Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 167
5 Alternative Mobile-Source Emissions Mocleling Techniques IN ADDITION TO MOBILE, numerous other computer models that estimate vehicle emissions have been developed over the years. Several of these models are regional in nature, simulating emissions over large areas, and others are far more microscale, simulating emissions along a corridor, at an intersection, or from an individual transportation project. This chapter briefly describes these alternative mobile- source emissions models, with a focus on defining key differences between MOBILE and the alternatives. Further, this chapter builds upon the transportation and emissions-model integration issues introduced in Chapter 2, defining how mobile-source emissions inventories can be generated at different levels of detail. CALIFORNIA AIR RESOURCES BOARD MOTOR-VEHICLE EMISSIONS INVENTORY SUITE The Clean Air Act allows California to adopt more restrictive automo- bile and fuel standards. This motivated the California Air Resources Board (CARB) to develop a mobile-source emissions model that more close- ly reflected their standards. The California model also integrates data sets for region-specific fleet characteristics and travel, allowing it to con- tain the activity data necessary to estimate both emissions factors as well as inventories. Below is a brief description of their motor-vehicle emis- 767
OCR for page 168
7 68 MODELING MOBILE-SOURCE EMISSIONS signs inventory (MVEI) suite, as well as a discussion of the differences be- tween the California approach and MOBILE. Overview and Recent History of MVEI The current version of the CARB mobile-source emissions inventory model is designated as MVEI7G (CARB 1996b). This is actually a suite of models consisting of the following components, their relationship is shown in Figure 5-1: · CALIMFAC is used to compute the basic emissions rates for light- and medium-duty gasoline-powered vehicles. The output of CALIMFAC is a set of regression equations giving the emissions rates as a function of calendar year for these vehicles. · WEIGHT calculates the distribution of vehicles, starts, and vehicle miles traveled MAT) by model year and vehicle category. · EMFAC calculates all emissions rates (exhaust, evaporative, and tire and brake wear) for a specified calendar year for all vehicle types. These rates are computed as a function of vehicle speed and temperature. · BURDEN calculates the emissions inventory (tons/ day) for a speci- fied county, air basin, or the entire state. One significant difference between the structure of the MOBILE pro- gram and the CARB model is the addition of an emissions inventory mod- ule, BURDEN. The U.S. Environmental Protection Agency's (EPA's) MOBILE model is designed to compute the emissions rates from vehicles. This is the function performed by CARB's EMFAC module. MOBILE has default distributions for the fraction of VMTs by various vehicle classes and model years. The new version of the CARB emissions factor mode] is EMFAC2000. This model has a large amount of area-specific data to compute vehicle activity in various areas of the state. Preliminary results from this model show significant increases in emissions, with statewide inventories for on- road vehicles increasing 68% for CO, 78% for VOCs, and 93% for NOX. Emissions Categories MVEI provides estimates of gaseous and particulate emissions. Gas- eous species are volatile organic compounds (VOCs), carbon monoxide (CO), oxides of nitrogen (NOX), and sulfur dioxide. VOCs can be expressed as total hydrocarbons (HC), nonmethane hydrocarbons, or reactive organic
OCR for page 169
769 l ~ .o ~ to 0) Q_ .m ~ _ a' a' Q cn .> Q ~ a, .0 ~ 3! m E ~ At: a U) o ._ U) U) ._ U) a, 0 e~ ~ . O ~ ~ o .~ _. g E ._ o lo, _ In l:5 o o LL ~: m . - U) Q o ¢D _ . ~ .0 tn ~n O .`n ~ LL] 41.0 tr a) tn CC m ~ ! _ LL ~: U' ~ ) · . .o iL~ o o 111 O U' a) t' E a Ct o ._ * cn ~ . - c] a) ~: a, a) tn —(D ' - (' ¢ v a~ c) o ~Q o cd o a~ c) 5- o a ¢ 'e ·—l ' c~ v o a c~ ~Q 1 l_
OCR for page 170
~ 70 MODELING MOBILE-SOURCE EMISSIONS gases. Only total hydrocarbons are actually computed; adjustment factors are used to obtain the other VOCs measures. (The differences among the various terms used for classifying organic compounds is shown in Appen- dix B.) MVEI computes emissions of lead, total exhaust p articulate s, and particulate emissions from tire wear and brake dust. It also computes fuel consumption. It does not compute refueling emissions. In California, refu- eling emissions are considered a stationary source and are handled as a separate part of the inventory. EMFAC2000 will include computations of carbon dioxide emissions. In EMFAC2000, emissions are computed for passenger cars, eight weight classes of trucks, school buses, urban buses, motorcycles, and motor homes. Vehicle classes are subdivided into gasoline-fueled, diesel- fueled, and electric. Gasoline-fueled vehicles are further subdivided into catalyst and noncatalyst. Basic Operation of CARB Moclels The overall approach for EMFAC2000, MOBILES, and MOBILES is very similar. All these models rely upon regression analyses of data sets to get basic emissions rates and correction factors. There is some sharing of data between the two models, but generally the main data sets for exhaust emissions of light-duty vehicles (LDVs) have been kept separate because of the differences in emissions standards between California vehicles and those sold in the rest of the country. Because the approaches of the California model are similar to MOBILE, the potential accuracy limitations are the same. The Coordinating Re- search Council has sponsored a detailed study of the accuracy of the EMFAC module in MVEI7G (Pollack et al. l999b). That review has noted several data and analysis limitations, which are similar to the criticisms leveled against the MOBILE models. Those include the need for better data on high emitters, heavy-duty vehicles (HDVs), start emissions, partic- ulate emissions, and evaporative emissions. It also noted the need to im- prove estimates of the effects of air-conditioning, inspection and mainte- nance (I/M), and on-board diagnostic (OBD) on emissions. Key Technical Differences Between EMFAC2000 and MOBILES Starting with EMFAC2000, the CARE model for light-duty vehicles (LDVs) will be based on a new cycle called the "unified" or LA92 cycle (Carlock 19991. The data for this cycle have been obtained on recent sur-
OCR for page 171
ALTERNATIVE MOB/[E-SOURCE EMISSIONS MODELING TECHNIQUES ~ 7 ~ veys, which have measured emissions on both the unified cycle and the Federal Test Procedure (FTP) cycle. Users of EMFAC2000 will have the option of obtaining emissions based on the FTP or the unified cycle. The unified cycle emissions should be closer to real-world driving results. MOBILES will continue to report FTP emissions, adjusted by IM240 data. The new CARB model will continue to use five emissions categories (normal, moderate, high, very high, and super) for light- and medium-duty vehicles. This contrasts with the decision to reduce the emissions catego- ries from four in MOBILES to two in MOBlLE6. The emissions category boundaries for MOBILES and EMFAC2000 are compared in Table 5-1. The category definitions in EMFAC2000 provide more distinction among emissions. However, these definitions increase the data requirements for getting a reasonable sample in each category. Much of the development of EMFAC2000 has been devoted to the inclu- sion of area-specific activity data for various regions of California. Area- specific vehicle registration data, mileage accumulation data, vehicle age distributions, and data on VTM are provided for different areas of the state. Area-specific data on temperature and humidity profiles are also provided. EMFAC2000 will use trip-based speed-correction factors (SCFs) rather than facility-specific speed correction factors that are planned for MOBILES. The trip-based SCFs are appropriate for an emissions inven- tory, but link-based SCFs, which should be given by the facility-specific cycles in MOBILES, are more appropriate for applications such as confor- mity determinations (Neimeier et al. 1998~. The use of trip-based SCFs can lead to reduced emissions estimates because the VMT distribution for a complete trip, at an overall average speed, is different from the VMT distribution for a particular link at the same overall average speed. There are various other differences in the detailed implementation of EMFAC2000 and MOBILE6. These can lead to significant differences in the emissions estimates from the two models. However, the two models share the same overall approach and neither model provides any guidance for an innovative approach to the estimation of on-road mobile-source emissions. MOBILE-SOURCE EMISSIONS MODELING IN THE FEDERAL REPUBLIC OF GERMANY The Federal Republic of Germany has developed a set of mobile-source emissions models for a number of purposes (UBA/SAEFL 1999~. There are three primary models that have different levels of complexity:
OCR for page 172
172 to to to v ¢ Cal CO H ~4 o a) o ED Ct V ~0 CO a) o o ·_1 Ct o V ¢ EN V V O O O V ¢ of Z O V V CS o 43 V V V A Cal to Cat ~ ~ ~ ~ O 48 V V V A 9 3) 48 ~ ~ Ad V V V A Cal a) ~) s~ =o cad A CC Cal cod A ~ C~ c~ A — o . ~o a - ct a) 5- o o ~ a~ ·~ ~ c~ a) ~,. - a) a) ~ o 5- -,= cd u) (Q a) a o ~ ~o a ct o ~ · - c~ ~ o · o ,~ ·-l ct .m (Q o ~ · - cO ct · - ~ a) O c) y a ~ · - · · a) ~ 0 a'
OCR for page 173
A1TERNAT/VE MOB/IE-SOURCE EMISSIONS MODEI/NG TECHNIQUES 7 73 Hand book of Emissions Factors (Hand book)7 This is a detailed model for calculating mobile-source emissions factors for a variety of driv- ing conditions and vehicle types; the level of detail for this model is high compared with other German models. This model serves as the foundation of many of the emissions assessments made. CITAIR This model is used to predict emissions levels and pollut- ant concentrations for a variety of microscale control measures; essen- tially, this is a dispersion model combined with a set of emissions factors (not too dissimilar from CALINE4 or CAL3QHC). . . TREMOD This is more of an aggregate model that is used to calcu- late the total emissions inventory for the entire country based on emis- sions from the entire transportation sector (i.e., road traffic, rail, air, and ship) (IFEU 19971. The three different models listed above are highly interrelated, with data files shared between the different models. For example, many of the emissions factors used in TREMOD come directly from the Handbook. This set of emissions models is used for a number of purposes, including assisting in the development of standards of emissions protection; performing environmental assessment studies; road planning; and establishing permits for construction; . Over the years, the German mobile-source estimation techniques have become more and more refined (similar to the incremental improvements to the MOBILE series of models). It is expected in future years to be even more sophisticated. Of most relevance to the MOBILE model are the Handbook and TREMOD models, which are briefly described below. Handbook of Emissions Factors The Handbook of Emissions Factors (UBA/SAEFL 1999) is essentially a database program that is capable of accepting a number of user inputs, combining appropriate data sets, and predicting emissions factors for sev- eral situations. The Handbook is programmed in Microsoft ACCESS, a iThe Handbook was developed with participation from Switzerland and Austria; thus, it is used in all three countries.
OCR for page 174
~ 74 MODELING MOBIIE-SOURCE EMISSIONS flexible, programmable database application and is contained entirely within a single CD-ROM. The Handbook is used to provide emissions factors for a variety of appli- cations. It is often used in conjunction with traffic-simulation models in which emissions are estimated for different roadway sections in a trans- portation network. The Handbook also provides key emissions factors for the other emissions models, CITAIR and TREMOD. The Handbook database contains information on several aspects of mobile-source emissions: (1) different vehicle categories, (2) different traf- fic compositions of those vehicle categories, (3) different traffic scenarios, (4) cold- and warm-start emissions factors, (5) evaporative emission fac- tors, (6) different years of reference, (7) ambient temperature profiles, and (8) different functions for different species of emissions. In creating the Handbook, two primary components were developed: an emissions-behavior component, and a driv~ng-behavior component (see Figure 5-2~. To characterize driving behavior, a number of instrumented vehicles were used to measure real-world driving patterns. A large set of velocity-time profiles were collected for a wide range of driving conditions, ranging from high-speed Autobahn conditions to stop-and-go traffic in ur- ban centers. Statistical analysis was then performed on the large driving-behavior data set, resulting in a number of representative driving cycles for differ- ent road types and different congestion conditions. All together, 43 repre- sentative driving cycles were created: 13 urban, 3 rural, 14 highway, and 13 special driving conditions. In the parallel component, continuous (i.e., second-by-second) tailpipe emissions measurements were made during various chassis dynamometer tests. These dynamometer tests were applied to a wide variety of vehicle types, using several standard driving cycles (e.g., the FTP, the New Euro- pean Driving cycle (NEDC), the U.S.-Highway cycle, and the German Au- tobahn cycles). Approximately 300 LDVs were tested in the basic program, represent- ing 15 different gasoline-fueled and 6 diesel-fueled vehicle types. The second-by-second emissions data for these different categories were matched with their corresponding instantaneous velocity and acceleration values, and velocity-acceleration indexed lookup tables were created to represent the emissions.2 These lookup tables were Fred out using inter- polation techniques. For HDVs such as trucks, a similar methodology was applied, using the European transient cycle and steady-state emissions measurements made on engine dynamometers. 2This type of instantaneous or "modal" emissions modeling is described in more detail in a following section.
OCR for page 175
Representative driving cycles for different road categories 1 1 ALTERNATIVE M OBILE-SOURCE EMISSIONS M ODELING TECHNIQUES 7 75 Real-world-driving Modal emissions behavior measurements measurements over (highways, arterials, etc.) various driving cycles r: ~ Instantaneous Corrective functions emissions model (velocitylacceleration) 1 Emissions factors for different applications FIGURE 5-2 Database development for the Handbook of Emissions Fac- tors. Source FRG-FEPA 1993. The representative driving patterns derived as part of the driving-be- havior component were then combined with the instantaneous emissions functions (i.e., lookup tables) representing the different vehicle types. For every second in a representative driving cycle, the emissions for a particu- lar vehicle type can simply be looked up, and all the second-by-second emissions values for the specific driving cycle can then be summed to- gether to represent an emissions factor. In addition to the hot-stabilized emissions factors, supplementary test- ing was performed to provide additional correction factors for changes in road grade and for cold- and warm-start effects. By combining the driving- behavior database, the emissions-function database, and the added correc- tion factors, emissions of CO, HO, NOx, particulate matter (PM), CO2, and a few other emissions species can be predicted. Compared with the MOBILE model, Germany's Handbook of Emissions
OCR for page 176
7 76 MODELING MOBl[E-SOURCE EMISSIONS Factors is somewhat more detailed and disaggregated in its emissions pre- dictions. One of the key differences is that the emissions factors repre- sented in the Handbook have been derived from the ground-up, using in- stantaneous emissions models developed specifically for producing emis- sions inventories for a wide variety of vehicles. MOBILE, in comparison, derives its emissions factors from integrated certification emissions test- ing, with additional correction factors for speed. Instead of using a global set of speed correction factors for all types of driving, the Handbook has also derived and established a wide range of representative driving cycles, something that MOBILE is now attempting to do in MOBILES with its facility and congestion cycles (see Chapter 3~. Other key differences are that the German Handbook also has corrections for road grade, something MOBILE does not have. TREMOD "The Traffic Emissions Estimation Model" In addition to the detailed Handbook, another macroscale emissions model was created for the entire transport sector of Germany. The TREMOD model was developed in 1993; it also uses the Microsoft AC- CESS program. TREMOD was designed to compute emissions of CO, VOCs, NOX, PM, and other species of emissions from all vehicles in Ger- many, including motor bikes, cars, trucks, airplanes, ships, buses, tractors, and trains. In addition, fuel consumption is also computed. The model is capable of predicting the transport sector emissions inventory for base years ranging from 1980 to 2020. The model uses extensive fleet charac- teristics and activity patterns (for past, present, and future years) for all transport modes. TREMOD has been validated by comparing its overall predicted fuel consumption with collected fuel sales data. For gasoline, TREMOD predic- tions match very well. For diesel fuel, the match was not as good, primar- ily because diesel fuel is used in many different parts of the transport sec- tor (e.g., military, agriculture, and stationary generators) where it is par- ticularly difficult to estimate fuel consumption. FUEL-BASED EMISSIONS INVENTORIES The majority of regional emissions models in the United States, such as MOBILE and EMFAC, use travel-based models that combine gram-per- mile emissions factors with activity data in the form of VMT to estimate emissions. In contrast, fuel-based emissions inventories can also be calcu- lated by normalizing emission factors to fuel consumption rather than VMT. Typically, fuel-based emissions factors are calculated from on-road emissions measurements (e.g., from remote sensors and tunnel studies).
OCR for page 177
ALTERNATIVE MOB![E-SOURCE EMISSIONS MODELING TECHNIQUES 7 77 The activity in this case is a measure of the amount of fuel consumed (Singer and Harley 1996~. This methodology assumes that a precise fuel- use data set is readily available from records such as fuel taxes. Results of fuel-based emissions estimates are contained in the model evaluation sec- tion of Chapter 4. Here we will briefly describe the approach as well as some of its limitations. In recent years, much has been learned about on-road vehicle emissions through the use of remote-sensing instruments. These instruments use an infrared source of light, and when the beam travels through an exhaust plume, it is possible to measure the spectral absorption. Measurements are made of the following ratios: CO to CO2, VOC to CO2, and in the newer sensors, NO to CO2. With these measurements, it is possible to relate the amount of pollutant emitted to the amount of fuel burned using carbon- balance equations (Singer and Harley 1996~. Further, since it is possible to obtain vehicle information (e.g., make, model, and vehicle type) by read- ing the license plate and applying it to a vehicle registration database, the fuel-based emission factors can be disaggregated within the vehicle fleet. Vehicle activity is given by fuel-use data, which can be derived from tax records of fuel sales within each state. Spatial apportionment can be de- termined by tracking fuel shipments and performing filling station sur- veys. To determine the fuel-use activity of disaggregated vehicle sub- groups, it is necessary to calculate the relative fuel economies between the subgroups and the travel fractions of the subgroups. The travel fractions can be determined by measuring the frequencies at which vehicles of each subgroup pass a remote sensor. The accuracy of a fuel-based inventory depends highly on two factors: How wed the entire vehicle fleet is represented by the remote-sensing measurements. Remote-sensing measurements are sensitive to a number of factors, including site location, speed and acceleration of vehicles, and road grade. The remote-sensing sites should be well distributed geographi- cally within the area of study. In general, large numbers of measurements from each remote-sensing site are required to ensure that average emis- sions factors are determined accurately for all vehicle model years. . How well the fuel-use activity data is accurately and correctly appor- tioned within the area of study. In summary, the use of gram-per-gallon instead of gram-per-mile emis- sions factors is claimed to be a simpler method to calculate an emissions inventory, as long as sufficient remote-sensing and fuel-sales data is readi- ly available. Many remote-sensing studies are taking place around the world and the use of remote-sensing in I/M programs are providing addi- tional data. As newer remote sensors are used, VOCs and NOx inventories might also be calculated.
OCR for page 186
186 O CO Z O CO CO LU At: O In o o In .`n ~ / / ~ ~ I ° /os ~ ~ ~ ° ~ ~ ° / ~ U. ~ o / ~ lo, a = = c = ' 1 2 = In ~ ~ ~ . — c,, . . ._ ~ _ . ·- u' s ~ O ~ O.0 Q. Q O ~7 o I ~ I E E ~ 0 E o, I o be\ ~ _ ~ ·, ~ o ~ 5 Q At, I ~ > \C\5 \ I e in O O a \- a' C) a) - a) ~5 o ~Q o . - · , - a) Cot o . - Ct 5- o ~Q Ct E~
OCR for page 187
ALTERNATIVE MOB`LE-SOURCE EMISSIONS MODELING TECHNIQUES 787 vehicles given specific driving cycles, in lieu of performing expensive dyna- mometer tests. Also, vehicle-velocity patterns collected by instrumented vehicles, laser guns, and video-based computer vision can be directly input into an instantaneous emissions model to determine the total emissions associated with its activity. As described in previous chapters, it is recognized that the conventional emissions models (e.g., MOBILE) have a number of limitations when pro- ducing a regional emissions inventory. As a result, version 6 of MOBILE is making a step in the right direction by disaggregating its representative driving patterns with its new facility and congestion cycles (see Chapter 3~. Further, there have been several research efforts to develop models that produce regional emissions inventories at the mesoscale level. For example, the MEASURE model (described later) falls into this category. It is important to point out that modal or instantaneous models can be used as the foundation for more aggregated emissions-factor models. An accurate instantaneous emissions model can be used to essentially replace expensive dynamometer testing. A driving cycle is simply applied, and the emissions associated with the cycle are produced. Therefore, with a modal emissions model providing the foundation, emissions factors can easily be created for models that have a wide variety of representative driving pat- terns. This is essentially what has been done with Germany's Handbook of Emissions Factors. Representative driving patterns were determined in a separate program from the emissions model component, and the emis- sions factors produced for these driving patterns were derived from a modal emissions model in the form of their velocity-acceleration-indexed lookup tables. Further, SCFs used in MOBILE and EMFAC can also be improved with the use of a modal emissions model. SCFs have been created by perform- ing emissions testing using a variety of driving cycles that have different average speeds. These emissions factors are then used to create the speed- correction curves as a function of average cycle velocity. When created these SCF functions in MOBILE and EMFAC, only a limited set of emis- sions testing has been carried out. With the use of a modal emissions model, many more factors could be produced for a wide range of driving cycles for many different vehicle types. Thus, the SCF functions would have a much stronger foundation, as long as a reasonably accurate modal emissions model was used in deriving them. Microscale Traffic-Simulation Mode' Integration with Emissions Factors At the microscale level of detail, traffic-simulation models can be com- bined with modal or instantaneous emissions models to predict emissions
OCR for page 188
7 BB M ODELING M OBILE-SOURCE EMISSIONS inventories. Second-by-second vehicle trajectory data are generated by the traffic-simulation model that can be used as input to the modal emissions model.4 The resulting emissions data from all vehicles can then be inte- grated to provide a total emissions inventory. The easiest form of a modal emissions model to be applied here is the velocity-acceleration-indexed lookup table. In fact, the majority of microscale traffic-simulation models already have the built-in ability to predict emissions, given these emis- sions lookup tables. Keep in mind however, that the lookup-table form of model lacks the ability to handle road grade and vehicle operational his- tory effects. FHWA's TSIS suite of microscale models (i.e., FRESIM, NETSIM, and CORSIM) are capable of estimating emissions using this technique. In these models, the movement of individual vehicles are tracked on a second- by-second basis at intersections (NETSIM), corridors (CORSIM), and free- ways (FRESIM) (FHWA 1998~. Other examples of microscale transportation models that operate in this fashion include the following: . INTEGRATION A microscale traffic-simulation and dynamic-as- signment model that traces movement of individual vehicles on freeways and arterials to a temporal resolution of 1 sec. Incorporating a built-in traffic-assignment algorithm, the model tracks the spatial and temporal activities of up to 500,000 vehicles operating on a subarea with a maxi- mum of 10,000 links. INTEGRATION's ability to combine arterial and freeway movements sets it apart from most conventional traffic-simulation models5 (Van Aerde & Associates 1995~. · PARAMICS A suite of high-performance software tools for micro- scale traffic-simulation. Individual vehicles are modeled in fine detail for the duration of their entire trip, providing very accurate traffic flow, tran- sit time and congestion information, as well as enabling the modeling of the interface between drivers and intelligent transportation system (ITS) technology. The Paramics software is portable and scalable, allowing a unified approach to traffic modeling across the whole spectrum of network sizes, from single junctions up to national networks. Key features of the Paramics model includes direct interfaces to macroscale data formats, so- phisticated microscaTe car-following and lane-change algorithms, inte- grated routing functionality, direct interfaces to point-count traffic data, 4It is important to note that current traffic simulation models may not provide accurate vehicle speed/acceleration data (velocity vectors and/or speed/acceleration probability distributions) due to inadequate car-following equations. There is still a great deal of work that needs to be done in the traffic simulation arena. 5The name INTEGRATION comes from the model's ability to combine movements on arterials and freeways.
OCR for page 189
ALTERNATIVE M OBILE-SOURCE EMISSIONS M ODE[ING TECHNIQUES 7 8 9 batch model operation for statistical studies, a comprehensive visualiza- tion environment, and integrated simulation of ITS technology elements (Paramics 1998~. Another microscale transportation model (in the sense that it tracks individual vehicles every second) is the TRANSIMS model, a large-scale program being developed under the sponsorship of the FHWA, EPA, and the U.S. Department of Energy. The details of this model and its emis- sions module are described later. One of the key challenges for all of these microscale models is how to match the different vehicle types represented in the traffic-simulation component with the vehicle types represented within the emissions compo- nent. Traffic-simulation models typically have different vehicle types that are based on how they operate within a roadway network. In addition to the obvious divisions of vehicle types (i.e., motorcycles, passenger cars, buses, and heavy-duty trucks), categories are often made based on vehicle performance (e.g., high-performance cars and low-performance cars) that can be closely related to traffic-simulation parameters. For heavy-duty trucks, transportation models and data sets typically categorize their vehi- cles based on their configuration and number of axles. In all cases, a straightforward approach to handling the vehicle matching is to create an appropriate mapping between the vehicle types defined in the traffic-simu- lation model, and the vehicle types defined in the emissions model. New Generation Research Transportation-Emissions Models MicroscaTe models track individual vehicles every second as they travel through a predefined roadway network. Because of the this detailed anal- ysis, computer time and storage requirements can be high, depending on the size of the network. Therefore, a number of new generation research models are being developed that are not as aggregated as MOBILE, nor are they as detailed as the microscale models. These models are often re- ferred to as "mesoscale models." MEASURE MEASURE is a model based on Geographic Information System (GIS) that uses an aggregate modal emissions model described earlier. The GIS framework allows for facility-level aggregations of microscale traffic-simu- lation, or disaggregation of traditional macroscale four-step travel-demand forecasting models to develop emissions-specific vehicle-activity data (Guensler et al. 19981.
OCR for page 190
7 90 MODELING MOBI1E-SOURCE EMISSIONS The MEASURE model estimates both spatially and temporally vehicle activities that result in emissions. An emissions rate per unit of activity is defined for each of the activities. Several variables are addressed: vehicle parameters, operating conditions, fuel parameters, and environmental conditions. The model is GIS-based to take advantage of the generation of spatial database management tools already being employed by state and metro- politan planning organizations for the management of municipal assets, resources, and activities. The GIS framework has been shown to be ex- tremely versatile, allowing emissions estimates to be properly allocated spatially. Several key attributes of this model include the following (GuensTer et al. 1998~: Modular A modular approach has been taken so that individual model components can be independently assessed and validated. Stochastic- Because of the high degree of variability in emissions, a stochastic modeling approach has been taken. . Vehicle fleets- Vehicle fleets are characterized by identifying distri- butions of different vehicle technology groups across space and time. . Vehicle activity Both on-network activities and off-network activi- ties are considered. Off-network activity (i.e., local roads) are handled on a zonal basis. Modal activities The model uses an aggregate modal modeling approach in combination with speed and acceleration distributions. . Running-emissions rate~Running-emissions rates are divided into two categories: hot-stabilized operation and enrichment conditions. Uncertainty An assessment of uncertainty is given with the model predictions. The Integrated Transportation-Emissions Modeling (ITEM) Suite In 1995, researchers began developing an integrated set of analytical tools that allows users to better assess the complex relationship between different traffic scenarios and emissions. This modeling suite is referred to as the Integrated Transportation-Emissions Model (ITEM) (Barth et al. 1995~. ITEM was designed to incorporate highly time-resolved modal emissions data that are directly related to vehicle-operating modes, such as idle, various levels of acceleration and deceleration, and steady-state cruise. The modal emissions modeling component is the CMEM model described earlier. ITEM's transportation component is being developed on a hybrid macro- scale and microscale approach. On the one hand, emissions data that are related to vehicle-operating characteristics such as acceleration and decel-
OCR for page 191
ALTERNATIVE MOB!LE-SOURCE EMISSIONS MODELING TECHNIQUES 7 9 7 oration necessitate the detail found in microscale transportation models. On the other hand, a macroscaTe model is better suited for a large, regional traffic network. The computational requirements for a large, regional microscale model would be prohibitively high and would result in a model that is not very useful. By combining a macroscaTe traffic-assignment model with a set of microscale simulation models (organized by roadway facility type), both regional (i.e., wide-area network) and local (e.g., inter- section) emissions inventories can be produced. Emissions are estimated as a function of vehicle congestion on particular roadway facilities, includ- ing freeway sections, arterials (with intersections), rural highways, and freeway on-ramps. Each microscale traffic-simulation model is tightly cou- pled with the macroscale traffic-assignment model, which can dynamically reroute traffic as network capacities change. A travel-demand model drives the traffic-assignment, thus a regional emissions inventory can be produced by using statistical emissions rates (as a function of roadway facility and congestion level) derived from the microscale components, and applying them to the individual links of the macroscale traffic-assignment model. The macroscale and microscale components are set up to run in parallel, so that users of the model can simulate real-time events (such as a traffic accident) and see the effect on traffic dynamics and emissions at both macroscale and microscale levels (Barth et al. 1995~. TRANSIMS The Transportation Analysis SIMulation System (TRANSIMS) is a ma- jor effort aimed at fully integrating transportation and emissions models. TRANSIMS is being developed at the Los Alamos National Laboratory (LANL), funded by the U.S. Department of Transportation, FHWA, EPA, and the U.S. Department of Energy as part of the Travel Model Improve- ment Program. The overall goal is to deploy a large-scale transportation- simulation effort that integrates components of (LANL 1999) activity-based travel demand; intermodal trip planning; traffic microsimulation; and air-quality and other macro analyses. The overall, unified architecture is shown in Figure 5-~. The impetus for developing TRANSIMS stems from issues derived from the Intermodal Surface Transportation Efficiency Act, the CAAA90, and the introduction of various ITS implementations. New technical ap- proaches are introduced in TRANSIMS to handle transportation-planning issues such as congestion pricing, alternative development patterns,
OCR for page 192
7 92 MODELING MOBILE-SOURCE EMISSIONS Households and Activities .< Routes and Plans 1 ~1 ~ ~ 1~. Microsimulation 1 FIGURE 5-5 The four major modules of TRANSIMS include household and activity generation, intermodal router, traffic simulation and an emissions estimator. Note: Feedback loops are provided between the modules to re- plan and modify demand based on the results of traffic simulation. Source: LANE 1999. transportation-controT measures, and their effect on motor-vehicle emis- sions. TRANSIMS has several key features: The identity of individual synthetic travelers is maintained through- out the entire simulation and analysis architecture, with activity times and locations computed for each individual. The simulation output can provide a detailed, second-by-second his- tory of every traveler in the system over a 24-hr day. Second-by-second dynamics of the traffic system can be observed in both local and global con- ditions. As illustrated in Figure 5-1, feedback paths are provided between modules in the simulation framework. These feedback paths provide sta- bility in the results. Thus also allow for the simulation of various ITS strategies, such as simulating the movement of traffic information to se- lected travelers. TRANSIMS is highly modular. The individual modules can be re- placed or modified without disturbing the overall TRANSIMS framework. Further, new modules can be introduced.
OCR for page 193
AITERNAT`VE MOBI1E-SOURCE EMISSIONS MODELING TECHNIQUES 7 93 Framework The flow among the different TRANSIMS modules is determined by a set of scripts. Intermediate data are collected in an iteration database to be used by other modules. In general the flow is summarized as follows: Given sufficient demographic data, synthetic household populations are created (at the desired level of detail) and distributed to match ob- served development patterns; typical demographic data include U.S. Cen- sus Bureau Public Use Microdata Samples and STF-3A data (data from the Census long form). Various activities for each traveler in the system (and freight move- ment) are generated. Activity patterns and mode-choice preferences are derived from surveys. Activity locations are determined based on stan- dard gravity model methods. Individual travel plans are then produced for every individual and freight shipment. The intermodal planner computes a shortest or least- cost path for each traveler. The planner estimates the time that it takes to make a trip based on link traversal-time estimates contained in the overall network. Individual travel plans are then simulated on the network, on a second-by-second basis. The 1-see update interval ensures that dynamic vehicle behavior is captured with a high degree of temporal fidelity. . The environmental module then uses results of the microsimulation to predict tailpipe emissions for LDVs and HDVs. Evaporative emissions are also estimated. A total emissions inventory is produced and is used as input to various air-quality models (e.g., the MODELS-3 framework devel- oped by EPA) to assess ambient concentrations of criteria pollutants at the regional or local level. Environmental Module The objective of the TRANSIMS environmental module is to translate vehicle behavior into consequent air-quality effects and energy consump- tion standards (Williams et al. 19991. Four major computational modules are required: emissions, atmospheric conditions, local transport and dis- persion, and chemical reactions. The last three modules are handled using an air-quality model. The emissions module consists of an evaporation module, which treats emissions associated with rest- ~ng losses, running losses, hot soaks, and diurnal pressure changes; an LDV emissions module, which includes aspects such as malfunc-
OCR for page 194
7 94 MODELING MOB/LE-SOURCE EMISSIONS tioning vehicles, emissions from cold- and warm-starts, normal driving, and off-cycle (i.e., non-normal) driving (when enrichment and enleanment events tend to occur); and an HDV emissions module, representing trucks and buses. The evaporation module uses information from the microsimulation to determine the location of each vehicle and whether it is presently operat- ing or has operated in the previous hour. If the vehicle has not been oper- ating in the last hour, resting losses and diurnal evaporative emissions are calculated using the same formulation found in MOBILE6. While the ve- hicle is operating, running-Ioss emissions are calculated using the MOBILES formulation. If the vehicle has operated in the last hour, hot- soak start emissions are calculated based on the MOBILES formulation. For LDV emissions, the comprehensive modal emissions model (dis- cussed in an earlier section) is currently being integrated into the model. For the calculations, three sets of data need to be developed: . Fleet composition- This is determined from vehicle registration data, and I/M testing. Techniques have been developed to categorize vehi- cles into the appropriate CMEM category using vehicle registration infor- mation (Barth et al. 19981. Fleet status—The status of each individual vehicle is tracked throughout the microsimulation. It is relatively straightforward to deter- mine whether the vehicle is in a cold- or warm-start mode by simply track- ing it through the network. Fleet dynamics One of the major challenges of the emissions mod- ule is to determine the dynamics of each vehicle as it is simulated in the traffic network. The key problem is that the microsimulation component of TRANSIMS predicts second-by-second velocities at "quantum" steps, due to the cellular automata nature of the model. Each vehicle can occupy a 7.5 m spatial bin at any 1 see; therefore, velocity can only assume one of several speed bins. To predict emissions due to vehicle dynamics (particularly during enrich- ment and enleanment events), the emissions module relies on additional empirical data of velocity-acceleration probability distributions (the MEA- SURE model described previously was a similar empirical approach). Us- ing massive data sets from instrumented vehicles, cumulative distribution of accelerations have been derived as a function of the velocity-acceleration product. Three groups of acceleration are then determined: hard accelera- tion, insignificant acceleration, and hard deceleration. The acceleration rate for each vehicle is chosen based on the cumulative probability distri- bution. In addition, different roadway types and congestion levels are de-
OCR for page 195
ALTERNATIVE MOBI[E-SOURCE EMISSIONS MODELING TECHNIQUES 7 95 termined from the microsimulation output, and using an additional empir- ical data set of typical velocity patterns for these roadway types and con- gestion levels, the fraction of the vehicles that undergo hard acceleration, insignificant acceleration, and hard deceleration are determined for the given context. The result is a continuous trajectory that can be fed into the modal emissions model to predict the emissions Williams et al. 19991. For HDVs, the fleet composition is broken down into buses and trucks. Various categories are then considered, based on engine size, chassis size, and model year. The fleet dynamics for the HDVs is less important with regard to emissions compared with LDVs. Buses and trucks typically have Tow accelerations and are usually driven at full throttle whenever the speed is less than desired and when there is adequate headway to acceler- ate. Thus, the modes of operation for heavy-duty-vehicles in TRANSIMS include full throttle, constant speed, and deceleration. The maximum ac- celeration is a function of engine size, road grade, and total vehicle weight. HDV emissions functions for TRANSIMS are derived from emissions test- ing performed at West Virginia University (CIark et al. 1999~. Status A TRANSIMS deployment strategy has recently been developed to make the transition of TRANSIMS technology from a research and devel- opment project to a commercial product that can be used by transpor- tation-planning agencies. The latest release of the TRANSIMS computer code is called TRANSIMS-LANI~. LANL is currently seeking commercial developers for the code, and has released the code for evaluation purposes. The product commercialization process includes initiating licenses and contracts with vendors and developers to build product shells that package the TRANSIMS-LANE technology with user interface enhancements and other modules. TRANSIMS-LANL is also being released to various uni- versities for research, development, evaluation, and demonstration pur- poses. It is expected that a commercial TRANSIMS product wiB be re- leased by developers sometime in the year 2001. SUMMARY In summary, MOBILE is not the only motor-vehicle emissions model that exists. Various other vehicle emissions models have been or are be- ing developed in other countries, at other regulatory agencies, and at dif- ferent research organizations. These other modeling activities approach vehicle emissions estimation in a variety of ways. Some are very macro-
OCR for page 196
7 96 MODELING MOBI[E-SOURCE EM`SS/ONS scale, similar to MOBILE. Others are much more microscale, looking at the vehicle emissions process at greater detail both temporally and spa- tially. One of the key points of this chapter is that some models are more appropriate in terms of their spatial and temporal resolution than others for a given application. It is clear that MOBILE cannot satisfy many of the applications that it is currently used for; therefore, the committee rec- ommends consideration of an emissions modeling toolkit that incorporates a variety of emissions models for different applications. This is described in the next chapter. It should be noted, though, that the upcoming version of MOBILE, known as MOBILES, has undergone extensive peer review and now pro- vides considerable documentation of the methodologies used in it. Any model that is used to replace MOBILE for specific applications must un- dergo a similar level of peer review and needs to provide in-depth docu- mentation to any potential users. In addition, model validation must be the foundation for new model adoption. Validation efforts for all new mod- eling methods should be conducted with vehicles and test conditions not reflected in the data used to develop the model and be undertaken at the scale (or scales) for which a model is designed.
Representative terms from entire chapter: