Expanding Metropolitan Highways: Implications for Air Quality and Energy Use -- Special Report 245 (1995)

The Clean Air Act Amendments of 1990 (CAAA) and complementary provisions of the Intermodal Surface Transportation Efficiency Act of 1991 (ISTEA) introduced new constraints on the transportation sector to help ensure that transportation activities do not delay expeditious attainment of national health standards for air quality. As a result highway projects that expand capacity have come under particular scrutiny in many metropolitan areas for their potential to increase motor vehicle traffic and emissions—a primary source of air pollution. The issue is already at the center of legal challenges or threats of litigation in several metropolitan areas, potentially stalling local highway construction programs.

For years transportation agencies have responded to traffic growth by expanding highway capacity to maintain reasonable levels of service. Capacity expansions ranging from small-scale signal-timing improvements to construction of major highways were expected to relieve congestion without substantial negative effects on air quality. In fact, capacity enhancements that raised travel speeds and smoothed traffic flows were believed to reduce vehicle emissions and improve

energy efficiency. Further, it was widely accepted that new highway capacity was essential to the continued economic growth and competitiveness of major metropolitan areas.

These views are now being challenged. Some analysts and environmental groups argue that adding highway capacity will result in more traffic, higher emission levels, and greater energy consumption in the long run by stimulating motor vehicle travel and encouraging dispersed, automobile-oriented development. In addition they see continued highway expansion as antithetical to a more environmentally oriented and resource-conscious future that stresses the revitalization of older urban and inner suburban neighborhoods and supports transit and nonmotorized forms of transport.

These issues are part of a larger debate over the appropriate direction of metropolitan growth and the role of transportation in that process. This debate involves value judgments about the relative importance of mobility, economic growth, environmental protection, and energy conservation. It considers a broad range of policies, from investments in transportation supply to demand management and pricing strategies.

This study is focused on a more specific topic: the effects of investment in highway capacity on air quality and energy use in metropolitan areas. Its primary audience is metropolitan planning organizations (MPOs), state officials, legislators, and courts with oversight responsibilities. These agencies and officials are being asked to meet the regulatory requirements of the CAAA by making judgments about the environmental effects of highway capacity expansion on the basis of their interpretation of the best available evidence. Energy issues do not convey the same urgency or require the same regulatory analysis, yet transportation 's increasing consumption of the nation's petroleum resources is of concern. To the extent that energy efficiency and energy use are affected by changes in traffic flow characteristics and travel volume from highway capacity expansion, these effects are considered in this study.

The purpose of this study is to review the current state of knowledge, evaluate the scientific evidence, and narrow the areas of disagreement about the impacts of highway capacity additions on traffic flow characteristics, travel demand, land use, vehicle emissions, air quality, and energy use. The state of modeling practice is also exam-

ined to assess the reliability of forecasting tools available to planning agencies; research, modeling improvements, and data collection are recommended to help narrow the gap between regulatory requirements and analytic capabilities.

International experience relevant to the study charge was considered, and alternatives to highway capacity expansion, such as “traffic calming,” were examined to the extent they shed light on the effect of changes in traffic flow characteristics on vehicle emissions and energy use. In most cases, however, the experience of other countries, particularly European countries, is not directly applicable to the United States because of considerable differences in land availability and cost, population density, mode choice, pricing structures, and institutional governance of regional growth. Wholesale adoption of European transport strategies might produce some reductions in vehicle energy use and emissions and concomitant improvement in metropolitan air quality, but, in the committee's judgment, it would also impose substantial social and economic costs and raise questions about institutional and political feasibility.

Under the CAAA the U.S. Department of Transportation (DOT) and MPOs are directly accountable for demonstrating the compatibility of transportation investments with timely attainment of national air quality standards. According to the current interpretation in Environmental Protection Agency (EPA) regulations, the act provides a grace period for states to revise their air quality attainment plans, which set emissions limits by source. In the meantime, transportation agencies in metropolitan areas that have not attained national standards must meet strict interim conformity requirements to prevent air quality degradation. Specifically, nonattainment areas are required to prove that (a) projects in regional transportation improvement programs and plans will not lead to motor vehicle emissions higher than in a 1990 baseline year and (b) by building these projects, emissions will be lower in future years than if the projects are not built (the “build–no-build” test).

Once EPA approves new state air quality attainment plans the conformity test changes: the emissions that result from carrying out regional transportation improvement programs and plans in nonattainment areas must not exceed target emissions caps for transportation sources established in the EPA-approved state air quality plans. The conformity test still requires a regional emissions analysis, but the criterion for comparison is less demanding: predicted regional emissions from transportation sources must be within state-determined emissions budget caps. The most analytically demanding comparison—estimating changes in emission levels in future years from marginal expansions of regional transportation networks relative to maintaining the status quo—would no longer be required.

The committee did not limit its examination of the effects of highway capacity additions to a specific time frame. However, particular attention was paid to the 20-year time frame established by the CAAA for attainment of air quality standards, because this represents the planning and forecasting horizon within which local planning agencies are required to make judgments about the air quality effects of highway projects.

Providing a precise assessment of the net effects of expansions of highway capacity on air quality and energy use is not straightforward. Addressing the questions raised in this study requires modeling a complex sequence of interrelated events—from initial impacts on traffic flows to longer-term consequences on travel demand, automobile ownership, and residential and business location in a metropolitan area. It also requires modeling the emissions generated by the predicted travel impacts and forecasting their effects on air quality at or near the points of emission and throughout the region. There is significant uncertainty in predicting precise quantitative outcomes at each stage in the analysis. In addition, different levels of analysis are required to distinguish project-level from regional effects.

Determination of initial effects is complicated by the network character of the transportation system. The ability of drivers to change their route, time, and mode of travel means that adding highway ca-

pacity and thus reducing travel time at one location in the system will affect other locations as users take advantage of the new capacity. Modifications in traffic flows and volumes on this broader system of facilities, and their effects on emissions and energy use, must be taken into account.

User responses also change over time. The response to large reductions in travel time is likely to be greater in the long run than initially, as highway capacity additions influence decisions about location and automobile ownership. The key uncertainty is at what point, or whether at any point, the emissions increases from the new development and traffic growth stimulated by the capacity addition will offset the initial emission reductions gained from smoothing traffic flows. Assessment of net project effects on emissions and energy use depends on the length of time over which the effects are analyzed and the value placed on long-term versus more immediate effects. The longer the prediction period the more likely other, often unpredictable factors such as changes in demographic or economic conditions are to intervene and diminish forecasting accuracy.

Forecasts of the net effects of adding highway capacity also involve comparisons with alternatives. Not investing in highway capacity or undertaking other investments or demand management strategies has consequences for air quality and energy use. A comparison of outcomes requires estimating how sensitive users are to changes in travel time and cost and to what extent travel by transit or other modes can be substituted for automobile travel. It also requires understanding how growth would be distributed both within and outside the metropolitan area if highway capacity expansions were restricted. Paradoxically, because of past and ongoing efforts to reduce vehicular emissions, it will become more difficult to discern differences in emissions resulting from different transportation and demand management strategies. These alternative strategies all represent small changes to a declining base of emissions from highway sources.

After examining the considerable literature on the relationships among transportation investment, travel demand, and land use as well

as the current state of the art in modeling emissions, travel demand, and land use, the committee finds that the analytical methods in use are inadequate for addressing regulatory requirements. The accuracy implied by the interim conformity regulations issued by EPA, in particular, exceeds current modeling capabilities. The net differences in emission levels between the build and no-build scenarios are typically smaller than the error terms of the models. Modeled estimates are imprecise and limited in their account of changes in traffic flow characteristics, trip making, and land use attributable to transportation investments. The current regulatory requirements demand a level of analytic precision beyond the current state of the art in modeling.

The state of emissions modeling illustrates the problem well. In theory the initial effect of a highway capacity addition on traffic flow characteristics and the resulting changes in vehicle emissions should be measurable for the current fleet. Despite considerable research and vehicle testing, however, no definitive and comprehensive conclusions can be reached. Virtually all motor vehicle testing has been based on a limited set of driving test cycles that inadequately represent current urban driving conditions. In addition, current emission models rely on average trip speed as the sole descriptor of traffic flows. Variability in speed, road grade, and other factors that strongly influence emissions is not explicitly incorporated into the models. Changes in vehicle emission rates thus cannot be predicted reliably for a wide range of changes in average trip speeds, many of which are in the range of average speed changes expected from highway capacity additions.

Current emissions models were developed to estimate motor vehicle emissions at the regional level. They cannot be appropriately applied to estimate the emissions effect of changes in traffic flow patterns at specific highway locations. Even for regional estimates, current models significantly underpredict emissions of some pollutants. Models can be developed that are more sensitive to vehicle operations and traffic conditions; some research and testing have already begun. However, development and incorporation of new models into the regulatory process will take substantial time and investment and require close coordination between the transportation and the regulatory communities.

The initial effects on energy use from highway capacity additions can be predicted more reliably than effects on emissions because fuel

economy is not as sensitive as emissions to traffic flow conditions, particularly speed variation. However, energy and emissions estimates must both be linked with reliable data on the likely impacts of highway capacity additions on traffic, travel demand, and location decisions. Here, too, the available travel demand and land use models provide imprecise and limited estimates of likely outcomes. Improvements in modeling capability will also require substantial investment.

More research and improved models can help narrow the gap between regulatory requirements and analytic capabilities. The conformity tests required by current regulations will themselves change as the build–no-build test is phased out once state air quality plans are approved by EPA. However, the complex and indirect relationship between highway capacity additions, air quality, and energy use, which is heavily dependent on local conditions, makes it impossible to generalize about the effects of added highway capacity on air quality and energy use, even with improved models.

On the basis of current knowledge, it cannot be said that highway capacity projects are always effective measures for reducing emissions and energy use. Neither can it be said that they necessarily increase emissions and energy use in all cases and under all conditions. Effects are highly dependent on specific circumstances, such as the type of capacity addition, location of the project in the region, extent and duration of preexisting congestion, prevailing atmospheric and topographic conditions, and development potential of the area.

Nevertheless, some general findings do apply. Within developed areas, traffic flow improvements such as better traffic signal timing and left-turn lanes that alleviate bottlenecks may reduce some emissions and improve energy efficiency by reducing speed variation and smoothing traffic flows without risking large offsetting increases from new development and related traffic growth. However, the cumulative effect of multiple small improvements in traffic flows may attract increased traffic, at least in the vicinity of the improvements. In lessdeveloped portions of growing metropolitan areas—where developable land is available and most growth is occurring—major highway capacity additions such as a freeway bypass or a major interchange reconstruction are likely to attract further development to these locations and increase motor vehicle travel, emissions, and energy use in these areas. Whether these outcomes lead to a net increase in regional

emissions and energy use depends on whether the highway expansion redistributes growth that would have occurred elsewhere in the region or whether it stimulates productivity gains that result in net new growth.

Despite the considerable uncertainties in predicting the effects of expanding highway capacity on air quality given the current state of knowledge and modeling practice, policy makers and planners must comply with current environmental regulatory requirements and make decisions on the basis of the best available information. Thus, the committee thought it should provide its best judgment on the likely payoffs of pursuing current policies.

In its opinion, the current regulatory focus on curbing growth in motor vehicle travel by limiting highway capacity is at best an indirect approach for achieving emissions reductions from the transportation sector that is likely to have relatively small effects, positive or negative, on metropolitan air quality by current attainment deadlines. Historically, measures to control traffic demand have had limited effects (Apogee Research, Inc. 1994). According to estimates from local studies using current emission models, these traditional transportation control measures (TCMs), which include traffic flow improvements among others, are likely to yield changes of 1 or 2 percentage points individually in regional emissions of key pollutants by attainment deadlines (DOT and EPA 1993, 9). The effects of traffic flow improvements could be positive or negative, depending on offsetting increases in traffic. These are small changes on a declining base, given EPA projections of continuing emission reductions for carbon monoxide and volatile organic compounds (DOT and EPA 1993, 9).

In the committee's opinion, major highway capacity additions are likely to have larger effects on travel and to increase emissions in the affected transportation corridors in the long run unless some mitigating strategy is implemented in conjunction with the capacity addition. However, because of the large investment implicit in current metropolitan spatial patterns, it may be years before changes in land use and related traffic patterns induced by the added capacity make a significant difference in regional emission levels and air quality.

Curtailment of all highway capacity expansion that has any potential for increasing emissions risks pitting environmental against economic concerns. In the past, when environmental goals have conflicted with economic objectives, the response has been to delay or reassess environmental regulations. It is easy to envision these pressures building again. In addition, requiring policy decisions to hinge on uncertain model outputs leaves the entire process vulnerable to error and manipulation.

In the committee's view, a more constructive approach is to look for ways to reconcile air quality with economic goals. The committee believes that technology improvements can yield more significant benefits for air quality relative to the current focus on curbing travel growth. Catalytic converters, electronic fuel injection, and unleaded gasoline have resulted in substantial reductions in vehicle emissions in the past 20 years (Nizich et al. 1994). EPA predicts further emission reductions for major pollutants on the order of one-quarter to one-third from 1990 baseline levels by attainment deadlines simply from continued vehicle fleet turnover and implementation of CAAA-required vehicular and fuel standards and enhanced vehicle inspection and maintenance programs (Nizich et al. 1994, 5-4, 5-6).

Market solutions also have promise, although the feasibility of some approaches is untested. For example, potential increases in traffic from new highway capacity such as added expressway lanes might be reduced by imposing tolls varied by time of day (i.e., congestion pricing) and collected electronically to control travel growth on the expanded facility. A recent National Research Council report on congestion pricing (NRC 1994) has examined in depth the technical and political feasibility of this approach. Applied in a limited setting, it would not require major changes in current highway finance patterns. It would allow highway capacity to be provided where it is needed but could mitigate negative effects on emissions from travel growth.

In the long run, stronger measures, such as pricing motor vehicle travel to better reflect the full social costs of highway travel and the introduction of areawide, time-of-day tolls, may be necessary to provide direct incentives for reducing or shifting travel demand in ways that use highway capacity more efficiently and with less cost to the environment. Local land use and zoning measures that increase building density and support mixed-use development could be introduced

more widely. Land use measures may reduce areawide automobile travel and emissions, but the changes are likely to occur gradually and will have more significant effects if they are implemented in conjunction with pricing measures. Finally, radical advances in vehicle technology could produce cleaner transportation, substantially reducing the level of vehicle emissions.

These more radical solutions are neither new nor easy. Major technological improvements require substantial investment and time to produce results. Changes in pricing or land use policies require significant institutional changes—such as more powerful regional institutions to coordinate areawide pricing schemes and land use strategies—and more public acceptance than has been demonstrated in the past. In the judgment of the committee, however, as long-run alternatives to current policy, they offer better prospects for reconciling economic and environmental goals.

DOT

U.S. Department of Transportation

EPA

Environmental Protection Agency

NRC

National Research Council

Apogee Research, Inc. 1994. Costs and Effectiveness of Transportation Control Measures (TCMs): A Review and Analysis of the Literature. National Association of Regional Councils, Bethesda, Md.

DOT and EPA. 1993. Clean Air Through Transportation: Challenges in Meeting National Air Quality Standards. Aug.

Nizich, S.V., T.C. McMullen, and D.C. Misenheimer. 1994. National Air Pollutant Emission Trends, 1900–1993. EPA-454/R-94-027. Office of Air Quality Planning and Standards, Research Triangle Park, N.C., Oct., 314 pp.

NRC. 1994. Special Report 242: Curbing Gridlock: Peak-Period Fees to Relieve Traffic Congestion, Volumes 1 and 2. Transportation Research Board and Commission on Behavioral and Social Sciences and Education, National Academy Press, Washington, D.C.

The Clean Air Act Amendments of 1990 (CAAA) (Public Law 101-549, 42 U.S.C. 7401, et seq.) and complementary provisions of the Intermodal Surface Transportation Efficiency Act of 1991 (ISTEA) (Public Law 102-240, 49 U.S.C. 101 note) tightened controls on the transportation sector to help ensure that transportation activities contribute to timely attainment of national health standards for air quality. As a result traditional transportation projects, such as additions to highway capacity, have come under close scrutiny as potential contributors to air pollution.

For years transportation officials have relied on adding highway capacity—building new expressways and adding lanes to existing freeways —to accommodate travel demand in growing metropolitan areas. Such increases in capacity were thought not only to bring congestion relief but also to improve air quality and fuel efficiency by contributing to freer-flowing traffic conditions.

This conventional wisdom has been challenged by environmental planners and other analysts who take a long-term perspective on the

effects of additional highway capacity. They concede that adding highway capacity may initially reduce some vehicle emissions and improve fuel efficiency by smoothing traffic flows and reducing stop-and-go traffic, although the benefits may not be as significant as were once believed. However, these positive effects may be eroded over time by growth in travel stimulated by the new capacity. Improved levels of highway service may encourage shifts from less polluting modes of transportation and induce new or longer trips once discouraged by congested conditions. As traffic volume grows, traffic operations may deteriorate, producing levels of congestion comparable with previous conditions but at higher traffic volumes. In the long run, these analysts maintain, new highway capacity will improve access and may encourage development in low-density areas not amenable to transit. Low-density development requires more frequent and longer trips, increasing emission levels and energy use and further degrading air quality.

The issue of highway capacity and air quality is already at the center of legal challenges brought by environmental groups to many additions to highway capacity in metropolitan areas. This issue is likely to receive more attention as metropolitan areas grapple with the stricter requirements of the CAAA. The act allows citizen suits to be brought for the first time against the U.S. Department of Transportation (DOT) and the Environmental Protection Agency (EPA) for noncompliance with legislative requirements and timetables, which opens the door to increased litigation.

Assessment of the precise consequences for air pollution and energy use of any particular addition to highway capacity is uncertain, given the current state of knowledge and modeling practice. Travel demand forecasting models—the basis for determining the effect of increased highway capacity on travel demand—were originally developed to help determine the appropriate size of new capital facilities. They are not well suited to providing the detailed data, such as speed data and travel data by time of day, needed for modeling and analysis of vehicle emissions and air quality impacts. Nor can most current travel demand models adequately measure the effect of improvements in highway service on the amount of travel or on land use patterns, which in turn could affect future demand for travel and its distribution in a region.

Emissions models, which measure the polluting effects of motor vehicle travel, inadequately represent the emission performance of

in-service vehicles. Current data on the relationship between vehicle speeds and emission levels—critical to analyzing highway capacity projects that will change the distribution of traffic speed levels and variability of speeds—are based on averages that mask wide variances across individual vehicle performance, roadway conditions, and driving behavior. Moreover, the models do not adequately capture major suspected sources of emissions from vehicle accelerations and high speeds, although some research is under way to understand these phenomena. Planning agencies often apply current speed-emission relationships as if they were precisely known.

Despite the limitations of the existing knowledge base, engineers and scientists are being pressed to provide reliable estimates of the likely effects of adding highway capacity on emissions and energy use to assist legislators, state officials, metropolitan planning organizations (MPOs), and judges in reaching decisions on these issues. Thus a review of the current state of knowledge has been undertaken to evaluate the scientific evidence concerning these effects and to narrow areas of disagreement. The specific questions at issue are described and, where possible, research or analyses that could be conducted to speed their resolution are recommended. More specifically, the study committee

• Critically reviews existing research of the links among highway capacity, traffic flow characteristics, travel demand, land use, vehicle emissions, air quality, and energy use in metropolitan areas;

• Identifies the conditions most likely to affect emissions and energy use;

• Reviews the reliability of models and analyses that regional and state planning agencies use to forecast travel demand and land use, emission levels, and energy consumption; and

• Recommends research strategies, modeling improvements, and data collection efforts to improve analytic capabilities.

The enactment of the CAAA has refocused attention on the effects of transportation activity on air quality, but legislation to set standards for alleviating air pollution and improving the fuel efficiency of motor vehicles dates back to the 1970s.

The harmful effects of air pollution on public health were formally recognized by the requirements of the Clean Air Act Amendments of 1970 (Public Law 91-604, 84 Stat. 1676), which mandated establishment of national ambient air quality standards (NAAQS) for six pollutants: carbon monoxide (CO), lead (Pb), nitrogen dioxide (NO₂), ozone (O₃), particulates (PM-10), and sulfur dioxide (SO₂) (Curran et al. 1994, 19). There is no standard for carbon dioxide, the principal greenhouse gas, because carbon dioxide is not toxic and therefore has no direct negative health impact.

Since 1970 substantial gains have been made in reducing pollution from transportation sources, primarily through technological improvements such as catalytic converters and electronic fuel injection, which are now standard equipment on cars,¹ and through the use of lead-free gasoline. Between 1970 and 1993, for example, highway vehicle emissions of CO and volatile organic compounds (VOCs)² (which are a precursor of ozone) declined by 32 percent and 53 percent, respectively; highway vehicle emissions of lead were effectively eliminated (Nizich et al. 1994a, 3-11, 3-13, 3-16; Figure 1-1). However, highway vehicle emissions of nitrogen oxides (NO_x)—the other ozone precursor—remained nearly constant (Nizich et al. 1994a, 3-12).³ Although the emissions models on which these estimates are based typically have underestimated absolute emission levels, the trend has been downward.

Despite more than two decades of progress in improving air quality, EPA estimated that in 1993 about 59 million people lived in counties that violated one or more of the NAAQS (Curran et al. 1994, 14). ⁴ Thus, the CAAA contain stringent requirements for further reductions in emissions from transportation sources, backed up by strict monitoring and sanctions for nonperformance, to bring nonattainment areas (i.e., those areas not attaining NAAQS) into compliance.

FIGURE 1-1 Highway vehicle emissions of selected pollutants: 1970, 1980, and 1993 (Nizich et al. 1994a, E-11–E-13).

The act requires additional technological advances, such as tougher tail pipe standards, enhanced vehicle inspection and maintenance programs, and cleaner fuels programs, which should result in further reductions in emissions as newer, cleaner vehicles replace older ones and as technology-oriented programs are implemented. These measures, however, may not offset growth in emissions from motor vehi-

cle travel. For example, if vehicle miles traveled (VMT) grows at a relatively conservative rate of 2 percent per year⁵ and no additional technology advances are made other than those included in the CAAA, EPA estimates that tail pipe emission gains could be offset by 2002 for CO and VOC and by 2004 for NO_x (Figure 1-2).⁶ Thus, the CAAA mandate measures to limit automobile trips in the most severely polluted areas and require strict monitoring of VMT growth in less severe nonattainment areas.

Deadlines for reaching attainment vary with the severity of air quality problems.⁷ Areas classified as marginal (40 for ozone) have 3 years from the baseline year, 1990, to attain NAAQS; areas classified as moderate (29 areas for ozone, 37 for carbon monoxide) have 6 years. The 12 areas classified as serious for ozone (and 1 for carbon monoxide) have 9 years, whereas the areas classified as severe (9 for ozone) have 15 to 17 years. Los Angeles, the only area classified as extreme (for ozone), has 20 years. Eighty-three areas have been designated as nonattainment for PM-10. Los Angeles is also the only area that does not meet the NO₂ standard.

Levels of effort also vary with the severity of air quality problems. Areas with ozone classifications of moderate or worse must submit revisions to State Implementation Plans (SIPs) (plans that codify a state's CAAA compliance actions) showing that within 6 years the pollutants that create ozone will be reduced by at least 15 percent from 1990 baseline emissions net of any growth in emissions during that period. These areas must achieve an additional 3 percent per year reduction in emissions until attainment is achieved. In addition to the latter requirements, areas classified as severe or extreme must adopt transportation control measures (TCMs) aimed at decreasing automobile travel.⁸

Areas with carbon monoxide designations only, of moderate or worse, must comply with somewhat different requirements. They must forecast VMT annually beginning in 1992, and if actual VMT exceeds forecast, they must be ready to adopt contingency TCMs that must be included in SIPs. TCMs are required for carbon monoxide areas designated as serious.

Finally, the CAAA provide strict sanctions for noncompliance. For example, approval of federally assisted highway projects can be withheld for failure to submit a SIP, EPA disapproval of a SIP, failure to

make a required submission, or failure to implement any SIP requirement. ⁹ Once such sanctions have been imposed—and they must be imposed if EPA determines that the deficiency has not been corrected 18 months after being identified—DOT can only approve highway safety projects or projects that would not increase single-vehicle automobile travel.¹⁰

The requirements and timetables for meeting CAAA goals and sanctions for noncompliance are stringent, but not unlike mandates of earlier clean air legislation (e.g., in 1970 and 1977). To help ensure attainment of legislative mandates, regulations implementing the 1990 CAAA strengthen existing conformity requirements. Conformity is a determination made by MPOs and DOT that transportation plans, programs, and projects in nonattainment areas are in accord with the compliance standards contained in SIPs (FHWA 1992b, 12). The new conformity rules under the CAAA hold MPOs and DOT directly accountable for demonstrating that transportation activities will not cause or contribute to any new violation of air quality standards, increase the frequency or severity of existing violations, or delay timely attainment of standards (Federal Register 1993b, 62,188).

EPA regulations provide a grace period for states to revise their SIPs; revised plans were due November 1994. However, stringent interim conformity guidelines are in effect until EPA has approved these plans to ensure that nonattainment areas do not fall behind in meeting target deadlines for achieving air quality standards. The salient feature of the interim conformity process is project-level review. Specifically, MPOs in nonattainment and maintenance areas must show that (a) all federally funded and “regionally significant projects,” including nonfederal projects,¹¹ in regional transportation improvement programs (TIPs) and plans will not lead to emissions higher than in a 1990 baseline year; and (b) by building these projects, emissions will be lower than if the projects are not built. Once SIPs are approved by EPA,¹² the conformity test becomes less demanding: MPOs must demonstrate through a regional analysis that the emissions produced by implementation of transportation plans and TIPs will not exceed target levels (known as emission

budgets)¹³ for motor vehicle emission sources from nonattainment and maintenance areas contained in the SIPs.

The conformity regulations thus provide the teeth to help guarantee compliance with air quality mandates. MPOs must reconcile emission estimates from transportation plans and TIPs with those contained in the motor vehicle emission budgets in SIPs, conduct periodic testing to determine whether actual emissions are in line with estimates, and take remedial action if they are not. Court challenges have already been issued by environmental groups against several state transportation departments to ensure that conformity requirements and attainment deadlines are met.¹⁴

ISTEA complements the CAAA by reinforcing air quality conformity requirements. The act allows local areas flexibility to shift highway funds to transit and other cleaner modes, enhances the planning re-

sponsibilities of MPOs that are responsible for conformity analyses, and creates a new Congestion Mitigation and Air Quality Improvement Program to direct funds to projects and programs in nonattainment areas that will contribute directly to attainment of the NAAQS (FHWA 1992c, 16–17). Metropolitan transportation planning regulations, in particular, were revised in part to ensure greater consistency and coordination between development of transportation and air quality plans (Federal Register 1993c).¹⁵

Legislation similar to the CAAA has not been passed recently in the energy area. In response to the oil embargo imposed by the Organization of Petroleum Exporting Countries (OPEC) in 1973 and because the automotive sector is a major consumer of petroleum, Congress passed the Energy Policy and Conservation Act of 1975, which set fuel economy standards requiring that automotive manufacturers increase the corporate average fuel economy (CAFE) of automobiles and light trucks sold in the United States to 11.7 kpl (27.5 mpg) in the 1985 model year and thereafter (NRC 1992, 12).

Between 1970 and 1992, in-use fuel economy for passenger cars rose from an average of 5.7 kpl (13.5 mpg) to 9.2 kpl (21.6 mpg), a 60 percent improvement (Davis 1994, 3–24). The introduction of CAFE standards, technology improvements such as electronic fuel injection and more efficient engines and transmissions, and the reduced weight of passenger vehicles were largely responsible.

Although the standards have been achieved,¹⁶ during the past decade the low price of gasoline, growing motor vehicle ownership, and increased motor vehicle travel have resulted in a steady increase of transportation's share of total petroleum consumption in the United States to 65 percent in 1992 (Davis 1994, 2–7). Also, other sectors have substituted alternative energy sources. Since 1989, U.S. energy dependence on foreign oil sources has reached levels only exceeded in 1979; oil imports accounted for 46 percent of U.S. petroleum consumption in 1992 (Davis 1994, 2–5). Thus, energy officials are seeking ways to improve fuel efficiency and reduce vehicle travel to cut back energy consumption.

The conformity requirements of the CAAA and, to a lesser extent, concern for energy use will place transportation projects—particularly highway projects in the nation's most polluted areas—under great scrutiny regarding their potential for stimulating automobile travel, raising emission levels, and further increasing dependence on fossil fuels. Local planning agencies, who are responsible for programming highway projects in urban areas and certifying their positive or neutral effects on air quality, are expected to have the analytic and modeling capabilities to forecast project impacts. Such requirements provide the impetus and the focus for this study. The committee did not limit its examination of the effects of highway capacity additions to a specific time frame. However, particular attention was paid to the 20-year time frame established by the CAAA for attainment of air quality standards, because this represents the planning and forecasting horizon within which local planning agencies must make decisions about the air quality effects of highway projects.

The size of the investment in capacity enhancements also warrants close examination of potential impacts. Although construction of new highways has tapered off in recent years, the combination of new construction with capacity additions to existing highways continues to represent a large fraction of the public investment in transportation infrastructure by state and local governments. For example, according to the most recent estimates, capacity improvements to roads and bridges totaled $15.4 billion,¹⁷ accounting for nearly two-fifths (38 percent) of total public capital outlays for highways, bridges, and transit combined (U.S. Congress. House. Committee on Public Works and Transportation. 1993, 4, 76).

Alternative ways of meeting urban transportation needs are not examined in this study, nor are the costs and benefits of alternative approaches to improving air quality and reducing energy use in a metropolitan region analyzed. Instead the study is focused on examining the scientific basis for estimating the impacts of highway capacity enhancement projects, both favorable and unfavorable, on emissions, air quality, and energy consumption.

Highway capacity projects are defined broadly in this study (see accompanying text box). Additions to highway capacity are often associated with major construction projects, such as building a new highway where none existed before, bypassing an existing route, or adding one or more lanes to an existing highway. Nonetheless, smaller-scale projects, such as traffic signal timing improvements or removal of onstreet parking, should also improve traffic flows and thus are included here. Less traditional measures that add capacity but attempt to restrict or manage travel as part of the facility improvement, such as high-occupancy-vehicle (HOV) and express bus lanes or variable tolls by time of day (congestion pricing), are also covered in the definition of highway capacity projects. Finally, many of the new technologies that fall under the category of intelligent transportation systems should also be characterized as capacity enhancement measures because their primary objective is to improve the efficiency of traffic flows on existing facilities.

Highway engineers define capacity of a facility as the maximum hourly rate at which persons or vehicles can reasonably traverse a segment of roadway during a given time under prevailing roadway, traffic, and control conditions (TRB 1992, I-3).¹⁸ The concept of capacity thus involves throughput of people and goods as well as the more traditional notion of vehicle throughput. Both measures are considered in this study, although lack of data on the former (e.g., vehicle occupancies) hampers analysis of emission and energy impacts on a person- or goods-throughput basis. Both freight movement on commercial vehicles and passenger transport are of interest because both types of traffic will be affected by capacity enhancements, although their emissions, energy use, and travel pattern changes will differ. Separate papers to review the likely effects of highway capacity additions on heavy-duty diesel vehicles, particularly heavy-duty trucks, were commissioned for this study.

A concept closely related to capacity is level of service. Although capacity refers to a facility's maximum carrying capacity, highways are rarely designed or planned to operate in this range. Instead, traffic engineers are concerned with estimating the maximum traffic that can be accommodated while maintaining certain operating conditions (TRB 1992, I-3). These conditions, or levels of service, are defined as

ranges in traffic density and are related to qualitative measures, including speed and travel time, freedom to maneuver, traffic interruptions, motorist comfort and convenience, and safety (TRB 1992, I-3). Six levels of service are defined for each type of facility for which analysis procedures are available (see text box). They have letter designations from A to F, with Level-of-Service A representing the best operating conditions and Level-of-Service F the worst (TRB 1992, I-3–I-4). Many capacity enhancements are directed toward improving traffic flow and level of service on a particular facility, and safety is also a consideration in some capacity projects (e.g., left-turn lanes).

The study is focused on additions to highway capacity in metropolitan areas, because the nation's air quality problems are primarily concentrated in its large urban centers. However, the effects of projects may extend beyond current borders. Over time, capacity additions at the urban fringe, which expand development opportunities, may change and expand the geographic boundaries of areas that do not meet air quality standards. Moreover, the decision to add or restrict highway capacity in metropolitan areas could have impacts beyond regional boundaries, affecting the competitive position of one metropolitan area relative to another or of large metropolitan areas generally relative to smaller areas, although systematic evidence of these effects may be difficult to document. For example, businesses could move from congested urban areas that are unable to provide new highway capacity to communities that can; if large metropolitan areas generally cannot provide the additional capacity, growth could be encouraged in medium-size and small cities.

Estimating the effects of highway capacity additions in metropolitan areas on air quality and energy use involves analyzing a lengthy chain of factors (Figure 1-3). In general, the need or demand for travel arises from the distribution of residences and businesses in a region and their activity requirements. Residential and business location in a region are, in turn, determined by historical development patterns, availability of land, and zoning and land use policies. The amount and frequency of travel are affected by regional economic conditions, area

demographic and income characteristics, and the cost and availability of local transportation services.

Regional travel is distributed as vehicle and passenger flows on the supply of transportation facilities—the transportation network —which may include more than one mode (e.g., transit and highways) and more than one form of transport (e.g., automobiles, trucks, bicycles, and walking). Motor vehicles operating on specific links of the network emit pollutants and use energy in varying amounts depending on the type of vehicle, its speed and operating condition (i.e., whether it is warmed up), and the length of the trip. In addition, other factors, such as local topography, meteorological conditions, and other

sources of emissions, interact with vehicle emissions to affect regional air quality.

The primary objective of this study is to examine how an increasein the supply of highway facilities affects each of these factors (Figure 1-3). Initial impacts involve a change in traffic flows on the affected links (e.g., changes in the distribution of vehicle speeds, changes in speed variation, and shifts in traffic volumes) and the resulting change in emissions and energy use from the vehicles on those links. Over time the effects could involve a change in the amount of travel and in land use patterns, with corresponding effects on emissions and energy use, as households and businesses react to the network addition. The incremental effects of all these changes on emissions, air quality, and energy use must be estimated to determine the full effects of the capacity increase.

There is both a spatial and a temporal dimension to these analyses. A change in supply may involve adding capacity and improving traffic flow at only one location on the highway system. However, the network character of the system is likely to affect travel patterns at other locations. Traffic may be diverted from alternate routes, or travelers may shift their time of travel to preferred travel times to take advantage of the new capacity or change their mode of travel if they are encouraged to reduce automobile trips by using transit or bicycle or by walking. If the addition to capacity is sufficiently large, it can induce new or longer trips, influence automobile purchasing decisions, and cause residents or businesses to change their location to take advantage of the improved access. Similarly, emissions from these changes are not confined to the location of the project, but depending on local atmospheric conditions (e.g., heat and wind patterns) may have broader effects on the air quality of the region and beyond. Individual highway projects may not have measurable effects on regional air quality, but the cumulative impacts of many projects could. Thus, whereas a highway capacity enhancement project may be localized, its effects are unlikely to be.

Project impacts will change over time. If a highway capacity addition stimulates travel demand or encourages dispersed development patterns, these outcomes will erode initial gains from some capacity projects. However, some of these impacts, such as decisions about changes in the location of residences and businesses and the traffic

generated by these changes, are likely to take place over a period of years or even decades. Assessing the net effects of highway capacity projects on emissions and energy use depends on the length of time over which impacts are analyzed and how the flow of future effects is valued (i.e., discounted back to the present). Examination of project effects at a single point in time is misleading.

The size and direction of the effects and the certainty of the predictions depend on several factors. The context in which the capacity enhancement is made is critical: whether existing highways are heavily congested and travel demand is constrained; whether the area is heavily developed and thus the potential for growth is limited; or, conversely, whether the area is relatively undeveloped but conditions for growth are present and thus the rate of growth in development and related traffic is likely to be accelerated. Each of these conditions will result in different project outcomes for travel behavior, land use, and urban form.

The scale of effects also depends on the unit of analysis. Highway capacity additions may have significant effects on travel patterns and development activity in a particular corridor or subregional area. However, the effects may be small at the regional level, at least in the 20-year planning horizon required for metropolitan transportation planning and conformity purposes.¹⁹ Highway capacity additions represent small increments to large existing metropolitan transportation networks and to land use patterns that have evolved over many years.

The type of project also matters. For example, minor capacity improvements on arterials—such as signalization improvements and exclusive turn lanes—may smooth traffic flows, initially reducing emissions and energy use, without major offsetting increases in motor vehicle traffic. ²⁰ However, the size of the benefits will be commensurate with traffic levels on these local routes and may have little measurable impact on regional air quality. A series of relatively minor traffic flow improvements can have a cumulatively larger impact on motor vehicle traffic growth. Major capacity enhancement projects, such as a new freeway or a major new bypass, are likely to have more significant initial effects on emissions and energy use (although effects can be negative as well as positive). They may have adverse effects in the longer term because of increased motor vehicle travel stimulated by the project.

The effects of not adding highway capacity also must be considered in any assessment of net effects on highway emissions and energy use. Is there sufficient capacity in the existing highway network for travelers to change routes and times of travel without significantly adding to highway congestion levels and thereby increasing emissions and energy use? Can other investments in transit facilities or bikeways provide alternative modes of transport with sufficient incentive for travelers to switch from automobile travel to keep highway congestion from growing? If highway congestion increases, will residents and businesses respond by moving to less congested suburban or exurban areas or to smaller, less congested metropolitan areas? Or will congestion simply become worse, making the capacity addition more valuable in the future relative to the no-build option?

Finally, the level of certainty of the predictions will vary. Forecasts of emissions resulting from the initial adjustments in traffic flows and travel patterns because of a highway capacity project should be relatively straightforward, although technical issues involved in modeling vehicle emissions and systems issues (i.e., route shifts and changes in time of travel) must be addressed. Forecasting longer-term effects such as shifts in travel demand and land use changes from capacity additions is more complex and requires predicting behavioral responses that are not fully understood. The longer the time horizon for estimating impacts, the more other, often unpredictable, factors such as changes in demographic or economic conditions are likely to intervene, reducing the ability to forecast accurately.

The difficulty of reaching consensus on the direction and size of effects was demonstrated in a recent court case filed in 1989 by the Sierra Club Legal Defense Fund and Citizens for a Better Environment against the Metropolitan Transportation Commission (MTC), the MPO for the San Francisco Bay Area (U.S. District Court 1990a).²¹ Environmental groups sued MTC for noncompliance with the obligations contained in the Bay Area's 1982 SIP to meet federal air quality standards by 1987.

A major issue of the court case concerned the extent to which large highway capacity expansions would adversely affect regional air qual-

ity and MTC's capacity to model and analyze these impacts. Court testimony by experts provides a good summary of the key arguments put forward by local planners and modelers, environmental groups, theoreticians, and practitioners concerning the effect on air quality of increased highway capacity.

Environmental groups argued that adding highway capacity in a congested system would increase vehicle use by making automobile travel easier and more convenient, thereby offsetting at least some of the initial reduction in emissions from smoothing traffic flows.²² Specifically, drivers respond to improved levels of highway service by concentrating work trips more in traditional peak periods; shifting from transit and higher vehicle occupancies to automobile and lower, or single, vehicle occupancies; taking more individual trips rather than combining trips (trip chaining); and taking longer trips or trips that might otherwise have been forgone, ultimately producing levels of congestion comparable with previous conditions but at higher traffic volumes.

In the long run, the argument continued, adding highway capacity in a congested system would encourage further decentralization of households and jobs and longer commuting by making travel to outlying areas easier and faster. Because most major highway capacity investments are being made in suburban areas, these projects would change the distribution of development within the region; low-density, automobile-dependent development at the suburban fringe would be encouraged over redevelopment of core areas that could better support public transit and nonmotorized travel modes. Moreover, if highway transportation investment stimulates economic growth, as project proponents frequently claim, then capacity enhancements should increase overall levels of regional growth, drawing new jobs and households to the area and making it even more difficult for the region to meet environmental standards.

Those who supported the MTC argued that capacity additions under most conditions would result in emission reductions and greater fuel efficiencies from freer-flowing traffic.²³ They conceded that adding highway capacity could increase the number and length of trips and that in the long run this could increase levels of automobile ownership and affect the location of population and housing in the region. However, given the scale of most highway projects in the built environment of major metropolitan areas, they argued that adding

new highway capacity should not induce new travel at levels that would overwhelm initial emissions and energy benefits. Moreover, they said, there is no empirical support for the proposition that highway capacity additions in regions that are already substantially developed make a measurable difference in overall levels of population or employment growth. Transportation is not the primary determinant of regional development; business and residential location decisions are based on a host of factors (e.g., national and regional economic conditions, local labor force skills, tax levels, and public services) of which transportation is one, but not the major, factor. Thus, providing new capacity would simply accommodate more efficiently the inevitable new development.

After 3 years of litigation, the court found that MTC's adoption of additional TCMs met the “reasonable further progress” requirement toward emission reductions outlined in the Bay Area's 1982 SIP (U.S. District Court 1992). In addition, the court ruled in favor of MTC's proposed modified computer modeling techniques for conformity assessment and lifted a temporary ban on several highway expansion projects (U.S. District Court 1991c). The judge did not require MTC to determine whether the new highway capacity would contribute to regional growth and increase total pollution levels, arguing that EPA has not yet made this a requirement and that the capability to model this relationship does not currently exist (U.S. District Court 1991c).

These issues are explored more systematically in the following chapters. Background information on the effects of motor vehicle transportation on air quality and energy consumption is given in Chapter 2, and the main analytic tools used to model these effects are introduced. The impacts, both initial and long term, of additions to highway capacity on traffic flows, travel demand, and land use and urban form are examined in Chapter 3, Chapter 4 through Chapter 5, respectively. In each area, what is known from theory, empirical studies, and statistical models is summarized; an assessment of the validity and certainty of the evidence is provided; and recommendations are made for improving the state of knowledge. In Chapter 6, major findings about the relationships between highway capacity additions, emissions, air quality, and

energy consumption are summarized, highlighting what is known, what is knowable, and what is not, and conclusions are drawn about the current focus and requirements of the CAAA.

1. Catalytic converters, which control for CO, VOC, and NO_x emissions, were introduced in 1978, and carbon canisters, which control hydrocarbon, or VOC, emissions from the fuel system that occur while the car is in operation or while parked by adsorbing the vapors, have been in use since 1975 (NRC 1992, 70).

2. VOCs are also referred to as nonmethane hydrocarbons (NMHC) and as reactive organic gases (ROGs) or total organic gases (TOGs), the terms used in California.

3. Comparative data on emissions from all other sources cannot be provided because the methodologies used to estimate emissions from all sources except highway vehicles changed after 1985. The differences in methodologies result in as much as a 20 percent difference in EPA estimates of CO emission levels in 1985 (Nizich et al. 1994b, 3, 6). Highway vehicle emissions, however, were estimated using a consistent methodology for 1970 through 1993.

4. This population estimate is based on a single year of data and only considers counties with monitoring data for that pollutant (Curran et al. 1994, 14). The data can vary substantially from year to year primarily because of changes in meteorological conditions.

5. VMT has grown at annual rates of 3.5 percent over the past decade (U.S. Congress. House. Committee on Public Works and Transportation. 1993, 37). There is some evidence, however, that future increases in VMT levels may not be as great as in the past because of saturation in vehicle ownership, aging of the driving population, and the like (U.S. Congress. House. Committee on Public Works and Transportation. 1989, 87–91).

6. EPA estimates assume a small increase in the share of diesel vehicles, particularly heavy-duty diesel vehicles, relative to gasoline-powered light-duty vehicles over the 20-year period (personal communication, Natalie Dobie, Technical Support Branch, EPA, Nov. 21, 1994).

7. The following two paragraphs draw heavily from background material on the impact of highway congestion on air quality prepared by Elizabeth Deakin for TRB in March 1991 (Deakin 1991). Data on nonattainment areas were updated using EPA's 1993 National Air Quality Emissions Trends Report (Curran et al. 1994).

8. TCMs are activities intended to decrease automotive travel or otherwise reduce vehicle emissions. TCMs are most closely associated with actions

to improve transit, support ridesharing, and encourage employer-based trip reduction programs (FHWA 1992a, 3–4).

9. In a Notice of Proposed Rulemaking (Federal Register 1993a), EPA proposed an automatic sanctioning procedure. An EPA finding of noncompliance by a state regarding its SIP would trigger an 18-month clock for the imposition of a two-for-one mandate for industrial emissions, requiring an offset of twice the increased emission from a new or modified point source. If the SIP remains deficient after another 6 months, EPA would then cut off federal highway funds.

10. Such projects include transit, HOV lanes, employee trip reduction programs, emission-reducing traffic flow improvements, fringe parking, pricing programs to restrict vehicle use, and accident management programs.

11. Regionally significant projects include any facility with an arterial or higher functional classification and any other facility that serves regional travel needs and that would normally be included in the modeling for the transportation network (Federal Register 1993b, 62,211). Projects that are not regionally significant must also be included in regional emissions analyses, but the effects of these projects, which cannot normally be modeled with a transportation network demand model, may be estimated in accordance with reasonable professional practice (Federal Register 1993b, 62,211). Finally, nonfederal regionally significant projects must have been included in a conforming plan or TIP, or included in the original regional emissions analysis supporting the adoption of the plan or TIP, or a new regional emissions analysis must demonstrate that the plan and TIP would still conform if the projects were included and implemented (Federal Register 1993b, 62,204).

12. During the period after SIPs have been submitted but have not been approved by EPA, MPOs must demonstrate that TIPs pass the build –no-build test and also that the build scenario does not exceed the emissions budget for motor vehicle emissions contained in the submitted SIP (Federal Register 1993b, 62,191).

13. Motor vehicle emission budgets are the motor vehicle-related portions of the projected emission inventory used to demonstrate reasonable further progress milestones, attainment, or maintenance for a particular year specified in a SIP. They establish a cap on emissions that cannot be exceeded by predicted highway and transit vehicle emissions (Federal Register 1993b, 62,194).

14. Six environmental groups filed notice of their intention to sue the transportation departments of New York, New Jersey, and Connecticut for approving transportation improvement programs that fail to meet the 1990 CAAA pollution reduction requirements (AASHTO Journal 1992, 3–4); suits have been filed against Connecticut and Rhode Island (AASHTO Journal 1993, 7–9). The Environmental Defense Fund and other groups

have already brought suit against EPA and DOT to force issuance of various regulations associated with the legislation.

15. For example, among other provisions, metropolitan transportation plans must cover a period of at least 20 years, contain all regionally significant projects, consider the likely effect of transportation policy decisions on land use and development, and, for nonattainment or maintenance areas, contain only conforming projects (Federal Register 1993c).

16. The sales-weighted fuel economies of automobiles and light trucks have, on average, met the fuel economy standards set by the federal government. This does not mean, however, that each manufacturer is meeting the standards each year. Some manufacturers still fall short, whereas others exceed the standards (Davis 1994, 3–51).

17. This does not include signalization and other traffic flow improvement projects, which are not broken out separately.

18. The time period used in most capacity analysis is 15 min, the shortest interval during which stable flow exists. Capacity is defined for prevailing roadway conditions (which refer to the geometric characteristics of the facility), traffic conditions (which refer to the characteristics of the traffic stream using the facility), and control conditions (which refer to the types and specific design of control devices and traffic regulations on a given facility); good pavement and weather conditions are assumed.

19. ISTEA requires the metropolitan transportation plan to address a period of at least 20 years (Federal Register 1993c, 62,210).

20. To the extent these projects degrade bicycle and pedestrian travel, they may induce some shifts to cars, thus increasing emissions.

21. The lawsuit was filed in the Federal District Court of Northern California. The citation for the consolidated cases is Citizens for a Better Environment et al. v. Peter B. Wilson et al., Civil No. C-89-2044-TEH, and Sierra Club vs. Metropolitan Transportation Commission et al., Civil No. C-89-2064-TEH.

22. The arguments of those who view highway capacity improvements as having a negative effect on air quality are drawn from three sources: (a) background material prepared for TRB by Elizabeth Deakin (Deakin 1991), (b) testimony by Dr. Peter Stopher, Professor of Civil Engineering, Louisiana State University, and Director of the Louisiana Transportation Research Center, in the MTC case (U.S. District Court 1990b), and (c) a paper by Greig Harvey and Elizabeth Deakin (Harvey and Deakin 1991) on regional transportation modeling practice.

23. The arguments of those who view highway capacity improvements as having a positive or neutral effect on air quality are drawn from the background material provided by Deakin (1991), from the Harvey and Deakin 1991 paper, and from the testimony by Elizabeth Deakin, Assistant Professor of City and Regional Planning and member of the research staff of

the Institute of Transportation Studies at the University of California, Berkeley (U.S. District Court 1991a), and Dr. Raymond Brady, Research Director for the Association of Bay Area Governments (U.S. District Court 1991b) in the MTC case.

EPA

Environmental Protection Agency

FHWA

Federal Highway Administration

NRC

National Research Council

TRB

Transportation Research Board

AASHTO Journal. 1992. States May Face Clean Air, ISTEA Suit. Sept. 25, pp.3–6.

AASHTO Journal. 1993. Connecticut, Rhode Island Hit with Environmental Suits. March 26, pp. 7–9.

Curran, T., T. Fitz-Simons, W. Freas, J. Hemby, D. Mintz, S. Nizich, B. Parzygnat, and M. Wayland. 1994. National Air Quality and Emissions Trends Report, 1993. 454-R-94-026. U.S. Environmental Protection Agency. Research Triangle Park, N.C., Oct.

Davis, S.C. 1994. Transportation Energy Data Book: Edition 14. ORNL-6798. Center for Transportation Analysis, Energy Division, Oak Ridge National Laboratory, Tenn., May.

Deakin, E. 1991. Scoping Study: Impact of Highway Congestion on Air Quality. University of California at Berkeley, March, 22 pp.

Federal Register. 1993a. Application Sequence for Clean Air Act Section 179 Sanctions. Vol. 58, No. 189, Oct. 1, pp. 51, 270–51, 279.

Federal Register. 1993b. Criteria and Procedures for Determining Conformity to State or Federal Implementation Plans of Transportation Plans, Programs, and Projects Funded or Approved Under Title 23 U.S.C. or the Federal Transit Act . Vol. 58, No. 225, Nov. 24, pp. 62, 188–62, 253.

Federal Register. Part II. 1993c. Statewide Planning. Metropolitan Planning. Vol. 58, No. 207, Oct. 28, pp. 58, 040–58, 179.

FHWA. 1992a. Transportation and Air Quality: Searching for Solutions: A Policy Discussion Series. No. 5, FHWA-PL-92-029. U.S. Department of Transportation, Aug., 30 pp.

FHWA. 1992b. A Summary: Transportation Programs and Provisions of the Clean Air Act Amendments of 1990. FHWA-PD-92-023. U.S. Department of Transportation, Oct.

FHWA. 1992c. A Summary: Air Quality Programs and Provisions of the Intermodal Surface Transportation Efficiency Act of 1991. FHWA-PD-92-022. U.S. Department of Transportation, Aug.

Harvey, G., and E. Deakin. 1991. Toward Improved Regional Transportation Modeling Practice. Deakin Harvey Skabardonis, Inc., Berkeley, Calif., Dec., 68 pp.

Nizich, S.V., T.C. McMullen, and D.C. Misenheimer. 1994a. National Air Pollutant Emission Trends, 1900–1993. EPA-454/R-94-027. Office of Air Quality Planning and Standards, Research Triangle Park, N.C., Oct., 314 pp.

Nizich, S.V., W.R. Barnard, and D.Y. Linderman. 1994b. Preparation of a National Emission Data Base for Trends and Other Analyses. EPA and E.H. Pechan & Associates, Inc. Presented at the Emission Inventory Specialty Conference, Air and Waste Management Association, Nov. 1–3, Raleigh, N.C., 12 pp.

NRC. 1992. Automotive Fuel Economy: How Far Should We Go? National Academy Press, Washington, D.C., 259 pp.

TRB. 1991. Special Report 232: Advanced Vehicle and Highway Technologies. National Research Council, Washington, D.C., 90 pp.

TRB. 1992. Special Report 209: Highway Capacity Manual (2nd edition revised). National Research Council, Washington, D.C.

U.S. Congress. House. Committee on Public Works and Transportation. 1989. The Status of the Nation's Highways and Bridges: Conditions and Performance and Highway Bridge Replacement and Rehabilitation Program 1989. Committee Print 101-2. Government Printing Office, June.

U.S. Congress. House. Committee on Public Works and Transportation. 1993. The Status of the Nation's Highways, Bridges, and Transit: Conditions and Performance. Committee Print 103-2. Government Printing Office, March.

U.S. District Court for the District of Northern California. 1990a. Memorandum of Points and Authorities in Support of Revised Conformity Assessment Procedures. Civil No. C-89-2044-TEH and C-89-2064-TEH [consolidated], July 3, pp. 9–10.

U.S. District Court for the District of Northern California. 1990b. Declaration of Dr. Peter R. Stopher in Support of Sierra Club's Objections to MTC's Proposed Conformity Assessment. Civil No. C-89-2044-TEH and C-89-2064-TEH [consolidated], Aug. 20.

U.S. District Court for the District of Northern California. 1991a. Declaration of Elizabeth Deakin in Support of MTC's Proposed Revised Conformity Assessment Procedure. Civil No. C-89-2044-TEH and C-89-2064-TEH [consolidated], Feb. 25.

U.S. District Court for the District of Northern California. 1991b. Declaration of Raymond J. Brady in Regard to Metropolitan Transportation Commission's Proposed Conformity Assessment Procedures. Civil No. C-89-2044-TEH and C-89-2064-TEH [consolidated], Feb. 26.

U.S. District Court for the District of Northern California. 1991c. Henderson Court Order Approving MTC Procedures. Civil No. C89-2064-TEH (Consolidated Cases), March 11.

U.S. District Court for the District of Northern California. 1992. Order. Civil No. C89-2064 TEH (Consolidated Cases), May 11.

Motor vehicles run on fossil fuels, emitting pollutants that are a major cause of poor air quality in metropolitan areas and consuming a large fraction of the nation's petroleum resources. In this chapter, the impacts of motor vehicle transportation on air quality and energy consumption are described and the models that are commonly used to analyze these impacts are introduced.

The four principal sources of polluting emissions from man-made sources are transportation (primarily highway vehicles), stationary fuel combustion (especially electrical utilities), industrial processes such as chemical refining, and solid waste disposal (Horowitz 1982, 21). Emissions that either directly cause or combine to form pollution (called primary and secondary pollutants, respectively) may also be classified as stationary, area, or mobile, depending on the magnitudes and geographical distributions of their emissions (DOT and EPA

1993, 2).¹ Pollutants from motor vehicle transport, the focus of this study, are commonly referred to as mobile source emissions.

To comply with the requirements of the 1970 Clean Air Act, the Environmental Protection Agency (EPA) developed national ambient air quality standards (NAAQS) that set allowable concentration and exposure limits for six pollutants considered harmful to public health. The NAAQS are expressed as average concentrations of pollutants over some period of time (see box).² EPA tracks both the emissions or flows of harmful materials from polluting activities, such as factories and transportation, on the basis of the best available engineering and modeling estimates, and the accumulation of these emissions in the air as concentrations of pollutants, which are directly measured at selected sites throughout the country (Curran et al. 1994, 20; Horowitz 1982, 3–4). Regulatory activities are directed toward attainment of NAAQS (i.e., concentration standards), because health effects are directly related to public exposure to pollutants at specific concentrations.

According to current estimates, transportation sources account for about 45 percent of nationwide emissions of EPA's six criteria pollutants. The range is considerable for each pollutant source (Table 2-1) and there is a high degree of uncertainty with respect to many of the estimates. Ground-level ozone is the most pervasive of the transportation-related pollutants; in 1993 approximately 51 million persons lived in counties that exceeded the ozone standard. Nearly 12 million persons lived in counties that did not meet the carbon monoxide (CO) standard in the same year (Curran et al. 1994, 15).

Highway vehicles are the largest source of transportation-related emissions for nearly every type of pollutant (Table 2-1). In total, they contribute slightly more than one-third of nationwide emissions of the six criteria pollutants.

The primary sources of motor vehicle emissions are exhaust emissions from chemical compounds that leave the engine through the tail pipe system and the crankcase and evaporative emissions from the fueling

system [mainly volatile organic compounds (VOCs)] (NRC 1992, 69). For most motor vehicles (i.e., those powered by gasoline), exhaust emissions are formed in a two-stage process: emissions originate as a result of the combustion of fuel in the engine (engine-out emissions) and are then reduced by passing through a catalytic converter (tail pipe or exhaust emissions). For diesel-powered vehicles, the process of producing exhaust emissions is simpler, because there is presently no aftertreatment (i.e., catalytic converter).

Carbon monoxide and VOCs are the product of incomplete combustion of motor fuels and, in the case of VOCs, of fuel vapors emitted from the engine and fuel system (NRC 1991, 257). Oxides of nitrogen (NO _x) are formed differently; they are the product of high-temperature chemical processes that occur during the combustion process itself (NRC 1991, 261). Particulates, another compound mainly found in diesel exhaust, are formed primarily from incomplete combustion of diesel fuel and lubricating oil (Weaver and Klausmeier 1988, 2–7; Conte 1990, 58).

The air/fuel (A/F) ratio, which is controlled by the carburetor or fuel injection system, is the most important variable affecting the efficiency of catalytic converters and thus the level of exhaust emissions (Johnson 1988, 40). Because concentrations of key emissions are not at a minimum at the same A/F ratio (CO and VOCs are highest under fuel-rich conditions and NO_x is highest under fuel-lean conditions), manufacturers must optimize catalytic converter operation within a narrow A/F ratio range, known as stoichiometry, to achieve the greatest control efficiency for all three pollutants (Figure 2-1).

Transportation is the dominant source of U.S. CO emissions, and highway vehicles contribute nearly two-thirds of the total (Table 2-1). Carbon monoxide is an odorless gas that forms from incomplete combustion of motor fuels. The higher the share of fuel in the air-fuel mixture, the more CO is produced (NRC 1992, 69). Fuel-rich operations occur under cold-start conditions, when the vehicle has been turned off for some time and the catalytic converter is cold, or under heavy engine loads (e.g., during rapid accelerations, on steep grades, or at high speeds). CO concentrations tend to be high on and near con-

gested roadways and at other locations where traffic densities are high. These concentrations are often referred to as CO hot spots. However, CO can also be viewed as a regional problem, with frequent reported exceedances of the 8-hr average concentration standards. Finally, CO contributes indirectly to greenhouse gas emissions (Gordon 1991, 60).³

Motor vehicles are also a major contributor to smog, the haze that hangs over many large urban areas, which has harmful health effects, contributes to the greenhouse problem, and adversely affects crops and vegetation (MacKenzie and Walsh 1990, 7). Ground-level ozone, an important constituent of smog, is not emitted directly into the atmosphere. Rather it is formed as a secondary pollutant through a chemical reaction between the ozone precursors, VOCs and NO_x, which is stimulated by heat and sunlight (Gordon 1991, 62). Highway vehicles account for about one-quarter of total VOC emissions and about one-third of total NO_x emissions (Table 2-1). EPA's estimates of VOC emissions, in particular, have been challenged in a National Research Council report as understating actual emission levels by a factor of 2 to 4 (NRC 1991, 7).⁴

Because it is a chemically reactive pollutant, ozone behaves quite differently from CO. The relation between ozone concentrations and VOC and NO_x emissions is both nonlinear and synergistic; thus, changes in VOC and NO_x emissions can have impacts on ozone that are difficult to predict. For example, ozone concentrations often are lower near large sources of motor vehicle emissions, because exhaust emissions of nitrogen oxide (NO) break down the ozone molecule.⁵ This is referred to as ozone scavenging. Also, spatial variations in ozone concentrations tend to be much more gradual than in CO concentrations (Horowitz 1982, 63).

The role of NO_x in urban ozone pollution has received attention recently from the scientific and regulatory communities. NO_x emissions from motor vehicles consist of a mixture of NO and nitrogen dioxide (NO₂) (NO being the dominant constituent), which is formed by high-temperature chemical processes during the combustion of fossil fuels (Horowitz 1982, 17; NRC 1991, 261). High concentrations of NO₂, which are responsible for the yellowish-brown color of the sky in many smoggy areas, are caused primarily by the oxidation of NO from engine exhaust and other sources to NO₂ through the chemical

processes that produce ozone (Horowitz 1982, 77). The NRC report (1991) argued that it is the balance between ambient levels of VOCs and NO_x that determines ozone levels in a particular area, and that efforts to reduce NOx may be the most effective ozone abatement strategy in many of the nation's most polluted urban areas (NRC 1991, 7).⁶ In its final conformity regulations following the Clean Air Act Amendments of 1990 (CAAA), EPA requires that transportation improvement programs proposed by metropolitan planning organizations (MPOs) show reductions in NO_x as well as VOCs from a 1990 baseline scenario (Federal Register 1993, 62,226).⁷

Particulates from diesel-fueled vehicles, primarily trucks and buses, contribute to pollution from inhalable particulate matter (PM-10). ⁸ Overall, tail pipe emissions of highway vehicles account for less than 1 percent of total PM-10 emissions (Table 2-1). The major source of particulates is road dust, which is a function of vehicle traffic, wildfires, and agricultural activity (Curran et al. 1994, 53). Particulate emissions are raising renewed concern because of medical evidence of their contribution to lung cancer (Dockery et al. 1993, 1753).

Transportation no longer accounts for a large share of pollution from lead; use of unleaded gasoline has resulted in a 99 percent reduction in total lead emission levels from highway vehicles since 1970 (Nizich et al. 1994, 3-16).

Finally, transportation is not a major contributor to sulfur dioxide (SO₂) (Table 2-1). Although heavy trucks and buses emit oxides of sulfur because of the high sulfur content of diesel fuel, coal-fired electric utilities are the dominant source of SO₂ emissions (Curran et al. 1994, 12).

Emissions of specific pollutants vary by vehicle and fuel type. The primary emissions of gasoline-powered, passenger vehicles—the most common vehicle on the road—are CO, followed by much smaller emissions of VOCs and NO_x. Most heavy-duty diesel vehicles—combination trucks and buses—use diesel fuel.⁹ Their primary emissions are NO_x, followed by smaller emissions of CO, PM-10, SO₂, and VOCs (Figure 2-2).

FIGURE 2-2 Comparison of national emission estimates (measured in short tons) for gasoline-powered, light-duty automobiles and diesel-powered heavy-duty vehicles (Nizich et al. 1994, A-4, A-8, A-15, A-19, A-24).

Heavy-duty diesel vehicles produce about 5 percent of total emissions from all highway vehicles, roughly proportional to their share of highway travel but small compared with the emissions of gasoline-powered passenger vehicles, which represent nearly two-thirds of total emissions from highway vehicles.¹⁰ Diesel-powered vehicles, however, contribute a disproportionate share of total highway vehicle emissions of PM-10, SO₂, and NO_x: 72, 47, and 27 percent, respectively (Nizich et al. 1994, A-8, A-19, A-24).

High levels of NOx emissions from heavy-duty vehicles are caused by the characteristics of diesel engines. Diesel engines typically run at higher combustion chamber pressures and temperatures than gasoline engines (Lilly 1984 in Guensler et al. 1991, Appendix A). Both conditions are conducive to high NO_x emission levels.

PM-10 and SO₂ emissions are also higher for heavy-duty diesel vehicles than for gasoline-powered automobiles. Catalytic converters

have not been used with diesel engines because of particulates and concentrated sulfur gases in the exhaust gas, which could clog or deactivate the catalyst (Guensler et al. 1991, Appendix A).

Particulates in diesel exhaust originate mainly from unburned fuel and oil (Weaver and Klausmeier 1988, 2–7; Conte 1990, 59, 61).¹¹ However, introducing higher combustion temperatures to burn the fuel more completely and reduce particulates leads to higher NO_x emissions. The challenge facing diesel engine manufacturers is to reduce emissions of both pollutants at the same time to meet NO_x and particulate standards.

Emissions of SO₂ are also substantially higher for diesel than for gasoline engines because of the high sulfur content of diesel fuel. However, mandatory use of low sulfur or “clean” diesel fuel, which began in October 1993, should substantially reduce SO₂ emissions as well as PM-10 emissions¹² from heavy-duty, diesel-powered vehicles.

Actual emission levels from transportation sources are a function of several variables that can be grouped under four main categories: travel-related factors, driver behavior, highway network characteristics, and vehicle characteristics. Highway projects that add capacity and smooth traffic flows should affect emissions related to travel, driving patterns, and physical characteristics of the highway itself.

Emissions are a function of trip taking as well as distance traveled. Trips matter because emissions vary depending on the share of the trip associated with different vehicle operating modes. Exhaust emissions, one of the major sources of emissions from motor vehicle operation, include vehicle start-up emissions (start-ups are classified as cold or hot starts depending on how long the vehicle has been turned off ¹³) and running emissions, which occur when the vehicle is warmed up and operating in a hot stabilized mode (Sierra Research 1993, 18, 19). Evaporative emissions, the other major source, consist entirely of

VOCs. They include running losses, which occur when the vehicle is operating in a hot stabilized mode; hot soak emissions, which result from fuel evaporation from the still-hot engine at the end of a trip; and diurnal emissions, which result from evaporation of fuel from the gasoline tank whether the vehicle is driven or not (Sierra Research 1993, 19, 20).14

Vehicle technology improvements have been focused primarily on reducing running emissions, which are a function of vehicle miles traveled (VMT). However, vehicle emissions from a cold start when the catalytic converter is not functioning at optimal temperatures, which are a function of trip making rather than VMT, can account for more than half of total CO and VOC emissions (FHWA 1992, 6).15

The importance of trips relative to VMT is most evident for VOC emissions as illustrated by the following example of a prototypical 32-km (20-mi) trip (Figure 2-3). In this example, vehicle start-up contributes approximately one-third of total VOC emissions and trip end contributes one-sixth. Neither of these emissions is a function of VMT, but together they account for about half of the total VOCs emitted.16 Thus, the impact of highway capacity additions on trips as well as VMT is of interest in assessing the effect on emissions.

Emission levels depend not only on the number of trips taken and VMT but also on the speed and acceleration of the vehicle and the load on the engine over the distance of the trip.17 In current emission models, vehicle speed and acceleration are combined into a single average speed for various trip types (i.e., drive test cycles) so that emission levels vary with average trip speed. The severe limitations of this approach are discussed in Chapter 3. Engine loads are generally not varied to reflect different vehicle operating or highway conditions (e.g., road grade) in modeling emission estimates.18

Figure 2-4, Figure 2-5 through Figure 2-6 show current model estimates of emission factors for key pollutants expressed in grams per mile for light-duty, gasoline-powered automobiles representing the 1990 fleet mix, for a range of average trip speeds. 19 The data are based on the most recent emission models—MOBILE5a developed by EPA and EMFAC7F developed by the California Air Resources Board (CARB) and approved

FIGURE 2-3 Sources of VOC emissions by type for a prototypical trip (Outwater and Loudon 1994, 17). Based on a 32-km (20-mi) round-trip in 1990 for a light-duty automobile traveling in the Bay Area of California at an average speed of 64 kph (40 mph).

by EPA for use in that state. A more detailed discussion of the test procedures and models used to estimate emissions is included in the final section of this chapter and in Chapter 3.

According to the estimates, emissions are generally highest in low-speed, congested driving conditions. In intermediate-speed, freer-flowing traffic conditions, emissions fall. They rise again at high

FIGURE 2-4 Comparison of MOBILE5a and EMFAC7F emission factors for carbon monoxide as a function of average trip speed, 1990 fleet average for light-duty gasoline vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

speeds, but not to initial levels. NO_x emissions are the exception; they rise at relatively low speeds and are highest at high speeds. Confidence in current speed-emission relationships, however, is severely limited by test data that poorly represent urban driving conditions and large variances in emission rates for a wide range of changes in average trip speeds. These topics are discussed further in Chapter 3.

Emission factors for heavy-duty diesel vehicles (also representing the 1990 fleet mix) as a function of average trip speed are shown in Figure 2-7, Figure 2-8 through Figure 2-9. The data show emissions of CO and VOC falling as speeds rise, but at high speeds they remain relatively constant (Figure 2-7 and Figure 2-8). Emissions of NO_x have the same characteristic U-shaped curve as for automobiles ( Figure 2-9). Esti-

FIGURE 2-5 Comparison of MOBILEa and EMFAC7F emission factors for volatile organic compounds as a function of average trip speed, 1990 fleet average for light-duty gasoline vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

mates for particulate emissions as a function of speed are not available. Particulate emissions are believed to follow the same trend as VOC emissions because both result from incomplete combustion of motor fuels (see Appendix A). However, the behavior of diesel particulates at high speeds is not well understood (see Appendix A). The final section on modeling describes the test procedures for estimating diesel emissions; the certainty of these estimates is discussed in Chapter 3.

Because highway capacity additions will increase average vehicle operating speeds, at least initially, they will have a direct effect on emission levels. However, the applicability of current average speed-emission relationships to assessing the emission effects of specific

FIGURE 2-6 Comparison of MOBILE5A and EMFAC7F emission factors for oxides of nitrogen as a function of average trip speed, 1990 fleet average for light-duty gasoline vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

capacity-enhancing projects is problematic. The emission factors relate to average trip speed, not to the portion of the trip on which the speed improvement is being made. Furthermore, by focusing on average speed, many of the variables of interest, such as the emission effects of traffic flow smoothing from a capacity project, cannot be analyzed directly. These limitations will be addressed more fully in the following chapter.

Recent research suggests that emissions are also affected by smoothness and consistency of vehicle speed, which are heavily affected by

FIGURE 2-7 Comparison of MOBILE5a and EMFAC7F emission factors for carbon monoxide as a function of average trip speed, 1990 fleet average for heavy-duty diesel vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

driving behavior and traffic conditions. Sharp accelerations from passing or changing lanes, merging onto a freeway from a ramp, or leaving a signalized intersection impose heavy loads on the engine that result in high emission levels.

Vehicle accelerations produce emissions because of an engine operating strategy called power enrichment. When heavy loads are placed on the engine by acceleration, vehicles are designed to operate with a richer fuel-to-air mixture (more fuel than air) to prevent engine knock and damage to the catalytic converter (EPA 1993, 19). This provides good driving performance but causes the catalytic converter to be overridden (there is insufficient oxygen for efficient performance of the catalyst), thereby producing high levels of emissions (EPA 1993,

FIGURE 2-8 Comparison of MOBILE5a and EMFAC7F emission factors for volatile organic compounds as a function of average trip speed, 1990 fleet average for heavy-duty diesel vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

19). Carbon monoxide emissions are most affected, followed by VOCs; there is little effect on NOx emissions.²⁰

Research is under way to measure the effects of accelerations on emission levels. Although widespread testing of a range of vehicle types and model years has not been completed, testing of individual vehicles suggests that the effects can be large. The staff of CARB, for example, have reported that for some vehicles, one heavy acceleration may produce more VOC emissions than the remainder of a 16-km (10-mi) trip (FHWA 1992, 29). Sierra Research has reported that aggressive driving (with many accelerations) results in CO emission levels 15 times higher, and VOC levels 14 times higher, than those resulting from “average” driving. These results were obtained by

FIGURE 2-9 Comparison of MOBILE5a and EMFAC7F emission factors for oxides of nitrogen as a function of average trip speed, 1990 fleet average for heavy-duty diesel vehicles (data from Sierra Research, June 1994). Note: confidence intervals around point estimates may be large and encompass positive and negative values. See discussion of uncertainty about emission estimates from EMFAC by Guensler (1994, Chapter 13) and in Chapter 3. 1 mi = 1.6 km.

comparing time-speed-emission traces for the same 11-km (7-mi) trip from downtown to an outlying area (Figure 2-10 and Figure 2-11).

Highway capacity additions should initially smooth traffic flows; researchers have found reductions in heavy accelerations and greater frequency of cruise-type driving at higher speeds (Effa and Larsen 1993, 3). Capacity additions should thus reduce emissions from accelerations. In the long run, it is likely that extreme impacts from power enrichment will gradually be ameliorated as new emission standards are established. Both EPA and CARB are presently working on certification test cycles for new vehicles that will include heavy accelerations. However, these new standards are not expected to be in place until the late 1990s.

The physical characteristics of the highway network itself can affect emission levels. The presence of highways with long grades, freeway ramps, signalized intersections, major arterials with numerous driveways and significant volumes of traffic entering the traffic flow, and rough pavement are all network conditions that can increase emission levels primarily because of engine enrichment from accelerations as described above (Meyer et al. 1993, 227). Certain highway capacity additions—including improvements to freeway ramp geometry, intersection reconstruction, grade separation of major crossings, synchronization of traffic signals, and other measures to improve traffic flow such as the application of intelligent transportation system technologies —should reduce these emission-creating situations.

Although not directly affected by highway capacity additions, other factors such as vehicle characteristics, fuels, and temperature can interact with changes in highway capacity and alter their effect on emissions. Future changes in vehicle technologies, fuels, and inspection and maintenance programs to reduce emissions will also reduce the emissions impact of traffic flow pattern changes due to highway capacity additions.

Emissions vary with vehicle age. Older carbureted vehicles have less precise fuel control than newer fuel-injected vehicles, resulting in higher average emissions during normal operation and vehicle starts and a greater incidence of malfunctions (Enns et al. 1993, 7–8). In addition, older vehicles have not met the more stringent emissions standards for the newer vehicle fleet. For example, 1990-model vehicles emit VOCs and CO at only one-third the rate of 1975-model vehicles (DOT and EPA 1993, 33).

Vehicle condition—whether the vehicle is well maintained, whether the catalytic converter has been tampered with or is malfunctioning—is even more critical than vehicle age in determining total emission levels. In conducting surveys of vehicle condition, researchers have documented the extent of tampering, particularly in areas without inspection and maintenance programs, and its effect on

emission levels (Greco 1985 in Johnson 1988, 50–51). More recently, using a roadside measuring device, researchers have been able to test the emissions of vehicles in operation, confirming that a few vehicles, commonly known as superemitters, account for a large share of emissions of CO and VOCs (Stedman 1991; Naghavi and Stopher 1993, 9–10).²¹ Two additional speed-time-emission traces from Sierra Research ( Figure 2-12), which can be compared with Driver B in Figure 2-10 and Figure 2-11, show the effect on CO and VOC emissions of driving with the catalytic converter disconnected; emissions are approximately 7 to 10 times larger, respectively. The CAAA requirements for enhanced vehicle inspection and maintenance programs in areas that do not meet air quality standards—annual or biennial high-technology emission testing supplemented with on-the-road testing—should help reduce the number of vehicles that are out of compliance with emission standards.²² Because emissions of superemitters are always high, however, their emission levels are generally less sensitive to changes in speed and sharp accelerations than are those of other vehicles.

As discussed earlier, the mix of vehicles, particularly the amount of heavy truck traffic, affects the types of pollutants emitted. Substantial truck traffic can affect the behavior of other vehicles and their emission levels. For example, slower-moving trucks, particularly in locations with grades, can contribute to more frequent accelerations and decelerations by passenger vehicles as they attempt to navigate around the trucks.

The type of fuel used also has important ramifications for emission levels. The nearly universal use of lead-free gasoline has effectively eliminated lead as a pollutant from transportation sources. The use of reformulated and oxygenated fuels as well as low-sulfur diesel fuel, and the introduction of clean fuels programs for vehicle fleets required by the CAAA in the nation's most polluted areas, will also help reduce emission levels.

Finally, ambient temperature affects both exhaust and evaporative emissions. Exhaust emissions increase below 24°C (75°F); at colder temperatures the engine and emission control system take longer to warm up, increasing cold start emissions (Sierra Research 1993, 122). Evaporative emissions increase above about 24°C, with higher emis-

sion rates the higher the maximum temperature (Sierra Research 1993, 125).²³

Examining the changes in vehicle emissions that will result from highway capacity additions is only the first step in understanding how these emissions are likely to be dispersed in the atmosphere and affect the air quality of a metropolitan area.

Meteorological conditions (e.g., wind and temperature) have a major effect on the transport and dispersion of emissions and hence their concentration in a region, which is the primary criterion of concern from a public health perspective (Horowitz 1982, 39). For example, the highest concentrations of CO generally occur in the winter, when atmospheric conditions tend to be more stable and wind speeds are lower, causing reduced dispersion and increased concentrations of CO (Horowitz 1982, 45). Because the chemical reaction that creates ozone is stimulated by heat and sunlight, ozone concentrations tend to be higher from midspring to midfall than during the rest of the year (Horowitz 1982, 74). The atmospheric mixing and transport of ozone concentrations long distances from the sources of the precursors is a well-known phenomenon (Horowitz 1982, 74).

Local topography—both man-made (e.g., tall buildings near a highway) and natural (e.g., mountains)—can also affect the rate of dispersion of emissions from their source.

The dominant energy source for the transportation sector is petroleum (96.6 percent), and nearly two-thirds of the petroleum consumed in the United States is in this sector (Davis 1994, 2-7, 2-10). The highway mode accounts for nearly three-fourths of total transportation energy use with about 80 percent from automobiles, light trucks, and motorcycles and about 20 percent from heavy trucks and buses (Davis 1994, 2-16). Petroleum energy users are heavily dependent on imported oil; nearly half of all petroleum consumed in the United States comes from foreign sources (Davis 1994, 2-5).

In addition to energy dependence, concerns about global warming have stimulated interest in improving motor vehicle fuel efficiency. Global warming occurs from the emission of carbon dioxide (CO₂) and other gases into the upper atmosphere, which trap heat and warm the earth; hence the term greenhouse effect (Gordon 1991, 55; Greene et al. 1988, 215). Although there is great uncertainty about the likely climatic changes from the greenhouse effect and the timing of these changes, transportation's contribution to the problem is well understood. Carbon dioxide, the principal greenhouse gas, is a by-product of any engine that burns carbon-based, fossil fuels. The U.S. fleet of gasoline-powered automobiles and light trucks contributes about onefifth of total U.S. CO₂ emissions (NRC 1990 in NRC 1992, 71). Improvements in fuel economy will result in reduced CO₂ emissions. For example, a 10 percent reduction in the fuel consumption of the vehicle fleet will result in an estimated 2 percent reduction in total U.S. emissions of CO₂ (NRC 1992, 71).

Improved fuel economy could also reduce other automotive emissions, such as VOC emissions, which affect ground-level pollution. Improvements in fuel economy may or may not reduce tail pipe VOC exhaust emission levels, depending on corresponding manufacturer changes to emission control systems (DeLuchi et al. 1993, 8).²⁴ If demand for gasoline is reduced, total VOC emissions are likely to be lower because of reduced motor vehicle evaporative emissions and reduced emissions from fuel extraction, refining, processing, distribution, and vehicle refueling (DeLuchi et al. 1993, 3).

The primary factors affecting fuel economy fall into the same categories as those affecting emissions. Highway projects that add capacity, increasing average vehicle speeds and smoothing traffic flows, will have the greatest effect on those fuel economy factors related to travel conditions, driver behavior, and the physical characteristics of the highway.

FIGURE 2-13 Fuel economy as a function of cruise speed (Davis 1994, 3-56) and average trip speed (An and Ross 1993b, 76). Note: 1 mi = 1.6 km; 1 gal = 3.8 liters.

Fuel consumption is a function of different traffic characteristics. Fuel efficiency under steady state, cruise-type driving conditions, measured in miles per gallon for a sample of automobiles and light trucks, peaks at speeds of 56 to 72 kph (35 to 45 mph) and then rapidly declines at higher speeds (Figure 2-13). At lower speeds, engine friction, tires, and accessories (e.g., power steering and air conditioning) reduce fuel efficiency; at high speeds the effect of aerodynamic drag on fuel efficiency, which increases exponentially as a function of speed, dominates (An and Ross 1993a, Figure 2).

Fuel efficiency under start-and-stop traffic conditions shows somewhat different patterns. The data points in Figure 2-13 show modeled estimates of fuel economy as a function of the average trip speed of various drive-test cycles, which are based on different trip

types, locations, and traffic conditions (An and Ross 1993b, 76). Fuel economy is somewhat poorer at lower average trip speeds, reflecting greater amounts of acceleration and stopping. At higher average trip speeds, the modeled estimates tend to converge with the cruise speed results, reflecting more cruise-type driving and less acceleration and stopping at these speeds. The results confirm earlier research findings that of all the travel-related factors affecting fuel economy, average vehicle speed explains most of the variability in fuel consumption and is a good predictor of fuel economy for most urban trips (Evans et al. 1974, 16; Murrell 1980, 132). The effect on fuel economy of highway capacity additions, which initially raise average vehicle speeds, depends on average traffic speeds before and after the project.

The modal operation of the vehicle also affects fuel consumption. Vehicles get lower fuel efficiency when started cold than when fully warmed up (Murrell 1980, 142).

Fuel economy is sensitive to driving behavior, including accelerations (Murrell 1980, 156), braking, and gear shifting. Aggressive braking and accelerations are both associated with reduced fuel economy. Energy is lost from braking as it is dispersed into heat. Repeated braking can account for as much as 15 percent of fuel use in an urban driving trip (An et al. 1993, 5). Aggressive accelerations result in higher engine speeds and greater fuel consumption than constant “cruise” driving.²⁵ Researchers have estimated that, in a congested urban setting, aggressive driving with rapid accelerations will result in a 10 percent increase in fuel use (An et al. 1993, 4). This is a considerably smaller impact, however, than the impact on emissions (Figure 2-10 and Figure 2-11).

Thus, highway capacity additions that smooth traffic flows, reducing the incidence of sharp accelerations and rapid braking, should improve fuel economy initially.

The characteristics of the highway itself can affect fuel economy. For example, steep grades and rough roads reduce fuel efficiency, the former from increased fuel use as a function of heavy loads on the en-

gine and the latter from increased rolling resistance and aerodynamic drag (Murrell 1980, 119, 121). To the extent that highway capacity additions improve these conditions, fuel efficiency can be gained.

The primary vehicle-related factors affecting fuel economy are vehicle weight and technology. In general, larger and heavier passenger vehicles, vehicles with automatic transmissions, and vehicles with more power accessories (e.g., power seats and windows, power brakes and steering, and air conditioning) all consume more fuel on the average (Murrell 1980, 126, 199, 202).

Vehicle maintenance is also a factor. For example, wheels that are out of alignment and tires that are underinflated increase rolling resistance and hence degrade fuel economy (Murrell 1980, 176, 179). Out-of-tune engines as well as inadequate lubrication to reduce engine friction also take their toll on fuel economy (Murrell 1980, 84, 180).

The mix of vehicles, particularly the amount of heavy truck traffic, affects total fuel consumption. The fuel economy for the U.S. fleet of heavy combination trucks currently averages about 2.4 kpl (5.6 mpg), whereas the fuel economy for the population of passenger vehicles averages about 9.2 kpl (21.6 mpg) (Davis 1994, 3-24, 3-39). However, relative to their weight, heavy-duty diesel-powered trucks are very fuel efficient (O'Rourke and Lawrence 1993, 11).²⁶

Finally, climate is a factor. Fuel economy is poorer at low temperatures (Murrell 1980, 107) and with high winds, which result in aerodynamic losses (Murrell 1980, 114).

A range of models are available to estimate the effects of motor vehicle transportation on air quality and energy use. However, many of them are not well suited to the more specific task of estimating the effects of facility-specific highway capacity enhancement projects.

An introduction to the key models of interest, an overview of the state of the practice, and a brief discussion of the key limitations of current modeling practice are provided in this section. More detailed discussions of the problems with individual modeling approaches

as well as recommendations for improvements in practice, including research, are included in each of the following three chapters.

Modeling the effects of additions to highway capacity on air quality and energy use requires a chain of different models—from land use and travel demand models used to generate trip and traffic volume data to emission, dispersion, and energy models used to estimate the impacts of changes in travel activity on emission levels, regional air quality, and energy use, respectively (Figure 2-14).

Many of these models were not developed to provide the type of analysis required of them today. For example, travel demand models were originally developed to help size and locate highway and transit facilities in a region. Travel volume forecasts could be approximate for estimating capacity requirements, but such forecasts are inadequate for providing the facility- and time-specific travel volume and speed data necessary to estimate the emission impacts of specific projects (Harvey and Deakin 1993, 3–6; DeCorla-Souza 1993b, 6). Similarly, mobile emission models were designed to predict macro-level emissions for input into emissions inventories,²⁷ not to provide precise estimates of emission rates for vehicles traveling on individual highway links (Guensler 1993, 12).

Land use and travel demand models are critical to forecasting changes in travel demand from capacity additions to the highway network in a regional area. Only a few MPOs have developed fully integrated, state-of-the-art, land use–transportation models that would enable them to model the impacts of highway supply changes on travel demand and land use. More MPOs are expressing interest in developing this capability as a result of the CAAA and ISTEA (Putman 1994, 2).

A growing number of metropolitan areas have formal land use modeling capabilities; however, their use is still restricted to a minority of

regions in the United States (Harvey and Deakin 1993, 3-14–3-15). The available models generally use time series data on population, employment, household income, land availability,²⁸ and accessibility to allocate regional population and employment forecasts to geographic subareas of a region (Harvey and Deakin 1993, 3-15).

The main allocation modeling approach in use in the United States ²⁹ is the DRAM-EMPAL (Disaggregated Residential Allocation Model–Employment Allocation Model) components of the Integrated Transportation and Land Use Package (ITLUP) (Putman 1991).³⁰ The DRAM-EMPAL models allocate projected employment and households in a region on the basis of the current distribution of jobs and population, travel times, and multivariate measures of the attractiveness of various locations (Putman 1993). For example, the attractiveness of specific household locations depends on such factors as the amount of vacant developable land and the characteristics (e.g., income range) of households already living in specific areas (Putman 1993). The DRAM-EMPAL models can be used alone or linked to a travel demand model. A variety of other land use models have been developed and used overseas, many of which are complex and data intensive, but none are currently in use in the United States (Cambridge Systematics, Inc. 1991, 9; Wegener 1994).

One criticism of DRAM-EMPAL is that it does not describe all of the major factors (e.g., housing and land values, tax rates, and crime rates)³¹ that determine the location decisions of firms or households (Harvey and Deakin 1993, 3-15). This limits the precision of its predictions. However, sensitivity tests of the DRAM-EMPAL models demonstrate the models' ability to reproduce large regional patterns (Putman 1993). Thus, the models should be able to provide a sense of the direction and magnitude of locational changes from major additions to regional highway capacity.

Many MPOs have a travel demand modeling capability.³²Figure 2-15 provides an overview of the typical four-step sequential approach used to forecast regional travel demand; each step is a model representing a different aspect of the traveler's decision (Shunk 1992, 109). The first step in the process, trip generation, is a function of exogenously

FIGURE 2-15 Four-step travel demand process (DeCorla-Souza 1993b, 1).

determined demographic patterns and economic activity in a region (DeCorla-Souza 1993b, 3). The remaining three steps, which are followed sequentially, simply allocate the trips among alternative destinations in trip distribution, alternative modes of travel in mode choice, and alternative highway (and transit if appropriate) routes in trip assignment (DeCorla-Souza 1993b, 2). Trip assignment is based primarily on minimizing travel time through an iterative process that feeds back to mode choice, and sometimes to trip distribution, in an effort to equate initial with final travel time estimates (Harvey and Deakin 1991, 10-11). The outputs of the process are vehicle and passenger volumes on highway and transit routes, respectively (DeCorla-Souza 1993b, 2) (Figure 2-15).

A primary limitation of current travel demand modeling practice is the lack of integration with land use models. With few exceptions ³³ the models do not provide feedback loops so that analysts can examine the long-term impact of changes in the transportation network and network performance on travel demand and land use patterns and the trips generated by these effects. Because trip generation is exogenously determined in most conventional travel demand models, few MPOs can examine how highway additions might affect the regional distribution of population and economic activity, or regional growth in general (Deakin 1991, 11). In fact, there are no operational models that formally analyze the impact of transportation improvements on aggregate regional growth, in part because of lack of a theoretical basis for determining the relationship between transportation improvements and population and economic growth (Deakin 1991, 11).

In addition, conventional travel demand models do not provide the level of accuracy or detail on vehicle or travel activity by time of day that is needed for accurate emission and air quality modeling of the impacts of additions to highway capacity. The reliability of data on speed and traffic volume forecasts is poor,³⁴ and other detailed data needed for emission and air quality analyses —such as temporal distribution of travel volumes and trips, vehicle type, and vehicle operating mode—are not direct outputs of the models. ³⁵

Traffic models, as mathematical representations of real-world phenomena, comprise the tools used by traffic engineers to assess the relationships between capacity, levels of service, speed, and delay. Simulation models, as computer implementations of traffic models, provide a more detailed representation of traffic flow, including information on speed, acceleration, and deceleration of individual vehicles. ³⁶ Several of the traffic and simulation models in current use have been adapted to provide estimates of emissions and fuel use.

Why are these models then not used more widely to address the questions raised in this study? One reason is that many of the models are extremely data intensive and are thus beyond the data capabilities of many MPOs (Harvey and Deakin 1993, 3-77). Another reason is that, although traffic simulation models provide sophisticated

analyses of the impacts of traffic flow improvements on roadway and vehicle performance, they are not well equipped to make similarly sophisticated assessments of changes in traffic volumes that will accompany these measures (Horowitz 1982, 212). Thus, they fall short in their ability to analyze the longer-term impacts of traffic flow improvements of interest in this study.

Under the auspices of the jointly funded federal Travel Model Improvement Program,³⁷ a multimodal traffic simulation model is being developed at the Los Alamos National Laboratory. The model as planned will provide detailed estimates of household trips and travel and vehicle movements in a metropolitan area, which can then be linked with emissions, airshed, and energy use models (presentation by G. Shunk, fourth study committee meeting, July 11, 1994). Although more detailed estimates of the physical effects of transportation activities on emissions, air quality, and energy use will be available, the model will not have any forecasting capacities to examine behavioral effects; forecasts of land use and trip generation patterns must be prepared separately and input into the model (Shunk, July 11, 1994).

Emission and atmospheric dispersion models are critical to assessing the impacts of motor vehicle travel on pollutant emissions and concentrations.

Two main emission models are currently in use in the United States: (a) the EPA MOBILE model, which is the most widely used emission model, and (b) the CARB EMFAC model, which is used in California. The structure of the models is the same: activity-specific emission rates estimated by the models are multiplied by emission-producing vehicle activities to provide emission outputs by pollutant (i.e., grams per vehicle-mile for MOBILE and, in addition, grams per vehicle-hour and per vehicle trip for EMFAC) (Guensler 1993, 3) (Figure 2-16). In both models, emission rates are a function of vehicle type and age, vehicle speed, ambient temperature, and vehicle operating mode (DeCorla-Souza 1993a, 2).

FIGURE 2-16 Emission modeling process (DeCorla-Souza 1993a, 1).

Baseline emission rates are derived from a laboratory test procedure known as the Federal Test Procedure (FTP), which has been used to determine compliance of light-duty vehicles and light-duty trucks with federal emission standards since the 1972 model year (EPA 1993, 11). The FTP driving cycle consists of a sequence of accelerations, decelerations, cruise speeds, and idles based on actual home-to-work commute trips in the 1960s on Los Angeles freeways and surface arterials. A review of the FTP required by the CAAA has found that it underrepresents the high-speed driving and high acceleration rates that are common features of today's driving patterns and are key causes of high emissions (EPA 1993, 3).³⁸ Several of the 11 other driving cycles used to develop speed correction factors, which adjust emissions rates for vehicles traveling at average speeds other than the FTP baseline speed, are also not considered to be adequately representative of today's urban driving conditions.

Procedures for determining heavy-duty vehicle emissions differ significantly from those for determining light-duty vehicle emissions. Heavy-duty vehicle testing is generally performed on an engine dynamometer³⁹ rather than the full chassis dynamometer, which is used for light-duty vehicles, with the result that conversion factors must be used to produce emission rate data on a grams per vehicle-mile basis equivalent to light-duty vehicles. However, the speed correction fac-

tors for emissions of heavy-duty diesel vehicles, which are embodied in the MOBILE model, are based on chassis dynamometer testing of 22 1979-vintage heavy-duty diesel trucks conducted in the early 1980s (see Appendix A).

Emission models have several shortcomings as analytic tools for assessing the impacts that are of interest in this study. Among these are reliance on a limited number of drive cycles that do not adequately represent current driving on urban highways and understate key sources of high emissions; limited testing of passenger vehicles at high speeds and of heavy-duty vehicles at all speeds; and emission rates that reflect average trip speeds rather than average link speeds. These shortcomings are critical because emission estimates from current models are sensitive to assumptions about average vehicle speeds and the underlying drive cycles on which they are based.

Atmospheric dispersion or diffusion models translate emission levels into atmospheric concentrations of pollutants. Regional models take input data on grid-based rates and locations of emissions in a region and on the region's topography and meteorology (i.e., wind speeds and directions) to determine the atmospheric transport, diffusion, and chemical reactions of pollutants. Data on concentrations of pollutants at particular locations and times are the result (Horowitz 1982, 311; NRC 1991, 303).

Under the CAAA, ozone nonattainment areas designated as extreme, severe, serious, or multistate-moderate are to use photochemical, grid-based air quality models, such as the Urban Airshed Model, to demonstrate attainment of NAAQS (NRC 1991, 81). These models require detailed travel activity data that are spatially and temporally allocated (e.g., by link by grid; by season, month, and weekday or weekend day) to reflect the episode being modeled⁴⁰ (DeCorla-Souza 1993a, 1).

Dispersion models are also used to estimate concentrations of pollutants near particular facilities, such as intersections. For example, air quality intersection models estimate CO concentrations using data on traffic, intersection geometry, vehicle characteristics, emission fac-

tors generated from emission models, and meteorological conditions (O'Connor et al. 1993, 1). The CAL3QHC model, which is recommended by EPA guidance for intersection modeling, is used in most states except California, where use of the CALINE 4 model is recommended (O'Connor et al. 1993, 1).

Model uncertainties about chemical mechanisms, wind-field modeling, and removal processes as well as problems with input data—difficulties in obtaining accurate measures of VOC emissions and detailed spatial and temporal data on travel activity from conventional travel demand models—limit the effectiveness of dispersion modeling for ozone (NRC 1991, 10-11). Similar data gaps as well as placement and location of receptors are problems for CO modeling. For example, a recent survey on current practices in intersection air quality modeling (O'Connor et al. 1993, 29) found that critical determinants of predicted concentrations of CO, such as data on queuing, vehicle operating modes, fleet age and composition, speeds in highly congested flows, accelerations, and turning movements, are not commonly available for specific intersections.

A major source of highway vehicle fuel economy estimates is the simulation model developed by the Oak Ridge National Laboratory for the Federal Highway Administration in the mid-1980s. The model, which is based on on-road vehicle tests (12 gasoline-powered and 3 diesel-powered light-duty vehicles) and laboratory (dynamometer) testing, estimates fuel consumption as a function of vehicle speed and acceleration (McGill 1985, 3). The simulation model provides a good approximation of actual fuel use. The variance in individual vehicle performance, particularly the effect of speed variation at lower speeds, is less for fuel economy than for emissions. At high speeds, the test results converge, reflecting more steady-state, cruise driving (McGill 1985, 49). More recently, simplified analytic models have been developed that can estimate fuel economy within a 5 percent error range on the basis of vehicle parameters that specify the make and model of vehicle and driving patterns that reflect the level of traffic congestion (An and Ross 1993a, 1).

The key limitations of current models can be summarized under three broad categories. More in-depth discussion, particularly of the uncertainties of the models, can be found in the following chapters.

• Appropriateness of the models: The models are being used to solve problems for which they were not originally designed. Many are ill suited to provide the detailed analyses of the impacts of link-specific highway capacity additions on travel behavior and vehicle performance that are the focus of this study.

• Validity of the models: There are large uncertainties in the models themselves, which are manifested in wide variances around some model estimates. These reflect the limited state of the knowledge of the underlying phenomena the models are attempting to capture. A better understanding is needed of travel and driving behavior, which are critical to accurate travel demand and emission modeling, respectively. Models are validated, that is, a comparison of estimated with actual measured data is made, when sufficient survey data are available.⁴¹ Too often the information is dated or limited.

• Links between the models: There is a mismatch of detail between the outputs generated and the inputs required by several of the models in the modeling chain. Lack of adequately detailed data from travel demand models for input into emission and atmospheric dispersion models is a particular problem for accurate estimation of the impacts on air quality of highway capacity additions.

Motor vehicles are a major source of air pollution in the nation's metropolitan areas, and they are major users of petroleum. According to currently available estimates, transportation sources account for about 45 percent, and highway vehicles slightly more than one-third, of nationwide emissions of EPA's six criteria pollutants. However, the range is considerable for each pollutant source and for different vehicle and fuel types. Moreover, there is a high degree of uncertainty with respect to many of the estimates.

Gasoline-powered passenger vehicles—the most common vehicle on the road—are the primary source of CO highway vehicle emissions and contributors to the ozone precursor emissions from highway vehicles (VOCs and NO_x). Heavy-duty diesel vehicles contribute a disproportionate share of total highway vehicle emissions of PM-10, SO₂, and NO_x.

The transportation sector accounts for nearly two-thirds of the petroleum consumed in the United States. The highway mode accounts for about three-quarters of the transportation total. The gasoline-powered motor vehicle fleet also contributes about 20 percent of total U.S. CO₂ emissions, the principal greenhouse gas.

Vehicle emission levels are a function of trip taking as well as distance traveled, because emissions vary depending on whether the vehicle is warmed up. Emission levels are sensitive to average vehicle speed over the distance of the trip and vary as a nonlinear function of average trip speed. In addition, emissions are affected by smoothness and consistency of vehicle speeds, which vary by trip type. Sharp accelerations, in particular, are an important source of CO and VOC emissions, which are not well reflected in current emissions models. Thus, average trip speed alone is not a good predictor of emission levels.

Fuel economy is also sensitive to average vehicle speed but somewhat less so to aggressive accelerations and braking. Thus, average trip speed is a good predictor of fuel economy for most urban trips. Highway capacity additions, which will increase average trip speeds and smooth traffic flows, should directly affect emission levels and fuel economy.

A range of models are available to estimate the effects of motor vehicle transportation on emissions, air quality, and energy use. However, many were developed to predict macro-level, regional effects; they are not well suited to assessing the impacts of link-specific highway capacity enhancement projects at the level of precision that is being required of them today. Nor were the different types of models (e.g., land use models, travel demand models, emission models) designed to be easily integrated in their operations. Data requirements —both the currency of the data and the detail needed for impact analyses —are also problems. Finally, the models are based on a limited understanding of the underlying relationships. Greater knowledge of

travel and driving behavior, in particular, is critical to improved modeling of the travel demand and emission effects of highway capacity enhancement projects.

1. A stationary or point source is a large, geographically concentrated emitter, such as a coal-fired electrical power plant, whose emissions rates are large enough to be significant by themselves even if no other emission sources are present (Horowitz 1982, 7). An area source is a collection of small, geographically dispersed emitters that are not significant individually but that are important collectively, such as dry cleaning establishments (Horowitz 1982, 7). A mobile source, such as an automobile, is characterized as not emitting from a fixed location.

2. Short-term (24-hr or less) averaging times were designated for some pollutants, such as CO and O₃, to protect against acute, or short-term, health effects; long-term averaging times (annual average) were established for other pollutants to protect against chronic health effects (Curran et al. 1994, 20).

3. CO contributes to the buildup of tropospheric (ground-level) ozone (the principal ingredient of smog) and methane, both major greenhouse gases. First, CO helps convert nitric oxide to nitrogen dioxide, a crucial step in ozone formation. Second, CO reacts with the hydroxyl radical (OH), which eventually removes CO from the atmosphere; however, OH is also the principal chemical that destroys ozone and methane. Thus, if carbon monoxide levels increase, OH concentrations will fall, and regional concentrations of ozone and methane will rise (MacKenzie and Walsh 1990, 8).

4. After the report was completed, EPA recomputed emissions from highway vehicles using the most recent version of its emission factor model, MOBILE5a (Curran et al. 1994, 22), but EPA estimates are still thought to be low by many in the scientific community.

5. Nitrogen oxide (NO), the dominant constituent of vehicle exhaust emissions of NO_x, combines with ozone (O₃) to form NO₂ and O₂. However, ozone is subsequently regenerated by further chemical reactions stimulated by the presence of sunlight (see NRC 1991, 168 for a more detailed discussion).

6. The problem is the consistent underestimate of VOC emissions, which leads to estimates of relatively low VOC to NO_x ratios. The nation's ozone reduction strategy has been based largely on the premise that VOC/NO_x ratios in the most polluted areas, where VOC control is more effective, are low (i.e., in the less-than-10 range). An upward correction in VOC emission inventories could indicate the need for a fundamental change

in ozone abatement strategies to greater use of NO_x controls in many geographic areas (NRC 1991, 7).

7. This regulation, however, is an interim requirement, that is, it applies to all projects contained in new or revised transportation improvement programs until EPA approves state implementation plans. In the latter, a state could choose to accommodate NO_x emissions generated by new transportation projects by reducing emissions from other sectors (AASHTO Journal 1994, 10–11.)

8. In 1987 EPA replaced earlier standards for particulate matter with the more stringent PM-10 standard, which focuses on the smaller particles likely to be responsible for adverse health effects because of their ability to reach the lower regions of the respiratory tract (Curran et al. 1994, 10).

9. The 1987 Truck Inventory and Use Survey reported that 67 percent of combination vehicles used diesel fuel, but most trucks operating only in local areas (96 percent) are fueled by gasoline (Bureau of the Census 1990, 37, 48).

10. In 1993 emissions from all highway vehicles, gasoline and diesel powered, for CO, NO_x, PM-10, SO₂, and VOC were 74,155 thousand short tons. Emissions from diesel-powered, heavy-duty vehicles for the same pollutants were 3,986 thousand short tons, or 5.4 percent of the total; emissions from light-duty passenger vehicles were 46,941 thousand short tons, or 63 percent of the total (Nizich et al. 1994, A-4, A-8, A-15, A-19, A-24). In the most recent year for which data are available (1992), all motor vehicles accounted for 2,239,828 million vehicle miles of travel; combination trucks and buses accounted for 104,771 million vehicle miles of travel, or nearly 5 percent of the total, and passenger vehicles accounted for 1,595,438 million vehicle miles of travel, or 71 percent of the total (FHWA 1993, 207).

11. The primary components of diesel particulates are soot formed during combustion (40 to 80 percent of the total); particulate sulfates, which depend on operating conditions and the fuel's sulfur content (5 to 10 percent of the total); and heavy hydrocarbons condensed or adsorbed on the soot from the fuel and lubricating oil and also formed during combustion (the remainder) (Weaver and Klausmeier 1988, 2-7–2-8).

12. Even if all fuel were burned in the combustion process, thereby eliminating particulates from incomplete combustion, impurities in the fuel would burn and appear in the exhaust as particulates; the primary offender is sulfur (Conte 1990, 61). With lower sulfur levels in fuel, this source of particulates should also be reduced.

13. EPA considers a cold start for a catalyst-equipped vehicle to occur after the engine has been turned off for 1 hr. For noncatalyst vehicles, a cold start occurs after the engine has been turned off for 4 hr (Sierra Research 1993, 18).

14. Refueling losses and crankcase emissions are also generally considered in the evaporative emissions category as is a new category, resting losses. The latter was previously included under the hot soak and diurnal categories (Sierra Research 1993, 20).

15. VOC and CO emissions are higher when a cold engine is first started, because a fuel-rich mixture must be provided to achieve adequate combustion during warm-up and the excess fuel is only partially burned. In addition, the catalytic converter does not provide full control until the vehicle is warmed up (Sierra Research 1993, 18). It takes between 1 and 3 min for modern, properly operating vehicles to warm up. Catalysts also cool off faster than engines and are completely cold in 45 to 60 min (EPA 1993, 115; Enns et al. 1993, 3). Preheated catalytic converters may ameliorate the problem. The California Air Resources Board estimates that they would decrease cold-start emissions by half or more but would not eliminate the problem (FHWA 1992, 28).

16. The number would be even larger if running loss evaporative emissions, which are included under running emissions, were separated out.

17. Loads are a function of vehicle operating conditions (e.g., number of passengers, whether a trailer is being towed, whether the air conditioning is on), highway conditions (e.g., road grade), and driver behavior (e.g., aggressive driving with sharp accelerations). The latter two conditions are described in subsequent sections.

18. The exceptions are air conditioning and towing corrections, which can be input by the user in running the emissions models.

19. Emissions were calculated under hot stabilized operating conditions.

20. CO emissions, which are a product of incomplete combustion of motor fuels, are most affected. Engine-out CO emissions increase because of incomplete fuel combustion under fuel-rich conditions and exhaust emissions increase because the catalyst is overridden (personal communication, John German, EPA, Feb. 4, 1994). VOCs are affected but to a lesser extent. They result from unburned fuel in the engine. As fuel is increased with the richer air-fuel mixture, the level of engine-out VOC emissions goes up proportionately; these emissions are not handled by the catalyst, which is overridden under fuel-rich conditions, thereby increasing exhaust emissions (personal communication, John German, EPA, Feb. 4, 1994). NO_x engine-out emissions decrease under rich operation, but NO_x reduction efficiencies in the catalyst also drop (EPA 1993, 19). Overall, there may be a slight increase in exhaust NO_x emissions under rich operation, but the effect is relatively minor and varies from vehicle to vehicle (EPA 1993, 19).

21. From a sample of 24,000 emissions measurements made over a 4-day period, Naghavi and Stopher (1993) found that more than half of the CO was emitted by 6.9 percent of the vehicles and that about half of the VOC was emitted by 20 percent of the vehicles (p. 1).

22. These programs are required in areas designated “serious” or above for ozone and “high moderate” or above for CO. EPA estimates that innovative inspection and maintenance programs could yield a 28 percent reduction in emissions (DOT and EPA 1993, 33).

23. Exhaust emissions of CO and VOC also increase at temperatures above 24°C (75°F), but not as sharply as at lower temperatures. The increase is primarily the result of an increase in vapors purged from the evaporative emission control system, leading to rich operation (Sierra Research 1993, 122).

24. For example, if VOC emission formation in the engine (engine-out emissions) is reduced because less fuel is being delivered to the engine chamber per engine cycle, these gains will show up as lower tail pipe exhaust emissions only if manufacturers do not cut back on catalyst emission-control systems (tail pipe emissions) and do not take advantage of some of the savings (DeLuchi et al. 1993, 7-8).

25. The problem is not the quick acceleration, but the delay in gear shifting. Drivers with manual transmissions shift later (at higher engine speeds); with automatic transmissions, the system delays shifting up, both with the same result—high engine speeds and high fuel consumption (An et al. 1993, 4).

26. A fully loaded diesel truck realizes 3 to 3.4 kpl (7 to 8 mpg) on the highway, or approximately 108.9 to 123.4 metric ton-km per liter (280 to 320 ton-mi per gal). A car weighing 2268 kg (5,000 lb) can realize 11 to 12.7 kpl (26 to 30 mpg), or 25 to 28.8 metric ton-km per liter (60 to 75 ton-mi per gal) (Duleep 1992 in O'Rourke and Lawrence 1993, 11).

27. Emission inventories contain the relative contributions, current and projected, of emissions and pollution levels from mobile and stationary sources, drawing upon regional models and data (Harvey and Deakin 1993, 2-1–2-2).

28. Population and employment forecasts for a region may be provided by econometric models or derived from federal or state sources (Harvey and Deakin 1993, 3-11). Land use data are obtained from local land use plans. Local development policies are important to understanding potential constraints on land availability and development intensity (Shunk 1992, 107).

29. Putman reports, for example, that there are 14 MPOs in various stages of implementing DRAM-EMPAL for regional forecasting and policy evaluation efforts (Putman 1994, 1). Eleven have completed preliminary calibrations of both models using their own region's data and four are working on developing direct linkages between their transportation and land use models (Putman 1994, 2). Other land use models in use in the United States include POLIS in the San Francisco Bay Area, EMPIRIC in Atlanta, and PLUM in Washington, D.C. (Shunk 1992, 107).

30. In the early 1980s the entire ITLUP was distributed as a supplement to the Urban Transportation Planning System (UTPS) package, a travel demand modeling system package developed by the Federal Highway Administration (Harvey and Deakin 1993, 3-16).

31. The DRAM model, however, includes the income distribution of residential households, which is a proxy for several of these factors.

32. Many MPOs use the UTPS software package (Harvey and Deakin 1993, 3-5).

33. The MTC of the San Francisco Bay Area and the Puget Sound Council of Governments in the Seattle area have formal land use models, which are integrated into their regional travel demand models in a manner that allows for feedback between transportation and land use over time (Cambridge Systematics, Inc. 1991, 42). Cambridge Systematics, Inc., has been working with the Portland, Oregon, metropolitan area through the Land Use Transportation Air Quality Connection project to develop this capacity. Finally, the Southern California Association of Governments has recently completed a full test run of an integrated land use (DRAM-EMPAL)–transportation model (Putman 1994, 3).

34. Direct estimates of travel speed are not an output of travel demand models. Instead, link speeds are adjusted through an iterative process of assigning trips to the shortest network path to arrive at travel volume estimates (Meyer and Ross 1992, 6). Traffic volumes are often calibrated with actual traffic counts, but no attempt is made to check travel speeds or travel time against observed speeds (DeCorla-Souza 1993b, 5-6), with the result that model-derived speeds tend to overestimate actual link speeds, particularly under congested conditions (Meyer and Ross 1992, 6; Harvey and Deakin 1993, 3-63). Estimated traffic volumes on specific links may be in error by as much as 15 to 50 percent, depending on the total traffic volume on the link (DeCorla-Souza 1993b, 2).

35. Travel demand models provide no information on cold starts, because trips are not chained and travel is not tracked by time of day (Ducca 1993, 3). Travel demand models typically provide data on average weekday traffic levels by traffic zone. Hourly data on episode days by grid square, however, are needed for photochemical modeling (DeCorla-Souza 1993a, 3; Ducca 1993, 3). Information on vehicle type and age and vehicle operating mode cannot be directly obtained from travel demand model output (DeCorla-Souza 1993a, 4-5).

36. Some of the most common models for simulating traffic flows on freeways and estimating the effects of bottlenecks and ramp metering are FREQ, TRAFLO, and INTRAS. NETSIM was designed to simulate traffic changes from traffic signalization and intersection design improvements. TRANSYT and PASSER simulate traffic flows on arterials and changes in performance, such as travel times and delays that result from traffic flow improvement measures (Harvey and Deakin 1993, 3-77).

37. The U.S. Department of Transportation and the Environmental Protection Agency have already authorized $3 million; Department of Energy support is also being sought for a long-term program total of $25 million.

38. An analysis of instrumented vehicles driven in the Baltimore area for the FTP review found that about 18 percent of total Baltimore driving time was composed of higher speeds and sharper accelerations than those represented on the FTP [i.e., maximum speeds of 90.7 kph (56.7 mph) and maximum acceleration rates of 5.3 kph/sec (3.3 mph/sec)] (EPA 1993, 3-4).

39. An engine-based test procedure was adopted because engine manufacturers are distinct from truck manufacturers and because the same engine can be used with a wide variety of trucks with different transmissions and axles (see Appendix A).

40. The models are validated for predictive purposes on the basis of their ability to simulate adequately a base-year episode day of high concentrations of ozone (NRC 1991, 308).

41. Validation of emissions data is further complicated, as is discussed in the following chapter, by identifying what actual data (e.g., what drive cycle) to measure.

DOT

U.S. Department of Transportation

FHWA

Federal Highway Administration

EPA

Environmental Protection Agency

NRC

National Research Council

TRB

Transportation Research Board

AASHTO Journal. 1994. Browner Responds on NO_x Issue. Feb. 18, pp. 10–11.

An, F., and M. Ross. 1993a. Model of Fuel Economy and Driving Patterns. Presented at 72nd Annual Meeting of the Transportation Research Board, Washington, D.C., 22 pp.

An, F., and M. Ross. 1993b. A Model of Fuel Economy and Driving Patterns. No. 930328. Society of Automotive Engineers, International Congress and Exposition , Detroit, Mich., March 1–5, pp. 63–79.

An, F., M. Ross, and A. Bando. 1993. How To Drive To Save Energy and Reduce Emissions in Your Daily Trip . RCG/Hagler, Bailly, Inc., Arlington, Va., and The University of Michigan, Ann Arbor, 10 pp.

Bureau of the Census. 1990. Truck Inventory and Use Survey. 1987 Census of Transportation, TC87-T-52. U.S. Department of Commerce, Aug., 166 pp.

Cambridge Systematics, Inc. with Hague Consulting Group. 1991. Making the Land Use Transportation Air Quality Connection: Vol. I: Modeling Practices. Oct., 84 pp.

Conte, F. 1990. Trucking in the '90s: Emissions. Owner Operator. Sept., pp. 58–65.

Curran, T., T. Fitz-Simons, W. Freas, J. Hemby, D. Mintz, S. Nizich, B. Parzygnat, and M. Wayland. 1994. National Air Quality and Emissions Trends Report, 1993. 454-R-94-026. U.S. Environmental Protection Agency. Research Triangle Park, N.C., Oct., 157 pp.

Davis, S.C. 1994. Transportation Energy Data Book: Edition 14. ORNL-6798. Center for Transportation Analysis, Energy Division, Oak Ridge National Laboratory, Tenn., May.

Deakin, E. 1991. Scoping Study: Impact of Highway Congestion on Air Quality. University of California at Berkeley, March, 22 pp.

DeCorla-Souza, P. 1993a. Travel and Emissions Model Interactions. Presented at Transportation-Air Quality Conference, Washington, D.C., Feb. 23–26.

DeCorla-Souza, P. 1993b. Travel Forecasting Process. Presented at Transportation-Air Quality Conference, Washington, D.C., Feb. 23–26.

DeLuchi, M., D.L. Greene, and Q. Wang. 1993. Motor-Vehicle Fuel-Economy: The Forgotten Hydrocarbon Control Strategy? UCD-ITS-RR-93-3. Institute of Transportation Studies, University of California, Davis, Jan., 25 pp.

Dockery, D.W., A. Pope, X. Xu, J.D. Spengler, J.H. Ware, M.E. Fay, B.G. Ferris, and F.E. Speizer. 1993. An Association Between Air Pollution and Mortality in Six U.S. Cities . New England Journal of Medicine, Vol. 329, No. 24, Dec. 9, pp. 1754–1808.

DOT and EPA. 1993. Clean Air Through Transportation: Challenges in Meeting National Air Quality Standards. Aug.

Ducca, F.W. 1993. Future Directions in Travel Forecasting. Federal Highway Administration, U.S. Department of Transportation , 9 pp.

Duleep, K.G. 1992.Analysis of Heavy Duty Fuel Efficiency to 2001. Presented at 71st Annual Meeting of the Transportation Research Board, Washington, D.C.

Effa, R.C., and L.C. Larsen. 1993. Development of Real-World Driving Cycles for Estimating Facility-Specific Emissions from Light-Duty Vehicles. Presented at the Air and Waste Management Association Specialty Conference on The Emission Inventory: Perception and Reality, Pasadena, Calif., Oct. 18–20, 20 pp.

Enns, P., J. German, and J. Markey. 1993. EPA's Survey of In-Use Driving Patterns: Implications for Mobile Source Emission Inventories. Office of Mobile Sources, U.S. Environmental Protection Agency.

EPA. 1993. Federal Test Procedure Review Project: Preliminary Technical Report . Office of Mobile Sources, May, 161 pp.

Evans, L., R. Herman, and T. Lam. 1974. Multivariate Analysis of Traffic Factors Related to Fuel Consumption in Urban Driving. GMR-1710. Research Laboratories, General Motors Corporation, Warren, Mich., Oct.

Federal Register. 1993. Criteria and Procedures for Determining Conformity to State or Federal Implementation Plans of Transportation Plans, Programs, and Projects Funded or Approved Under Title 23 U.S.C. or the Federal Transit Act . Vol. 58, No. 225, Nov. 24, 62, 188–62, 253.

FHWA. 1992. Transportation and Air Quality: Searching for Solutions: A Policy Discussion Series. No. 5, FHWA-PL-92-029. U.S. Department of Transportation, Aug., 30 pp.

FHWA. 1993. Highway Statistics 1992. FHWA-PL-93-023. U.S. Department of Transportation, 235 pp.

Gordon, D. 1991. Steering a New Course: Transportation, Energy, and the Environment . Island Press, Washington, D.C., 244 pp.

Greco, R. 1985. Motor Vehicle Tampering Survey—1984. Office of Air and Radiation, U.S. Environmental Protection Agency , July.

Greene, D.L., D. Sperling, and B. McNutt. 1988. Transportation Energy to the Year 2020. In Special Report 220: A Look Ahead: Year 2020. Transportation Research Board, National Research Council, Washington, D.C., pp. 207–231.

Guensler, R., D. Sperling, and P. Jovanis. 1991. Uncertainty in the Emission Inventory for Heavy-Duty Diesel-Powered Trucks. UCD-ITS-RR-91-02. Institute of Transportation Studies, University of California, Davis, June, 146 pp.

Guensler, R. 1993. Transportation Data Needs for Evolving Emission Inventory Models. Institute of Transportation Studies, University of California, Davis, 27 pp.

Guensler, R. 1994. Vehicle Emission Rates and Average Operating Speeds. Ph.D. dissertation. University of California, Davis.

Harvey, G., and E. Deakin. 1991. Toward Improved Regional Transportation Modeling Practice. Deakin Harvey Skabardonis, Inc., Berkeley, Calif., Dec., 68 pp.

Harvey, G., and E. Deakin. 1993. A Manual of Regional Transportation Modeling Practice for Air Quality Analysis. Deakin Harvey Skabardonis, Inc., Berkeley, Calif., with Cambridge Systematics, COMSIS, Dowling Associates, Gary Hawthorne Associates, Parsons Brinckerhoff Quade & Douglas, and Ann Stevens Associates, July.

Heywood, J.B. 1988. Internal Combustion Engine Fundamentals. McGraw-Hill Book Company.

Horowitz, J.L. 1982. Air Quality Analysis for Urban Transportation Planning. MIT Press, Cambridge, Mass., 387 pp.

Johnson, J.H. 1988. Automotive Emissions. Air Pollution, the Automobile, and Public Health. Health Effects Institute. National Academy Press, Washington, D.C.

Lilly, L.R.C. (ed.). 1984. Diesel Reference Book. Butterworths, Boston, Mass.

MacKenzie, J.J., and M.P. Walsh. 1990. Driving Forces: Motor Vehicle Trends and Their Implications for Global Warming, Energy Strategies, and Transportation Planning. World Resources Institute, Washington, D.C., Dec., 49 pp.

McGill, R. 1985. Fuel Consumption and Emission Values for Traffic Models. FHWA/RD-85/053. Oak Ridge National Laboratory, Tenn., May, 90 pp.

Meyer, M., and C. Ross. 1992. Transportation and Air Quality Modeling: Fitting a Square Block into a Round Hole. Georgia Institute of Technology, Atlanta, 11 pp.

Meyer, M.D., M. Rodgers, C. Ross, F.M. Saunders, and C.T Ripberger. 1993. A Study of Enrichment Activities in the Atlanta Road Network, pp. 225–233.

Murrell, D. 1980. Passenger Car Fuel Economy: EPA and Road. U.S. Environmental Protection Agency, Jan., 305 pp.

Naghavi, B., and P. Stopher. 1993. Remote Sensing, Means, Medians, and Extreme Values: Some Implications for Reducing Automobile Emissions. Presented at 72nd Annual Meeting of the Transportation Research Board, Washington, D.C., 28 pp.

Nizich, S.V., T.C. McMullen, and D.C. Misenheimer. 1994. National Air Pollutant Emissions Trends, 1900–1993. EPA-454/R-94-027. Office of Air Quality Planning and Standards, Research Triangle Park, N.C., Oct., 314 pp.

NRC. 1990. Confronting Climate Change: Strategies for Energy Research and Development . National Academy Press, Washington, D.C.

NRC. 1991. Rethinking the Ozone Problem in Urban and Regional Air Pollution. National Academy Press, Washington, D.C., 489 pp.

NRC. 1992. Automotive Fuel Economy: How Far Should We Go? National Academy Press, Washington, D.C., 259 pp.

O'Connor, K., L.L. Duvall, and R.G. Ireson. 1993. Intersection Air Quality Modeling: Review of Ambient Data and Current Modeling Practices: Vol. 2: Survey on Current Practices in Air Quality Modeling. Systems Applications International, San Rafael, Calif., 49 pp.

O'Rourke, L., and M.F. Lawrence. 1993. Strategies for Goods Movement in a Sustainable Transportation System . Presented at Transportation and Energy Strategies for a Sustainable Transportation System, Asilomar Conference System, Pacific Grove, Calif., Aug. 23, 35 pp.

Outwater, M.L., and W.R. Loudon. 1994. Travel Forecasting Guidelines for the Federal and California Clean Air Act. Presented at 73rd Annual Meeting of the Transportation Research Board, Washington, D.C., 24 pp.

Putman, S. 1991. DRAM/EMPAL ITLUP: Integrated Transportation Land-Use Activity Allocation Models: General Description. S.H. Putman Associates, Philadelphia, Pa., Jan.

Putman, S. 1993. Sensitivity Tests with Employment and Household Location Modes. Presented at Third International Conference on Computers in Urban Planning and Urban Management, Georgia Institute of Technology, Atlanta, July 23–25.

Putman, S. 1994. Integrated Transportation and Land Use Models: An Overview of Progress with DRAM and EMPAL with Suggestions for Further Research. Presented at 73rd Annual Meeting of the Transportation Research Board, Washington, D.C., 28 pp.

Shunk, G.A. 1992. Urban Transportation Systems. In Transportation Planning Handbook (J.D. Edwards, Jr., ed.), Institute of Transportation Engineers, Prentice Hall, N.J., pp. 88–122.

Sierra Research, Inc. 1993. Evaluation of “MOBILE” Vehicle Emission Model. Report SR93-12-02. Sacramento, Calif., Dec. 7.

Stedman, D. 1991. Presentation at The Transportation-Land Use-Air Quality Connection: A Policy and Research Symposium. Public Policy Program, University of California Extension at Los Angeles, Lake Arrowhead, Calif., Nov. 6–8.

Weaver, C.S., and R.F. Klausmeier. 1988. Heavy-Duty Diesel Vehicle Inspection and Maintenance Study. Final Report, Volume II: Quantifying the Problem. Radian Corporation, Sacramento, Calif., May 16.

Wegener, M. 1994. Operational Urban Models: State of the Art. Journal of the American Planning Association, Vol. 60, No. 1, Winter, pp. 17–29.

Highway capacity enhancement projects have long been viewed as providing emission reductions and energy efficiencies by contributing to freer-flowing traffic conditions. In this chapter, the current knowledge about the initial effects of highway capacity additions on emissions and energy use is presented. Gaps and uncertainties in current understanding are identified. Finally, a summary appraisal of the factors most likely to affect outcomes is given, and recommendations for improving the state of knowledge are made.

The shaded box in Figure 3-1 illustrates the fundamental relationships involved in analyzing the initial impacts of a highway capacity addition. The primary impacts are changes in traffic flow patterns on the facilities affected by the improvement. Expansion of highway capacity should reduce the probability of stop-and-start traffic, raise average ve-

hicle speeds, and reduce speed variability (i.e., smooth the traffic flow). All else being equal, stop-and-start traffic, low speeds, and highly variable speeds are all associated with high emission levels and poor fuel economy. Expansion of highway capacity should also reduce traffic density¹ and improve levels of service (LOS) on the affected facilities.

As capacity is added and congestion is eased on one link in the system, the effects will spill over onto other routes. Travelers who divert from other routes or shift their time of travel to take advantage of the new capacity will affect traffic flow patterns and traffic volumes on the broader network of highways (Figure 3-1), making the task of assessing the net effects on emissions and energy use more complex.

Highway capacity additions can also affect other modes of travel. New capacity that reduces highway congestion and commuting time may encourage mass transit riders to switch back to their cars. Bicycle and pedestrian travel may be discouraged if the capacity addition (e.g., intersection widening) improves traffic flow patterns for automobiles and trucks to the detriment of slower-moving modes.

The effects on travel demand of all these network changes—route shifts, shifts in time of travel, and mode shifts—are considered in Chapter 4; this chapter is focused primarily on the effects of these changes on the physical traffic flow. For the purpose of analyzing the initial impacts of highway capacity additions, travel demand is assumed to remain constant. Any additional traffic volume on expanded highway links is presumed to be a reallocation of existing traffic.

Determining the net energy and emissions effects of changes in traffic flow patterns from a highway capacity addition also requires taking into account the construction phase of the project. Construction of certain capacity projects may cause traffic delays and create stop-and-start traffic conditions conducive to high emission levels.

Finally, determining net effects depends on the consequences if the project is not undertaken: whether congestion will worsen if the new capacity is not added, whether there is adequate capacity in the system to avert congestion if travelers shift their travel routes and times of travel, or whether investments in other modes (e.g., transit and bicycle facilities) can accommodate travel needs. These outcomes must be compared with those of the “build” option to determine net effects.

A motor vehicle trip consists of a beginning and an end connected by a series of accelerations, decelerations, constant speed cruising, and idling. To measure emissions and fuel consumption, trips are simulated on laboratory dynamometers and motor vehicles are tested using standardized drive cycles. Each drive cycle is composed of a unique profile of stops, starts, constant speed cruises, accelerations, and decelerations and is characterized by an overall average speed (Guensler 1994, 24). All vehicle trips at the same average speed are assumed to have the same underlying drive cycle regardless of facility type, roadway conditions (e.g., grade), or driver behavior.

Different drive cycles are used to represent driving under different traffic conditions (Guensler and Geraghty 1991, 5–6). For example, the New York City cycle with an average trip speed of 11.4 kph (7.1 mph) reflects city driving at very heavy levels of congestion; vehicle speeds are low and vehicle movements consist primarily of stops, starts, and accelerations (Figure 3-2). The Highway Fuel Economy Test cycle with an average speed of 77.3 kph (48.3 mph) was originally created to reflect rural highway driving. Vehicle speeds are higher and there is more cruise-speed driving, although accelerations are still evident (Figure 3-2). As discussed in the following sections, however, current drive cycles do not adequately represent real-world driving conditions.

Current estimates of emission rates embodied in the Environmental Protection Agency's (EPA's) MOBILE model and the California Air Resources Board's (CARB's) EMFAC model are expressed as functions of average speeds and are based on vehicle testing on a limited number of drive cycles.

Baseline emission rates for light-duty motor vehicles are derived from the Federal Test Procedure (FTP) cycle. Emission rates at other average speeds are calculated by testing automobiles on 11 other drive

cycles and heavy-duty trucks on 4 drive cycles, including a baseline composite cycle (Table 3-1); adjusting baseline rates with the appropriate speed correction factors; and applying them to the appropriate vehicle class (see accompanying text box).²

The resulting average trip speed–emission curves (Figure 2-4, Figure 2-5, Figure 2-6, Figure 2-7, Figure 2-8 to Figure 2-9) can be used to give a preliminary estimate of a highway capacity addition's effect on emissions by comparing the distribution of traffic speeds on the affected highway links before and after the project (Figure 3-1). These curves can also be used to estimate predicted emission levels for the affected highway links both with and without the project. The comparisons are not exact, however, because average trip speeds are not equivalent to link-specific speeds for portions of vehicle trips.

According to current models, if traffic is heavily congested and contributes to low average trip speeds [i.e., below 32 kph (20 mph)], capacity enhancements should initially reduce emissions of carbon monoxide (CO) and volatile organic compounds (VOCs) for automobiles and emissions of CO, VOC, and oxides of nitrogen (NO_x) for heavy-duty diesel vehicles (HDDVs). Reductions in NO_x emissions are likely to be small for automobiles; these emissions increase with speed at low average trip speeds [i.e., above 32 kph (20 mph) according to the MOBILE model and above 56 kph (35 mph) according to the EMFAC model]. If traffic is moderately to lightly congested and average trip speeds following a capacity addition exceed about 80 kph (50 mph), emissions initially should increase for all pollutants for automobiles and for NO_x for HDDVs. Emissions of CO and VOCs should remain relatively flat at high speeds for HDDVs.

If congestion is moderate and initial and final average trip speeds are in the intermediate range where the curves flatten out [i.e., between 32 and 80 kph (20 and 50 mph)], then capacity enhancement projects should only result in small emissions changes for both automobiles and HDDVs; emissions of CO and VOCs should be slightly reduced but partially offset by rising emissions of NO_x.

Modeled emissions estimates for particulates as a function of speed are unavailable, but industry data suggest that diesel particulate exhaust emissions follow the same trend as VOC emissions (i.e., they decline) up to about 80 kph (50 mph). Particulate exhaust emission levels at higher speeds are not well understood (Appendix A).

Unfortunately, estimates of the emission effects of changes in traffic flow patterns from a highway capacity addition are not nearly as precise as implied by the complex functions used to derive the speedcorrected emission rates. First, current model estimates are based on a limited set of drive cycles (Table 3-1) that inadequately represent specific traffic flow conditions. Many of the drive cycles on which emission estimates are based were developed decades ago (the FTP is more than 20 years old), when laboratory testing equipment capabilities were more limited.³ They are not believed to be representative of current real-world driving conditions (Effa and Larsen 1993, 1).⁴ For example, a comparison of three cycles with approximately the same average speed—the baseline FTP cycle and two cycles recently developed from driving on arterial roads and freeways in the Los Angeles area—shows the extent to which the baseline FTP drive cycle underestimates driving at higher speeds and accelerations, both of which are believed to be sources of high emissions (Figure 3-3 and Figure 3-4).⁵

Another problem is that little effort has been devoted to the development of driving cycles that represent high-speed operation. The only cycle developed from actual data on in-use operation is the Highway Fuel Economy Test, with an average speed of 77.3 kph (48.3 mph) (Figure 3-2, bottom). Because vehicles routinely travel at substantially higher average speeds on freeways, CARB developed a series of higherspeed cycles (Figure 3-5). Each uses the same portion of the Highway Fuel Economy Test cycle except that more time is allowed at the beginning and end for vehicles to accelerate to a higher initial cruising speed and to decelerate to a stop. The vehicles are required to perform the same accelerations at successively higher speeds, increasing engine power demands and the likelihood of higher emissions. The problem with this approach is that the cycles have not been validated with actual on-road data from customer service, and thus the resulting emissions estimates may be biased and may not represent real-world operation.

A second issue related to the certainty of emission estimates is that current models predict the effects of changes in traffic flow characteristics on emissions solely on the basis of changes in average speed. This one-dimensional approach cannot adequately describe the un-

derlying distribution of speeds and accelerations, which vary by type of facility and level of congestion and produce potentially large variations in emission levels. For example, existing emission models do not distinguish between driving on arterial highways and driving on freeways (Figure 3-6), although each has quite distinct frequency distributions for speed and acceleration, likely to result in different emission levels (Effa and Larsen 1993, 19).

Finally, current models cannot accurately predict emission rate changes for light-duty vehicles for a wide range of changes in average trip speeds. As will be shown, many of these speed changes are within the range of average speed changes expected from highway capacity additions. Guensler (1994) examined the confidence intervals⁶ for changes in predicted emission rates of CO, VOC, and NO_x for late-model-year, fuel-injected, light-duty vehicles as a function of changes in average trip speeds using the EMFAC model.⁷Figure 3-7a shows the results for speed changes in 8-kph (5-mph) increments for CO, VOC, and NO_x; the initial average trip speed is shown on the x axis. Speed changes in 16-kph (10-mph) and 32-kph (20-mph) increments are shown in Figure 3-7b and Figure 3-7c.

With the exception of NO_x, variances are large relative to predicted changes in emissions. In many cases the lower bound of the confidence interval is negative and the upper bound is positive, so that it is impossible to determine the direction, much less the magnitude, of the expected change (Guensler 1994, 167). For example, the 95 percent confidence interval for the reduction in CO emissions associated with an increase in average trip speeds from 64 kph (40 mph) to 72 kph (45 mph) encompasses a 40 percent reduction and a 19 percent increase in emission rates (Figure 3-7a). Even for NO_x emission changes, which have small variances, the questionable representativeness of the drive cycles on which the estimates are based leaves the validity of the results uncertain.

The committee did not conduct an in-depth review of the statistical approaches used to generate the confidence intervals, but it believes that the analysis illustrates the uncertainty of emission rate estimates based on current models. The analysis also suggests that current models do not reflect important explanatory variables that can significantly affect emission levels, such as the incidence of sharp accelerations at lower and moderate speeds.

Diesel emissions have not received the same scrutiny as emissions from gasoline-powered light-duty vehicles. Although emission estimates for HDDVs are based on a very limited number of drive cycles, the variance in their emission performance is believed to be considerably less than that of light-duty vehicles (a complete discussion of the emissions and energy characteristics of heavy-duty, diesel-

powered trucks and buses is presented in Appendix A). Diesel engine emissions can be predicted with greater certainty because the process of generating and controlling them is less complex than is the process for cars (i.e., HDDVs do not yet have exhaust aftertreatment; modeling their emissions thus does not require predicting catalyst efficiency). Diesel engines also require little cold start or acceleration enrichment, which significantly increases the variation in emission levels for light-duty vehicles (Appendix A).⁸

Important exceptions to these generalizations exist. The absence of data on particulate matter exhaust emissions as a function of speed is troublesome. HDDVs are the primary highway vehicle source of these emissions, and particulate concentrations are a risk factor for lung cancer. In addition, particulates are the most sensitive of all regulated diesel emissions to deterioration and malperformance of vehicle components from poor maintenance or tampering (Appendix A). Engineering analysis suggests that estimates of other diesel exhaust emissions may be incorrect at high speeds. In particular, diesel emissions of NO_x at very high speeds [e.g., 112 kph (70 mph)] are probably overstated, whereas diesel emissions of VOCs at similar speeds are probably understated by current model estimates (Appendix A).

In summary, although considerable research and vehicle testing have been performed, no definitive conclusions can be reached about how highway capacity additions and their effects on traffic flow characteristics change vehicle emission levels. The analysis requires examination of relatively small changes in average trip speeds using emissions data with large variances. In addition, the data are based on a small number of test cycles that do not adequately represent real-world driving conditions. Finally, the application of models based on average trip speeds to facility-specific speeds reflecting portions of trips is inappropriate. Emission models were not developed to provide microscale detail on vehicles traveling on particular highway links; their original purpose was to provide macroscale data on transportation emissions for input into emission inventories.

Analyzing the effects of highway capacity additions on emissions also requires accurate baseline data on traffic flow patterns to which emission rates can be applied. Traffic engineers have developed speed-flow-density relationships for different types of highway facilities on the basis of field surveys and have linked these critical traffic flow variables to operational levels of service. The speed-flow curve for free-ways (Figure 3-8) shows that average travel speeds are relatively high and stable for a wide range of service conditions, averaging between 88 and 112 kph (55 and 70 mph) for LOS A through C.⁹ Average travel speeds fall only under more heavily congested conditions (i.e., LOS D and E) and then only by a modest amount—to between 80 and 96 kph (50 and 60 mph) until breakdown levels (LOS F) occur. Less extensive data for urban arterials, which link average speeds with LOS, indicate much lower average travel speeds in general and a wider divergence between free-flow speeds (LOS A) and speeds under more heavily congested conditions (LOS D and E). Table 3-2 displays these differences.

These speed-capacity relationships by type of facility can be used to predict the likely distribution of traffic speeds and speed variation before and after different types of capacity projects, and the likely ef-

fect of these changes on emissions. The accompanying text box shows typical before and after speeds for a range of capacity projects on free-ways and arterial roads, which range from about an increase of 8 kph (5 mph) for capacity-enhancing, traffic engineering improvements to an increase of 32 to 48 kph (20 to 30 mph) for new freeway lanes. Many speed changes are small although the variance in emission changes is large (Figure 3-7a).

Traffic simulation models are generally used to estimate traffic flow changes on affected facility links as the result of highway capacity expansions. As with emission models, however, most currently available traffic simulation models provide data only on average vehicle speeds. For several of the microscale models—which involve tracking individual vehicles—vehicle flow profiles showing the distribution of vehicle accelerations, decelerations, cruise speed operations, and idling could be made available with modest revisions to the programs (Table 3-3). Traffic flow data also could be made available by vehicle type (Table 3-3). However, modeled estimates of vehicle velocities and accelerations should be validated with actual on-road vehicle operation data if they are to be used to provide inputs to emission models. Because variance in vehicle operation has a significant effect on emissions, the accuracy of these inputs is critical.

Some traffic simulation models (e.g., NETSIM and FRESIM) already provide data on emissions and fuel use for light-duty vehicles. However, the emissions estimates are based on a limited and dated sample of vehicles (15 light-duty, 1981 to 1984 model-year vehicles) and reflect steady-state operating conditions.¹⁰

Research that should improve understanding of the emission characteristics of motor vehicles under real-world driving conditions has begun. A significant investment is required to expand and accelerate this effort so that a better emission rate model can be developed and incorporated into the regulatory process.

The research to date has focused on the effects of such driving patterns on emissions as sharp accelerations and high speeds. Neither is well represented by current drive cycles, and they are suspected of being major reasons for current underestimation of emission levels (LeBlanc et al. 1994, 2). The research was initiated in response to a requirement of the Clean Air Act Amendments of 1990 (CAAA) that EPA review current vehicle testing procedures to ensure that they re-

flect actual driving conditions. EPA has conducted surveys of driving behavior in selected cities, and CARB has sponsored similar research in the Los Angeles area (EPA 1993, 1–2).¹¹

The research sponsored by EPA has confirmed that sharp accelerations and high speeds are not well represented in the baseline FTP drive cycle. The current maximum acceleration rate of 5.3 kph/sec (3.3 mph/sec) is frequently exceeded in on-the-road driving (EPA 1993, 3). Similarly, the current maximum FTP speed of 90.7 kph (56.7 mph) is routinely exceeded (EPA 1993, 3).

Preliminary results suggest that sharp accelerations,¹² which cause a vehicle to operate in a fuel-rich mode, contribute significantly to high emission levels for CO and VOCs (Sierra Research 1993; Kelly and Groblicki 1993; LeBlanc et al. 1994). For example, Sierra Research found CO emission levels 15 times higher, and VOC levels 14 times higher, from aggressive driving (with sharp accelerations) on the same trip (Figure 2-10 and Figure 2-11). Using instrumented vehicles in Atlanta, researchers found that most vehicles spend less than 2 percent of total driving time in this mode, but this small fraction of the trip can account for up to 40 percent of total CO emissions (LeBlanc et al. 1994, 11).

Sharp accelerations that command fuel enrichment have little effect on NO_x emissions. Engine-out NO_x emissions decrease under fuel-rich operation, but reduction in catalyst efficiency can produce offsetting effects (EPA 1993, 19). Tail pipe emissions of NO_x may increase slightly overall under fuel enrichment conditions, but the effect is relatively minor and varies from vehicle to vehicle (EPA 1993, 19). In contrast, limited testing of 28 new-technology vehicles by EPA suggests that mild accelerations, which do not cause fuel enrichment, increase NO_x emissions because of the higher loads placed on the engine. NO_x emissions under conditions of mild accelerations can be 2 to 3 times higher than under cruise-type driving (personal commu-

nication, John German, EPA, Jan. 11, 1995). Thus, traffic-smoothing projects, which keep speeds constant, should initially reduce emissions of all three pollutants.

Sharp accelerations appear to be most common at lower and moderate speeds, where speed variability is greatest (LeBlanc et al. 1994, Figure 3a). Mild accelerations occur at all speed ranges. A related finding from drive cycle research in Los Angeles suggests much greater frequencies of sharp accelerations and much lower frequencies of cruise-type conditions on arterial roads than on freeways (Effa and Larsen 1993, 3).13 Sharp accelerations at high speeds, although less frequent, can also cause emission increases.14 The situations most conducive to sharp accelerations at high speeds include entering expressways from metered ramps, maneuvering within merging lanes, and passing slower vehicles in freeway-like conditions (Cicero-Fernández and Long 1993, 7).

Evaluations of traffic-calming projects—projects designed to slow cars and make streets more accessible to bicycles and pedestrians in downtown commercial areas and residential neighborhoods15—suggest the emission benefits from reducing accelerations and speed variation. In Buxtehude, Germany, for example, traffic calming in the 2.5-km² historic core of the city resulted in reduced emissions for all of the major pollutants.16 The primary reason for the improvement was the encouragement of steady driving speeds—reducing accelerations, decelerations, and idle time—and the areawide implementation of the scheme, which prevented traffic diversions to other routes (Pharoah and Russell 1989, 16, 48).17

CARB is testing 125 light-duty vehicles on newly developed drive cycles—seven freeway cycles and three arterial cycles—that were constructed from actual highway driving (Effa and Larsen 1993, 9). Preliminary results from testing 86 vehicles on the freeway cycles, which do not include the artificially sharp accelerations of the old high-speed drive cycles (Figure 3-5), show no increase in CO and VOC emissions up to 96 kph (60 mph) (personal communication, Bob Dulla, Sierra

Research, Oct. 14, 1994). These results have called into question existing model estimates, which show CO and VOC emissions of light-duty vehicles increasing above approximately 80 kph (50 mph) (Figure 2-4 and Figure 2-5). CO and VOC emissions will increase with higher loads on the engine from aerodynamic drag, but additional research is required to understand at what speed this occurs and by how much emissions are increased.

Preliminary results for NO_x show a steady but slow increase in emissions as a function of increasing speed (personal communication, Bob Dulla, Sierra Research, Oct. 1994). However, as illustrated by changes in the speed correction factors for NO_x for each of the last three versions of the MOBILE model (Figure 3-9), there is still considerable disagreement about the speed at which NO_x emissions begin to increase and the magnitude of the increase. Specifically, it is unclear to what extent increasing NO_x emissions under cruise-type conditions are offset by decreasing NO_x emissions from reduced accelerations and smoother traffic flow conditions. Further research and testing are required to resolve these uncertainties.

The preliminary results confirmed previous results indicating that emissions of all pollutants are highest at very low average trip speeds.

Current research to develop new drive cycles that better represent contemporary driving conditions should provide much of the empirical data needed for more valid estimates of emission rates.18 Considerable data have been collected on individual vehicles for a range of different driving conditions. It is too early to determine how vehicle emissions will be affected in the aggregate (i.e., for the current vehicle fleet) and thus how current emission models should be modified.

EPA's efforts are focused on using the results of the drive cycle research to develop an alternative to or modify the FTP drive cycle for certification purposes. The agency has conducted a limited test program on current technology vehicles (Enns et al. 1993, 4), but further testing has not been planned and major modifications to the MOBILE model are unlikely.19 EPA's Office of Research and Development, which

is responsible for the agency's long-term research program, is pursuing research on drive cycles toward development of a modal emissions model.²⁰ However, implementation of a fully operational model is several years away (personal communication, Carl T. Ripberger, EPA, Aug. 19, 1994).

CARB, by comparison, is using the drive cycle research results not only to certify new vehicles but also to test a representative sample of vehicles to quantify differences in emissions from alternative drive cycles. The results will then be used to modify the speed correction curves contained in the EMFAC model.

Perhaps the most comprehensive effort to date is the state-initiated research project to develop and verify a modal emissions model within 3 years (NCHRP 1994, 1). The states have pooled funds through the National Cooperative Highway Research Program (NCHRP) to develop a model that accurately reflects the impacts of speed, engine load, and start conditions on emissions under a comprehensive variety of driving characteristics and vehicle technologies (NCHRP 1994, 1). A primary purpose of the model is to reliably estimate emission impacts from changes in driving characteristics associated with traffic operations and transportation system improvements. The model will also be responsive to the regulatory compliance needs of transportation and air quality agencies (NCHRP 1994, 1).

Controlling the variability of emission results so that modal models more accurately and reliably predict emission levels for different speeds and acceleration rates is a significant challenge. Modal emissions are affected by vehicle condition and state of repair. More to the point, the new models must reflect transient vehicle operations that are representative of on-road driving.²¹ The inherent variability of vehicle emissions is one of the difficulties in developing predictive models that can estimate with precision the emission impacts of highway capacity additions.

The multimodal traffic simulation model being developed at the Los Alamos National Laboratory will be able to provide second-by-second data on vehicle movements in traffic combined with detailed

modeled estimates of emissions and pollutant concentrations. This system of integrated models is several years from being fully operational, much less widely available to metropolitan planning organizations. ²² Moreover, a complementary emission factor model must still be developed to drive the emissions element of this program.

The Federal Highway Administration has contracted with the Oak Ridge National Laboratory (ORNL) to improve the accuracy of emissions and fuel consumption estimates used in traffic simulation models. ORNL will collect modal emissions and fuel consumption data for a limited sample of vehicles and develop a model to estimate emissions and fuel use for alternative traffic patterns.²³

The aggregate emissions impact of highway capacity additions depends on the traffic volume on the affected link as well as the changes introduced in traffic flow patterns. The magnitude of effects on traffic volume in turn relates to the type of capacity improvement and facility. The emissions effect of many traffic flow improvement measures (e.g., intersection improvements, improved signalization, or reduced on-street parking) is apt to be modest because such relatively small-scale improvements affect a relatively small amount of traffic volume for a relatively short distance. For example, an intersection improvement typically affects traffic for only 1/4 mi on either side of the intersection. In contrast, large capacity additions (e.g., new freeway links or lanes added to existing freeways) are likely to have major aggregate effects on emissions, affecting traffic for a substantial distance. Depending on initial and post-improvement distributions of speeds and accelerations, these more significant effects could be either positive or negative.

The traffic volume on the link affected by the capacity expansion is also likely to change during the course of the day. The effects will be greatest for facilities that are congested for most of the day.²⁴ For facilities that experience only peak-period congestion—a more common occurrence—the amount of traffic volume affected by the capacity addition will decline during off-peak periods. Whether the

overall effect on emissions will be positive or negative depends on the initial and post-improvement distributions of speeds and accelerations during peak and off-peak travel times.

Impacts of highway capacity additions may spill over onto neighboring routes, with mixed effects on emissions. Traffic may be diverted to the newly enhanced link, easing traffic congestion on other routes but increasing the traffic volume (and reducing speeds and increasing speed variability again) on the improved link. Route shifts can result in large emission reductions if traffic is diverted from heavily congested facilities. New freeway capacity, for example, may relieve very slow, heavily polluting, stop-and-start traffic on parallel arterials.

Travel times may also be affected. Commuters may take advantage of the new capacity to schedule their trips at more desired peak travel times, again reducing some of the freer-flow traffic conditions generated by the capacity addition at certain times of day.

Finally, highway capacity expansion may encourage mode shifts. Mass transit riders may find it desirable to drive to work if peak-period traffic congestion is alleviated by the capacity addition. If highway capacity expansions, particularly intersection and signalization projects, are designed to improve automobile and truck traffic flow without adequate attention to pedestrian and bicycle traffic, these other modes may be discouraged. These mode shifts will increase automobile traffic, reducing levels of service on the facilities affected by the capacity addition.

The magnitude of these network effects depends on the type of capacity improvement and, to a lesser extent, on its design. Minor traffic flow improvements are likely to have limited spillover effects on other routes, but mode of travel may be affected if the project design does not accommodate system users other than motorists, such as bicyclists and pedestrians. Major capacity additions are likely to have spillover effects on other roads that may be as important as the change in traffic flow patterns on the improved facility itself.

The effects of highway capacity additions on emissions are much less certain than generally believed.

To the extent that capacity enhancement projects smooth traffic flows, reducing speed variability and the incidence of sharp accelerations, they should initially reduce emissions of CO, VOCs, and NO_x. The effect is most pronounced for CO and VOC emissions from light-duty vehicles, particularly if traffic smoothing reduces the incidence of sharp accelerations. Exhaust emissions of CO and VOCs from HDDVs will be less affected because diesel engines do not require acceleration enrichment. Emission reductions from traffic smoothing should occur when capacity is added to heavily congested facilities, such as urban arterials or congested freeway segments. The effects are greater for projects enhancing capacity on roads that experience heavy congestion for a large part of the day, that affect large travel volumes for significant distances, and that divert significant traffic from other heavily congested facilities. Of course, the potentially negative emissions effects of traffic delays and stop-and-start traffic during the construction phase of the project must also be considered.

The initial effect on emissions of highway capacity additions that result in free-flow traffic operations at freeway speeds [i.e., above 80 kph (50 mph)] is less certain. The effect on light-duty vehicle emissions of CO and VOCs is not well understood. At high speeds, increased power demands on the engine cause CO and VOC emissions to increase, but it is unclear at what speed this occurs and by how much emissions are increased. NO_x emissions also increase, although there is considerable uncertainty about the speeds at which this increase begins and the rate of the increase. NO_x emissions are thought to increase gradually at speeds well below free-flow freeway speeds. Exhaust emissions of VOC and NO_x from HDDVs also rise at high speeds. The increase in NO_x emissions from automobiles and HDDVs could pose a serious problem for those ozone nonattainment areas

where NO_x emissions are already high because of the stringent requirements for NO_x reduction in conformity regulations.

For a broad range of intermediate conditions, highway capacity additions are likely to have modest effects on smoothing traffic flows and result in small upward shifts in the distribution of traffic speeds. The effects on emission levels are unknown; the changes are too small, and the variances around emission rates too large, to determine effects with any degree of confidence. Each of these conditions must be evaluated with reference to predicted changes in traffic flow characteristics and emission levels if the project is not undertaken, to determine net effects.

Changes in vehicle technology and new regulations for vehicle testing over the next decade may reduce the traffic-related effects of highway capacity additions, both positive and negative, on emission levels. For example, electronically controlled, multipoint fuel injection technology, which has largely replaced the older carbureted fuel delivery systems,²⁵ can be programmed to reduce emissions of CO and VOCs during sharp accelerations as easily as they now are programmed for power enrichment. Several automobile manufacturers (e.g., Ford Motor Company and Mercedes Benz) have developed vehicles that rarely go into the power enrichment mode. If EPA changes the FTP certification cycle to capture high-speed, high-acceleration driving that is currently outside the FTP envelope, automobile manufacturers will be required to take these factors into account in optimizing vehicle performance (personal communication, John German, EPA, Dec. 13, 1993). These changes will only reflect certification of new vehicles; it will take 10 to 15 years of vehicle turnover for them to work their way through the vehicle fleet. However, in the long run, they will weaken the emission-reduction potential of many highway capacity enhancement projects.

Understanding the effects of highway capacity additions on regional air quality requires looking beyond changes in motor vehicle emis-

sions to determine how concentrations of pollutants in the atmosphere will be affected. Highway capacity additions should help relieve localized concentrations of CO, or hot spots, by eliminating bottlenecks where congestion builds and traffic speeds are highly variable.

Effects on ozone concentrations are less clear. Many highway capacity additions are likely to have a neutral or negative effect on NO_x, one of the major ozone precursors, unless congestion is near breakdown levels before the capacity expansion. VOCs, the other ozone precursor, are likely to be reduced by traffic flow–smoothing projects, but such projects will affect only running emissions. Because most vehicles are fully warmed by the time they reach an arterial street or freeway —the location of most highway capacity projects—the projects will not affect cold start or evaporative emissions, which account for more than half of total VOC emissions on automobile trips.²⁶ Finally, regional concentrations of ozone are as much a function of local meteorological conditions and topography as of local travel patterns (Horowitz 1982, 76–77).

A simulation model developed by ORNL in the mid-1980s (described in Chapter 2), which used data from on-road and laboratory testing of light-duty motor vehicles (McGill 1985, 50), provides the basis for estimating the initial impacts of highway capacity projects on energy use of automobiles. Model estimates suggest that measures that increase vehicle speeds from about 32 kph (20 mph) to 56 to 72 kph (35 to 45 mph) should improve fuel economy (Figure 2-13). At higher speeds, fuel economy degrades rapidly.

Improvements in fuel economy are largest in the intermediate speed ranges—from about 32 kph (20 mph) to about 72 kph (45 mph)—suggesting that the greatest benefits for fuel economy will come from highway capacity projects that raise the distribution of traffic speeds

within this range. Traffic flow improvements on urban arterials, where average vehicle speeds fall within this range, are most likely to meet these conditions.

Automotive fuel economy is also affected by speed variation, but not as dramatically as are emission levels. Aggressive driving with sharp accelerations increases fuel consumption by about 10 percent (An et al. 1993, 4). Fuel use is minimized in cruise-speed driving conditions and worsens as speeds slow and stop-and-start driving conditions become prevalent. Of course, all improvements in fuel economy will be accompanied by reductions in CO₂ emissions (NRC 1992).

The fuel economy of heavy-duty diesel trucks varies with truck design and load. Heavy-duty trucks with gross vehicle weights (GVW) in the range of 6350 to 14 969 kg (14,000 to 33,000 lb) are less sensitive to speed changes than over-the-road heavy trucks in the GVW range of 14 969 to 36 287 kg (33,000 to 80,000 lb), which are more highly optimized for highway use (Appendix A).

Few data on how fuel economy varies with speed are publicly available from tests of HDDVs conducted under controlled conditions on a chassis dynamometer (Appendix A). However, selected data from simulation models developed by truck manufacturers²⁷ and on-road data collected by DOT in 1974 indicate that fuel economy for HDDVs improves as speeds increase up to about 80 kph (50 mph) but declines sharply at higher speeds largely because of the rapid rise of aerodynamic drag (Appendix A).

How reliable are fuel economy predictions? Automotive fuel use is believed to be relatively well approximated by current models, in part because fuel economy is not as sensitive as emissions to changes in traffic flow conditions, particularly speed variation. The largest source of variance in estimating fuel economy is differences in vehicle design and technology. An analysis of the relationship between fuel consumption and speed, using data on 350 automobiles manufactured between model years 1967 and 1977, found that fuel economy decreased

with increasing speed and engine size (Greene 1981, 441). The estimates from the ORNL simulation model corroborated these results. Fuel economy for the 12 light-duty test vehicles varied by a factor of nearly 3 at intermediate speeds, but the direction is the same: an improvement in fuel economy. At higher speeds [i.e., above 88 kph (55 mph)], fuel economy estimates begin to converge in a negative direction; that is, fuel economy declines sharply (Davis and Hu 1991, 3-68). ²⁸ The same pattern is evident for heavy-duty diesel trucks (Appendix A)

Given the importance of differences in vehicle type on speed-fuel economy estimates, prediction ability could be strengthened considerably by a sample that better reflects the current mix of vehicles in the fleet.

Fuel economy estimates as a function of average speed are based on many of the same drive cycles commonly used in modeling motor vehicle emissions (An et al. 1993, 3), thus raising similar questions concerning the validity of the simulations. However, testing on a broader range of drive cycles has not been a research priority and is unlikely to become one, because of the relatively minor effects of accelerations on fuel use (An and Ross 1993, 12)²⁹ and the fact that fuel economy estimates converge at higher speeds.

Highway capacity additions that raise vehicle speeds and reduce speed variation should improve fuel economy as long as speeds do not exceed approximately 72 kph (45 mph) for automobiles and 80 kph (50 mph) for HDDVs on the affected facilities. The greatest improvements in fuel economy for automobiles should be realized by highway capacity additions that raise vehicle speeds from about 32 kph (20 mph) to about 56 to 72 kph (35 to 45 mph). The magnitude of the improvements depends on the traffic volume affected by the project and the extent of adjustments on other routes. Net effects also depend on predicted changes in fuel economy had the project not been undertaken.

Adding highway capacity will not necessarily produce across-the-board reductions in emissions and fuel consumption, as generally has been believed. Although the current state of knowledge does not allow for precise estimates of the emissions and energy effects of traffic flow changes from capacity enhancement projects, a basic understanding of the major factors contributing to high emission levels and poor fuel economy for the current vehicle fleet suggests where benefits are most and least likely.

Currently available information indicates that the highest levels of emissions and fuel consumption are associated with heavily congested traffic conditions, in which traffic is moving at low and highly variable speeds with frequent stops, idling, and accelerations. These conditions are prevalent on urban arterials with very low speeds and stop-and-start traffic. High levels of congestion are also found on some freeways, particularly at freeway-to-freeway interchanges. Capacity enhancement measures that result in significant smoothing of traffic flows and reductions in speed variability on these facilities, ideally at off-peak as well as peak-period travel times, should initially reduce emissions of all pollutants and conserve energy. The effects will be greater for emissions than for energy use, for CO and VOCs than for NO_x, and for light-duty vehicles than for HDDVs. All else being equal, the more traffic volume affected by the improvement, including traffic on routes parallel to the improved facility, the greater the impact.

Highway capacity additions under moderate or lightly congested conditions at which traffic speeds are less variable will have limited effects on smoothing traffic flows and may result in free-flowing traffic at freeway speeds that will increase energy use and emissions of some pollutants. Fuel use is known to increase rapidly at speeds above 56 to 72 kph (35 to 45 mph) for automobiles and above 80 kph (50 mph) for HDDVs. NO_x emissions for automobiles, and NO_x and VOC emissions for HDDVs, will increase although there is considerable uncertainty about the speed at which these increases begin and the magnitude of the increase for light-duty vehicles. Knowledge about

the behavior of CO and VOC emissions at free-flow freeway speeds is uncertain.

For a broad range of intermediate traffic conditions, highway capacity additions are likely to have modest effects on traffic smoothing and to result in small upward shifts in the distribution of traffic speeds. The effects of such relatively small changes on traffic flow characteristics and related emissions and energy use cannot be predicted with confidence by current models.

Evaluating the net effects of highway capacity expansion projects also requires taking into account any negative effects on emissions and energy use from traffic disruptions during project construction. Finally, it requires predicting changes in traffic flow characteristics and related effects on emissions and energy use had the project not been undertaken.

Understanding how highway capacity additions are likely to change emission levels is only one step in determining how the projects ultimately will affect air quality. CO concentrations, which tend to be localized in a region and exacerbated by congested traffic conditions, are likely to be reduced by capacity enhancement projects that eliminate bottlenecks and smooth traffic flows on major highway links. The effects on ozone are likely to be more problematic. Emissions of NO_x, an ozone precursor, will increase as a result of highway capacity projects that produce free-flow traffic conditions, particularly at freeway speeds. This could pose serious problems in ozone non-attainment areas where conformity requirements mandate reductions in NO_x emissions. Although VOCs, the other major ozone precursor, would be reduced by traffic flow smoothing, other major sources of VOC emissions (i.e., cold starts and evaporative emissions) would remain largely unaffected by capacity enhancement projects.

The ability to predict effects, particularly emission reductions for light-duty, gasoline-powered automobiles, is severely limited. Current emission models rely on average trip speed as the sole descriptor of traffic flow. Variability in speed, road grade, and other factors that strongly influence emissions are not explicitly dealt with. In addition, virtually all emissions testing has been based on a limited set of test cycles with questionable representativeness for specific traffic

flow conditions. Finally, model estimates of predicted changes in emission levels tend to have large variances for a wide range of changes in average trip speeds that are typical of many highway capacity additions.

Diesel emissions can be predicted with greater certainty. They can be measured more directly (no aftertreatment is involved yet), and diesel engines require very little cold start or acceleration enrichment, which are large sources of variance for light-duty vehicle emissions. However, detailed data on diesel particulate emissions as a function of speed are not available. This is a troubling omission, because HDDVs are the primary source of highway vehicle particulate exhaust emissions, and particulate concentrations pose a significant health risk.

Fuel economy estimates from simulation models are relatively reliable and valid indicators of actual in-use vehicle fuel consumption. In part, this reflects the fact that fuel economy is not as sensitive as emissions to changes in traffic flow conditions, particularly speed variation.

Research has begun that could significantly improve understanding of the emissions characteristics of motor vehicles under real-world driving conditions. In response to a requirement of the CAAA, EPA has collected data on driving behavior in selected cities to develop new drive cycles more representative of current driving conditions. CARB has conducted parallel surveys in Los Angeles, has developed new drive cycles, and is testing them on a representative sample of 125 light-duty vehicles. The states have pooled resources to develop and verify a modal emissions model. However, federal leadership and funding are required to expand and accelerate these efforts. The most pressing needs are for the development of an emission rate model sensitive to a wider variety of driving patterns than current models and truly representative of the range of vehicles on the road today and for the collection of vehicle activity data to use as inputs to this model. EPA should become an active participant in developing these models and incorporating them into the regulatory process.

Because the greatest source of variance in estimates of fuel economy is differences in vehicle design and technology, estimates of the

in-use fuel economy of the vehicle fleet should be periodically updated. Data and models for representing the effects of speed and acceleration on motor vehicle fuel economy should also be updated from time to time to be representative of the current vehicle fleet. This could be accomplished as part of the vehicle testing described above.

Traffic simulation models should be modified to provide baseline data on travel speed and speed variation for individual vehicles, in addition to the more traditional average speed data for all vehicles that are currently reported. Differentiation should be made between automobiles and trucks. Such disaggregated data are needed to provide better estimates of emission levels. Current modeled estimates of vehicle velocity and acceleration rates should be validated if they are used to provide inputs to emission models. Traffic simulation models should also be integrated with regional forecasting models so that the long-term impacts of highway capacity additions on travel demand and emissions can be estimated.

Finally, as is typical of many interdisciplinary efforts, lack of a common terminology hampers development of appropriate models. The terminology and definitions used by traffic engineers for key traffic flow parameters (e.g., delay) are not always consistent with terms used by air quality analysts. A strong effort is needed to remove such inconsistencies from research projects and bring traffic and air quality into consistent reference frames.

Additional research and vehicle testing should help reduce the uncertainties associated with modeling emissions and energy use. However, it will take substantial time and investment for the results of this knowledge to be implemented in practical models. Although the models can be improved, they still cannot predict outcomes with absolute certainty. Good policy should allow for some variance in modeled results.

In this chapter the current knowledge of the initial effects of highway capacity additions on emissions and energy use has been identified. If capacity enhancement projects stimulate new travel over time or encourage relocation of residences or businesses at more dispersed, automobile-dependent locations, the initial gains from some projects will be eroded. The potential for these outcomes is explored in the following chapters.

1. Density is defined as the number of vehicles occupying a given length of a lane or roadway, averaged over time. It is generally expressed as vehicles per mile (TRB 1992, 1–6). See discussion of level of service in Chapter 1 of this report.

2. EPA has not developed separate emission factors or speed correction factors for diesel buses. Available data suggest that bus emissions are significantly higher than truck emissions (Appendix A).

3. The old belt-driven chassis dynamometers could not handle acceleration or deceleration rates that exceeded 5.3 kph/sec (3.3 mph/sec) (EPA 1993, 13).

4. Another flaw of the drive cycle tests is that not all vehicles were tested at all speeds.

5. CARB is in process of testing 125 vehicles representative of the current fleet (reduced from an initial testing program of 250 vehicles because of budgetary constraints), to determine emission levels for these and several other cycles (personal communication, L. Larsen, CARB, April 14, 1995).

6. Confidence intervals measure the likelihood, frequently at a 95 percent level of confidence, that the predicted mean for a random sample will lie within the confidence interval bounds. A large confidence interval indicates a wide range of variability around the mean.

7. Guensler uses the same vehicle test data base that CARB used in developing the speed correction factors in the EMFAC model. He reran the regressions and developed confidence intervals using disaggregated test results. By contrast, the data aggregation techniques used by CARB did not retain the data variability necessary to establish confidence intervals. Guensler is undertaking the same kind of analysis for the MOBILE model, but the results are not yet available.

8. Transient and cold start effects account for 10 percent or less of VOC and NO_x emissions for heavy-duty diesel trucks (Duleep 1994, 16). Thus, it is possible to develop emission estimates for almost any arbitrary driving cycle with reasonable accuracy from a steady state emissions map of the engine (Appendix A).

9. The actual level of the average speed depends on the free-flow speed of the facility, that is, the speed of a passenger car traveling at low traffic volume conditions (LOS A).

10. The emissions and fuel consumption data were developed by the Oak Ridge National Laboratory for the Federal Highway Administration in the mid-1980s (see discussion in Chapter 2). These data are currently being updated.

11. Two methodological approaches were used. The first used instrumented vehicles recruited from inspection and maintenance stations to collect second-by-second data on driving behavior. This approach allows for collecting data on the full vehicle trip but may be biased because of the driver's knowledge of the instrumentation. The second approach involved chase cars, which followed randomly selected target vehicles. The approach is nonintrusive but does not allow data to be collected for complete trips (EPA 1993, 36, 43).

12. Decelerations, another characteristic of stop-and-start driving, were not found by EPA to increase emissions significantly. However, testing was performed on current technology (multipoint fuel injection) vehicles. With older carbureted technology, some increase in VOC emissions could occur (personal communication, John German, EPA, Feb. 23, 1994).

13. On arterial roads, 18.7 percent of the driving was at acceleration rates greater than 4.8 kph (3 mph) per second, versus 2.3 percent on freeways. Twenty-eight percent of the driving was at cruise-type conditions on arterials versus 53 percent on freeways. The freeway driving, however, did not include driving on ramps to enter the freeway (Effa and Larsen 1993, 3).

14. Although these events are less common, sharp accelerations at high speeds [i.e., above 80 kph (50 mph)] contribute a disproportionate share of total CO emissions (LeBlanc et al. 1994, Figure 3b). Duration of accelerations also appears to matter; CO emissions increase rapidly as the duration of a severe acceleration event increases (LeBlanc et al. 1994, 10).

15. Traffic-calming techniques fall into two general categories: traffic management strategies (e.g., signalization system improvements, transportation system and parking management, truck restrictions, and speed limits and enforcement) and physical design (e.g., narrowing and curving roadways, speed bumps, and pavement design) (Project for Public Spaces 1993, 15).

16. Emissions were estimated by driving instrumented cars both aggressively and calmly through the traffic-calmed area. These estimates are subject to the same limitations as is instrumented vehicle drive cycle research in the United States: potential bias because the driver is aware of the instrumentation and representativeness of individual driving patterns of

the entire traffic stream (personal communication, Carmen Hass-Klau, consultant, June 3, 1994. Hass-Klau believes that if traffic calming is introduced in a widespread way, driving behavior will be affected).

17. Schemes designed to encourage steady driving speeds are believed to be more effective in reducing emissions than those designed to encourage slow speeds per se. A comprehensive evaluation of traffic-calming projects (Pharoah and Russell 1989, 48–49) found that schemes that have resulted in very low speeds increase emissions of CO and VOCs because they caused more acceleration and deceleration and greater use of second gear. However, there was generally no net increase in air pollution because of the overall reduction in traffic volume.

18. Driving cycles more representative of current driving patterns are being developed. Sierra Research, for example, under contract to EPA, has developed three cycles in addition to the FTP to represent the range of in-use vehicle operation: (a) a start cycle that represents driving that occurs during the first 4 min after the start of the vehicle, (b) a non-FTP cycle that represents the distribution of speeds and accelerations outside the boundary of the FTP, and (c) a remnant cycle that represents everything else (Enns et al. 1993, 3, 4). CARB has developed seven freeway cycles and three arterial cycles, which are discussed in the text.

19. EPA may add an option to account for some “off-cycle” events (e.g., high speeds and high accelerations) in its next revision of the MOBILE model (person communication, Dave Brzezinski, EPA, Aug. 17, 1994). In fact, the agency has introduced a proposed rulemaking to add a supplementary cycle to the FTP, which would capture some of these off-cycle events (Federal Register 1995).

20. The work is being performed under contract to the Georgia Institute of Technology. No extensive vehicle testing is planned, however (personal communication, Carl T. Ripberger, EPA, Aug. 19, 1994).

21. Modal models predict second-by-second emissions for a wide range of engine speeds and loads under steady-state operations. The problem in predicting emissions at a particular engine speed and load is that the results are affected by vehicle operation in the period preceding the desired predicted value. For example, a modal model could predict emissions for a vehicle operating at a speed of 40 kph (25 mph) and an acceleration rate of 4.8 kph/sec (3 mph/sec). However, predicted emissions at 40 kph (25 mph) would differ depending on whether the vehicle has accelerated from 32 to 40 kph (20 to 25 mph), whether it is operating at a steady-state speed of 40 kph (25 mph), or whether it has decelerated from

48 to 40 kph (30 to 25 mph). The challenge is to capture these transient operations. An alternative approach—an “event-based” modal model—could partially address these concerns. In an event-based model, modal emissions are predicted for a complete driving event representative of actual driving conditions, such as a 15-sec acceleration from 0 to 80 kph (0 to 50 mph). This approach captures the variability in emissions introduced by prior operations.

22. The Volpe National Transportation Systems Center has developed a PC-based model framework to estimate the relative impacts on emissions (and fuel economy) of intelligent transportation system user services. The models include a regional planning model (SYSTEM II), two traffic simulation models (FREQ, a freeway/ramp model, and TRANSYT-7F, an arterial model), and VEHSIME, a modal model that produces second-by-second estimates of emissions and fuel consumption under hot stabilized vehicle operation (Sierra Research 1994).

23. Researchers at the University of California, Riverside, are also developing a modal emissions model that can be integrated with traffic simulation models to more accurately portray the emissions effects of dynamic vehicle activities (e.g., accelerations and decelerations) on traffic networks. Although the researchers are using data from their own instrumented vehicle, they are depending primarily on outside sources for vehicle modal emissions data (CE-CERT 1993, 12).

24. However, morning congestion may be more important for the ozone precursors, because the pollutants have more time to be exposed to sunlight, which drives the chemical reactions that form ozone (Horowitz 1982, 70).

25. In 1993, approximately 90 percent of the fleet was equipped with the newer multipoint injection technology (personal communication, John German, EPA, Dec. 13, 1993). The technology enables each engine cylinder to operate with a more consistent air-fuel ratio.

26. Cold starts are not a large problem for diesel-powered heavy-duty trucks. VOC and particulate emissions are about 10 percent higher in the cold start mode (Duleep 1994, 12).

27. The Navistar model is well documented; its representatives claim that the model has been validated to within ±5 percent in real-world tests. The model has three built-in cycles to represent city, suburban, and highway driving conditions with average real-world tests. The model has three built-in cycles to represent city, suburban, and highway driving conditions with average speeds of 32, 64, and 88 kph (20, 40, and 55 mph), respectively (Appendix A).

28. The range between the highest and lowest estimates of fuel economy for the 12 light-duty, gasoline-powered test vehicles is as follows: 40 kph, 15.0 kpl; 56 kph, 12.8 kpl; 72 kph, 5.6 kpl; 88 kph, 6.1 kpl; 104 kph, 5.1 kpl; and 120 kph, 4.4 kpl (25 mph, 35.4 mpg; 35 mph, 30.1 mpg; 45 mph, 13.1 mpg; 55 mph, 14.4 mpg; 65 mph, 12.1 mpg; and 75 mph, 10.4 mpg).

29. An et al. (1993) found that three variables—average speed, free-flow speed or maximum attempted speed when there is no congestion, and the percentage of total trip time the vehicle is stopped—explained more than 90 percent of the variations from using different drive cycles (p. 3). Diesel engines do not require acceleration enrichment, and the air-fuel ratio during a transient acceleration/deceleration is more carefully controlled than in a gasoline engine (Appendix A).

CARB

California Air Resources Board

CE-CERT

College of Engineering–Center for Environmental Research and Technology

EPA

Environmental Protection Agency

NCHRP

National Cooperative Highway Research Program

NRC

National Research Council

TRB

Transportation Research Board

An, F., and M. Ross. 1993. Model of Fuel Economy and Driving Patterns. Presented at the 72nd Annual Meeting of the Transportation Research Board, Washington, D.C.

An, F., M. Ross, and A. Bando. 1993. How To Drive To Save Energy and Reduce Emissions in Your Daily Trip . RCG/Hagler, Bailly, Inc., Arlington, Va., and the University of Michigan, Ann Arbor, 10 pp.

CE-CERT. 1993. The Development of an Integrated Transportation/Emissions Model to Predict Mobile Source Emission. South Coast Air Quality Management District Contract AB2766/C0004 . University of California, Riverside, May.

Cicero-Fernández, P., and J.R. Long. 1993. Modal Acceleration Testing on Current Technology Vehicles. Presented at Conference on the Emission Inventory: Perception and Reality, Pasadena, Calif., Oct. 18–20.

Davis, S.C., and P.S. Hu. 1991. Transportation Energy Data Book: Edition 11. ORNL-6649. Center for Transportation Analysis, Energy Division, Oak Ridge National Laboratory, Tenn., Jan.

Duleep, K.G. 1994. Briefing on Heavy Duty Diesel Vehicle Emissions. Energy and Environmental Analysis, Inc., Arlington, Va.

Effa, R.C., and L.C. Larsen. 1993. Development of Real-World Driving Cycles for Estimating Facility-Specific Emissions from Light-Duty Vehicles. Presented at Conference on the Emission Inventory: Perception and Reality, Pasadena, Calif., Oct. 18–20, 20 pp.

Enns, P., J. German, and J. Markey. 1993. EPA's Survey of In-Use Driving Patterns: Implications for Mobile Source Emission Inventories. Office of Mobile Sources, Certification Division, U.S. Environmental Protection Agency, Ann Arbor, Mich.

EPA. 1993. Federal Test Procedure Review Project: Preliminary Technical Report . Office of Air and Radiation, May, 161 pp.

Federal Register. 1995. Proposed Regulations for Revisions to the Federal Test Procedure for Emissions from Motor Vehicles. Vol. 60, No. 25, Feb. 7, pp. 7404–7424.

Greene, D.L. 1981. Estimated Speed-Fuel Consumption Relationships for a Large Sample of Cars. Energy, Vol. 6, pp. 441–446.

Guensler, R., and A.B. Geraghty. 1991. A Transportation/Air Quality Research Agenda for the 1990s. Presented at the 84th Annual Meeting and Exhibition, Air and Waste Management Association, Vancouver, British Columbia, Canada, June 16–21, 32 pp.

Guensler, R., D. Sperling, and P. Jovanis. 1991. Uncertainty in the Emission Inventory for Heavy-Duty Diesel-Powered Trucks. UCD-ITS-RR-91-02. Institute of Transportation Studies, University of California, Davis, June, 146 pp.

Guensler, R., S. Washington, and D. Sperling. 1993. A Weighted Disaggregate Approach To Modeling Speed Correction Factors . Presented at the 72nd Annual Meeting of the Transportation Research Board, Washington, D.C., 44 pp.

Guensler, R. 1994. Vehicle Emission Rates and Average Operating Speeds. Ph.D. dissertation. University of California, Davis.

Horowitz, J.L. 1982. Air Quality Analysis for Urban Transportation Planning. The MIT Press, Cambridge, Mass., 387 pp.

Kelly, N.A., and P.J. Groblicki. 1993. Real-World Emissions from a Modern Production Vehicle Driven in Los Angeles. Air and Waste, Vol. 43, Oct., pp. 1351–1357.

LeBlanc, D., M.D. Meyer, F.M. Saunders, and J.A. Mulholland. 1994. Carbon Monoxide Emissions from Road Driving: Evidence of Emissions due to Power Enrichment. Presented at the 73rd Annual Meeting of the Transportation Research Board, Washington, D.C., 23 pp.

McGill, R. 1985. Fuel Consumption and Emission Values for Traffic Models. FHWA/RD-85/053. Oak Ridge National Laboratory, Tenn., May, 90 pp.

NCHRP. 1994. Research Problem Statement: Development of a Modal-Emissions Model . NCHRP Project 25-11. Transportation Research Board, Oct. 6, 4 pp.

NRC. 1992. Automotive Fuel Economy: How Far Should We Go? National Academy Press, Washington, D.C., 259 pp.

Pharoah, T., and J. Russell. 1989. Traffic Calming: Policy and Evaluations in Three European Countries . Occasional Paper 2/89. Department of Planning Housing and Development, South Bank Polytechnic , London, United Kingdom, 67 pp.

Project for Public Spaces. 1993. The Effects of Environmental Design on the Amount and Type of Bicycling and Walking. National Bicycle and Walking Study. FHWA Case Study 20. FHWA-PD-93-037. Federal Highway Administration, U.S. Department of Transportation , April, 40 pp.

Sierra Research, Inc. 1993. Evaluation of “MOBILE” Vehicle Emission Model. Report SR93-12-02. Sacramento, Calif., Dec. 7.

Sierra Research, Inc. 1994. Development of an Emissions, Fuel Economy, and Drive Cycle Estimation Model for IVHS Benefits Assessment Framework. SR94-07-01. Sacramento, Calif., July 9.

TRB. 1992. Special Report 209: Highway Capacity Manual. 2nd edition revised. National Research Council, Washington, D.C.

An argument against adding highway capacity is that the roads will simply fill up again with traffic as service levels improve. Traffic conditions, the argument continues, will become congested again but with larger traffic volumes, thus increasing total emissions and energy consumption. In this chapter the current knowledge about the stimulative effects of highway capacity projects on motor vehicle trips and travel is reviewed (Figure 4-1). An introductory section on the primary determinants of travel demand and recent travel trends is provided.

Travel is a derived demand; that is, it is derived from the activities of households and businesses that locate in a metropolitan area. Travel activities can be classified as commercial, involving goods movement and distribution, and personal. Personal travel is typically categorized as work travel and other personal travel (e.g., for shopping or recreation).

The primary factors affecting metropolitan travel growth are demographic and economic. Metropolitan areas have grown tremendously in population. In 1990, nearly half the population (124 million people) lived in metropolitan areas with populations of more than 1 million, a 60 percent increase since 1960 (Rossetti and Eversole 1993, ES-2). Growth has been particularly rapid in the South and West, fueled in part by high rates of immigration (Pisarski 1992a, 4). During the same period household size has dropped sharply, from 3.24 to 2.65 persons per household (Rossetti and Eversole 1993, 2-1). To the extent that there are economies of scale in household trip making, persons organized in smaller households make more trips on the average than if they were part of larger households.

Metropolitan area economic growth has accompanied population increases. Job growth in metropolitan areas outpaced the national average between 1960 and 1990, reflecting in part the entrance of the baby boomers into the work force (Rossetti and Eversole 1993, 2-1). The number of workers per household also grew. Women in the work force increased from one-third to nearly one-half of the metropolitan area work force, becoming an important factor affecting travel demand (Rossetti and Eversole 1993, 2-1). During the same period, growth in personal income resulted in increased household automobile ownership and travel.

These growth trends have led to substantial increases in both trips and travel in urban areas.¹ According to the Nationwide Personal Transportation Survey,² between 1983 and 1990 household vehicle trips increased by 22 percent and household vehicle miles traveled (VMT) by 34 percent³ (Vincent et al. 1994, 10). Although travel for commuting purposes still represents the largest share of urban VMT (36 percent), travel for family and other personal business other than shopping was the fastest-growing element of household VMT. Trips for this purpose exceeded work trips for the first time in 1990 (Vincent et al. 1994, 1-1, 2-1). The implications of this shift for travel demand are important because a larger portion of nonwork trips are likely to be discretionary and thus more responsive to changes in the cost of travel.

The location of population growth and economic activity in metropolitan areas has also affected the amount and type of travel. Most

population and job growth has taken place in the suburban portions of metropolitan areas (Mieszkowski and Mills 1993; Chinitz 1993). The trend toward decentralization has increased average trip lengths for all travel purposes between 1983 and 1990, with commute trips increasing the most (Vincent et al. 1994, 3-3). However, commuting trip speeds have remained relatively stable,⁴ reflecting more travel on higher-speed, suburban roads and a shift from slower to faster commuting modes (i.e., a decrease in transit use and carpooling and an increase in driving alone).

Low density suburban growth is correlated with more automobile travel and higher automobile trip rates than occurs in core central cities where alternative modes of travel are available (Dunphey and Fisher 1993). For all urban areas in the United States, the share of urban trips made by automobile is 87 percent. The corresponding figure for urban areas with rail transit systems and populations greater than 1 million is only 78 percent (Vincent et al. 1994, 4-4, 6-11). These locations provide a variety of transportation services and have residential and employment densities that promote less reliance on the private vehicle for trip making (Vincent et al. 1994, 2, 4-2).

Most personal travel in urban areas is by automobile, reflecting both low-density development that does not support alternatives to driving for most travel activities and the travel speed advantages of private vehicles.⁵ In 1990, 87 percent of all urban person-trips were made by private vehicle, 2 percent by public transportation, and the remainder by all other modes⁶ (Vincent et al. 1994, 4-4). Commercial travel in major metropolitan areas is primarily by truck, although some rail shipments are brought to key distribution centers in urban areas.

The cost of transportation to the individual and to businesses also affects the demand for travel. Although the provision of highway capacity in urban areas has tapered off in recent years,⁷ it has resulted in high levels of accessibility and substantial reductions in the time-related cost of motor vehicle travel. Out-of-pocket costs of motor vehicle travel have also declined as a share of household budgets with the rise in personal income and the drop in oil prices over the past decade. Consumer surveys suggest that about 9 percent of typical household budgets is spent on gas and oil and other vehicle expenses (BTS 1994, 2).⁸ In the manufacturing and retail sectors, transporta-

tion costs typically account for between 1 and 4 percent of total production costs (Appendix C). All else being equal, the lower the cost of highway travel, the greater the propensity to travel and the less priority residents and businesses will give to transportation relative to other preferences and costs of doing business.

The link between the low cost of and increased demand for travel suggests that raising the price of travel should reduce demand. Empirical evidence suggests that motorists respond to increases in bridge and turnpike tolls, transit fares, and parking fees (Harvey 1994). The decline in demand depends on the size of the price increase, the current costs of travel, and the capacity of alternative roads and transit systems (Bhatt 1994).

Are current travel trends likely to continue? There is evidence that the rate of travel growth may be slowing. Historical trends indicate some reduction in the rate of growth of urban VMT (Figure 4-2). In addition, automobile ownership patterns may be reaching a saturation point. In 1989 there were 0.95 vehicles per person of driving age (Lave 1992, 5-7). As the proportion of household members who are eligible to drive and have access to a car approaches 100 percent, trip making per household and VMT per vehicle increase, but at a less rapid rate, with the purchase of additional vehicles (Hu and Young 1992, 30; Vincent et al. 1994, 1-16). Moreover, although the number of households with more than one vehicle has grown, households with three or more vehicles are mostly located in rural states (Pisarski 1992b, 19–20).⁹ Finally, some researchers argue that the potential for substantial additions to the driver pool is limited by the aging of the driving population and a reduction in the rate of growth of women in the work force.

Offsetting these trends are the rise in young immigrant populations and the potential for increased travel by older drivers. For example, individuals 65 years old or older took 6 percent more trips and traveled 25 percent more on the average in 1990 than in 1983. However, travel in this age group in 1990 was still 30 percent below the average annual number of trips and 42 percent below the average annual miles traveled of all other age groups (Hu and Young 1992, 42).

In summary, regional and metropolitan differences in immigration levels, work force participation levels, and overall rates of population and economic growth are likely to determine travel and trip demand levels in specific metropolitan areas.

FIGURE 4-2 Changes in the rate of growth of urban VMT, 1945–1990 (FHWA 1987, 225-228; FHWA 1992, 193).

By reducing the onerousness of travel as perceived by individuals, expansion of highway capacity can affect decisions about where, when, and how to travel. The primary impact of adding highway capacity is to reduce travel times in the corridor in which the improvement is located.

Highway capacity additions can also improve the reliability of travel by diminishing the day-to-day variability in the time required to make a trip. Vehicles using congested facilities experience daily delays and unanticipated delays that can occur because of disabled vehicles and crashes. Adding capacity to these facilities reduces the likelihood¹¹ and severity of delays from both recurring congestion and nonrecurring incidents.

Finally, highway capacity additions can also affect the out-of-pocket cost of highway travel to the extent that fuel consumption and other

vehicle operating costs vary with speed and level of congestion. These effects are generally much smaller than travel time effects and are not perceived by most drivers.

Potential direct effects of highway capacity improvements on travel patterns include the following:

• Changes in the route used to make a given trip: Construction of a new highway or the addition of capacity to an existing highway will allow some automobile users to reduce their travel times by shifting their route to the new or improved facility. Route shifts can increase or decrease total highway system use as measured in VMT by increasing or decreasing the circuity of trips. In most cases these effects are minor. Of more importance from an environmental perspective, route diversion can result in traffic shifts toward or away from areas that are more sensitive to local air quality and noise impacts.

• Changes in the time of day a given trip is made: Travelers may alter the time of day during which a trip is made in response to congestion levels. For example, individuals may deliberately avoid periods of congestion in scheduling shopping or other nonwork trips. Also, to the extent that their work requirements permit, individuals may schedule journeys to and from work outside of the period of most severe congestion. Adding capacity to a highway can reduce the severity and duration of congestion experienced on that facility. When peak-period congestion decreases, some of the trips that were shifted in time to avoid congestion in peak periods may be shifted back to the peak period.

• Changes in the mode used to make a given trip: Capacity additions can increase speeds and reduce automobile travel times, enhancing the attractiveness of highway use relative to other modes that do not benefit from the capacity improvement. The importance of mode shifts as a source of additional highway use depends heavily on the presence, type,¹² and use of alternative modes such as transit¹³ in the corridor where capacity additions are made. Bicycle and pedestrian travel may also be discouraged if capacity

additions improve traffic flow for motor vehicles to the detriment of slower-moving modes.

• Changes in trip destinations: Capacity additions can increase the relative attractiveness of some trip destinations by reducing travel times to those destinations. Capacity additions can thus cause individuals to shift the destinations of their trips. This effect is more important for shopping and recreational travel (for which individuals have some flexibility in choosing destinations) and less important for work-related travel (for which individuals have little or no flexibility in choosing destinations in the short term). The substitution of one trip destination for another can increase or decrease total highway system use measured in VMT, depending on which destination is closer. The net effect of changes in trip destinations because of highway improvements is usually to increase VMT (i.e., new destinations are typically further away than the destinations they replace).

• Increases in the number of trips made: Because travel is a derived demand that reflects the need to carry out other activities, highway capacity additions are unlikely to significantly affect the number of home-to-work trips made. However, capacity additions could influence travel associated with more discretionary activities, for which the cost of travel might be an appreciable part of the total cost of the activity. In addition, improved travel times may reduce driver incentives to link trips (e.g., stopping for gas on the way home from the grocery store).

The time periods over which these changes occur will likely vary. Most route diversion (which merely involves a shift from one route to another with no change in destination or time of day) may occur soon after a new or expanded facility opens, as drivers learn about the new route and the time savings it offers. Changes in the time of day for shopping trips may also occur soon after the new or expanded facility opens. Changes in the mode used for work travel, shifts in trip destinations, and new trips could occur more gradually because they involve more significant changes in travelers' activity patterns.

In addition to the direct effects on traffic noted above, highway capacity additions can influence traffic levels by affecting other types of decisions, as illustrated by the following: