characteristics are those parameters that describe the vehicle population and traffic flow. In this category, the inventory must con- tain data on • Vehicle mix by age and type (e.g., numbers and ages of light duty vehicles and heavy duty diesel and gasoline-powered vehicles). • Vehicle speed (by facility type and location). • Distribution of origin-destination (0-D) locations. Because vehicle emissions cannot be measured directly in situ, they need to be determined with the aid of an emissions model using traffic and roadway data in the manner of Roberts' model2 or Ludwig's model.4 (A number of other sources discuss the requirements and availability of roadway and traffic data. )s2 > 3S? 24; 2Bj>36 As indicated above, we recommend the use of the average-speed emissions model on a link-by-link basis as the primary mobile emissions format; in this manner, area-averaged emissions can be obtained appropriate to the dispersion model structure or individual- link data could be retained for the larger roadways for line-source models. We further suggest that O-D data be used to identify cold-start emission contributions. Lastly, a multimodal emission model should be used for individual microscale studies. But the collection of the appropriate data and the operation of the model to generate emissions data should not be a part of the basic inventory; rather it may be considered as a separate task of the individual miscroscale study. Lastly, we note that the use of multimodal models19 is still in a preliminary stage pending the availability of emissions test data. The average route speed model requires inputs of traffic volume on the links and link speed. Diurnal traffic factors and emission factors for the vehicle mix on the link are also parameters of the model. There is enough similarity of utilization (volume) on most urban streets that estimates based on periodic samples, supplemented by data from fixed sensors on high-volume freeways and on selected arterial streets will serve the purpose. These periodic samples are routinely taken by the traffic department as part of monitoring the adequacy of the street system. The volume data from these surveys are expressed in average daily traffic and require a diurnal pattern to convert the daily into hourly traffic. For this reason, monitors capable of hourly or finer resolution should be installed on typical streets representative of different usages: arterials at various distances from the center of 149 ------- town, arterials with various functions (downtown distributor; suburban collector; shopping center, university, or airport service). These different functions may imply different diurnal variation patterns and, hence, should be monitored. In addition, fixed monitors in such places will help to determine the variation by day, week, or season; from these data, factors for similar facilities may be developed and used for the emission models. As in the hourly variation patterns, speed (and hence emissions) vary with volume of traffic, but speed distribution will be similar from day to day. Speed data are needed for the same kinds of facilities as those described above, and both fixed and portable monitors capable of providing speed information should be used. Special consideration should be given to the inclusion of cold-start emission effects. Martinez, Nordsieck and Eschenroeder27 have evaluated the comparative impacts of cold- and hot-start emissions (vehicular HC, NO, and CO) for Los Angeles in the time period from 0600 to 0900. Their findings are shown in Table 25 expressed as the ratio of emissions with cold start to emissions without cold start. Table 25 RATIO OF COLD-START EMISSIONS TO HOT-START EMISSIONS Species NO CO Reactive hydrocarbons Year 1968 1.15 1.48 1.08 1971 1.16 1.57 1.10 1974 1.16 1.53 1.11 1980 1.01 2.30 1.11 We see that the neglect of the cold-start emissions con- tribution would have a moderate impact on the estimation of reactive hydrocarbons and a severe impact on CO by 1980. The emission inventory must therefore provide cold-start emission factors. Martinez37 provides values that may be used for light-duty vehicles through 1976; these are summarized in Table 26. 150 ------- Table 26 COLD-START EMISSIONS FROM LIGHT-DUTY VEHICLES ;_ 1968 1969 1970 1971 1972 1973-74 > 1975 Cold-Start Emissions (gm/start) N0x (as N02) 5.66 6.56 5.89 5.36 4.02 3.78 -1.04* HC 10.73 9.54 10.88 6.18 4.25 4.25 2.68 CO 254.1 143.6 143.3 92.53 76.74 76 . 74 43.06 The negative factor is, of course, used in con- junction with the running emission factor, thereby effectively reducing net NOX emission. Under special conditions, as when forecast values of air quality are compared with measured values, continuous and detailed mon- itoring of vehicle behavior in the immediate environment of the measure- ments may be required with emissions being forecast with a multimodal emission model such as Ludwig's (reviewed in Appendix B). Such monitoring would not be done area-wide or continuously, because it is not appropriate to mesoscale studies. The multimodal model contains emission factors for the standard vehicle maneuvers: steady driving, slowing and speeding, stopping, and accelerating from a stop. Hence, the resolution of the model in space can be adjusted to the location and needs of the experiment. That is, a downtown block might be divided into several links, each a few tens of meters long. Vehicle modes such as acceleration, steady driving, congested driving, or stopping, would be assigned to each link. Volume in each link would be measured on a time scale comparable with the needs of the experiment. 151 ------- Special microscale studies of vehicle-generated emissions may require subhourly inputs of vehicle parameters. This is especially important when the distribution of inert pollutants in heavily traveled areas is evaluated (e.g., CO in the central business district or in the corridor of major freeways). For such situations, 15-minute resolution may be necessary. In the earlier discussion on special measurements of stationary emissions, specific details of special mobile emissions re- quirements must await formalization of the plans for individual research studies. c. Water Vapor Emissions A special aspect of the overall emissions modeling program (both for stationary and mobile sources) is the treatment of aqueous emissions. As noted earlier, a comprehensive regional model for water has not been developed although emission factors are available at various levels of refinement. For example, Benson20 presents emission factors for water (and carbon dioxide) emissions from the combustion of gasoline, fuel oil, and coal. For gasoline and fuel oil, emissions of water are readily specified on the basis of spatial and temporal distributions of fuel consumption. These will generally be available directly from the basic inventory of the other pollutants (e.g., SO2, CO). Two exceptions may be surface vehicles and major point sources. For the former, the link inventory will provide facility type and vehicle volume, mix, and speed information; additional factors (or submodels) will therefore be required to obtain gasoline consumption. Direct monitoring of water emissions from large point sources may not be the optimum method. Rather, grab samples as discussed in Section IV may be more appropriate both be- cause of economies of the method and because of relative invariability of the exhaust water content. These would then be related to the more available information on fuel and process features. The man-induced emissions of water from air conditioning (ac) units, small cooling ponds, sewage treatment plants, and so forth should be aggregated into gross composite emissions for specified areas of the order of one to two kilometers on a side. For ac units, statis- tical relationships based on an inventory of ac capacity and type and the ambient conditions of wind, humidity, and temperature would be adequate to estimate the corresponding emission rates. Similar procedures could be used for cooling ponds and the like; for sources of this type, the inventory should contain information on the area and volume of the ponds, input water temperature, and degree of mechanical mixing. The emission rate is then determined for the appropriate time and location using an analytical or statistical model of potential evapotranspiration. 152 ------- However, rather than calculate the emissions of water from the measured or prescribed inputs, we recommend that such "meteorologically-controlled" aqueous emissions from man-induced sources be determined outside the RAPS emissions inventory. We view these sources as those in which man's activities provide the potential (i.e., a water supply) for aqueous emissions to the atmosphere but in which meteorological processes regulate the magnitude of the vapor flux. In summary, aqueous sources in this category include the following: • Cooling ponds • Sewage treatment ponds • Irrigation • Space cooling units. Lastly, anomalous sources of aqueous emissions, in par- ticular, wet cooling towers with their enormous outputs of water vapor and their significant impact on visibility, gaseous reactions, and such, must be considered. For example, moisture discharge rates for very large power plants (e.g., 2, 000-MW fossil steam plants) are of the order of 1.5 x 10^ Ib/min—a rate equivalent to the daily deposition of a 1-cm column of water on a lO-km^ area. To simulate these emissions accurately, the inventory should contain the design performance standards and power generation rates for each source. The environmental parameters required in the inventory are the ambient wet bulb temperature and relative humidity. Figures 15 and 16 illustrate performance curves for a natural-draft cooling tower that dissipates 8 x 10^ Btu/h; note that the performance curves are also used to estimate the sensible heat output of the tower as required for the thermal emissions inventory. 3. Emissions Model Verification - Mobile Sources In view of uncertainties in determining mobile emissions, an experimental program of emissions model verification is advocated. The uncertainties arise from imprecisions both in the specification of process input parameters and in the accuracy of the postulated input-output relationships. Recalling previous discussions on the nature of emissions models, we may conveniently define two principal types. In one, the input (I) is known as a function of time (t), and its relationship to the emissions is 153 ------- 110 LL 105 LU § 100 tr m Q. 95 * 90 85 45 50 55 60 65 WET BULB TEMPERATURE SOURCE: Bowman and Biggs, 1972. 70 — °F 75 SA-2579-6 FIGURE 15 PERFORMANCE CURVES FOR COOLING TOWERS GIVING THE EXIT TEMPERATURE AS A FUNCTION OF THE AMBIENT WET BULB TEMPERATURES FOR VARIOUS EXIT RELATIVE HUMIDITIES 154 ------- 160 110 45 50 55 60 65 WET BULB TEMPERATURE SOURCE Bowman and Biggs, 1972. SA-2579-7 FIGURE 16 PERFORMANCE CURVES FOR COOLING TOWERS GIVING THE MOISTURE DISCHARGE AS A FUNCTION OF THE AMBIENT WET BULB TEMPERATURE FOR VARIOUS EXIT RELATIVE HUMIDITIES 155 ------- specified by an input-dependent emission factor (EF); the functional form for the specification of the time-dependent emission rate E(t) is E(t) = I(t) • EF(I) In the second type, the input rate is also unknown and, hence, a time- dependent input factor (lF(t)) is also required. The emission rate for this case is then given as E(t) = I • IF(t) • EF(I) Implicit in these formulations is a knowledge of the spatial distribution of the inputs. Mobile sources (and certain stationary area sources) are especially complex in that the time-dependence of the input factor must be expressed as a function of location. Thus the so-called emissions model may actually be composed of three components: (1) the apportionment of inputs on a temporal basis, (2) the apportionment of inputs in space, and (3) the relation of emissions to inputs. Similarly, the field (i.e., in situ) verification program also must serve to evaluate the emissions model performance in these three areas. Parameterization of the inputs is particularly relevant to the generation of the mobile emissions. The recommended procedure is to use a Statistical database (e.g., ADT) together with continuous measurements of vehicle volume, speed, and mix on selected key links to generate a dynamic link-by-link mobile inventory. Evaluation of the adequacy and representativeness of this procedure should be undertaken. Two methods appear feasible for this purpose: (1) aerial photography, and (2) side- looking aerial radar. The first method could easily provide spatial inventories of vehicle volume and mix by facility type, location, and time of day; less easily obtainable, although possible, is the deter- mination of vehicle speeds. The second method can provide volume and speed distributions, but not vehicle mix. Again, these evaluations are recommended for selected hours, days, and seasons; we do not advocate continuous usage of either method throughout RAPS. A similar but less difficult problem is the parameterization of temporal and spatial variations of fuel consumption by distributed area sources. 156 ------- ...i^sion !-u ,'Vdora] dv . s<.ruti >i,< ios shiniLi •-. \,.- ! cept of the mass >ud^> ; , i the field can be ; : , ; •.. i cent are requit ( {-.- into and out of . •. <>ni. •. • to the pollutant .-out : Q(gm ni~l sec"1) i ',ro the top (H) i^, gi'-i-n ,i. ' to evaluate parameterization • ilready discussed treatment iaiter, it remains to verify /n a p<;/'-l ink basis. Both :,r, ht- ii.->e(i. and both are '•vie (FDC) . What is not • il< , 1) aj:, and so ,.l<.ri ..ki ;i to evaluate the iyj>t-r. (such as freeways^ r-) in congested areas. •-\ • i 1 iinie, mix, speed, and ii'uld be del cri dned as o.i Lin- J ink. A mass The con- ntation in (' to 20 per- il i e I f 1 a v i a Iux is equal Horizontally density i ilux through Here, x^2-^ ~'L" t^lt ^cJ 1 and 2 refer to the determination of the the following: ii i'•• iiiv; c>uuscripts J CK .1 t ii>n . Thus, i rate) requires X7aluf mounted on i'j M (. .• r sensors may be riu,.. capabilities, bin ih. < i i i I OP , Remote ... 11)-averaging .-a : i ' , iently ------- advanced for this purpose. One possible exception is the correlation spectrometer which may have application in the evaluation of emission rates from elevated point sources. In this case (i.e., tall stack), the pollutants are emitted into an atmospheric regime where the wind shear is small. Hence, the mass flux equation may be rewritten, where Q a- u jffy (z)dydz - (T^ (z)dydz) J J Al J J ^2 ( The correlation spectrometer provides the vertical mass loading; cross- wind (y) traverses up- and downwind of the plume can thus be used together with the mean transport wind to determine Q (gm sec"-1-) . In conclusion, we stress the need to evaluate the parameteriza- tion of both inputs and subsequent output emissions by models, thus ascertaining the representativeness of the parameters that are input to the inventory and evaluating the precision with which the emissions are subsequently modeled. 4. Resolution of Source Location A major practical concern in developing an emissions inventory is the degree of resolution required in specifying the location of the point sources. Normally, point sources are resolved to 100 meters in operational inventories; however, subsequent errors introduced in predicting concentrations from such sources may be intolerable in a research study. A 10—meter resolution in source location has been proposed. Before adopting such a specification, it seems desirable to objectively evaluate errors in the predicted concentration field resulting from a mislocation of the source. Perhaps the most difficult task in the evaluation process is the definition of a meaningful measure of the error (i.e., the difference in predictions with and without an error in the source location). We recognize that this error can be a function of many parameters: source strength, stack height, lateral and longitudinal receptor position, and atmospheric stability. For this evaluation, a worst-case meteorological situation should be used in order to determine the most stringent source- location resolution required. However, a single stability category cannot be used to satisfy this criterion as the worst-case stability for a low stack can differ from that for a tall stack. We have chosen, there- fore, two stability regimes—slightly stable (Class E) and slightly unstable (Class C). Source strength and stack height will be seen to be 158 ------- inputs in the assessment features are retained. ' natives exist: (1) ih. i a particular source .<' air quality monitor!"- impact of a partiou],.v- we define, as tin m- u both the ] .it era i • •. . longitudinal d i ^ • ••• or to the do, iiv. i >. : r. Hi individual source location, two alter- '..i/i'or associated with ' ocation (e.g.^ • . to f /a lua te the i-oi;'ion. Therefore^ iii'' 01 { ii<; error over . •. ii'U ' d , rccti ons ; the citht - »• to the receptor source is (.-(ai i v,. similarly mi.slocn ' . . the absolute valu. ot correct location ^nd tance away. A ba^ic f is assumed ,' . • • .. i , • •!) of a : . , '.1 ' ir -A <•)•(' - ••• .1 t N i; , t i or to be • u 01,1 rot ion y at the iit -1.1:10 <;rosswind dis- r atirospheric dispersion 11 I ) I I '-1 (.it v i a t ion i , :• ; < -i t L velv ------- b a = ax . y d a = ex . z a, b, c, d = experimentally determined constants dependent on stability conditions. x = distance along the mean wind direction from the source (m). We define the integrated measure of error, IE, as = JJ U -XE\| x y Since Q and iT are constant over the integration, it is possible to define an integrated normalized concentration error, INE, which is independent of these quantities. (IE) = — (INE) . (2) TTU To determine the allowable source location error, Ay, for known values of Q, H, u, stability and downwind distance, Equation (2) is solved for INE, which is then used to obtain Ay from Figures 17 through 24. How- ever, Equation (2) contains two unknowns, IE and INE. In order to solve Equation (2), a value for IE that represents a tolerable level of concen- tration error must be defined as a reference. The magnitude of this ref- erence value may be arbitrarily chosen; the error limit presented here is specified on the basis of (1) a reference source strength, Qreference» and (2) an allowable source location error for that strength (from which (INE)reference is found using Figures 17 through 24). Thus, IE in Equa- tion (2) becomes IE, the reference integrated error at downwind distance x for a specific source height and stability. Q * reference (IE) = = (INE) . (3) TTU reference Combination of Equations (2) and (3), where IE now equals IE, yields an equation for INE: Q reference (INE) = (INE) . (4) reference Q 160 ------- The tolerable rn:,. • lu 1 : on . y is 17 through 24) on the nn ••-, i > ••( j'\h ! ogt-t source height, and the .s.-U c U-fi dounwim! 20 apply to stable a in->spix'/-K' uruMHon apply to unstable on< t>. -M^IC • ci: >i i . used in the figures, winie .,.,- v.nues arc interpolation mus- b. u^." , ' . • - , -••'• is recommended Ks. i .HI ing in 1 imi t lug t > t- •. ' • ,• ' ' • •n do; i mined (from Figures i'!i atmospheric stability, '. at:. !• igures 17 through 1 !, ! u'uro.s 21 through 24 ., ;>'"i, and 100 meters are i i, :•>', 60, and 100 meters; \ stcp-by-step pro- i ' . : .: ir'u , . \ ';! 10 meters '! •••!<•» , "(' initial mix- Step 1. Select slabil;i height (H), dov. iy re ferrm . unstable), stack , and • ronce I :-i I, ,, -, . i - i . t' lg',1 i «-'b i i Li) i i , • i (i ; I'' 'I . \. IK .: ^ .id 1J1 b is "reft-1 en< <•' Federa 1 Ki-e '..'<, point or' a r. the p res en t > i take _/,' L '• year sou < •_ \ plant st.K.'k : ceptor pr i:i-t . 409 from t1 , .1 ! \ ! ':'< . ' . ) I * ' j )>j-. i I U iiil 01 '•) ( :•! le .. ' • > - . ,'-n, : • I • V'die of f . ' ct. as a • ' • i . .c. ep ted at ineier^, so we shall / an r!5,000 tons-per- . ,v pj.i;d i ' •[ a power •i>nriii ions, at a re- !•• found to be ! •"!('<.• •)U . . i = , (!) then ------- 10a 10* 10J -II Ay = 100 — H = 0 S = Stable — I 500 1000 1500 2000 2500 INE 3000 3500 4000 4500 SA-2579-2 FIGURE 17 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS AT THE SURFACE AND ATMOSPHERIC CONDITIONS ARE STABLE 162 ------- FlGURL !« vAHU!"::\i Ul >N 1 t(.«A 11D NORMALIZED ERROR ,VITH !.; IxiGiTUOINAL DI'^AfJCt- ' MOM THE SOURCE v\ntN THE- ;..,ii)R(;t HEIGHT iS f iVfc METERS AND /MMOJPhEidC. CONDI] IONS ARt" SIABL.E ------- FIGURE 19 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS 30 METERS AND ATMOSPHERIC CONDITIONS ARE STABLE 164 ------- 103 10" 10° 10 H = 100 m S = STABLE ?(' /b 100 125 150 175 200 225 250 275 300 325 INE SA-2579-20 FIGURE 20 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS 100 METERS AND ATMOSPHERIC CONDITIONS ARE STABLE 165 ------- H = 0 S = SLIGHTLY UNSTABLE FIGURE 21 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN THE SOURCE HEIGHT IS AT THE SURFACE AND ATMOSPHERIC CONDITIONS ARE SLIGHTLY UNSTABLE 166 ------- 10° 10" — 10° H = 5 S = SLIGHTLY UNSTABLE FIGURE 22 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS FIVE METERS AND ATMOSPHERIC CONDITIONS ARE SLIGHTLY UNSTABLE 167 ------- H = 30 m S = SLIGHTLY UNSTABLE 80 90 100 110 SA-2579-23 FIGURE 23 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS 30 METERS AND ATMOSPHERIC CONDITIONS ARE SLIGHTLY UNSTABLE 168 ------- 10° 10" 10° 10" H = 100 m S = SLIGHTLY UNSTABLE FIGURE 24 VARIATION OF INTEGRATED NORMALIZED ERROR WITH LONGITUDINAL DISTANCE FROM THE SOURCE WHEN SOURCE HEIGHT IS 100 METERS AND ATMOSPHERIC CONDITIONS ARE SLIGHTLY UNSTABLE 169 ------- The resulting Ay for this INE, obtained from Figure 19, is very small, suggesting that such a large source must be precisely located, i.e., within 10 meters. With a source strength of 1,000 tons per year and other variables as above, Ay is approximately 18 meters. A significant feature of this method is its flexibility: The error measure can be taken to most any selected downwind distance or distance interval. In addition, rather than specifying a single resolu- tion required of all stacks, the method yields a recommended spatial resolution for each source based on the stack height and emission rate of the individual source. C. Consideration of Natural Emissions Pollutant emissions from natural phenomena require evaluation in the development of an accurate inventory for the region. The natural emissions or production of CO, NO, NOg, SC^ nonmethane hydrocarbons, and particulate material in the Saint Louis area are generally small compared with other pollution emissions, but some are large enough to produce a measurable background. There is enough known about the pro- duction of the primary pollutants in the biosphere or atmosphere to estimate the production rates or source strengths within a selected area and volume. For the purposes of RAPS, the Saint Louis source area can be defined as a circle of 35 kilometers radius centered in the business district of Saint Louis. Urban and suburban areas include East Saint Louis, Granite City, Alton, and Belleville (all in Illinois). The area of this circle is about 4000 km2, the estimated vegetated source area is half of this, or 2000 km2. The source strengths of each of the five emission components will be estimated according to its potential production mechanism or mechanisms active in this particular area. Calculated atmospheric concentrations in the Saint Louis area resulting from these emissions must also include background concentrations of the air transported into the Saint Louis air space if a realistic value is to be obtained. Carbon monoxide is produced naturally in the atmosphere by the oxidation of methane.31 Atmospheric methane is largely naturally produced from decay of vegetation. The best estimated global production rate is about 2xl09 tons per year. All of this methane is oxidized in the atmos- phere to produce more than 3x10^ tons of CO annually. The CO, in turn, is oxidized to C02. The 4000 km2 area chosen is 8xlO~6 of the global 170 ------- area; therefore, the average daily CO production rate is estimated at 70 tons per day. There are no known natural sources of NO2 but, NO is generated by vegetation.32 The global NO natural production rate is 500 x 10^ tons per year, from an approximate 1x10^ km2 vegetation covered area. If we use the 2000 km2 area for trees and shrubery, NO production in the region is about 30 tons per day. All of the S0r> produced naturally in the area would result from ozone oxidation of H2S, again generated by the decay of vegetation. The total annual production of H2S from the IxlO8 km2 of land areas of the earth is about 70xl06 tons.33 Using the same area ratio as for NO gives an estimated production of 4 tons of H^S per day. If we consider the 4 tons H2S to be oxidized in situ, it produces 8 tons of S02 per day. The nonmethane hydrocarbons emitted naturally are all terpenes or isoprene. Pipperton estimates worldwide natural emissions of these substances to be 10^ tons per year.34 Hence, 50 tons per day of non- methane hydrocarbons are produced in the area. The diurnal variation in terpene vapor emission rates is extreme; nearly all of these vapors are released from plants during the hours of daylight. There is also a seasonal variation with a summer maximum and a winter minimum. All of the vapors are released through the foliage. Through the year, the hourly rate of emission will vary from essentially zero (winter night) to about 10 tons per hour (summer daylight maximum). Natural particulates in the atmosphere of the Saint Louis region are mostly the product of chemical reaction. The oxidation of S02 to sulfate and NOX to nitrate and the formation of polymeric organic par- ticles from terpenes produce most of the background. The only direct natural particulate in the Saint Louis area is dust. The dust loading will vary widely with wind speed. The average dust loading is 5 (j,g/m^ . The secondary or condensed aerosol from natural sources adds about 15 to 20 (ag/m^ to the background. The yearly global average for dust generated, 200 x 10^ tons,33 is reasonable to apply to the Saint Louis area; thus, the average dust pro- duction in the Saint Louis region is about 25 tons per day. Table 27 summarizes the estimated emissions of CO, NO , S02, H2S, nonmethane HC, and particulates from natural sources alone and all sources in the Saint Louis region. Carbon monoxide, SO2 and NOX pro- duction from natural sources is only a small part of the total regional production and can be expected to have little impact on the RAPS studies. 171 ------- Particulates from natural sources are of marginal significance. Nonmethane HC, on the other hand, may be significant during summer daylight periods. Therefore, we recommend that a procedure be developed for estimating the contribution of natural HC for incorporation in the RAPS Emission Inventory. Table 27 POLLUTANT EMISSIONS IN THE SAINT LOUIS AREA* (Tons/Day) Type CO NO X S0x H2S Nonmethane hydrocarbons Particulate (dust) Natural 70 30t 0 4 50* 25 Total 5100 810 2040 — 790 520 The area is 4000 km2 in size. Expressed as NO. Maximum rate = 10 tons/hour. D. Summary of Conclusions and Recommendations The specific conclusions and recommendations that result from the studies carried out under this task are summarized below: No e\ ."ting models adequately meet the requirements of RAPS; to provide emissions data with the necessary high resolution in space and time, it is recommended that • Direct information be acquired of emissions or of factors determining emissions to the extent possible. • Where not possible (i.e., for small point sources, and area sources), the most suitable of the existing models should be used in adapted and improved form. Specifically, for stationary sources, 172 ------- - Direct data from emissions monitoring should be used wherever available (probably only for the largest sources, such as power stations). - For other large sources, information on hourly fuel consumption rates, process operations, and such, should be obtained and used to derive emissions. - For small point sources and area sources, modeling procedures based upon the models noted below (but, with refinement of the input data wherever possible) should be used. Recommended models are - The Argonne model8 for hourly SC>2 from distributed residential, commercial, and institutional sources and hourly SC>2 and heat from major point sources. The Systems Applications, Inc. (SAI) model1?3 for hourly NOX and HC from stationary point and area sources. For automotive mobile sources, we recommend that an average route speed model be used (with a link or line source geometry) supplemented with measured data of traffic flow from fixed sensors on high volume freeways and on selected arterials. The following models are recommended: - The SAI model1;3 for hourly CO, NOX and HC with modified inputs derived from Stanford Research Institute's model4 for spatial and temporal distribution of vehicle number and speed on a link basis for primary traffic and area basis for secondary traffic. - Recommended for other mobile sources, are the Geomet model6 for diurnal emissions from river vessles and railroads and the Northern Research and Engineering model,5 as revised by Geomet (in preparation), for aircraft emissions. - For bo tii ^ * ti tionarv and mobile sources, the Ontario iJe.artment oi the Environment (1<)71) model for Toronto should be referred to for overall guidance and planning methodology. For microscale studies under the RAPS program the basic inventory will need to be supplemented with special emission inputs. These should be obtained on an experimental or project basis. Detailed data can be acquired from small stationary sources or, for mobile sources, from con- tinuous and detailed monitoring of vehicle behavior in a given area with the application of a multimod-i] emission model to forecast emissions. 173 ------- In view of the limitations of existing mobile emission source models and the impossibility cf resorting to direct information (as is done in the case with major stationary sources), it is recommended that special steps be taken to improve and verify such mobile emission models. Spe- cifically, • An experimental evaluation should be made of mobile emissions models to refine the specification of process input parameters and the accuracy of postulated input- output relationships. • Parameterization of inputs for mobile emissions (statistical database plus continuous measurements of vehicle volume, speed, and mix on key links to generate a dynamic link-by- link inventory) should be verified on a selective basis via aerial photography or side-looking aerial radar or both. • Temporal and spatial variations of fuel consumption by distributed area sources should be spot metered in fine scale. • Parameterization of outputs from mobile sources should be evaluated in various models to verify representation of time-averaged emissions on a per-link basis taking into account types of link and the impacts of deterioration, mix, road grade, drag, and so forth. • Of the pollutant emissions (CO, NOX, S02, H,,S, nonmethane HC, and particulates) from natural phenomena in the Saint Louis area, nonmethane HC may contribute a significant background during summer daylight hours and should be included in the RAPS emission inventory. Suitable pro- cedures to estimate these should be developed early in the RAPS program. • In the context of the RAPS program, the accuracy of point source locations is best considered in terms of the effect of errors throughout the study area. Accordingly, we propose that a measure of error in source location be defined as the integral value of the error over both the lateral (cross wind) and longitudinal (downwind) directions. 174 ------- REFERENCES 1. P. J. Roberts, P. M. Roth, and C. L. Nelson, "Contaminant Emissions in the Los Angeles Basin—Their Sources, Rates, and Distribution," Report No. 71, SAI-6, System Applications, Inc. (March 1971). 2. P. J. Roberts et al., "Chicago Air Pollution Systems Analysis Program: A Multiple-Source Urban Atmospheric Dispersion Model," Report No. ES-CC-007, Argonne National Laboratory (May 1970). 3. P. J. Roberts et al., "Extensions and Modifications of a Contaminant Emissions Model and Inventory for Los Angeles," Report No. R73-15, Systems Applications, Inc. (January 1973). 4. F. L. Ludwig et al., "A Practical Multipurpose Urban Diffusion Model for Carbon Monoxide, " Project 7874, Stanford Research Institute, Menlo Park, California (September 1970). 5. R. C. Koch and S. D. Thayer, "Validation and Sensitivity Analysis of the Gaussian Plume Multiple Source Urban Diffusion Model," Final Report, Contract No. CPA 70-94, Geomet, Inc. (November 1971). 6. M. Platt et al., "The Potential Impact of Aircraft Emissions Upon Air Quality," Report No, APTD-1085, Northern Research and Develop- ment Corporation, Cambridge, Massachusetts (29 December 1971). 7. "RAPS Series No. 1 Study Plan," fifth draft (October 1972). 8. H. M. Benedict, C. J. Miller, and R. E. Olson, "Economic Impact of Air Pollutants on Plants in the United States," Project No. LSD-1056, Stanford Research Institute, Menlo Park, California (November 1971). 9, G. r, Hilst, 'Sensitivities of Air Quality Prediction to Input Errors and Uncertainties, " Proc. Symp. on Multiple-Source Urban Diffusion Models, Environmental Protection Agency Research, Triangle Park, North Carolina (1970). 175 ------- 10. W. F. Dabberdt and P. A. Davis, "Observations and Analysis of Char- acteristics and Energy Budget Components of Urban-Rural Surfaces in the Greater Saint Louis Area," Final Report 2322, EPA Contract No. 68-02-1015, Stanford Research Institute, Menlo Park, California (January 1974). 11. "Compilation of Air Pollutant Emission Factors (Revised)," EPA Report No. AP-42 with supplements, Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina (1973). 12. D. H. Fair, J. B. Clements, and G. B. Morgan, ''SAROAD Parameter Coding Manual, " EPA Office of Air Programs, Research Triangle Park, North Carolina (July 1971). 13. G. Ozolins and R, Smith, "A Rapid Survey Technique for Estimating Community Air Pollution Emissions, " Public Health Service Publication No. 999-AP-29, U.S. Department of Health, Education, and Welfare, Washington, D.C. (1966). 14. "Guide for Compiling a Comprehensive Emission Inventory (Revised)," APDT-1135, Monitoring and Data Analysis Division, Office of Air and Water Programs, Environmental Protection Agency (1973). 15. "Universal Transverse Mercator Grid," Publication No. TMS-241-8, U.S. Department of the Army, Washington, D.C. (July 1958). 16. J. C. Sturm, "Railroads and Air Pollution: A Perspective," Report No. FRA-RT-73-33, U.S. Department of Transportation Federal Railroad Administration, Washington, D.C. (May 1973). 17. D. B. Turner and N. G. Edmisten, "St. Louis SO2 Dispersion Model Study—Basic Data," unpublished report (November 1968). 18. L. J. Shieh et al., "The IBM Quality Diffusion Model with an Appli- cation to New York City," Report No. G320-3290, Palo Alto Scientific Center, IBM Corporation (June 1971). 19. F. L. Ludwig, A. E. Moon, and R. C. Sandys, "A Preliminary Study of Modeling the Air Pollution Effects from Traffic Engineering Alter- natives, " APCA Journal, Vol. 23, No. 6, pp. 499-504 (June 1973), 20. C. S. Benson, "ice Fog: Low Temperature Air Pollution," Report No. UAG R-173, Geophysical Institute, University of Alaska (November 1965). 176 ------- 21. R. L. Mancuso and F. L. Ludwig, "Users Manual for the APRAC-1A Urban Diffusion Model Computer Program, " Technical Report, Contract CAPA- 3-68 (1-69), Stanford Research Institute, Menlo Park, California (1972). 22. "Compilation of Air Pollutant Emission Factors (Revised), " EPA Report No. AP-42, Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina (1973). 23. S. D. Berwager and G. V. Wickstrom, "Estimating Automotive Emissions of Alternative Transportation Systems, Final Report, Contract DOT 05- 2004, Department of Transportation Planning, Metropolitan Washington Council of Governments (1972) . 24. J. L. Beaton, E. C. Shirley, and J. B. Skog, "Traffic Information Requirements for Estimates of Highway Impact on Air Quality," Interim Report No. FHWA-RD-72-35, California Department of Public Works, Sacramento, California (1972). 25. J. L. Beaton, J. B. Skog, and A. J. Ranzieri, "Motor Vehicle Emission Factors for Estimates of Highway Impact on Air Quality," Report No. CA-HWY-MR 6570825(2) 72-10, California Division of Highways, Sacramento (1972). 26. T. D. Wolksko, M. T. Matthies, and R. E. Wendell, "Transportation Air Pollutant Emissions Handbook," Report No. ANL/ES-15, Argonne National Laboratory, Argonne, Illinois (1972). 27. J. R. Martinez, R. A. Nordsieck, and A. Q. Eschenroeder, "Morning Vehicle-Start Effects on Photochemical Smog," p. 917, Environmental Science and Technology, Vol. 1, No. 10 (1973). 28. W. A. Bowman and W. G. Biggs, "Meteorological Aspects of Large Cooling Towers," Paper 72-128, 65th Annual Meeting of the APCA, Miami Beach, 18-22 June 1972. 29. W. B. Johnson et al., "Field Study for Initial Evaluation of an Urban Diffusion Model for Carbon Monoxide," Report 8563, Stanford Research Institute, Menlo Park, California (June 1971). 30. F. L. Ludwig and W. F. Dabberdt, "Evaluation of the APRAC 1A Urban Diffusion Model for Carbon Monoxide," Report 8563 Contract No. CAPA- 3-68(1-69), Stanford Research Institute, Menlo Park, California (February 1972). 177 ------- 31. J. McConnel, M. McElrory, and S. Wofey, "Natural Sources of Atmos- pheric CO," Science,, Vol. 233, No. 187 (1971). 32. L. A. Pipperton, J. B. Worth, and L. Kormrich, "NO2 and NO in Nonurban Areas/' paper 68-122, 61st Annual Meeting APCA, June 1968. 33. E. Robinson and R. C. Robbins, "Emissions, Concentrations, and Fate of Particulate Atmospheric Pollutants, " Final Report 8507, Stanford Research Institute, Menlo Park, California (May 1971). 34. L. A. Pipperton, 0. White, and H. E. Jeffries, "Gas Phase Ozone- Pinene Reactions Division Water-Air-and Waste Chemistry, " 147th Meeting ACS, Chicago, Illinois, September 1967. 178 ------- Appendix A SAMPLE EMISSION INVENTORY PRINTOUTS This appendix consists of 12 figures listed on the following page. A-l ------- CONTENTS A-l Typical Source Printout from IPP Inventory A-5 A-2 Typical Area Source Printout from IPP Inventory A-6 A-3 Typical Point Source Printout from IBM Inventory .... A-7 A-4 Typical Emission Summary by Counties from IBM Inventory A-ll A-5 Summary of Air Pollutant Emissions for the Study Area from the NATO Inventory A-12 A-6 Summary of Air Pollutant Emissions, Saint Louis City, Missouri, from the NATO Inventory A-13 A-7 Combustion of Fuels by Stationary Sources from the NATO Inventory A-14 A-8 Annual Fuel Consumption by Counties from the NATO Inventory A-15 A-9 Population and Housing Units by Grids from the NATO Inventory A-16 A-10 Emission Densities from the NATO Inventory A-17 A-ll National Emission Data System Point Source Listing . . . A-18 A-12 National Emission Data System Condensed Point Source Listing for Particulate for All Values Greater Than or Equal to Zero A-19 A-3 ------- i ^ <-f ^ M J- I- I [J m I a. u >-j k> u I I1: j; ' CI LO or cf cc. L/) i LJ to. • i_ I t_» 5 'LLI UJ i UJ 1 i. I i i i i" O ,0 J ~ O. — Q. I U • I l_i O CC O Q- Q. O CC LL O CC Q_ Z) o CO o D. CC =) A-5 ------- •1 j jj ^ 3 J ! i i i i •i •1 i 4 1 I 1 1 U > " U, f— Uf L. U. C- i , —• •±. J U (.r L-, LL LJ Q. — ' ^ LJ -1 ~ M \r cr 3 «w> X U •— < H- -t o -1 PI u ^ o t— < h- u» ,-« (X _J 0 t-« ~~ LJ LX 2 X LL) _J a: _J >- X L? X ^ LO QL •a X i a. >• 0 5 i/> ix x 1 u. X V) 'X _.,- l- t— O O UJ ?* CX UJ I— lij •a UJ ^ ij co .j UJ ,1 1-4 \ >~* t~ X ^ LU X v: i j t - LO u. LU UJ _l U t- -J ^J no LJ i/l ^ i—i t/i UJ UJ ^ UJ ^ u ^ 1—t »~ « IJ UJ if <1 LC UJ Q. UJ h~ V u. LL U_ LU CC ^~ t— UJ £ " ^ 5 . a. ;r H^ V) • LL U, aj _j LJ cr 1— 0 u o -; -« K V LxJ vki ^ h- »-. t— CY LJ ^ (J a. L) L- LI t t™ UJ :^ LJ L/l o V— O") _1 X -< u, t/) X t — J !j ^ ^ '- JJ ^ f— J. r - t^ . -« a t~ o UJ ( • \— OL IX LT LJ ~ U. u Q LJ LJ O U- U r. K. ' ' J.) .^ »- ~> "- oL L1J -_3 iu a ^ t-t nr tu iL. < ^ t- a UJ i~< * UJ ^- (M to LU tx u t— 1 t/) UJ _J ^ ~^ ' — 1 -^ LU -. iX UJ t— — t— ~-i Lt. f— ; 2-- iX UJ i. ZJ /J u LJ Uj - 0 •-4 i/l UJ CJ 1 UJ U"> o " ^ LLJ QC LU 2. a_ ^ to (X c_ UJ Lit u I— « CO LU LU * aj ~ i— .T. I a ri ' c5 i^. i cO > i. ij; LU • DC p g LU ^ — Q_ a. — O DC LL ^_ ID 0 h- cc Q. LU Q DC —^ O CO LU DC 1 ^j. CJ Q_ \| — CN 1 ii i fn - LL A-6 ------- rr-^* X r o —I- L. U. -T 0 3 O O Inu, H 1 M. DC O. > QC O • z UJ i •>. 3 J oc 1 t CO ' A-7 ------- ^ - i - ., ~ 'Jl 'SI t 0 J 0 s- il .j j ~ -r •v t/i > i.1 J t. 7> J r- ^ "X j ^ ^ ^ V- " U J ^^ J, J •J >~ JL 1. v/1 * I i_ j ~: vj •; i> ,- - - -4 -P ry 1-1 ^ 'I Z '" ,--, .„, 7 0 _"» J U -". ,, -. -. - J 5 o -, " -> -1 - £o -\| .,., ,, _• - — 0" -I'o ,1 f - 'J A -J ' "* - ^ • - 5 S - '- -- -• ' .. ^. -.--", 'JT - '- ;; - C," -' i~) J 1 - -•• -' ' - - -;• : ,0 -1 '_^ •fc •" - -• 7 " > -i .-- »•« T - .; j ,,. - ." • ^ :; --* ~ ,» •> •-> - » ^ , _•- - -, ^ T J , '-' -J J '.' r~ ', - —* —4 -.. ^ . :,~ F . ; r ,- -) ' 3 0 ? " J" •> '7 "1 3 ^1 .- T J -' J ' = - -,** ~ t J J " - J ' CJ -J '.t rJ O 3 —i "8 D C 4-" C u DC /-\ NVENTC s co S O cc. U- H PRINTOU LU O CC O C/5 . _1 O Q- M LU or LL A-8 ------- i! 7- z • u* U. M ^^ O X "* iJ • J f r- J 4U VI X » ef .J 4£ ^ J. -i ..J -I- tU -J -J fi -a '3 — J '/ '_! ac <-— j TJ ii. O — J •jf •* l>* JL _J O .TJ --J J I •J li- v/> .' 0 ' t, r •C - X1 V ,5, ,-, i" ~~ -, -J } ^. U. XL X. O X 3 ' 9° , 3 * a J - T ~> X _ . o i ;' 1 ' -,v * -' •"' -•• '. O '.J J 3 -•• c- '"> ™ "t •> ^ J " -3 : J 3 ,a -0 ~3 CJ 1 i CJ CJ 1 s. •i ^ - n / < ' -^ ' -. o -i .; -> u 'J -J T V ' °° -J I - j ;? w "J •3 O ~. o --! . i » ,3 ., . . ^ - -j - o -• J - 5 " s '-> -^ o -,« - 0 -5 a " 3 0 J - "> ^ -, o „,, ', 'J -» .J J ... j . 1 " < J • " J " ' " J - . _, /~, ^ - o ri -• -J - r^ ? 'P ,0 — — ! 1 '1 1 i ->. -JO -"-• ->,. ii ] -1 m .1 -J -J ~> X J O C u r "" - •~ - ,^J lj 5" 00 ;3 o ~> 'J j i ;; i ,- . - ^•7 J-. .- J ."- ., 0 , "S ~> ""' "> ' J "a 5 •-- ^ r> -> -; 1 '- ,0 5 -i\| 2 \| ,. ' •• J - •' •> , J. ^ ~> 1 i H "" -> -' "2 ~ S - -•- T.-, 7 . '^i p __^ O O ' f > •; 'J J ^ j •VI J r ~_ ~ '^ ^ 0 •* J 1C "* ^ p , '.J 0 -1 !-) I - • "°! ., - _„ " .j " 3 j >r j- ^ j j ^ ^ '• ' • -,:• ,. i ... . - '- J -i % - c- ^ ~ •- ^ ' ' J w ^~ ''i >•- 7 -^ 1 ••-' '•< 1 - -J , . -t ^ , ' i •* \ } C- '- -.1 •> T 3 O j^- r- 1 ~ ;" ,-• ^ --i > : i • i j^! -V •• J -, tl - J ... » J J T ; _- * •* -: CT .. :- " " -. ,' -i — . :/ •= ,-., -,, '- . -, , " -1 - „ j i -• >u. " - f J tl - y > > >j ^ „ -v ^g - -" O 'J -,., r j^ -,o iL ,- -' -' -7 7 - .i - -' ,, T -) fr; _, j j " - ^ j '-- f U •"? V O --i \ •"1 ' I , - ' ' 3 J -' •• - .- . '.} -, ~ - 5 ' ', "~ -> a -' J -. 0 .-c; ^ ~ ,, ,. •> j -* > -j > -' ? ^j j Q T - = 0 J ~ 3 ,"• -5 i J C " "0 ^ o i . - j - , 1 •' y l -j -J • * '/ 'J "" '-J ,, j :• ,„ "J "> -J tj V T ^J -^ J "5 v-, >« "-• j 3 ' ••' 3 ~. -, - '; t? "; 3 -- '3 "j "J ,. 7 'J rj n -3 u 3 'J "' 0 J J-, -J3 -3 -:• o • • • _> - •.' j .j -> j -• J V 3 r - .- ; 7 n •• 3 •> o 5 'J -^ -> ?." -,, ) O '- J O I, ,„ •» ' J J ',' I-J f IJC- •'" .- I. 3 1 J " ^ - ,, -J - J ~ -j J " ^* ~) ._, ,» o-. C< r> -, „. - T -, ; "/ J .' • ~* -^ " ' ' J -, -^1 "I <~) J rj J J V 'O -J O J •-,' 3 - r» /"^ J -; ,, Jl T r-1 -0 t/1 tr _ j • j -1 i - , .-_, ~ -i "• -> - - y O 30 r •» -i o > -j ~ J -, -J 1 '- '-- -> C -C <> '.. ><: ->'. UJ OQ O ac u. t- o tr a. o ac I o. >- CO Ul ac D o A-9 ------- -* I'M „' • ', W-r,* ^•'¥' ;•?'?, *r" «— Jftl^ ct- ..k^. '^4! ft."; ' * - >,.-*«.',• - ,L- ,, ,,'iffi, •',$} ,,#4'* -7 Vk #v;" " '' " -JL, ' ' ' ' "'* .\|"4 '~K. " t) '--.- J%il f * . ' ' ""'•;'.' \ • -.. >f< "V i ^ Vjp ' V/ vT\|- C.tAS&Ftc^nck' , -. r- rf JP _ ^^ ^ K ««^; __;X. - - ' -» ;^- , A j. i ' 1 CE^ict l:o£»iiriciT. o u. •-1 u* 'J j/¥ t A.t AtLGb/ABL -^ 1 •^ o '-) ~J LJ 1 X X „ 1 -J "5 a -• ^ ~K f 0 ^ £ — o -. u •M .'! r >u a r Q. ^1 4\| iff a . ' u. FHiILLL <• -j -j u X cues C 3 O • rs: '.J w T> •, -J v3 ? > J O O 3 C.' 05 u> ^ 6- 0 V. J -^ a ' J sr ^ •ti W 3 "> J .^ -^ sr tn m •3 O 3 O '- o -> "- T> K. < J " ^ 0 "3 O .0 •? ^ * .-C x; ^ N TO r* '-^ - - - ^ •O '^* ~^ X ^ ay O 0 ^ J ^ r •r ~* 1 o o - o , ^ J -* — * .A jo ,r 41 m 07 r ITi un j c.- x- - ro * UL, r V -f 7 '-3 1 j ~ "* ^ « -» T ~» O 00 », _? J ^ - LJ -J « •- a s. o o j 3 J ^ l»" i-> "•" ^ - r rj -j - j -» C7 o cr -^ i - J T u -J -3 n a ,0 ^ J •! 0 J ,c, •J J LJ '~J 3 CJ J O J» 4V ' W ,-! JC "-1 3 C1 r-> ^3 ^J 0 O O •> J n v '., ^ J -3 J ~> a -3 'j ------- o e . o •r 1 < 1 UJ a, 0 o o aj »- ' O * ^J £ X l/l LJ z o o 0 u (/) J! 1 * l/l aj I --5 , ^ ?I ^ -' ^ Cj « 2 t -3 _> lo U"" p < f-I t— 5-» o r —-— LlJ ,2 j^ a. hYDf/CCAft!3 CTLAL ) c ^ - a • u j j~j J-4 O. CJ U CJ~~ PL .< < j ^ < i O — ' j 1 i ' i i i \| c \|j - \| op) r, p lo •-. \|; o i i 1 1 opo o [ ) 1 f. 0 D \|O 0 JO 0 ! I 1 1 , i j i j ... .1. . . . . . . <-• ]o o iJ \|o " p M '- ^ M l ' i i ! ' ; i i , ( -u i ' : »- f ; ' •yj u! ,U O- ! i \| 7, * ! ^ ' ^ ' ! ' i T i1 j;! J; ,-%>-" ; ,'/:'> ' -it--'' ; _) V-1 ' ' ! ' -I v> ' ^ ;\ ' i t >r \ b .. .j < -j a. >• ; V-J ** " - ; i ; ^ i : i i i i , i ! 1 i ' ! i : i , j i U lr' M -J ^1 1 ,_J 00 """ ^i^ bi'~J -iH^i "^ * "*' k_) « [t u i O ' * ' la: ti t_j f- la. -« "d LJ JLJ jj »— - ! ' ' ! O l\J 0^ •3 -1 --» X £ 3 ^. - M lj 0 °°rj i :j r-« -* ° p o L> . > i 1 U' f« r-j HM ij ^ LJ o ,-J 'J iX. ^ "-* M ^ ^ •«-* LI j. ' «, ^ . • Uj "i • - 1- t I ' ; JD (J •n ** " :;<^ -T C3 0 0 O 3 J J 0 o j j I J 0 1 t t ' r- - ; , 1 1 -'"i i 3C 3f, Jl Ifl -•J 4- .J ex O H a • -* i u) ' 1 ! ! i j i i~J O f J O i o o O L» cj 1 i 1 1 1 f \| j 1 I ' 1 OC f^ 1 -^ UJ z CQ ^~ 2 o CC LL OUNTIES o m cc 5 13 t/) Z O C/5 t/5 P LU _j ^ 0 a. >- ~ t ! -»• ; ; i 1 ... , j i I o o J O i o u o (_? i ! i 1 i ' . -. I ---'-I--' I i j ! \| 1 j CC C5 _ LL. A-11 ------- u- 2 "? tX »— 3£j UJ K 5 tl I> O C. c= < CL a £ i HI ^ a; u V 1/1 t- • LU - o t^ IX' K 0 r" sj IT r<- ^ Is- O Is- , fx r- £ C •c ^- r- Lf (V cr «-- IT ty LL '2 — C c: p— •- »— • 3" t- C C C- «J (^ c 2 a z Ci t- cn -^ IT C p— ^~ (f' \C •C f^~ ^~ f~ Is- cc c ~ u Is- cc e cc u^ o C- ^3• f cr O f^ r<" Or Cr rv or Is- O cv cv Is- ^_ ^ t- r c i- u 1 T C" t- r cr i/- ^ - f~ „. <-- cr ^ f\ C r c. rj- r^ ' t~- ^ cr IT •- ^ ^~ 00 _l a ^ LL U. C *• i^ Cr C K •- o- h- " t^' C" ^ a: >~ 5" C V. I" CC CV: Is- xO cr o- tf o r-4 r c ir. p"' r-t •— t Is- Is- f\ r^ C. (X f IT r~ — f ^ vr C sj rf^ if1 Is- IT r-, ex ^ ^3" Is- (V 5 7 ^ (—4 Or' - ». ^-" * LL <- « >— LL1 ^ 1 U. r c <" i — LL r\. -J IP Ct fM Is- .-J- •" CC C~ C c- 1— ' 5 c Is- rs. f s^- ^ -* IT _J H' •^ C t/" »— C 1- c. <- t/ Cx " LL. c r »„/ I- t. i— sO o •c cc ^_ fv L," ir. Is- cc -• ir c j- ^ «- c c ir. CO r- f., •£. •a £ _j t- C C E- ^ cr C , rf LU CC >. 0 D CO LU I h- cc 0 LL CO z o 8? LU 1- rf CC HO 31- -JZ -1 LU O> "- Z cc ~ < ° O^, > "J QC i < so DOC CO LL LO < UJ QC o LL A-12 ------- i- V C_- u-. \s V- ^_ u 1- — • 1 r C a L C X 2 J * 1— c c_- 1— ^ ^' C r.- __i ** LL •— ^. w — ? C •— k- X C c. ^ C l_ >— c* «- a- X c c C u * u ^_ £. ~" c "" I 1- ~ r- •»• i O • — < (\i i •C ^f C i t— f t • f^1 C" • — ' r^ C O C" ': <• I C l~ ff C ! CC 1 • I-M a c\* c. •V. . r ir f^ CC (X • C or (^ — ir- ^ r- i ^ t- c ^> i CT-. ^* "" (_ U- U C i?" c >— t— t/ ~ c ** c U' c- c cc c: <">- <\ -C (N vT f c. r~ CC i — c - ^^ C; U- f- 1 l^ 2 ~" f I" tL i t— •— V '<" IT C x^ •-1 >r 0 C vf O f C sC c ir <^ ir. C V C1 i • cr L- f^~t r'' H- t/ ___ ^ <• »— »— i- r «c ^ U. « C »» ^ ts~ ^ u. c or i_ f <• ^ r r- <- vT C r- : < c t/t »- r- >- c <. L^ Cl >- U C 2, »- u c; (^ ^ Z: •- u u. a: c p C 1- c r- ir cr 0 a Lf ^-< st ir r- -r rr <" cr cr~ C- sf f c ^¥ — ^ — Cf «'. >-- tt C V— ~" •», U) C c. ~ C i^* (\ r. CL r c. ir r° C ct tx f 0 CO D O h- z CO GO" ll ^> z ~ < 0 O LU o- I < 5 LL. DC O LL 3 SUMMARY MISSOURI, FIGURE A-( A-13 ------- ' i I i i i 1 i i ) 1 i i i i i i i i i/~ u_ <_ C IS N_ ?~ b t , i ~ i i i i i i i t i ! i t t 1 i t , i i \- <~ I- t/ ^ ^ I/ LL LL C c - h- t/" CC C L. . i i i 1 < L C" — <• 0 < > 11 ^ > ~ V 1- if u z C U- t- I- ^— 1 1 i 1 1 * C r" u c <- >- cr <• d X C l/~ > c. c c? U- >- <" l_ C- u I/" _J u. u 1 1 i i \| i i • • • c rr c a , C C — (V. 1 — f . N+ 0 vt M r^- • • • c' h-j r^ c C >? r1"' c v- Cf (V ir r rr ^ C IT • • C vf c~ r tv r^ •c L- •" <- -* <_ H vC NT •£ C — If • 1 ir if NT P- -" e- c c? ^-> \~ u _. M «• -I «C Uj — -Ht r D L_! »: * •- '! £ ! r «; C- Li\| r - b»- t/ l_ «r C — r «.- 5 a- cr c. v_ 1 1 1 •i • — 1 N^ vT r^ c ^- f\ IT or cr • c Nt C pr r • r "• L^ f\ rJ ^ c r ^_ <: •— t i— u ^; C - c _J u u "f • i o r C" r' • < IT C - _j h- u. r 1 LL i- c? < - LI l^ t- U f c. i 1 1 1 r t 1 ' * <" ! h- •£ cr . • »I r • (V • f\ r^ • a r~ t- . 5 C (_ f (<• i r\ r-- w « C r — • — CC ^- r- j\ J ^ ^. • • c r- ir >r ^ • « r cc r^ ^-i c- c NT" vC • f^ r~ ^ . •3- C c c r^ • L" C >c If •f. _l < I— t— ^; •z «r C' C LU I H- .=- ^ O cc LL CO LU O CC ID 0 co > cc < o I— CO > QQ CO _l LU 3 "- > LL. LT 00 z z O LU c: > co E CD O S h- 0 < 0 Z r-v i LU DC ^ (3 i i LL. A-14 ------- =1 I A a 01 c t— 5 c ^1 1 c c o c o o c c — .-, o c i- fv. cc ^ rvj c r~ e 0 C1 0 C c .— , ,-. tr ir soooo, 4SPPP. c c- r- o rsj * %t rv c c. o c 0 C ^ c rvj K, a — or 1 IT or oca.r-.cri ,r - c\ — c u Cl u\ C IT c ,. IT IT O f~ C C q c rl c i s •4 c o c el d c °i c J !? 1 -T _J C V i/ - c f <: Or U <. C _ ™ " C (- tr ,£ I tr c C) c c~ - c c c- ^J \J - t- c 1 1 — a •s. _i < 0 o O o 1 '-' 1 "" _J c • tr i» C ; i fv. T 0 r^ C v* r ^ c c- o' — f t\ IT c c c c rv, a cr f- c c xl c c »- f" r cj C c ^ 1 ~ ^ - 'C tr, tr _ , C C t- f- 1 P1 c _ Cf < i_ T U LL l~ i 1 i ! i I 0 N rv c a? c ir X J H ir o Q r- r\, r<~! IT cc (N r cc a. c c'c c rv r^ S. -J C <: c cy < o u <• C -j r C 1- <" c r. C f\ f^- u- U -J k- c »-* a' ;>• _j o o o jo . ~4 r — / c • ' c to LJL1 a: , t ! O C C O .C C 0 C- c ^ < rf (\, c o c c C 0 c c: c c- o r- O C _rt r" r c c I- •— c u: <_ i/" tr "~ rr c c •- »- x* o c 2" LL C _ (,' Q Q < a a i— c o c c e •C. O r- -• 0 C c c c r r- ~ c- o •o 2T _j C ^ IT •f> ~z —l C. ^e tr <• c o ------- SM. I N fl N/£> HOU31WG V NOilVTf 0- O P- Ct u C C n c. X -J c I/* c c X r\ C C. c r 0 C C u- o c O b r »J (N- C o in r~ IT O C C C ir sr •C P- «-^ f-^ r- L- c <. r" r C r- c c c c u- ir r- C c- c o c c- c C -3- I- C . . C' c- f C c c 0. sc r\ -- rv O C ^ o N- - tr — ' c r 0 c h- cr r- j ^r c c c c Nf -3 C C c * c- e c c — * r O tf c c c c < J (N, IT -C »? c c IT. < CC O cc r- c c c c C 0 c c. r- (r P » if if IT >r c c c. c 0^ C cr <- o c- — o c e If «T •C tf c c c c 0 C -«• If < -J- IT r if- r^ — - •• r c o c 0 C c f < If 1" •»• J- or cr r cc r O -T If s£ (V ic f- C -sr r f fv, «? _ r c c c u c: r^ o c C If r. C If V IT cc o O < o QC 9 cc o CQ to o Q- O a. UJ cc D a A-16 ------- o o 0 o o 0 0 o o o 0 o o o o -c 0 o IT o o o c c c o o c c o c c o o 1- c o o c -* (f o o o o o o c o o o o c c 0 o .c c c < c- o f- Q c 0 c c c o o o c c o o o o o c c c 0 ^ I" rvj c o 0 * o o o o o o o o o o o c c o o 0 o c o c c o c c o 0 0 c c 0 c o c c c r~ f o o o o o o o c o c o c 0 c c IT 0 o ir- c IT c o o o o c c 0 0 a o o 0 c o o c 0 c •* ^ ec <7 o tt irv I" t^ IT, r* »> oc ° c c o cr o o \r> p o cc c o o cr* c. c c c c c r^ f\ cr ir rv cr IT fv If IA m o o (" o o m 0 o in c o IT o c c o c o- (•* o f- c o o o c c o o c fv. e c IT o c c c J vD o 0 o t-t o o 0 c o o 0 c o c r<* 0 0 r* o rf o c c c c o o o o c c c r~ c c c o o <• r^. o 0 o o o o c c r-t o c o o o o .0 c o ^ 0 J- c c c c- c c. c 0 c o c C' o \' c~ c c c c •t er 0 0 o 0 o o o 0 o g o c c o c o o rv c o r» c M o o 0 c c c r c c \c f\, o ir c c f o c: c 0 o c c o c c c c 0 o r- c c o 0 -t !• o o o o o C\J o o f^ o 0 £ o o c o c c c o c c c o o c c o c o c- c c c c 0 r~ tr> o o 0 0 c o 0 o o o c- c o c o c c o CX 0 0 r o r\ o o o c c c c c r c o c c c o c * IT- o o •» o o 1^ o 0 o- o c tf o 0 c 0 rsj r- o r* c CO r- o ^_ c c c c c c c c c c- c o c c rr r- •» 0 c « o o IT o o 0 o 0 oc o . 0 o c a c o ^c o in r- c ^ c c. z 0 c o o o c o o o c o f 1C 0 o «• o 0 o c o c o o c c o r> c o rr c c re o 0 c c o c o c c c 0 c c. c c c- c c c •» e o oc I HI > z o < O tr CO ai ai O o I LU QC O A-17 ------- z o STATED): ILLINOIS \ocsio7o): METROPOLITAN ST. LOUIS ULL-MOJ CITYI8520): WOOD 'RIVER COUNTY(4680irMADISON CO " PLANT ID: 0004 "POINT ID: 18 •MME-j'OPR^SS: AMOCO OIL CO, 400 STATE 610 PD.WOOD RIVER SIC{?911): PETROLEUM REFINING PERSONAL CONTACT: F K WEB. SCCO-06-00! -04) : INDUSTRIAL PROCES -PETROLEUM. INOPY -PROCESS HEATER -GAS ALLOWABLE EMISSIONS 1 ***************** j IMATES ** ATED EMISSION £ ST k****** *** 1 HAND CALCUl \| ******< "~UT~M GRID COHRDINATPS A*******-**** GENERAL INFORMATION ******** **w oc oc ce. a. a. PARTICULATE: 1 TONS/M SCX: 2 TONS/M NOX: 2 TONS/\ HC: 0 TOflS/M CO: TONS/I «•_«_ .» — — — — _ _ , I OL CL CX OL CC co c/> to co to ^ Z Z Z Z 0 0 C1 0 C K (- (- \|_ \|_ O •? (V O O rj 1 PARTlCUUie 1 Sl> I ~ N0> 1 HC !_-____„__£: UTM ZONE: 15 HO^IZ^NTAL: 752.4 KM VERTICAL : 4,306.6 KM Y^AP OF RECORD: 1972 "OWNERSHIP: "PRIVATE IPP 'PROCESS: i COMPUTE? CALCULATED EMISSIONS 1 ************************* \| TIMATIQN METHODS k ************ EMISSION ?! \| *******< 1 h- U- o o .# 1 I 6A/SNOJ 1 1 MA/SNOJ 1 i V N £ X a. o < CO a. >ION FACTOR(AP-42 OR PENDING) IAL BALANCE £ ENGR KNOWLEDGE ION FACTORIAP-42 OR PENDING) ITN FACTORIAP-42 OR PENDING) CO LU l/> CO T < x y - X X o Ct C 0 I o on 2: a : U. LL a- o • r^ O O CL a rONS/YR I 1 rONS/YR 1 CM >H V X 0 o x z 1 CO: NOT APPLICABLE ~. _ ~_ E 11. O O o o u A****-.*! STACK HCTGHT STACK 01 AMP-[ GAS TE^Pi--KATl GAS FLOW RAT PLUME H-MNO ! SOUPCE: COMBUSTION CO * 2 # '2 J «T * or * u t ,c * ', * !< * S- * tt * r * z # HOURS/DAY: 7A c L »• . ; _i t o > a L Z u 0 " O L FONS/YR 1 1 o o u » - * 0 ******* PART: 00 INSlMd . lN£wd *** NOUV31 u. z c c 1- u r LL > ' \ C 1- g o ?S :b»3A/SX3_3M___ 0 0 IU LU o r o o PRIMARY PART SECOND. PART ""f ANNUAL THROUGHPUT *****«<«****** 0 . 0 0 0 0 0 X X o a CO Z 1- t- 1- (- z z z z U U. U' UJ JT r i r o o o c Ul UJ UJ UJ c c c~ c O 3L d. CL C. C CO 0 C« C C3 XX XX o o go or c ii o •-to — 0 OC UJ 0 U. a co ,0. to K K 9* ». j W fvj CM ifvJ O > u1 U'! ~) CO\| Ul 1 "a \ - ! 1 1 i t i j 0 0 0 0 0 0 o c X 0 1- .- (- (- z z zz UJ U-' U • '_ T 5C IT y 33 33 O O O O LU U' UJ UJ c- r c c a: a ££ ------- o o X o z. X O OC O LL £ o oc te IJ N oi O o I- I! Q ^ 01 ^: CO F < »~ < o z h-l ( CO ,_, 2. , UJ -J CO t- Z Z -10 r> < •• — o a: a: & o o .. u: . a oc X U, CO O ct tU QC ! Z < 7 U1 a >- i j 1 : O 1 $" (SJ i "O J 1 5! 1 14 ' o .. o ^_ 1 Z r- •- 0 O 0 0- 1 o cl ' * \ o II 0 00 U. I "- uj CM; ^4 m ' i H Oi to 1 i i O UJ >-i h- > •• 1-4 O a a CL C1 o .. UJ CL OC X LL CO O at uj a z < ^ UJ o > «J3 •" 0V LU «J H' O O .. o* 1- 1 fi o a o CL t O » »-« O O£ CtC CL O • • LU a. at X u. to Q oc UJ tt Z < 3 UJ C > -0 >• (V. »o LL) — J -4 O -« o 0 0 a. i o •• ^^ o a ac CL O .• u- a CL X u. to o CL UJ C* 3C U' a > 1 a rO (M ° 1 t z i-4 o (SJ •t O 1— I z o -^ fsj o o CL \| o w I fo N O H CJ LL U. UJ r-t ^ Sf -< H O CO at o UJ H H > • .-. Q CL CL D, O .. u- CL OC I u. a. UJ QC Z < 3T UJ 0 > , 0 pf, fv. •• M O (X OC CL r u • • U' CL ac I LL co o oc UJ CL z < 2 UJ 0 >- ' IT •o fs. >0 m o ^4 •• 0 t- i z r- - o o o 0. \| CM o »• 1 m B U II U tsi U. U. UJ r4 O K U t/1 IE o < O 2i ^— QC ^2 »Q UJ ^"^ CO i S^_ ULJ CO _j ^f ^c p— *"^ Q__J 1 z< '•' CC co O CO 1 1 S LU ^r J" 1 ^ "" ^ z 0 2c r> o 3 «»i Q 2, a % 3 »-i "-) ifl 't (SJ \o UJ 5! z oc o 1 ^jj t— • , 1 ^ J f^ u. o a ct a u> LU CO CO UJ t- O O o to I/) a: U h- z C 1 < -J U. O -• l/l z •* z o o cc to •-. »- to o o z »-* rvj -J LU a: o » CL - »-. D CO ^J LU O CO Z UJ 0 fr~ -J — • M* O fl -1 QC — co O i— -0- 2T > o a o CD Z ac O •— « H- UU -LL- O t~ vO ^- (NJ LU > 2^ OC O ~z~ bx _J ^; 4 a: LL. O 0 or Q. U' UJ t- co CO u. 1— C O --. O x: t _j r >" o CD CO >M o cj 2; fg -j uj O • O- H- •— ' 1/1 3 O 2! LL, t- -J - C* -J CL CO C f- •- Q- 2T COO Z OC O »— ' h- -J U' C -J S Z <-> cn CO (X O CJ u, CO "Z o to CO r~t f_ UJ 0 UJ CO r- m ^f >. cd UJ H- 3Ti " >** X O o o •• O O H •4- h- O co FH O O UJ i O O O , • • O 0 • o (v. CO 1 i LLt [ > ' Z i CO ty a LD <»^ ' O •• CJ, J- Z If CJ >-• 1 J -J _J LL ^ _j i: D »-• to •a Z — . Z ct O ' O U. CO CO M -* to X •• 13 H CO o cj _j ai z: O iTi IE UJ ct: o • a tx v- •-• o co r> LU -J O CO LU Z it' D UJ < »- t- _J — _J a — CO CO O h- LU >-• CL ^ >• K C O C CD O LU CJ CL Z UJ ^ 1 1 -» -i O l^ -J or — C. (- 0. Z > C O CD of LJ H U' UJ O K S- Z -3 s: 0 O K- r o oo O O LU 0 fl cv. •c. 1 7\ -jc «« UJ O t/J 7 O i" 1 K C> _l J_ 1 i- i: -i _J t-< O 00 a o o -t _J o -t 0 -< i H- >• f> oc oc z ^3 ^_ o »- ^J LU VI O z ^ a. TOO u-' z a 01 _J ItJ -J I •• " o .. o O -J- h- O r-t O ! u CO or a (_> u_ zz o CO CO ^H 3T 1- U' z UJ -J < a ^ •~- 0 3 UJ a a. UJ CO J •" LJ ir\ ot — H- z: >- o tc Ui C 1- z < r O h- O 00 O U' 1 IT •o CV •0 00 1 _J t_l 3 X CM p~* c o z ^ QC o z u; D ^ CE u z •• o c 0 ; u LO -> UJ I CD c. i 2 K- _J •— ^ »-, _J -• 0- CL -1 0. ^p 0 0 Z 0 ^t — 1 LU Z !• H ~4 • a. LU 1- «- 0 oo rs o o u Z UJ < t- _l — «-« a —J ac — 00 0 H •- a. z > o n o to z or o •- H- UJ _iuj a t- » ff ..0 1- "•f1 f- co »-. O UJ i CN < OI OC o A-19 ------- Appendix B SUMMARY OF EMISSION MODEL REPORTS B-l ------- CONTENTS LIST OF ILLUSTRATIONS B-7 INTRODUCTION B-9 I Argonne—Chicago Air Pollution Systems Analysis Program: A Multiple-Source Urban Atmospheric Dispersion Model . . B-ll II Argonne—Transportation Air Pollutant Emissions Handbook B-19 III ESSA—The Diurnal and Day-to-Day Variations of Fuel Usage for Space-Heating in Saint Louis, Missouri B-27 IV FHA—APRACM0D, Modification of APRAC-1A B-31 V Free University of Berlin—Numerical Simulation of Temporal and Spatial Distributions of Urban Air Pollution Concentration B-35 VI GE—Study of Air Pollution Aspects of Various Roadwaj' Configurations B-39 VII GRC—Evaluation of a Diffusion Model for Photochemical Smog Simulation B-47 VIII Geomet—Validation and Sensitivity Analysis of the Gaussian Plume Multiple-Source Urban Diffusion Model . . B-49 IX IBM—The IBM Air Quality Diffusion Model with an Application to New York City B-61 X IBM—Source Emissions and the Vertically Integrated Mass Flux of Sulfur Dioxide Across the New York City Area B-67 XI Metropolitan Washington Council of Governments— Estimating Auto Emissions of Alternative Transpor- tation Systems B-71 B-3 ------- XII National Air Pollution Control Administration— Saint Louis G02 Dispersion Model Study, Basic Data B-77 XIII National Air Pollution Control Administration— Emission Factors B-81 XIV NR&E—The Potential Impact of Aircraft Emissions Upon Air Quality B-87 XV Ontario Department of the Environment— Information System Descriptive Manual B-93 XVI Pacific Environmental Services—Controlled Evaluation of the Reactive Environmental Simulation Model B-97 XVII Rutgers University—Comparison of Air Pollution from Aircraft and Automobiles B-103 XVIII Sacramento Regional Area Planning Commission— General Inventory of Air Pollution Sources and Emissions B-107 XIX SRI—A Practical, Multipurpose Urban Diffusion Model for Carbon Monoxide B-109 XX SRI—Procedures for Estimating Highway User Costs, Air Pollution, and Noise Effects B-117 XXI SRI—A Preliminary Study of Modeling the Air Pollution Effects from Traffic Engineering Alternatives B-125 XXII SAI—(1) Contaminant Emissions in the Los Angeles Basin—Their Sources, Rates, and Distribution. (2) Extensions and Modifications of a Contaminant Emissions Model and Inventory for Los Angeles B-131 XXIII SDC—Development of a Simulation Model for Estimating Ground Level Concentrations of Photochemical Pollutants B-145 XXIV TRC—Sensitivities of Air Quality Prediction to Input Errors and Uncertainties B-149 XXV TRW—Prediction of the Effects of Transportation Controls on Air Quality in Major Metropolitan Areas . B-153 B-4 ------- XXVI University of Alaska—Ice Fog: Low Temperature Air Pollution B-161 XXVII University of California (Davis)—The Impact of Highways on Air Quality B-165 XXVIII USWB—A Simple Diffusion Model for Calculating Point Concentration from Multiple Sources B-167 B-5 ------- ILLUSTRATIONS B-l Flow Diagram for Calculation of Pollution Levels .... B-18 B-2 Speed Adjustment Factors for Hydrocarbon Exhaust Emission Factors and Carbon Monoxide Emission Factors . . B-23 B-3 Profile of Carbon Monoxide Versus Traffic Flow Rate at FDR Drive B-40 B-4 Relationship Between Vehicular Pollution Factors Taken at Exhaust Plane and Traffic Speed B-42 B-5 Flow Chart of Auto Emissions Model B-73 B-6 Exhaust Emission Flow Diagram B-84 B-7 Flow of Source Subroutine B-101 B-8 Automobile Hydrocarbon and Carbon Monoxide Emissions Added per 1,000 Stops B-120 B-9 Automobile Hydrocarbon and Carbon Monoxide Emissions per 1,000 Miles of Driving at Uniform Speed B-120 B-10 Automobile Emissions Added from Speed Changes per 1,000 Vehicle-Miles B-121 B-ll Factor to Convert Reference Year Emissions to Emissions in Year Y B-122 B-12 Relationship Between Carbon Monoxide Emissions and Steady-Speed Driving B-127 B-7 ------- INTRODUCTION In this Appendix we have reviewed and summarized twenty-seven reports on the treatment and parameterization of man-made emissions. These summaries are intended to provide the potential user with a con- cise overview of the scope and content of each of the models. Each summary identifies the author and his or her affiliation, includes the author's abstract, arid provides a synopsis of the emission model method- ology together with a statement of the model's format and availability. For convenience, the reports are listed alphabetically by affilia- tion. The depth and merit of the various models are reflected in the summaries. While our review is extensive, it should not be interpreted as all inclusive. Many reports were reviewed initially and later deleted from this final listing either because they were of no direct use to RAPS or because the methods they contained were more appropriately documented or developed elsewhere. B-9 ------- I ARGONNE NATIONAL LABORATORY Chicago Air Pollution Systems Analysis Program; A Multiple- Source Urban Atmospheric Dispersion Model. J. J. Roberts, E. J. Croke, A. S. Kennedy, J. E. Norco, and L. A. Conley, Argonne Natl. Lab. Rept. No. ES-CC-007, prepared for Chicago Dept. of Environmental Control and National Air Pollution Control Administration, May, 1970. A. Author's Abstract This report is a comprehensive documentation of the dispersion- model development phase of the Chicago Air Pollution Systems Analysis Program. A multiple-source, urban atmospheric dispersion model has been developed which described transients such as morning transitions in atmospheric stability and mixing layer height. The dispersion model has been validated by comparison with over 10,000 hourly averages of sulfur dioxide monitored by the Department of Environmental Control of the City of Chicago. For example, the model accounts for 50% of the variance in 6-hr averages of observed data and 70% of the variance in 24-hr averages. Of particular significance in the capability of the model to describe "area sources" as volumetric clouds of pollutant and thus to evaluate the effect of these sources on dose points within, as well as external to, the area. The atmospheric transport kernel in the model describes the instantaneous release (delta function) of pollutant, advection according to piecewise constant hourly wind vectors, and Gaussian diffusion about the centroid. Continuous plumes are simulated by integration in time of this point-source Green's function. This report details the transport theory and all other computerized algorithms that influence the dispersion problem and presents statistical results of extensive validation studies. B-ll ------- B0 Summary An inventory of average hourly emissions of sulfur dioxide was developed for power plants and industrial, residential, commercial, and institutional sources in the Chicago area. Hourly SO2 emissions from power plants were derived from actual power output logs; hourly emissions from other sources were obtained through reduction of annual fuel-use data according to seasonal and diurnal patterns. Major emitters were con- sidered individually as point sources, with the remaining emissions grouped into area sources. 1. Treatment of Area Sources The area source overlay grid is input to the model. It is specified by five parameters: the x, y coordinates of the origin, the number of grid squares in the x and y directions, and the length of the side of each grid square. The grid used in the validation computation consisted of l.Oxl.0-mile squares at the center of the array, with surrounding 2.0x2.0-mile, then 4.0x4.0-mile grid squares. Representative pollutant emission rates and stack heights are assigned to each grid square for each source: (1) low-rise space heaters; (2) high-rise space heaters; and (3) industry. a. Low-rise Residential and Commercial Emissions Hourly emissions from low-rise residential and commercial sources are assumed to be proportional to the number of heating degree days, with a fixed fraction of annual emissions uniformly prorated to hot water heating. The heating cycle is assumed to occur between the hours TON and TOFF, with emissions equal to the hot water rate at all other times. A "janitor function" is synthesized to account for the greater than usual fuel demand during the first two hours of the heating day. Hourly emissions are given by B-12 ------- Q (1-PCTHW) (65-TEMP) TE TEMP £ 65°F and Q = annU TON S time of day heating DDAVG (TONOFF) s=> l ------- a simple method for calculating air pollutant emissions from roadways. A fairly complete set of emission factors is tabulated in Appendix B, so that transportation emissions can be calculated without using either of the computer programs. B. Summary Two computer programs were written to relieve the environmental impact analyst of the emission calculation burden. The first program, TREFACT, calculates the emission rates for vehicles or mixes of vehicles based on vehicle usage (speed, vehicle miles, type of trip). The second program, TREMISS, was developed to calculate emissions for the complex traffic simulation grid used by urban regional transportation planners. The outputs from a typical regional traffic simulation model (origins, vehicle miles, speed) and TREFACT outputs are used to calculate grid- dependent emissions for the entire urban region. A detailed description of each of these programs is provided, along with input descriptions and sample problems. Complete listings of both programs are contained in an appendix. The computer programs are used to calculate grid-dependent emissions for large urban transportation networks. A method for manually calculating air pollutant emissions from transportation sources is described in another section of the report. This method is designed to calculate emissions from individual transportation links (e.g., highways). The emission factors calculated by TREFACT are for the standardized urban and rural speeds. TREMISS has speed adjustment data to modify the emission factors for the speed-dependent transportation network. The speed adjustment curves from the figures accompanying this summary have been transformed into a set of linear equations that will be used to adjust the emission calculations for individual roadways. Urban and rural speed adjustment factors should be used with urban and rural emission B-20 ------- factors, respectively. It should be remembered that this speed adjust- ment data are representative of typical traffic and does not represent free-flowing traffic at that average speed. TREFACT calculates emission factors for transportation systems with widely varying vehicle types and operating conditions. Emission factor can be calculated for any year between 1960-1990 for six classes of vehicles. A vehicle-class-weighted emission factor is also calculated. Options are available to include the effects of cold-start operation, retrofitting control devices on pre-1968 vehicles (Class 1 only), and consideration of diesel-powered vehicles. Urban regional transportation planning involves an enormous computer traffic simulation system. The typical Chicago Area Transportation Study (CATS) traffic simulation network has approximately 1,800 transportation activity grids which cover the eight-county Chicago metropolitan area. A computer program was written to calculate grid-dependent air pollutant emissions (CO, HC, NO ) from transportation sources for the CATS planning X area. This program, TREMISS, uses output data files from the CATS traffic simulation model, but data from any traffic simulation program can be used. Regional traffic simulations are quite costly and there may not always be data available for the study year. Therefore, TREMISS has the additional capability of linearly interpolating between simulation output data for two study years. This is extremely helpful when traffic data is not available for a particular year. TREMISS is structured to interpolate between 1965 and 1985 simulations for any intervening year. Emissions can be calculated for each grid using either one set of "weighted" emission factors or six sets of emission factors, one for each class of vehicles. (Diesels [Class 6] may or may not be used in the calculation.) Additionally, the complete set of grids may be grouped into B-21 ------- any number of subsets. Each such subset has its own set of emission factors. However, the mixing of "weighted" and "class" emission factors is not permitted. Emission summaries for each grid and the totals for all grids are the major outputs. If class-dependent emission factors are used, summaries by class are also given. Emission summaries for any specified set of grids representing a county, municipality, or problem area (e.g., Cook County, Chicago, Chicago Business District) can also be calculated. B-22 ------- 2.0 tr o o i- LU 3 1.0 o o \ 2.0 IT O o z LU o X URBAN TRAVEL 1.1 cc e o u- 1.0 I- z 3 0.9 \ -\ o 0 0.8 0.7 RURAL TRAVEL 10 15 20 25 30 35 40 45 50 55 60 65 SPEED — mph 1 URBAN TRAVEL \ CC O Z LU Q 1.1 1.0 0.9 O 0.8 0.7 RURAL TRAVEL 10 15 20 25 30 35 40 45 50 55 60 65 SPEED — mph SA-2579-26 FIGURE B-2 SPEED ADJUSTMENT FACTORS FOR HYDROCARBON EXHAUST EMISSION FACTORS AND CARBON MONOXIDE EMISSION FACTORS B-23 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User From Program Supplies AP-42 Proxies/Comments Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor x X X X X X X X X X X X Factors available in report for use as desired Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode x X NP NP x x NP NP NP x NP NP Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area X NP NP x X NP NP NP x NP NP NP X NP NP NP For heavy duty truck calculations NP = No provision B-24 ------- Output Available Pollutants -CO Yes -HC Yes -NOX . Yes -Particulates NO Availability of Program Documentation Yes No Included in report x Referenced in report x Language FORTRAN ; PL/1 Equipment IBM Validation/calibration: Chicago area and EPA data Time resolution; Hourly (peak/off-peak); daily Spatial resolution; Subareas; grids B-25 ------- Ill AIR RESOURCES FIELD RESEARCH OFFICE, ESSA The Diurnal and Day—to-Day Variations of Fuel Usage for Space-Heating in St. Louis, Missouri. D. B. Turner, Atmospheric Environment, 1968. A. Author's Abstract Data on the wintertime emissions of SO from residential and ^4 commercial space-heating sources by 2-hr periods were needed for use in a diagnostic dispersion model. Analyses were made of hourly steam- output data from a centralized heating plant and hourly gas-sendout data for December 1964 at St. Louis, Mo., to determine dependence upon temperature and other factors. Methods were then developed to determine the rate of fuel use from residential and commercial space-heating sources for each hour of the day from values for the hourly temperature, the hour of the day, and the day of the week. Relations developed from December 1964 data were tested on data for January and February 1965. B. Summary 1. Description of Algorithm A mathematical representation of the diurnal and day-to-day variations of fuel consumption resulting from residential and commer- cial space heating was developed for St. Louis, Missouri. Available data included hourly temperatures and hourly gas and steam output rates. Steam output rates are taken as representative of commercial sources, and gas send-out rates are taken as representative of residential sources, To obtain equations relating daily output and daily temperature, graphs were constructed to give the least-squares linear regressions for weekdays and weekends. For weekdays, 20 data pairs were used to give B-27 ------- DS = 8.40 - 0.1078 DT DG = 30.00 - 0.3498 DT where DS - daily steam output rate DG - daily gas send -out rate DT = daily temperature The effect of short-term temperature variation on the diurnal cycle of space heating was investigated by determining the deviation of the hourly temperature from the daily value for both gas and steam, and then averaging these deviations for each hour for all weekday data. Plots of both gas and steam deviations versus temperature deviation were made, and the finite difference forms of the daily send-out equations were applied to give output deviations with zero temperature deviation. These hourly values, calculated for weekdays, Saturdays (only four data days were used) and Sundays (only five data days were used) , represent diurnal variation due to all causes other than temperature. To make these variations applicable to other space Seating sources, the values were divided by the slope of the regression line, thus con- verting to units of temperature. Saturday and Sunday values were adjusted so that the same total daily heat is required for a day vitii tae same daily temperature, regardless oi whether it is a Saturday, Sunday, or workday . Hourly steam output or gas-send-out is given by A = a- where A = hourly output a,b = regression constants in daily output equations T = hourly temperature H = diurnal value (calculated above) for appropriate hour of day, day of week, gas or steam B-28 ------- It is assumed that no lag time exists between a change in temperature and its effect on space heating requirements. 2. Validation The above equation was derived from data recorded in St. Louis during the month of December, 1964, and was used to calculate hourly output for January and February, 1965, for comparison with actual data. Error was measured in degrees Farenheit. For steam, 71% of the cal- culated values were less than 6 F in error, and 95% were less than 12°F in error. For an hour with a temperature of 30 F, an error of 12°F results in about 25% error in steam output. Overall, 80% of the calculated data were within ± 7°F of the actual values. The author states that this degree of accuracy is commensurate with the accuracy of the emission inventory. It is assumed that derived diurnal factors for gas will be nearly the same for oil and coal, and that steam factors will reflect patterns of commercial space- heating with any fuel. C. Model Availability A computerized treatment for calculation of hourly emission estimates is not referenced in the paper. B-29 ------- IV FEDERAL HIGHWAY ADMINISTRATION APRACM0D, Modification of APRAOIA. P. W. Blow, FHWA. A. Author's Abstract None given. B. Summary APRACM0D is a modified version of the APRAC-1A urban diffusion model reported on in A Practical, Multipurpose Urban Diffusion Model for Carbon Monoxide. The purpose of the model is to estimate carbon monoxide con- centrations at arbitrary points in an urban region resulting from pro- jected vehicular travel. The APRAC-1A urban diffusion model was modified to accept a standard FHWA Urban Transportation Planning Battery BUILDHR data set with carbon monoxide emission words added on each link record as input rather than the link card inputs for APRAC-1A. The modification further allows more links to be processed than the original program. The program is composed of 17 separate decks and is written in FORTRAN. B-31 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User From Program Supplies AP-42 Proxies/Comments Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed x x X X X X X X NP NP x NP NP NP x NP NP NP NP NP Average CO of vehicle mix for year of study interest Speed and Distance Spatial Distribution -Links -Grinds -Area NP x x NP NP NP NP = No provision B-32 ------- Output Available Pollutants -CO Yes -HC No -NOX No -Particulates No Availability of Program Documentation Yes No Included in report x Referenced in report x Language FORTRAN Equipment IBM 360 Validation/calibration: Based on APRAC-1A Time resolution; i hour Spatial resolution; Radial segments-22.5° out to 16 km from receptor point B-33 ------- V FREE UNIVERSITY OF BERLIN Numerical Simulation of Temporal and Spatial Distributions of Urban Air Pollution Concentration. H. D. Fortak, Proc. of Symposium on Multiple-Source Urban Diffusion Models, Chapel Hill, N.C., APCO Pub. No. AP-86, 1970. A. Author's Abstract A multiple-source diffusion model for the simulation and prediction of long-term (climatological) ground-level sulfur dioxide concentrations in urban areas is described. The computer input consists of data from an emission source inventory together with statistics on relevant diffusion parameters. Because of the capacity of available computers, only a limited number of the largest emission sources (approximately 150) can be treated indivi- dually. Smaller industrial emission sources are treated as residential sources. These are represented by a large number of stacks (about 150) of the same dimensions, distributed over areas of 1 square kilometer, for which the mean area emissions have been estimated. The meteorological input consists of data on wind direction, wind- speed, and Pasquill-Turner stability classes. These parameters are assumed to be spatially homogeneous throughout the metropolitan area. Low-level emissions (residential) are correlated with low-level windspeeds and Pasquill-Gifford diffusion parameters, whereas high-level emissions (industrial) are correlated with extrapolated windspeeds and Brookhaven diffusion parameters. The program also uses corresponding statistics for urban boundary layer depths and values for parameters affecting absorption at the earth's surface. The diffusion model used is basically Gaussian. It is modified, however, such that turbulent diffusion is restricted exclusively to the B-35 ------- depth of the urban boundary layer. This is true for all sources having effective emission heights less than the height of the upper limit of the boundary layer. The rate of decay of sulfur dioxide is taken into account, as well as the experimentally determined absorption at the earth's surface. The model calculates fields of steady-state ground-level concentra- tions that correspond to a given spatial distribution of emission sources and to any possible combination of relevant meteorological diffusion para- meters. Knowledge of frequency distributions of these meteorological diffusion parameters permits the derivation of frequency distributions of ground"-level concentrations for any location within or outside of the metropolitan area. The computerized experiments simulate frequency dis- tributions of ground-level concentrations for a great number of regularly arranged grid points (up to 2500 with a mesh size of 500 by 500 meters) and for a variety of time periods (months, heating period, seasons, year, etc.). The frequency distributions are characterized by a limited number of parameters (mean, percentiles, etc.). Each parameter is plotted as a system of isograms on a map of the metropolitan area. Experiments to validate the model were conducted during the heating period in 1967-68 at four continuously monitoring stations that had been installed at special locations within the limits of the metropolitan area of Bremen. During the sampling period, the assumption of a suffi- ciently homogeneous wind field was validated by wind measurements at the same locations. The calculated frequency distributions of half-hourly mean values of concentrations generally agreed fairly well with those derived from observed values. Comparison, however, shows that the model does not simulate ground-level concentration fields in the vicinity of industrialized areas very well, because uncontrollable low-level emissions from industrial plants could not be taken into account in the diffusion mode1. B-36 ------- B. Summary Sources in the city of Bremen, Germany, are classified in three groups. First are individual stacks with emission rates greater than one kilogram of SO per hour, which are treated as point sources. Records £ for these sources contain information as to output by volume, maximum emissions, and mean winter and summer emissions. Whenever possible, daily emission variation data and emissions for holidays were obtained. The second source category includes individual stacks from small industries that contribute less than 0.02% each to the total emission rate. Since these sources contribute such a small amount to the total emissions, they are treated in the same manner as emissions from space heating. Emissions from space heating are apportioned as area sources in a 500 X 500-meter grid. From the sulfur content of the fuels, the total fuel consumption and the total number of dwelling units, a mean emission rate of 8 g SO /hr per dwelling unit is obtained. This rate assumed the £ daily mean temperature remains constant throughout the heating period. To make emissions from space heating a function of time, a relationship between daily mean temperature and daily emissions is used in some cases. Multiplying 8 g/hr (or the corresponding temperature corrected value) by the number of dwelling units in each area gives the mean emission rate for that area. This is then divided by the number of individual stacks representing the area source. C. Model Availability A computer code for the emission inventory methodology is not pre- sented. B-37 ------- VI GENERAL ELECTRIC COMPANY Study of Air Pollution Aspects of Various Roadway Configurations; APTD-1146; New York City Dept. of Air Resources, EPA, FHWA, New York Division Office; Prepared by General Electric Company, 3198 Chestnut Street, Philadelphia, Pa., 1971. A. Author's Abstract This experimental study was undertaken to (1) ascertain the current air quality in the immediate vicinity of various urban roadway con- figurations; (2) determine how the selected urban roadway configurations aid or hinder the diffusion of the pollutants emitted by urban traffic; and (3) develop mathematical relationships between traffic, traffic speed, pollutant concentration, meteorological parameters and roadway configuration. It is possible to assess the impact of the 1970 Clean Air Act by future monitoring programs which determine the extent to which the clean air goals are achieved. It is possible to identify those areas in and around a given configuration that do not meet the National Air Quality Standards under certain meteorological and traffic conditions. It is possible for urban and transportation planners to compute in advance the pollution anticipated for a proposed roadway design in order to insure that dwellings adjacent to the proposed design will not be exposed to pollutant concentrations in excess of the National Air Quality Standards. B. Summary One section of the report presents nomographs which provide highway designers with gross estimates of the CO concentration to be expected for various roadway configurations. Figure A presents representative relationships of CO concentration versus traffic flow rate for one of B-39 ------- 90 80 70 60 50 40 30 20 10 I 2000 4000 6000 TRAFFIC FLOW RATE — N vehicles per hour 8000 SA-2579-27 FIGURE B-3 PROFILE OF CARBON MONOXIDE VERSUS TRAFFIC FLOW RATE AT FDR DRIVE B-40 ------- the ten sites monitored during this study. The CO values shown in the curves are those of the vertical plane of probes reflecting the maximum values. The traffic values shown are those of the total traffic involved. Regression and correlation analyses were performed to determine whether the relationship between environmental variables and carbon monoxide and hydrocarbon concentrations could be predicted mathematically. The resultant simple correlation coefficients and multiple correlation coefficient, R, show large variations. It must be concluded, therefore, that the postulated regression equation does not adequately describe the effects of the wind parameters on the carbon monoxide and hydrocarbon concentrations . It is assumed that any given roadway can be divided into independent line sources and each probe located at the roadway measures the concen- tration from its immediate line source plus the concentration emanating from the other (x-1) line sources. Furthermore, it is assumed that the concentration decreases exponentially with distance from each probe location. The model was further generalized by defining the vehicular pollution factor, , by dividing the unperturbed concentration, CC^, by the traffic flow rate in the immediate vicinity of the probe location. The vehicular pollution factor, tf> , CO N (1) For eight of the configurations evaluated, it was found that & shows a strong linear relation with traffic speed (correlation of 0.81) as shown by Figure B. The viaduct site exhibited pronounced meteorological effects which resulted in the two lower points to the extreme left in the figure. The low velocity points in the lower right hand is data collected B-41 ------- 00 CM 8 b UJ > oc o o i- Z o - 3 o Q. O OC t» i UJ CO DC O O < LL O a. D Q o UJ - UJ X a. UJ CO >0 UJ g ^c to n CM < DC m LU CO Q- ib S3 _l I LJJ x cr uj m LU oc D a LL — Q33dS OlddVdl B-42 ------- in a city street. Data from the tunnel sites was not included in Figure B. The linear relation between <£ and traffic speed was found to be 0 = [-0.51 (T ) + 26.9] = 10~3 (2) i i for T: 15 mph ^ T ^ 49 mph (PPM-Hr) ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User From Program Supplies AP-42 Proxies/Comments Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode NP NP NP NP NP NP NP NP Experimentally measured at roadside for CO NP )Average speed at point of fmeasurement Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area NP NP NP NP x NP NP NP NP NP NP Hourly traffic count at point of measurement. ^Determined for area I bounded by roadway )configuration NP = No provision B-44 ------- Output Available Pollutants -CO Yes -HC No (Could not obtain statistically "good" results) -NOX No -Particulates No Availability of Program Documentation Yes No Included in report x Referenced in report x Language NP Equipment NP Validation/calibration; Calibrated/validated for 10 different roadway config- urations in NY Metropolitan area Time resolution; 1 hour Spatial resolution; Area occupied by roadway configuration B-45 ------- VII GENERAL RESEARCH CORPORATION Evaluation of a Diffusion Model for Photochemical Smog Simulation. A. Q. Eschenroeder, J. R. Martinez, and R. A. Nordsieck, Final Report, General Research Corporation, prepared for Environmental Protection Agency, Contract No. 68-02-0336, October, 1972. A. Author's Abstract Extensive improvements have characterized this evaluation of the GRC Photochemical/Diffusion model. Despite the limitations of smog chamber experimental data, they have served an essential purpose toward updating the kinetics portion of the model. Consistency of rates and reactivities is now achievable using recently measured coefficients for a wide variety of systems. Model methodology revisions have enhanced the realism of the advective and diffusive descriptions. Previous assumptions regarding transverse (cross-streamline) horizontal diffusion have been confirmed by an exhaustive series of parametric tests. Photochemical/ diffusion validations were successful for trajectories occurring during four days of the 1969 smog season in Los Angeles. Our measure of success is concentration-history fidelity with a minimum of adjustments of diffusion parameters. (Chemical coefficients were scaled from the smog chamber studies and held fixed for the simulations carried out to date.) Future directions for air pollution model development are discussed in detail in an appendix as information supporting the experimental recom- mendations . B. Summary General Research Corporation's diffusion model uses hourly emissions of carbon monoxide, ozone, hydrocarbons, nitrogen dioxide and nitric oxide B-47 ------- from both fixed and moving sources on a 2-mile interval grid to predict pollutant concentrations. The emissions inventory used is that compiled by Systems Applications, Inc., (see Review No. XXII). B-48 ------- VIII GEOMET Validation and Sensitivity Analysis of the Gaussian Plume Multiple-Source Urban Diffusion Model. R. C. Koch and S. D. Thayer, Geomet, Inc., Final Report, Contract No. CPA 70-94, prepared for Environmental Protection Agency, November, 1971. A. Author's Abstract This report, submitted by GEOMET to the National Environmental Research Center, presents the analysis and results of a program of validation and sensitivity analysis of the steady-state Gaussian plume- type of urban diffusion model. The report develops a careful definition of the fundamental short- term steady-state model and its various modes of implementation, in terms of emission and environmental input parameters, and of calculational modes. A set of computer programs developed especially for validation and sensitivity study purposes is described. The validation study consists of a variety of comparisons of short- term and long-term concentration predictions from the model, with com- parable measured SC>2 concentrations covering three months of 2-hour values at ten locations in St. Louis, and one month of 1-hour values at eight locations in Chicago. The predictions use hourly estimates of meteorological and emission parameters. The atmospheric stability is estimated from hourly weather observations from an adjacent airport using the McElroy-Pooler diffusion parameters based on Turner's definitions of stability categories. The mixing layer ceiling is estimated from radiosonde observations taken twice daily from remote locations (100 to 200 miles away). The wind speed and direction are B-49 ------- hourly averages of several continuous records. The emission rates of the largest sources are identified and located individually. For other sources a mean emission rate per unit area is estimated for a square gridwork of points with a one-mile spacing between adjacent points. Each emission rate is related to hourly estimates of space heating and other operating requirements. No "calibration" or other adjustment of the inputs or output concentrations is employed anywhere in the analysis. Individual short-term model results (1- to 2-hour periods) show large deviations from the observed concentrations, but the frequency distributions of calculated short-term concentrations over a month or a season compare quite well with the comparable frequency distributions of observed concentrations. No single factor could be found which accounts for a significant fraction of the individual deviations. Predicted long-term concentrations show consistently good agreement with observations, as contrasted with the significant overestimation usually found in other model implementations. A technique is proposed for calculating the long-term estimates which obviates the need to calculate every short-term concentration in the period. A sampling process is used in which a statistically selected set of as few as five to ten percent of the short-term periods is employed, and the representativeness of the distribution is maintained. In the sensitivity analysis, the insensitivity of the model con- centrations to the parameters of wind speed profile parameter value, and the distribution of area source emission heights, is demonstrated. Quantitative description is given of the sensitivity of the model to the following parameters: changes in spatial variability in emissions, vertical diffusion parameter, pollutant half-life, wind speed, mixing ceiling, wind direction, and downwind variation in emission rates. B-50 ------- Finally, recommendations are made on implementation and use of the model described herein, and for further study in special problem areas highlighted by the study. B. Summary The GEOMET multiple-source diffusion model designates individual sources with emission rates significantly higher than the average rate as point sources, e.g., power plants, large industrial plants, and large commercial and municipal emitters. For the remaining emitters (area sources such as residential homes, high-rise apartments, and var- ious smaller commercial and industrial establishments), it is assumed that an average emission rate is representative of a small region. Hourly estimates of SO point and area source emission rates are required, £i Two methods of parameterizing temporal and spatial variations of these rates have been used in this study, one for the St. Louis emissions data and one for the Chicago data. 1. Treatment of St. Louis Emission Data The following algorithms for estimating S02 emissions as a function of time of day, day of week, location, and temperature were adapted from methods developed by D. B. Turner.* a. Area Sources Area source emissions were divided into 5000 X 5000-foot grid squares, and included data from residential, commercial, river vessel, automobile, railroad, backyard burning, and industrial sources. Turner suggests the following algorithm for area sources: * Personal communication to GEOMET, Inc. B-51 ------- Q(t) = Q + q D (t) + Q F (t) + q D (t) + Q + Q F (tl + Q + Q rrr cc cc v a a wb +q D (t) + Q F (t)F (t) x c p d h D (t) = 65-T(t) - A (t) r r D (t) = 65-T(t) - A (t) c c where Q(t) = SC>2 emission rate 0 = base residential S00 emission rate r z q = residential heating S00 emission rate per degree r ^ T(t) = ambient air temperature A (t) = residential temperature correction factor r A (t) = commercial temperature correction factor c Q = base commercial SO emission rate F (t) = commercial diurnal variation factor c q = commercial heating SO emission rate per degree c 2 Q = river vessel SO emission rate v 2 Q = base automotive SO emission rate a 2 F (t) = automotive diurnal variation factor a Q = railroad SO emission rate w 2 Q = backyard burning SO emission rate D &i q = industrial heating SO emission rate per degree x 2 Q = base industrial process emission rate P F (t) = industrial day of week variation factor d F (t) = industrial diurnal variation factor h B-52 ------- Introducing subscript notation, let i,j appear on parameters that vary from one area source to another, let k denote fuel type, a denote annual or national average, and & refer to the hourly period. Then H R. . K s /N \ < \ _ a 13 V^ _k\ k/ ij qr) ij D R Z-f E H / * * . k k where H = average annual U.S. household space heating energy requirement a D = average annual U.S. degree days a R = average number of rooms per U.S. household a R = average number of rooms per dwelling unit in grid square (i j) ij ' S = S09 emitted per unit fuel k E = heating efficiency of fuel k k H = heat content of fuel k k ( N ). . = number of residential dwelling units using fuel k in grid \ k/ij x . .\ square (i,j) F r/ij ~ 1-F ( r/ij w s K - number of fuuls where F = summer day fuel consumption as fraction of average winter day s D = average winter degree day w c _K 1 j At k=l where At = duration of season s C. . = number of commercial sources in grid square (i,j) J. J W = annual quantity of fuel k used by JLth source B-53 ------- (Fs)> fraction of annual quantity of fuel k used by $th source in summer season \ qcj ' S = SO emitted per unit of fuel k ' 2 c K - 'F ij D ) At w k=l \ W w/k.e •(«=)« } where D = winter season degree days w fraction of annual quantity of fuel k used by £th source in winter season I K w ij F ^ W wk-6 k=l where I = number of industrial sources in grid square (i,j) Substituting, the original algorithm becomes F D s w 1-F + 65-T(t) -V" H R a a E H k=l k k F c At 65-T(t) - D 65-T(t) - A- (t) c w K D w k=l K ij . k=l C ij + Q + Q F (t) + Q v a a w 65-T(t) - A (t) c b D w K ij £ kr k=l £=1 ,W B-54 ------- I ij + F (t)F. ------- n 1 W S J=l J J or n q = — ) (F \ W S X °w fe HJ J J where W . = annual quantity of fuel j used «J S = SO emitted per unit fuel j j 2 D = annual number of degree days a D = winter season degree days w = fraction of annual quantity of fuel j used in winter season w J The SO emission rate equation then becomes & n Q(t) = Q F (t)F (t) + [65 - T(t) - A (t) ] ^- } W S . p d h c D f—' j j a j=l In addition to emission rate parameters, data inputs include location coordinates of the source, physical stack height and the plume rise, wind-speed product. Power Plant Sources. According to the available data, one of two methods of estimating power plant emission is used. For some plants hourly power outputs were available. The information included graphs of fuel weight flow rate, stack temperature, and stack exit gas volume flow rate as functions of power output. A linear relationship was found between fuel flow rates and power output: F = A + AL B-56 ------- where F = fuel weight flow rate L = power output A ,A = empirical parameters Emission rates are S Q = FE 100 where Q = SO emission rate F = fuel flow rate E = SO emission factor 2 S = percent sulfur content of fuel Stack temperature and volume flow rate are also linearly related to output, but it was found necessary to divide the range of power outputs from zero to a peak value into three equal parts and to use a linear approximation over each part or class of the range. This gives T = T + s JL L where T = stack temperature s L = power output L = lower limit of power output class J6 L = peak power output P T = stack temperature for power load of Ln Ju T = stack temperature for power load of Lfl + 1/3 L u I and B-57 ------- V = V + 1— IV - V s SL L P^ T u P/3 V where V = stack gas volume flow rate S V = stack gas volume flow rate for power load of L V = stack gas volume flow rate for power load of L + 1/3 L u 1L t p For another plant, estimates were made for each two-hour period of the day of average emission rates, wind speed and plume rise product. These estimates were derived by government personnel based on discussions with the plant operator, and are assumed representative of all days in the data period. 2. Treatment of Chicago Emission Data The algorithms employed by GEOMET to obtain hourly SO emission £i rates from the Chicago emission data were those developed by Argonne National Laboratory and described previously in Section II of this appendix. The area source grid consists of one mile squares. Additions to the algorithms include Turner's temperature correction factor in the residential and commercial area source treatments and the following equation for describing heat emission rates for the three classes of additional point sources: H =0.15^ 100 3680 N where H = heat emission rate e Q = SO emission rate £t S = percent sulfur content of fuel N = number of stacks B-58 ------- 3. Model Sensitivity Sensitivity to the vertical distribution of area source emissions was investigated. It was concluded, for the cases tested, that in general the differences resulting from the use of various vertical distributions were negligible. To study model sensitivity to diurnal variation in emission rates, the factors(represented by algorithms) which affect emission rate estimations were analyzed. The inputs to the algorithms are temperature, electric power load at generating stations, and hour of the day, but errors associated with the measurement of these inputs are small. There- fore, the assumptions made in formulation of these algorithms were investigated. In the emissions model, annual emission rates act as scaling factors for diurnal variations; an error in one of these estimates creates systematic error in model predictions. However, the authors feel that if concentrations are considered at a number of widely dispersed locations, the errors tend to cancel each other. Sulfur dioxide emissions are characterized as arising from three classes of operations: (1) electric power generation, (2) industrial processing and (3) space heating. Hourly power plant emissions are estimated by means of linear relationships with hourly electric power load of individual generating units. Error in this regard is small, but uncertainty would be increased for proposed power plants where hourly loads would need to be predicted. Diurnal emissions from industrial processing are allocated on the basis of utilization factors concerning day of the week, hour of the day, etc. Variations between actual and estimated emissions may be considerable, since industrial operations must respond to fluctuations in demand and breakdowns in equipment. The authors state that errors B-59 ------- in emission rate estimated in the immediate vicinity of individual sources result, but a random distribution of these over an urban area tend to balance. Algorithms for emissions from space heating operations contain correction factors duch as Turner's temperature correction factor and Argonne's "janitor function." It was felt that random errors associated with individual sources of these types are probably small, and a more serious type of error is associated with the assumputions in the algorithms of no time lag between outside temperature change and fuel consumption. This type of error is critical to short-term model sensitivity, sometimes resulting in an emission error of a factor of 2; it is not important in calculating long-term concentrations. Systematic errors due to unseasonable temperatures can result when thermostats are controlled only partially by temperature, since the algorithm will immediately respond to the temperature change. In these cases, the emission rate error will persist for the duration of the unseasonable weather. The resulting concentration error will be minimal over the long-term, but short-term concentration error will be directly proportional to the emission error. A careful study of the magnitude and duration of heating emissions is required to determine the nature of heating system response to rapid and extreme temperature fluctuations. Co Model Availability Computer program listings of the emissions treatments for both Chicago and St. Louis data appear in the GEOMET report. B-60 ------- IX IBM The IBM Air Quality Diffusion Model with an Application to New York City. L. J. Shieh, P. K. Halpern, B. A. Clemens, H. H. Wang, and F. F. Abraham, IBM Corp., Palo Alto Scientific Center, Report No. G320-3290, June, 1971. A. Author's Abstract An experimental multisource air pollution diffusion model based on the Gaussian plume formulation is described. The model is capable of incorporating point and area sources, time and space dependence of source strengths, and time and space dependence of meteorological variables. Numerical experiments sumulating the SO concentration 2i distribution for New York City are presented for January 11, 1971. The numerical results agree favorably with experimental measurements. B. Summary The IBM diffusion model requires 2-hourly SO emissions data in £1 order to predict the average ground level concentrations for a 2-hour interval. The emissions treatment recognizes individual point sources, which include power generation, municipal incinerators, manufacturing plants, industrial combustion, chemical and oil refinery and mineral smelting, large hospitals, etc; and area sources, which account for commercial and domestic space and hot water heating. 1. Treatment of Area Sources The grid square size for area sources is variable, allowing for specification of different grids for urban, sub-urban, and rural areas. These specifications are based on receptor locations relative to the sources and the distribution of source strengths. Computations were B-61 ------- made using a 1.0 x 1.0-mile square grid in the urban area, 2.0 X 2.0 in the sub-urban area, and 4.0 X 4.0 in the rural area. Necessary inputs include the number of different area source sizes used and the number of area source grids in each area source size; the model groups source grids (for any grid size) in a rectangular region, thus also requiring as input the number of rectangular regions for the entire source grid; and, for each region, the x,y coordinates denoting the limits of the region and the size of the source grid for that region. A source emission height for each area source size must be input to the model, and it is taken as the average building height within the area source grid. Total annual SO emissions are first broken down into daily, 2t then 2-hourly emissions. One method used to generate daily patterns was to relate daily emissions to the average daily temperature, where Q Q annual annual Q = v + V (DD) daily Yl 365 Y2 TDD and Q = daily total SO output Q = annual total SO output annual 2 DD = degree days = 65 - daily mean temperature in °F if the latter is 65°F or less; otherwise, it is zero TDD = annual total degree days y =: fraction of sources due to hot water heating y = fraction of sources due to space heating ^ and YI + Y2 = 1 An alternate approach divides the calendar year into three seasons and allocates fractions of the annual source emissions to each season. Let Q'1> ot and a represent these fractions. The seasonal J- 2 3 B-62 ------- emissions may be written as Q = a Q season 1 1 annual where a + 01 + a =1 Daily emissions are found by dividing by the number of days in the season in question. Estimates of daily emissions using this method are not as precise as those calculated by the former method. To reduce daily SO £ output to 2-hourly source emissions, statistically derived coefficients representing diurnal variations of emissions as a function of ambient temperature are introduced. It is assumed that for each 10°F interval of daily mean temperature a typical diurnal emission pattern exists, from which these coefficients can be derived. Two-hour source emissions are calculated with the expression: 12 12 \"~* Q/2 hour \ = \ ^ 3 Q Z-f \ V 2-f * daily where 3. (i=l, 12) equals the fraction of daily output at a particular 2-hour interval. Values of 3. are given in the table. Small point sources are included with the area source emissions if certain criteria are satisfied: (1) the total annual emissions are less than one-tenth of the total annual emissions of the associated area source; (2) the stack height of the point source is within 20 m of the associated area source emission height; and (3) the point source has no appreciable plume rise. All other point sources are considered individually. 2. Treatment of Point Sources The model requires bi-hourly emissions data from point sources, and if measured data are not available, the following simulation technique B-63 ------- CD £ •H H r— \| Oj o 3 CU r< 3 -P cd CU P, p! CU H O O O O CM CM O • t>. O o CM CM 1 0 CM O • o rH O 0 o CM 1 00 r-H O • O rH o o 00 rH 1 CO O • O rH rH o o CO H 1 rH O • cn o o rH 1 CM o cn i~H o o CM rH 1 O rH o • o rH O O O rH 1 00 O O 0 rH o 0 00 o 1 CD 0 O • O rH O o CD O 1 ^ o O • t> o o o 1 CM O O • CO o o CM O 1 0 o 0 in CU bfi a a! OS cF o to A! O • CD O • cn 0 • o rH O • cn o • c« o t- o • cn 0 CM' rH O • ^ rH O H rH 0 CO o CO 0 cn m o m o in 0 o rH O • o rH O o rH O • cn o o rH o c rH O rH rH 0 o" r-4 O cn o CO o CO 0 cn i o o o • • CD t> o o • • cn cn 0 0 • • o o rH H O O 0 • o cn rH O O • • O 00 rH O O o cn rH O O • • o cn rH O O o cn rH O O • • O O rH rH 0 0 O 0 0 CO CD 0 0 co in O cn o co cn 1 N O CO VI in in in CU rH O -P S3 in ai a - O in •rH !H -P nl tsi V ?H >> 0 0. r< S-i CU O Si O -P O S • j rj •H • 3 CD TD CD cn in rH CU •H •H O IH 3 rH •a cu HL r3 •P 4-> O •H 0 0 -p ?H a 0 0 >H 'H •p S ri CU O rH c a •H a d •a CU bJO rt a •H -H ai -P •P pi O CU oi -P •p oi ' 03 r>> •a c o CU oi •O J3 !H CU £ 3 •P O •H CU O e ^ as •H oi rH O r •H tuO CU CU S3 B Si •H CD -P IQ in 3 -P ?H * O •a cu > SH -P CU • -P CO B-64 ------- (i.e., emission model) is suggested by the authors. Point sources are classified according to industry, and operational modes for each industry are established. Patterns for daily output will vary according to mode of operation; that is, operation on weekdays versus weekends, seasonal trends, shift factors, etc. Several patterns for a particular industry are distinquished, and diurnal emissions are obtained through the application of the various factors derived from the patterns appropriate to the mode of operation under consideration. C° Model Availability The computer program for the IBM model is considered experimental and is not available outside IBM. B-65 ------- IBM Source Emissions and the Vertically Integrated Mass Flux of Sulfur Dioxide Across the New York City Area. P. Halpern, C. Simon, and L. Randall, Journal of Applied Meteorology, 1971. A. Author's Abstract Sulfur dioxide concentrations (in ambient air) , obtained from helicopter soundings and traverses, together with wind data from pibals, were used in a kinematic box model to determine hourly average three-dimensional fluxes of SO for the New York City area. Continuous £t dry-bulb temperature and pressure height records were obtained con- currently and utilized in the analysis and interpretation of the flux data. The SO fluxes were compared to degree-day-dependent emissions ^ from residential, industrial and utility sources. Results indicate that the vertical structure of the fluxes is related to diurnal variations of the temperature lapse rates. Furthermore, the emission rates determined from the measurement of flux in 13 tests agree with those derived from the New York City sulfur dioxide emission inventory within limits which have been considered acceptable forusein air pollution modeling. B. Summary 1. Treatment of Emissions The emissions treatment described by Halpern et al., considers hourly SO emissions from residential, utility, and industrial sources. £ Hourly values of power plant emissions were obtained from the utility companieso Hourly manufacturing and industrial emissions are estimated by uniformly distributing annual emissions data over an 8-hour day, B-67 ------- 5-day week, and 50-week year. Daily residential emissions are obtained in the manner described in the IBM model summary previously given. Diurnal variations in residential rates are estimated using data on duration of hourly operation times for apartment house boilers using fuel oil. Total boiler operating time for each day is correlated with degree day values: t = 17.2TT + 192.0 where t equals the total operation time of the boiler in minutes and TT represents the degree days. Since space-heating boilers receive fuel oil at a constant rate when operating, knowledge of t and the fuel sulfur content is sufficient to describe the hourly SO emission rate. £ 2. Model Validation Sulfur dioxide fluxes are calculated from concentration measure- ments taken by a helicopter and using wind measurements obtained from pibal anci radiosonde soundings. A kinematic box model based on conser- vation of mass is used to verify the hourly emission rates and to demonstrate a mass balance of SO . & For thirteen test cases, the vertically integrated SO flux was ^2 greater than the flux estimated from the emissions inventory. The average value of the ratio of source emissions flux to the vertically integrated flux was 0.74 for tests during May and 0.83 for tests in November. The authors state that this indicates that the source emission estimates were more responsive to the degree-days dependence in the colder November tests than in the warmer May tests. The average value for the ratio for all tests was 0.78. The mass balance of SO has been verified by comparing calculated £i SO flux to source emission data. ^ B-68 ------- C. Model Availability A computer program for the emission inventory methodology presented is not included in the paper. B-69 ------- XI METROPOLITAN WASHINGTON COUNCIL OF GOVERNMENTS Estimating Auto Emissions of Alternative Transportation Systems; Office of the Assistant Secretary for Environment and Urban Systems, Washington, D. C. 20590; Prepared by S. D. Berweger and G. V. Wickstrom, Dept. of Transportation Planning, Metro- politan Washington Council of Governments, 1972. A. Author's Abstract This report discusses the development and application of a model which can estimate the relative magnitudes of carbon monoxide, hydro- carbons, and oxides of nitrogen automobile emissions for alternative regional transportation systems. The computation of auto emissions is accomplished by means of a computer program which accepts travel and facility data, together with assumed emission rates, and calculates speed and vehicle-miles of travel by type of facility by sub-area; applies the rates (for peak and off- peak travel speeds and volumes) and calculates the amount of emission by sub-area. Emission rates are a function of the speed of travel and the age distribution of vehicles in the year under study. A portion of the program logic is based on a model developed by the Tri-State Transportation Commission for estimating highway facility requirements, modified for the purpose of calculating auto emissions. These findings are based on the emission factors supplied by E.P.A. If these factors are revised as a result of further research, it is quite possible that some of the results obtained would change. Revisions to the emission factors can be readily accommodated by the model developed in this study. B-71 ------- B. Summary The methodology makes use of vehicle trip forecasts, along with highway network information, to estimate future travel, the speeds at which this travel will occur, and the emission levels produced. The methodology does not require trip distribution and traffic assignment model procedures. It is aimed at providing a regional overview of the relative magnitude of the air pollution problem in an area. The method- ology is not intended to provide air quality forecasts as no diffusion models have been used. For each system, estimates of average daily and peak-hour emissions of carbon monoxide (CO), hydrocarbons (HC), and oxides of nitrogen (NO ) X were made and compared. Computer programs have been developed to do much of the work and are available. The three-stage Auto Emissions Model developed in this study is shown in the figure. For each system tested, a trip generation sub-model is used to determine automobile vehicle trip origins, a travel description sub-model is utilized to convert those trips into travel characteristics and, finally, an emission sub-model is used to convert these travel parameters into estimates of pollutant emissions. The emission sub-model utilizes average speeds, to determine the CO, HC, and NOX emission rates, in pounds per vehicle mile, for each facility type. This rate is then multiplied by the VMT on each facility type to determine the total pollutants emitted for each of the sub-areas. This process is carried out for both peak-hour and daily conditions. The emission factors used in this study were supplied by the E.P.A. CO and HC emissions per vehicle-mile of travel decrease with increasing speeds while emissions of nitrogen oxides are assumed to be constant for all speeds. B-72 ------- (PHASE 1 I PHASE 2 IPHASE 3 DATA REQUIREMENTS \|TRIP GENERATI ON MODEL iTRAVEL DESCRIPTI ON MODEL iPOLLUTANT EM ISSIONS MODEL SOCIO-ECONOMIC FORECASTS HOUSEHOLD INCOME, DISTRI3UTIONOFPOPULATION AND EMPLOYMENT TRANSIT SYSTEM ALTERNATIVE HIGHWAY SYSTEM ALTERNATIVE ENVIRONMENTAL PROTECTION AGENCY EMISSIONS FACTORS HIGHWAY NEEDS MODEL /VEHICLE^ /MILES-OF- ITRAVEL AND V SPEED BY . X^FACIIITY TYPE EMISSION MODEL EMISSIONS OF OXIDES OF NITROGEN SA-2579-29 FIGURE B-5 FLOW CHART OF AUTO EMISSIONS MODEL There was concern over combining trip specific speed-emission factors with facility specific speeds. At the time the study was initiated, however, there were no data available to resolve this inconsistency. The most important relationship used in the travel description sub-model is between vehicle-miles of travel (VMT) density, vehicle trip origin density, and expressway supply. In the Washington, D.C. area, the relationship was found to be the same as that found in the New York region, that is: where: VMT = vehicle-miles of travel per square mile B-73 ------- VTO = vehicle trip origins per square mile FE/FO = proportion of total roadway surface area made up by expressways It was found that the expressway speed estimates calculated for the New York region were too high for use in the Washington area; therefore, the source deck was recompiled using a speed equation specifically calibrated for the Washington region. To obtain an estimate of peak-hour speeds, the 24 hour VMT was factored to peak hour volumes per lane by facility type, using observed peak-to-daily and directional flow factors. The Technical Advisory Committee eventually decided that the EPA emission factors published in the EPA document, "Air Pollution Emission Factors"were the best available for current use. These factors were modified somewhat by EPA to account for 1976 conditions and the fact that the age distribution of automobiles in the Metropolitan region was different (younger) than the national akjfc average. This resulted in a slightly lower base emission rate for the region, but the proportional changes in rates for speed variations remained identical to that published. Preliminary data from a current five-car study in California indicates that the impact of control devices might well be to flatten out the CO and HC curves, that is, emission rates may not decrease as rapidly for a given speed increase as they did under the old curves. M. Ja McGraw, and R. L. Duprey, "Compilation of Air Pollutant Emission Factors" (Preliminary Document), Environmental Protection Agency, Research Triange Park, N. C., April, 1971. "Motor Vehicle Emission Factors for Metropolitan Washington, B.C.," Environmental Protection Agency, Office of Air Programs Memo, November 4, 1971. B-74 ------- EMISSION MODELS CHECKLIST Input Data Requirements Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Within User Program Supplies x From AP-42 x NP NP NP Proxies/Comments Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area NP NP NP NP NP NP NP NP NP NP NP NP NP Average speeds for each mode based on total VMT by mode NP = No provision B-75 ------- Output Available Pollutants -CO Yes -HC Yes -NOX Yes -Particulates No Availability of Program Documentation Yes No Included in report x Referenced in report x Language Basic FORTRAN IV Equipment IBM 360/30 Validation/calibration; Calibrated/validated on emissions for Metropolitan Washington, B.C. Time resolution; Peak/off-peak Spatial resolution; Area specified by zonal configuration B-76 ------- XII NATIONAL AIR POLLUTION CONTROL ADMINISTRATION St. Louis SOp. Dispersion Model Study—Basic Data. D. B. Turner and N. G. Edmisten, Unpublished report, November, 1968. A. Author's Abstract '•'••« I'1:1'ft con- , 'AS ck'S' riptions -f busio data lor- ,i (iirco , o.it'i period used with a research study formulating a dispersion model for the St. Louis Metropolitan Area. This includes emission information ....] --.our •..-• 11 ?-,'\-" (iio'-itii ,i:; well .;s i.iet ion , r... i "• t j. ., ia.' o;. sums for each 2-hours for the three-month period, hourly meteorological data from routine and special instrumentation, and 24-hour sulfur dioxide concentrations for 40 locations and 2-hour concentrations for 10 locations. B. Summary This unpublished report outlines methods of computing the number of dwelling units per grid square and the population per grid square, from basic census data. 1. Estimation of the Number of Dwelling Units per Grid Square When the census tracts are larger than a grid square, the pro- cedure is to first calculate the total number of grid squares per census tract by dividing the number of acres in the tract by the number of acres in a grid square. The number of inhabited grid squares in the tract is estimated, deleting rivers, lakes, parks, etc. Weights are then assigned individually to the squares to adjust their dwelling unit density to that of the total census tract. The number of each type of heating unit (coal, gas, or oil) is estimated for an average grid square from census data. B-77 ------- Finally, the number of each type of heating unit per average grid square is multiplied by the assigned weights, giving the number of each type of unit in the grid square in question. When the census tract is smaller than a grid square, the fraction of the total census tract in each grid square of which it is a part is estimated, and weights are assigned to each portion of the tract to compensate for nonresidential areas. The total number of each type of unit in the census tract is multiplied, individually, by the fraction of area and by the assigned weight. These products are summed for each of the census tracts within the grid square, yielding the number of each type of unit in the square. 2. Estimation of the Population per Grid Square When the census tract is larger than a grid square, the total number of grid squares per census tract is calculated, and the number of inhabited grid squares in the tract is estimated. Weights are individually assigned to the squares to adjust their population density to that of the total census tract. By dividing the population of the total census tract by the number of grid squares per tract, the number of people per grid square is found. Finally, multiplying the average grid square population by the assigned weight gives the population of the grid square in question. If the census tract is smaller than a square, the fraction of the total census tract in each grid square of which it is a part is estimated, and a weight is assigned to each portion of the tract to compensate for nonresidential areas. The total population of the tract is multiplied, individually, by both the fraction of area and the assigned weight, and the sum of these products, for each census tract within the grid square, equals the population of the square. B-78 ------- C. Model Availability A computer code for the methodology presented is not included in the report. B-79 ------- XIII NATIONAL AIR POLLUTION CONTROL ADMINISTRATION Emission Factors. A. H. Rose, Jr., W. D. Krostek, National Air Pollution Control Administration, June, 1969. A. Author's Abstract A method has been developed for the prediction of gaseous pollutants emitted into the national environment from on-the-road gasoline-powered vehicles. The method involved two quantities, the emissions emitted per mile traveled, and the total mileage traveled. Vehicular emissions were characterized by engine type and load under actual operating conditions. Five vehicle types and two use classes were used to define characteristic vehicular emissions. Vehicle types were automobile (auto), light duty truck (LDT), and three weight classes of heavy duty trucks (HDT). The two use classes were urban and rural. The method allows comparison among the possible degrees of control, methods of control, and implementation dates. The method could be utilized on a regional rather than national basis, and could include particulates as well as gaseous pollutants. B. Summary 1. Inputs The method requires knowledge of: (1) the mean annual road emissions of a characteristic vehicle in mass per distance traveled, and (2) the total distance traveled during the calendar year by that portion of the total vehicular population characterized by the vehicle. Classi- fication of vehicles was by use: urban or rural driving; and by general type: auto, LDT [GVW less than 6,000 Ibs], HDT(II) [GVW of 6,000-10,000 Ibs], HDT(III) [GVW of 10,000-19,500 Ibs], HDT(IV) [GVW over 19,500 Ibs]. Vehicles in a class were categorized by the model year (the model year is the calendar year the vehicle entered the population). This is basically B-81 ------- characterization of vehicular emissions by engine type, load, and age. Reference characteristic emissions were obtained for each of the vehicular types and use classifications. The deterioration factor used to account for vehicle age was considered to be solely a function of accumulated road mileage, thus accommodating changes among vehicles in driving usage. It was assumed that vehicular deterioration will not change and that the mileage distribution with age will also remain constant. HDT vehicles were assumed to deteriorate as automobiles, assuming that the increase in mileage traveled is offset by more durable construction. Three gaseous pollutants were considered, hydrocarbons (based on a hydrogen/carbon ratio of 1.85 to 1.00), carbon monoxide, and nitrogen oxides (as NO ). Emissions from evaporative sources and crankcase blowby ^ were presumed to consist entirely of hydrocarbons. Rural driving was characterized by hot start 45 mph (average route speed) operating conditions; urban driving, by cold start 25 mph (average) operating conditions. The relationships among the various operating conditions are themselves independent of mileage. REFERENCE AUTOMOTIVE EMISSIONS N0n CO HC Type - exhaust - exhaust - exhaust Value gm/mile ppm gm/mile % gm/mile ppm 1962 5.757 1500 87.234 3.7 5.678 456 1963- 1967 5.757 1500 87.234 3.7 5,678 456 1968- 1969 5.757 1500 35.103 1.5 3.428 275 1970 5.757 1500 23.000 1.0 2.200 176 1971- 1974 5.757 1500 23.000 1.0 2.200 176 1975- 1990 0.960 250 11.500 0.5 0.611 49 HC - evapora- gm/mile tive HC - crank- case gm/mile 2.767 3.151 2.767 2.767 2.767 0.490 Concentration values are based on exhaust volume of 70.68 standard cubic feet/mile and pollutant densities of 16.33 gm/scf for hydrocarbon, 33.11 gm/sfc for carbon monoxide, and 54.30 gm/scf for nitrogen dioxide. Emissions are for federal composite cycle. B-82 ------- 2. Outputs National or regional gaseous pollutants emitted from on-the-road gasoline powered vehicles. 3. Resolution Annual national or regional. 4, Validation None B-83 ------- Composite Cycle HC . 0.9101 CO 0.8217 NO0 . 0.8743 2 1 Hot Cycle HC CO NO, 0.911 0.679 0.8742 Hot 25 mph Road HC CO 1.266 1.392 1.000 Cold 25 mph Road HC CO 1.120 1.085 1.260 Composite Cycle to Hot Cycle Hot Cycle to Hot 25 Road Temperature HC CO NO,, 0.679 0612 1.000 Speed Hot 45 mph Road HC CO NO,, 1.000 0.920 1 295 Seasonal (Summer to Annual) Additional (Regional) Mean Annual Mean Annual Exhaust From Exhaust From Urban Vehicle Rural Vehicle FIGURE B-6 EXHAUST EMISSION FLOW DIAGRAM SA-2579-30 B-84 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies From AP-42 Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed x x X X X X X X NP NP NP NP NP NP x NP Proxies/Comments Uses vehicle mix for year of interest in study Average rural and urban routes NP NP NP NP Spatial Distribution -Links -Grinds -Area NP NP NP NP NP NP = No provision B-85 ------- Output Available Pollutants -CO -HC -NOX -Particulates Yes Yes Yes No Availability of Program Documentation Included in report Referenced in report Language Equipment Validation/calibration: Time resolution: Spatial resolution; Yes NP NP No x None Annual National or regional B-86 ------- XIV NORTHERN RESEARCH AND ENGINEERING The Potential Impact of Aircraft Emissions Upon Air Quality. M. Platt, R. C. Baker, E. K. Bastress, et al. , Northern Research and Engineering Corp., Cambridge, Mass., 02139, Rept. APTD-1085, December 29, 1971. A. Author's Abstract Three categories of emissions were presented: aircraft operations, non-aircraft airport operations, and airport surroundings. In general, each of the sources in these categories can be accurately represented as either point, line, or area sources of time-dependent strength (i.e., emissions per unit time). However, to reduce the complexity of analysis in this study, each emission source is represented by one or more con- tinuous point sources of constant strength over the time period being considered. Each point source is a specific location within the airport or its environs. The strength Q of a point source was evaluated as the total emissions E associated with the point during the time period, divided by the length T of the period. B. Summary 1. Aircraft Operation The modes of operation of a typical aircraft which were con- sidered in this study have been established as: (1) approach, (2) landing, (3) taxi [after landing and before take-off], (4) idle and shutdown, (5) maintenance, (6) start-up and idle, (7) idle at runway, (8) take-off, (9) climb-out, and (10) fuel venting. B-87 ------- Operational modes (4) idle and shutdown (6) start-up and idle (5) maintenance (7) idle at runway are characterized by pounds of pollutants per aircraft class. The total is calculated by using characteristic times periods for each mode for each aircraft class. Operational modes (3) taxi (1) approach (9) climb-out (2) landing (8) take-off are characterized by pounds of pollutants per aircraft class. Dis- tribution of traffic among runways, average taxi times, characteristics of the airport under study, and the times and positions along represen- tative flight paths are used to calculate the amount and distribution of the pollutants. Operational mode (10) fuel venting is characterized by total pounds of hydrocarbon per aircraft class and is associated with a point source at 600-meter altitude in take-off. 2. Automobile Travel Automobile travel within the airport is characterized by a constant emission of pollutant per automobile per unit automobile travel. The total emissions due to automobile operation for a time period were calculated using: B-88 ------- a. Average travel per automobile within the airport b. Fraction of automobiles entering the airport per passenger c. Average number of passengers per aircraft d. Aircraft activity data. The first two items are considered characteristic of the airport, while the third item is considered characteristic of an aircraft class. The total emissions were distributed along the airport roadway system, which in addition to highways, included roads within the parking areas. The distribution is accomplished by locating point sources at stations which divide each roadway into a number of equal segments. The emissions of each segment were equally divided between the two bounding stations. This resulted in the same type of distribution as would be obtained for trapezoidal integration of a uniform line source of emissions. Service vehicle emissions were calculated in an approximate manner as 50% of automobile emissions. They are then included in the distribution of the automobile emissions. 3. Model Sensitivity The mathematical model of airport emissions and the dispersion of those emissions used to predict pollutant concentrations involves two categories of physical parameters: a. Meteorological parameters, which consist of atmospheric stability, wind speed, wind direction, mixing height, and for short-term calculations, wind persistence. b. Source parameters, which consist of the strength and location of emission sources with respect to a receptor. Parameters in the first category were the subject of a sensitivity analysis, with parameters in the second category investigated indirectly in the analysis of control techniques. B-89 ------- 4. Accuracy of the Model Airport emissions are modeled by a large number of continuous point sources of constant strength distributed throughout the airport. This results in two limitations on the validity of the model. First, the model is limited to time periods that are much larger than character- istic times of individual aircraft activity. Second, the model is limited to receptors that are not in close proximity to the emission sources. Since an airport has many emission sources which would be best represented as lines or areas, the assumption of point sources results in an important limitation of the model. In providing a check of the results obtained using the model, a comparison was made with emissions and concentrations reported for Los Angeles International Airport. While the emissions for carbon monoxide are in excellent agreement, agreement for the other pollutants is poor. The aircraft emissions of particulates, nitrogen dioxide, and total hydro- carbons for Los Angeles International Airport are greater than those obtained using the model by factors ranging up to 6.7 for particulates. Since the activity data were similar for the two cases, the disagreement can be traced directly to the emission factors. Concerning the other pollutants, it is significant that agree- ment is so good for carbon monoxide which can be measured most accurately in engine emission tests and so poor for the other pollutants where different measurement techniques can give very different results. The measured concentration of carbon monoxide is a factor of 2.8 larger than the analytical value. It is felt that the imprecision with which the surroundings were modeled is the likely cause for this discrepancy (for instance, emissions of automobiles on roadways adjacent to the airport not directly included in the model). Therefore, concen- trations due to aircraft emissions alone should be predicted more accurately than concentrations due to all emissions. B-90 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed x x x x NP x x NP NP NP NP x x X X NP From AP-42 Proxies/Comments NP NP NP NP Spatial Distribution -Links -Grinds -Area x x x NP NP NP NP = No provision B-91 ------- Output Available Pollutants -CO Yes -HC Yes -NOX Yes -Particulates Yes Availability of Program Documentation Yes No Included in report x Referenced in report x* Language NP Equipment NP Validation/calibration: Los Angeles Airport - 1970 data Time resolution: 1 hour Spatial resolution; Airport boundary * M. Platt, K. M. Chug, and R. D. Siegel, "Computer Program for the Air Quality Analysis of Airports" (NREC Report No. 1167-2), Northern Research and Engineering Corporation, Cambridge, August 1971. B-92 ------- XV ONTARIO DEPARTMENT OF THE ENVIRONMENT Information System Descriptive Manual, Air Management Branch, Department of the Environment, Province of Ontario, Canada, January, 1971. A. Author's Abstract None given. B. Summary The Toronto emissions-inventory methodology considers three source categories: (1) major intense and discontinuous sources (point and line sources); (2) numerous minor sources (area sources); and (3) sources beyond the grid system boundary. If emissions are greater than 100 Ibs per day, the source is con- sidered a point source; otherwise, its emissions are included in area source calculations. The 100 Ibs per day criterion may be modified. The Toronto area is divided into 1 kilometer grid squares for area sources, with point sources located on a 5 X 5-kilometer grid to a tenth of a kilometer. Area sources include emissions from motor vehicles, shipping, railroads, aircraft, and gasoline storage, from space heating and waste disposal of apartments, schools and universities, industrial and commer- cial sources, and public buildings, and from space heating of residential dwellings. All minor sources within one grid square are summed, and the total is distributed evenly over that square. Line sources are treated as special cases of area sources. Sources beyond the grid boundary are treated as providing a flux through the border, with assigned source strength and initial vertical B-93 ------- distribution,, These parameters maintain constant values along the length of the border. The five pollutants considered are sulfur dioxide, nitrogen oxides, particulates, carbon monoxide, and hydrocarbons. Provision is made for sources which, under certain conditions, must switch from their regularly used fuel to another type, thus changing their emission patterns. The information system is designed to accept new data and can modify old data. The capability exists to modify classes of emissions by a fixed percentage. 1. Automotive Emissions • Roadway emissions are divided into five classes: (1) super- highway (line source), (2) other highway, (3) major artery, (4) residential, and (5) business street. Information is collected such as the vehicle count per day for all roads in each grid square (for residential streets a count of 500 is assumed, slightly higher for residential streets with large apartment buildings), and road lengths in each grid square. Yearly emission quantities are found by multiplying the vehicle count per day by road length and by 365. and then applying the emission factor appropriate to the emission class. Trie lourly o daily emissions are computed using nie appropriate diurnal, daily and seasonal factors that have been tabulated. 2. Po ,t_ an_d Oi '>r>r Are i Sonrr-o E~ "•s^ion,0 Patterns for point source operations, process and nonprocess, were developed based on days of the week in operation and the number of shifts per day. Seasonal and daily space heating variations for both point and area sources were developed using degree day fluctuations, with hourly variations derived from natural gas demand records over 24-hour periods during different seasons. Other area source emission variations were determined in a similar fashion, based upon usage or frequency variation data. B-94 ------- To obtain hourly emissions from the total annual amount of pollutant, the procedure outlined below is applied. The annual total is: a. Adjusted by a seasonal factor. Depending on the season, one of four seasonal percents (which add to 100) is mulitplied by the annual pollutant total and then divided by 25, b. Adjusted to a daily rate, by dividing by 365, c. Adjusted for day of week, using one of the seven percentages associated with a particular day, and d. Adjusted for weather if a heating emission. For area sources, the factors depend on dwelling type and fuel type. For point sources, fuel combustion emission are adjusted by a space heating percent. This is done by multiplying by the amount of the degree day total and dividing by the degree year total. e. Only area sources also adjusted by hour of the day. A table of diurnal codes, each containing 24 numbers which add to 100, are applied so that emissions are spread over the hours of the day, based on the components of the code. C. Model Availability As stated earlier, the Toronto emissions inventory methodology was developed by the Department of the Environment, Ontario Province, Canada. The emission model is in the form of a computer code, but it is not generally available for public dissemination. Accordingly, the model can only be obtained through a formal request to the Department. B-95 ------- XVI PACIFIC ENVIRONMENTAL SERVICES Controlled Evaluation of the Reactive Environmental Simulation Model. L. G. Wayne, A. Kokin, and M. I. Weistaurd, Final Report, Pacific Environmental Services, Inc., Contract No. 68-02-0345, prepared for Environmental Protection Agency, EPA R4-73-013a, February, 1973. A. Author's Abstract The development and validation of an operational version of the Reactive Environmental Simulation Model (REM) were completed, REM was specifically designed to handle large chemical mechanisms to assess the impact on air quality of air pollution control devices, fuels, propul- sion systems, stationary sources, and transportation systems where thorough evaluation of emissions, emission constituents and reaction rates are required. The version delivered to the U.S. EPA under this contract contains a mechanism involving 32 reactions, 12 accumulating species, and 12 nonaccumulating species. Larger or smaller mechanisms, also, can be readily inserted into REM for specified purposes. REM is based on a Lagrangian moving coordinate system which enables the numerical simulation of the chemical reactions that take place in a parcel or column of air moving along a dynamic wind trajectory. The tra- jectory approach gives REM considerable flexibility and is adaptable for use in a number of practical operational situations. These include short- term air quality forecasting; analysis of the impact of sources of air pollution at various designated locations on the air quality at specified receptor points; interpolation of contaminant concentrations and dosages at locations not covered by air monitoring stations, and control strategy evaluations. B-97 ------- REM contains such user features as reverse and forward trajectory routines; automatic and objective interpolation from input emission inventory, meteorological and air quality data bases; a chemical dynamics routine capable of accommodating mechanisms based on elementary chemical reactions; and automatic estimation of mixing depth and solar irradiance based on input of local weather and sun angle data. REM has attained a running time which makes it cost-effective for practical use. On an IBM System 370/155, the real-time-to-simulation ratio for the 31-step .mechanism is 150:1; for carbon monoxide alone it is 3000:1; for some of the shorter mechanisms available it should run more than 300:1. The program, also, is user-oriented in that it provides simple input procedures, ^er documentation, receptor point and time-of- day selectivity, flexibility in treating specific problems, and ability to conveniently select any of an infinite number of trajectories cm any number of days o± interest. 'i.ie modular construction of KEIU, also, makes it easy to add, replace, or delete individual modules. REM can be used as an unlimited receptor point model and to chronicle emission inputs contributing to the air quality at any receptor point. The validation record of REM over a large number of runs may be summarized as follows: • CO, less than a factor of 2 in more than 80% of the comparisons; 40% agreement within one parr per million. & 0 , within a factor of 2, 75% of the comparisons. O • NO, within a factor of 2, 75% of the comparisons; 90% agreement within 0.02 ppm. • NO2, within a factor of 2, 60% of the comparisons. The shapes of predicted time profiles of contaminant concentrations are generally similar to those observed at air monitoring stations. B-98 ------- The ease and economy at which REM can be utilized under a variety of data availability situations make potential improvements in accuracy and application possible on a cost-effective basis. B. Summary 1. Treatment of Source Emissions A subroutine (module) of the REM program computes emissions of NO, CO, and reactive and less reactive hydrocarbons from freeway and street traffic. Emissions of NO and reactive and less reactive hydro- carbons are calculated for stationary area sources. All urban grid squares are chosen as four square-mile blocks, although the kilometer may also be used as the standatd. Emissions within these squares are viewed as essentially uniform and continuous. Inputs to the module include time, position of the pollutant column, the distribution of daily traffic on streets and freeways, emission factors, and diurnal usage curves. Traffic and grid system distributions and diurnal curves for the Los Angeles Basin are taken from the emissions treatment developed by Systems Applications, Inc., a review of which appears in this appendix. Traffic emissions are computed as a function of time by applying the emission factors for the respective pollutants to the traffic levels at the air parcel location, qualified by the daily traffic usage curve. Area source emissions as a function of time are found by applying a con- stant diurnal usage curve to grid system distributions of average hourly pollutant emissions. 2. Module Method of Operation Following is the procedure by which pollutants are injected into an air parcel. B-99 ------- 1. Initialization (background concentration). 2. Traffic distribution curves are evaluated. 3. The grid in which the trajectory point lies is found. 4. The three closest adjacent grids are found (those with the closest midpoints). 5. The distance from the trajectory point to each grid midpoint is calculated. 6. Is the trajectory point on the midpoint of the grid square in which it lies? If yes, go to step 9; if no continue. 7. The contribution to the parcel from each traffic grid for each pollutant is calculated (the contribution is inversely proportional to the square of the distance from the trajectory point to the grid midpoint). 8. If area sources are to be added, the area source con- tribution from each grid square for each pollutant is calculated (inversely proportional to the distance squared). Skip to step 11. 9. The traffic contributions of the trajectory grid only are calculated. 10. If area sources are to be added, the area source con- tributions of the trajectory grid only are calculated. 11. The traffic and area contributions are then modified by the appropriate diurnal usage curves and added. C. Model Availability A computer listing of the emission inventory is included in the supplemental User's Guide of the report. B-100 ------- DAYLIGHT SAVINGS TIME SET TIME TO DAYLIGHT TIME 1 ! FIND THREE CLOSEST ADJACENT GRIDS SET MIDPOINTS OF THE FOUR GRIDS i FIND DISTANCES OF TRAJECTORY POINT TO GRID MIDPOINTS EVALUATE TRAFFIC DISTRIBUTION CURVES CALCULATE CONTRIBUTIONS OF TRAFFIC GRIDS (1/DIST2) FIND GRID IN WHICH TRAJECTORY POINT LIES SA-2579-31a FIGURE B-7 FLOW OF SOURCE SUBROUTINE B-101 ------- ADD TRAFFIC CONTRIBUTIONS OF TRAJECTORY GRID ONLY ADD AREA SOURCE CONTRIBUTIONS OF TRAJECTORY GRID ONLY CALCULATE CONTRIBUTIONS OF SOURCE GRIDS (1/DISJ2) DAYLIGHT SAVINGS TIME RESET TIME TO STANDARD TIME SA-2579-31b FIGURE B-7 FLOW OF SOURCE SUBROUTINE (concluded) B-102 ------- XVII RUTGERS UNIVERSITY Comparison of Air Pollution from Aircraft and Automobiles (Pro.-ject Eagle) . C. Bright et al. , Rutgers University, New Brunswick, New Jersey, AD713913, September, 1970. A. Author's Abstract None given. B. Summary Emissions from either aquadrome or roadway areas are considered to constitute a surface of randomly distributed multiple sources. Tiiis surface is assumed to be a uniform area source with respect to a receptor located a variable distance downwind from the edge of the aquadrome or roadway segment. Emissions are calculated by assuming aircraft passenger load factors and the number of passengers per commuting automobile. The number of passengers and vehicle road or air miles from each source to central locations are used to determine the total number of auto and aircraft miles. This mileage is used to calculate emissions using the following emission factors: B-103 ------- Hydrocarbons Carbon monoxide Nitrogen oxides Particulates MOTOR VEHICLE EMISSIONS (gin/passenger mi) 1970* Emissions 8.81 38.95 10.35 0.40 197 of Actual 12.10 72.35 S.84 0.47 1975* Goals 0.40 8.87 0.73 0.08 19751" Actual 8.51 45.24 7.84 0.38 •* 1980 Goals 0.20 3.79 0.32 0.02 1980^ Actual 3.11 18.87 3.52 0.17 TOTAL AIRCRAFT VEHICLE EMISSIONS: APPROACH, CRUISE, TAKE-OFF AT 100% LOAD FACTOR (gm/passenger mi) 1970 1975 1980 Hydrocarbons Particulates Carbon monoxide Nitrogen oxides 0.3696 1.0107 1.4717 0.8151 0.3696 1.0107 1.4717 0.8151 0.3696 1.0107 1.4717 0.8151 Based on HEW estimates of emission-controlled vehicles that will be in production at that date. In converting the grams per vehicle mile to grams per passenger mile, an average automobile passenger occupancy of 1.24 persons is used. For aircraft, occupancy load factors of both 100% and 50% were con- sidered. The 100% load factor was demonstrated to be feasible in the Project Eagle Study. As shown in the tabulation above, the values for aircraft in grams per passenger mile is computed allowing separate emission rates for approach, cruise, and take-off. In converting the grams per vehicle mile to grams per passenger mile, a trip length of 20.8 miles is used for the 53-passenger aircraft. This 20.8-mile trip length represents the average air passenger distance traveled in an urban air transportation system during 7 a.m. to 9 a.m. and 4:30 p.m. to 6:30 p.m. between Manhattan and the 11 transportation centers in Connecticut (Bridgeport, New Haven, and Stamford), New Jersey (Linden/Rahway, New Brunswick, Paterson and Newark), and New York (Farmingdale, Hempstead, Mt. Vernon, and White Plains). 43ased on a methodology considering automobile age and vehicle usage corresponding to automobile age. This procedure is currently being used by the National Air Pollution Control Administration, Durham, North Carolina. B-104 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies From AP-42 Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area x x X X X X X X X NP X NP NP NP NP x NP NP NP x Proxies/Comments By mix on road for year of study interest NP NP NP NP NP NP NP NP = No provision B-105 ------- Output Available Pollutants -CO -HC -NOX -Particulates Yes Yes Yes Yes Availability of Program Documentation Yes Included in report Referenced in report Language Equipment Validation/calibration; None NP No x Aircraft emission factors based on 1968 NRC work - no program given. Auto emission factors based on NAPCA data. Time resolution; 4 hours Area; Tri-Stav.-. (Connecticut, New Jersey, and New York) B-106 ------- XVIII SACRAMENTO REGIONAL AREA PLANNING COMMISSION General Inventory of Air Pollution Sources and Emissions. The Air Pollution Threat, Sacramento Regional Area Planning Commission, NTIS No. PB191382, October, 1969. A. Author's Abstract The Sacramento Region has three major sources of pollution: transportation, agriculture waste burning, and other stationary sources. The report was directed at estimating the amounts of organic gases, carbon monoxide, nitrogen oxide, sulfur dioxide and aerosols that are emitted from their sources. The predominant emission is carbon monoxide with an estimated 1,215 tons being produced daily. Carbon monoxide is produced mainly from gasoline powered motor vehicles. The other emissions that are produced in the region are organic gases(267 tons/day), nitrogen oxide (87 tons/day), aerosols (46 tons/day), and sulfur dioxide (9 tons/aay). B. Summary An emission inventory of the Sacramento, California region was compiled for organic gases, carbon monoxide, nitrogen oxides, sulfur dioxide, and aerosol emissions. Sources were classified as transportation, agricultural waste burning, and stationary sources. The transportation category is divided into gasoline and diesel powered motor vehicles, railroads, and aircraft. Pollutant emission is based on estimates of the different fuels used. Gasoline usage is based on an estimate of 1.1 gallons per day per person. Emission factors for gasoline and diesel powered vehicles take into account existing smog control devices. B-107 ------- Agricultural emissions arise from agricultural waste burning and crop processing, the latter being included with stationary sources. In the Sacramento region 15% of total agricultural waste is burned. Emission figures represent tons of emissions produced on a straight daily average. Seventy-five percent of the burning is done from October to February, with emission figures higher during this season. Stationary sources include agricultural product processing, where 1% of the total amount of crops processed is estimated as the amount of particulate lost due to processing. Annual figures are divided by the number of days per year to obtain daily values. Emissions from petroleum handling result when tank trucks, service station tanks, and automobile tanks are filled. A daily figure for gallons handled is found by multiplying the population of a region by the estimate of gallons per day per person used. Emissions estimates for solvent use, metal, and mineral processing, and chemical manufacturing are obtained from regional agencies. To estimate emissions from the incineration of solid waste, the total amount of refuse burned daily is found by assuming an average figure of 2 Ibs per person per day and multiplying by the population. This amount is divided between municipal dump burning (75%) and backyard burning (25%). Backyard burning is further divided into incinerator, open, and landscape burning. Heating of homes and industrial fuel use are the main sources of emissions from fuel combustion. Natural gas figures are obtained from the utility company, and bottled gas is accounted for either directly or by increasing the yearly utility company figure by 10%. Total daily gas is divided between domestic (75%) and industrial (25%) sources. C. Model Availability A computer program for the emissions model is not included. B-108 ------- XIX STANFORD RESEARCH INSTITUTE A Practical, Multipurpose Urban Diffusion Model for Carbon Monoxide. F. L. Ludwig, W. B. Johnson, A. E. Moon, and R. L. llancuso, Stanford Research Institute, Menlo Park, California, September, 1970. A. Author's Abstract This report describes the development and current status of a receptor-oriented diffusion model that can be applied to urban areas to give the following outputs: (1) carbon monoxide (CO) concentration isopleths for a given set of meteorological conditions and times of day, (2) sequences of hourly CO concentrations at specific locations for given sequences of meteorological conditions, and (3) climatological summaries of CO concentration for specific locations if a historical record of meteorological data is available. The model can be used to obtain the frequency distributions of concentrations averaged over various time intervals for specific hours of the day or days of the week. B-109 ------- B. Summary Because nearly all urban carbon monoxide emissions are from internal combustion vehicles, the model assumes emissions to be at ground level. Furthermore, the emissions within each segment are assumed to be uniformly distributed. The inventory of vehicular emissions has two components: (1) primary network link emissions from vehicles traveling on the network of major arterial streets and freeways, and (2) secondary background emissions from vehicles traveling over the less densely traveled local and feeder streets. The primary network links for which the emissions are computed are the sections of major arterial streets and freeways between inter- sections with other major arterials and freeways. The average length of the links on the primary network is approximately one mile, although the links may be much shorter in densely traveled downtown areas, and more widely spaced in outlying areas. Traffic volumes on links vary hour-by-hour over a day and are different for weekend days than for weekdays. Further, there are smaller seasonal variations. Highway engineers and planners average these variations into a quantity that they call average daily traffic (ADT). For use in the synoptic model, the emissions must be expressed as hourly averages, so the average daily traffic is expressed as a function of time. The amount taken to occur within any given hour is based on the hourly distribution of trips compiled by the traffic study agencies of many areas. To calculate the contribution of the emissions from traffic links within the five closest segments, the links are first identified with the segments through which they pass. Then the length of the link that lies within a given segment is determined. The emissions within the four segments farthest from the receptor are calculated by a different technique than that used for the closer segments. B-110 ------- Before other computations begin, the city is divided into a grid of one-mile (1.6 km) squares. The traffic link data are used to deter- mine the average daily CO emission in each of these squares. This is accomplished by dividing each link into small increments, less than 0.08-km long. The average daily emission from each of these small pieces is determined by assuming emissions are spread uniformly along the link. Then, the emissions from all the small pieces of link within a square are added together to give total daily emissions for that square. This total is paired in the computer memory with the coordinates of the point at the center of the square. These calculations are done only once for the city. A trapezoidal grid system is used in the four outer emission segments. As the location of the receptor changes or the wind direction changes, these trapezoidal grids are superimposed on different parts of the fixed city emission grid described in the preceding paragraph. The average emissions at each point on the trapezoidal grid are determined by interpolation between the values at the points on the fixed grid. The average of the emission values determined for the points on the trapezoidal grid defines the emission rate for the segment. Historical link-volume data are obtained from traffic departments in the cities, towns, and counties in the region being studied. Because traffic varies according to seasonal, weekly, and daily cycles, an observation of volume for one day must be adjusted for the weekly and seasonal fluctuations. The resultant corrected value is recorded as the average daily traffic (ADT) for that location. The number of vehicle-miles traveled on streets not represented by the primary network is computed from an estimate of the total vehicle- miles traveled in the area and the total vehicle-miles on the links of the primary network. The local street mileage is taken to be the difference between the two. B-lll ------- Link speeds for use in the emission rate calculation are determined from averages of peak and off-peak travel hours on various kinds of route facilities. Speed data were obtained from the traffic survey. For most locations, the peak traffic hour speeds were taken as equal to 80% of the off-peak speeds. Peak-hour speeds were generally assigned to the four heaviest traffic hours of the day. The emission rate, e (g/vehicle-mile), is determined from the equation e = c SP where S is the average speed over the link, in mi/hr, and c and \|3 are constants. For vehicles in use before exhaust control systems, c = 1121 and P = -0.849, as determined by Rose et al. (1969) from observations on a number of vehicles in several locations. For autos reflecting the level of emission controls mandated for 1968-70, one hundred thirty-nine emission values were calculated from emission data presented by Beckman et al. (1967)2 and from actual observations of speeds, speed changes, and stops for a variety of road types and traffic congestion conditions. Regression analysis was used to examine the relation between emissions and average speed, both with and without controls. The following power function proved to be the best fit for 1969 model year automobiles: -0.48 e = 245 S 1 A. H. Rose and W. D. Krostek, "Emission Factors," U.S. Dept. of Health, Education and Welfare, National Air Pollution Control Administration, p. 5, 1969. 2 E. W. Beckman, W. S. Fagley, and 0. Sarto, "Exhaust Emission Control by Chrysler—The Cleaner Air Package," Air Pollution—1967 (Automotive Air Pollution), Hearings before the Subcommittee on Air and Water Pollution of the Committee on Public Works, United States Senate, Ninetieth Congress, First Session, Part 1, pp. 411-424, 1967. B-112 ------- Consideration of emission control progress yielded the emission model for 1980 and later model year automobiles: -0.48 e = 34 S C. Model Availability The APRAC model is fully documented in a recent report by Mancuso and Ludwig3. The program is available in two versions: one for the SRI CDC G400 computer and another for the EPA IBM 360/50. 3 R. L. Mancuso and F. L. Ludwig, "User's Manual for the APRAC-1A Urban Diffusion Model Computer Program," CRC/EPA Contract No. CAPA-3-6S(l-69),SRI Project 8563, Stanford Research Institute, Menlo Park, California, 1972. B-113 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area x x X X X X X X NP NP X NP NP NP x NP NP x x From AP-42 Proxies/Comments Average CO of vehicle mix for year of study interest Speed and distance NP NP NP NP NP NP NP NP = No provision B-114 ------- Output Available Pollutants -CO Yes -HC No -NOX No -Particulates NO j Availability of Program Documentation Yes No Included in report Referenced in report Language Equipment Validation/calibration; Calculated data were low compared with CAMP-observed data in St. Louis Time resolution; 1 hour Spatial resolution; Radial segments-22.5° out to 16 km from receptor point See Section C, Model Availability B-115 ------- XX STANFORD RESEARCH INSTITUTE Procedures for Estimating Highway User Costs, Air Pollution, and Noise Effects. D. A. Curry and D. G. Andersen, NCHRP Report 133, prepared by Stanford Research Institute, 1972. A. Author's Abstract: None given B. Summary Because the researchers were unable to locate data in the proper form, the approach to the estimation of emissions is to compute emissions for a vehicle of a particular type and characterize the emissions from that vehicle as "typical" of the vehicles that will be built during the next few years. The assumption is that the control technology will not change the patterns. The emission estimating procedure assumes that during the next few years vehicles will achieve carbon monoxide and hydrocarbon control through use of the engine mondification package and a catalytic afterburner, and will achieve nitrogen oxides control through exhaust gas recirculation. Emission characteristics of a vehicle using the assumed type of engine modification package were obtained for a vehicle manufactured by the Chrysler Corporation, Because most manufacturers are using this type of control, it is assumed that the vehicle is typical in that regard. The emissions for stopping and changing speeds were computed under the assumption that slowing down and speeding up would occur under con- stant acceleration conditions. The emissions added due to speed changes were computed by applying the emission characteristics of the test vehicle to the pattern of speed changes analyzed elsewhere in the report and by characterizing the results by the best straight line that could be drawn B-117 ------- through the emission values plotted against volume to capacity (v/c) ratio for each type of facility. The reference emissions provide an estimate of the pounds of each pollutant generated by travel over the analysis segments if all of the traffic consisted of automobiles meeting the exhaust emission control requirements for the 1969 model year. The table below gives the emission limits in grams per mile and their relative ratios for various model years as used in the conversion procedure (the ratios do not always track the emission limits exactly, owing to changes in test procedures). These relative values were combined with the fraction of the vehicle population represented by each model year as predicted by averaging the fractions for the years 1966-1970. AUTOMOBILE EMISSION LIMITS FOR VARIOUS MODEL YEARS Actual or Permissible Emission (gm/mi) Model Year 1967 and earlier 1968 and 1969 1970 1971-72 1973-74 1975 1976-79 1980 and later HC 13.6 5.1 3.4 3.4 3.4 0.41 0.41 0.20 CO 117 59 39 39 39 3.9 3.9 2.0 NO* 4.0 4.0 4.0 4.0 3.0 3.0 0.4 0.2 Emission Factor Used (1970=1.0) HC and CO 3.5 1.5 1.0 1.0 1.0 0.1 0.1 0.05 NOx 1.0 1.0 1.0 1.0 0.75 0.75 0.1 0.05 Use of the conversion factor assumes that the future controls will reduce emissions proportionally for all operating conditions. Use of a catalytic afterburner for control of carbon monoxide and hydrocarbons will result in emissions that approximate this assumption. B-118 ------- The emissions for single-unit gasoline trucks were estimated to be 2.5 times those of the reference automobile under all conditions. The last step was to incorporate a degradation factor for the pre- dicted increase in vehicle emissions due to wear, derived from data of Rose and Krostek. The degradation factor increases at a decreasing rate with average mileage drive, and hence with age: Vehicle Age (yr) 0 1 2 3 4 5 6 7 £ 8 Degradation Factor 1.00 1.05 1.12 1.18 1.20 1.22 1.23 1.24 1.25 A. H. Rose, Jr. and W. D. Krostek, "Emission Factors," Department of Health, Education, and Welfare, June, 1969. B-119 ------- a § 80 o 8 LU Q X O Z o 5 z o CO DC < O § .. — "" ^ / r / / X / I/ / / / / / / / / /co HC 0 20 40 60 80 1C I.U 0.8 I u I 0.6 I c c 0.4 C a a < c. Q 0.2 > 0 )0 SPEED STOPPED FROM — mph SA-2579-32 FIGURE B-8 AUTOMOBILE HYDROCARBON AND CARBON MONOXIDE EMISSIONS ADDED PER 1,000 STOPS 100 •n 80 60 UJ a x o z o ca oc < o 40 20 0 \ V \ 1 \ 2 ^ \ Ni 0 4 ' — • . 0 . -. 6 ^•""' ^ - — — 0 C CO 8 0 1 I.U u j no c c L !c O.b Z \| C U'4 c cc tt 4 c 0.2 g > Z 00 UNIFORM SPEED — mph SA-2579-33 FIGURE B-9 AUTOMOBILE HYDROCARBON AND CARBON MONOXIDE EMISSIONS PER 1,000 MILES OF DRIVING AT UNIFORM SPEED B-120 ------- n. d « OOCU/sPunod — NOaUVOOHQAH in (N Z o CO tc. cc o LU _i CJ I LU C O o a: LU a. to LJJ (D 2 < I CJ Q. CO O cc Ul Q X o o i u Q G < CO Z O UJ CO O 5 O m cc rs a g ooOl/sPunod o CM Noayvo B-121 ------- >- DC LU oc < CO CO CO CO z q i LU DC LU LLJ CJ tr. in u. LU QC I- oc 8 DC o I m DC D dOlOVd B-122 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies From AP-42 Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed Spatial Distribution -Links -Grinds -Area x X NP NP NP NP x x NP x NP x NP NP Proxies/Comments Based on 1969 reference auto, conversions supplied for other years NP NP NP NP NP NP NP NP = No provision B-123 ------- Output Available Pollutants -CO Yes -HC Yes -NOX NO -Particulates No Availability of Program Documentation Yes No Included in report x Referenced in report x Language NP Equipment NP Validation/calibration; None Time resolution; 1 hour Spatial resolution; Link B-124 ------- XXI STANFORD RESEARCH INSTITUTE A Preliminary Study of Modeling The Air Pollution Effects From Traffic Engineering Alternatives. F. L. Ludwig, R. C. Sandys, and A. E. Moon, Stanford Research Institute, Menlo Park, California, Air Pollution Control Association Journal, Vol. 23, No. 6, pp. 499-504, June, 1973. A. Author's Abstract Three separate mathematical models were combined to calculate the changes in carbon monoxide (CO) concentrations that might result from traffic engineering changes. The three models used were: (1) The Dynamic Highway Transportation Model (DHTM) which relates traffic flow patterns to physical parameters and traffic signal characteristics of a network; (2) an emission model that predicts CO emissions from traffic flow parameters such as number of stops, idling time, etc.; and (3) the APRAC-1A urban diffusion model which calculates CO concentrations from source distribu- tions and meteorological factors. The composite model was applied to traffic in downtown Chicago for a specific set of meteorological conditions, Results are compared for two traffic signal control schemes. In those blocks where concentrations were highest, the model indicates a 20% reduc- tion is possible through improved traffic signal controls. The model should be useful for testing other traffic control measures. B. Summary Moon has presented a methodology for calculating the amount of CO generated by a "reference" automobile under four conditions: idling, 1 D. A. Curry and D. G. Andersen, "Procedures for Estimating Highway User Costs and Air and Noise Pollution Effects," Final Report, Highway Re- search Board Project 708, Stanford Research Institute, Menlo Park, California, 1971. B-125 ------- steady cruising, stopping, and starting. The third and fourth conditions can be combined, with the results expressed as the amount of CO emitted per vehicle stop. Because of continual changes in the pollution control devices required for cars, and the changing mix of vehicle-model years, it is necessary to convert emissions for "reference" cars to averages for a particular year. The appropriate conversion factors have been provided in Reference 1. According to Moon, 0.15 gram per second (gm/s) of CO are released by an idling vehicle. The amount of CO released by a vehicle traveling at a steady speed is a function of that speed. Figure 1 shows an approximation of the function; the curve has been fitted with exponential functions to simplify the calculations. At the low speeds, below about 7 or 8 mph, the curve shown is based on the assumption that a slowly moving vehicle emits CO at the same rate as an idling vehicle; thus, the emissions per vehicle- mile will be inversely proportional to the speed. The emissions for each vehicular stop (and subsequent start) are an exponential function of the steady driving speed from which the stop was made. A factor is used to convert emissions of CO from a vehicle made in the reference year (1968) to emissions for years after 1971. The factor accounts for the mixture of vehicles manufactured in different years, based on national averages of vehicle life expectancy. It is based on existing legislation for the 1970 to 1975 era and on expected further limitation on emissions through 1980, and assumes no retrofit requirements. Moon's emissions model can calculate the emissions of CO on a given street segment, given: 1. Number of vehicle-miles of steady-speed driving on the link 2. The speed achieved during steady-speed operation 3! The number of stops and starts during the hour 4. The number of vehicle-seconds spent stopped and idling. B-126 ------- 700 500 300 200 > 100 I 8 70 "o E 50 30 b! 20 CO CO 5 LU LU~ 10 l I i E = 540 S E = 2350 S -1.72 E = 86 S -0.55 5 7 10 20 30 S, SPEED — mph 50 70 100 SA-2579-41 FIGURE B-12 RELATIONSHIP BETWEEN CARBON MONOXIDE EMISSIONS AND STEADY-SPEED DRIVING B-12 7 ------- The Dynamic Highway Transportation Model can provide the above items of information. For this study, an intermediate computer program was written to accept the magnetic tape output of the DHTM and create a punch card deck for input to the APRAC-1A Diffusion Model; this is not the most effi- cient method of treating the problem. This model is offered for central business district applications. For the outlying streets, the method using the formula developed by Rose 2 and Krostek to relate emissions to average speeds and traffic volumes is considered adequate. 2 A. H. Rose, Jr. and W. D. Krostek, "Emission Factors," Department of Health, Education, and Welfare, June 1969. B-128 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User From Program Supplies AP-42 Proxies/Comments Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed x x NP NP NP NP NP NP x > NP x x NP 1968 Reference; factored to later years NP NP NP NP Starts; stops; idling; steady-state speed; vehicle miles/ link Spatial Distribution -Links -Grinds -Area x NP NP NP NP NP NP = No provision B-129 ------- Output Available Pollutants -CO -HC -NOX -Particulates Yes No No No Availability of Program Documentation Included in report Referenced in report Language Equipment Validation/calibration: Time resolution: Spatial resolution; Yes No NP NP None 1 hour Intersection B-130 ------- XXII SYSTEMS APPLICATIONS, INC. (l) Contaminant Emissions in the Los Angeles Basin—Their Sources, Rates, and Distribution. P. J. Roberts, P. M. Roth, and C. L. Nelson, Appendix A., "Development of a Simulation Model for Estimating Ground Level Concentrations of Photochemical Pol- lutants," Systems Applications, Inc. Report No. 71 SAI-6, Contract No. CPA 70-148, prepared for the Environmental Protection Agency, March 1971. (2) Extensions and Modifications of a Contaminant Emissions Model and Inventory for Los Angeles. P. J. Roberts, M. Liu, S. D. Reynolds, and P. M. Roth, Appendix A, "Further Development and Validation of a Simulation Model for Estimating Ground Level Concentrations of Photochemical Pollutants," Systems Applications, Inc., Report No. R73-15, Contract No. 68-02- 0339, prepared for the Environmental Protection Agency, January, 1973. A. Author's Abstract (1) Perhaps the most tedious and mundane aspect in the development and validation of a simulation model of reaction and dispersion processes in the atmosphere is the compilation of a complete contaminant emissions inventory. Yet, such an inventory is a sine qua non in model validations and, if done properly, the emissions estimates probably constitute the most precise segment of requisite input data. Contrast, for example, the relatively low magnitudes of errors in emissions estimates with the imprecision of wind speed and direction estimates, both at the surface and aloft, as well as with the uncertainties in estimates of the variation in mixing depth with location and time. Furthermore, as emissions inventory need be carried out but once to serve as an adequate representation of a region, whereas meteorological data must be collected for each validation day, and for the purposes of this modeling venture, represented through hourly variation in wind field and mixing depth. It thus seemed wise B-131 ------- to put a considerable effort into the establishment of an accurate emissions inventory for the Los Angeles Basin. Particular emphasis was placed on developing a detailed representation of the spatial and temporal traffic distribution in the Basin, as vehic- ular emissions account for approximately 97% of CO; 85% of reactive hydro- carbon, and 62% of NOX emissions. Attention was also given to those sources which, while responsible for only a small proportion of emissions on an area-wide basis, contribute heavily to pollutant concentration levels in their own locale—airports, power plants, and refineries. In this Appendix, we present in detail the emissions inventories developed for the major moving and fixed sources of pollution in the Los Angeles Basin. (2) During the last months of 1970, we prepared a pollutant emissions inventory for the Los Angeles Basin for use in the modeling of the transport, diffusion, and reaction of atmospheric contamination. Pollutant sources were grouped into five categories—automobiles (and other motor vehicles), aircraft, power plants, refineries and distributed fixed sources. Emissions rates for a 2x2-mile grid system covering the Basin were compiled for nitrogen oxides, carbon monoxide, and hydrocarbons. Temporal variations in emissions rates were also determined. The complete inventory is reported in "Contaminant Emissions in the Los Angeles Basin— Their Sources, Rates, and Distribution," by P.J.W. Roberts, P. M. Roth, and C. L. Nelson (1971). Early in 1972, we had the opportunity to make a number of modifica- tions and extensions for the emissions inventory. The changes which affected all segments of the original inventory, were motivated by a variety of factors, but most heavily by a desire to improve the accuracy or the resolution of the inventory, or to correct errors. It is the purpose of this report to document all modifications and extensions that B-132 ------- were implemented. In general, this report is segmented similarly to its predecessor, the exceptions being that (1) changes applicable to all portions of the inventory are included in an introductory general section and (2) the one modification to the refinery inventory as a matter of convenience, is included in the section dealing with distributed fixed sources. Finally, we wish to point out that only changes are reported here; we have not attempted to present a final version of the inventory, either in summary or in detail, in this document. One must read both this report and the original to construct the complete inventory. B. Summary Tiie emissions model developed by SA1 considers both fixed and moving sources of air pollution in the Los Angeles Basin. Fixed sources are characterized as area sources, with the exception of point source powci plant emissions, which are treated as volurue sources. Both auto- motive and aircraft emissions are distributed over an array of grid squares. Concentrations of carbon monixide, hydrocarbons and nitrogen oxides are predicted, using hourly emissions estimates. 1. Treatment of Automotive Emissions Automotive emissions are estimated for 2.0x2.0-mile grid squares and are assumed to be attributable to exhaust, crankcase leakage (blow-by) and evaporation. The Federal Driving Cycle (FDC) is adopted. a. Exhaust Emissions Surface Streets. Average emission rates for surface streets are estimated by calculating Q.a(t) = y(t)Q.C + [l-y(t)]Q.h 111 where y(t) equals the fraction of cars started at time t that are "cold- started" (superscript c); the relation between cold-start and hot-start B-133 ------- (superscript h) emissions is: Q = Q :L h zP + (l-z)T. zP + (l-z)T i i-l where z = fraction of vehicle registration that are automobiles h c P. ,P. = hot-running and cold-start auto emission rates T. = emission rates from trucks and buses c h Both Q. and Q. are based on the FDC, which is defined to simulate a trip having an average speed of about 19.6 mph. A variation exists in the emissions rate due to the non- uniform distribution of cold vehicle starts during the day. To account for this, a correction factor, \|3.(t), is formulated. An integral representing the average emissions rate for each vehicle is derived, relating the emissions from a vehicle during c a time interval after a cold-start, e. , to those after a hot-start, h c n e . Functional forms for e an- e are establishes., giving a Q linoar Decrease in e. during die j-irst 7-1/2 minutes of operation and 1 h a constant emissions rate thereafter, equal to e. . The integral is approximated by an expression for the average emissions rate at the midpoint of a 15-minute time interval that assumes that the trip start time is uniform during the time interval. The correction factor is defined as this approximation divided by the average emissions rate. This treatment considers the effect of vehicle start variation only for the period 6 a.m. to 9:23 a.m. The "midday bump" and evening rush hour variations are not considered because the model was not validated for that time period. Also, it should be noted that large quantities of data are needed to properly evaluate \|3.(t). Pertinent data include driving patterns in an area, trip length, average speed, time between trips, emissions data as functions of time for both hot- B-134 ------- and cold-starts, etc. Since large quantities of this data are not generally available, the authors have made simplifying assumptions. The correction factor is E.S(t) P.(t) = — - Q.S(t) s where Q. (t) is the average emissions rate as defined previously, and . c n-2 E St 30 e r i c r/ \ he / \ h - n (l-y ) h . + n-1 l-^.J^ + V,, «j + jl "/^ * c n-2 n 30 where n = a fifteen-minute time interval E. [t )= average emissions rate of species i evaluated at the mid- point of time interval n c = total number of trips started in a time interval n y = ratio of cold-starts to total starts during a time interval n e (t) = emission from a vehicle during a time interval after a hot-start e. = average emissions rate of species i from an automobile over the time period between 7-1/2 and 22-1/2 minutes after a cold start 'e'. = average emission rate of species i from an automobile over the time period between zero and 7-1/2 minutes after a cold-start h c The e. and e. values are derived on the basis of the FDC, and e and e follow from these. Values of c are based on an analysis i i n of data by Kearin et al. (1971), where the weekday was segmented into D. H. Kearin, R. L. Lamoureux, B. C. Goodwin, "A Survey of Average Driving Patterns in Six Urban Areas of the United States: Summary Report", Report TM-(L)-4119 Vol. 7, System Development Corp., Santa Monica (January 1971). B-135 ------- eight time periods, each with an assigned constant value of y. The total emissions from surface streets for a particular grid square are g where d (t) = fraction of daily nonfreeway traffic count assignable to an hourly period s M = nonfreeway vehicle mileage per day for the grid square in question c; Values of d (t) ire hasecl on an average of 52 j nndf'ily bflccted L- Greets ^ (the sample being stratified according to the magnitude of daily traffic flow), the counts on individual streets being weighted in proportion to the magnitude of trafiic flow on the street. Traffic counts were takc-n from state and city data. Freeways . In calculating freeway emissions, all vehicles are assumed to be hot-running. It has been found that average emissions rates for freeways in Los Angeles correlate well with average route speed alone, although many other factors influence these rates. Freeway emissions rates for species i for a particular grid square are given by E.f(t) = — — b ot = a (v) i i i where v ,v = average speed in the fast and slow directions, respectively f s n ,n = number of vehicle m?.les driven per hour in the fast and slow directions, respectively B-136 ------- a. = hot-running emissions rate of species i a ,b = empirically determined constants i i "Fast" and "slow" refer to an assignment of names to the two opposing directions of flow on a freeway. f n = f 1+x n = s 1+x f f f N = d M a a where d = fraction of daily freeway traffice counts assignable to an hourly period f M = freeway vehicle mileage per day for a grid square x = n /n s f The average vehicle flow as a function of time for both directions on all freeways in the Los Angeles Basin is calculated from 15-minute count data for a 24-hour period at 31 locations. Values of x are computed using these data. Average freeway speed as a function of time for both freeway directions was obtained from state division of highways data. To determine the constants a and b , average emissions i i rates for hot-running conditions at a known average speed are necessary. These rates were estimated in the same manner as those for surface streets. Also required are the slopes of emissions rate/average speed curve. Data for these are very scarce, and the best such data available are somewhat out of date (less than 50% of the vehicles on the road in B-137 ------- September, 1969, are represented by the test group). Correlations based on the data give estimates of average emissions rates at high speeds under hot-running conditions that appear to be rather low. The authors decided to modify the values, basing their modifications on the premise that automobile manufacturers design their systems with the expectation that they will be tested at low average speeds. Thus, it may be expected that average emissions rates at low speeds have decreased more rapidly from year to year than rates at high average speeds. Since there is no data on which to base estimates of revised slopes b., they are estimated using two points, the hot-running emissions rates at 19.6 mph and the rates at 60 mph, the latter computed from h \|~ h h Qi (60) + 1/3 Q. (19.6) - Q (60) where Q. (19.6) are the FDC values, modified for hot-running conditions, ^ -^ and Q. (60) are computed from the data described above. The factor one-third is purely a guess and could range from 0.1 to 0.5. Total Exhaust Emissions. Total vehicle emissions for a particular grid square are given by E±(t) = E±f ------- that these losses occur mainly during the hours 7 a.m. to 7 p.m. PDT, and that they are distributed evenly over this period. Evaporative emissions in each square are assumed to be proportional to the number of nonfreeway vehicle miles in that square. / number of \ / \ / \ /evaporative \ automobiles 1 I / evaporative \ \ / \ losses / / \ \ registered / \ / I emissions for 1 - m \ / i j /total nonfreeway \ \ square i,j / I I \ vehicle mileage / where m.. equals thousands of nonfreeway vehicle miles per day driven in square ij. 2. Treatment of Aircraft Emissions Hourly emissions from airport operations are classified as arising from ground and airborne operations. For each class of aircraft it is assumed that pollutants are emitted at a uniform rate during each of five operating modes; the amount of contaminants injected into a cell is thus proportional to the length of the flight path in that cell. Air- craft emissions are treated as volume sources, well-mixed in their cells, and are assumed to occur at a uniform rate over each 1-hour time period. Emissions indices measured from one aircraft are assumed representative of all aircraft of that class. a. Ground Operations Emissions Ground operations are assigned three modes of operation: taxi, landing, and take-off. The taxi mode includes taxi between runway and satellite upon landing, between satellite and end of runway and awaiting clearance for take-off, and idle at satellite. The landing mode includes emission from touchdown on the runway to turn-off from the runway, with the take-off mode being the reverse. B-139 ------- Ground operation emissions rates into ground cell ij.£ for an hourly period are -37 3 1O da. . ^-^ ^-^ Qijm,,e,m=l = K£- nuMu ^ fgu Cgu u-i where i,j are indices for the x,y coordinates m is an index for the z coordinates (= 1 for ground cells) H is an index for the hourly period k refers to the type of pollutant (CO, HC, NO, and N02> g refers to ground operation mode (taxi, landing, and take-off) u refers to aircraft class (long-range jet transport, medium-range jet transport, business jet, turboprop transport, piston engine transport, piston engine utility, and turbine engine helicopter k Q ijm£ = emissions rate d = fraction of total daily flights in an hourly time period Ju A . = fraction of airport area belonging to a ground cell K = 60 (min/hr) n = number of flights per day of aircraft class u M = average number of engines per aircraft class u u k f = pounds of pollutant k emitted per 100 Ibs of fuel consumed by gu aircraft class u operating in mode g C = pounds of fuel consumed per engine of aircraft class u operating gu in mode g b. Flight Operations Emissions The modes of operation for airborne operations are an approach mode and a climb-out mode. It is assumed that arrival and B-140 ------- departure rates within each time period are equal; that aircraft follow straight-line flight paths while in approach and climb-out modes; and that ascent and descent angles for all aircraft at a particular airport are equal to those of the aircraft class having the highest fraction of total operations at that facility. Other assumptions are that flight paths originate and terminate at the most frequently used runway at each air- port; that a fixed proportion of aircraft of a given class arrive and depart from each airport; and that an unknown temporal distribution of flight operations at an airport may be represented by the temporal distribution from another airport having a similar mix of aircraft by class. Emissions rates for airborne flight operations are given by -3 7 k 10 d P V"N k t Q = H ijm X n M f C u V"-\ > / j j u u u up u=l u where t = time spent in descent from inversion height to touch-down by aircraft of class u t = time spent in descent from 3000 feet above ground elevation u to touchdown by aircraft of class u k f = pounds of pollutant k emitted by aircraft of class u per 1000 Ibs fuel consumed during descent C = pounds of fuel consumed per engine of aircraft of class u during descent from 3000 feet P. = fraction of the length of the flight path assignable to cell ijm The above equation also describes the emissions rate during climb-out, jj with the appropriate changes in the definitions of t , f , t' and C . u u u u B-141 ------- 3. Treatment of Fixed Source Emissions a. Power Plants Emissions from point source power plants are distributed evenly as volume sources in the column of cells up to the inversion base, under the assumption that the plume is well mixed in these cells. Since diurnal variations in NO emissions were substantial, hourly emissions X averages for the two seasons in some plants and for one day in others are used as source strengths and apportioned as described above. b. Oil Refineries Refinery emissions are treated as area sources and are assumed to be well mixed in the cell into which they are injected. Total daily emissions for nitrogen oxides and low and high reactivity organic gases and the individual refinery crude capacities were available from the Los Angeles County Pollution Control District. The total emissions are distributed uniformly over 24 hours and in proportion to the crude capacity of each refinery. c. Distributed Sources Additional sources are treated uniformly as area sources. Since some of the emissions from these sources are spread over highly populated areas, the modeling area was divided into regions of high and low population density. Petroleum Marketing, Domestic, Ship and Railroad Emissions- Nitrogen Oxides. It is assumed that half of the total daily domestic emissions and the total daily emissions attributable to petroleum marketing operations occur between 6 a.m. and 6 p.m. PST. Half of the daily ship and railroad emissions are assumed to occur at the Port of Los Angeles. The emissions cited above are distributed uniformly over the high-population grid square. All emission rates are assumed to be uniform between 6 a.m. and 6 p.m. PST. B-142 ------- Incineration—Nitrogen Oxides. The authors assume that the total daily emissions occur between 6 a.m. and 6 p.m. PST, and that emission rates are constant. In Los Angeles County, emissions assignable to a grid square are proportioned to the number of permit units issued for the 5-mile square area containing that grid square. Due to lack of information on the spatial distribution of industrial plants in Orange County, emissions are assumed to be released in the highly populated squares. Mineral Processing Plants and Metallurgical Plants— Nitrogen Oxides. These emissions are apportioned for each county in the same manner as incineration emissions. Petroleum Production and Other Industries—Nitrogen Oxides. Petroleum production emissions are also apportioned in the same manner as incineration emissions, but are distributed uniformly over 24 hours. Other industries contribute emissions which are distributed uniformly between 6 a.m. and 6 p.m. PST, mainly over the south central portion of Los Angeles County, with smaller amounts in other areas. All Fixed Sources—Organic Gases. Organic emissions from petroleum marketing, dry cleaning, degreasing, and other organic solvent users are assumed to be uniformly distributed over the highly populated squares between the hours of 6 a.m. and 6 p.m. PST. Surface painting and coating operations emissions are distributed uniformly between 6 a.m. and 6 p.m. PST in proportion to the number of paint bake oven permits in each square. Petroleum production organic emissions are spread uniformly over 24 hours and in proportion to the number of petroleum processing equipment permit units in each square. Incineration, mineral processing plant, power plant and other industrial organic emissions are not considered significant enough to be included. B-143 ------- All Fixed Sources—Carbon Monoxide. The total fixed source carbon monoxide emissions in the modeling area are considered negligible. C. Model Availability Although the emission inventory methodology has been thoroughly discussed and the appropriate equations listed, no operational computer code is included in the report. B-144 ------- XXIII SYSTEM DEVELOPMENT CORPORATION Development of a Simulation Model for Estimating Ground Level Concentrations of Photochemical Pollutants. M. Weisburd, L. G. Wayne, R. Danchick and A. Kokin, Final Report. System Development Corporation, prepared for Environmental Protection Agency, Contract No. CPA 70-151, January, 1971. A. Author's Abstract Modeling photochemical air pollution for simulation on the computer must, at the present time, be based on the limited operational and experi- mental data available. Computer simulation dictates that the gaps and uncertainties which occur in our present understanding of photochemical pollution be dealt with in order to simulate the photochemical system as a functional whole. All modeling, therefore, must involve the making of assumptions and the application of theoretical considerations. Modeling strategies largely differ in the type and amount of theory applied. The approach taken by the System Development Corporation (SDC) stays relatively close to the state-of-the-art in that (1) while the conveniences of theoretical mechanisms are necessary in some instances, the modeling is largely based on the operational and experimental data available, (2) the first iteration of the concept of the model is based on present understand- ing, and (3) the basic model is flexibly constructed so that it can be modified, expanded, and refined as new information becomes available. SDC, accordingly, has taken an approach which emphasizes those mechanisms and functions which through experience and observation, in themselves, arid in combination, appear to have the greatest potential for predictive accuracy. These include, in their order of assumed importance: (1) chemical mechanisms; (2) the trajectory (and history) of moving air parcels; and (3) the effect of vertical stability (mixing depth). B-145 ------- The basic assumption taken is that the chemical mechanism largely governs, both quantitatively and qualitatively, the ground level concen- trations of primary and secondary contaminants at particular times and lo- cations. This accuracy is dependent upon, and is maximized by the ability to follow in a detailed manner the cumulative build-up of pollution in samples of air as they move across the metropolitan area. The 33-step, 25-species mechanism, and the calculation of trajectories based on a dense network of meteorological stations, contained in the current version of the model, reflect the degree of confidence placed in this approach. The calculation of mixing depth is of similar importance as it affects the concentrations of primary pollutants. B. Summary The SDC model is an early version of the REM transformation model, the emissions treatment of which has been previously discussed. It should be noted that an earlier version of the model contained a subroutine to include pollutants from large point sources. However, initial runs of the model with this routine indicated no significant difference in air quality from that obtained using area source emissions alone, although an insuf- ficient number of trajectories were run to conclusively establish the impact of the rout ine. 1. Sensitivity to Changes in Emission Factors One sensitivity run consisted of doubling the emission factors for all contaminants. The results were an approximate 20% increase in ozone, no appreciable effect on NO , nearly an 11% increase in CO, and ^ a 25% reduction in nitric oxide concentrations. The effect of doubling emission factors must be taken as the doubling of emissions in the imme- diate vicinity of the trajectory only, as changes in emission factors do not affect the value of background concentrations. B-146 ------- Reduction of the emission factors by one-half decreased ozone concentration at the trajectory terminus by only 1 pphm, CO by approximately 5%, and both oxides of nitrogen by less than half the control values. The emission factor for nitric oxide alone was increased by 25%, with a negligible effect on concentration. 2. Sensitivity to Changes in Background Contamination The effect of reducing background contamination was explored in a series of runs using different trajectories. Results confirm that the relative effect differs "substantially" from one location to another. 3. Sensitivity to Changes in the Diurnal Traffic Curve The diurnal traffic density curve used in the model assumes a peak at 8 a.m. PDT and a constant lower value between 1 p.m. and 6 p.m. PDT. Two runs were made with different trajectories, where the assumed traffic density curve was altered. For one run the curve was shifted so that the morning peak occurred at 7 a.m. PDT, resulting in minimal changes in concentrations from those of the control run. The other case tested utilized a trajectory leading to downtown Los Angeles that was much more directly affected by the morning traffic peak. The curve was shifted by 2 hours, resulting in a 50% reduction in ozone, a one-third reduction in CO, little change in NO , and approximately a 100% increase in NO. The & effect of change in the curve is highly dependent on trajectory location and time of day. B-147 ------- XXIV TRAVELERS RESEARCH CORPORATION Sensitivities of Air Quality Prediction to Input Errors and Uncertainties. G. R. Hilst, Proc. of Symp. on Multiple-Source Urban Diffusion Models, Chapel Hill, N.C., APCO Pub. No. AP-86, 1970. A. Author's Abstract With the advent of computer-oriented simulation models of the physical and chemical system that produces varying levels of air quality, it is both possible and desirable to assess the model's sensitivities to errors and uncertainties in the input variables. From such analyses, it is possible to derive more explicitly the levels of accuracy that must be observed in the specification of the inputs for source strengths and distributions, wind velocity, horizontal and vertical diffusion rates, and pollutant chemical reactions or physical decay and loss rates, if the air quality predictions derived from these inputs are to remain within useful limits of accuracy and uncertainty. For the present report, a single case study is utilized to test the sensitivity of the Travelers Research Corporation Regional Model to random and systematic errors in the source-strength input and to systematic errors in the remaining input variables. It has been found that random errors in the source strengths do not produce comparable errors in the air quality prediction. Beyond this, and in order of decreasing sensitivity, the TRC model has been found to be highly sensitive to systematic regional wind- direction errors and moderately sensitive to systematic errors in source- strength estimates, decay or loss rates, and vertical diffusion rates. The model is insensitive to errors in lateral diffusion rates, at least for the multiple-source distribution encountered in Connecticut. These Sensitivities are quantified for the case studied. B-149 ------- The case used for the sensitivity analyses also provided an oppor- tunity to determine the effect of changes in the input variables on the verification of the model, since observations of SO were available from £t the verification program. Significant improvements in the TRC model verification, over that obtained from the original, independently chosen, input variables, were found when either the source strengths or the decay or loss rates were adjusted appropriately. This result suggests the need for much better understanding of the decay and loss of airborne SO . £ B. Summary Using the Travelers Research Center Gaussian plume model, an assess- ment was made of the model's sensitivity to errors in various input param- eters, including misestimates of source strength and positional errors. 1. Source Strength Error The model was found to be relatively insensitive to large random errors in the specification of source strength. Evidently, among multiple sources random errors tend effectively to cancel each other. Assuming a Gaussian distribution of fractional error in concentration X - X T E XT where X = true value of concentration T X = erroneous value of concentration E the range of the fractional error increases by only a factor of 2, while the RMS value of the error in source strength, Q, increases by a factor of 8. B-150 ------- While no systematic error in concentration is introduced by random errors in Q, the standard deviation of concetration error is broadened as random errors in Q increase. The model was much more sensitive to systematic errors in Q. Since it is prescribed that the concentration, X, is directly proportional to Q, there must be a one-to-one correspondence between systematic errors in Q and errors in X. Thus, =[(QT-QE)/QT1 The standard deviation for systematic error is zero. 3. Positional Error Mislocation of sources, receptors, etc., automatically invokes errors in all of the input parameters. Therefore, a systematic error in the direction of the general wind field was utilized for the assessment of sensitivity to the compounded errors of position. Noting that the refer- ence values for all input parameters were arbitrarily chosen, results show that mean error is less than or equal to 50% and individual error is greater than 200%. The mean fractional concentration errors are not extreme, but the spread of these errors is. B-151 ------- XXV TRW Prediction of The Effects of Transportation Controls on Air Quality in Major Metropolitan Areas. Prepared by TRW, Inc., McLean, Virginia, APTD-1363, EPA, Office of Air Quality Planning and Standards, Research Triangle Park, North Carolina, 27711. A. Author's Abstract Transportation data in the form of vehicle miles traveled (VMT) by geographic areas have been used as the basis for calculating emission rates and air quality. The methodology involved: Step 1. Assignment of VMT and speed to elements of a grid network. This required superimposing a rectangular grid network consisting of one mile squares or one kilometer squares over a base map of the metropolitan area and summation of VMT from each of the indi- vidual roadways which may fall within one of the small grids to obtain the total VMT for each element or small grid in the grid network. The speed for each grid of the grid network is obtained by averaging the speeds from each element or roadway within the grid for each element of roadway as a proportioning factor. In this manner the many vehicular sources moving within an individual grid can be represented by a single^stationary source, which produces the same amount of emissions, equal to the size of the individual grid. This equivalent source is called an area source because the emis- sions from the grid or area source are now considered as evenly distributed or evenly produced over the entire area of the individual grid. Step 2. Use of vehicle emission factors to calculate emissions on a per-grid basis. B-153 ------- Step 3. Conversion of emission rates to pollutant concentration on a per-grid basis by the method of Gifford and Hanna. Step 4. Application of transportation control strategies to the data base to obtain predicted concentration patterns for each con- trol strategy for the year 1977. The resulting concentrations are presented graphically as isopleths over a map of the metropolitan area. Pollutants considered were carbon monoxide, hydrocarbons, and oxides of nitrogen, with carbon monoxide being of primary interest. B. Summary Transportation data in the form of vehicle miles traveled (VMT) by geographic areas have been used as the basis for calculating emission rates and air quality. The methodology involved: Step 1. Assignment of VMT and speed to elements of a grid network. This required superimposing a rectangular grid network consisting of one mile squares or one kilometer squares over a base map of the metropolitan area and summation of VMT from each of the individual roadways which may fall within one of the small grids to obtain the total VMT for each element or small grid in the grid network. The speed for each grid of the grid network is obtained by averaging the speeds from each element or roadway within the grid for each time period required, using the VMT along each element of roadway S. R. Hanna and F. A. Gifford, Jr., "Urban Air Pollution Modelling,' Presented at 1970 International Air Pollution Conference of the International Union of Air Pollution Prevention Associations, ATDL Contribution No. 37. B-154 ------- as a proportioning factor. In this manner the many vehicular sources moving within an individual grid can be represented by a single stationary source, which produces the same amount of emis- sions equal to the size of the individual grid. This equivalent source is called an area source because the emissions from the grid or area source are now considered as evenly distributed or evenly produced over the entire area of the individual grid. Step 2. Use of vehicle emission factors to calculate emissions on a per-grid basis. 1. Specific Approach From urban transportation demand models (the traffic assignment portions), VMT and speed data were obtained. Wherever possible, VMT and speed data were obtained by zone. If no zonal information was available, zonal estimates were made. However, the emission levels are actually used as average zonal values and are rough approximations rather than point measures. In addition, the factors used to convert VMT and speed data into emission levels themselves introduced errors which are of at least equal magnitude to those introduced by allocation of link VMT into zones. Most of the VMT and speed information supplied was for a 24-hour average weekday (in some cases peak-hour data were available) and, there- fore, had to be adjusted to estimate the eight-hour and one-hour maximum requirements of the project. These estimates were made utilizing daily hourly traffic counts made available by the traffic departments and trans- portation planning agencies in each of the cities. Most of the traffic assignment data provided had base years in 1968 and 1969. These same models generally provided projected values for VMT (and very often speeds) for the year 1980 and beyond. Interpolations, therefore, had to be made for the year 1977. In making these interpolations B-155 ------- employment, residential, and population data were used whenever available. When these data or growth models were not available, interpolations were generally based on straight line assumptions. In a number of cities speed information had to be estimated for each of the zones. When this was done, the speeds were based upon link- node data using weighted averages weighting the respective speeds using link volumes as weights. The program input and output are oriented to a rectangular grid network which overlays the city of interest. The results are presented in tabular form as the average concentrations for each element of the grid network. Punched card output was used to produce concentration isopleths or concentration density maps which were overlaid on a map of the metropolitan area to give a visual representation of the spatial distribution of the pollutant of interest. 2. Program Input Three groups of input data are required: traffic data, meteoro- logical data, and emission factors for each pollutant. Traffic Flow Data. The resulting grid network of VMT and speed from above provide the basic traffic flow data required for emission calculations. To correspond with the time periods specified by the National Ambient Air Quality Standards, peak 1-hour and peak 8-hour VMT are required, For comparison with the ambient air quality standards for hydrocarbons and nitrogen dioxide, a Larsen transformation was used to convert estimates to 3-hour and annual averages, respectively. When these data were not available, they were estimated as percentages of the 24-hour VMT by using traffic count information at various stations within the area. The pro- gram applied this percentage whenever necessary to obtain the appropriate data format. Particular streets may vary significantly from the mean B-156 ------- values obtained by this procedure; however, the allocation of VMT to grids as described above tends to minimize the effect of individual streets on the predicted concentrations values for the individual grid. Traffic speed data on a per-grid basis are required as an adjust- ment to the average vehicle emission factors. Carbon monoxide and hydro- carbon emissions decrease with increasing average vehicle speed; therefore, the emission factors must be adjusted to be more realistic. EPA has pub- lished emission factors curves as a function of average vehicle speed for carbon monoxide and hydrocarbons. These curves are incorporated in the program. The modal split of VMT must be considered, i.e., what is the proportion of cars, buses, and trucks comprising the total VMT. Data on the modal split are not typically available on a detailed basis; therefore, this factor is most readily accounted for in the formulation of area emission factors as discussed below. Emission Factors. EPA has developed a computer program to pro- duce emission factors for the entire vehicle population for a given metropolitan area for any calendar year from 1960 to 1990 for which vehicle age-type distribution data are available. Emission factors are subject to revision as required by changes in vehicles, control devices, "representative" driving cycles, and measurement techniques. 3. Outputs 1. Annual emissions (tons/year) for each grid and total emissions for a grid network. 2 2. Emission rates (mg/m /sec) for each grid. 3. Total 8-hour and 1-hour emissions for each grid. B-157 ------- 4. Resolutions Generally, the data available for each city did not permit classification into fine-grained geographic areas (e.g., the eight rings used for Washington, D.C.). Since the impact of flow controls was based on general experience and Judgment, for the purposes of project analysis, the respective cities (except Washington, B.C.) were divided into three broad geographic areas: (1) the CBD or core, (2) an area surrounding the core but still within the political boundaries of the city, and (3) the remaining area outside the boundaries of the city. This three-way division allowed relatively easy division of zones while still permitting our analysis to be somewhat more fine grained than would otherwise be possible. B-158 ------- EMISSION MODELS CHECKLIST Input Data Requirements Within User Program Supplies From AP-42 Vehicle Emission Factors -By model year -By size class -By power plant -Deterioration factor Vehicle Operating Modes -Hot/cold start -Route average -Link average -Mode-by-mode Traffic Features -Vehicle mix -Engine operating time -Vehicle miles traveled -Fuel consumed x x X X X X X X NP X NP NP NP NP x NP Proxies/Comments Uses vehicle mix for year of interest in study Speed and distance NP NP NP NP Spatial Distribution -Links -Grinds -Area NP x X NP NP NP NP = No provision B-159 ------- Output Available Pollutants -CO Yes -HC Yes -N°x Yes -Particulates No Availability of Program Documentation Yes No Included in report x * Referenced in report x Language NP Equipment NP Validation/calibration; Compared with Washington, D.C., camp data; CO CO data show best agreement Time resolution; 1 hour; 8 hour Spatial resolution; 1 mile or 1 kilometer square * Uses EPA computer program for emission factors for entire vehicle population for a given metropolitan area for years 1960 to 1990. B-160 ------- XXVI UNIVERSITY OF ALASKA Ice Fog: Low Temperature Air Pollution. Prepared by Carl S. Benson, Geophysical Institute, University of Alaska, UAG R- 173, November, 1965. A. Author's Abstract Stable pressure systems over interior Alaska sometimes produce prolonged extreme (below -40°C) cold spells at the surface. The rate of radiative cooling of the air is enhanced by suspended ice crystals, which are themselves a result of the initial cooling. Radiation fogs formed during the onset of cold spells are generally of short duration because the air soon becomes desiccated. These fogs consist of supercooled water droplets until the air temperature goes below the "spontaneous freezing point" for water droplets (about -40°C); the fog then becomes an ice crystal fog, or simply "ice fog." During the cooling cycle, water is gradually condensed out of the air until the droplets freeze. At this point there is a sharp, discontinuous decrease in the saturation vapor pressure of the air because it must be reckoned over ice rather than over water. The polluted air over Fairbanks allows droplets to begin freezing at the relatively high temperature of -35°C. Between -35 and -40° C the amount of water vapor condensed by freezing of supercooled water droplets is three to five times greater than the amount condensed by 1°C of cooling at these temperatures. This results in rapid and widespread formation of ice fog, which persists in the Fairbanks area as long as the cold spell lasts. The persistence of Fairbanks ice fog depends on a continual source of mois' from human activities within the fog. depends on a continual source of moisture (4.1 x 10 Kg H^O per day) B-161 ------- Ice fog crystals are an order of magnitude smaller than diamond dust, or cirrus cloud crystals, which in turn are an order of magnitude smaller than common snow crystals (0.01, 0.1 and 1 to 5 mm respectively). The differences in size are shown to result from differences in cooling rates over five orders of magnitude. Most of the ice fog crystals have settling rates which are slower than the upward velocity of air over the city center. The upward air movement is caused by convection cells driven by the 6°C heat island over Fairbanks. This causes a reduced precipitation rate, which permits the density of ice fog in the city center to be three times greater than that in the outlying areas. The inversions that occur over Fairbanks during cold spells begin at ground level and are among the strongest and most persistent in the world. They are three times stronger than those in the inversion layer over Los Angeles. Thus, the low-lying air over Fairbanks stagnates and becomes effectively decoupled from the atmosphere above, permitting high concentrations of all pollutants. The combustion of fuel oil, gasoline, and coal provides daily inputs of 4.1 x 106 kg C02; 8.6 x 103 kg S02; and 60, 46 and 20 kg of Pb, Br and Cl respectively, into a lens-like layer of air resting on the surface with a total volume less than 3 x 10 m . The air pollution over Fairbanks during cold spells couldn't be worse, because the mechanisms for cleaning the air are virtually eliminated while all activities that pollute the air are increased. B. Summary Aqueous emissions are an important pollutant in Fairbanks, Alaska, in cold weather, when the capability of the air to hold water in solution is reduced. The Geophysical Institute, University of Alaska, estimates aqueous and C02 emissions from combustion, power plant cooling ponds, and miscellaneous sources, such as sewage treatment plants, mine tunnels, people and animals breathing. B-162 ------- 1. Combustion Emissions Complete combustion of the representative molecular species is assumed, with average molecular weight ratios of 1.38 (H O/fuel) and 3.10 (C02/fuel) for gasoline and 1.33 (H20/fuel) and 3.13 (CO2/fuel) for fuel oil. For coal, the calculation of these ratios is more complex because the water contained in coal is exhausted directly to the atmosphere during combustion. It is assumed that 20 percent by weight of the coal is water. Total molecular weight ratios for coal are 0.68 (HgO/fuel) and 2.37 (C02/fuel). The above ratios are applied to the quantities of each fuel type burned to give monthly and daily emissions of H O and CO . ^i Zj 2. Power Plant Cooling Emissions Given the water flow rate to and from the cooling pond and the net temperature change of water, the required heat loss may be calculated. This heat loss is accomplished by three processes: (1) evaporation to the atmosphere; (2) radiation to the sky; and (3) conduction through the bottom of the pond. Assuming an extreme winter vertical temperature gradient in the pond, no convection in the water, and the condition of all heat con- ducted downward being conducted away through the soil below, the calculated vertical heat flux from conduction for a large plant at Fort Wainwright, Alaska, is less than 0.1 percent of the total required flux. Therefore, conduction can be neglected. Radiative cooling is maximized using Q = 0.105 + 0.0018 T B-163 ------- and multiplying Q by the area of the pond. A temperature T over the open water surface must be assumed. The value of maximum radiative heat flux at the plant cited above is slightly over 5 percent of the total flux and is negligible when ice fog forms over the pond. Therefore, during ice fog conditions, the evaporative heat loss approaches the value of the total heat loss. To estimate the amount of water introduced into the atmosphere from power plant cooling waters, each plant must be analyzed separately in a manner similar to that described for the Fort Wainwright plant. 3. Miscellaneous Emissions Aqueous emissions from miscellaneous sources such as sewage treatment plants are treated individually by calculating heat loss from the warm water discharge flow rate, the amount of open water area, the amount of heat dissipated by mixing with river water, and so forth. For tunnels, the moisture flow rate through the fans (and therefore the emission rate from tunnels) are estimated using the air flow rate past the fans and the relative humidity and temperature of the air. Another important source of aqueous emissions in very cold climates is the breathing of people and animals. Saturated air is exhaled at 35°C. This exhalation process together with respiration rates and the populations of each sex and species gives the amount of water vapor released to the environment. It has been shown that the loss of water vapor through perspiration exceeds the loss from respiration; therefore, the moisture output from breathing can be doubled to give a combined perspiration and respiration loss. Other miscellaneous sources include moist air discharge from laundries and air leaks from buildings and houses. B-164 ------- XXVII UNIVERSITY OF CALIFORNIA AT DAVIS The Impact of Highways on Air Quality. L. O. Myrup, unpublished report, August, 1972. A. Author's Abstract None given. B. Summary Carbon monoxide emissions for Sacramento, California, input to the pollution transport and diffusion model developed at the University of 2 California at Davis are distributed into 1.7 mi areas. Total emissions for each area are formulated by first measuring the total miles of resi- dential streets, arteries, and highways in each area. The average traffic volumes for each of the three types of streets are obtained, and car-mile factors computed. Vehicle emission factors at appropriate vehicle speeds are then applied, yielding the total amount of pollutant for each kind of street, and after summation, the total emissions for the grid square in question. C. Model Availability No computer program for the emission inventory methodology presented is available. B-165 ------- XXVIII U.S. WEATHER BUREAU A Simple Diffusion Model for Calculating Point Concentrations from Multiple Sources. J. F. Clarke, Journal of the Air Pollution Control Association, September, 1964. A. Author's Abstract None given. B. Summary The application of a method designed by Hagan, et al. (1964) resulted in the source inventory for the city of Cincinnati used in Clarke's calcu- lations. Two source categories for sulfur dioxide and nitrogen oxides are presumed: (1) combustion sources, which include space and process heating, power generation, transportation, and waste incineration sources; and (2) industrial process losses, which include only the losses from processes, not from combustion. Emissions are allocated on the basis of unit-area emission rates for each of 19 areas suggested by population density maps and modified subjectively to more clearly define commercial and industrial areas. Among the assumptions made in apportioning emissions are that the amount of fuel used for space heating is directly proportional to the degree day value, and that this fuel is used only during days registering a degree day value. It is assumed that industrial and commercial sources that use fuel for power and waste incineration and process heat operate at a uniform daily rate. Fuel used for transportation is taken as con- stant for days without a degree day value and inversely proportional J. E. Hagan, III and G. A. Jutze, "A Rapid Technique for an Air Pollution Emissions Survey," Public Health Service Manuscript (1964). B-167 ------- to that value when one is registered. This assumption is based on seasonal variations of gasoline sales. The industrial process losses used are engineering estimates derived from knowledge of the characteristics of the individual processes, as sufficient quantitative data was not available. One power plant was thought significant enough to be included as a point source. The plant does not operate at a uniform daily rate, and fuel usage is not a simple function of season or time of day, since the output of the plant is used to both heat and cool, according to season. Since operational data from the plant were not readily available and the precision of the general inventory did not seem to warrant the effort required to calculate otherwise, emissions from this plant were taken as invariant in time. C. Model Availability A computer code for the emissions treatment used for Cincinnati is not referenced in the paper. B-168 ------- TECHNICAL REPORT DATA (Please read Instructions on the reverse before completing) 1. REPORT NO. EPA-450/3-74-030 2. 3. RECIPIENT'S ACCESSION-NO. 4. TITLE AND SUBTITLE 5. REPORT DATE A Regional Air Pollution Study Preliminary Emission Inventory January 1974 6. PERFORMING ORGANIZATION CODE 7. AUTHOR(S) Fred E. Littman Svlvan Rubin 8. PERFORMING ORGANIZATION REPORT NO, Konrad T. Semrau Halter F. Dabberdt 9. PERFORMING ORGANIZATION NAME AND ADDRESS Stanford Research Institute Menlo Park, California 94025 10. PROGRAM ELEMENT NO. 11. CONTRACT/GRANT NO. 68-02-1026 12. SPONSORING AGENCY NAME AND ADDRESS Environmental Protection Agency Research Triangle Park, North Carolina 13. TYPE OF REPORT AND PERIOD COVERED Final Report 27711 14. SPONSORING AGENCY CODE 15. SUPPLEMENTARY NOTES 16. ABSTRACT This report presents an operational plan for providing emissions data for the Saint Louis Regional Air Pollution Study (RAPS). It also describes existing emission inventories for the Saint Louis area and reviews in detail emission models that have in the past been used to provide emissions data. KEY WORDS AND DOCUMENT ANALYSIS DESCRIPTORS Regional Air Pollution Study (RAPS) Inventory Emissions UTM Coordinates Pollutants Fortran Emission Modeling Area Sources Point Sources b.IDENTIFIERS/OPEN ENDED TERMS c. COSATI Field/Group 8. DISTRIBUTION STATEMENT Release Unlimited 19. SECURITY CLASS (ThisReport) Unclassified 21. NO. OF PAGES 187 20. SECURITY CLASS (This page) Unclassifipd 22. PRICE EPA Form 2220-1 (9-73) B-169 ------- INSTRUCTIONS 1. REPORT NUMBER Insert the EPA report number as it appears on the cover of the publication. 2. LEAVE BLANK 3. RECIPIENTS ACCESSION NUMBER Reserved for use by each report recipient. 4. TITLE AND SUBTITLE Title should indicate clearly and briefly the subject coverage of the report, and be displayed prominently. Set subtitle, if used, in smaller type or otherwise subordinate it to main title. When a report is prepared in more than one volume, repeat the primary title, add volume number and include subtitle for the specific title. 5. REPORT DATE Each report shall carry a date indicating at least month and year. Indicate the basis on which it was selected (e.g., date of issue, date of approve!, date of preparation, etc.). 6. PERFORMING ORGANIZATION CODE Leave blank. 7. AUTHOR(S) Give name(s) in conventional order (John R. Doe, J. Robert Doe, etc.). List author's affiliation if it differs from the performing organi- zation. 8. PERFORMING ORGANIZATION REPORT NUMBER Insert if performing organization wishes to assign this number. 9. PERFORMING ORGANIZATION NAME AND ADDRESS Give name, street, city, state, and ZIP code. List no more than two levels of an organizational hirearchy. 10. PROGRAM ELEMENT NUMBER Use the program element number under which the report was prepared. Subordinate numbers may be included in parentheses. 11. CONTRACT/GRANT NUMBER Insert contract or grant number under which report was prepared. 12. SPONSORING AGENCY NAME AND ADDRESS Include ZIP code. 13. TYPE OF REPORT AND PERIOD COVERED Indicate interim final, etc., and if applicable, dates covered. 14. SPONSORING AGENCY CODE Leave blank. 15. SUPPLEMENTARY NOTES Enter information not included elsewhere but useful, such as: Prepared in cooperation with, Translation of, Presented at conference of, To be published in, Supersedes, Supplements, etc. 16. ABSTRACT Include a brief (200 words or less) factual summary of the most significant information contained in the report. If the report contains a significant bibliography or literature survey, mention it here. 17. KEY WORDS AND DOCUMENT ANALYSIS (a) DESCRIPTORS - Select from the Thesaurus of Engineering and Scientific Terms the proper authorized terms that identify the major concept of the research and are sufficiently specific and precise to be used as index entries for cataloging. (b) IDENTIFIERS AND OPEN-ENDED TERMS - Use identifiers for project names, code names, equipment designators, etc. Use open- ended terms written in descriptor form for those subjects for which no descriptor exists. (c) COSATI FIELD GROUP - Field and group assignments are to be taken from the 1965 COSATI Subject Category List. Since the ma- jority of documents are multidisciplinary in nature, the Primary Field/Group assignment(s) will be specific discipline, area of human endeavor, or type of physical object. The application(s) will be cross-referenced with secondary Field/Group assignments that will follow the primary posting(s). 18. DISTRIBUTION STATEMENT Denote releasability to the public or limitation for reasons other than security for example "Release Unlimited." Cite any availability to the public, with address and price. 19. &20. SECURITY CLASSIFICATION DO NOT submit classified reports to the National Technical Information service. 21. NUMBER OF PAGES Insert the total number of pages, including this one and unnumbered pages, but exclude distribution list, if any. 22. PRICE Insert the price set by the National Technical Information Service or the Government Printing Office, if known. EPA Form 2220-1 (9-73) (Reverse) ------- NO 30 — (0 -o^£ 5? o 3 o 3- 3 a. n> ~ fD T 3- CD DJ CD Q. — Q. — CD ~~ CO •< co O co ci. a ° 2 2T CD (-* Q. <-> D a. S}5 (Q q O » 3D 2-0 1° 2 i/> m -( Z *> O O m -4 VI O m O ~~-------~~


EPA450/3-74-030
January 1974
 A REGIONAL AIR POLLUTION
 STUDY (RAPS) PRELIMINARY
          EMISSION INVENTORY
    U.S. ENVIRONMENTAL PROTECTION AGENCY
        Office of Air and Water Programs
    Office of Air Quality Planning and Standards
    Research Triangle Park, North Carolina 27711

-------
                                  EPA-450/3-74-030
A REGIONAL AIR  POLLUTION
STUDY  (RAPS)  PRELIMINARY
     EMISSION INVENTORY
                    by

          Fred E. Littman, Sylvan Rubin,
        Konrad T . Semrau, Walter F . Dabberdt

            Stanford Research Institute
           Menlo Park, California 94025
             Contract No. 68-02-1026


         EPA Project Officer:  Charles Masser


                 Prepared for

       ENVIRONMENTAL PROTECTION AGENCY
          Office of Air and Water Programs
      Office of Air Quality Planning and Standards
         Research Triangle Park, N. C. 27711

                 January  1974

-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers.  Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - as supplies permit - from the
Air Pollution Technical Information Center, Environmental Protection
Agency, Research Triangle Park, North Carolina 27711, or from the
National Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22151.
This report was furnished to the Environmental Protection Agency by
Stanford Research Institute, Menlo Park, California 94025,  in fulfillment
of Contract No. 68-02-1026. The contents of this report are reproduced
herein as received from Stanford Research Institute. The opinions,
findings, and conclusions expressed are those of the author and not
necessarily those of the Environmental Protection Agency.  Mention of
company or product names is not to be considered as an endorsement
by the Environmental Protection Agency.
                   Publication No. EPA-450/3-74-030
                                     11

-------
                                CONTENTS


LIST OF ILLUSTRATIONS	     vii

LIST OF TABLES	      ix

ACKNOWLEDGMENTS	      xi

  I  INTRODUCTION 	       1

     A.   Saint Louis Regional Air Pollution Study (RAPS)  ....       1

     B.   The Emission Inventory  	       2

     C.   Organization of the Report	       3

 II  SUMMARY	       5

     A.   Basic Concepts  	       5

     B.   Results of Specific Tasks 	       8
          1.   Task A	       8
          2.   Task B	       9
          3.   Task C	      12
          4.   Task D	      13
          5.   Task E	      14

     C.   Recommendations	      18

III  TASK A:  DEFINITION OF POTENTIAL USERS AND USES	      21
     A.   General	      21

     B.   Summary	      21

     C.   RAPS Requirements	      22

          1.   Atmospheric Structure Models 	      24
          2.   Transformation Processes  Models  	      25
          3.   Mobile Sources	      27
          4.   Other Studies	      29

     D.   State and Local Control Agencies  	      30

          1.   City of Saint Louis	      30
          2.   Saint Louis County 	      32

-------
Ill  TASK A:   DEFINITION OF  POTENTIAL  USERS AND USES  (Continued)
          3.    Missouri  Air Conservation Commission  	
          4.    Illinois  Environmental  Protection Agency  .  .  .  .
          5.    Federal Highway Department DOT  	

     E.    The Planning Agencies  	
          1.    East-West Gateway Coordinating  Council  	
          2.    Industrial-Waste Control Council  	
     F.    Research Programs 	
          1.    Metromex  	
          2.    Community Health and  Environmental Surveillance
               System (CHESS)  	
          3.    University Research   	
          4.    Plant  Damage 	

     G.    Conclusions 	

 IV  TASK B:   EMISSION INVENTORY CONTENT   	  ,
     A.    General Principles  	
          1.    Introduction 	  ,
          2.    The Emission Inventory  System   	
     B.    Precision of Emission Estimates  	
     C.    Inventory Resolution   	
          1.    Temporal  Resolution   	
          2.    Spatial Resolution  	
     D.    Pollutants	,
          1.    General Discussion  	  ,
          2.    Nonreactive Gases   	  .
          3.    Reactive  Gases  	  .
          4.    Particulate Matter  	  ,
          5.    Other  Pollutants  	  ,
          6.    Heat and  Water Vapor  Releases   	
     E.    Source Categories 	  .
          1.    Classification Scheme  	
          2.    Stationary Sources  	  .
          3-    Mobile Sources  	
     F.    Emission or Data Conversion  Factors  	

     G.    Physical Aspects of Emission Sources  	
32
33
34

35

35
35

36

36

36
37
38

39

41

41

41
43

46

48

48
48

50

50
50
51
53
55
56

56

56
58
66

67

68
                                   iv

-------
IV  TASK B:   EMISSION INVENTORY  CONTENT  (Continued)

    H.    Units of  Measurement	     68

    1.    Specification of  the  RAPS  Emission  Inventory	     69

         1.    General	     69
         2.    Point Sources	     70
         3.    Point Combustion Sources  .  	     75
         4.    Point Noncombustion Sources   	     77

    J.    Stationary Area Sources	     77

         1.    General	     77
         2.    Area Combustion  Sources	     77
         3.    Area Noncombustion Sources	     79

    K.    Mobile Source Procedures   	     79

    L.    Natural Background  Emissions   	     80

    M.    Particulates	     80

    N.    Scheduling and Scale  of Effort	     81

 V  TASK C:   EMISSION INVENTORY  FILE SYSTEM	     83

    A.    General	     83

    B.    System Environment	     84

    C.    Inventory Operation 	     87

    D.    Data Structure and  Formats	     93

         1.    Source Categories   	     93
         2.    Data Item Formats	     94
         3.    Other Data	     97
         4.    Data Element Formats	     97
         5.    Summary Table	     100
    E.    Format Extensions for Batch Data Entry   	     102

    F.    Implementation	     104

    G.    Specifications for  Computer Programs   	     105

         1.    Data Management  System	     105
         2.    Input and Output Subroutines for Emission Model
              Programs .	     106
         3.    Verification Program  	     106

-------
 VI  TASK D:  SURVEY AND EVALUATION OF EXISTING EMISSION
     INVENTORY DATA	    107

     A.   Introduction	    107

     B.   Existing Emission Inventories for the Metropolitan
          Saint Louis Interstate Air Quality Control Region .  .  .    108

          1.   IPP Inventory—1968	    108
          2.   IBM Inventory—1970	    109
          3.   DAQED Emission Inventory—1971 	    109
          4.   NATO Emission Inventory—1971	    109
          5.   NEDS Inventory—1973	    113

     C.   Traffic and Transportation Inventories  	    127

          1.   Streets and Highways	    127
          2.   Railways and Vessels	    129
     D.   Summary	    133

VII  TASK E:  EMISSIONS MODELING	    135
     A.   Introduction	    135

     B.   Modeling	    137

          1.   Review of Existing Models  	    137
          2.   Specification of Emissions 	    146
          3.   Emissions Model Verification--Mobile Sources .  .   .    153
          4.   Resolution of Source Location  	    158

     C.   Consideration of Natural Emissions  	    170

     D.   Summary of Conclusions and Recommendations  . 	    172

REFERENCES	    175

APPENDICES
  A  SAMPLE EMISSION INVENTORY PRINTOUTS  	    A-l

  B  SUMMARY OF EMISSION MODEL REPORTS  	    B-l
                                   vi

-------
                            ILLUSTRATIONS
 1   Metropolitan Saint Louis Interstate  Air  Quality Control
     Region	     31
 2   Emission Inventory System .  	     44

 3   Schematic of Inventory System 	     92

 4   Sample Point Source Item  	     95
 5   Sample Area Source Item	     96
 6   Emission Inventory System Concept  	     115

 7   Number of Point Sources and  Emissions  	     120
 8   Particulate Emissions for Saint  Louis  Air Quality Control
     Region	     122
 9   SOg Emissions for the Saint  Louis  Air  Quality  Control
     Region	     123
10   NOX Emissions for the Saint  Louis  Air  Quality  Control
     Region	     124
11   Hydrocarbon Emissions for the Saint  Louis Air  Quality
     Control Region  	     125
12   CO Emissions for the Saint Louis Air Quality Control
     Region	     126
13   1969 Traffic Map of Saint Louis  Metropolitan Area
     Interstate and Freeway System 	     130
14   Schematic Illustration of Classification Matrix for
     Emission Models 	     136
15   Performance Curves for Cooling Towers  Giving the Exit
     Temperature as a Function of  the Ambient Wet Bulb
     Temperatures for Various Exit Relative Humidities 	     154
16   Performance Curves for Cooling Towers  Giving the
     Moisture Discharge as a Function of  the  Ambient Wet Bulb
     Temperature for Various Exit  Relative  Humidites 	     155
17   Variation of Integrated Normalized Error with  Longitudinal
     Distance from the Source When Source Height is at the
     Surface and Atmospheric Conditions Are Stable  	     162

                                  vii

-------
18   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is
     Five Meters and Atmospheric Conditions Are Stable 	     163
19   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is 30 Meters
     and Atmospheric Conditions Are Stable 	     164
20   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is 100 Meters
     and Atmospheric Conditions Are Stable 	     165
21   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is at the
     Surface and Atmospheric Conditions Are Slightly
     Unstable	     166
22   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is Five Meters
     and Atmospheric Conditions Are Slightly Unstable  	     167
23   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is 30 Meters
     and Atmospheric Conditions Are Slightly Unstable  	     168
24   Variation of Integrated Normalized Error with Longitudinal
     Distance from the Source When Source Height Is 100 Meters
     and Atmospheric Conditions Are Slightly Unstable  	     169
                                 viii

-------
                                TABLES
 1   Emission Inventory Requirements 	    23

 2   Emission Inventory Input Requirements for
     Atmospheric Structure (WATPIC)  Model  	    25

 3   Emission Inventory Input Requirements for the
     Transformation Processes Model   	    26

 4   Emission Inventory Input Requirements for
     Vehicular Sources Model 	    27

 5   Emission Inventory Input Requirements for
     Mobile Sources (Airports)  Model 	    28

 6   Emission Inventory Input Requirements of the
     Air Pollution Control Office, City of Saint Louis 	    32

 7   Emission Inventory Input Requirements of the
     Air Pollution Control Division, Saint Louis County
     Health Department 	    33

 8   Emission Inventory Input Requirements of the
     Illinois Environmental Protection Agency  	    34

 9   Emission Inventory Input Requirements for CHESS Studies  ...    37

10   Classification of Sources  for Emission Inventory  	    57

11   Major Noncombustion Sources of  Pollutants 	    64

12   Pollutant Sources by Class	    72
13   Tentative Workload for the Monitoring of
     SO  and NO  Sources	    73
       ?       x
14   Development Schedule  	    82
15   Sample of DAQED Inventory  SO Emission Summary  	   110
16   Sample of DAQED Inventory  Current and Trend Data,
     Ambient Concentrations  ...  	   Ill
17   Comparison of Emission Inventories for the
     Metropolitan Saint Louis Interstate Air Quality Region   .  .  .   118

18   Number of Point Sources	   119
                                   IX

-------
19   Number of Sources Monitored 	   121
20   Breakdown of Area Sources	  .   128

21   Classification of Possible Emission Inventories 	   131

22   1967 Emissions from Rail Operations Within
     100 Miles of Saint Louis	132
23   Summary of Content of Emission Models Reviewed  	   138

24   Emissions Models Input/Output Specifications  	   139

25   Ratio of Cold-Start Emissions to Hot-Start Emissions  ....   150
26   Cold-Start Emissions from Light-Duty Vehicles 	   151

27   Natural Pollutant Emissions in the Saint Louis Area 	   172

-------
                            ACKNOWLEDGMENTS
     This study was carried out by a project team led by Dr. F. Littman
of the Operations Evaluation Department of the Engineering Systems Divi-
sion at Stanford Research Institute.  The project supervisor was Mr. R.
Rodden, Assistant Director of that department.   R.T.H. Collis, Director
of the Atmospheric Sciences Laboratory of the Electronics and Radio Sci-
ences Division, acted as project advisor and assisted in the preparation
of the final report.

     rlhe principal authors were as follows:
     Section III (Task A)
     Section IV (Task B)
     Section V (Task C)
     Section VI (Task D)
     Section VII (Task E)
Dr. F. E. Littman

K. T.  Semrau, Chemical Engineer,
Physical Science Division

Dr. S. Rubin, Information Science
Laboratory, Information Science
and Engineering Division

Dr. F. E, Littman

Dr. W. F. Dabberdt, Atmospheric
Sciences Laboratory
     Other contributors were as follows:

          P.  A.  Buder
          R.  C.  Robbins
          P.  J.  Mai-tin
          •\.  E.  Moon
          G.  1.  Thompson
                                   XI

-------
                              INTRODUCTION
     This report presents an operational plan for providing emissions
data for the Saint Louis Regional Air Pollution Study  (RAPS).  It also
describes existing emission inventories Tor the Saint Louis area and
reviews in detail emission models that have in the past been used to
provide emissions data.
A.   Saint Louis Regional Air Pollution Study (RAPS)

     The Saint Louis Regional Air Pollution Study is the most ambitious
study of its type ever attempted.  Focused principally upon the verifi-
cation and development of air quality models, it  is also designed  to de-
velop a better understanding of atmospheric transformation processes and
to provide a basis for studying various air pollution control strategies.
Material advancements in the technology and methodology of air quality
monitoring and other aspects of air pollution control, including partic-
ularly the improvement of emissions inventory procedures, are expected
also.

     Considerable attention has already been paid to measuring meteoro-
logical conditions and air quality in the test area, and a network of
25 automated data collection stations will be installed, stations which
will continuously telemeter their readings to the data center.  Data of
this quality will be valuable to RAPS, particularly for diffusion model-
ing on an extended scale.

     For ihe lull reali^atJon ol such purposes,  however, especially lor
the modeling and prediction of air quality,  it is essential to know the
emissions from the air-polluting sources.   Without effective assessment
ot sunh emissions to an appropriate degree of resolution in time and
space,  the experiments and studies of the RAPS program will be severely
handicapped no matter how well other measurements are made or other fac-
tors assessed.   To collect data on critical  emissions for the multitude
of different sources in an area the size of Saint Louis is a very tall
order.   It can only be met by an effort of considerable magnitude and
cost for exceeding any previous undertakings of a similar nature.   Indeed,
it must be recognized that the requirement demands and warrants the appli-
cation of detailed data collection procedures considerably different from

-------
those appropriate to a less rigorous determination of high resolution
data and that many of the estimation or modeling techniques used hitherto
cannot be used.
B.   The Emission Inventory

     To determine how to meet RAPS' needs for emission data, the problem
was considered under five tasks.  These logically follow each other se-
quentially, and they were addressed, at least initially, in that order.
As the study proceeded,  however, the interrelationship of the problems
addressed under each task became ever more apparent, and, although the
separate submission of the task reports in draft form served the useful
purpose of checking and correcting progress, it was agreed that the re-
sults of the project were most useful when integrated as they are in this
final report.

     The tasks and their objectives were

     •  Task A.  To ascertain the uses and users of emission data in
                 the RAPS program (and elsewhere).

     •  Task B.  To specify the content and nature of the emission
                 inventory for RAPS.

     •  Task C.  To specify the data formats and a data handling
                 system needed for the inventory.

     •  Task D.  To locate and assess existing sources of data suitable
                 for incorporation in the RAPS inventory.
     •  Task E.  To review the emission models available for developing
                 emission data for studies of the type to be carried
                 out in RAPS.

     As noted, the full value of these tasks is only realized if they
are allowed to interact, and a failure to recognize this until late in
the project led to difficulties in formulating and presenting our conclu-
sions.

     In the final form,  summarized below, the findings of the studies
made under Tasks D and E have determined, to a certain degree, the con-
tent of Task B, which in turn reflects certain constraints and opportu-
nities of the data handling system described in Task C.

-------
C.    Organization of the Report

     The first part of this report, the Summary, is an overview of the
project that summarizes the results.  The second and largest part includes
sections on the five separate tasks, each section reporting the appropri-
ate detailed results.  Appendix A is an assembly of data in printout form
that pertains to Task D,  and Appendix B is a lengthy summary of emission
model reports.

-------
                             I I  SUMMARY
A.   Basic Concepts

     The requirements of the RAPS program are wide ranging, diverse, and,
in many cases, stipulate degrees of spatial and temporal resolution  that
have not hitherto been achieved.  For the multiple purposes of  the RAPS
program, it is thus more appropriate to consider data on pollutant emis-
sions in terms of a system of data resources than in terms of a single
all-serving emission inventory.  Alternatively, one can look upon an
emission inventory for RAPS as being a multifaceted data source that
comprises a number of elements, which are capable of providing  needed
information as required.  In either case, we are led to a system in
which varying criteria are applied to determining the nature and extent
of the inventory content.

     A major distinction can be drawn between the requirements  of RAPS
for high resolution data (in space and time), for such purposes as the
verification and development of air quality models, and the more general
needs of the program, such as, for example, investigations of economic,
social,  or political processes, or the investigation of control or abate-
ment strategies not immediately served by the air quality modeling studies.
This distinction leads immediately to a concept in which a special high
resolution inventory (for pollutants of significance) is provided to the
air quality modeling studies and other investigations (such as  transpor-
tation processes or improved inventory methodology) where such  high  res-
olution data is appropriate.  Those pollutants are S00, NO , CO, and
                                                     ^    !X
hydrocarbons (HCs).  For other pollutants, the high costs and difficul-
ties of providing such high resolution emission data on a routine basis
are not justified.  For such other pollutants, we recommend that the
National Emissions Data System (NEDS) Inventory be used.

     When special information on emissions is required—for example, for
extra precise high resolution data for limited  'maximum effort" periods
of model validation or for intensive transformation process investiga-
tions—we recommend that each such emission data collected be treated
as a special case.  This will avoid placing excessive requirements on
the basic emission inventories and save cost and effort.

-------
     It is convenient, therefore, in speaking of the RAPS emission inven-
tory, to consider it primarily as the specially developed, high resolution
data source concerned initially with SC> ,  NO ,  CO,  and HC.  This inventory,
                                       A    A
as noted,  must be flexible and capable of serving many purposes.

     To ensure this, it is necessary to retain the basic or prime data
from which it is built in its compilations in order to make revisions or
extensions readily possible.   Data should not be lumped in a form that
applies to a single, or narrowly perceived, requirement.

     The most satisfactory approach, wherever possible, is to provide
files of emissions data for each pollutant in terms of direct statements
of weight of pollutant emitted by each source or class of sources as a
function of point or area location for every hour.   Although this is a
substantial undertaking, as described below,  in most cases it can be ac-
complished to a useful level of precision within the RAPS program.  This
is mainly because the bulk of the emissions is a result of a limited num-
ber of sources.  In a program of the scope of RAPS, it is possible to
treat such sources individually and to use measured data providing direct
or closely related information on emissions as the basis for hourly emis-
sion estimates.  The frequency and quality of the measurements will be
proportional to the magnitude of each source.  This approach differs
from that commonly used in developing emission data for purposes similar
to the RAPS main objectives,  which has perforce relied heavily upon emis-
sion modeling techniques to interpolate from gross annual data to provide
information with high nominal temporal resolution.   In the proposed RAPS
inventory, such modeling techniques would only be applied to estimating
hourly emissions from minor point sources or from area sources, although
in certain of these cases it is recommended that the required precision
demands the development of improved modeling methods for providing hourly
data.  In the case of mobile sources, however,  where considerable progress
has been made in assessing emissions with high spatial and temporal reso-
lution, the current modeling approach is suitable for the RAPS inventory;
the records of mobile sources to be incorporated in the inventory have a
special form.

     The term "emission modeling" sometimes leads to confusion, and it
is necessary to clarify its significance.   Strictly speaking, any compu-
tational manipulation of raw data can be described as modeling, for ex-
ample,  the combination of pollutant concentrations and flow rates to pro-
vide emission estimates.  More often, modeling applies to the derivation
or estimations of emissions from limited prime data, such as the estima-
tion of emissions from residential space heating on an area basis (from
a consideration of population density and daily temperature).  In partic-
ular, the apportionment of emissions from any source or class of sources

-------
on, say, an hourly basis, from data such as gross annual  tonnage  of  fuel
burnt, is commonly performed by modeling using expressions  (which can  be
more or less arbitrary) to describe the patterns of variability that oc-
cur by season, by day of  the week, or by hour of the day.

     In a more comprehensive sense, various emission models have  been
developed for specific purposes.  These may consist of series  of  sub-
elements integrated into a single system for deriving required emissions
data.   Such models are generally intended to provide emissions data  in
a specified form suitable for a particular use, deriving  their estimates
from relatively limited prime data (usually that relating  to gross annual
fuel consumption).  A typical example is the Argonne Hourly SC>2 Model.
This model,  described in detail in Section VII, provides  hourly estimates
of SOg and thermal emissions from power plants, industrial  point  sources,
and one-square-mile areas of residential, commercial, and  institutional
sources.  It derives such estimates from a range of input  parameters
based largely upon fuel consumption statistics as well as  upon other
factors such as temperature and time.

     In considering such comprehensive emission models, it  is  important
to emphasize their limitations for the purposes of RAPS.   Adequate though
they may be for the uses for which they were developed, they reflect the
expediency and economy that was appropriate to such use.   They seek  to
provide the best emissions estimates possible from available (i.e., mini-
mal) data.   They are not the best approach to developing  emission inven-
tories if better data can be specially acquired--as it can  in  the case
of RAPS.  Another aspect of emission models is the way that the specific
output requirements determine the manner of developing the  emission  esti-
mates from the prime data.  This generally results in a lack of the  flex-
ibility needed to meet the multiple and varied demands of  RAPS.   The RAPS
Emission Inventory should make emissions information available in such a
manner that  any desired selection of data may be extracted  and presented
in any desired form.   This requirement is met by providing  emission data
in the form noted above,  i.e.,  as direct statements of weight  of  pollutant
emitted by each source, or class of sources as a function  of location  for
every hour.   In this approach,  the emission modeling techniques,  where
used,  are limited to the production of a common database,  i.e., the gen-
eral purpose high resolution emission inventory.   The task  of  extracting
emissions data for any purpose and of presenting it in any  appropriate
form then becomes a separate function.

     It is recommended, therefore, that the RAPS inventory  be developed
upon the basis of specially collected data,  rather than upon current
emission models (with the exception of mobile source data).  Further,  it
is recommended that the inventory provides flexibility and  universality

-------
by being a simple compilation of all required data in the lowest common
denominator form, i.e..  hourly emissions by weight of each pollutant
(SO ,  NO , HC, and CO).
   £    X

     For other pollutants,  it is recommended that the National Emissions
Data System (NEDS) be used.
B.   Results of Specific Tasks

     1.    Task A

          To determine the potential users and uses of the emission in-
ventory for RAPS, a survey was made.  Users range from the research groups
directly involved in the RAPS program to city, county, and other planning
organizations and air pollution control agencies in the area.   Research
programs other than RAPS were also considered.  The principal uses of
emissions data will be in the verification and development of transport
and diffusion models and in studies of chemical transformation processes,
the main interest of the RAPS program.  Other uses will be in the analysis
of economic, social, and political impact of air pollution and its control
and abatement and in the development of such control and abatement strat-
egies (particularly episode control).  The RAPS emissions data could also
be useful to programs such as the Community Health and Environmental Sur-
veillance System (CHESS) or research studies such as METROMEX (a joint
investigation of meteorological effects of air pollution) or other locally
based investigations requiring detailed knowledge of pollutant emissions.

          Two factors emerge from this survey.  First, the requirements
for these purposes are multiple and diverse, covering a wide variety of
pollutants and very different degrees of temporal and spatial resolution.
Second,  the needs of the air quality modelers are the most demanding, and
they dominate the requirements for high resolution data in space and time.

          The needs for emissions data in the RAPS program can,  thus,
best be met routinely by

          •  Relying on NEDS for data on the broad range of pollutants
             for which high resolution information is not essential.
          •  Developing a special RAPS Emission Inventory capable of
             providing the high resolution data needed by the modelers
             and of serving other purposes.

          For special purposes, such as particular experiments to inves-
tigate chemical transformation processes or studies of the dispersion of

                                   8

-------
specific pollutants for which the above mentioned  inventories  are inade-
quate, appropriate special data collections  should be  made  to  determine
emissions on an experimental basis.   (Such data  may be collected,  pro-
cessed, and stored within the KAPS Emission  Inventory  data  handling sys-
tem, but are not an integral part of  this  inventory.)

          The specification of various air-quality model  input require-
ments that act as common denominators  thus determines  what,  is  required
of the RAPS Emission Inventory.  Specifically,  the requirements are

          •  Pollutants:  SO , CO, NO  , hydrocarbons (by  types),  and
                            2         x
             particulates.
          •  Resolution:  Temporal, hourly,  for  each hour;  spatial,
             0.01 km point sources, 1  km  grid  squares [area sources
             using Universal Transverse Mercator (UTM)  coordinates
             in each case].
          •  Area covered:  The Saint  Louis  Air  Quality Control
             Ken ion (AQC'K).
          •  Period i. overed:  Continuous hour by hour  throughout  the
             years of full RAPS operations.
          •  Units:  Emissions by metric weight; distance in kilo-
             meters, in UTM coordinates.
          •  Output iormat;  Variable, but in l-OKTRAX  compatible  form.
          •  Other information:  Data  on sources (e.g., stack  height,
             exit temperature, velocity) where  appropriate; data  for
             mobile sources, traffic  flow data,  and aircraft movements
             as required for tise  in appropriate  emission  models.
          Jn this task (specification of  the content  of  the  RAPS  Emission
Inventory) an extensive analysis was made of the  problems  of developing
emission data with the necessary spatial and temporal  resolution  and  ac-
curacy to meet  the requirements of RAPS.

          After an initial review of the basic  form the  inventory system
should take to meet the multiple purposes of RAPS, we  discussed each
aspect of such a system in detail.

          Because of the need for precision in  estimating  the  emission
and the requirements for spatial and temporal resolution,  a  quantitative

-------
basis can only be provided by analyses made for specific air quality or
dispersion models (as discussed in Section VII).   The emission inventory
can potentially provide the most complete data input to such models,
depending upon the cost and effort applied, although even in the most
careful direct assessments of emissions at individual sites, uncertain-
ties of ±10 percent must be expected.  We described the pollutants of
interest, distinguishing between the reactive gases, nonreactive gases,
and particulates.  The role of the emission inventory of such pollutants
is discussed in the context of RAPS.  While data can fairly readily be
obtained for the key gaseous pollutants, particulate emissions are ex-
tremely difficult to monitor.  Existing data on such emibsions are often
inaccurate and give no information on the size distribution of the parti-
cles, which is so important in the modeling context.  In the circumstances,
it is proposed that particulates not be included in the high resolution
] n\ < MI t orv on .in hourly basis, and that I he needs •>[' RAPS for parliculate
data for dispersion model studies or research on transformation processes
be dealt with on a special experimental or project basis—including, spe-
cifically, the use of tracer techniques.

          The study team described and analyzed sources categorized ac-
cording to identity,  function, and pollutant emitted.  Each category is
discussed in detail,  necessary for the data collection procedures to be
recommended later.

          The importance of emission and data conversion factors is then
noted.   Such factors are critical to any inventory and especially to an
inventory of the type needed for RAPS.  The economics of the problem,
however, are such that, while it would be desirable to test every size-
able source in the inventory on an individual basis, wherever reasonable
confidence can be placed in standard emission factors, these should be
used in practice.  Finally, the physical characteristics of the sources
needed for the inventory are noted,  and recommendations are made regard-
ing the metric units to be used.

          The second part of Task B was to specify the RAPS Emission In-
ventory system outlined below.

          While NEDS is proposed as a source of information on annual or
long-term emissions from a broad range of pollutants, it is recommended
that a special, high-resolution RAPS Emission Inventory be developed to
provide hourly data emissions of SOg, CO, NOX, and HC in the form of
direct statements of weight of pollutant emitted in kilograms each hour
from each point or line source or area element for the Saint Louis AQCR.
Locational data will be provided in UTM coordinates with a resolution of
at least 0.1 km for point sources.  Mobile sources will be specified in
terms of traffic links (with a resolution of approximately 0.2 km)  for
major routes and in areas for secondary routes.  Both for stationary and
                                   10

-------
mobile sources,  the area  elements will  reflect  either the spatial resolu-
tion of the basic data  from which the emissions  are estimated  (e.g.,  hous-
ing developments) or one  kilometer  grid  squares,  whichever is  the smaller.
Information on  the physical characteristics  of  point sources and the prime
data used  in compiling  the  inventory would also be provided.  It is not
considered practicable  to develop comparable data for participates.

           For initial planning  purposes  we propose that  only point sources
emitting over 100 tons  per  year of  each  pollutant be considered individu-
ally.  The estimation oi  emissions  from  such sources will be accomplished
in classes according to the magnitude of each source.   In general,  we rec-
ommend that emission data be  based  upon  direct  monitoring of emissions
for  the largest  sources.  For other major sources,  direc!  monitoring of
hourly fuel consumption,  operating, or  process  data should be  used as the
basis for  estimating emissions.  For smaller point sources,  we propose
that emissions  be estimated on  the  basis of  generalized  patterns of tem-
poral variation  of emission as  a function of source type applied to the
best available  data on  fuel consumption,  operating,  or process data (per
shift, daily, or weekly).  This approach approximates current  practice
in which annual  totals  of such  data are  used as  a basis  for modeling.
For area sources, we recommend  the  use  of modifications  and adaptions of
already available emission models.  Similarly,  for a first approach to
assessing  mobile source emissions,  we recommend  trie use  of  available
models.  Improved mobile  source models  should be employed,  as  they become
available, to provide higher quality data.   A simple new model should be
developed  to estimate natural emissions  of nonmethane hydrocarbons on
summer days.   in all cases, ue  propose  thai,  special testing efforts be
carried out to  ensure that  the  emission  and  data conversion factors used
to derive  the emission  estimates are as  reliable as possible,  and have
regard for the  high temporal resolution  required.   Standard emission fac-
tors should only be employed  if  a review of  the  source on a  case-by-case
basis indicates  that such a factor  may be used  with confidence.

           The concluding  portion of Task B is a  description of methodology
and detailed procedures lor developing  the inventory in  the form proposed.
Point sources using the NEDS dam are classified according to  size to
faciii iale (he description ol how emission data  would be derived in four
bio.id grounh accoichng  1 <,> the ija^ic records  used to estimate emissions—
Group I,  continuous emissions monitoring records;  Group  II,  continuous
records of (ueL consumption, operating,  or process data;  Croup III,  short
term, periodic records  of fuel  consumption,  operating,  or process data
(per shift, daily,  or weekly);  Group IV,  long-term (semiannual  or annual)
records or estimates of iuel consumption, operating,  or  process  data.
Within Groups II and 111, a further  distinction would be  made between
sources for which separate conversion or emission lac tors are  established
                                    LI

-------
and those for which representative factors will be established.   Data
from Groups I, II,  and III will be collected on an ongoing basis.

          In principle,  it would obviously be desirable to deal with
all sources under Group I.  In practice,  sources should be treated in
the highest ranking group possible within limits of cost and difficulty.
Some 95 percent of the S02 emissions and 70 percent of the NO  emissions
can be derived from direct data (Groups I, II,  and III) on the basis of
routine data inputs from 101 sources with a high confidence based upon
the testing of 65 individual sources.   Similarly,  in the case of CO
sources,  over 90 percent of the emissions can be estimated on an hourly
basis by the treatment of direct data from only six sources.  On the
other hand, it would be necessary to obtain data from and test some ten
or more sources to achieve hourly estimates from direct data for only
50 percent of the HC emissions.  The proposed approach, however, is in-
tended to provide considerable flexibility, so that optimum results can
be achieved within budgetary and other limitations.  Thus, the number of
sources to be treated on an individual basis and the numbers of tests to
be made should be decided upon the basis of an initial survey on a case-
by-case basis.

          In the case of area sources, the shortcomings of existing emis-
sion models require that special efforts be made to improve and verify
these,  for, although such sources contribute only a small proportion of
the total burden, their local effect is significant.  Similarly, in the
case of mobile sources,  while a simple form of existing models will suf-
fice initially (with improved traffic data inputs), the products of more
sophisticated models should be used as they become available.  Finally,
it is noted that a method for estimating emissions of nonmethane hydro-
carbons from natural sources on summer days is needed.

          Task B includes a tentative schedule aimed at providing initial
data on SO2 and NOX by the first quarter of the second year of full activ-
ities,  complete data by the third quarter of that year; effective initial
data on CO and HC would become available by the end of the second year,
and complete data would be available by the third quarter of the second
year.
     3.    Task C

          Task C is the specification of a system for handling, storing,
and processing the relevant data.  Specification is based upon certain
assumptions regarding the nature and capability of the data management
system that will be available as part of the RAPS overall data handling
                                  12

-------
system.  These assumptions either  indicate facilities  that  are  normally
provided by such a system or could be readily provided by appropriate
adaptation or additional programming of a conventional type.

          In order to specify the data formats and structures,  it  is
necessary to determine how the inventory files will  be used as  a system
for source data preparation and entry, for processing  data  through emis-
sion models, and finally, for production of  the data files  for  the vari-
ous user activities.  The system is designed to minimize manual transfor-
mations of the raw data to the form in which it is entered  into the
computer.  Input data are thus based upon the sources, and  all  pertinent
data (e.g., pollutants, fuels, consumption) are entered in  direct  form
as reported.  Rather than using integer codes such as  the SAROAD manual,
we propose the use of direct, mnemonic identifiers.  We also  specify the
use of a hierarchical, list type structure to enter  such data and  similar
necessary information.  This leads to data structures  that  comprise ex-
tendable data items within an appropriate data reference system allowing
maximum flexibility both to accommodate disparate types of  data and to
allow ready extension to cope with new or additional data forms that might
need to be added after the establishment of basic lists or  thesauri.

          The master inventory file will consist of  data items, each of
which describes a source or a set of identical sources.  Each item will
include source identification data, source operating parameters, the set
of primary measurements from which emissions will be computed,  and the
set of computed hourly emission values for each pertinent pollutant.
This inventory file is desU.iu.vi as the core of the whole inventory sys-
tem.  It will be supported and maintained bj a gt-neral-purpose  data
management system and by such special-purpose programs as may be required
for verification, emission computations, and production of  data files
properly formatted for use a.s input to air quality models.
     4,   Task JD

          Task D, i"he review of the existing emission data sources  in  the
Saint L-,;uis area, has been disappointing.  Despite the fact that Saint
Louis ha:- beta the site of a number of advanced air quality studies and
has active and well-organi'/ed local air pollution control agencies, the
available emissions inventorv data is quite inadequate for the purposes
of RAPS.  Although the emissions data that have been collected in the
area place it in the forefront of urban areas in the availability of such
information and the existing data fairly well meets (or met) the require-
ments for which it was collected,  it falls far short of the new and highly
demanding requirements of RAPS.   Spec!fica11v, the data do not provide
for hourlv resolution.

-------
The five major inventories considered are the IPP (1968), IBM
(1970), DAQED (1971), NATO (1971), and NEDS (1973) inventories. Of these,
the first four are now essentially of historical interest only. The NEDS
inventory contains the best current estimates and constitutes a valuable
source of general data for the broad requirements of RAPS. For the prin-
cipal purposes of RAPS (model verification and development and study of
atmospheric transformation processes), NEDS will serve as a basis for
developing the high-resolution RAPS Emission Inventory necessary to meet
the requirements of the modelers and others.

Thus, NEDS provides good information on point sources that is
of use in planning and implementing the RAPS inventory. In addition, it
provides basic inputs on area and mobile sources for the advanced pur-
poses of that inventory.
5. Task E

In this task, 28 existing emission models relevant to the ob-
jectives of RAPS were reviewed in detail. In \ppendix B, complete infor-
mal i
-------
gross, and much reliance is placed upon extensive interpolation procedures.
Since these modeling procedures are clearly of limited precision for use
in the context of RAPS requirements and since, in the absence of extensive
verification, confidence in their use must be guarded, it was not possible
to recommend the use of such models unreservedly for the RAPS program.
Certain models appear to have potential utility for the purposes of RAPS
depending upon the degree to which interpolation procedures will be needed
(as distinct from the preferable procedure of deriving emissions directly
from time resolved related data). These models are noted below.

In addition to the review of existing models, a brief analysis
is also made of the nature of and need for verification procedures. In
particular, a procedure is set out for evaluating the validity of mobile
source models by combining specially collected experimental data on traf-
fic volume with basic statistical data.

Another point addressed in detail is the degree of resolution
required for the specification of location of point sources. The inter-
play of source strength, stack height, lateral and longitudinal receptor
position, and atmospheric stability makes the definition of locational
precision complicated and complex. Accordingly, a new methodology is
developed to enable the varying interrelated factors to be considered
for any given circumstances, enabling the effects of positional error
to be assessed for any downwind distance or distance interval, and pro-
viding a recommended spatial resolution for each source specification
based upon stack height and emission rate. Finally, the role of natural
mechanisms in adding significantly to the man-made pollutant burden of
the Saint Louis region is considered.

The specific conclusions that result from the studies carried
out under Task E are listed here.
i
• No existing models adequately meet all the requirements
of RAPS.
• To provide emissions data with the necessary high resolu-
tion in space and time, it is recommended that

- Direct information of emissions or of factors deter-
mining emissions be acquired to the extent possible.
Where this is not possible (e.g., for small point
sources and area sources) the most suitable of the
existing models should be used in adapted and improved
form.
15
-------
• Specifically, for stationary sources

Direct data from emissions monitoring should be used
wherever available (probably only for the largest
sources, such as power stations).

Information on hourly fuel consumption rates, process
operations, and so forth should be obtained and used
to derive emissions for other large sources.

- Modeling procedures for small point sources and area
sources, based upon the models noted below, should be
used, but, with refinement of the input data wherever
possible.

- Recommended models are the Argonne model (for hourly
SC>2 from distributed residential, commercial, and in-
stitutional sources; and hourly SC>2 and heat from major
point sources) and the Systems Applications, Inc. (SAI)
model1'3 (for hourly NOV and HC from stationary point
.A.
and area sources).

For automotive mobile sources, we recommend that an
average route speed model be used with a link or line
source geometry supplemented with measured data of traf-
fic flow from fixed sensors on high volume freeways and
on selected arterials.

- The recommended models are the SAI model1'3 (for hourly
CO, NC> and HC with modified inputs derived from Stanford
j\.
Research Institute's model4 for spatial and temporal dis-
tribution of vehicle number and speed on a link basis for
primary traffic and area basis for secondary traffic).
- For other mobile sources, the recommended models are the
Geomet model (for diurnal emissions from river vessels
and railroads) and the Northern Research and Engineering
model as revised by Geomet (in preparation, for aircraft
emissions).

For both stationary and mobile sources, the Ontario De-
partment of the Environment 1971 Model for Toronto should
be referred to for overall guidance and planning method-
ology.

• For microscale studies under the RAPS program, the basic
inventory will need to be supplemented with special emis-
sion inputs. These should be obtained on an experimental
or project basis (e.g., the acquisition of detailed data
16
-------
from small stationary sources or for mobile sources, con-
tinuous and detailed monitoring of vehicle behavior in a
given area with the application of a multimodal emission
model to forecast emissions).

In view of the limitations of existing mobile emission
source models and the impossibility of resorting to direct
information, it is recommended that special steps be taken
to improve and verify such mobile emission models. Spe-
cifically,

- An experimental evaluation should be made of mobile
emissions models to refine the specification of process
input parameters and the accuracy of postulated input-
output relationships.

Parameterization of inputs for mobile emissions (statis-
tical database plus continuous measurement of vehicle
volume, speed, and mix on key links to generate a dy-
namic link-by-link inventory) should be verified on a
selective basis via aerial photography and/or side-
looking aerial radar.

- Parameterization of outputs from mobile sources should
be evaluated in various models to verify representation
of time-averaged emissions on a per-link basis taking
into account types of link and the impacts of deteriora-
tion, nix, road grade, drag, and so forth.

Of the pollutant emissions (CO, NOV, S00, H0S, nonmethane
x ^ £
HC, and particulates) from natural phenomena in the Saint
Louis area, nonmethane hydrocarbons may contribute a sig-
nificant background during summer daylight hours and should
be included in the RAPS emission inventory. Suitable proce-
dures to estimate these should be developed early in the
RAPS program.

In the context of the RAPS program, the accuracy of point
source locations is best considered in terms of the effect
of errors throughout the study area. Accordingly, we pro-
pose that a measure of error in source location be defined
as the integral value of the error over both the lateral
(crosswind) and longitudinal (downwind) directions.
17
-------
C. Recommendations

The following recommendations provide an outline of an operational
plan to derive the RAPS Emission Inventory.

• An initial survey should be made of point sources emitting over
100 tons per year of S00, NO , CO, or HC. The survey should be
£ X
based upon the information of the NEDS inventory, and as many
as possible of the sources should be inspected. An initial
assessment, based upon the nature and quality of the records
available, should be made on how each source should be treated
by group (according to the data collection procedure).

• With regard for budget and the results of the survey, detailed
plans should be made as to the number of sources to be treated
in each group, especially the number of individual source tests
to be made.

• Meanwhile, the proposed data handling system should be developed
and coordinated with the specific data management system actually
procured for the RAPS project.

• As soon as possible, effort should be applied to the development
and refinement of emission modeling procedures to meet the needs
of the RAPS inventory. Specifically, improved techniques of
handling shorter term input data for both point source and area
sources should be applied to the Argonne model and the SAI model.
For mobile sources, similar improvements should be made on the
basis of the SAI model with modified inputs derived from the
SRI model. A model must be developed to estimate summer day
emissions of nonmethane hydrocarbons from natural sources.

• Following the detailed plan developed from the initial survey,
an intensive program of source testing should be undertaken to
develop a reliable basis of emission factors or conversion fac-
tors to relate the routinely collected input data to the emis-
sions of specified pollutant on an hourly basis.

• Arrangements must be made for the routine periodic collection
of input data from the sources to be treated on that basis or
additional data for modeling purposes not acquired in the source
of the initial survey. Arrangements should also be made to in-
stall any necessary monitoring equipment provided under the RAPS
program for point sources or measuring traffic flow on selected
high volume freeways or arterials.
18
-------
Finally, the system should be operated on a routine basis follow-
ing the sequence noted in Task !; to provide data first on S()(
and \O and later <», CO and 11C.
x
In support of (he inventory development effort, tin1 follouin<;
inputs should he sought from otlur elements ot HAPS ,is e.irlv as
poss i h i e;

- Tests •)[ specific air qualit\ or d i spi >rs i on modi-Is should
be wad e to ascertain '-ens i t i \r i t \ ti> the pri-c i s i on ruid reso-
lution 01 emissions data, in (he Saint Louis ati',. (usit,^ \i.i)h
'ia t a a ;: a ba s j s ) .

- The \'icinitv oi identilied so that consid-
eration can b( 5'iven to t i'ea , i :ii; tlu>ni on an individual basis
even though they do noi em 11 ovc.'r 100 tons of pollutants prr
veai .
-------
Ill TASK A: DEFINITION OF POTENTIAL USERS AND USES
A. General

An emission inventory is a fundamental part of any air pollution
study. The Regional Air Pollution Study (RAPS) in Saint Louis is ex-
pected to be unique in scope and depth. Consequently, the emission
inventory contemplated for this study needs to be far more detailed and
accurate than those currently available.

The actual scope, content, and format of the RAPS inventory is de-
termined by its proposed use, and, in order to ensure that the inventory
will correspond as closely as possible to the demands placed on it by
future users, a survey was made to determine the characteristics desired
by potential users both within EPA and elsewhere. Primary consideration
was given to RAPS requirements. To complete the picture, the needs of
other potential users were then considered.
B. Summary

Although an emission inventory is a key information base for all
air pollution studies and control strategies, it is of direct use to
RAPS investigators concerned with the relationship between ambient
concentrations of pollutants and their source, that is, the modelers.
Other groups directly concerned with emission inventories are (1) the
regulatory agencies charged with the achievement and maintenance of the
National Air Standards (this group has the responsibility and authority
to collect and store emission data) and (2) other workers in the field
concerned with transportation problems, health effects, plant damage,
and the like (these groups are primarily concerned with ambient concen-
trations of pollutants of interest).

The requirements of these users are quite different. The modelers
are concerned primarily with those pollutants for which air quality
standards have been promulgated, that is, sulfur dioxide, carbon monox-
ide, hydrocarbons, nitrogen dioxide, and suspended particulates. The
information they require is very detailed (accurate location of the
sources, detailed information on emissions as a function of time, and
precise estimates of quantities emitted are all essential). The period
21
-------
for which such high resolution data is needed will vary considerably.
In many cases, model verifications will be limited to relatively short
periods, of, say, 48 to 72 hours, selected for representative meteoro-
logical conditions. Since such periods may not be readily determinable
until a later time, when the best combination of circumstances is deter-
mined, post facto, it is necessary to collect appropriate emissions data
on a continuing basis, parallel with the even more comprehensive contin-
uous measurements of air quality and meteorological factors.

The requirements of the people working with health effects or plant
damage, on the other hand, are quite different. The range of pollutants
of interest is very broad, including not only compounds of known toxic
effects such as mercury or cadmium but also an almost infinite variety
of compounds suspected of having an effect on plants or people. For
these purposes, detailed time resolution of emission is seldom required
(except in cases of accidents or unusual meteorological events) since
most effects are cumulative over long periods of time. Therefore, long-
time historical records on many substances are of main importance in de-
termining causative relationships of air pollution to health or vegeta-
tion effects.

State and local regulatory agencies are in a sense caught between
these two requirements. They have to maintain inventories adequate for
the enforcement of legal standards, and they would like to have records
of emissions that are potentially hazardous.

As RAPS may eventually include a number of investigations that cur-
rently are not well defined, it is essential to include in the inventory
design the flexibility required for the addition of other pollutants.
For the moment, however, the primary users of the RAPS emission inventory
are the modelers, and it is their requirements that have to be met. In
the following section the details of these requirements are delineated.
Table 1 summarizes the requirements.
C. RAPS Requirements

The Saint Louis emission inventory is of prime concern to the RAPS
research groups engaged in the development, evaluation, and refinement
of air pollution models. Several types of models are currently under
consideration or active study as part of one RAPS study plan;7 in this
plan, the four basic model types are emission, atmospheric structure,
transformation, and removal process models. Also being considered are
existing models that encompass one or more of the basic models, such as
some models that treat the generation and dispersion of pollutants from
certain transportation sources.

22
-------
g.
fr
s t> 3
a 8
S 3 «
M X X X
X X K X
£ S
Hazardou
Material
1
C ^
ydr
bon
ydr
rbo
Gro
XX XX
tu
oc
Agency
aint Louis
is County
State APC
Air Contro
n
r
f
L
i
gul
Cit
Sai
Ill
Mis
Com
23
-------
Upon examination of the four basic models, it can be seen that only
the emission model is self-contained. The others exhibit various degrees
of interdependence. For example, a model of boundary layer structure
will need to account for the meteorological effects that result from
emissions of, say, heat and water vapor. Similarly, transformation
models must consider both emissions and atmospheric structure models,
while removal process models consider all three models. Thus, in assess-
ing the requirements of the various specific models, we need to recognize
that, while the primary emphasis of the particular model may be one thing,
it is quite likely and indeed probable that many of its input require-
ments derive from the structure of the other model components.

Up to now little effort has been devoted to the development of com-
prehensive or nonspecialized emission models. In fact, the development
of a framework for a RAPS emission model program is an objective of this
study. In the absence of an established emission model that the emis-
sion inventory would serve, we have surveyed the various model require-
ments for emission inputs so as to specify criteria and objectives for
such an inventory. These models and their requirements are discussed
below.
1. Atmospheric Structure Models

Atmospheric structure models describing transport and diffu-
sion within the planetary layer on a regional scale (<30 km) are being
developed at the Model Development Branch, Meteorology Laboratory, NERC,
Research Triangle Park, North Carolina. The specific requirements re-
lated to emissions are shown in Table 2. The model is currently under
development, and, hence, these requirements reflect our best judgment
based on available information. Three additional models used at the
Meteorology Laboratory of NERC are

• PTMAX, an interactive program that performs an analysis
of the maximum, short-term concentration from a point
source as a function of stability and wind speed.

• PTDIS, an interactive program that computes short-term
concentrations downwind from a point source at distances
specified by the user.

• PTMTP, an interactive program that computes, at multiple
receptors, short-term concentrations resulting from mul-
tiple point sources.
24
-------
Table 2
EMISSION INVENTORY INPUT REQUIREMENTS
FOR ATMOSPHERIC STRUCTURE (WATPIC) MODEL
Content of inventory
Time resolution
Spatial resolution
Preferred medium
Units
Source descriptors
Temporal variations
Level of confidence
SO , CO, particulates, heat, water vapor
£

1 hour

Point sources on ITM grid (.10.«L km)
Area sources on variable1 grid from 1 km"

Tape

Grams/second, watts

Stack height (meters), exit velocity
(meters/second), temperature (°K), stack
diameter (meters) '

Diurnal, daily, seasonal

Depends on source of data (stack sampling,
questionnaires)
We believe that the emission inventory input requirements of these
models are met by the specifications in Table 2. However, when the
manuals for these models become available, we will review those require-
ments.
2. Transformation Processes Models

.vhiny materials emitted into the atmosphere undergo not only
physical dispersion but chemical changes as well. We need to know the
whole life history of a pollutant, from emission through chemical and
physical transformation to eventual removal, before accurately predict-
ing the relationship of emission to air quality.

At present, the fate of six major pollutants is under investi-
gation by modelers at Research Triangle Park and outside contractors under
the guidance of the Chemistry and Physics Laboratory. The pollutants are
25
-------
sulfur oxides, nitrogen oxides, hydrocarbons, oxidant, carbon monoxide,
and particulates. The photochemical transformations have received the
most attention so far. The emission inventory inputs into this model
are listed in Table 3.

Table 3

EMISSION INVENTORY INPUT REQUIREMENTS
FOR THE TRANSFORMATION PROCESSES* MODEL
Content of inventory

Time resolution

Spatial resolution
Preferred medium
Units
Source descriptors
Temporal variations
Level of confidence
S0
NO
v,, iiu , GO, CO, HC, particulates (NO by com-
& X O A
pounds; HC by categories, e.g., paraffins, ole-
fins, aromatics)

1 hour

Point sources on UTM grid. Line sources (end
points) in UTM coordinates. Area sources on
2
variable-size grid from 1 km . Eulerian models
utilize a fixed rectilinear grid, while La-
grangian models may require a flexible grid
aligned along a given wind trajectory.

Tape

Grams, meters, seconds

See Table 1

Diurnal, daily, seasonal

Depends on source data
Principal candidate-model EPA contractors:

1. Systems Applications, Inc.
Beverly Hills, California 90202
2. General Research Corporation
Santa Barbara, California 93105

3. Pacific Environmental Services, Inc.
Santa Monica, California 90403

4. Systems, Science & Software, Inc.
La Jolla, California 92037
26
-------
3.
Mobile Sources
a.
Surface Vehicles
Emissions from mobile sources such as automobiles, trucks,
railroads, and vessels constitute an important segment of urban pollution.
They account for virtually all of the carbon monoxide, the bulk of hydro-
carbons, and a significant share of nitrogen oxides.

Since emissions from these sources are a function of the
speed, number, and type of vehicles and the traffic patterns for given
highway segments, they are best represented as line sources for heavily
traveled streets. Lightly traveled areas can be considered area sources.

Emission inventory inputs into transportation-related
models will require items indicated in Table 4. Emission requirements
of models that treat vehicular sources have been limited to the require-
ments of mesoscale transportation modes. Microscale models—those treat-
ing near-source effects—may have more stringent emission requirements
both in quantity, types, and precision of the data. Consequently, these
Table 4

EMISSION INVENTORY INPUT REQUIREMENTS
FOR VEHICULAR SOURCES MODEL
Content of inventory

Time resolution

Spatial resolution
Preferred medium

Uni ts

Source descriptors

Temporal variations

Level of confidence
CO, hydrocarbons (by types such as olefins,
paraffins, aromatics), NO , particulates
x
1 hour

Street segments. Endpoints (in UTM coordi-
nates), width (meters), center strip width
(meters). Type of roadway at grade, elevated,
o
cut section, street canyon. 1 km (minimum)
blocks for area sources.

Tape

Grams/second, meter

Mean speed (meter/second), traffic volume by
mix

Diurnal, daily, seasonal

Depends on source data (sampling, traffic
counts, simulated traffic patterns, and the
like)
27
-------
types of highly specialized requirements are not within the scope of the
basic inventory and will need to be obtained as a special task in conjunc-
tion with the particular microscale model experiments and evaluations.
b. Aircraft and Airports

Models dealing with pollutant emissions from planes on
the ground level and above airports are also being developed. These
are essentially microscale models dealing with the movement of individual
aircraft during landings, taxiing, and takeoff. The degree of detail of
emission data required for such a model (Table 5) is probably outside the
scope of a regional inventory.

This work is coordinated by the Model Applications Branch
of the Meteorology Laboratory, NERC, Research Triangle Park.
Table 5
EMISSION INVENTORY INPUT REQUIREMENTS
FOR MOBILE SOURCES (AIRPORTS) MODEL
Content of inventory

Time resolution

Spatial resolution

Preferred medium
Units
Source descriptors
Temporal variations
Level of confidence
CO, hydrocarbons, NO , particulates
X
Every flight: taxiing time, takeoff and
landing

Each runway, taxiway, gate, direction of
takeoff

Tape

Grams/second, meter

Type of plane
c.
Railroads and Vessels
The contribution of railroads and vessels to the air pol-
lution burden of the Saint Louis area is under investigation by the Trans-
portation Center, Department of Transportation, Cambridge, Massachusetts.
28
-------
The Center is conducting a study that will include a sur-
vey, an experimental program to obtain emission data and some dispersion
modeling studies. In the course of this work, the Center will be both a
user of and a contributor to the RAPS emission inventory.

The main pollutants of concern are SO and particulates.
4. Other Studies

a. Particulate Emissions

Visibility loss caused by suspended fine particles is one
of the most obvious manifestations of air pollution. Particles also may
be involved in gas phase reactions and provide one of the mechanisms for
the removal of nitrogen and sulfur oxides. Thus, the formation and re-
moval of fine particles is part of our understanding of atmospheric re-
actions and will ultimately have to be accounted for in models describing
such reactions.

Particles also play an important role in health studies,
since those fine enough to remain suspended for indefinite periods (2 to
3 microns) are breathed in and can deposit in the lungs. Thus, there is
a particular interest in this fraction of particulate emissions.

Studies are under way under the auspices of the Chemistry
and Physics Laboratory, NEHC-KTP, in the Saint Louis area to measure par-
ticle size distribution and their chemical analysis. Data on the emission
of particles are needed, including data on size distribution of major
sources. At the very least,, da+a on the coarse (greater than 3pJ and
fine (less than 3^) fractions should be obtained.
b. Various Analyses

In addition to the modeling studies dealing with the phys-
ical characteristics of air pollution, such as emissions, air movement,
or ambient concentrations of pollutants, the RAPS data are expected to
be used for a number of studies arid analyses dealing with the social,
economic, and political impacts of air pollution and its control. Most
of these studies will be based on long-term averages, and they do not
require the fine resolution the RAPS inventory will provide. Still,
they will profit from the increased accuracy of the new inventory.
29
-------
The development of an inventory of the scope and detail
of the Saint Louis study is in itself no mean achievement. On the basis
of the experience gained in this study, guidelines for improved emission
inventory procedures will no doubt be developed.
D. State and Local Control Agencies

The Metropolitan Saint Louis Interstate Air Quality Control Region
comprises two states—Missouri and Illinois—and 12 counties. Five
counties are in Missouri (Saint Louis City, Saint Louis, Saint Charles,
Jefferson, and Franklin); seven are in Illinois (Bond, Clinton, Madison,
Randolph, Saint Clair, and Washington). The Missouri counties are under
the jurisdiction of the EPA Region VII office at Kansas City, Missouri.
The Illinois counties are the responsibility of the EPA Region V office
at Chicago, Illinois. Saint Louis City and Saint Louis County have sepa-
rate air pollution control offices; the other Missouri counties are ad-
ministered directly by the Missouri Air Conservation Commission at Jef-
ferson City. The seven Illinois counties are under the jurisdiction of
the Illinois Environmental Protection Agency. The area is shown in Fig-
ure 1. The crosshatched areas are those covered more completely by mon-
itoring stations.

The ultimate purpose of the state and local control agencies is to
achieve and maintain the air quality that corresponds to at least the
national primary and secondary air standards, although states may estab-
lish more stringent standards. Thus, control agencies are concerned not
only with pollution episodes (emergency conditions), but also with long-
term pollution control programs aimed at a gradual reduction of air pol-
lution levels and with community growth planning as it relates to added
pollution. Ultimately, they are responsible for the effect of pollution
on the health of the community, its vegetation, and perhaps even weather
modifications.
1. City of Saint Louis

The Air Pollution Control Office, Department of Public Safety,
City of Saint Louis, has a source evaluation section in charge of the
emission inventory; it has identified some 16,000 point sources, but
reliable data are available primarily for the large sources. The sec-
tion's primary mission is enforcement rather than data collection.

The emission inventory is in the form of reports gathered by
inspectors. The data were not computerized. See Table 6.
30
-------
I

I
/ \
PIKE
L.
0 10 20 30 40 50

SCALE — km
EMISSION RATE — tons/day

• 10 - 100 of any single
± > 100 pollutant
SA-1365-33

FIGURE 1 METROPOLITAN SAINT LOUIS INTERSTATE AIR QUALITY CONTROL REGION
31
-------
Table 6
EMISSION INVENTORY INPUT REQUIREMENTS
OF THE AIR POLLUTION CONTROL OFFICE,
CITY OF SAINT LOUIS
Content of inventory

Time resolution

Spatial resolution

Preferred medium

Units

Source descriptors

Temporal variations

Level of confidence
SO (HC, CO), NO , particulates
^ X
Yearly average output

Point sources, UTM coordinates

Reports

Engineering

Stack height, exit velocity, temperature

Inspectors' reports
2. Saint Louis County

The Air Pollution Control Division, Saint Louis County Health
Department, like its Saint Louis City counterpart, is primarily an enforce-
ment agency. It has identified the major pollution sources (about five
that are emitting more than 100 tons per day, about 20 with emissions in
excess of 25 tons per day). Some modeling of sulfur dioxide concentra-
tions has been done, and some CO modeling is contemplated. The inventory
(Table 7) is based on inspectors' reports. The control of additional
pollution resulting from the growth of the area through proper land use
is under consideration.
3. Missouri Air Conservation Commission, Jefferson City,
Missouri

This agency has responsibility for the air quality of the state
of Missouri. It operates through county or local air pollution control
boards where such agencies exist, and it operates directly in other areas.
Its function is to promulgate and enforce air pollution standards for the
state of Missouri.
32
-------
Table 7
EMISSION INVENTORY INPUT REQUIREMENTS
OF THE AIR POLLUTION CONTROL DIVISION,
SAINT LOUIS COUNTY HEALTH DEPARTMENT
Content of inventory

Time resolution

Spatial resolution

Preferred medium

Units

Source descriptors
Temporal variations

Level of confidence
SO . NO , CO, EC, particulates
2 x
Yearly averages

Point sources on UTM coordinates

Printout or tape

Engineering

Stack height, exit velocity, temperature
None

Major sources have been updated and some
stack sampling is performed every six
months
At present, the commission maintains data on particulates, SO ,
id
NOX, total hydrocarbons, and CO. The data are based on an original sur-
vey carried out in 1968 (for particulates and 862 only), which has been
updated in 1970 by IBM to include the other three pollutants. It was
further updated in 1972 by Radian, Inc., and is now in National Emissions
Data System (NEDS) format.

Emission input requirements are similar to those shown in Table
7 for Saint Louis County.
4. Illinois Environmental Protection Agency, Division of
Air Pollution Control, Springfield, Illinois

The Illinois side of the Saint Louis Air Quality Control Region,
which is heavily industrialized, is under the jurisdiction of the Illinois
Environmental Protection Agency through its regional office (State Region
IV, Collinsville) and the central office at Springfield. The Surveillance
Section is in charge of the emission inventory, which is continually re-
vised and enlarged. Currently it contains information on about 2,500
sources, but it is expected to reach 14,000 sources. The inventory is
in the form of a computer printout, and it is used constantly as a guide
33
-------
for issuance of permits (surveillance), as a source of information to
other agencies and interested parties, and as an aid in modeling. The
modeling is contracted to Argonne National Laboratories, the State Im-
plementation Plan is based on the Argonne work. Emission inputs are
shown in Table 8.

Table 8

EMISSION INVENTORY INPUT REQUIREMENTS
OF THE ILLINOIS ENVIRONMENTAL PROTECTION AGENCY,
AIR POLLUTION CONTROL DIVISION
Content of inventory
Time resolution

Spatial resolution

Preferred medium
Units
Source descriptors
Temporal variations
Level of confidence
SO , CO, NO , HC, particulates
£ X
Quarterly, seasonal
Daily during air pollution episodes

Point sources: UTM coordinates ±0.1 km
Area sources by counties

Printout, tape

Engineering

Stack height, exit velocity, temperature
5. Federal Highway Department, Department of Transportation,
400 - 7th Street, N.W., Washington, D.C.

The Federal Highway Department, in cooperation with the Air
Data Office of EPA, will conduct a study designed to provide information
on driving patterns on highways in the Saint Louis area. This informa-
tion, in combination with emission data based on different driving modes,
will permit an accurate assignment of the contribution of automotive
traffic to air pollution in Saint Louis. The emission inputs necessary
for this study will have to be developed from dynamometer runs under
simulated driving conditions for the pollutants of interest (CO, hydro-
carbons, NO ).
x
34
-------
E. The Planning Agencies

1. East-West Gateway Coordinating Council

The East-West Gateway Coordinating Council (Laclede Gas Build-
ing, Saint Louis, Missouri) is a coordinating agency created to facili-
tate the land use and transportation planning for the Saint Louis metro-
politan area. II helps to coordinate the planning efforts of the Missouri
State Highway Commission, the Illinois Department of Transportation, and
various county and city planning dep;i tments. It has also received sup-
port from the U.S. Department of Housing and Urban Development and the
Federal Urban Mass Transportation Administration.

The council has produced a comprehensive Transportation Study
Report analyzing present problems and suggesting long- and short-range
solutions to transportation needs based on socioeconomic projections,
analyses of present traffic patterns, and estimates of future travel
demands. The council has generated an "Average Daily Traffic Volume
Summary" for the Saint Louis metropolitan area (1970), which is updated
yearly and constitutes a valuable source of information on mobile sources.

The council, thus, is both a source and a user of data on mo-
bile sources. Additional traffic data are available from

• The Illinois Department of Transportation, 930 St. Clair
Avenue, Fairview Heights, Illinois (Mr. Robert Kronst,
District Engineer),

• The Missouri State Highway Department, State Highway
Building, Jefferson City, Missouri (Mr. Carl Klam,
Highway Engineer).

• The Saint Louis County Department of Highways, 120 N. Gay
Street, Clayton, Missouri (Mr. Richard Daykin, Director).
• The Saint Louis City Traffic Department, City Hall,
Saint Louis, Missouri (Mr. Jim Bauman).
2. Industrial Waste Control Council

The Industrial Waste Control Council is an association of pol-
lution control officers from the major industrial firms in the Saint
Louis area. As the problems they face are to some extent common to all
of them, they share technical information and develop public relations.
Their members are probably in the best position to develop emission
35
-------
inventories for their respective companies, inventories which they need
to satisfy legal requirements. The data thus developed are generally
confidential and available only to organizations empowered to receive
them.
F. Research Programs
1. Metromex
Metromex is a joint program involving groups from Argonne
National Laboratory, Battelle Pacific Northwest Laboratories, University
of Chicago, Illinois State Water Survey, Stanford Research Institute,
and the University of Wyoming. It is basically a federation of individual
scientists with diverse sources of funding, but common or at least compli-
mentary research interests in the inadvertent weather modifications pro-
duced by large urban complexes through the interaction of their emission
of pollutants and their contribution to the energy balance of the area.
The study is centered in the Saint Louis area. Pollutants of interest
to Metromex are primarily suspended particulate matter. Thus, major
sources of particles, their physical properties and chemical analysis,
and their size distribution are all parameters of interest.
2. Community Health and Environmental Surveillance System

The Community Health and Environmental Surveillance System
(CHESS) is a national program that relates community health to changing
environmental quality. It consists of a series of standardized epidemi-
ologic studies designed to measure simultaneously environmental quality
and sensitive health indicators in sets of communities representing ex-
posure gradients for common air pollutants. The program is conducted
by EPA in cooperation with local public health agencies, universities,
and private research institutes. The purpose of the CHESS program is

• To evaluate existing environmental standards

« To obtain health intelligence for new standards

• To document health benefits of air pollution control.

One of the areas under study is Saint Louis.

As mentioned earlier, CHESS, like other health effect studies,
requires primarily ambient concentration data rather than emission data.
At present, it obtains such data from its own network of air monitoring
36
-------
stations sited within each CHESS community. Eventually, RAPS data may
be used. Since historical data on ambient concentrations are generally
not available, estimates of past pollutant concentrations have to be
based on past emission data from which estimated ambient concentrations
are derived via meteorological dispersion models. Thus, CHESS has some
need for emission data. The CHESS program is administered by the Human
Studies Laboratory at Research Triangle Park. The Bioenvironmental Mea-
surements Branch is responsible for obtaining exposure data for the use
of epidemiologists. Compounds of interest to these studies are shown
in Table 9.

Table 9
EMISSION INVENTORY INPUT REQUIREMENTS
FOR CHESS STUDIES
Content of inventory
Time resolution
Spatial resolution
Preferred medium
Units
Source descriptors
Temporal variations
Level of confidence
SO,
NO
particulates, especially sulfates,
nitrates,and organics (size fractionated, if
possible), hydrocarbons, Oo, CO, trace metals,
e.g., Cd, Zn, Pb, Hg, Cr, V, Ni, Cu, Mn, as-
bestos, benzo-a-pyrene, PCB, pollens

For acute studies: 1 to 4 hours
For chronic studies: cumulative exposures

School districts

Printout

Gram/cent imeter/second
Seasonal
3. University Research

Individual researchers at several of the universities in the
study area are engaged in work requiring knowledge of pollutant levels.
For example, at Washington University, Dr. Richard Gardener is conducting
37
-------
work in the area of thermal pollution— heat islands '—and constructing
an aerosol submodel in conjunction with other agencies. Dr. Christopher
Hill is studying the total (mining, processing, use, and disposal) envi-
ronmental impact of certain industries. At the Center for the Biology
of Natural Systems, Dr. Charles Lee is establishing pollution burdens
in rain and soil samples and in human tissues.

At Saint Louis University, Dr. Slavin is studying the relation-
ship of air pollution to health problems. There seem to be strong cor-
relations between airborne sulfates and asthma attacks, between S0? and
emphysema, and between carbon monoxide and heart attacks. His department
is collecting a daily pollen count and could make it available to the
RAPS inventory.

At Southern Illinois University, Dr. Al Kahn is carrying out
studies of the carbon monoxide and hemoglobin reactions.

As mentioned earlier in this report, these activities have
only a peripheral requirement for emission data at this point. However,
to a certain extent, the format of the inventory should be designed with
the potential to accommodate such data if they should become necessary.
4. Plant Damage

There are a number of compounds in the atmosphere that cause
vegetation damage either by reducing yield and growth or by affecting
the quality of the crop. Well over 90 percent of the losses are caused
by the following pollutants:

• Ozone
• Peroxyacetyl nitrate (PAN)
• Oxides of nitrogen
• Sulfur dioxide
• Fluorides

Out of 65 standard metropolitan statistical areas, the Saint Louis metro-
politan area is rated by Benedict8 16th with respect to potential damage
due to hydrocarbons, 24th with respect to oxides of nitrogen, and 18th
with respect to sulfur dioxide on a scale that takes into consideration
total emissions, meteorological factors, area, and stagnation periods.

Thus, plant damage constitutes a significant aspect of air
pollution in Saint Louis, something that should be taken into account
in setting up the emission inventory. Data on oxidants—ozone, PAN, and
38
-------
oxides of nitrogen—and sulfur dioxide would normally be covered by the
inventory. It may be advisable to make provisions for a fluoride inven-
tory as well. The main known sources of fluorides are aluminum, phosphate,
and ceramic plants.
G. Conclusions

The emission requirements data for RAPS are such that inventories
taken for other purposes—such as the Implementation Planning Program
(IPP) or air pollution control agencies—are inadequate mainly in the
degree of detail required for modeling studies. Since the verification
of atmospheric dispersion models is an early and major goal of RAPS, the
requirements of the modelers constitute the most important criterion of
the adequacy of the inventory.

The most important requirements for modeling are

• Pollutants

- From stationary sources: SO . CO, particulates, NO .
2 x
- From mobile sources: CO, NO , particulates, hydrocarbons
X
(by types).

• Resolution

Temporal: 1 hour
Spatial: Point sources—0.01 km.
Area sources—variable size grid from
2
1 to 16 (km) .

• Units. Depends on model, but always metric.

• Output programs. Depends on model. Inventory should provide
FORTRAN interface. Output to include source descriptors such
as location (UTM coordinates), stack height, exit temperature,
and velocity. For traffic models, description to include de-
scriptors of roadway.

• Area. The area to be covered coincides with the boundaries
of the Metropolitan Saint Louis Interstate Air Quality Region.
Alternatively, it could be resolved to a smaller, more central
area covering only the City of Saint Louis and the Counties
of Saint Charles, Jefferson, Saint Louis, Monroe, Madison, and
Saint Clair (Figure 1).
39
-------
• Time. The period will vary considerably. In some cases, model
verifications will be limited to relatively short periods (48
to 72 hours).

The requirements of other study groups, such as the health study
workers, are more extensive, but generally less intensive. They require
the system to store and retrieve information on many compounds, but the
demands for spatial and particularly temporal resolution are much less
critical.

Special programs, such as Metromex, have specific requirements for
certain categories of pollutants, which may or may not be satisfied by
a general emission inventory.
40
-------
IV TASK B: EMISSION INVENTORY CONTENT
A. General Principles

1. Introduction

As is shown in Section III, there are numerous agencies and
groups--both inside and outside the RAPS program—that have actual or
potential needs for emission inventory data from the Saint Louis Air
Quality Control Region (AQCR). Wide differences exist in the specific
data requirements of these agencies, particularly with respect to the
time resolution of the estimated emissions. However, the most demanding
requirements are set by the need to verify experimentally various multi-
source atmospheric dispersion (air quality) models. In particular, the
emission inventories used for this purpose must provide a high degree of
time resolution, giving emission estimates at one-hour intervals. A high
degree of spatial resolution is also required.

Other possible users of emission inventories surveyed under
Task A have generally far less demanding requirements. In particular,
they do not commonly need the refinements of time resolution required for
short-interval dispersion modeling. If they use the emission data for
air quality modeling at all, it would probably be to estimate relatively
long-term, average pollutant concentrations. Others may use inventory
data only for surveillance and for determining the priorities for regu-
latory programs. In any case, the high-resolution emission inventory
designed for use in dispersion model verification will be more than ade-
quate for any of the other uses.

Attempts have been made previously to verify experimentally
predictions of short-term air quality made from dispersion models. How-
ever, the source emission data fed into the dispersion models have been
derived primarily from long-term average emissions (e.g., annual emis-
sions), at least for stationary sources. Furthermore, most of the annual
emission data have been derived by applying emission factors to survey
information on fuel consumption and materials flow, rather than by making
direct measurements. Estimates of hourly emmissions have been derived or
extrapolated from the annual rates by more or less elaborate calculation
procedures (see Section VII). The resulting compilation of extrapolation
procedures, algorithms, and correlations for simulation of short-term
41
-------
emissions may be described, for lack of a better term, as an emission
model. Unfortunately, the errors and uncertainties, both in the basic
data and in the assumptions required for their extrapolation, are so
great (or so indefinite) that estimates of hourly emissions derived from
gross annual emissions cannot be relied upon for the critical tests of
short-term air quality predictions to be carried out in RAPS.

At best, the predictions of pollutant concentrations from an
air quality model can be no more precise than the source emission data
used as input to the model. In addition to errors in emission data, other
sources of error in the predictions, besides the inherent limitations of
the dispersion model itself, include the description and measurement of
the atmospheric structure, the measurement of the actual pollutant con-
centrations, and the representativeness of the pollutant concentrations
actually measured. Assuming that sufficient care is exercised, we can
estimate source emissions more accurately than any of the other input
data to the air quality model. The level of effort needed to ensure the
accuracy of the emission estimates rises rapidly as the required accuracy
is increased, particularly as the time resolution is extended to shorter
and shorter periods. Nevertheless, the emission estimates, being the
least equivocal of the inputs to the air quality model, should be made
accurately enough to ensure that they do not become the limiting factor
in the testing of the model.

There is no practical way by which the emissions from all sources
can actually be measured during the particular hours in which the test of
an air quality model is carried out. Consequently, it is necessary to
employ simulation techniques to estimate the hourly emissions from most
sources, a process that also can be described as emission modeling. How-
ever, such emission modeling should involve as little extrapolation oi data
as possible. In particular, the emission model for a particular source
should, where possible, contain an independent variable to be evaluated
from data collected in real time at the source. Such data can, as in an
extreme case, indicate whether the source was actually in operation during
the period of time under consideration. In the general case, the data
should indicate the value of some variable having a functional relationship
to the emission. Most directly, in sources having continuous monitoring
systems, the variable may be the concentration of the specified pollutant
in the waste gas. In other cases, it may be a quantity such as the rate
of firing of fuel. At all important sources, the relationship between the
emission and the chosen independent variable should be established from
actual measurements. Such measurements essentially constitute a calibra-
tion of the emission model. It will, of course, be necessary to redeter-
mine the relationship between the independent variable and the emission
(recalibrate the model) whenever there are significant changes in the mode
of operation of the source.

42
-------
It is feasible to carry out an emission inventory of the fore-
going type for the Saint Louis AQCR and to store hourly values of the
chosen independent variables and of the corresponding emission estimates
for an appropriate period (for two or three years if desired). However,
the process is necessarily difficult, time consuming, and expensive. It
should therefore be carried out only for the minimum number of pollutants
necessary for verification of atmospheric dispersion models; as discussed
in Task A, these are considered to be SO , NO , HC, and CO. The inventory,
x x
which should be adequate for use in verification of mesoscale atmospheric
dispersion models, may be designated as the RAPS Emission Inventory.

It should be appreciated that the RAPS Emission Inventory will
be essentially separate and distinct from the existing National Emissions
Data System (NEDS) Inventory, which was compiled under far less demanding
criteria. Some information from the NEDS Inventory will be useful in the
RAPS Inventory; for example, the NEDS Inventory locates and identifies
stationary point emission sources and indicates their relative magnitudes.
Some of the data in the NEDS Inventory relative to area sources and some
small point sources may be, if carefully evaluated, adequate for inclusion
in the RAPS Inventory. In addition, NEDS will provide the needed data
on emissions for purposes that do not require high resolution data of the
principle pollutants covered by the RAPS Emission Inventory.

For some anticipated microscale atmospheric dispersion studies,
very high degrees of temporal and spatial resolution of emissions may be
required for specific, small areas. Again, for particular studies, higher
resolution data may be required for a pollutant and included in the RAPS
Emission Inventory. The precise requirements for emissions data in such
cases will vary considerably, and cannot be anticipated at this time.
Specification and execution of these special emission data collections are
basically research tasks that will be best handled by the research project
staff carrying on the dispersion studies.
2. The Emission Inventory System

The following discussion is intended to provide an understanding
of the general process by which information is generated and flows through
the proposed RAPS Emission Inventory system and is intended to present
definitions of terms (see Figure 2).

It should be appreciated that, whereas the eventual emission
inventory is organized primarily according to pollutants, the execution
of the inventory is necessarily organized primarily according to sources
(the manner in which the original data must be obtained). The process
43
-------
DATA
SURVEYS
INTERMITTENT
EMISSION
MEASUREMENTS
CONTINUOUS
EMISSION
MONITORING
PROCESSED EXPERIMENTAL
AND ANALYTICAL DATA
ON EMISSIONS
EMISSION SOURCE
OPERATING
VARIABLES
BASIC SOURCE
DATA FILE
CONTINUOUS
EMISSION
MONITORING

EMISSION
MODEL
\

EMISSION SOURCE
OPERATING
VARIABLES
1
EMISSION
ESTIMATES
EMISSION
INVENTORY
Output

FIGURE 2 EMISSION INVENTORY SYSTEM
SA-2579-42
44
-------
by which the RAPS inventory is to be carried out is essentially conven-
tional; however, the special emphasis on and treatment of certain steps
is determined by the high degree of temporal and spatial resolution
required to meet the specific requirements of RAPS such as verification
of atmospheric dispersion models.

Original information from emission sources is to be compiled
from data surveys and, as far as possible, directly from intermittent
measurements or continuous monitoring of the sources themselves. Sub-
sidiary data on the operating variables and parameters of the sources
must be collected so that correlations can be made between the variables
and parameters and the actual measured emissions. The source variables
are to be quantities that are--or can be--routinely measured as part of
the source operation. Hence, once a correlation is established, the
measurement of the variable can be substituted for the more difficult
operation of measuring the actual pollutant emission.

For a relatively small number of the largest point sources, it
is anticipated that continuous analyzers will be installed to monitor the
concentration of key pollutants (SOg and NOX). For the remaining point
sources as far as practicable, reliance must be placed on intermittent
manual measurements of emissions correlated with continuously measured
process variables as described above, rather than upon the use of exist-
ing emission factors, although the latter can be used in certain cases.

For the smaller point sources, and for area and line sources
(stationary and mobile), individual emission measurements will not be
made, but more accurate emission factors than are currently available for
typical sources must be derived from a sampling program.

Processing procedures for the raw data will depend on the nature
of the information. At least initially, computation of the emissions from
raw experimental data will probably be most conveniently carried out
manually. Once the processed data is suitably organized, it is to be
stored in the Basic Source Data File, which then is the source of infor-
mation required for creation of the Emission Model for each source cate-
gory of sources.

The emission model in this context is essentially a collection
of correlations or algorithms that permit calculation of estimated emis-
sions from readily or routinely measured operating variables and parameters
of the various emission sources (as discussed above). The relationships
must be worked out from the basic source data; they may range from simple
to complex, depending on the nature of the emission sources.
45
-------
To generate specific emission estimates, it will be necessary
to supply to the emission model particular values of the independent
emission source variables pertinent to the desired time interval. These
input data must be collected from the sources on a continuing basis over
whatever time period is required. The Emission Estimates thus generated
provide the input to the Emission Inventory.

The Emission Inventory will consist of the compiled and stored
estimates of the hourly emissions of the specified pollutants on an
hourly basis for the time period (two or more years) of interest to the
RAPS program. The emission estimates corresponding to any chosen sequence
of one-hour intervals will then be withdrawn from the Emission Inventory
to form input for the air quality models and other purposes.
B. Precision of Emission Estimates

Since the requirements for the emission inventory are to be set by
the needs of dispersion modeling, it is necessary to specify, within some
limits, what degree of agreement between predicted and measured pollutant
concentrations will be considered to constitute a verification of a dis-
persion model. While the work of Hilst9 discussed below approaches this
issue, no such specification of model precision has been made. Modelers
have thus far simply striven to obtain the best results that they could
get. In tests of models made to date, it has been common to encounter
deviations of ±50 to 100 percent or more between predicted short-term
pollutant concentrations and the measured values. Better results are
reported for prediction of long-term average concentrations.

An ideal multisource atmospheric dispersion model would permit
accurate prediction of the concentration of a specified pollutant at any
desired point in the geographical region of interest over as short a time
interval as might be of practical interest. In practice, it is necessary
to compromise the requirements for the model with respect to both temporal
and spatial resolution. In addition to the inherent limitations of a
model itself, the precision of measurement of emissions and meteorologi-
cal variables and the practicability and precision of measuring atmospheric
pollutants in the field set practical lower limits on the time intervals
and the receptor areas that can be considered.

The predicted pollutant concentration is assumed to represent the
average value within one square kilometer, which is the smallest element
of the standard geographical coordinate grid. The pollutant concentra-
tion measured by a single monitor will bear a relation to the average
concentration that will be determined by the distribution of the pollutant

46
-------
over the specified area. The distribution of the pollutant will depend
on the geographical and topographical characteristics of the area. The
distribution will therefore vary not only from one grid area to another,
but also within any one grid area under different circumstances and at
different times.

Other experimental studies in the RAPS program are expected to address
the problem of the spatial variation of pollution concentrations within the
receptor area. However, where it is necessary to employ a single monitor
for one square kilometer area, the relationship of the measured concentra-
tion to the true average concentration for the area will necessarily be
statistical and subject to a greater or lesser degree of uncertainty.

For such studies, or the development of microscale submodels designed
to couple the regional scale models to specific localities or points where
measurements are made or air pollution effects are significant, special
efforts must be made to collect detailed emission data on an experimental
basis. The routinely collected d;i t a of the RAPS Emission Inventory should
be designed to serve the broader scale needs of the modelers on a con-
tinuing basis.

It will be important, however, to make a special check in the vicinity
of each permanent monitoring station (say within a radius of one kilometer)
to ensure that any point or other source that could significantly effect
the representativeness of readings made at the monitoring station, are
taken into account. For example, it may be found necessary to treat an
office building heater system efflux as a specially considered point source,
rather than as part of a nonspecified area source. Alternatively, if such
consideration cannot be made in connection with the emission inventory
system, steps should be taken by those responsible for air quality moni-
toring to develop weighting or calibration factors for the station in
question, or even to set its readings aside for certain purposes under
particular wind conditions

As stated above, it is essential that the precision of the pollutant
emission data not limit the precision of the dispersion model. The dif-
ficulty of attaining highly precise measurements or estimations of pollut-
ant emissions increases rapidly as time intervals are shortened. Hence,
it is necessary to make compromises between precision and the cost and
difficulty of measurement and estimation.

The question of the significance of errors in specifying the strength
or location of sources has been addressed in Section VII, especially in
terms of the precision needed in defining position. The sensitivity of air
quality models to such errors is a complicated question and specific to

47
-------
each model and each area. The work of Hi 1st" in a case study of the
TRC Model in Connecticut is indicative of what can and should be done
for the major models to be tested in the RAPS program. Depending upon
the resolution in space and time of the predictions made by an air
quality model, Hilst's findings in New England are encouraging because
the cancelling effect of random uncertainties minimizes the errors due
to inaccurate point source data. Most serious errors can result from
systematic inaccuracies, such as those introduced by incorrectly esti-
mating area sources, and the smaller the proportion of the total inven-
tory reported in this way the better. It is recommended that as early
as possible, specific studies (based initially on NEDS data) of the type
reported by Hilst be performed to provide information on the effects of
varying the degree of precision of emissions data in Saint Louis, espe-
cially in terms of hourly resolution. In the meantime, it is highly
desirable to develop the most detailed and accurate emissions database
economically feasible in Saint Louis, both to ensure the best support
of the air quality model studies and to provide the base data for re-
search in emission modeling methodology.
C. Inventory Resolution

1. Temporal Resolution

As is indicated above, the difficulties both of formulating
an adequate dispersion model and of testing it increase rapidly as the
time interval is shortened. Ideally, a model should be capable of pre-
dicting the atmospheric concentration of a pollutant over the shortest
interval for which an air quality standard is specified. In fact, it is
probably impractical to specify an interval shorter than one hour. One
limitation is that it is practically impossible to specify significant
values for the pollutant emissions from most sources for periods of less
than one hour. Indeed, it is difficult even to make precise estimates of
emissions for periods as short as one hour, at least without making direct
measurements in the particular time interval of concern. In view of these
considerations, the minimum time interval to be considered may be set at
one hour. This degree of time resolution has been accepted by most
modelers.
2. Spatial Resolution

Stationary sources fall into two categories: point sources
and area sources. The distinction between the two is generally arbitrary;
point sources are those having emissions of a specified pollutant in excess
of some selected minimum value, and they are treated individually. Smaller
48
-------
sources within a selected region are aggregated, and their total emission
is assumed to arise uniformly from the area.

Choice of the minimum emission rate for specification of a point
source obviously depends on the nature of the pollutant and the ratio of
the emission from the particular source to the total emission from all
sources.
x
Specification of the location of a source, whether a point or
an area, is to be made by reference to UTM coordinates- The precision
required in specifying the position of a source is also partly related to
the position of the receptor. Some dispersion modelers have asked that
the position of point sources be determined within +10 meters. From the
standpoint of a nearby receptor, the physical size of the stack or vent
may be appreciable; large stacks may be 10 meters or more in diameter,
and some building vents may be still larger in their principal dimension.
An objectivi method for determining the spatial resolution required is
presented in Section VII.

The method enables one to estimate the allowable error Ay, in
the cross wind position of a source (for any specified stack height,
source strength, wind velocity, and stability conditions), as it affects
pollutant concentrations at given downwind distances. "Allowable error"
is defined arbitrarily by comparison with a reference source strength—
e.g., the 100 tons/year commonly used as the lowest value for point
sources, and a reference positional error—e.g., 100 m, the resolution
of most current inventories.

Although such a high degree of precision in specifying location
may only be necessary in a limited number of applications, it is recom-
mended that wherever possible the location of point sources be reported
with a resolution of ±10 meters. Although existing data (in NEDS) is only
specified to ±100 meters, additional surveys of significant sources will
have to be made in any case, and it would therefore be readily possible
to obtain the further locational data at that time.

As noted above, verification of microscale dispersion models
will generally be special exercises. They will require finer spatial
resolution and more detailed specification for local sources than will be
required for tests or use of the mesoscale models. This finer degree of
resolution can be incorporated into the general emission inventory for the
entire area covered by the major dispersion models. However, the addi-
tional time and cost required may not be justified, since the increased
resolution will probably never be required except in relatively few small
areas being treated in the microscale dispersion studies. It is more
practical to make special, fine-resolution inventories for the particular
limited areas of the microscale studies.
49
-------
D. Pollutants

1. General Discussion

Initially, inventories must be carried out for those pollutants
that are to be used in model verification. In principle, it would be
desirable to carry out the inventories simultaneously for as many of the
pollutants as possible, particularly where surveys of the same sources
are involved. But in practice, it may not be entirely feasible to do this,
because the basic data may simply not be available for all of the pollu-
tants at the same time. Acquisition of adequate data on emissions of
nitrogen oxides and hydrocarbons will probably take much longer than will
the gathering of equivalent information for sulfur oxides.
2. Nonreactive Gases

Although none of the major pollutant gases that may be used in
tests of dispersion models are actually nonreactive, both sulfur dioxide
and carbon monoxide are relatively unreactive in relation to the time of
dispersion that is of primary interest to dispersion model verification.
Carbon monoxide is the more inert of the two. However, sulfur dioxide
has the advantage that its reaction products (sulfuric acid mist and
sulfates) can be traced by chemical analysis, whereas the identity of
carbon monoxide is lost after reaction.

Carbon monoxide is best suited to testing models of dispersion
from automotive sources, since these are by far the major sources of this
pollutant. It will be necessary, nevertheless, to inventory stationary
sources, since their contribution to the ambient concentration of carbon
monoxide may be appreciable in relation to the model testing.

It has been found that large quantities of carbon monoxide are
produced in nature by the oxidation of methane in the atmosphere. Carbon
monoxide is removed from the atmosphere by microorganisms in the soil.
The background concentration of carbon monoxide in the atmosphere is
determined by the balance of the natural removal processes and of the
natural and anthropogenic formation of the compound. It is unlikely that
either natural formation or natural removal is important on the time scale
of concern to the dispersion studies and within urban areas, where high
concentrations of carbon monoxide may be generated. Task E presents a
more detailed discussion of natural emissions-

Sulfur dioxide is best suited to use in testing of dispersion
models for pollutants from stationary sources, since relatively little is

50
-------
emitted from mobile sources. It is more reactive than carbon monoxide
and more subject in the atmosphere to the action of various removal
processes. The oxidation of the sulfur dioxide to sulfur trioxide re-
sults in conversion to an aerosol (sulfuric acid mist), and subsequent
or coincident reaction of the sulfuric acid with ammonia or other basic
materials in the atmosphere results in formation of solid aerosols. If
the time interval of interest to the dispersion study is no more than a
few hours, these reactions will probably not be of great consequence.
Fortunately, it is possible to sample for both sulfur dioxide and the
sulfate aerosols and, thus, to deduce the probable history and dispersion
pattern of the total sulfur originally being discharged to the atmosphere
as sulfur dioxide. (About two to three percent of the sulfur emitted
from combustion processes is usually already in the form of sulfur tri-
oxide . )

Sulfur dioxide and its reaction products may be removed, or
scavenged, from the atmosphere by a variety of mechanisms. The sulfur
dioxide may be absorbed by vegetation or reactive solid surfaces or be
adsorbed by a variety of solid surfaces. It may also be absorbed by rain
drops. The sulfuric acid or sulfate aerosols may be rained out or washed
out or may be mechanically deposited by inertial, diffusional, or gravita-
tional mechanisms. In time, the aerosols will also coagulate, increasing
the particle size and accentuating particle deposition by the inertial and
gravitational mechanisms. The coagulation mechanism will be strongly
affected by the total concentration of particulate matter present in the
atmosphere

In general, the influence of the various removal or scavenging
processes on sulfur dioxide should be much greater than the corresponding
influence on carbon monoxide. Allowances must nevertheless be made for
the depletion of both pollutants by the various removal processes, but
determining these adjustments is, of course, a task separate from the
emission inventory. Estimating the depletion of the emitted pollutants
is, in fact, an inherent part of the formulation and development of the
dispersion models. It will call for an extensive supporting experimental
program.
3. Reactive Gases

The most important transformation process involving major air
pollutants is the photochemical reaction of the nitrogen oxides with hydro-
carbons to form PAN and ozone. The nitrogen oxides are contributed in
roughly comparable amounts by both stationary and mobile sources, which
51
-------
also contribute the hydrocarbons. Consequently, it is necessary to
devote comparable attention to an inventory of emissions of both of these
types of pollutants from both classes of sources.

As they are present in waste gases at elevated temperatures, the
nitrogen oxides are primarily in the form of nitric oxide (NO). Partial
conversion of the NO to nitrogen dioxide (NO2) takes place primarily after
the waste gases are cooled and mixed with ambient air. There appears to
be no compelling reason why the distribution of NO and NOg in the hot
waste gases need be determined or specified as part of the emission inven-
tory. The subsequent distribution of the two species in the atmosphere
is, of course, of critical concern in the transformation models, but it
will have to be determined from separate experimental and analytical
studies.

Developing an inventory for nitrogen oxides presents special
problems. Most of the nitrogen oxides are formed by fixation of atmo-
spheric nitrogen at elevated temperatures in the combustion processes, and
the quantity formed is dependent on the temperature reached, the residence
time at high temperature, the size and configuration of furnace or other
combustion devices, the amount of excess air present, and any other fac-
tors that may influence the time and temperature history of the combustion
gases. The nature of the fuel also influences the formation of nitrogen
oxides, primarily through those fuel properties that affect the time and
temperature relationships--flame temperature and flame emissivity. Any
nitrogen compounds in the fuel may also be partly oxidized. Hence, the
quantity of nitrogen oxides emitted from a given piece of combustion
equipment cannot be estimated from first principles by any currently
available methods. Strictly, it must be determined experimentally for each
device and each operating condition.

Fortunately, the range of variation in nitrogen oxides formation
is not so great that reasonably representative values cannot be obtained
by testing specific types of devices over appropriate ranges of operating
conditions. Nevertheless, extensive source sampling will be required to
obtain adequate information for a precise emissions inventory.

Inventory of the hydrocarbon gases and vapors presents a poten-
tially complicated case because of the very large number of different
organic compounds that may be emitted and that may also vary widely in
photochemical reactivity. It is completely impractical even to attempt
to inventory emissions compound by compound, except in instances where,
for example, a relatively pure, known compound is lost by vaporization.
Probably the most that can be done in a practical way is to identify the
organic compounds by major groups or classes (e.g., paraffins, olefins,

52
-------
diolefins, aromatics, alcohols, ketones, and so on). For the most part,
even this will require a major source-sampling program. Identification
and quantitative determination of the classes of compounds, alone, will
call for a large analytical effort. Hence, it is essential to keep the
scale of the sampling and analysis to a minimum consistent with attain-
ment of the program objectives.
4. Particulate Matter

The development of dispersion models for particulate matter, as
well as efforts at verification, are well behind comparable developments
for dispersion models for gases. The dispersion of particulate matter is
more complex than that of gases, since the dynamics of the particles are
strongly affected by the particle size, as are the removal processes that
result in a decay of the particulate cloud. In addition, the nonspecific
nature of most particulate matter makes it difficult, if not impossible,
to identify the source of the material, trace its path from source to
receptor, and allow for losses resulting from removal processes.

In using simple dispersion models, it has been common for workers
to assume that all dust emitted from emission sources remains airborne and
contributes to the resulting atmospheric concentration of suspended par-
ticulate matter at the receptor. Such an assumption is highly unrealistic,
since relatively coarse particles will, in fact, tend to settle out. Rel-
atively fine particles (about 10 microns in diameter and smaller) are
frequently assumed to behave more or less like a gas; this is a fairly
safe assumption, but one still subject to somewhat indeterminate errors.

Making a realistic test of models concerned with particulate
formation or dispersion would require the following information:

• The quantity of particulate material emitted from each
source and the distribution of effective particle sizes
of the particulates. It should be noted that the effec-
tive particle size distribution, which may reflect
particle flocculation, is not necessarily the same as
the distribution that may be indicated by the measuring
techniques commonly employed.
The particle size should, of course, be expressed in
terms of the Stokes diameter (i.e., the diameter of
the equivalent spherical particle of the same density
that would have the same terminal settling velocity as
the particle in question).
53
-------
• The extent to which particles of each size are de-
posited by removal processes during passage over the
terrain between the source and the receptor. No
essential quantitative information is available on
these phenomena. The technical problem is similar
to that of determining the efficiency of dust col-
lection equipment in the collection of particles of
various sizes. However, the studies of particle
deposition from the open atmosphere are necessarily
less controllable than experiments with dust collec-
tors, and work in even the latter field is not well
advanced from the standpoint of basic science and
technology.

Determination of the mass emissions and particle-size distri-
butions of particulates emitted from even the major sources alone would
require a major experimental program. The experimental methods and
procedures for determination of the effective particle size of the mate-
rials as emitted are still poorly developed. In addition, the study of
particle deposition from the atmosphere calls for a far greater experi-
mental effort. If a comprehensive effort is to be made at confirmation
of particulate dispersion models using particulates normally discharged
from various sources, a lengthy and costly preliminary experimental and
theoretical investigation will be required. Although information on
particulates is noted as a requirement by a number of users identified
in Task A, there is no evidence that the implications of including such
data have been seriously considered.

It is therefore open to question whether an attempt to verify
particulate dispersion models should be made using the actual particu-
lates emitted from various sources. An alternative procedure using
particulate tracers may be usable with less preliminary work and less
expense and at an earlier date.

Use of particulate as well as gaseous tracers has already been
proposed as a part of the RAPS program for tracing wind movements and
plume dispersion. Conversely, it is possible to use synthetic particu-
lates prepared from tracer materials (e.g., dyestuffs) to test predictions
of particulate concentrations at various receptor areas, and also to get
a relatively direct measurement of the depletion of the particulate cloud
in its movement from source to receptor. The synthetic particulate matter
can be produced within reasonable ranges of desired particle size, and it
can be discriminated from other particulate matter in the atmosphere by
its chemical or physical characteristics. The most convenient materials
of this kind are those that can be measured by colorimetric or fluoro-
metric techniques.
54
-------
If the synthetic particulates can be used for the testing of
the dispersion models, there will probably be no need to develop a par-
ticulate emission inventory that has a high degree of resolution with
respect to time intervals and particle-size distributions. This will
result in major savings of both cost and time.

For the other purposes of RAPS, e.g., research on transformation
processes, there may be a need for detailed information on particulate con-
centrations with high temporal and spatial resolution. Because of the
difficulties noted above, however, it is suggested that research studies
for these purposes be planned to avoid, as far as possible, dependence
upon general source emissions data. Better results can be achieved by
special monitoring of only the critical sources in a carefully selected
area. The alternative is an extremely costly and difficult inventory
development program.

In summary, it is proposed that data on particulates should
not be included in the high-resolution RAPS inventory emission, since the
means are not currently at hand either to acquire the full range of data
that may in principle be required or to employ the information effectively
if it were obtained.
5. Other Pollutants

In the RAPS area, only a few agencies have indicated an interest
in obtaining inventories of pollutants other than the major ones discussed
above, and none of the agencies that did were concerned with short-term
dispersion modeling. Local and state regulatory agencies were interested
in emission inventories for designated hazardous air pollutants, including
mercury, beryllium, cadmium, lead, and asbestos. They called only for
quarterly or yearly average emission rates or, at most, for daily average
emissions during air pollution episodes.

The CHESS program includes consideration of a wider number of
trace materials as pollutants including cadmium, zinc, lead, mercury,
chromium, vanadium, nickel, copper, manganese, asbestos, benzo-a-pyrene,
and polychlorinated biphenyls. Although CHESS relies mainly on air con-
centration measurements obtained from its monitoring network, it will
probably require estimates of relatively long-term average concentrations
in the years before the air monitoring network was established. Thus, it
may be necessary to prepare estimates of emission rates for these earlier
periods, so that the atmospheric concentrations at those times can be
calculated from use of dispersion models.
55
-------
In general, there is little point in attempting to carry out
emission inventories for pollutants other than the major ones discussed
above until such time as the precise requirements are specified by the
potential users. Variations in the required precision and the time and
space resolution can greatly change the costs and time required for the
inventory.
6. Heat and Water Vapor Releases

Although not pollutants, sensible heat and water vapor releases
are of interest to meteorology and dispersion modeling because of their
effects on atmospheric stability. Studies are now being made of the
effects of the urban "heat island. The sensible heat and water vapor
content in waste gases can generally be calculated readily from data that
will be collected in the course of the emission inventories for sulfur
oxides, nitrogen oxides, and particulate matter. The heat transferred to
the atmosphere by convection and radiation from the surfaces of buildings,
paved areas, and the earth must, of course, be estimated by separate pro-
cedures not related to the emissions inventory.

The temperatures and volumetric flow rates of the waste gases
will be part of the information collected during the inventories. Hence,
it will be possible to calculate the sensible heat content of the gases
above the ambient temperature level. For combustion processes, it will
generally be possible to calculate the water vapor emission from the
quantity and the hydrogen content of the fuel burned. For some other
fuel-burning operations, such as direct-fired drying, the water vapor
present in the waste gas will greatly exceed that formed by burning of
the hydrogen in the fuel. However, the water vapor content of the gas
may be obtained during the sampling for various pollutants. Sometimes
data on the process may permit calculation of the water content of the
gas.
E. Source Categories

1. Classification Scheme

A general classification scheme for emission sources is pre-
sented in Table 10. It is broad enough to encompass all emissions of
pollutants from all sources. The source categories (stationary and
mobile) and subcategories (area, line, and point) are convenient from
the standpoint of dispersion modeling as well as from inventory orga-
nization. The division of source processes into combustion and
56
-------
o

in
0
CJ

0
CQ

CU
H
X!
o

en
0
O
f*i
3
o
CQ

ra
c
o
>H
p
ra
P
CQ

0 r->
o o
r. be
3 0
O p
CQ CO
U

O)
CD
O

o
CQ

0
C

•J

CO
0

0
CJ
3
0
CQ

P
C
•H
o
OH

to
CD
y

0
CO
CO
0

*
ource
:atego:
w 2
3

in
3

d
0
H
-p
01
&
0
u

c
0

0",
3
XI
E
O

q
o
H
J-J
in
3

0
u

c
o

01
XI

u
o

K
D
X!

ource
rocess

•D
C T3
iH CO
in 0
0) 3 in
<-H CH 0
O rH
H bC > ^ Xl

CQ CO

01
in CD
rH CJ
in o H
0 H .q
M in o: ,c 0 oi
0 rH 3 CD > P 0
P-H co x; > c CM
CJ CJ p*i 0 H 0
H "C rH ca E be c
,C IH q -T in ? ft CH

C 01 r-i 0 --H H a* PCD

4Hcoi- 0
p o c: o o
tn >> H P SH
M 3 IH t4 P ro
ra -O O T3 C
H C *— ' H 0 bo
O -H bC cfl C C --»
^ q in p. H H c
0 -H H JH i—l rH O

O E0cadracara
CJ CQ>>HbC^3rH

l
in i/i i
—1 r. !H rH

H P p Ul 01
rH S-H ffl CO O rH
CO P 0 CD ft O

rH-P QJ -Hp,njOPH
QJ rH 13 i— < 0CQ3^CQ MO
E P H rn 3 ra C
§ffi en ra &- £ H
d CD E
O M « CQ
en
p
C
o

0
CQ

ca
in
C 0 CJ
OP -H
_Q ra u C
JH rH H rt
ro 3 c he
CJ CJ CO rH
0 -H be o
rH P f-l C
13 rH O rH

53 &(

•a «*-*
c
en ra X! > 0 01 > 01
rH -H P 0 H 0
•O ra P co (H P >
d cj ca r-" 0 ca H

tnHrHCJ.M'a HH

0 OOO >10'D-IJ £ 0 H 0T!
W cO^UE'OO rHCQ^OQ^
ra ca
O &H

in
P
c
p
3

rH
rH
<
CJ
H

H
CJ
CD in
ft P
in c
T3 0
c en c
M CO P 0
i- H ft
0 to S
ft q 01 T3 o
CO O0 to C U
> Xi > CD ra
in H P r-f
*O ra P ra _C ra
c o ca rH to o
ra 0 > M 3 co H

in XX -ofi-H o-H>i0o

ca ca E
O a, CQ

01
rH
03 tn o
t. ^ in
ao o
ft f-
B ca - /-• in 0
> > in 0 0 ra
PC P
T3 cj c H ra cj
C H O rH rH -H

ra ^ M [H cj ra X
in bcoco o H boo
0 rHtflhCTJP rH £
in o ^ O ^ o w
ra ra
O ft

•o ra
C 01 O
to ra P H

am 0
q in -o rC
co 00 to c u tn
> X! > O CO P
t, H P o d
T; ra P ra -C -H o>
q u ra •-< w «H c
ra o> oi3ra-no
^ H ;•< cj cj a
in xx -OrHrH OH>.a)ea>
0OOO t>i0tJfe'^''*-)r~'ri"O'*

ra co E
O CU co

in
p
c
ra
P
3

-H
O
ft
57
-------An error occurred while trying to OCR this image.
-------
Industrial power and steam plants will usually display
load patterns substantially different from those of the public utility
plants serving the general area. Where the manufacturing plants are
operated on a 24-hour basis, both the electrical power and the steam
required for process use should in general represent fairly steady loads.
On the other hand, the power and steam plants will be essentially if not
completely shut down when the manufacturing plant is out of operation.
The power and steam required for heating and air conditioning the plant
will superimpose a fluctuating load on that resulting from process re-
quirements.

The nature of the load cycle in an industrial power plant
will also be affected if the plant is tied into the public utility power
grid. In such cases the plant will usually send power into the public
system when power demand is low in the manufacturing plant and draw power
from the public system when demand is high.

Depending on the type of manufacturing, electrical loads
may exhibit cyclic behavior, sometimes over relatively short intervals
(e.g., electrical power loads in a steel rolling mill). Batch chemical
processes may have a cyclic steam demand.

To obtain a significant emission inventory from such
sources, particularly an inventory from which it is possible to simulate
one-hour emissions, it is necessary to acquire detailed knowledge of
the operating cycle of each installation as well as of the characteris-
tics of the installation itself, the fuel or fuels used, and so on. A
sampling program to acquire information for derivation of representative
emission factors will have to be carried on.

Although emissions from such sources are commonly esti-
mated by emission modeling techniques (as described in Section VII)
using information on load and operating patterns discussed above, we
recommend that, wherever possible, emission data from the point sources
responsible for the major proportion of emissions be acquired by direct
methods using information on actual emissions, loads or fuel consumption.
Details of this approach are given below.

Incinerators constitute another point combustion source.
The sulfur contents of common wastes are low, so that incinerators are
only rarely significant sources of sulfur dioxide. The variability of
the composition of wastes will, however, make it much more difficult
to correlate the measured emissions with operating variables than with
fuel-burning combustion sources.
59
-------
The design of most existing incinerators has not been
favorable to efficient combustion, so that emissions of carbon monoxide
and hydrocarbons tend to be much higher than in fuel-burning operations.
This problem is frequently aggravated by improper operating conditions.
The formation of nitrogen oxides can vary widely depending on incinerator
operating temperatures as they are affected by the amount of excess air
used in combustion.

Since the variations between different incinerators may be
extreme, emissions are relatively unpredictable. Hence, any large units
(municipal incinerators and possibly some industrial incinerators) should
be tested individually, i.e., on the basis of operating data, such as
daily tonnage of throughput. Smaller industrial and commercial inciner-
ators may be treated by modeling with only representative units being
sampled, since their emissions will be small enough to be estimated with
adequate precision from such data.
Noncombustion Sources—Noncombustion point sources of
pollutants are practically exclusively industrial in nature. They com-
prise such a diverse collection of manufacturing operations that no
generalizations can be offered concerning either their emissions of
pollutants or the factors affecting them. Most noncombustion point
sources will emit one or more of the pollutants of concern to the dis-
persion model studies, as is shown in the following discussion under
each pollutant.

The principal point noncombustion sources of SO,, are found
in petroleum refineries, nonferrous smelters, sulfuric acid plants, and--
possibly—integrated steel mills.

In petroleum refineries, the process sources are the Claus
sulfur plants, the regenerators of catalytic cracking units, and flares
burning waste sour refinery gases. Refinery process heaters or boiler
furnaces burning sulfur-containing fuels (including sour refinery gases)
fall in the category of point combustion sources discussed above. The
flare represents only a combustion source, but it is one of such special-
ized nature that it is best classified with the process sources. Because
of its irregular operation, the flare presents special problems of emis-
sion estimation, which must be worked out in reference to process measure-
ments available and with consideration of the actual level of S02 emmissions
for which the source is responsible. Claus sulfur plants are likely to
be fitted with continuous monitors for measurement of S02 in the tail gases.
In the event that they are not, manual measurements of S02 emissions must
be made and correlated with an available measured process variable, such

60
-------
so that the source location is rather diffuse. Control of such emissions
has been made difficult by the small sizes of many of the individual leaks.
However, vaporization losses from storage tanks has been a large and spe-
cific source of hydrocarbon emissions. Losses from evaporation occur not
only in refineries, but also in subsequent handling and storage of gasoline
and other hydrocarbons liquids; estimates of such emissions should be made
wherever possible.

Other point sources of emissions of hydrocarbons and hydro-
carbon derivatives will include large surface coating operations, such
as the spray painting of automobiles at assembly plants. Data will have
to be developed on the actual consumption of coatings and solvent in these
operations and on the rate of consumption. Even the operators may not
currently have reliable data on anything but the gross long-term con-
sumption.

Chemical analyses should be made to classify the hydrocar-
bons and hydrocarbon derivatives emitted according the major groups
(paraffins, olefins, diolefins, aromatics, alcohols, ketones, and alde-
hydes). Where possible, the actual material vaporized, not the liquid
stocks, should be sampled.

To summarize, Table 11 shows the major noncombustion
sources and the basis of the principal pollutants emitted.
b. Area Sources

Combustion Sources—The stationary area sources are com-
posed of:

• Residences.

• Commercial establishments.
• Institutional establishments (e.g., schools,
hospitals, public buildings).
• Some small industrial (manufacturing) estab-
lishments.

Most of the pollutants arise from the combustion of fuels used for cook-
ing, space and water heating, and steam generation. Some may come from
combustion of wastes in small incinerators. Industrial establishments
in this category are composed mostly of small manufacturing operations
that employ fuels primarily for space heating or that burn small quanti-
ties of waste.

63
-------
Table 11
MAJOR NONCOMBUSTION SOURCES OF POLLUTANTS
Pollutants
Sources
S°
NO
x
Particulate matter
Hydrocarbons and derivatives
Contact sulfuric acid plants
Claus sulfur plants
Coke oven gas production
Copper, lead, and zinc smelters
Sulfonation of organic compounds

Nitric acid plants
Nitration of organic compounds
EOF steel furnaces
Secondary lead smelters
Grey iron foundries
Flour mills
Cement plants
Coke quenchers
Organic chemicals manufacture
Paint and varnish manufacture
Coating of fabrics with rubber
or plastics
Venting of storage tanks for
petroleum products and organic
chemicals
The rate of fuel consumption by the area combustion
sources exhibits pronounced diurnal and seasonal variations as well as
variations related to ambient temperatures. Fuel consumption peaks
appear over periods of one to three hours- The consumption patterns of
residences differ from those of commercial, institutional, and small
industrial operations, which may operate at night and over weekends as
well as during the day. The different types of sources may also use
different kinds of fuel. Individual residences and small apartments
tend to use natural gas, whereas large apartment houses and commercial,
institutional, and industrial operations are relatively more likely to
use oil or coal.
64
-------
The amounts and nature of the emissions from fuel combus-
tion are related not only to the type of fuel burned but also to the type
and the size of the combustion equipment. If waste incineration is used,
the practices employed in incineration may vary widely.

The only feasible way to develop an emission inventory for
the area combustion sources is to apply emission factors characteristic
of the combustion equipment to the quantities of fuels consumed. Unfor-
tunately, there will seldom be data that will give directly the quantities
of various fuels consumed by particular types of equipment and by specific
types of users. The flow of natural gas to various sections of a city may
be measured at metering stations. However, the measurements will indicate
only the temporal pattern of the total flow to all users. Data on sales
of oil and coal indicate only the average rates of consumption over rela-
tively extended periods of time.

For significant area sources, it will be necessary to deter-
mine, by survey and experiment, the amounts and the patterns of fuel usage
by typical components of area sources. For instance, the consumption of
fuel by a typical apartment for cooking, water heating, and space heating
could be measured and fitted to a model incorporating diurnal and seasonal
influences and the effects of ambient temperature. Pollutant emission
factors should be determined experimentally for the combustion devices
unless they are already available. Similar determinations should be made
for other representative source units: individual homes, offices or office
buildings, retail stores, and such.
Noncombustion Sources--The most important noncombustion
area sources are those that emit organic vapors and may make appreciable
contributions to the total emission of hydrocarbons. Prime examples
include vapors from (1) gasoline storage and handling by distributors
and service stations, (2) dry cleaners, (3) painting, (4) degreasing,
and (5) application of asphalt to roofing and pavement.

The total quantity of vapors released can be estimated
fairly readily by applying appropriate emission factors to the quantity
of gasoline handled, the amount of cleaning fluids consumed, the amount
of paint applied, and so on. However, attainment of adequate spatial and
temporal resolution will require a fairly extensive survey of the sources
and a detailed study of operations at a representative sample of estab-
lishments or activities of each type.

The emissions of gasoline vapors from automobile service
stations will vary with the day of the week and the hour of the day, as
65
-------
the volume of business fluctuates. The emissions of cleaning fluid
vapors from dry-cleaning establishments are probably relatively constant
during actual working hours. Solvent vapor emissions from the painting
of products in small manufacturing establishments and from solvent de-
greasing operations should have a similar temporal pattern. However,
the emissions of solvent vapors from the painting of buildings (interior
and exterior) are irregular with respect to quantity, time, and the loca-
tion of the activity. Nonmethane hydrocarbons are also emitted in sig-
nificant quantities in summer daylight hours. (See Section VII.)
Specific estimates for the Saint Louis area are needed before that con-
tribution can be properly evaluated.

Again, for the major contributing sources of this type we
recommend the direct approach to determining emissions data described
later.
3. Mobile Sources

The determination of emissions from mobile sources presents
special difficulties. The problem has received considerable attention
in connection with air quality modeling specifically related to the
nature and variability of traffic sources. In studies of this type (as
described in detail in Section VII) a considerable degree of sophistica-
tion has been achieved in some cases. In these however, it has been con-
venient and appropriate to consider the relationships between the behavior
of the sources and the ultimate air quality conditions in models which
integrate and combine the various parameters involved in ways specific to
the purpose for which the model is designed. In many respects, current
modeling capability in this area is inadequate, and its improvement will
be an important aspect of the RAPS program. The basic RAPS Emission
Inventory may properly be expected to provide basic data for the use of
such modeling development, and thus data on traffic flow and traffic
links will be part of the data collected and stored. For the primary
needs of the principle users of RAPS, however, the inventory must provide
data on the emissions of the key pollutants from mobile sources, which
contribute to overall input of emission data to air quality modeling or
transformation process studies. Accordingly, we recommend that, until
more refined inputs are available as a result of progress in mobile
source emissions modeling studies, the basic traffic data be used with
simple models based upon average speed to develop the initial informa-
tion on emissions of the key pollutants.
66
-------
F. Emission or Data Conversion Factors

The derivation of emission estimates from continuously monitored
operating variables is a critical step that will determine the validity
of the high resolution inventory data. Emission factors as currently
used in developing emission inventories are generalized factors that are
most valid when applied to a large number of sources or processes espe-
cially over extended time periods. They have limited applicability for
use with single sources, certainly in the estimation of hourly emissions,
and cannot therefore be used indiscriminately in deriving an inventory
for RAPS in the manner proposed. The whole point of this inventory is
to have high precision in terms of space and time, and for this reason,
the best possible conversion factors must be used to develop the emissions
data from the monitored variables. In general, this means that specific
factors must be developed on an individual basis for each source for
which the critical relationships are not already known with a high degree
of confidence. (Even in the latter cases, some testing and verification
would be desirable.) On the other hand, such specific source testing is
a costly and time-consuming activity to be performed only when necessary.
It is recommended, therefore, that consideration be given to using pub-
lished, established emission factors for each source,* where the type of
equipment or activity is sufficiently typical that the published emission
factors can be identified as being truly representative of the case in
question. It should be noted that, in general, the published factors
have the most credibility for the major combustion sources (particularly
in the case of SOg emissions). It is important to recognize, however,
that in the case of the RAPS inventory, the role of minor sources is of
great importance in local effect, i.e., at near-surface levels, with the
spatial resolution required for RAPS. Lower confidence factors are
assigned to emission factors for such lesser sources (quality factors
B and lower in the categorization of Reference 11). Considerable caution
must therefore be exercised in treating such lesser sources. This is
especially the case for area sources, which, although minor contributors
to the total burden, are important on the local scale. Special care is
necssary to improve the accuracy of emission factors from such sources
for use in refined modeling techniques.
*
For example, Reference 11.
67
-------
G. Physical Aspects of Emission Sources

For purposes of dispersion modeling, it is essential to know not
only the quantity of a pollutant emitted to the atmosphere but also the
manner in which it is emitted. It is necessary to know the effective
height from which the emitted pollutant can be assumed to start dispers-
ing. Therefore, there must be sufficient information from which it will
be possible to estimate the effective height of the plume of gas. The
effective height of the plume is the sum of the physical height of the
stack and of the plume rise, which is the distance that the plume ascends
under the influence of its initial momentum and its buoyancy.

For small source units, such as those that constitute area sources,
the water gases are usually discharged effectively at ground level. Even
where the gases are discharged a short distance above the roof of a build-
ing, they will generally be carried quickly down to ground level. Hence,
it is commonly assumed in dispersion modeling that the waste gases are
discharged at ground level and that neither their initial temperature nor
velocity influences subsequent dispersion.

For point sources, and particularly for the larger ones, the emission
inventory data must be supplemented by data on the gas flow rate, tempera-
ture, pressure, and specific gravity, and on the physical height and
diameter of the stack. Calculations of the plume rise require knowledge
of the gas exit velocity and the stack inside diameter at the plane of
discharge, and of the temperature and specific gravity of the gas. Cal-
culation of the specific gravity of the gas requires knowledge of the
composition of the gas stream including the amount of water vapor.

In addition to the data noted above, the inventory should include
information on the layout of the source stack with respect to other
structures. As a rule of thumb, a stack should be at least 2.5 times
the height of the nearest structure; otherwise, the proximity of the
other structure; may lead to "downwash" of the gas from the stack and a
consequent rapid descent of the plume to the ground. Short stacks pro-
jecting from the roof of a building are particularly subject to this kind
of problem, and the role of the stack in securing dilution and dispersion
of the gas is largely lost.
H. Units of Measurement

Many agencies continue to use English engineering units, and it is
virtually certain that many of the data coming into the inventory will
continue to be expressed in engineering units despite a growing preference
68
-------
for the metric system. It is undesirable to convert such data into
metric units manually before storage in the inventory because of the
large amount of time it would take plus the time required for checking
for conversion errors.

It is recommended that the data be entered into computer storage in
the same units in which they are originally received. The computer should
then be programmed to convert the data to metric units when they are re-
called and printed out. This procedure will diminish the amount of manual
processing that the data must undergo and still retain flexibility.
I. Specification of the RAPS Emission Inventory

1. General

While NEDS as a source of information on annual or long-term
emissions from a broad range of pollutants, it is recommended that a spe-
cial high resolution RAPS Emission Inventory be developed to provide
hourly data emissions of SC^, CO, NOX, and HC in the form of direct state-
ments of weight of pollutant emitted in kilograms each hour from each
point or line source or area element, for the Saint Louis (AQCR). Loca-
tional data will be provided in UTM coordinates with a resolution of at
least 0,1 kilometers for point sources. Mobile sources will be specified
in terms of traffic links (with a resolution of approximately 0.02 kilo-
meters) for major routes and of areas for secondary routes. Both for
stationary and mobile sources, the area elements will reflect either
the spatial resolution of the basic data from which the emissions are
estimated (e.g., housing developments) or one kilometer grid squares,
whichever is the smaller.

Information on the physical characteristics of point sources
and the prime data used in compiling the inventory would also be provided.
Although it is not considered practicable to develop comparable data for
particulates, it is recommended that additional data on particulate
emissions be collected to supplement the NEDS inventory, specifically
data to distinguish between coarse and fine particulate emissions from
the major sources.

For initial planning purposes we propose that only point sources
emitting over 100 tons per year of each pollutant be considered individ-
ually. The estimation of emissions from such sources will be accomplished
in classes according to the magnitude of each source as specified in
detail below. We recommend that emissions data be based upon direct
monitoring of emissions for the largest sources. For other major sources,

69
-------
direct monitoring of hourly fuel consumption operating or process data
should be used as the basis for estimating emissions. For smaller point
sources, we propose that emissions be estimated on the basis of general-
ized patterns of temporal variation of emission as a function of source
type and using the best available data on fuel consumption, operating or
process data (i.e., per shift, daily, weekly). In the limit, this ap-
proach approximates current practice in which annual totals of such data
are used as a basis for modeling. For area sources, we recommend the use
of modifications and adaptions of already available emission models.
Similarly, for a first approach to assessing mobile course emissions, we
recommend the use of available models. Improved mobile source models
should be employed as they become available to provide higher quality
data. A simple new model should be developed to estimate natural emis-
sions of nonmethane hydrocarbons on summer days. In all cases, we pro-
pose that special testing efforts be carried out to ensure that the
emission and conversion factors used to derive the emission estimates
are as reliable as possible and have regard for the high temporal reso-
lution required. Standard emission factors should be employed only if
a review of the source on a case-by-case basis indicates that such a
factor may be used with confidence in the particular case.
2. Point Sources

a . Classification by Size

It is convenient to classify point sources by size accord-
ing to the amount of pollutant emitted annually.

• Class I, over 100,000 tons per year.
• Class II, over 10,000 but less than 100,000 tons
per year.

• Class III, over 5,000 but less than 10,000 tons
per year.
• Class IV, over 1,000 but less than 5,000 tons
per year.

• Class V, over 100 but less than 1,000 tons per
year.

• Class VI, over 1 but less than 100 tons per year.

The numbers of sources comprising these classes for SO , CO, NO , and
2i x
HC and their contribution to the total burden in each case is shown in
70
-------
Table 12 based upon NEDS data that is discussed in greater detail in
Section VI.

As indicated b\ these classifications, a relatively few
sources account for the bulk of the emissions.

The annual emissions, obtained from the NEDS inventory,
although adequate for initial source classification, are not entirely
representative of the probable contribution of the various sources to
air quality at specific time periods. For example, units supplying
steam for space heating will be operated only in periods of cold weather,
and the actual emissions at that time will be higher than indicated by
the annual average. Furthermore, the emission rate will follow a diurnal
cycle even within a single day.

The foregoing considerations are important because reports
from atmospheric dispersion studies have indicated that ground-level pol-
lutant concentrations are disproportionately affected by local low-level
pollutant emissions, even though the latter are individually small and
contribute—individually or even collectively—only a minor percentage
of the total pollutant emission from all sources. Thus, the smaller emis-
sion sources cannot be ignored even as part of the general RAPS inventory
that is intended to supply emission estimates for use in mesoscale atmo-
spheric dispersion models. They must be appropriately evaluated even
though the effort required is more than proportionate to their relative
contribution to the total emission. Within these reservations, the sta-
tistics noted in Table 13 provide a useful basis for assessing the gen-
eral magnitude of the problems involved in developing a program of point
source emission estimates and describing the approach recommended.
b. Data Acquisition Groups

We propose that point source emission data be derived in
four broad groups according to the basic records used to estimate emis-
sions.

Group I. Continuous emissions monitoring records
Group II. Continuous records of fuel consumption,
operating, or process data

Group III. Short term, periodic records of fuel con-
sumption, operating, or process data (per
shift, day, or week)
71
-------
Ol 0-
01 O
r- O
o — o ^
72
-------
w
w
CJ
o
w
O
LO

Cn
O

o
z

5
o
H
.a
co
O
fa
K
O
>
I—I
H
W

CD
p
CO
I-H
3
E
CJ

•ogram
-p
CD
H

tH
O
p
CD
CJ
*H
CD

rH
•H ft
•a E
a cs
hH CO

CO
p
CO
T3

01
01
CD
CJ
o
rf-i
ft

c
bo
b/) C
S -H
•H ^
P O
CO P
^ -H
CD C
ft C
O E

co O
CO t^

^p in
C71 CT1

o
co
in
01
CO
r-H
CJ
in ^ ^H o
•* CM O -H
01 01 /^ O 01 ^-v
CD P m CM p m
o 01 ^ in ^
r< CD >> CD
3 +J 01 r-H +J m
0 P CD P
tO — 1 0! -P r-l 01
CO CD CO CO CD
> 3 P E 3 P
hH *O -H T3
•HO) X H CD
01 > hH O > hH
01 -H ft rl H ft
CO -C £ ft 73 £
•H C CO ft C CO
O hH CO < hH CO
d
0
•H CO
P P
ft CO
E -a
3
01 01
S 01
0 CD
CJ CJ
o
1 — ' ?H
CD ft
3
O
£ be
h< bo c!
CD C -H
hH p -H h-
hH IPO
hH p CO P
£-1 ^4 *H
ft C CD C
3 s: ft c
0 CO 0 E
o

m
CO
in
co

i — i

, — i

1 — i
CO
p
Q
H

CO
•c
W)

C
•rH
-P
01

CD
"H
o
73
-------
Group IV. Long-term (semiannual or annual) records
or estimates of fuel consumption, operating,
or process data.

Within Groups II and III, a further distinction should be made between
sources for which separate conversion or emission factors are established
and those for which representative factors will be established. Data
from Groups I, II, and III will be collected on an ongoing basis.

In principle, it would be desirable to deal with all
sources under Group I. In practice, sources should be treated in the
highest ranking group possible while maintaining regard tor the costs
and difficulties of so doing. A tentative work load is indicated in
Table 13 for S00 and NO sources to show how the scope of effort can be
Z> X
analyzed and assessed. This shows that, by collecting data routinely
from some 101 sources (of which 65 have been individually tested), ap-
proximately 95 percent of the hourly S02 emissions and 70 percent of the
hourly NO emissions can be estimated on a direct basis. Similarly, over
90 percent of the emissions of CO sources can be estimated on an hourly
basis by the treatment of direct data from only six sources. Still, it
would be necessary to obtain data from and test some 10 or more sources
to achieve hourly estimates from direct data for only 50 percent of the
HC emissions.

It is recommended that the method of grouping sources for
this purpose be determined after a survey has been made of all the sources
over 100 tons per year. This survey will produce a better understanding
of the data available and the difficulty of collecting useful direct data
on fuel consumption or operating or process data from which hourly emis-
sions may be derived. Such a survey should preferably be made in site
visits to obtain more details of the source and operations than are avail-
able from the NEDS review. It should ascertain the availability of direct
monitoring data on omissions or fuel consumption and of operating or pro-
ce<-s data that can provide direct information on hourly emissions. Alter—
n..iivt'ly it should reveal the nature of short-term data from which hourly
emission estimates can be derived. The survey should pay particular at-
tention to the form of the available records, and the optimum frequency
and method of collecting them.

The survey will provide a basis for determining how much
special source testing is necessary to provide reliable conversion or
emission factors for individual sources and possible categories of source.
This survey would also reveal the needs and opportunities for installing
appropriate monitoring devices at selected sources, an approach that could
usefully upgrade the inputs from certain sources and reduce data collec-
tion effort.
74
-------
It will be appreciated that the methodology proposed allows
considerable flexibility in the manner in which sources are treated. In
this way, optimum benefit may be derived from the available RAPS inventory
budget on a case-by-case basis. In any limitation of effort for budgetary
reasons, it would be preferable to reduce the numbers of sources treated
individually, rather than to restrict the source testing program, since
operating patterns may be more confidently determinable at the present
time than emissions or conversion factors.
3, Point Combustion Sources

Combustion point sources burning oil and coal are responsible
for most of the SOo emissions in the region, and are also major sources
of NOV. Those burning natural gas produce NO , but do not produce SOr,.
-*• x ^
Neither type of point source is a major source of CO or HC, particularly
when compared with automotive sources, but their emissions of these pol-
lutants are significant and must be considered in the context of RAPS.

The procedure for collecting emission data within the groups
described in Table 13 is described here.

For each source, appropriate reference data, such as type of
unit, nominal ratings, control equipment, and so forth, together with
physical characteristics of the emissions point (i.e., stack height,
diameter at the discharge, location of stack in UTM coordinates) must
be compiled. In addition, critical fuel consumption, operating, or
process data must be acquired in routine collections. These constitute
data as a function of time, ascertained in the initial survey to be basic
for the determination of emissions.

Typical of such data are

• Type of fuel (e.g., coal, oil, natural gas).

• Characteristics of fuel (e.g., sulfur and ash contents,
higher heating value).

• Fuel-firing rate (e.g., weight or volume rate, thermal
rate if available).

• For boiler furnaces, the steam generation rate at
specified temperature and pressure.
• For power boiler furnaces, the steam generation rate
and electric power output.

Then, the following procedures should be applied.

75
-------
Group I; Directly Monitored Emission Sources—Most of the
sources in Class I (over 10,000 tons per year of pollutants)
are expected to be equipped with continuous monitoring devices
for the measurement of SO and NO and possibly with a Pitot
^ x
element to measure flue gas flow continuously. Exit flue gas
temperatures are likely to be recorded continuously at such
large sources. The calibration of the permanently installed
recorders will have to be verified experimentally unless data
on such verification are available at the source. Similarly,
flow measurements have to be obtained or verified. If no per-
manently installed Pitot element is on hand, such flow measure-
ments will have to be correlated with a measured furnace oper-
ating parameter or variable. It may be advantageous to under-
take the determination of other pollutants at the time of
calibration. For example, from flue gas analysis, measurements
of CO and hydrocarbons should be made, if the source is a sig-
nificant (more than 100 tons per year) emitter of these pollu-
tants.

Group II; Directly Monitored Fuel Consumption, Process, or
Operating Data—Typically these are sources emitting less than
10,000 tons per year of SO2 or NOx but more than 5,000 tons per
year and any larger sources not equipped with continuous moni-
toring devices. These units are not expected to have continu-
ous monitoring systems for measurements of pollutant concentra-
tions, but they should have good records of fuel consumption
or other operating data with hourly resolution. In Group HA,
conversion factors are derived linking the continuously moni-
tored parameter with the emission rates are derived from manual
sampling at each source. This is carried out initially in the
program and thereafter repeated intermittently to ensure the
continuing validity of the results. In Group IIB, standard
emission factors may be assigned to categories of similar sources
after they are determined by testing representative stacks.

Group III: Short-Term, Measured Fuel Consumption, Processing,
or Operating Data Available—In this group, in which both Class
IV (1,000 to 5,000 tons per year) and Class V (100 to 1,000
tons per year) sources will be treated, records will be avail-
able on a shift, daily, or weekly basis for fuel consumption,
process, or operating data. (Direct hourly data must be inter-
polated.) Again two possibilities are open. In Group IIIA,
each plant is considered individually, and tests are made to
establish both the time dependence and the amount of pollutant
emissions that can be deduced from the recorded data. Appro-
priate conversion or emission factors are thus assigned to each

76
-------
source tested. In Group IIIB, factors are established for ap-
propriate categories of similar sources on the basis of tests
of representative samples.

Group IV: Long-Term Fuel Consumption, Processing, or Operating
Data—For sources over 100 tons per year, for which suitable
specific data are not available (or which must be treated in
this fashion for budgetary reasons), we recommend the use of
modeling procedures of the type described in Section VII. A
suggested approach for S02 is to employ the Argonne3 model, and
for NOX and HC the SAI model,1'3 but with adaptation and modi-
fication to ensure that the best possible input data on fuel con-
sumption are used and that conversion factors are verified by
testing.
4. Point Noncombustion Sources

Because of the extreme diversity in point noncombustion sources,
no general and detailed procedure for correlating emissions with source
operating variables can be oifered; appropriate procedures will have to
be worked out, case by case, for each type of source, following the gen-
eral principles outlined in Table 13 and in Section IV-E-2.

For the purpose of organizing the inventory, such sources should
also be considered according to the most important pollutant that they
emit. In this grouping, particulate emission^ will not be used as a basis
even though they may be the principal pollutant emitted by some sources.
J. Stationary Area Sources i
|
\
1. General j

Stationary area sources contribute minor proportions of the
total burden of SO0, NO , CO, or HC as described here. Their importance
i—i X
may be significant, however, in the context of RAPS, since they may con-
tribute significantly to the concentrations of the key pollutants at the
near-surface level on a local basis.
2. Area Combustion Sources

The NEDS inventory data indicate broadly that the emission
sources in this category are primarily those using natural gas or dis-
tillate oils. Consequently, the estimated SO^ emissions are minor (only

77
-------
about 1 percent of the estimated total). The relative contribution of
NOX from these sources is greater, but still a minor part of the total.
The NEDS inventory indicates that 28 percent of the total NOV emitted
x
comes from area sources, but these include vehicles that are known to
be, in the aggregate, a large contributor. The contributions ol CO and
hydrocarbons from the stationary area combustion sources are likely to
be minimal.

As discussed, the only practical way to estimate short-term
emission rates from the area sources is the use of simulation techniques
such as are discussed in Section VII. The two most advanced models to
date are the Argonne model2 for S00 and the SAI model1'3 for CO and NO .
* x
They are alone in their class in that they have received significant
field verification. As discussed in Section VII, the models have limited
precision for the estimation of area combustion source emissions for the
RAPS inventory. Accordingly, it is suggested that to improve emissions
estimates by such modeling techniques, more accurate and precise survey
data on fuel consumptions as well as more accurate emission factors be
obtained. Both will require that new experimental studies be made at
typical sources.

Data on fuel consumptions in specific, small areas of cities
are not readily available. Commonly, continuous metering of natural gas
is carried on only at a few points on lines supplying major sections of
the city. The flows metered at these points give no indication of the
breakdown of usage by type of source, and the temporal variations in
flow reflect only the summation of the demand patterns of different
users. The consumption of gas by individual customers can, of course,
be obtained from meter records. However, these will, at best, give only
the monthly totals of gas consumed; they will not show the different uses
by a given customer.

Data on fuel oil consumption by residential and commercial
users are largely restricted to data on fuel oil sales. Even users will
generally have no more than a qualitative idea of the actual rates of
consumption over short periods.

To obtain a high degree of both temporal and spatial resolu-
tion in the estimates of emissions from area sources, it will be neces-
sary to conduct extensive surveys of the patterns and amounts of fuel
usage by representative sources. Sufficient sources must be surveyed
to constitute statistically valid samples. It will probably be necessary
to install meters at these sources that will record both the rates of
fuel consumption at various times and the integrated total.
78
-------
Residential and commercial installations, in particular, will
show marked diurnal variations in their fuel consumption patterns as well
as variations in usage between week days and weekends or holidays. There
will also be seasonal variations in the magnitude and distribution of fuel
consumption. For example, the consumption of fuel for water heating may
be relatively constant throughout the year, but the consumption for space
heating will be highly seasonal. Because space heating will be the most
highly variable as well as the largest consumer of fuel, the rate of fuel,
consumption for this purpose should be correlated with ambient temperature.

Better emission factors must be obtained for the typical fuel-
burning units used in residences and commercial establishments. Some re-
quired data may be derived from other studies currently in progress, but
others will probably have to be made specifically to meet the needs of
the RAPS inventory.

Once emissions are determined for the representative sources,
the emissions from specific areas can be estimated from the number and
sizes of residences and commercial establishments in the areas.
3. Area Noncombustion Sources

The only area noncombustion emission sources that appear to be
significant are those emitting hydrocarbons and hydrocarbon derivatives
(gasoline and solvents). The nature of the sources and their emissions
was already discussed.

The release of solvent vapors from the painting of buildings
probably cannot be incorporated in a practical way into models for pre-
diction of hourly emission rates; the emissions are too irregular in
quantity, time, and location. Probably also, the solvent emissions from
this source are too small to be of concern in mesoscale dispersion studies.
However, a painting operation might be of concern to a microscale disper-
sion study being conducted nearby. If so, making allowances for the emis-
sion is best handled as a project task of the dispersion study.

The area hydrocarbon sources of consistent importance will con-
sist primarily of the smaller gasoline service stations and dry cleaning
establishments. It will be necessary to survey the typical operating
cycles of such establishments to establish the pattern of their emissions.
79
-------
K. Mobile Source Procedures

For automotive mobile sources, we recommend that an average route
speed model with a link or line source geometry supplemented with mea-
sured data of traffic flow (via fixed sensors on high volume freeways
and selected arterials) be used.

The SAI (Systems Applications, Inc.) model1'3 for hourly CO, NO .
X
and HC emissions is recommended. It can have modified inputs derived
from Stanford Research Institute's model4 for spatial and temporal dis-
tribution of vehicle number and speed on a link basis for primary traffic
and area basis for secondary traffic. As a basis for this approach, data
should be assembled on traffic volumn and the location of the primary
traffic links and of the areas of secondary traffic. Additionally, ar-
rangements should be made to install and collect data from fixed sensors
on selected routes, as noted above.

As soon as improved methodology is developed for refining the inputs
to such models, it should be used to update the inventory, and, in due
course, the simple average speed approach could be replaced by more com-
prehensive modeling techniques as they become available.

In any case, prime data relating to route links and such would be
available from the RAPS Emission Inventory records for use in such model
development or other purposes.

For other mobile sources the following models should be used to the
extent possible:

• Geomet model5 for diurnal emissions from river vessels and
railroads.
• Northern Research and Engineering Model,6 as revised by Geomet
(in preparation), for aircraft emissions.
L. Natural Background Emissions

Techniques of estimating the natural emissions of nonmethane hydro-
carbons in summer daytime should be developed so that such emissions may
be added to the inventory in time for studies in which hydrocarbon emis-
sions are considered.
80
-------
M. Particulates

As noted, it is impracticable to develop a high resolution inventory
of particulate emissions on a routine basis. For modeling studies it
would appear that such emissions must be treated on a project basis with
special collections of data (possibly limited in space and time) as re-
quired. In addition, it is suggested that the tracer technique be em-
ployed to investigate the dispersion of participates under various con-
ditions.
N. Scheduling and Scale of Effort

The development of a complete RAPS Emission Inventory is scheduled
to take up to three years.

It would be desirable, however, to aim for an earlier provision of
emissions data, especially for the first critical tests of air quality
and dispersion models. These will use SOg (emphasizing stationary
sources). Accordingly, the development of the inventory of these pol-
lutants should be given priority, particularly in the source testing and
modeling phases. (Since the majority of point sources of SC>2 also emit
NOX, the collection of data on this pollutant can conveniently be carried
on in parallel.)

A possible schedule is shown in Table 14. The aim is to provide
initial data on SO0 and NC) by the first quarter of the second year of
Z> X
full activities and complete data by the end of the third quarter of the
second year. For CO and HC, effective initial data should become avail-
able by the end of the second year and complete data by the third quarter
of that year.
81
-------
Table 14

DEVELOPMENT SCHEDULE
Inventory Operations
Years
Stationary sources

Survey of point sources and
assimilation of NEDS data

Development and adaptation of modeling
procedures

Source sampling

Organization of routine data collection

Routine data collection

Mobile sources

Survey of mobile sources and routes
and assimilation of existing data

Development and adaptation of modeling
procedure

Organization of routine data collection

Routine data collection

Data handling system and system operation

Development of programs for use with
data management system selected

Operation of system

Completed acquisition of data

SO2 and NOx
Initial
Full-seale

CO and HC
Initla 1
Full-scale
Additional effort if interface facilities of basic system need major extension.
82
-------
V. TASK C: EMISSION INVENTORY FILE SYSTEM

A. General

This section concerns the specifications for data formats in the
emission inventory. Because this inventory is intended primarily for
air quality modeling in the RAPS program, the data elements were chosen
to provide input data and supplementary documentation for that purpose.

The data structures specified below are extensible data items in a
data reference system that provides the facilities to manage such data
types. Data systems of this kind have been developed and are available.
We describe in this report the main functional features of the inventory
system, and the computer system interfaces needed to provide data to
modeling programs.

Although most of the expected categories of data are used to
demonstrate the data structures and formats, this chapter is not
intended to be a complete listing of the content of the emission
inventory. Because the data structures are specified to be extensible,
any data not explicitly shown can be attached to the structure.

Section B describes the computer system environment in which we are
assuming this inventory will be implemented. Section C discusses the
main operational aspects of the inventory. Both of these discussions
provide the basis for the design of the data structures described in
detail in Section D. Section E suggests a possible alternative format
for implementing these data structures in the absence of a supporting
data reference system. Section F lists the steps required for
implementation of the inventory system, and Section G is a tabulation of
the principal specifications of the computer programs needed.
83
-------
B. System Environment

In this section, we discuss the key features of the system and
software environment in which the emission inventory will be
implemented. A data management system will be needed for the operation
of the RAPS Emission Inventory, although it may be possible to perform
the inventory operations with systems not having exactly the features
described here,, Data management software is usually the responsibility
of the central computing group of an organization, however, because this
system must meet the needs of the Air Data Branch for the inventory
functions, specifications for the system are given in this report.

Either a general-purpose or a custom-designed data management
software system must be obtained or constructed that will run on the
computer facility at EPA-RTCC. This data management system should
provide the basic functions of file definition; addition, revision, and
deletion of data items; retrieval of selected items using selection
criteria on data content; hierarchic sorting by content; and output of
specified data components of selected items in specified formats to
construct files for use in other programs.

The system should provide for defining hierarchically structured
data items, with components consisting either of single data records or
lists of records, subordinate to other components of the data item. We
assume that the system will provide means for assigning names to each of
the components of a data item. The system should provide functional
records of any desired length to contain character strings.

The system should provide for data entry from punch cards, tapes,
and terminals (if it functions in an on-line or interactive mode). It
should also provide for merging of preliminary files from disk. Data
should be readable as ASCII character strings. Conversion of declared
numeric data to numeric internal representation may be provided, but is
not necessary. The system should not pad card input with blanks and
should provide for continuation of a data record to successive cards
until the record is terminated by an end-of-record character.

For the purpose of this emission inventory, all operations can be
carried out in batch operating mode. However, the RAPS Data Manager has
indicated that the RAPS program will have other data base requirements,
and that compatibility among these different data bases would be opera-
tionally and economically beneficial. One level of compatibility that
can be obtained is to use the same data management software to implement
all of the RAPS data bases. Some of the data base applications may
require on-line use of the data system for retrieval and perhaps even
for data entry. Therefore, it would be desirable for the data manage-
ment system to have the potential capability of being used in an on-line
mode, perhaps by means of additional software options.

All of the presently existing air quality models are written in
Fortran. The primary purpose of the emission inventory file is to
provide input data to these models. Therefore, a system that has an
interface for Fortran programs would be advantageous, because the input
data files could be constructed by means of Fortran programs that could

84
-------
access the inventory files through the system interface and write data
in the formats and data representations required by the models. Since
the inventory file formats include data embedded in character strings,
using various separator and operator characters, these records must be
decomposed by the Fortran interface programs, the numerical values
extracted as character strings, and converted by the interface into
proper internal number representation for arithmetic operations. This
translation of the internal formats will have to be added to any
interface routines that a data management system may already have.

The RAPS Data Manager has stated that EPA-RTCC will obtain the
Univac DMS-1100 data management system when their Univac 1100 computer
is installed. DMS-1100 is described in a catalog as a host language
system: it is based on Cobol and can be utilized only by Cobol programs
that issue calls to the data management system for file operations.
DMS-1100 is not in itself a complete data management system. It must
be implemented by writing Cobol programs to perform the functions of
defining and labeling data structures, entering data, sorting and
indexing, and formatting output data. There is no Fortran interface
to DMS-1100, so additional system programming work would be needed to
provide a facility for generating input files to models if the inventory
is based on DMS-1100. (DMS-2000 is also being acquired.)

At this time we do not have sufficiently detailed information about
the facilities and file structures provided by DMS-1100 to be able to
make any positive recommendation concerning its applicability to the
RAPS Emission Inventory.

It is estimated in Section IV on Task B that the total number of
point sources that should be included in the RAPS inventory will
probably not exceed 1000. The number of area sources will depend on how
the areas are defined. If each area is only 1 square kilometer, then
there may be several thousand area sources to be described. However,
the format described here allows for any size and shape of area source
and also for including sets of equivalent area sources in one item, so
the number of inventory items needed to describe the area sources can be
reduced to a few hundred if this feature is used. The line sources that
need to be described will be mostly major traffic routes, and they
should be relatively few in number.

For the purpose of specifying the capacity requirements of a data
management system for the RAPS inventory, we can probably put an upper
bound of about 2000 data items per file. For some point sources, the
data elements may be fairly detailed, but most source items will
describe only one or two products or pollutants, and only a few typical
patterns of time dependence can be predicted by emission models, so that
most of the data items will contain from 200 to 500 characters for a
total of about one million characters. The longest elements in the file
will be the hourly data on consumption and emission for about 50 point
sources, each of which will require up to 75,000 characters to describe
hourly consumption to 1% precision for a year. Then, if hourly emission
values are also stored for an average of three pollutants for each such
source, the total size of each of these point source items will be about
300,000 characters, or 15 million characters for 50 sources.

85
-------
The master inventory files will be easily accomodated on a single
large disk pack, such as the IBM 3330, or the Univac Fastrand drum,
which holds over 100 million characters per drum. Therefore, the
storage of emission values in the inventory file will not significantly
affect the computer facility requirements.

The accumulation of historical inventory data, over a period of
years, will most conveniently be accomplished by making a separate file
for each calendar year with the possible exception of the first one,
which will have to average several prior years experience. If this
system is eventually extended to maintain emission inventories for other
regions, each Air Quality Control Region (AQCR) should be handled as a
separate set of data files.
86
-------
C. Inventory Operations

To define the data formats for a system file, it is necessary to
have a view of how the file will be used as a part of an overall system
operation. In the case of this emission inventory, there are three
major operational activities: source data preparation and entry,
computation of emission values of sources and production of data file
inputs to air quality model programs.

A principle we have followed in designing these data formats is
that source data should be organized and filed with a minimum of manual
transformations from the form in which it is gathered to the form in
which it is entered into the computer. Thus we have chosen to organize
the file with sources as the primary items, and we are making provision
for entering the names of fuels, pollutants, and other materials in
readable standard nomenclature. Consumption data should be entered in
the units in which it is normally described to avoid manual computations
and errors. The emission modelling programs can compute emission values
in standard system units.

The SAROAD parameter coding manual of EPA has assigned integer code
numbers to many hundreds of compounds of interest in air pollution.
However, only a relatively few pollutants will be included in the high
resolution emission inventory for air quality model verification.
Furthermore, the SAROAD codes do not include materials other than air
pollutants, such as consumables that will be included in the emission
inventory as a basis for computing emissions. Finally, writing 'S02'
instead of '42401' to describe sulfur dioxide is more brief, more
readable, requires no catalog look-up by an inspector or data clerk, and
is far less error prone.

We suggest, in this report, that compounds and other materials be
labeled mnemonically and concisely. Of course, it is essential that the
labels be absolutely consistent. Since the number of materials is
expected to be relatively small, a standard thesaurus of material
identifiers should be easily maintained.

A considerable amount of data is needed to identify and describe
each source in addition to the set of values for emission or consumption
of a variety of possible pollutants or materials and a specification of
the time dependence of those values. The number of different kinds of
sources and different materials indicates that a hierarchic list type of
data structure is most applicable to this purpose.

In connection with model sensitivity studies or studies of control
strategies, it will be desirable to be able to include the effects of
simulated sources, or real sources operating under simulated conditions
or under actual but very unusual conditions. All such sources can be
entered into the Master Inventory in the standard way, but identified by
an appropriate category name so that they can be selectively included
with or substituted for real sources in data files for model runs.

It is expected that most of the data will be entered in a batch
mode via punched cards or using key-to-tape machines. In either case,

87
-------
the hierarchic structure of the data items must be represented in the
character stream by additional delimiter characters. An extended format
for this purpose is described in Section E on extensions for batch data
entry; however, any existing system will have its own format for batch
entry. This format is sufficiently complicated that it is recommended
that the data be written by a skilled data clerk, who can transcribe
original data from an inspector's notes or from data collection forms.
This clerk should also be qualified to proof read data entries both for
syntactic correctness and for data sense. Data preparation forms can be
used to aid in the structuring of the data and to make it easier to use
the correct syntax and punctuation characters. However, it is not
possible to depend only on well-designed data preparation forms for data
quality, because the content as well as the form of the data must be
verified.

A data management system generally will check its own data entry
syntax, and should be used for this purpose. However, the internal
formats of some of the data elements, such as consumption data, will be
transparent to the system and will need to be tested by a separate
emission inventory verification program. The same program should also
verify whether material names are included in the material name
thesaurus. This thesaurus therefore should be kept in the form of a
computer reference file available to the data entry program as well as
in paper reference copies in the data preparation rooms. For example,
the substitution of the digit-10' for the letter-'O1 in S02 by a typist
can be detected this way. The same kind of verification should be
applied to all other data descriptors for which there is only a small
number of correct choices. The verification program should make output
listings of all items in which it detects errors in a form in which the
erroneous data component or syntax is identified. Although the
verification program could include facilities for user-entered
corrections, it will be more practical to use the editing facilities of
the data management system for this purpose, and this is recommended as
the preferred approach.

Data should be entered first into a preliminary data system file,
so that they can more easily be further verified by printing out the
entire content of a small data set in an expanded easily readable format
for proof reading and further correction prior to adding these items to
the main inventory file. Another reason for this preliminary stage is
that most data management systems perform a significant amount of
behind-the-scenes processing on new entries into a directly accessible
file, such cS ".riding address tables, hash coding computations, making
inverted lit_» GJ. key data elements. The small preliminary file, on the
other hand, can be a simple sequential file with none of this extra
processing until after the data manager has determined that it is
accurate enough to merge it into the main file.

The retrieval facility of the data management system will be used
for the selection of groups of sources for which emission values can be
computed by the same tuodel or are selected for production of a
particular data file for model validation or for editing and updating of
selected items. The retrieval of source items should be accomplished on
the basis of any appropriate data in the item, such as source category,

88
-------
pollutant, location, or status of previous computation of emissions.

The retrieval facility can also be used to produce special files
organized in terms of any user-specified parameters, such as a
pollutant, sources using fuel at greater than some rate, location. This
can provide for a variety of special applications without requiring
additional file programming. A sorting capability in the data
management system should be provided to order and index these files by
their content.

In very few cases will source data include measured values of
emission. For most actual sources, the emission must be calculated from
consumption or other source data. Formats are provided in this
inventory for both emission data and consumption data, using the same
data syntax and structure for both. For those sources for which there
are no direct emission data, emission must be calculated using emission
factors or emission models applied to consumption or activity data. The
emission factors have been compiled by EPA into a table as a function of
the Source Classification Codes (SCC) for a wide variety of industrial
processes and consumables.

After a set of new data source items has been entered, the next
logical operation is to calculate the emission values for those sources
that have no pollutant data entered. The results of these computations
can be appended to each source item as the content of a pollutant
component using the same time interval description for each source as
was used for consumption if only the emission factor is applied. If a
specific emission model is used, the time interval description also must
be generated by the model program.

The emission inventory software system will require emission model
programs that can access the consumption data element for each item,
refer to the SCC code for that material, look up the appropriate
emission factor in an auxiliary Emission Factor File maintained as a
direct access file on the system, and compute the emission values for
each of a prespecified set of pollutants. If the programs have access
to the data system for file v,riling as well as reading, they can then
update each item by adding a pollutant list component with an entry for
each pollutant to be calculated. One of these will be a "no model"
program that will only multiply consumption data by emission factors to
obtain emission values. These emission models will generally also
require input of ambient atmospheric data, which can be obtained from
meteorolr.ylc,:J dr^a files based on the RAPS data network. These files
are not considered in this report, because they are not part of the
Emission Inventory system.

In this manner, the master emission inventory file will serve both
as a file of input source data and also as the repository for derived
emission values with no duplication of source information.

For the principal objective of the RAPS program the final function
is that of production of input data files for air quality models. In
general, the required emissior d?.*a will be that for a specified
pollutant for a specified time Interval or series of time intervals and

89
-------
also for specified positional coordinates. Many data management systems
have search and retrieval features for selecting data items that satisfy
specified data constraints or specifications. Some preliminary
selections may be made using the system facilities, such as selection of
a particular type of source, certain UTM coordinate ranges, and the
like. However, the time dependence data are coded in the emission data
element format in a special way, and a general system feature will not
be able to decode it.

If the general data management system includes the capability for
calling procedural language subroutines, such a subroutine can be
written to decode the time and emission data in the emission data
element and can be linked to the data system so that it will
automatically be invoked whenever the system is asked to search for
emission data within some particular time range. If that capability is
not available, then the data file production program must have a
subroutine included within it to perform the time interval selection
function.

After the items containing the desired data have been identified,
it is necessary to abstract the emission data for the particular
pollutant and write an input data file for the model program. The input
file must be written in exactly the format expected by the model; the
values must be in the units defined for that model program. Since the
model programs are written in Fortran, the data file must also be
written with the appropriate integer or floating point representation
for each quantity.

The generation of these input data files will require interface
programs designed to perform this task. These programs (or maybe one
general-purpose file-building program) must read the emission data
elements that are character-string data, decode them, compute new time
sequences if the model requires them, transform the string
representation of the emission value to internal numeric form, read the
name of the unit: used and multiply the values by the. proper conversion
factor, and write the file for the model.

Most of the air quality models operate on the basis of allocation
of emission values to a fixed geographical grid system. The data
interface programs will also have the function of translating the
geographical coordinates of point, area, and line sources in the
inventory j.-.tr the appropriate grid units for the model.

The dat-L file derived from the RAPS Emission Inventory can be
combined with similar data files derived from other source inventories,
such as NEDS, to obtain the most economical use of existing systems. It
is assumed that programs to generate input data files for atmospheric
models already exist in the NEDS system. If such programs do not exist,
then it may be more economical to transfer all the NEDS data relating to
the RAPS area into this RAPS inventory system.

The only sources that may be duplicated iu both systems should be
those for which time-resolved dat-a -in be obtained. Since there are
expected to be not more than one or two hundred such sources, it will

90
-------
not be efficient to provide a program for computer translation of NEDS
to RAPS data. It will be cheaper to manually copy these source data,
especially since much of the data for each source will be new. If,
eventually, the use of the RAPS inventory is extended to cover much more
data from other areas, then the possibility of computer transfer of data
between the two systems should be re-examined.

Figure 3 shows a schematic of the inventory system.
91
-------
INPUT AND
OUTPUT
PROGRAMS
DATA FILES
REFERENCE FILES

ORIGINAL \y
SOURCE DATA /

ERROR /
LISTING N^

ERROR /
MESSAGES >?

RETRIEVAL \
SPECIFICATIONS/"

EDITING \
CORRECTIONS/

SELECTED /
DATA LISTINGS*^

MODEL \
PARAMETERS /

DATA
ENTRY

1 r

DATA
MANAGEMENT
SYSTEM
PROGRAMS'
MERGE,
EDIT,
SELECT,
DELETE
SORT,
OUTPUT

EMISSION
MODELS:
"NO MODEL"
"£" .. "-7"

^ ^
^^

PRELIM.
cni IRPF
DATA

MASTER
EMISSION
INVENTORY

SELECTED
SUBFILES

( ,

THESAURUS OF
IDENTIFIERS

EMISSION FACTORS
AND SCC CODES

DATA FORMAT
SPECIFICATIONS
i
DATA
r
FILE
GENERATORS

INPUT DATA FILES
FOR AIR QUALITY
MODELS
SA-2579-8
FIGURE 3 SCHEMATIC OF INVENTORY SYSTEM
-------
D. Data Structures and Formats

1. Source Categories

The three principal categories of emission sources are point, line,
and area. Because they require sufficiently different identification
parameters, they should be filed by using different item data
structures. They can be stored in separate computer files or in the
same file in different item categories.

Point sources can be identified by the use of a group of
identification parameters selected from those given in the manual for
the NEDS inventory. Although this is a RAPS emission inventory, some of
the NEDS identifiers are included primarily to facilitate data
interchange between the two systems. The primary identifiers should be:

* Assigned source item number (sequence number assigned by
inventory system);
* Source category (e.g., power station, foundry, refinery);
* Source name (name of company, organization, or generic name
if it has no specific name);
* Plant No., Point No.;
* Street address (or nearest street identification);
* City, County, State, AQCR No.

The UTM coordinates are not included in the source identifiers of
point sources, because each stack of a multistack source will have its
UTM coordinate position in the data item. If some stacks at a plant
have different emission values or time sequences than others, more than
one data item will be needed to inventory that plant. The Point No. is
intended to identify such different sources within a plant.

The NEDS Source Category Code (SCC) is included as part of the
identification of each consumed or manufactured product, rather than as
a single-valued source identifier, because a source may utilize more
than one product. The code includes both the material and process
identification.

Location, source category, and sequence number identify area
sources. Predetermined area sizes or shapes need not be allocated. Any
closed simple area bounded by straight lines can be uniquely specified
by the coordinates of the intersections of these lines taken in order.
Thus the areas can be selected on the basis of convenience and
appropriate internal contents without size or shape restrictions. Area
sources can overlap; for example, an area identified for automotive
emissions may include or overlap an apartment house area. The
recommended identifiers for area sources are:

* Item number;
* Source category (e.g., automobiles, residential, dump);
* Area number (assigned sequence number);
* List of UTM coordinates of corners of boundary (starting
with most westerly or southwest corner and going clockwise);
* City, County, State, AQCR No.;

93
-------
* English language paragraph for general descriptive purposes.
This should always include a street or locality description
to supplement the UTM coordinate specification.

Identification of line sources will be similar to that for area
sources except for the significance of the UTM coordinate list. The
sequence of the UTM coordinates of the end points of a connected
sequence of straight lines approximating the position of the source will
locate each line source.

2. Data Item Formats
The items of the file will consist of point sources, area sources,
or line sources.

Many sources will have fuel-burning data or process rate data
instead of direct emission data. For some plants, the materials being
produced may be significant sources of air pollutants, and these
products can be included in the process category. Or, if data are
lacking on either, an activity may have to be identified to which an
average emission model can be applied (for example, automobile traffic).
An activity might be included for some point sources (like a freight
terminal), if the fuel or emission rates cannot be measured directly.
Thus, three different components of the source data that can be used in
combination for some sources are:

* Emission data
* Consumption and process data
* Activities.

In the case of consumables, the data element would describe rates
of consumption. For activities, the possibilities are so varied that no
single standard format for descriptive data is likely to be usable. One
example of an activity, automobile traffic, is discussed in the line
source category.

Except where direct emission data are available for a source, the
emission can be calculated from the consumption or activity data by
application of emission factors and emission models, and the result
entered into that source item in the inventory file as calculated
emission data.

a, roiut Sources

Any point source may emit one or more of several pollutants of
interest. If data capacity is provided in each item for all possible
pollutants, the file structure will be inefficiently used (mostly
empty). A more appropriate structure would use list elements
implemented such that each pollutant or consumable actually included in
the item list is associated with a pointer to an associated data
element, each of which is a logical record of variable length. This
will permit efficient use of computer storage and efficient retrieval
and selection of data for construction of model input files.

94
-------
In some data reference systems, the system automatically maintains
the pointers on the basis of the data linkages specified during data
entry. If the system does not provide internal pointers to associated
data elements, then explicit addresses (such as record number) must be
written in the file, and the user must maintain the necessary address
directory.

Figure 4 illustrates a sample point source item.
SOURCE IDENTIFIERS
ITEM NO., CATEG.,
NAME
PLANT NO.,
POINT NO.,
ADDRESS,
CITY, CO., STATE
AQCR NO.
STACK
1
POINTERS
CONSUMPTION
COAL
SCC CODE
POINTER
OIL
SCC CODE
POINTER
1
UTM COORD.
HEIGHT
-------
b. Area Sources

Because the format of an area or a line source will have to provide
for an indefinite number of UTM coordinates, these are placed in a
separate data element. The area and line sources are defined in terms
of a single principal pollutant; thus only that pollutant is identified,
instead of a list of them as in a point source item. The emission of
that pollutant will usually have to be computed from consumption data by
means of standard emission factors, so the consumption data should also
be included in each area source description in the inventory.

Figure 5 shows a sample area source item.

H1/V1, H2/V
SOURCE IDENTIFIERS
ITEM NO., AREA NO.
CATEGORY
ETC.

H1/V1, H2/V
0
2'

-
*
UNIT NAME | UNIT FACTOR
UTM COORD.
2
POINTERS
I

DESCR. COMMENT
LOCATION
APPROX AREA

EMISSION DATA | METHOD
POLLUTANT
NAME
POINTER

CONF. LEV. •*— J
STACK
HT.
FLOW
TEMP.

SA-2579-10
FIGURE 5 SAMPLE AREA SOURCE ITEM
This example shows how two separate areas can be included in one
area source item. An area source item can describe a set of
non-contiguous areas to avoid duplication of the complete description of
each of a set of area sources having identical emission characteristics.
The number in the UTM Coordinate component specifies the number of
different areas described in the item. For each area, there will be a
separate data element containing its coordinates.

The Stack data in an area source item refers to an equivalent stack
for purposes of dispersion modeling. The stack area is assumed to be
the whole source area; volumetric flow rate is estimated from fuel or
other consumption data. The. consumption data component is not shown in
this St-.'^p1-" it is strii^tut^d the same as the consumption component
shown in the _-.t a^jice sample.

Only a ,dry slight change in format is needed to use a single area
source item to describe the emission of more than one pollutant, as in
point sources. Therefore, although we are here suggesting a single
pollutant per area source item, that choice can be made by the emission
inventory manager at the time that the areas are defined.
-------
c. Line Sources

An important type of line source is automobile traffic. The format
of line source items will differ from area sources primarily in the
significance of the UTM coordinate list that will be the endpoints of
straight-line segments approximating the road route. The category name
"Road" can identify this type of source. The activity name can be
"Automobile", and additional elements should be appended to the item
under the names:

* Traffic (average vehicle flow, average speed, vehicle mix)
* Way (type, dimensions).

3. Other Data
File data describing the composition of fuels or other consumed
materials may be useful in the inventory, if they are variable for
different sources. A data element for composition can be attached to
any entry in the consumption list by a second pointer. The pointer
identification is: Pointer 1 to the consumption data element, Pointer 2
to the composition data element (if any).

The composition element can be a paired list of impurity element
symbol or compound formula and weight fraction for as many constituents
as are signifacant. The element or formula name can be terminated by a
blank, and each pair can be separated by a slash. A sample is shown
below.

S .02 / As .005
Another category of data that may be needed in some source items is
certain special emission factors, such as those for nitrogen oxides, for
sources where there are actual measurements of the factors. The
measured emission factors can be added to the data structure by a third
pointer under the appropriate consumable. The emission factor data
element will have to identify each pollutant calculated by each factor.
The data can be entered in the same format as that for composition,
above, except that the identifiers will be pollutants and the numerical
values will be emission factors.

Other source data may be added to any item by use of a labeled
pointer to an extension record. Any number of such extension records
may be attached to any item, but no additional storage space need be
allocated to items that do not include such data.

4. Data Element Formats
Most of the data elements will be numeric or text constants and
will require very little formal formatting. Real numbers must be
written in standard notations such as decimal or exponential (ie. 94.27,
1.03 E-4). Text elements can be written free-form. It is often
convenient to have one printing character such as "#" or "$" defined as
a text element terminator. A continuation character can also be defined

97
-------
to enable the entry of multiline text elements. On some systems, the
linefeed key is most effective for this purpose, because it initiates a
new line at the terminal.

The emission and consumption data elements, on the other hand, can
have a large variation in amount of data and should allow for different
levels of time resolution. These therefore require format rules to
provide exact identification of descriptors and data.

Since the data elements for consumption data may deal with a large
variety of consumable products, a single unit of measure (e.g.,
grams/second) may not always be appropriate. Therefore, the first
component of the data element is the name of the unit of measure and the
second is the conversion factor to convert that unit to the standard
inventory system unit.

For those sources for which a series of time-sequence measurements
are available, it will be useful to describe the emission or consumption
by those measurements. On the other hand, there will be many sources,
especially area sources, for which only approximate estimates can be
made, usually on the basis of some repetitive patterns over time. The
format rules described below can be used to describe either a
consecutive time sequence or repetitive time patterns or any convenient
combinations of both. This component of the data element may be very
long (up to 75,000 characters), but it will still be one logical record.

Each emission or consumption data element also includes a component
for Method, for Confidence Level, and an optional component for any
further comments that may be considered useful to add for documentation
of the data. Method will generally be described by means of a concise
standard descriptor or abbreviation of the name of the method used to
obtain the data, For example, "EF'1 can identify emission values
calculated from consumption data by application of the standard Emission
Factor table. Confidence level should be an estimate of the probable
error of the values given, based on a knowledge of the errors of the
method of measurement, or the reliability of the source data.

The emission element format is the following:

Data Element = Unit name + Unit factor + Data Part + Method +
Conf . Level 4 Comment(optional)
Da;. n""t = Time P^t'Data Part or Value
Ti.;.f- ?- - -- . ic...- Code:Time Range or (Time Part, Time Part, ...)
Time K.c.,
-------
Operator characters are:

: means a 'range/data1 sequence follows;
/ means a datum or a data set follows;
, Is a separator between components of a simple data set;
; is a separator between data sets of a multiple data set;
- is a connector between first and last time identifiers
(Inclusive) of a time range;
( ) enclose a multiple group of time pattern descriptions.

Some examples are:

M: 1/1.4,2/1.7,3/1.9, 12/1.3; is a set of values for each of
the 12 months.
H: 1/9.7,2/10.3, 24/8.2; is a set of 24 hourly values, for some
"average" day.
(M:ll-3/D:7-l/H:21-7, M:6-7/D:l/H:22-6)/1.4; specifies a repeating
pattern of 8 pm to 7 am on Saturdays and Sundays during November
through March, and 9 pm to 6 am on Sundays during June and July,
with an emission value of 1.4 units during each of these
intervals.

This looks complicated, but the notation is self-descriptive once
the user becomes familiar with the meaning of the punctuation symbols.
It satisfies two important criteria; it can concisely describe a large
variety of possible time-dependent data, and it can be read by machine.

There appears to be no way to design a preprinted, "fill in the
blanks", form for data to be entered in this format. The reason is that
the format is designed to provide a concise method of describing any
sequence or repetitive pattern of time-dependence using several
different time units, so that the syntax must be dynamic as in a
language, rather than a static syntax that can be preconstructed .

For each time interval code, the total period covered, and the
notation is shown in Table 9.

Table 9
Time Interval Codes and Ranges
Time Interval Code
Annual

Monthly
Weekly
Daily

Hourly
M:
W:
D:

H:
Period

range in calendar years to which the data
applies (e.g., 69-73)
1 year 1-12 values
1 year 1-53 values
1 week 1-7 values;
4=WED, 5=THU,
1 day 1-24 values
1=SUN, 2=MON, 3=TUE,
6=FRI, 7=SAT
All the time units refer to inclusive intervals (i.e., M:l-2 means
January and February). Counted hours, not clock time, is used (i.e.,
H:7-8 means from 6 am to 8 am). Zero (0) is avoided because it leads to
confusion in data entry and because it is not common usage. Since the
99
-------
week is defined to be Sunday through Saturday, a year will generally
begin and end with a partial week, so that there will always be 53
identifiable weeks in every year.

The emission data should correspond to the average emission during
the specified interval, for a current example of the period.
Multi-nested intervals can be used where sufficient data exist and the
emission changes considerably over the larger time intervals. For
example, hourly emission data for an average day during each of several
months may be useful, if the emission rates change considerably through
the year.

The large choice of time intervals permits the entry of detailed
time-dependence of emission if the information is available and less
detailed averages when that is all that is known. If data are available
only for portions of a data set, only those elements of the set need be
identified and entered in the inventory because the format includes a
numeric identifier for every time interval for which data are entered.

Since M: and W: interval sets both cover a period of one year,
they are mutually exclusive. The D: and H: intervals can be used with
any one of the first two in a multiple set, and D: and H: can also be
combined. By the use of these formats, each emission data element can
contain as few as one value, averaged over a typical year, or a value
for every hour of a year.
Although it uses a concise notation, this format is sufficiently
redundant to permit recognition of many entry errors. Manual
proofreading will detect many common typing errors, and the use of the
computer for checking the intern':! ; insistency of data and the syntactic
correctness of the data elements will also be feasible.

5. Summar}*- Listing

The following listing summarizes the form and arrangement of
identifiers in a source item. Generally, most of the entries will be
data relative to higher level identifiers and an identifier for lower
level data. They will only be entered into an item if they represent
actual information so that only a small fraction of the entries in this
table would occur in any r'r.gle inventory source item. Many of the
entries in •'" - table (e.,\,t names of pollutants) represent only a few
examples <_ £ u: ireitea set of possibilities and are included to
display their arrangement and nrtational style.
100
-------
Point
Identifiers
NO. Item number. For file maintenance.
Category Type of source
Name Ordinary name of facility
Street address
City, County, State, AQCR No.
Pollutants
S02 Chemical formula or abbreviated name
Unit Emission rate data Method C.L. Comment
HC
Unit Emission rate data Method C.L.
Consumption
Coal SCC code
Unit Consumption rate data Method C.L.
Activities
Trucks
Count/day
Helicopters
Count/day Comment
Stacks Count
UTM coord. Height Exit area Exit velocity Temp.
Area
Identifiers
NO.
Category
Area Number Assigned for map reference
City, County, State, AQCR No.
UTM Coord Area count
Coord list
Description of area
Pollutant Heat
Unit Emission rate data Method C.L.
Consumption Fuel Gas SCC code
Unit Consumption rate data Method C.L.
Activity
Autos
Flow Speed Veh. mix
Population
Persons Housing units Hous. mix
Line
(See Area—elements are essentially identical)
101
-------
E. Format Extensions for Batch Data Entry

To prepare the data items off-line for batch entry, it is necessary
to provide additional format conventions to indicate the logical
boundaries of data components and the hierarchic level structure of each
data item. If an existing data management system is obtained, it will
have its own formats for this purpose.

If a data management system is expected to be used to maintain the
RAPS Emission Inventory, but the scheduling of data processing requires
that data be prepared for machine entry before the entry formats for
that system are available, then this format can be used to enter the
data onto machine-readable media, and the form,.* -inverted later to
those required by the system. The special-pi^"" ~ .inversion program
will be a temporary program use; ,'.: ' once t> , ; :',-" data
previously entered. After that, a;. I new data 'i prepared directly
in the system format.

This extended format can a^so be used for the purpose of making
sequential files for temporary use without the use or support of a
hierarchic data management system. For this lain. •. > purpose, it is also
useful to include a code to show whether a given J.ata it en is a current
one in the file or has been replaced by an updated version, because the
actual deletion of items and compaction of the i il< ^ill require
processing by a file maintenance program.

The characters chosen as delimiters for this extended format have
not been used for the internal data formatting. These data structures,
together with all format defining ,-naracters, can be implemented within
the conventional 64-character ASCI! subset thai is available on most
keyboard data entry devices and pij.ut.ers:

[n Left bracket, followed by an integer, indicates
opening of a hierarchic data level, n = data level

n] Close data level n, and all inner levels.

< > Delimiters of datu element nanes

# Pound sign -• >*-j.td. c lament separ^i.-n .

Each - ' •-'at-'' itt-i:i •;> . , i t-c, n •> ith [1. • i id,, ate the status of
the item, tb in;. ;et I cat> ' t-. - linked. [1 means mat the item is a
current one jn the file; ' I;.LV,, i Vat it has been superseded. A
partial example of a data j i t-m > . •»» che niannei of u^e of these format
controls:

[1 [2 # 234f/ MFG?; UNIVERSAL HORSE-COi l.ARS# 999 WHINNY ST.,
ST. LOUIS, M0.# 70# 2| |2 -..POLLUTANTS: # S02 I i/' G/S # M:l/1.4,
2/1.7, 3/1.9 12/1.3* J| # HC [3/^ KG/HR « A:69-72/19.5# 2] [2
# 1 C3# 321.0,963. 1?> ':','=' 0.83/* '3V vji,,- H

The
denotes this as a paii.^ : (not no. -u .: unless all types of
sources are included on one fi.it). ": lie denotar.;.o£i :J is the

102
-------
abbreviated name for Source Identifiers, a level 2 data structure.
The denotation is the name of the next level 2 structure,
but it contains two level 3 data sets. The terms S02 and EC are not
names of data structures, they are data, but a level 3 data structure
depends from each of them. is the name of another level 2 data
component, containing the parameters of each stack.
103
-------
F. Implementation

The next steps in the implementation of this emission inventory are
the following:

* Specify the set of point sources, area sources, and line sources
in the St. Louis metropolitan area that will be used as the
initial basis for the RAPS emission inventory.

* Assemble the data applicable to these sources from other
inventories and from other data sources.

* If a general-purpose data management svstem is not available,
then write a system of programs for fil, <.**.-finition,
data entry, and updating of the emi.ss:

* Specify data t-ntr> : '«:.,. - rr :- 3 based joint
described in this < . |-;v. , •'. the capa'i.i:
Any conflict.-, must ;• • ';ol\-id r>v ci.ai.- u1.
or the choice of p' • c (,<.-)> : - ;i ra^l-. i •

* Enter a small set of :at-: .-.» verify \ • \
structure and the furiat-, r» the datu In,,.
ui t lie formats
) i, f t he system.
<^ f riuat structures
; e o e f- s a r y.

'-s of the
* Enter and proof-,:; ; ,

* Write and tei.r tr t
- Computing emibsu r.
- Accessing data '.t^
of time, omisfs. • ..
- Const rue ti.ig .1 ^..i.
i n- mi i
. f i um uo:
he ousi -.
-T SOUl-
* Produce input data files i\,r models u: '.i
procedures or separate cGrn;>iier progrj..ic.
-1-•" vlata;
^ t - . , i i:d ranges
(.at ions;
p •. • . models.

- tther data system
I (14
-------
G. Specifications for Computer Programs

1. Data Management System

These systems usually consist of a group of interconnected
programs. This specification concerns only the external features
related to the operation of the RAPS Emission Inventory.

* Data bases of 20 million character size, containing up to 2000
items, each including 10 to 100 data elements, in up to 4
hierarchic levels.
* Batch data entry from punch cards, magnetic tapes, or disk
files.
* Capability for defining hierarchic data structures, including
list components.
* Symbolic naming of data components.
* Unrestricted lengths of lists and records.
* Retrieval of items by content or partial content of any data
component, including logical combinations of retrieval
specifications.
* Editing and updating individual items or groups of items.
* Interface for external programs to access data items and data
components in the files.
* (Optional) Capability of calling special subroutines within the
data system for operations requiring recognition of specialized
formats within a data component.
* (Optional) Construction of data files in specified formats for
use by external programs.
* (Optional) Sorting and indexing selected subfiles by content.
2. Input and Output Subroutines for Emission Model Programs

These are intended to be standardized subroutines, callable by any
emission model program to provide access to the emission inventory file.

The input subroutine must:
- Access source item via file address;
- Read consumption data component from source item;
- Extract SCC code, unit, time ranges, and consumption
values;
- Read emission factor for SCC code from Emission Factor
File;
Convert values to the units required by the model.

The output subroutine must:
- Generate calculated emission data element for the pollutant
and for the time ranges defined by the model, in the format
required by the inventory;
- Append the emission element to the same source item from
which the consumption data were read.
105
-------
3. Verification Program

This can operate in three different ways. It can be part of a data
entry program; it can be a separate program to correct preliminary data
files prior to merging into the Master Inventory; or it can operate
within the data management system on data sets retrieved by the system.
It should perform these tasks:

* Verification of syntax of formatted data components for
logically correct usage of operator and delimiter characters.
* Comparison of material names with those in a thesaurus file to
verify that every name is one of an allowed set.
* Printing listings of erroneous items with bad elements marked.
4. Data File Generator Program

This capability may be included in a complete data management
system. It should do the following operations as a minimum, although
more extensive formatting capabilities may be useful.

* Accept user or file input of specifications for the exact
sequence of data values and data representations required by an
external program.
* Take its input from a subset of the emission inventory file
selected by the data system and extract data from data
components (the input subroutine for the emission model can be
used for this purpose).
* Convert coordinate locations to the grid identification code
used by the model.
* Interpolate emission values, if necessary, to the time intervals
required by the model.
* Write the output file on the disk, magnetic tape, or punch
cards, as specified.
106
-------
VI TASK D: SURVEY AND EVALUATION OF EXISTING EMISSION INVENTORY DATA
A. Introduction

Air pollution studies have been conducted in the Saint Louis area
for many years, and several emission inventories have been developed.
To avoid any unnecessary duplication of effort, it was important to
review the existing emission inventories, to compare their content with
the information needed for RAPS, and, if they appeared to contain useful
information, to transfer this information to the RAPS inventory.

The City of Saint Louis has been involved in air pollution control
since its first antismoke ordinance was passed in 1893. Smoke and fumes
were, of course, the first targets of pollution control efforts; even so,
50 or 60 years passed before effective pollution control measures were
applied to the largest pollution sources of the Saint Louis area, the
power plants and the steel mills. In 1964, an "interstate Air Pollution
Study, Saint Louis-East Saint Louis Metropolitan Areas" was undertaken
by the U.S. Public Health Service. Questionnaires were sent out to de-
termine fuel use and combustible waste disposal practices in the area as
well as manufacturing activities. A revised emission inventory, still
based on 1963 data, was published in December 1966 as Phase II of the
interstate study.

After the Metropolitan Saint Louis Interstate Air Quality Control
Region had been established, the first comprehensive inventory was taken
in 1968, to serve as a basis for the Implementation Planning Program (IPP)
This inventory included the jurisdictions of Jefferson, Saint Charles,
and Saint Louis Counties and Saint Louis City in Missouri; and Madison,
Monroe, and Saint Clair Counties in Illinois. Franklin County was added
in 1969. Four additional Illinois counties, Bond, Clinton, Randolph,
and Washington, ,vere added to the air quality control region in 1971.
Only sulfur dioxide (SO?) ond particulates were included in the initial
IPP inventory. The more recent inventories contain estimates of carbon
monoxide (CO), hydrocarbons, and oxides of nitrogen (NO ) as well. The
X
following inventories are described in some detail in Section II:

• IPP Emission Inventory-1968

• IBM Emission Inventory-1970

• DAQED Emission Inventory-1971

107
-------
• NATO Emission Inventory-1971

• NEDS Emission Inventory-1973

In addition, the following traffic and transportation inventories
are discussed :

• Streets and highways

• Railways and vessels

The emission inventories in current use by the Missouri and Illinois
regulatory agencies were recently (Summer 1973) acquired and transferred
to the NEDS files under another contract; therefore, they are not con-
sidered separately.

Contact was also established with research groups, such as the
Metromex consortium, active in the Saint Louis area. However, these
groups do not have any emission inventory data. T*'or were any data avail-
able from industrial sources, such as the Waste font]ol Council.

The data available from the various air pollution control organiza-
tions deal with pollution arising from industrial, commercial, residential
and traffic activities. There appear to be no direct emission data on
such transient sources as agricultural spraying or fugitive dust.
B. Existing Emission Inventories lor the Metropolitan Saint Louis
Interstate Air Quality Control Region*

1. IPP Inventory-1968

The Implementation Planning Program (II'P) emission inventory
covering sulfur dioxide and parti dilate matter in the six counties and
the City of Saint Louis originally making up the Air Quality Control
Region was assembled bv the EPA Region VII office in 1968. This inven-
tory VDP used by Argo, National Laboratory as a basis for diffusion
modeling is. the .u>\ <;lopinent of the IPP for Illinois and Missouri.

The 1968 inventory lists yearly emissions of SO and particulates
^j
from 116 major spot sources and 316 area sources, on the basis of ques-
tionnaires and letters of intent filled out by local agencies. Typical
printout pages are shown in Figures A-l and A-2 in Appendix A.
*
The pioneering interstate study by U.S. P,li,o. it> mentioned above on
page 107.
108
-------
2. IBM Inventory-1970

In 1970 the International Business Machine Company (IBM) compiled
an emission inventory for the Office of Air Programs. This inventory
included only three Missouri counties (Jefferson, Saint Charles, and
Saint Louis) and Saint Louis City; therefore, it is not directly compar-
able with other inventories in the area.

Though it covers a smaller area, the IBM inventory is consider-
ably more detailed than the IPP inventory. It contains data on 105 point
sources, giving their location; stack parameters; fuel consumption data;
and calculated S00, particulate, CO, hydrocarbon, and NO emissions in
^ x
tons per year (Figure A-3, Appendix A). From these data, emission
inventories were calculated for each of the political jurisdictions.

In the IBM inventory, area sources were treated by counties.
Each county was divided into source areas, varying in size from 4 to
700 square kilometers. For each area, emissions of particulates, S02,
CO, hydrocarbons, and oxides of nitrogen were calculated for domestic,
commercial, and industrial sources, as well as for incinerators and motor
vehicles. A summary of the data is shown in Figure A-4, Appendix A.
3. DAQED Emission Inventory-1971

The former Division of Air Quality and Emission Data (DAQED)
acquired an emission inventory for the Saint Louis area by using the
rapid survey technique described by Ozolins and Smith.13 This technique
is based on information available in most urban areas; it does not entail
extensive surveys or sampling procedures. The results are, however,
considerably more accurate than those derived by gross approximations
based on published production or consumption data.

The inventory included SO?, particulate matter, CO, hydrocarbons,
and oxides of nitrogen. The data are given in metric units. Summary
information on population, area, some background information, and air
quality data, IF shown in Tables 15 and 16.
4. NATO Emission Inventory-1971

The North Atlantic Treaty Organization (NATO) emission inventory
for the Metropolitan Saint Louis Interstate Air Quality Region was pre-
pared in 1969 to serve as a comparison and model for the development of
emission inventories in Frankfurt, Germany, and Ankara, Turkey. The so-
called NATO Emission Inventory covered the following pollutants:

109
-------

CJ
r;
1 -f
C
O rH
o d
o
0 C
J^
-fj ^
c
-r- d
d
Cfi
•u t.
O CJ
d TH
cj ,—
r-H rH
5"
'X 3
™ 0
cr.
ix x
rH H
5 "
hJ "C
c
. d
+j
CO rn
a
C-

^" IT
0 /.
rH rH
£— ' :X
< en

Ci *-*
o '
PL, O1
S5 -C
P*~j hH T-J

3 >. P ^
"S cd & ~
O E O O
g § S -.
1-4 X, W OJ
(0 03
r-H G G < r-H
td o ra —
a o* -H
rH < S.
£ OX
o w
w x

DH W
p;
<;
w

'"-
p
N--
r.
5
O -r-'
S£ CO
>-;
HH J"'
1— .11
1 rH
K £;
*2 *"
O ~
C/]
GO •
I-H 4-1
P1- .0

'J~l i-
i — i C
0 £

M_.

H 'P

CO
W
O BH
i i
C£ CJ
5-
CJ
•rJ
C
•H
~Z

CO
r-H
'rH

in
rH

£
d
J-
Cj
f-^

O
_r,
H

10
.
!— •

O
.C
•*->

i_j
^

M
CJ

G
O
CJ

, — 1
d
o
!S
a
^

fcx

r- '
rH
\_x

d
I 1
O

t
— J

n
o

d
E

•
™
rH
£
£

"C
C
d

4-
"
<~t
„
^
C
X
J_
o
~
£
"3
X
r-
G
73

o c
> r-H

j tn
^
c
C i — i
*c d

o —
/: C
o
t-* -t-j
3 C

Ji 0

^ £

O >

d I-H

^
C7

c;
-T-

^
W
W
*^*

m
r^i
00
t^
CD

, 1

^
O
I-H
a
a
0
g
"C C
o r=
b£
(H
d •
,£? X
0 CJ
rl ^
•C 3
0
O X
JH
d i— *

'X rH
C r"
c -.-
"j- 3
X 73
j- ___
^
~
Cj C
" d
•M
^ J]
c

3 rH
w-i a
^H

x o
^
O 0
^ a

-^.
C-H C
C
c
-r- 0

0 -H
P: X
3
X.!
•• £
CO 0
y, o
O
I — i b£
CO C
CO H
hH rH
S 3
W T:

W
H
< O
hH C
y o

"H '^
5

-
r-* i^

K 0
& s c
0 0
CO +J o
^,
O CT:
CO
^—. x
O
P- c/:

"T,

4->
ri C
O d
1— 1 rH

{JO CO QJ
s s &
o o o
rH U PH
CO
00 i-J
W E-u
r-H CJ
r-H -H
CO r-
3 3
C 'r-
C r-H
d 3
J-

O
E cl
3 C
CT n
C 4-

CJ O
£
X. 7
^
d C

X
o
£1 O
3 --•
— _3
r- C
'J
/
X
->-; O
0

C rH
C-

O rH
c a
G r-,
0 i~i
-i-J -H

G 3

rH G
i — ( r-H
rH
•rH -r-l
C. X
M U
• -M
t^ ^d

d i

-r- E—

C C
o c
c c

C CC

0 0
0 0
o o
LO -o
*y rH
r-|

rH E
d £
•H O
4-> CJ
G

"U
•rH
x -a
OJ G

7v
CJ
-4_i
+J
•H
E
o

;_
C

T--
CO
r-H
O
-rH

C
cx

J
o
•^
o
c
r-l

O
^

~
c
~z

cX
c
H
^,
3
+^
CJ
CO
SH
G
d

•H
^

_p;
X
d

^.
o
•^,
a

n
r;

.^
^

^
_-
r-;
Q
_

r-^
-
r-

LT
—
-C

J
^j

,_^

c
O
^,
• —

_
•H
CO

O
o
C;

O
O
o
J^.
o

CO
w
w
8
cc
OH

o
^
-t-*
o
o —
d r-

X r-i
CJ C
w
d £

3 r-«
O G
-- C
rH £
d c
"*" r—
__ r^
C d

J -r-
O C
-^ O
.b r
C O
7
^
o x

r" r-l
d rH

r- 1—
O O
rC O
4-J -i-1
O x

~
• ^
cv '-
C71 w

^ ^
C

O C
4-J H
G C

O t£

O
O
LO

10
[-S.
LO

<£
CO

ex
CO
1 — 1
o
w
CO
£
C5

*o
c
c

r;
4-J
c

^,
-

r-
L

C
~
r-
J
J-

;-_
r^

G
r-i JXH
J. H

O I-H
O .-»;
C ^L,
— O1

w
^ r-i
O <
~
'*.
_ 0

« ri
CO
™ <•

CQ &
rH OH
C- <

c
c
o

s
^

•^
0

r ,
t-f
H
OS
g
CO
i

^,
1-1
I—
d C
3 • •*"
~ c d
0 *C
_r; -«-H ^^
-- d c
~ 0
' C -
^ tp
^ i£ E
C O d
O "C
3 O

/. x; c
5 ? ~
~~ ~ ~
~ ,_
" — X
X G
'/EC
co C O
O
"" - -
— r-
^ " "
.T/, ^
-C — t-j

r-i -c "
j=. c d

O 3

- K
3 pG X
~ _ ,
O

'•~ C C
CO ^ 'rH
Q. -r-
X X rH
OCX
4-- E 0
, r- p
3 " 3
H ^
-r-1 -r- Jj
in rH
P3 w_< X3
C- C E—

O
0
C r-H
C
CT —
r- C

C 0
_, ^5
o
CD C

d
^
a
d

,_j
c o -
-*-' r-i 0
*- ^
O d C
d

X C pi

O r:
,-- C r-
o c

t-' 3
H- d C
d -r- O
t: r =
^ w ^
: , r^
C C 'J
c. ~
r_
0 r; C-
X
r -f /-
.-• 3 rH
™ / ~
d — '
rC /

'/ "1 ~
0 -^ E

'•— d
O "C
r- O
• d -^
!/ "J

C -^ rH
d r-l
-— G -i—
3 'l-H J",
P 1 F-H
1 — 1 ~ ~
c o
- l (
^ o o
C Y rH T-J

C ,£ C C-

p, .

„

X 0 i
-^ — —
5 T- 7;
c / S
_0 O C

OH'

OH
LO

3 "U OH
c a

CX CO -^
o. -
OH 0 J= • !H

0 >
CC -^ o £J X - H
C IB 0 b£ 0 0 OS
O ID s ai -^ o
•rH bjC « +J rH ,3
M (H SH SH . C 0
Ul C3 O O J-J O 0
•rH rH P O. CO > ~:
S
QJ 00

r-l LO
d
+-> d)
^H -M O
0
^H £ f-H
O 0 3

y
— .

s
1— H
r^
r-^
1
i '
[^
C

y
i — i
"

LJ
q

's
^

o
r--.
—

r. -
110
-------

aMaax ii ^ i ii

I
1

CT) CM CT: ^^ CO O
= CD n C CM LO l-i

b£ ' E-

x ^r CT; CD c CM x t=
rt CO 1-1 N X X CO 1- O
_ M _ — rl CO K
< " " O

!
t-
_; i- x CT: ~H *r "CO CT -r u;
^ CD ^r — n x M cr. r- r-i ss

H *

5; CD (^ :o 10 10 :^ — — cr. t;
Xcoxcxi ro'O CDcn 10— o

C " " 2S
^ CO X C T CD r- rH *~
co C
£2 i
O t

E-
W
O
C ^
O 0
E- C
H S
i> W

G cs

Q <
5 -

G E"1

(i. C5
O
G
w y
J W CO
ft CS
2 E-

c/i C
^
<
r^
SI C
W c
ca
Pi
D
0

e^
H
g

G
"^ > '"" > "^ >' "^
.
's*', 1ft ^2 ±->
•< *? -— v -M
r-l ^ O X O
3
X rH CM O — '
3 x x c tn
C -H H -H H
* i-4 Z3 "j 4-* p
c o q c o
•rl -t-J >, O

C3 -H -H -4-1
<
c^ c
C t^ X
c
<

f.
^
<^ ?* PJ t>-. ^-,
^ iJ J rJ
0 -H W

Cfl . +j +J 4_) +j
-H W DO CO r/) cn
CQ W

t—t
1-1
,_C

i— i
1
*— i
p~

O
xl
i—i

cr'
i—

.3:
r—
cr.

X
r:
S

•^
0
rw.
T—

"C
5_,
C
-;
• C
rH ™:
c3 C
fn ~
P
rH 1 I
1
K • -
™

— ^
Cj rH
— ~*^
F-l
rJ C
rC C

J~ I1
C
1
•- c
C 0
•g ~
= ~
," ?

~
""
X 1
0 +
-
^ cr.
CJ
G -c
o p
rH O
4-1
CJ C
U £
O 3
i-J r— 1
O
C w
Q
H ~
+- C

4-1 -H
CO t-i
P O D r-J ^ *
111
-------
• Sulfur dioxide
• Participates
• Carbon monoxide
• Hydrocarbons
• Oxides of nitrogen

Essentially, this inventory v.as an updating from the IPP and DAQKD emis-
sion inventories of all area and major point source emissions. In June,
1971, the NATO inventory was further updated by using nev. emission factors.

The NATO inventory is the most \ersatile of the inventories
discussed so far. The data are grouped in the following 23 sections:

(1) Summary if Air Pollutant l-missi ons, Study Area.

(2) Summary of Air Pollutant Emissions, Saint Louis (ity,
Missouri.

(3) Summary of Air Pollutant Emissions, Saint Louis County,
Missouri.

(4) Summary of Air Pollutant Emissions, Jefferson County,
Missouri.

(5) Summary of Air Pollutant Emissions, Saint Charles County,
Missouri.

(6) Summary of Air Pollutant Emissions, Eranklin County,
Missouri.

(7) Summary of Air Pollutant Emissions, Madison County,
Illinois.
(8) Summary of Air Pollutant Emissions, Monroe County, Illinois,
(9) Summary of Air Pollutant Emissions, Saint Clair County,
Illinois.

(10) Summary of Air Pollutants from Point Sources.
(11) Summary of Air Pollutants from Area Sources.

(12) Combustion of Fuels, Stationary Sources.

(13) Transportation Sources.

(14) Process Losses.

(15) Emissions from Evaporative Losses.

(16) Emissions from Solid \\astc Disposal.
112
-------
(17) Point Source Emissions.
(18) Area Source Km i s s i on s (272 grids).

(19) Annual Fuel Consumption (by counties).

(20) Housing Units (bv .jurisdictions).

(21) Population, Housing Units (by grids).
•~>
(22) Emission Densities (tons km").

(23) Flights per year (various airports).

Samples from the inventory are shown in Figures A-.i through A-10,
Appendix A.

5. NEDS Inventory-1973

The inventories discussed so far show the gradual evolvement
of the techniques for collecting and processing inventory data. The
earliest inventory, the IPP inventory, contained information on only two
pollutants, sulfur dioxide and particulates, while subsequent inventories
include also data on carbon monoxide, hydrocarbons, and oxides of nitro-
gen. The data collection methods have also become More roll able. \- or
the IPP imentory, information (Mi area source emissions was obtaimd
from gross estimates on fuel delivery and consumption for the entire
area under study. The use of the rapid survey technique for the DAQED
inventory, which breaks down the information into geographic areas of
the community and makes use of formulas using degree-days (a unit repre-
senting the deviation from the mean ambient temperature—G5"F—for one
clay), the number and size of dwellings, and other refinements, provided
considerably more accurate estimates of emissions from area sources.
Similar techniques were used for all subsequent inventories.

A comparison of the presentations of data shows an increasing-
degree of sophistication and usefulness indicated by the various ways
in which the data are broken down and presented, for example, in the
NATO inventory.

The actual emissions of pollutants shown in the tabulations
have been continuously updated to reflect current practices as new sources
came into being or to reflect the installation and effect of control
equipment. For this reason, the inventories currently maintained by the
local and state regulatory agencies, such as the Saint Louis City or
County Air Pollution Control Office, are the only ones which can be used
for the purposes of RAPS. The data in these inventories, some of which
113
-------
are in the form of computer printouts (others are on Inspectors' Report
Sheets), have recently (Summer 1973) been acquired and entered into the
National Emission Data Systems (NEDS) inventory. The characteristics of
this inventory system, presented in APTD-1135, "A Guide for Compiling a
Comprehensive Emission Inventory,"14 are discussed below.

The NEDS was created by the EPA Office of Air Programs to pro-
vide a uniform format for source inventory data on a national scale. It
consists of a number of separate files suitable for computerized process-
ing and has provisions for storing data for both point and area systems
as well as separate files containing various input/output, maintenance,
and editing programs. The concept is shown in Figure 6.

The system distinguishes between point sources (emitting at
least 100 tons per year) and area sources for the primary pollutants:
particulates, S02, CO, hydrocarbons, and NO . The decision to designate
sources emitting more than 100 tons per year of primary pollutants as
point sources is based on the estimate that a 100-tons-per-year source
would produce, under average meteorological conditions, an ambient con-
centration of about 5 to 8 micrograms of pollutant per cubic meter.
This is about 10 percent of the secondary annual federal air quality
standard for SO or particulates. (Page 10-2 of APTD-1135.14)
£t

In addition, about 70 compounds or categories of industrial
sources are designated as point sources regardless of quantities of
emissions.

A point source is defined, therefore, as any emitting point or
plant or facility whose summation of emitting points totals 100 tons or
more per year of any one of the five primary pollutants: CO, SO , NO ,
^ x
particulates, or hydrocarbons; or as any of the pollutant sources listed
below, regardless of quantity of emissions.

• Chemical process industries: adipic acid, ammonia, ammonium
nitrate, carbon black, charcoal, chlorine, detergent and
soap, explosives (TNT and nitrocellulose), hydrofluoric acid,
nitric acid, paint and varnish manufacturing, phosphoric
acid, phthalic anhydride, plastics manufacturing, printing
ink manufacturing, sodium carbonate, sulfuric acid, synthetic
fibers, synthetic rubber, terephthalic acid.
• Food and agricultural industries: alfalfa dehydrating,
ammonium nitrate, coffee roasting, cotton ginning, feed and
grain, fermentation processes, fertilizers, fish meal proces-
sing, meat smoke houses, starch manufacturing, sugar cane
processing.

114
-------
EMISSION INVENTORY
DATA SYSTEM
POINT SOURCE FILE
EMISSION CALCULATION
PROGRAMS
AREA SOURCE FILE
INPUT, OUTPUT
PROGRAMS
HAZARDOUS POLLUTANTS
SOURCE FILE
AREAWIDE INVENTORY
PROGRAMS
EMISSION FACTOR FILE
IPP MODELING CONVERSION
PROGRAM
GEOGRAPHICAL ID FILE
AREA SOURCE GRIDDING
PROGRAM
CONTROL EQUIPMENT
ID FILE
TREND PROJECTION
ANALYSIS PROGRAMS
IPP PROCESS ID FILE
POPULATION DATA FILE
SOURCE: "Guide for Compiling a Comprehensive Emission Inventory,"
Environmental Protection Agency (1972).
SA-2579-11

FIGURE 6 EMISSION INVENTORY SYSTEM CONCEPT
115
-------
• Metallurgical industries: aluminum ore reduction, copper
smelters, ferroalloy production, iron and steel mills, lead
smelters, metallurgical coke manufacturing, zinc (primary
metals industries) ; aluminum operations, brass and bronze
smelting, ferroalloys, gray iron foundries, lead smelting,
magnesium smelting, steel foundries, zinc processes
(secondary metal industries).
• Mineral products industries: asphalt roofing, asphaltic
concrete batching, bricks and related clay refractories,
calcium carbide, castable refractories, cement, ceramic
and clay processes, clay and fly ash sinlering, coal clean-
ing, concrete batching, fiberglass manufacturing, frit manu-
facturing, glass manufacturing, gypsum manufacturing, lime
manufacturing, mineral wool manufacturing, paperboard manu-
facturing, perlite manufacturing, phosphate rock preparation,
rock, gravel, and sand quarrying and processing.

• All petroleum refining and petrochemical operations.
• All wood processing operations.
• Petroleum storage (storage tanks and bulk terminals).
• Miscellaneous: fossil fuel steam electric power plants,
municipal or equivalent incinerators, open burning dumps.

» Hazardous pollutant sources.

Another category of point sources, regardless of quantities of
emissions, are pollutants designated as hazardous. A list of these is
still in preparation.

The location of all point sources is described using Universal
Transverse Mercator (UTM) coordinates. This system, designed by the U.S.
Army provides projections of square grid zones with convenient measur-
ing units. The system has the advantage of continuity and uniformity for
the country, and it is rapidly becoming the accepted system of base
coordinates for a body of technical information.

The NED System contains other files beside the source inventory.
These are
• The Population Data File, which contains such information
as number of household units in an area, number of manu-
facturing employees, number of retail and wholesale establish-
ments. This information is used to calculate emissions
116
-------
where no direct data are available, as for example, emissions
resulting from solid wastes, fuel consumption for space
heating, and others.
• The Emission Factor File, which contains conversion factors
that provide an estimate of the production of a given pol-
lutant by a process or activity. For example, it has been
estimated that two pounds of solvent per person per year are
emitted from dry cleaning plants in moderate climates. In
conjunction with appropriate population data contained in the
Population Data File, the total amount of hydrocarbons per
unit area can thus be calculated. The most complete collec-
tion of factors has been published by EPA in the Compilation
of Air Pollutant Emission Factors (AP-42).

• The Control Equipment File, which contains information on
types and efficiencies of various control devices.

As indicated in Figure 6, the NED System contains a number of
programs; these include not only "housekeeping" programs, such as those
for input and output, editing, and emission calculation, programs per-
mitting outputs by area, source classification, and pollutant, as well
as conversion programs selecting data for implementation planning pro-
grams and trend analysis.

The system is designed to provide data on annual emissions of
pollutants; estimates of shorter emission rates—weekly, daily, or hourly--
can be calculated automatically by applying appropriately resolved pro-
duction throughput, or other control data to the base emission figure.
Such an approach can provide the time resolution necessary for RAPS where
no actual hourly measurements are available. Typical printouts are shown
in Figures A-ll and A-12 in Appendix A.

Table 17 shows a comparison of the various emission inventories.
Only limited conclusions can be drawn from a comparison of these data.
The data in the IBM inventory, covering a much more limited area, are not
comparable at all; nor do the other inventories cover exactly the same
territory. The data shown for the NEDS inventory have been adjusted to
include only the counties included in the two NATO inventories; thus,
these three inventories are directly comparable. Good agreement is shown
for sulfur oxides (if a correction of about 100,000 tons per year,
corresponding to the SO emissions of Franklin County, is added to the
IPP and DAQED figures). The figures for the total emission of partic-
ulates and carbon monoxide are much higher in the NEDS inventory than in
the older inventories; the emission values for hydrocarbons and nitrogen
oxides again show quite good agreement (the contribution from Franklin
County is fairly small).

117
-------
cd
H
*

CD
O
W
g
W K
PJ M

O W
H E-
§
M p
•B, O
PH
O
O co
ns
o
o
o
K

0)
rH
cd '•^
-P £H
O cd
H CU

d \
C 01
•H C
01 O
to -p
•H * —
£
W

c
cu 01
fa£ CU
O 73
r. -H
-P X
-H °

i tr.
0 C
•c -5
>. t-(
E Cd
CJ

o
c -o
0 'H

I* 0
a c
CJ 0

i
O 01
•H O
+- -P
Cd rH
fl 01
3 CU
4H "G
rH -H
3 X
W O

K>>
rH
o
-p
c cu
CD 3

C TH
hH g
r*
C CJ
O CD
•H H
01
01
-H
g
pq

*
cd
cu
^
«jj

Cd CU !H
-p 01 cd
cd cd cu
Q P3 >H
O
-P CU
c e
cu cd
> X
c
~

C
c
X

CM
en
i — i
X
co

to
,O 0 rH
r-H
•P ^— v rH
3 E • ^
0 O 0
tn 3 -P
TJ IH •—' d
cu cu
r-H 01 I/I -P
rH CU CU C
• H -H rl 'H
=H CJ -H
C cd 'H
01 CU C O
E ho C
rH Cd 0 01
O -rl SH
'H rH -P CU
cd c/i -p
OH CJ OJ -P
OH 0 P CU
rH rH G1 rH
-P
C
3 -a
o cu
CJ T3
3
C rH
•H O
rH C

c
cd +->
IH O
PH C
X
CO
en
rH

1 1

QJ
r— 1
fQ
C3

C3
p-
E
0
O

•g
'^

cd
-p
cd
T3

cu
CJ ^^
^ -o
3 CU
O -P
01 cd
T3
-p p.
C 3
• H
0 .— -
P. 00
to
0< Cn
OH rH
rH i— '
>> •£'
-PC >.
C 3 -P T3
3 C ^H C CU
0 CJ -H T3 3 "D
CJ cd C O 3
C rH Cd CJ rH
C -H CJ CJ
O I-H >> CU C
01 r^ -P -P O -H
•H C C C! SH
•a cd 'M 3 c +->
cd h cd o O O
S PH CO O S C
o
J^
CJ)

L ,
CQ
M
c
X
t--
-
o
X
•"""
o
oo
CM
CO
OJ
CO

o
r^
to
..
^
^^T-
CO
^
p-.

c
en
O)
^
to
X
0
en
T

CM
to
to

^f.
cu'

r^
3
01

•c
•H
p.
cd
K
-p
c
3
0
O CU -P
0 0
e SH c -a
•H C CU
rH O >> -D
^J B -P 3
£.1 C i~-l
cd -a 3 cj
S« S O C
PH Cd CJ -H
c
•P X
to
^ en
tO rH
1 — 1

Q
W
O1

Q
C C
1-H ^
to co
-
x m
i — i en
CM CM
c c
m co
to *-
r- x
X X
rH CM

C C
CO CM
r- r-
~
i> in
r-H tO
\r x
~ ~
rH -— <

O C
rH f-
cn en
K *.
m T
x x
rH i — '
0 0
to co
rH CD

CO TF
^ ^
i- r-

0!
rH
O
.p
cd cd cj
-P -p cd
cd cd VH
T3 'O
C
OH OH O
Oj OH -rl
*-H i-r 01
0)
TJ "U -H
CU CU E
•P -P CD
td cd
•o -as
a P. o
P ^ ^

i i
i i
i i
i i

en en
to to
en en
rH rH

0 0
H I-H H CM
< 1 < 1
Z Z
, ,
en
T~
^
x
en
CM
CM
X
CD
CM
X
M

C
to
*T
*.
CO
o
X
..
CO

en
i^
CO
-
t^
rr
m
0
i — i

t^
m
^

cd
H-J
cd
•a

OH
OH
1 — 1

•Q
CD
-p
cd
TJ
a
D

i
i
F
1

CM
t>
cn

CO
Q
W
*
C
3
O
O
01
•H
J
^j
C
•H
cd
co

>,
c
0
CJ

01
cu
rH
SH
S
O

-P
C
•H
a
CO

K-
-P
C
6
CJ

c
0
01
0)

J -p c 3 0 o c 1—1 y c cd PH - -p O 01 •H 3 0 i-4 . -p W 01 .c -p C o g to >> cu > 0 TJ cu -p TJ a 3 01 •H CO c PJ ^ 0) rH cd cu cu • H rH rH Cd cu c o TJ CU 01 _Q CU rH CU ^ -P 3 •P fc co [^ en rH c •H CO Q H iz; o •p c •H -p 5. cu cu ^ cd -P cd TJ CU (H H 118

-------
          The inventories are too close in time to allow any clear trends
to show up.   The differences are probably better viewed as examples of
changes produced by better inventory techniques.   Even so, closer examina-
tion of NEDS data shows examples of values that suggest the desirability
of rechecking the data; for example, a CO emission of 1,340,000 tons from
a single blast furnace or evaporation of 14,000 tons from a single tank
suggests a check.  The former amounts to almost 35 percent of the total
estimated emission of carbon monoxide for the air quality control region;
if correct,  it will certainly affect the view of carbon monoxide as a
                                                               ~L.
pollutant whose origin is primarily related to traffic sources.

          As the NEDS inventory represents the current understanding of
emissions in the Saint Louis area,  it could and should be used for plan-
ning purposes.  For example, it can serve as a basis for the determination
of the number of point sources that should be monitored for the RAPS
emission inventory, since only the  most important point sources can be
monitored.   A review of the point sources listed in the NEDS inventory
shows the breakdown of point sources by size in Table 18 and Figure 7.
                                Table 18
                         NUMBER OF POINT SOURCES

Pollutant

Particulates
SO
X
NO
X
HC
CO
Emission Category
(tons/year)
>10*
271
345

352

269
93
1 - - 	
>102t
99
184

111

99
31
>io3t
28
67

24

23
11
>104t
6
13

7

1
6
          t
Data from NEDS printout of 19 December 1973.
some sources not in earlier printouts.
Data from NEDS printout of 28 August  1973.
                                                         Includes
 Both of these figures are erroneous (Personal Communication, Illinois
 Environmental Protection Agency,  December 1973).
                                  119

-------
H "2
co A
Z
o „
*~ o
Z A
n
CO CM
10 o
/\

co «„ ^
-------
This information is even more relevant if it is arranged to indicate what
percentage of total emissions can be evaluated by monitoring a given num-
ber of sources. By referring to Figures 8 through 12, we see that we can
monitor, say, 90 percent of all emissions from point sources greater than
1 ton/year by observing about 40 sources for SO2, about 5 for CO, 50 for
particulates, and 100 for hydrocarbons; oxides of nitrogen, however, would
require the monitoring of 330 sources to achieve the 90 percent level.
Alternatively, we could decide to use an emission of 100 tons per year
as a cutoff point. In that case, 184 sources for SO , 31 sources for CO,
99 sources for particulates and hydrocarbons, and 111 sources for NO
should be monitored. Table 19 summarizes.

As mentioned above (page 114) the 100 ton-per-year break point
is consistent with a 10 percent sensitivity for modeling or measuring con-
centrations near the levels of the national secondary air quality standards.
While this is probably equally suitable for the modeling verification
studies of RAPS, the sensitivity of such multisource models to this break
point should be investigated by specific studies as recommended above on
pages 48-49.

Table 19
NUMBER OF SOURCES MONITORED
Pollutant
so2
Particulates
CO
Hydrocarbons
NO
X
Type of Inventory
90 Percent of
Emissions from
Point Sources
40
50
5
99
330
100 Tons
per Year
Inventory
184
99
31
99
111
The above data can also be used to evaluate the relative im-
portance, and thus the appropriate treatment, of area sources.

For example, if we assume that all sources emitting more than
100 tons per year of any pollutant will be monitored directly, the remain-
ing point sources (that is, those emitting between 1 and 100 tons per year)
121
-------
§
h- <
Z 1-
3£
cc
LU LL
0- O
CO
LL W
°£
£ 0
g<"
*£
uj ^-
Q- 0
CL
DC
^
LU
CO
0
tr [2
"J o
CD Q;

i °
Z co

C r- CO
O O r-
O r- CO
*—

o q
8 CO
o>

CM •- r-
S28
in i- •-
in in tf
CO CM CM

O) O
in en
in

CO
O
DC
°£
o: Mp
8^0
< — 0
o -
-I Q. ,
< A
i j
O -I «?
1- < o.
t CO
«- CO
CD ^r

en co
CD CO
co co