Skip to contentU.S. Department of Transportation, Federal Highway Administration
FHWA HomeFeedback
Environment
FHWA > HEP > Environment > Air Quality > Conformity > Research
Previous Contents Next

2 Project Approach and Discussion

2.1 Background

Emission factor models are the underlying analytic tool for the development and evaluation of mobile source emissions in State Implementation Plans (SIPs), conformity determinations, and project-level impact analyses. As such, changes in these emission factor models will directly impact project-level analyses. The increased flexibility in specifying inputs to MOBILE6 gives the analyst increased capabilities to provide site-specific detail, but comes at the expense of additional resources needed in gathering and developing the additional data.

The MOBILE series of models have been developed using a fleet-wide average emission rate determined from individual vehicle type emission rates for each vehicle class and the fractions of vehicle miles traveled (VMT) for each vehicle class. This makes the models suitable for regional-scale modeling, but less appropriate for project-level analysis where site-specific real-time fleet emissions are needed. EPA has recognized this limitation and is conducting research in developing a more appropriate model for project-level analysis. However, until EPA develops an alternative, project-level analyses will need to be conducted using MOBILE6. Hence, the current need to evaluate MOBILE6 for project-level analyses.

Across the country, many Metropolitan Planning Organizations' (MPOs) and state Departments' of Transportation (DOTs) staff are familiar with the strengths and weaknesses of previous versions of EPA's MOBILE model. However, most have just begun using MOBILE6 and have limited familiarity with the model for use in project-level dispersion modeling.3 Therefore, one goal of this project was to inform practitioners in the transportation community about the most critical and sensitive input parameters in MOBILE6 affecting project-level transportation and air quality analyses so that resulting transportation decisions are improved.The results of this project provide transportation planners with information that can help them develop cost effective strategies for using MOBILE6 in project-level dispersion analyses and design mitigation strategies when hotspot modeling is needed.

The CAL3QHC roadway intersection model is typically employed to determine potential air quality impacts for project-level analysis. CAL3QHC uses both an idle emission factor and an appropriate emission factor based on each link's free flow speed. In determining the appropriate CO emission factor using MOBILE6, input parameters could reflect such site-specific factors as: fleet mix, age distribution, facility type, air conditioning usage, start distributions, soak time, starts per day, temperature, inspection and maintenance (I/M) program, fuel (% oxygenate), reid vapor pressure, weekday versus weekend, and vehicle speed distribution. How CAL3QHC will respond to these MOBILE6 emission factors will depend on how these parameters change the idle and free-flow speed emission rates inputted to CAL3QHC. This study examines how the most important of these parameters affect the CAL3QHC results.

2.2 Approach and Analysis

This study explores how the use of MOBILE6 impacts project-level analyses. The study is divided into three main components addressing the questions:

The first question is addressed through exploration and investigating of the following topics:

The third question is addressed through model applications for the following topics:

2.2.1 Impacts from the use of MOBILE6 in Project-Level Analysis

Base Emission Rate Changes from MOBILE5 to MOBILE6

Various studies on emission model runs conducted within EPA have indicated that MOBILE6 usually provides different emission factors than the MOBILE5 model. This is due to a variety of differences in model characteristics, such as updated information on in-use deterioration rates, new technologies (e.g., on-board diagnostics), updated emission base emission rates, and incorporation of new regulations.

A number of changes in MOBILE5 to MOBILE6 make it more complicated to directly compare outputs between the two models. For example, in MOBILE5, specification of a single average speed would return a composite emission rate that represents the best estimate of cumulative emissions for vehicle operations over the course of a driving cycle, such as the Federal Test Procedure (FTP) driving cycle. In MOBILE6, emission factors by speed are specified by facility type. Thus an appropriate facility type must be matched to each speed to provide the best comparison.

MOBILE6 contains a number of changes relative to the MOBILE5 version. These changes can be broadly classified as follows:

Because of these many changes between the two versions of MOBILE and the fact that many transportation agencies use national default values for a large number of the input variables, the baseline comparison between the two versions of the MOBILE model uses national default values for both models. In many cases, results between the models may be different because of more up-to-date information used internally in MOBILE6. Thus, two types of comparisons have been performed: 1) to identify CO emission factor changes as a result of switching models for a baseline (national default) and 2) to identify changes that would most likely be seen by the transportation air quality analyst in applying MOBILE6 for a project-level analysis.

MOBILE5 and MOBILE6 Using National Defaults for a Set of Project-Level Vehicle Speeds and Temperatures

A comparison between MOBILE5 and MOBILE64 models was made applying national default values for a base year, 2005, and 30-years in the future, 2035. The scenarios were performed over a range of ambient temperatures: 0, 10, 20, 30, 40, 50, 60, 70, 80 and 90 °F and a range of speeds: idle (0), 3.4, 7.1, 12.1, 19.5 and 35.9 miles per hour (mph). These speeds were selected based on the driving cycles used in MOBILE5 based on statistical analysis of emissions testing from the eight driving cycles used internally within MOBILE5 that are used to calculate speed-specific emissions. MOBILE6 has a similar makeup, but uses slightly different driving cycles, although the lowest two speeds are identical. To provide a direct comparison, the same speeds used in MOBILE5 were used in the MOBILE6 simulations.

Base input files were prepared for both models using national defaults. An I/M program was defined for the base case with the following specifications:

Other general specifications for the base file were as follows:

MOBILE6 does not directly model idle emissions. The MOBILE6 User Guide recommends that idle emissions be estimated from modeling the 2.5 mph speed bin and then multiplying over the hour to provide a grams/hour  emission rate. This approach was used to determine the idle emission rate from MOBILE6.

Tables A.1.1 through A.1.6 are presented in Appendix A with the CO emissions factors for MOBILE5 and MOBILE6 for each of the temperature/average speed combinations for 2005 and 2035, as well as percentage changes relative to MOBILE5 based on the formula:

Percentage change equals (the MOBILE6 emissions factor minus the MOBILE5 emissions factor) divided by the MOBILE5 emissions factor.

These results are referred to as the base case scenarios and are discussed as part of the project-level scenarios presented in the following section.

MOBILE5 and MOBILE6 Comparison for Project-level Scenarios

A review of the MOBILE6 changes that may impact the project-level analysis suggested that the scenarios described below will be of primary interest to the project-level analyst in assessing the change between using MOBILE5 and MOBILE6. These selected scenarios reflect both typical applications and/or potential changes from national distributions with anticipated significant impacts on CO emission factors. The emphasis has been placed on light-duty gasoline vehicle and trucks as they are the largest category of vehicle types, have significant differences in CO emission rates between vehicles and light-duty trucks, and have historically been the largest contributor to CO emissions. The scenarios evaluated were:

For the first scenario, without an inspection and maintenance program, new input files were created without I/M programs for each of the years to model. These results are shown in Appendix A in Tables A.2.1, A.2.2, and A.2.3 for year 2005 and A.2.4, A.2.5, and A.2.6 for year 2035.

  1. For the shift of plus and minus three years in the average fleet distribution for light-duty vehicles and trucks, the following steps were followed:

  2. National default registration distributions for MOBILE5 and MOBILE6 were obtained for all vehicle types.

  3. For light-duty vehicles and trucks only, the average age distribution was calculated for the year in which the median value (50%) is found in the accumulated fleet distribution. That year was considered to be the average age of the fleet. For light-duty vehicles and MOBILE5 that was around 6 years. For MOBILE6 and light-duty vehicles that age was around 7 years.

    Once the average age was found, the fleet was made three years younger or older by altering the distribution in a way that, while keeping original proportions between years, the average age of the fleet distribution would be three years newer or older. For example, for light-duty vehicles and MOBILE5, a fleet three years younger should have the same sum of accumulated vehicle fractions for the first three years of the distribution as it previously had for the first 6 years, and the same sum in the remaining 22 years as it previously had for the last 19 years. In order to do that, the sum of the fractions for the first 6 years was subtracted from the sum of the fractions of the first three years. That difference was then summed to each of the fractions in the first three years by multiplying that difference by the proportion of the vehicle fraction of each of the years with respect to the three years. This can be expressed mathematically in the general expression:

New distribution for years 1-3 equals the distribution for years 1-3 plus (the summation of the years 1-6 distributions minus the summation of the years 1-3 distributions) multiplied by the distribution for years 1-3 divided by the summation of the years 1-3 distributions.

Equation 1. Calculation of New Distribution for Three-Years-Newer Fleet, First 3 Years

For the last 22 years of the new distribution a similar procedure was followed but the difference was set between the last 19 years of the distribution and the last 22 years of the distribution:

New distribution for years 4-25 equals the distribution for years 4-25 plus (the summation of years 7-25 distributions minus the summation of the years 4-25 distributions) multiplied by the distribution for years 2-25 divided by the summation of the years 4-25 distributions.

Equation 2. Calculation of New Distribution for Three-Years-Newer Fleet, Last 22 Years

To age the fleet three years, the distribution for the first 9 years was changed following a similar procedure.

  1. The new registration distributions were incorporated directly into MOBILE5. For MOBILE6, two new external registration distribution files were prepared - one for the light-duty vehicles and the other for light-duty-trucks. The other vehicle types used the national defaults. These changes in age distribution are illustrated in Figure 2.2.1 through 2.2.3 for MOBILE6 for the three year shifts and for the national default for light-duty vehicles, light-duty truck type 1 or 2, and light-duty truck type 3 or 4. Similar patterns are seen for MOBILE5. The resulting emission factors and percentage changes are shown in Appendix A, Tables A.3.1, A.3.2 and A.3.3 for 2005 and A.3.4, A.3.5 and A.3.6 for 2035 for the three- year-newer fleet. The results for the three- years-older fleet are shown in Appendix A in Tables A.4.1, A.4.2 and A.4.3 for 2005 and A.4.4, A.4.5 and A.4.6 for 2035.

Figure 2.2.1. Cumulative Frequency Distributions of Light-duty Vehicles
as Used in MOBILE6

This chart depicts cumulative percentage of light-duty vehicle fleet for vehicle ages from 1 to 25, for the default MOBILE6 distribution, as well as average 3-year older and 3-year new fleets. The chart is used to illustrate the average default vehicle of around seven and a half years.

Figure 2.2.2. Cumulative Frequency Distributions of Light-duty Trucks (Type 1 or 2)
as Used in MOBILE6.  

This chart depicts cumulative percentage of light-duty truck fleet for vehicle ages from 1 to 25, for the default MOBILE6 distribution, as well as average 3-year older and 3-year new fleets. The chart is used to illustrate the average default light-duty truck (type 1 or 2) of around seven years.

Figure 2.2.3. Cumulative Frequency Distributions of Light-duty Trucks (Type 3 or 4)
as Used in MOBILE6.

This chart depicts cumulative percentage of light-duty truck fleet for vehicle ages from 1 to 25, for the default MOBILE6 distribution, as well as average 3-year older and 3-year new fleets. The chart is used to illustrate the average default light-duty truck (type 3 or 4) of around eight and a half years.

The two final scenarios involved changing the vehicle fleets mix. Neither model allows for the direct modification of the vehicle fractions within the model. In order to modify the fleet distribution, the vehicle miles fraction, which defines the fractions of miles traveled for each vehicle type in the vehicle fleet, is adjusted.

National default VMT fractions were used for each of the years to increase or decrease the proportion of light-duty vehicles in the fleet. To increase the light-duty vehicle percentage in the fleet by 30%, the corresponding VMT fraction for light-duty vehicles for each of the years was increased by 30% (see equation 3). The other fractions were proportionally decreased as indicated in equation 4.

New VMT fraction for LDV equals 1.30 multiplied by the old VMT fraction for LDV.

Equation 3. Calculation of VMT Fraction Increased by 30% for Light-duty Vehicles

New VMT fraction for vehicles other than LDV equals old VMT fraction for vehicles other than LDV plus (1 minus (the summation of the new VMT fractions from LDV plus the summation of the old VMT fractions from vehicles other than LDV)) multiplied by the old VMT fraction from vehicles other than LDV divided by the summation of the old VMT fractions from vehicles other than LDV.

Equation 4. Calculation of New VMT Fractions for Vehicles Other Than Light-duty Vehicles

Similarly, the VMT fraction was decreased by 30%, as follows:

New VMT fraction for LDV equals 0.70 multiplied by the old VMT fraction for LDV.

Equation 5. Calculation of VMT Fraction Decreased by 30% for Light-duty Vehicles

The other fractions were increased also using Equation 4.

The new VMT fractions were input to both MOBILE5 and MOBILE6. The resulting emission factors and percentage changes can be found in Appendix A, Tables A.5.1, A.5.2 and A.5.3 for 30% decrease in the light-duty vehicle (LDV) fraction in 2005; in Tables 5.4, 5.5 and 5.6 for 30% decrease in 2035; in Tables A.6.1, A.6.2 and A.6.3 for 30% increase in 2005; and in Tables A.6.4, A.6.5 and A.6.6 for a 30% increase in 2035.

Discussion of Results

For the 2005 base case files, the percentage changes are negative for all temperatures and for all speeds from idle to 19.5 mph, indicating the emission factors calculated by MOBILE5 for those speeds are larger than those calculated by MOBILE6. Also, higher speeds correspond with lower percentage changes. The negative percentage change decreases for low temperatures (0-20 °F), increases over medium temperatures (30-60 °F for speeds idle to 3.4 mph and 30-70 °F for speeds 7.1 to 19.5 mph) and decreases again for high temperatures (70-90 °F for speeds idle to 3.4 mph and 80-90 °F for speeds 7.1 to 19.5 mph), but the change is small  . Percentage changes are positive for the higher speed, 35.9 mph, indicating larger MOBILE6 factors than MOBILE5 factors. For this speed, there is an initial increase in the percent change for low temperatures (0-10 °F), a decrease for medium to medium-high temperatures (20 to 70 °F) and a final increase for the higher temperatures (80-90 F). In summary, there is a clear trend showing that MOBILE5 has larger emission factors for lower speeds and that MOBILE6 has larger emission factors for higher speeds.

For the 2035 base case, the trends are very similar to the 2005 base case; however, now all of the percentage changes are negative, indicating that MOBILE5 emission factors are larger are larger than MOBILE6 factors for all the speed and temperature combinations modeled. The changes are also larger than those seen for 2005, but similar to the 2005 results, percent changes decrease with increasing speed. Changes with increasing temperature are larger than those for 2005, but still small. Overall, these results show that, by 2035, MOBILE6 emission factors for even the higher speeds are lower than MOBILE5 emission factors.

For the no inspection and maintenance program for the 2005 scenario, the emission factors for both MOBILE5 and MOBILE6 increase with respect to the base, as expected. The percentage change values are similar to, but slightly lower than, those calculated for the corresponding base case scenario at low speeds. For speeds higher than 12.1 mph, the changes become slightly larger. With the application of similar I/M programs, MOBILE6 factors do not decrease as much as the corresponding MOBILE5 factors.

For the 2035 scenario with no I/M program, the percentage changes are lower than the corresponding percentage changes for the base case with inspection and maintenance for all speed and temperature combinations. Also, the difference is slightly larger. The difference between the base MOBILE5 and MOBILE6 emission factors is lower for 2005 than the difference for MOBILE6.  

For the three- year-newer average fleet distribution for 2005, the emission factors are lower for both models, with MOBILE5 and MOBILE6 generally showing a 20% lower emission rate. The trends for the percentage changes between the two versions of the emission factor models are similar to those observed for the base case (with I/M). Notice that, in this case, the percentage change with increasing years for speeds between idle and 19.5 mph decreases more than for the base case, and the differences grow larger as the speed increases. Also, for the larger speed of 35.9 mph, the percentage change increases significantly, around 5-10%, with respect to the base case for all temperatures. This trend indicates that both MOBILE5 and MOBILE6 integrate the change in the average fleet distribution for light-duty vehicles and trucks in a similar manner. For 2035, the trend is similar to the trend observed for 2005, with the exception of the larger speed, 35.9 mph. For that speed the percent change for 2035 is negative, and the trend of increasing difference with increasing temperature continues up to 80 °F.

For the 2035 three- year-older average fleet distribution for light-duty vehicles and trucks, the emission factors for MOBILE5 and MOBILE6 are higher, typically by about 20%, than those of the base case (with I/M). The trends for the percentage changes between the two versions of the emission factor models are similar to that observed for the base case. For the higher speed, 35.9 mph, the percentage changes are higher than the base case; the others behave similarly to the base case. There is also a small increase in the percentage change with respect to the base case, and that difference increases with increasing speed. This shows that by 2035, the emission factors for MOBILE6 decrease more than the MOBILE5 emission factors do, indicating that MOBILE6 projects that significant improvement in emission reductions will continue after 2005, while MOBILE5 does not. For 2035, the trend of increasing difference with increasing temperature continues up to 80 °F, which is similar to what was seen with the three- year-newer fleet.

For the 30% decrease in the light-duty vehicles VMT fraction for 2005, the emission factors increase only a few percent in value with respect to the base case. Thus, differences in fleet composition appear to have minimal impact on emissions. The percentage changes between the two versions of the emission factor models are large and generally decrease as the speed increases. However, the 35.9 mph speed scenario is an exception, in which the percentage change decreases significantly with respect to the base case. The trend with respect to the temperature is relatively constant. For the 30% decrease in the light-duty vehicles VMT fraction for 2035, the trend for the percent changes is similar to that explained for 2005. The differences in the percent changes for this scenario with respect to the corresponding base case scenario are slightly larger than those observed for 2005; however, they do decrease with increasing average speed.

For the 30% increase in the light-duty vehicles VMT fraction for 2005, the emission factors increase only a few percent in value with respect to the base case. Again differences in fleet composition appear to have minimal impact on emissions. The percentage changes between the two versions of the emission factor models are large and generally decrease as the speed increases. The 35.9 mph speed scenario is, again, an exception, in which the percent change increases significantly with respect to the base case. For 2035, the trend is similar to the trend observed for 2005. For temperature, both 2005 and 2035 show decreasing emissions with increasing temperature, however MOBILE6.2 shows a more rapid decrease in emissions with increasing temperature than MOBILE5. This same trend is also found in the other scenarios.

In summary the major findings from this two model comparison are:

MOBILE6 Impact on the Validity of CAL3QHC

Two studies have performed extensive monitored-to-modeled comparison of the CAL3QHC model. The first study, "Evaluation of CO Intersection Modeling Techniques Using a New York City Database" (Sigma Research, 1992), is the older study, with traffic and air concentration data collected in 1989. The study also used the then current emission factor model, MOBILE4.1. This study formed the basis for EPA's selection of the CAL3QHC model as the preferred guideline model for project-level analysis. The second study is NCHRP25-6, "Intersection Air Quality Modeling," which is a mid-1990s study, which used MOBILE5 emission factor model for three intersections in Tucson, Arizona; Denver, Colorado; and Sterling, Virginia. A review of how the CAL3QHC performed in these two studies and how the use of MOBILE6 will impact CAL3QHC's validity is discussed below.

Application of the MOBILE4.1 Emission Factor Model in CAL3QHC Using the New York City Model Evaluation Database

The MOBILE4.1 model analysis was limited to the three least complex intersections with the best quality data. These three intersections were #1 - West and Chambers, #2 - 34th Street and 8th Avenue and, #5 - 12thStreet and 34th Avenue. For all three intersections, the observed and CAL3QHC predicted values paired in time and location showed underpredictions with an average difference of 2.2, 2.8 and 2.6 parts per million (ppm), respectively. Similar results were found for all three intersections events paired only in time. For the 25 highest observed and predicted CO concentrations paired only by each intersection, the CAL3QHC results showed systematic underpredictions ranging between 2.7 to 5.0 ppm.

The fractional bias is a measure of the model's ability to simulate observed concentrations during the highest observed periods. The fractional bias (FB) is defined as

Fractional bias equals 2 multiplied by [(observed value minus predicted value) divided by (observed value plus predicted vaule)].

where OB and PR refer to the averages of the observed and predicted highest 25 values matched by rank for each intersection. A positive value of the fractional bias means the model is underpredicting. CAL3QHC has a positive value for all three intersections, but is within a factor of two of the observed concentration. The ambient conditions most important for regulatory applications are those which lead to the highest concentrations. These were categorized in the New York City study as wind speed less than 6 mph and neutral or stable atmospheric conditions. Under these conditions, the CAL3QHC model was found to predict values to within 50% at all three intersections, but systematically low. In combination across all three intersections, CAL3QHC was found to be the best performing of the intersection models tested. Thus, the overall finding for the CAL3QHC model was a general underprediction bias, but at higher concentrations, results were generally within a factor of 2. Under meteorological conditions favorable for high CO concentrations, CAL3QHC was found to be within 50% of the observed concentration, but again biased low for all three intersections.

Application of the MOBILE5 Emission Factor Model in CAL3QHC Using the NCHRP 25-6 Database

Three high-volume, suburban intersections in Tucson, Virginia and Denver were intensively monitored - fourteen locations at the Tucson intersection and 20 locations at the Denver and Sterling, Virginia, intersections. Seven random days of data were selected for simulation and comparison from the twelve-week winter monitoring period. The periods selected for evaluation were all weekday periods where complete traffic volumes and meteorological data were available for full 24-hour periods. The Tucson intersection was monitored in early 1994, and the Virginia and Denver intersections were monitored during the 1994-1995 winter period.

For the Denver and Virginia intersections, the observed and CAL3QHC predicted values (using MOBILE5) paired in time and location showed overpredictions with an average difference of 3.0 and 1.6 ppm, respectively. The Tucson intersection showed an underprediction of 0.6 ppm. For the 25 highest observed and predicted CO concentrations paired only by each intersection, the CAL3QHC results showed overpredictions of 5.6 and 5.5 ppm for the Denver and Virginia intersections, respectively. The Tucson intersection showed the reverse outcome, with an average underprediction bias of 5.5 ppm. Similar biases were found with the fractional bias for the top 25 observed concentrations. However, only the Denver intersection was, on average, within a factor of two of the observed concentrations.

The three intersections were multilane intersections, and all had high total traffic volumes. Both the Denver and Virginia intersections were dominated by flow volumes in particular directions. The Tucson intersection was evenly balanced between the north-south and east-west direction. This balance in traffic flow and signal cycle timing resulted in the CAL3QHC model predicting minimal queue lengths and concentrations being dominated by the moving emissions. CAL3QHC queue lengths for Denver and Virginia were large because of the unbalanced flow volumes and signal cycle timing, with a resulting overpredicition in concentration dominated by the idle emissions from excessive queue lengths.

Based on the findings in Section 2.2.1.1, application of the MOBILE6 model will likely improve the CAL3QHC model performance for Tucson, as moving emissions will increase, leading to higher predicted concentrations and a better match to monitored values. For Denver and Virginia, the contribution from idle will decrease, reducing the overprediction bias, resulting in improved model performance.

MOBILE4.1 versus MOBILE6.2 Emission Factor Comparison for Future Applications of the MOBILE6 Model for CAL3QHC

Emission factors from the mobile emission factor models are used as input for the roadway intersection model, CAL3QHC. This model was selected as the preferred roadway intersection model by EPA based on the Route9a evaluation study, using the then current model, MOBILE4.1, as input to CAL3QHC, for three key intersections located in Manhattan (Sigma Research Corporation, 1992). It is therefore important to determine how changes between MOBILE4.1 and MOBILE6.2 may have changed as a result of improved understanding of emissions. Studies have shown that CAL3QHC model simulation results are usually driven by queue length and number of lanes (queue density) for overcapacity conditions. CAL3QHC uses an internal queuing algorithm to estimate queue length. Queue emissions result from emissions produced while idling. In most cases, the highest CO concentrations occurred during overcapacity situations. Thus, the question of principal interest is how have idle emissions changed between the two versions of the model.

Based on a review of the "Evaluation of MOBILE Vehicle Emission Model" conducted by Sierra Research in 1994 to help in determining the most significant changes between MOBILE4.1 and MOBILE5 and the more recent studies on MOBILE5 and MOBILE6, as well as the efforts conducted in Section 2.2.1.1,theinput variables found likely to have had the most significant changes in the CO emission factors between MOBILE4.1 and MOBILE6.2 models are the start fraction and temperature. Fuel volatility effects are only important for VOC emissions.

To assess these levels of changes, a base emission scenario was modeled using MOBILE4.1 and MOBILE6.2 based on the parameters used in the Route 9A studies which are summarized as:

Four different scenarios were evaluated in comparison to the base simulation using the two emission factor models. These scenarios were:

Discussion of Results

Tables 2.2.1.1, 2.2.1.2 and 2.2.1.3 present the results of the models sensitivity test for idle, 20 and 30 mph average speed conditions for the base case and four sensitivity scenarios.

For the base case, the composite CO emission factor calculated for idling by MOBILE6.2 is much lower, 56% lower, than the factor estimated by MOBILE4.1. For 20 and 30 mph average speeds, the emission factor estimated by MOBILE6.2 is higher than MOBILE4.1 by 36% and 103%, respectively. This represents an important shift in emission contribution at intersections. Emissions from idle will typically now contribute 56% less with MOBILE6.2, and moving emissions will typically increase by 36% or more.

For the percent start modification scenario, MOBILE6.2 was held constant using the MOBILE6.2 national default engine soak times as EPA guidance states. Nearly all emissions are hot-stabilized, and unless it is a special modeling situation (such as a parking lot) which may require modeling of the effects of engine starts, it is strongly suggested to use the emission factors from MOBILE6.2 without any special adjustment for starts since MOBILE6.2 already includes vehicle idling in proportion to normal driving. Only intersection #1 from the Route9a study was near a parking facility, but the parking lot was reported to be relatively small, so no adjustments were made to MOBILE6.2. On the other hand, MOBILE4.1 was applied following the guidance developed specifically for this model, which suggests the user should use the site-specific estimate of the cold start fraction as input. In the Route9a study, the lowest and highest surveyed cold start percentages were 3.4% and 26%, respectively. These were used to define the lower- and upper- range in the estimated emission rates when using MOBILE4.1. For the idle condition, the lower-range for MOBILE4.1 yields similar results to MOBILE6.2, while the upper-range doubles the emissions. For the 20 to 30 mph average speeds, the change in the emission factor for the lower-range is much lower than for the base case, but with a smaller difference for the upper-range. This suggests that applying MOBILE6.2 for intersections that have been characterized as having a low percentage of start emissions, as is now the current understanding, as input to MOBILE4.1 will result in little change in idle emissions, but nearly double the contribution from moving emissions.

For the temperature range scenarios, the 90 °F temperature has almost no effect on the MOBILE4.1 emission factor, while the MOBILE6.2 factor increased by 60%. The overall effect is to narrow the differences for the base case between the two versions of the emission factor model. For the moving emissions, the changes are much smaller, with MOBILE6.2 showing slightly larger emission factors. For the 10 °F ambient temperature scenario, the emission factors calculated by both models increased with respect to the base case. In the case of MOBILE6.2, the idling factor increased by 36%, while the MOBILE4.1 factor increased 55%, resulting in a difference of about a factor of 2 between the two models. For the 20 and 30 mph moving scenarios, both models increased emissions, resulting in the relative differences remaining about the same, with MOBILE6.2 having the higher emission factors. Overall, for the cold temperature conditions, MOBILE6.2 will typically reduce idle emission contribution by 50% and increase moving emissions by 37%. The results indicate that MOBILE6.2 is less sensitive to cold temperatures and more sensitive to warm temperatures than MOBILE4.1.

Overall, the idle emissions decreased significantly for MOBILE6.2, while moving emissions increased. Historically, analyses of roadway intersections have found that high concentrations are a result of large queue emissions and hence, the idle emission factor. The tradeoff in emissions seen here will likely impact the CAL3QHC model by lowering concentrations in most situations where queue length is important. Since the model performance evaluation of CAL3QHC in the Route 9a study had a positive fractional bias (underprediction) for all three intersections, the model bias will likely increase using MOBILE6.2. However, this somewhat contrasts with the NCHRP25-6 study results, which suggest that MOBILE6.2 will improve model performance relative to its evaluation based on using MOBILE5 . Some of this difference may be the result of changes in engine technology since these evaluation studies were based on pre-1990 and pre-1995 vehicles. It is possible that differences between the two MOBILE models may be considerably different for a newer fleet of vehicles. If, however, today's fleet is analogous to the NCHRP's pre-1995 fleet, then the use of the MOBILE6.2 model in project-level analysis will likely improve model performance.

Table 2.2.1.1.New York Route9A: MOBILE4.1 and MOBILE6.2 Idle Emission Factors (g/hr)

 

Base

Lower-range of start activity

Upper-range of start activity

Tmax = 90 F

Tmin = 10 F

MOBILE4.1

554.70

359.40

624.56

531.77

852.22

MOBILE6.2

310.68

310.68*

310.68*

494.75

424.45

MOBILE4.1 - MOBILE6.2

244.02

48.72

313.88

37.03

427.77

Table 2.2.1.2.New York Route9A:MOBILE4.1 and MOBILE6.2 20 mph Emission Factors (g/mi)

 

Base

Lower-range of start activity

Upper-range of start activity

Tmax = 90 F

Tmin = 10 F

MOBILE4.1

30.72

21.40

34.06

30.65

45.72

MOBILE6.2

41.88

41.88*

41.88*

47.89

62.48

MOBILE4.1 - MOBILE6.2

-11.16

-20.47

-7.82

-17.24

-16.76

Table 2.2.1.3.New York Route9A:MOBILE4.1 and MOBILE6.2 30 mph Emission Factors (g/mi)

 

Base

Lower-range of start activity

Upper-range of start activity

Tmax = 90 F

Tmin = 10 F

MOBILE4.1

19.20

13.53

21.22

19.34

28.52

MOBILE6.2

38.96

38.96*

38.96*

43.70

58.95

MOBILE4.1 - MOBILE6.2

-19.77

-25.43

-17.75

-24.36

-30.43

*   Emissions rate held constant based on MOBILE6.2 formulation, which assumes that nearly all emissions are hot-stabilized unless strongly influenced by a nearby start location.

Emission Changes for Characterizing Start Situations, Starts per Day, Start and Soak Distributions

With the release of MOBILE6, EPA recommended that, in most instances, the model's emission factor estimates should be used without adjustment for start fraction, since nearly all emissions are hot-stabilized, unless the location is near a parking garage or shopping center with a large number of starts. In this section, ICF investigated under what circumstances a user may want to account for start emissions and identify possible approaches for estimating start fraction for a variety of settings. In locations where start emissions may be important, the potential impact on project-level results may be significant.

The emission rate of a gasoline-fueled vehicle will be at its lowest rate when the engine and catalyst are at their full operational temperature. When the engine temperature and catalyst are not fully warmed up, inefficiencies in combustion and catalytic conversion result in higher emission rates. The elevated emissions, which are a combination of fuel enrichment and increased engine and transmission friction, are termed "cold start" emissions. Cold start emissions are of particular concern, as a large proportion of total mobile source emissions are due to vehicles being driven under cold start conditions (TRL, 2000). The length of time a vehicle has been parked influences the temperature of the engine and catalyst on restart and hence, the cold start emission rate. Vehicles restarted shortly after being stopped are characterized as "hot starts" and have much lower emission rates.

In the 1970s, the introduction of the catalytic converter into the vehicle fleet shifted the focus on high CO emission rates to the time period before the catalyst was fully functional. During this time and into the 1980s, bag 1 of the FTP   cycle, the first 505 seconds (s), was used to define the cold start emission profile (Midurski and Castaline, 1977). Since that time, improved capabilities to measure second-by-second emissions, as well as improved combustion technology using fuel injection and on-board diagnostics, have reduced engine warm-up time leading to a shorter cold start emission period. Most recent studies, by measuring second-by-second emission rates, have now characterized the cold start emission period as the first 200 s following a 4-hour or longer period since the engine was last started (Boulter, 1997; Laurikko, 1996; and Singer et al., 1998; Rakaha et al., 2003).

Start Emissions Characterization

The three parameters that define the average start emissions for a region are

For project-specific locations, the key parameter is the soak distribution, the length of time parked before the start, which defines the much higher cold start emission rate versus the much lower hot start emission rate. Both the start distribution over the day and the number of starts per day are important for region-wide estimates of total CO emissions, but are much less important to defining the emissions at a project-specific location. As a result, the two primary objectives for start characterizations were to develop estimates for soak distributions for a variety of potential high emission ("hot spot") locations and investigate potential methods for collecting site-specific data. A secondary objective was to gatherinformation on the number of starts per day and start distribution for a subregion, as it may be useful for regional analysis or providing localized background concentrations.

A list of potential high start emission locations was developed for this investigation. Seven locations were identified where a high percentage of start emissions could be potentially found at a nearby intersection. The seven locations are

Other locations were considered, such as a theme park or amusement center, but were eliminated because start departures were not generally clustered during a particular period. Each of these seven locations/settings has sufficient differences in their soak distribution to warrant separate discussion.

Commuter Parking Garage

Vehicle activity levels were monitored at a three-level underground parking garage located in an office building at a CBD location in March 1997 (Singer et al., 1999). The second and third levels are used as employee only parking and only their activity levels were investigated. Two three-weekday periods were studied with similar results found for the soak distribution. Figure 2.2.4 shows the soak distribution for the vehicles parked in the CBD garage. The figure shows that over 90% of the vehicles were in cold start mode upon departure. Driving in the garage averaged approximately 41 s for the second floor parking and an additional 39 s for the third floor parking, which required travel through the second floor and a ramp. This would allow most vehicles to reach a nearby intersection and still be in start mode.This soak distribution is likely transferable to other CBD parking locations; however, some adjustments may be needed for in-time garage travel distances, which might limit the number of vehicles reaching nearby intersections in start mode. Appendix B contains a listing of the soak distribution input file formatted for MOBILE6, "soakdst.d," for the CBD garage soak distribution as translated from Figure 2.2.4.

Figure 2.2.4. Distribution of Soak Times for Vehicles Parked in a CBD Garage

This chart depicts the fraction of vehicles for various soak periods from over 10 hours down to less than 1 hour for vehicles parked in a CBD garage 3 days in March. This chart is used to illustrate that the majority of vehicles in this garage had 8-9 or 9-10 hour soak periods.

Shopping Center

Estimates of vehicle activity at shopping centers were made from the national survey studies,Parking Generation, 2nd Edition publication conducted by the Institute of Transportation Engineers (ITE) (ITE, 1987)5 and the Urban Land Institute'sParking Requirements for Shopping Centers (ULI, 2000). The soak distributions can be estimated by first estimating the potential parking capacity. The number of vehicles parked at a peak hour is based on the type of shopping center, ranging from a neighborhood to a super regional center. The busiest hour of parking demand falls between 1:00 and 3:00 pm on a Saturday.The ITE Study provides a mathematical expression between the gross square footage (X) of the retail center and the number of parked cars (P) as:

The natural logarithm of the number of parked cars equals 1.261 multiplied by the natural logarithm of the gross square footage minus 0.365.

This equation provides an estimate of the parking demand. However, this relationship is based on survey data for the average Saturday. The ULI report identified the 20th highest hour of parking demand based on 1997 survey data and found results generally about 10% higher than given by the above expression. While the parking reports do not estimate parking duration, they do estimate that the peak parking demand occurs between 1:00-2:00 pm and that about 20% of all parking is conducted by employees (ULI, 2000). The best estimate of the duration comes from TRL Report No. 469 (Green and Boulter, 2000), in which they surveyed vehicle activity for parking duration at a supermarket and golf course. Figure 2.2.2 shows the parking duration for a supermarket and golf course only. The supermarket parking duration distribution would best represent the neighborhood center (30,000 to 100,000 sq. ft.) and community center (100,000 to 350,000 sq. ft.), while generally longer shopping durations, characteristic of the time for a round of golf, would be anticipated for the regional (400,000 to 800,000 sq. ft.) and super centers (>800,000 sq. ft.). However, to apply the supermarket soak distributions to a shopping center, each hour should be proportionality reduced to account for the 20% of parking accomplished by employees and the 9-10 hour parking period should be increased for the 20% of employee parking.  The total number of vehicles leaving the parking facility and the subsequent percentage of starts reaching nearby intersections can then be estimated based on the peak parking demand and the estimated soak distribution. Appendix B contains the soak distribution input file formatted for MOBILE6, "soakdst.d", for the "supermarket" profile adjusted for employee parking and the "golf course only" as translated from Figure 2.2.5.

Figure 2.2.5. Percentage of Vehicles Parking Duration Distribution

This chart depicts the percent of vehicles for various parking durations from less than 1 hour to 13-14 hours for a supermarket, golf course, golf course and practice area, and a railway station. This chart is used to illustrate that the peak of vehicles had parking durations less than 1-hour for the supermarket and golf course/practice areas, while the railway station had a peak of 11-12 hours.

Hospitals

Estimates of vehicle activity at hospitals were made from the national survey study,Parking Generation, 2nd Edition, conducted by the Institute of Transportation Engineers (ITE)(ITE, 1987). The soak distributions can be estimated by first estimating the potential parking capacity. The ITE study provides a mathematical expression between the number of beds (X) of the hospital and the number of parked cars (P) as:

The number of trips equals 0.62 multiplied by the number of parking stalls minus 32.

This equation provides an estimate of the parking demand. The peak parking demand coincided with the mid-morning and mid-afternoon hours associated with employee shift change overlap. However, no survey data was available to estimate the parking duration distribution, and no appropriate surrogate distribution appeared readily available in the literature. The 2004Parking Generation manual update may contain some additional information to help in estimating park duration distribution.

Universities

Estimates of vehicle activity at universities may be made from the national survey study,Parking Generation, 2nd Edition, conducted by the Institute of Transportation Engineers (ITE)(ITE, 1987). However, the 1987 survey only contains data from a single university and cannot be considered representative of most situations. Thus, at this time no reliable estimate can be made on parking duration distribution at universities. It is anticipated that the 2004 ITEParking Generation manual will contain much more data for various sizes and locations of universities and be able to define the peak parking periods.

Park-and-Ride Lots

Vehicle inbound and outbound activity levels were monitored at seven park-and-ride lots for light rail transit in two different years in the City of Calgary, Canada (population 708,000), serving primarily the central business district (Kok, et al., 1994). The soak distributions can be estimated using the PM peak hour volume and assuming that approximately 45% of the travel is CBD-oriented, home-based work trips. The trip generation equations were developed from all seven of the park-and-ride lots and can be expressed mathematically as an expression between the number of parking stalls (X) for the lot and the number of trips (T) as:

The number of trips equals 0.62 multiplied by the number of parking stalls minus 32.

Of these trips, on average, 83% were outbound. Therefore, for example, a 1,500 park-and-ride lot would be estimated to generate (0.62*1500-32)*0.83*0.45 = 335 start vehicles during the PM peak. It is anticipated that these results would be transferable to US cities and could be applied as long as the home-based work trips fraction is readily available.

Railway Station

Green and Boulter (2000) studied a commuter-oriented railway station parking lot from data collected during December 1998 and February 1999. They observed two peaks in parking duration. A primary peak centered on the 11-12 hour, and a secondary peak occurred at parking duration hour 6-7. The hour-by-hour railway station parking distribution is presented in Figure 2.2.5. This soak distribution is likely transferable to other railway station locations; however, additional information on the number of vehicles using the parking lot would be needed to apply to other locations.

General Residential Areas

In some areas, high CO concentrations have been found to be associated with the general startup of residential area emissions. A household travel behavior survey conducted in Anchorage, Alaska, based on travel logs of some 1,548 households found that the AM peak (6-9 AM) for the general residential area had 51% of the starts associated with parking periods of over 12 hours and 73% of all peak AM starts having an eight-hour or longer soak (Municipality of Anchorage, 1993). This high start distribution could be used as a conservative first estimate for areas where no travel log surveys have been conducted.

Starts per Day and Start Distribution

In addition to the soak distribution, some literature was found on starts per day and start distribution, which may be useful for sub-regional analyses. While not specific to project-level analysis, this information may be useful in preparing background concentration levels, particularly in regions where the background concentration is high and where only moderate project-level activity levels may lead to potential CO violations.

The primary information on starts per day and start distribution is from the vehicle-instrumented study sponsored by EPA for Baltimore, Maryland, and Spokane, Washington. An alternative approach developed by Everett and Sacs (2001) uses a more economical approach employing a simple electronic data logger. In this study, done in the mid-sized city of Knoxville, Tennessee, the data logger was connected through the cigarette lighter of a vehicle and allowed to record whether the engine was on or off on a second-by-second basis. Data was collected from some 377 vehicles from 200 households during weekdays over a three-month period.

Analysis of the collected data from the Knoxville study showed that Knoxville had about 1.5 fewer starts per day than Baltimore or Spokane experienced for both weekday and weekend days. It was surmised that this difference was due to differences in study area characteristics. For start distribution results, the weekdays were found to be similar to those found in Baltimore and Spokane, except that no 3 PM peak was found in Knoxville. It is believed that this peak is associated with the pickup and transport of children from school. Weekend start distributions in Knoxville were the same as Baltimore and Spokane.

Methods for Collecting Project-Specific Start and Soak Distribution

In addition to a review of the available literature, methods were reviewed on collecting data on soak distribution using approaches other than fully instrumented vehicles. The primary focus was directed to collecting data for project-specific soaks, with extended capabilities to determine start information.

Parking Studies

One approach applicable to parking facilities is to collect information on parking activities. For example, for a commuter-dominated parking garage, the number of vehicles entering and exiting the facility may be recorded during AM and PM peaks. Departure and arrival times for each vehicle may be matched via license to determine length of time vehicles were parked. Vehicle trip times and lengths while in the parking facility may be recorded via stopwatch during AM and PM peak periods. Measurements made using this approach over a three-day period found similar results for each day (Singer, et. al, 1998). This suggests that a one-day sample may be sufficient in other applications. The approach also provides details on the trip time through the garage and the fraction of vehicles exiting in start mode since individual licenses are matched between entering and exiting. This approach can be extended to other garages, but it is imperative to carefully measure time spent in the garage following engine start-up before exiting to the street.

Remote Sensing

Another suitable technique is remote sensing, which uses open-path reflection of infrared radiation off the road surface to see the functioning state of the catalyst (Stedman, 2002 and Rendahl et al, 2003). By measuring other vehicle parameters, including vehicle speed, acceleration, mass and road slope, and additional emissions of water vapor and carbon dioxide (CO2) the technique can distinguish between the three conditions leading to high CO emitting conditions:

If the catalyst is operating properly, the operating temperature will be high (this is inferred through measurement of the water vapor and CO2), and if acceleration or road slope is high, then the vehicle is in "hard acceleration". If the catalyst operating temperature is low and emissions are high, then the vehicle is in start or has a malfunctioning catalyst. By measuring the other pollutants, ammonia, acetylene, ethylene and the ratio of total hydrocarbon to methane, the catalyst's operating state can be determined.

The advantage of this remote sensing technique is that it would provide direct information on start fraction at the precise location of interest. The limitation with the method is that it has not been fully developed and tested and needs two to four infrared sensors to collection the emission measurements.

Travel Demand Models

In addition, travel demand models may be used to provide sub-regional (not site-specific) estimates of soak distribution (Allen and Davies, 1993). This may be useful in estimating concentrations in areas with high CO background concentrations. Within a travel demand model, assignments may be made based on the start duration for each type of travel (e.g., home to work) and travel analysis zone. The number of start trips may be summed separately in the travel model output, with the resulting final assignment giving the percentage of vehicles on a link in start mode. This approach would have the added benefit of reducing the linear growth in emissions with increased VMT if growth increased through increased travel distances.

Simple Instrumented Vehicles

While not specific to a project-level analysis, a relatively inexpensive instrumented vehicle approach using an electronic data logger, at an approximate cost of $100 each, can be used to collect regional or sub-regional level (traffic analysis zone) data on vehicle starts, start distribution and soak distribution. The inexpensive data logger simply measures whether the vehicle is on or off and time stamps those events. By analyzing the duration between starts and by designing a representative household survey sample, a local estimate of the soak distribution may be made (Everett et. al, 2001). Analysis of the data collected can also provide information on the number of starts per day and the start distribution. If sufficient number of vehicles are instrumented at a sub-regional level, e.g. at the travel analysis zone, then characterization of the local soak distribution can also be estimated.

Impact of CAL3QHC Results by Examination of Typical High-End Project-Level Applications

In this section, an assessment is made on how the implementation of MOBILE6 will affect the results of project-level analysis through the modeling of changes in CO concentration from a typical high volume freeway segment and a typical high volume intersection while migrating from the MOBILE5 to MOBILE6 emissions model. Analysis involved the application of the CAL3QHC and CALINE3 dispersion models for a variety of emission scenarios, representing the expected range of differences between MOBILE5 and MOBILE6 models. Each of the scenarios was run for a base year of 2005 and a future year of 2035. In addition, the impact of start emissions was made for an urban and a suburban intersection for several levels of service where a high number of start emissions are possible (e.g., parking garages and park-and-ride lots).

Changes in CO concentrations in Migrating from MOBILE5 to MOBILE6

Air quality modeling was performed using the emissions changes identified in section 2.2.1.1 as the most likely to affect the simulated ambient concentrations of CO at the project-level. This effort was performed for three settings (urban intersection, suburban intersection and freeway); two years (2005 and 2035); and six emission scenarios, with emissions factors determined from both the MOBILE5 and MOBILE6 emissions models; and for three levels of service. For each of these combinations, a pair of emission factors was selected that showed the most significant differences between these two emission factor models for a given temperature and speed.

Three scenarios for dispersion modeling were prepared to assess the CO concentration changes: one suburban intersection, one urban intersection, and one freeway segment. Each of the simulations was applied for the peak traffic period with worst case screening meteorology to determine the maximum one-hour concentration. In all cases, it was assumed that settling and deposition velocities were zero.

The freeway setting occurred on a six-lane freeway (one link of three lanes in each direction), each lane about 12 feet in width, with a 12-foot median. Each link had a total mixing width of 56 feet, a length of 10,000 feet, and was at grade level. The freeway was oriented in a north-south direction, with four receptors located at the north-south midpoint of the link and at distances of 50, 100, 200, and 500 feet from, and perpendicular to, the freeway edge. All receptors were placed at a height of 1.8 meters (breathing level height). Worst case meteorology was applied, a near parallel wind direction (south-southwesterly wind at 190°) at 1 meter per second (m/s), with D stability. The surface roughness length was 175 centimeters (cm) (suburban), and the mixing height sas 1 kilometer (km). All emissions source heights were set at ground level. Traffic was divided evenly across all lanes, with a peak flow of 1,620 vehicles per lane per hour. Modeling was performed with CALINE3, the same dispersion module as CAL3QHC, but without the queuing algorithm.

The CBD intersection simulations occurred at a four-legged, symmetric intersection. Each of the four links corresponds to a principal compass direction, with each having two approach lanes, two departure lanes and a single left-turn bay. The approach lanes and the turn bay are queued, with the flow and signalization different for each of the three levels of service (D, E, and F). The signal type was set as actuated and the arrival rate as average. All lanes were 12 feet wide and at grade level. The total mixing width of each approach to the intersection was 44 feet. Sixty receptors were located symmetrically around the intersection, 15 in each quadrant. Receptor numbers 1, 16, 31, and 46 were located at the corners of the intersection, with receptors 1-15 in the NE quadrant, 16-30 in the NW quadrant, 31-45 in the SW quadrant, and 46-60 in the SE quadrant of the intersection(see Figure 2.2.6). All receptors were located parallel to the edge of the links, 10 feet from the roadway edge, at a height of 1.8 meters and with 40 feet (2.5 car lengths) of spacing between them. The surface roughness was 321 cm (urban), and the mixing height was 1 km. Worst case screening meteorology was assumed, with wind speeds of 1 m/s; D stability; and wind direction varying between 0° and 350°, inclusive, incremented at 10°. Although the total traffic volume varied by level of service, in all cases, 15% of the vehicles turned left, 80% went straight through and 5% turned right. Traffic volume by approach, as well as the signal cycle and average red time length, is given in Table 2.2.1.4. The saturation traffic flow for the intersection was 1,800 vehicles per hour per lane. Clearance lost time was 3.0 s for left turn queues and 3.5 s for through and right turns. All queues had an average rate of progression. All modeling was performed with CAL3QHC.

Table 2.2.1.4. Traffic Operations for Urban and Suburban Intersections

Modeling
Scenario

Approach Volume (veh/hr)

Average Red Time Length (s)

Control Delay
Per Vehicle
(s/veh)

Conservative
Level of Service
(LOS)

Cycle Length(s)

Left

Right/Through

Suburban LOS D

1100

87.9

72.1

 56.4

D

100

Suburban LOS E

1300

122.8

97.2

 84.9

E

140

Suburban LOS F

1400

128.7

98.8

103.9

F

145

Urban LOS D

 800

97.8

76.7

 52.1

D

110

Urban LOS E

1000

117.5

87.0

 87.1

E

130

Urban LOS F

1100

136.0

98.5

114.6

F

150

The suburban intersection simulations were set up similar to the CBD intersection. The main difference between the two settings is that the suburban intersection has three approach lanes, three departure lanes, and one turn lane in each direction. The total mixing width of each approach is 56 feet. The placement of the receptors relative to the roadway edge was the same as for the CBD intersection. The surface roughness was 175 cm (suburban), and the mixing height was 1 km. Traffic volume and signalization varied by level of service, but the same fractions for turning left, going straight and turning right were used as for the suburban intersection as were used for the CBD intersection. The saturation traffic flow was 1,800 vehicles per hour per lane. Clearance lost time was 3.0 s for left turn queues and 3.5 s for through and right turns. All queues had an average rate of progression. The same worst case screening meteorology and concentration averaging time were assumed in this scenario as were used for the urban intersection.

Figure 2.2.6. General Intersections and Receptor Layout

This figure depicts the general layout and receptor locations for the intersection. Each leg contains four 12-foot lanes on the approach; one exclusive left-turn lane, two through lanes, and one through/right-turn lane. Each quadrant formed by the intersection contains fifteen receptors located 10 feet from the road edge.

Simulated Emissions

Six scenarios for emissions calculations were identified to create "incremental" emissions factors that were used in the air quality modeling for the MOBILE5 to MOBILE6 change comparison. These were the same scenarios as described in Section 2.2.1.1.2:

Each of the scenarios above was used as input to create "incremental" emission factors for both the base (2005) and future (2035) years (note that in scenarios 5 and 6, above, the fraction is relative to the appropriate year), using both MOBILE5 and MOBILE6. For each of these 24 combinations, two pairs of idle and mobile emissions factors were chosen that represented the largest differences (i.e., the "incremental" emissions) between the emission models as a function of temperature and speed combinations 6(for the freeway modeling, only moving emissions factors were included). Each of these combinations was then run through each of the three model settings. For both the urban and suburban intersection settings, the pairs of emission factor values used were:

In addition to the tests performed above, another suite of CAL3QHC model runs were made to test the effects of cold start emissions on ambient concentrations of CO in migrating to the MOBILE6 emissions model. These additional runs were done in order to evaluate the potential for exceedances of the eight-hour CO standard for situations with heavy cold start emissions. The same 3 x 3 suburban and 2 x 2 urban intersection settings were used for these simulations as for the comparisons described above, but only LOS D was explored for each of the intersections. Real-world examples of these types of intersections include locations near an urban parking lot or a suburban park and ride-discharging traffic during the afternoon traffic volume peak on a cold winter afternoon. The emissions factors used in these simulations were created with the MOBILE6 model, producing cold start only and hot-stabilized idle only emission factors8. Three emission factor values were incorporated into the dispersion model for two temperature scenarios: cold start idle emissions, hot-stabilized idle emissions and running emissions. Idle emissions input to the model were determined from the cold start and hot-stabilized idle values by assuming that one-fourth of the vehicles were in cold start mode, i.e., within the first 200 s of ignition, while three-fourths of the vehicles were assumed hot-stabilized. This ratio was kept constant in all simulations. For these simulations, the suburban intersection was modeled at a temperature of 10 °F and the urban intersection at a temperature of 20 °F.

Air Quality Modeling Results

For thefreeway modeling, the emissions factors for the various scenarios described above were applied in the CALINE3 dispersion model. The outputs for each of the six scenarios, two temperature pairs, and two years were compared for the two emissions models at each of the four receptors. The concentration changes, along with the emissions change for each of the scenarios is given in Table 2.2.1.5. Note that negative percent changes indicate that MOBILE5 values are higher than MOBILE6.

Table 2.2.1.5. Percent Change in 1-Hour CO Concentration from the Freeway Modeling

Scenario

Year

Temp (F)

Change in Moving EF (M6-M5) (g/mi)

Average M6-M5 Concentration Change Across All Receptors (%)

M6-M5 Peak CO Change at 50 ft (ppm)

M6-M5 Peak CO Change at 100 ft (ppm)

M6-M5 Peak CO Change at 200 ft (ppm)

M6-M5 Peak CO Change at 500 ft (ppm)

1

2005

0

20.60

124.1%

4.70

3.29

2.13

0.99

1

2005

70

5.98

79.5%

1.37

0.96

0.62

0.29

1

2035

10

3.20

24.1%

0.73

0.51

0.33

0.15

1

2035

80

-0.66

-9.8%

-0.15

-0.11

-0.07

-0.03

2

2005

0

21.90

121.0%

5.00

3.50

2.27

1.05

2

2005

70

6.56

79.6%

1.50

1.05

0.68

0.31

2

2035

10

4.60

31.7%

1.05

0.74

0.48

0.22

2

2035

80

0.19

2.5%

0.04

0.03

0.02

0.01

3

2005

0

18.50

141.3%

4.23

2.96

1.91

0.89

3

2005

70

5.31

90.1%

1.21

0.85

0.55

0.25

3

2035

10

3.50

32.7%

0.80

0.56

0.36

0.17

3

2035

80

-0.25

-4.7%

-0.06

-0.04

-0.03

-0.01

4

2005

0

23.90

120.1%

5.46

3.82

2.47

1.14

4

2005

70

6.95

76.8%

1.59

1.11

0.72

0.33

4

2035

10

2.90

18.7%

0.66

0.46

0.30

0.14

4

2035

80

-1.08

-13.6%

-0.25

-0.17

-0.11

-0.05

5

2005

0

19.60

111.4%

4.48

3.13

2.03

0.94

5

2005

70

5.64

70.0%

1.29

0.90

0.58

0.27

5

2035

10

2.30

16.4%

0.53

0.37

0.24

0.11

5

2035

80

-0.95

-13.3%

-0.22

-0.15

-0.10

-0.05

6

2005

0

21.60

138.5%

4.93

3.45

2.24

1.03

6

2005

70

6.42

91.9%

1.47

1.03

0.66

0.31

6

2035

10

4.10

32.8%

0.94

0.66

0.42

0.20

6

2035

80

-0.34

-5.4%

-0.08

-0.05

-0.04

-0.02

The urban (CBD) intersection was simulated using the CAL3QHC model, as described above. The results of the simulations are shown in Table 2.2.1.6, which gives the changes in emissions factors between the MOBILE5 and MOBILE6 models, as well as the corresponding changes in peak ambient CO concentrations for each of the scenarios. Unlike the freeway modeling, the results are not presented at each of the receptors since the intersection is symmetric and there are too many receptors to show in one table. Instead, only the peak concentration from the full array of receptors is presented. It should be noted that the location of the peak was not always the same between the two emissions models, although, in cases where the locations were different, the concentration differences were typically small and/or occurred symmetrically about the intersection.

Table 2.2.1.6. Ambient 1-Hour CO Concentration and Emissions Changes
for the Urban Intersection Modeling

Scenario

Year

LOS

Pair

Temp (F)

Change in Idling EF (M6-M5)/M5

Change in Moving EF (M6-M5)/M5

% Change in Ambient CO Concentration (M6-M5)/M5

Change in Ambient CO Concentration (M6-M5) (ppm)

1

2005

D

1

10

-56.7%

46.5%

-42.7%

-4.7

1

2005

E

1

10

-56.7%

46.5%

-39.7%

-4.6

1

2005

F

1

10

-56.7%

46.5%

-38.7%

-4.6

1

2035

D

1

10

-79.3%

-21.3%

-69.3%

-7.0

1

2035

E

1

10

-79.3%

-21.3%

-68.8%

-7.5

1

2035

F

1

10

-79.3%

-21.3%

-66.7%

-7.4

1

2005

D

2

70

-56.5%

-37.6%

-50.8%

-3.2

1

2005

E

2

70

-56.5%

-37.6%

-49.3%

-3.4

1

2005

F

2

70

-56.5%

-37.6%

-50.0%

-3.6

1

2035

D

2

70

-79.7%

-70.2%

-75.4%

-4.6

1

2035

E

2

70

-79.7%

-70.2%

-75.4%

-4.9

1

2035

F

2

70

-79.7%

-70.2%

-76.1%

-5.1

2

2005

D

1

10

-55.4%

44.2%

-42.5%

-5.1

2

2005

E

1

10

-55.4%

44.2%

-39.1%

-5.0

2

2005

F

1

10

-55.4%

44.2%

-37.4%

-4.9

2

2035

D

1

10

-76.1%

-17.3%

-67.9%

-7.6

2

2035

E

1

10

-76.1%

-17.3%

-65.3%

-7.7

2

2035

F

1

10

-76.1%

-17.3%

-64.5%

-7.8

2

2005

D

2

70

-55.2%

-37.4%

-50.7%

-3.5

2

2005

E

2

70

-55.2%

-37.4%

-50.0%

-3.8

2

2005

F

2

70

-55.2%

-37.4%

-48.7%

-3.8

2

2035

D

2

70

-75.9%

-66.3%

-71.2%

-4.7

2

2035

E

2

70

-75.9%

-66.3%

-72.6%

-5.3

2

2035

F

2

70

-75.9%

-66.3%

-72.0%

-5.4

3

2005

D

1

10

-55.6%

57.0%

-38.4%

-3.3

3

2005

E

1

10

-55.6%

57.0%

-35.9%

-3.3

3

2005

F

1

10

-55.6%

57.0%

-33.7%

-3.2

3

2035

D

1

10

-78.4%

-14.9%

-69.5%

-5.7

3

2035

E

1

10

-78.4%

-14.9%

-64.0%

-5.5

3

2035

F

1

10

-78.4%

-14.9%

-64.4%

-5.6

3

2005

D

2

70

-55.1%

-34.3%

-52.0%

-2.6

3

2005

E

2

70

-55.1%

-34.3%

-47.2%

-2.5

3

2005

F

2

70

-55.1%

-34.3%

-43.6%

-2.4

3

2035

D

2

70

-78.6%

-68.1%

-74.5%

-3.5

3

2035

E

2

70

-78.6%

-68.1%

-74.5%

-3.8

3

2035

F

2

70

-78.6%

-68.1%

-75.5%

-4.0

4

2005

D

1

10

-56.1%

45.4%

-41.5%

-5.4

4

2005

E

1

10

-56.1%

45.4%

-40.4%

-5.7

4

2005

F

1

10

-56.1%

45.4%

-38.5%

-5.5

4

2035

D

1

10

-80.1%

-24.7%

-70.6%

-8.4

4

2035

E

1

10

-80.1%

-24.7%

-70.1%

-8.9

4

2035

F

1

10

-80.1%

-24.7%

-69.2%

-9.0

4

2005

D

2

70

-56.1%

-37.3%

-50.7%

-3.8

4

2005

E

2

70

-56.1%

-37.3%

-51.2%

-4.2

4

2005

F

2

70

-56.1%

-37.3%

-50.6%

-4.3

4

2035

D

2

70

-80.6%

-71.5%

-76.8%

-5.3

4

2035

E

2

70

-80.6%

-71.5%

-76.6%

-5.9

4

2035

F

2

70

-80.6%

-71.5%

-76.3%

-6.1

5

2005

D

1

10

-56.7%

40.7%

-42.9%

-4.8

5

2005

E

1

10

-56.7%

40.7%

-39.8%

-4.7

5

2005

F

1

10

-56.7%

40.7%

-38.8%

-4.7

5

2035

D

1

10

-79.4%

-24.9%

-71.2%

-7.4

5

2035

E

1

10

-79.4%

-24.9%

-69.1%

-7.6

5

2035

F

1

10

-79.4%

-24.9%

-67.9%

-7.6

5

2005

D

2

70

-56.3%

-37.9%

-52.3%

-3.4

5

2005

E

2

70

-56.3%

-37.9%

-48.6%

-3.4

5

2005

F

2

70

-56.3%

-37.9%

-49.3%

-3.6

5

2035

D

2

70

-79.5%

-70.3%

-75.4%

-4.6

5

2035

E

2

70

-79.5%

-70.3%

-75.4%

-4.9

5

2035

F

2

70

-79.5%

-70.3%

-76.5%

-5.2

6

2005

D

1

10

-56.7%

52.6%

-43.5%

-4.7

6

2005

E

1

10

-56.7%

52.6%

-38.6%

-4.4

6

2005

F

1

10

-56.7%

52.6%

-37.1%

-4.3

6

2035

D

1

10

-79.1%

-17.1%

-68.7%

-6.8

6

2035

E

1

10

-79.1%

-17.1%

-68.2%

-7.3

6

2035

F

1

10

-79.1%

-17.1%

-66.1%

-7.2

6

2005

D

2

70

-56.6%

-37.3%

-50.0%

-3.1

6

2005

E

2

70

-56.6%

-37.3%

-47.8%

-3.2

6

2005

F

2

70

-56.6%

-37.3%

-48.6%

-3.4

6

2035

D

2

70

-79.8%

-70.0%

-76.7%

-4.6

6

2035

E

2

70

-79.8%

-70.0%

-77.8%

-4.9

6

2035

F

2

70

-79.8%

-70.0%

-76.1%

-5.1

The suburban intersection was also simulated with the CAL3QHC model for the settings described above. The results of the simulations are shown in Table 2.2.1.7, which gives the relative changes in emissions factors between the MOBILE5 and MOBILE6 models, as well as the corresponding changes in peak ambient CO concentrations for each of the scenarios. The same caveats regarding location for the urban intersection also apply here.

Table 2.2.1.7. Ambient 1-Hour CO Concentration and Emissions Changes
for the Suburban Intersection Modeling

Scenario

Year

LOS

Pair

Temp (F)

Change in Idling EF (M6-M5)/M5

Change in Moving EF (M6-M5)/M5

% Change in Ambient CO Concentration (M6-M5)/M5

Change in Ambient CO Concentration (M6-M5) (ppm)

1

2005

D

1

10

-56.7%

46.5%

-41.1%

-6.0

1

2005

E

1

10

-56.7%

46.5%

-40.1%

-6.3

1

2005

F

1

10

-56.7%

46.5%

-40.7%

-6.6

1

2035

D

1

10

-79.3%

-21.3%

-70.4%

-9.5

1

2035

E

1

10

-79.3%

-21.3%

-70.7%

-10.4

1

2035

F

1

10

-79.3%

-21.3%

-69.3%

-10.4

1

2005

D

2

70

-56.5%

-37.6%

-51.2%

-4.3

1

2005

E

2

70

-56.5%

-37.6%

-50.5%

-4.7

1

2005

F

2

70

-56.5%

-37.6%

-51.1%

-4.8

1

2035

D

2

70

-79.7%

-70.2%

-75.9%

-6.0

1

2035

E

2

70

-79.7%

-70.2%

-76.5%

-6.5

1

2035

F

2

70

-79.7%

-70.2%

-76.4%

-6.8

2

2005

D

1

10

-55.4%

44.2%

-40.3%

-6.4

2

2005

E

1

10

-55.4%

44.2%

-39.5%

-6.8

2

2005

F

1

10

-55.4%

44.2%

-40.1%

-7.1

2

2035

D

1

10

-76.1%

-17.3%

-67.1%

-10.0

2

2035

E

1

10

-76.1%

-17.3%

-66.5%

-10.7

2

2035

F

1

10

-76.1%

-17.3%

-65.9%

-10.8

2

2005

D

2

70

-55.2%

-37.4%

-50.5%

-4.7

2

2005

E

2

70

-55.2%

-37.4%

-50.0%

-5.0

2

2005

F

2

70

-55.2%

-37.4%

-49.0%

-5.0

2

2035

D

2

70

-75.9%

-66.3%

-72.4%

-6.3

2

2035

E

2

70

-75.9%

-66.3%

-71.6%

-6.8

2

2035

F

2

70

-75.9%

-66.3%

-71.4%

-7.0

3

2005

D

1

10

-55.6%

57.0%

-41.4%

-4.8

3

2005

E

1

10

-55.6%

57.0%

-38.4%

-4.8

3

2005

F

1

10

-55.6%

57.0%

-37.2%

-4.8

3

2035

D

1

10

-78.4%

-14.9%

-68.8%

-7.5

3

2035

E

1

10

-78.4%

-14.9%

-69.0%

-8.0

3

2035

F

1

10

-78.4%

-14.9%

-68.6%

-8.3

3

2005

D

2

70

-55.1%

-34.3%

-48.5%

-3.2

3

2005

E

2

70

-55.1%

-34.3%

-51.4%

-3.7

3

2005

F

2

70

-55.1%

-34.3%

-50.7%

-3.8

3

2035

D

2

70

-78.6%

-68.1%

-73.8%

-4.5

3

2035

E

2

70

-78.6%

-68.1%

-75.0%

-5.1

3

2035

F

2

70

-78.6%

-68.1%

-75.0%

-5.4

4

2005

D

1

10

-56.1%

45.4%

-41.7%

-7.3

4

2005

E

1

10

-56.1%

45.4%

-40.4%

-7.6

4

2005

F

1

10

-56.1%

45.4%

-38.5%

-7.4

4

2035

D

1

10

-80.1%

-24.7%

-72.3%

-11.5

4

2035

E

1

10

-80.1%

-24.7%

-69.6%

-11.9

4

2035

F

1

10

-80.1%

-24.7%

-69.9%

-12.3

4

2005

D

2

70

-56.1%

-37.3%

-51.0%

-5.1

4

2005

E

2

70

-56.1%

-37.3%

-49.1%

-5.3

4

2005

F

2

70

-56.1%

-37.3%

-47.8%

-5.4

4

2035

D

2

70

-80.6%

-71.5%

-76.6%

-7.2

4

2035

E

2

70

-80.6%

-71.5%

-77.0%

-7.7

4

2035

F

2

70

-80.6%

-71.5%

-76.9%

-8.0

5

2005

D

1

10

-56.7%

40.7%

-41.6%

-6.2

5

2005

E

1

10

-56.7%

40.7%

-41.0%

-6.6

5

2005

F

1

10

-56.7%

40.7%

-41.2%

-6.8

5

2035

D

1

10

-79.4%

-24.9%

-71.7%

-9.9

5

2035

E

1

10

-79.4%

-24.9%

-71.3%

-10.7

5

2035

F

1

10

-79.4%

-24.9%

-69.9%

-10.7

5

2005

D

2

70

-56.3%

-37.9%

-49.4%

-4.2

5

2005

E

2

70

-56.3%

-37.9%

-51.6%

-4.9

5

2005

F

2

70

-56.3%

-37.9%

-51.0%

-4.9

5

2035

D

2

70

-79.5%

-70.3%

-75.9%

-6.0

5

2035

E

2

70

-79.5%

-70.3%

-77.0%

-6.7

5

2035

F

2

70

-79.5%

-70.3%

-75.3%

-6.7

6

2005

D

1

10

-56.7%

52.6%

-41.3%

-5.9

6

2005

E

1

10

-56.7%

52.6%

-38.6%

-5.9

6

2005

F

1

10

-56.7%

52.6%

-40.3%

-6.4

6

2035

D

1

10

-79.1%

-17.1%

-70.1%

-9.4

6

2035

E

1

10

-79.1%

-17.1%

-70.1%

-10.1

6

2035

F

1

10

-79.1%

-17.1%

-68.7%

-10.1

6

2005

D

2

70

-56.6%

-37.3%

-50.6%

-4.1

6

2005

E

2

70

-56.6%

-37.3%

-51.6%

-4.7

6

2005

F

2

70

-56.6%

-37.3%

-50.5%

-4.7

6

2035

D

2

70

-79.8%

-70.0%

-75.9%

-6.0

6

2035

E

2

70

-79.8%

-70.0%

-76.5%

-6.5

6

2035

F

2

70

-79.8%

-70.0%

-77.3%

-6.8

For the start emission scenario, both the suburban and urban intersections were simulated with the CAL3QHC dispersion model for the 2005 base and 2035 future years using LOS D signalization and traffic flow values. Worst case meteorology was included in the simulations, as described above, but the temperatures were taken as 10 °F for the suburban and 20 °F for the urban intersection, conditions that are conducive to high CO concentrations. Specific intersection parameters are given in Table 2.2.1.8, and results for each of the combinations of year and intersection scenario are shown in Table 2.2.1.9. In Table 2.2.1.9, the hourly peak CO concentrations have been reduced to an eight-hour average concentration by use of a 0.7 persistence factor for comparison to the 9 ppm eight-hour air quality standard. Background concentration was assumed to be zero.

Table 2.2.1.8. Intersection Parameters Used for Start Modeling

Year

Setting

Temp (F)

Idle EF (g/hr)

Mobile EF (g/mi)

total flow (vh/hr)

Left Turn V/C

Left Turn Queue Length

Through V/C

Through Queue Length

2005

Urban

20

728.98

28.5

800

1.05

8.4

0.76

7.3

2035

Urban

20

393.38

14.2

800

1.05

8.4

0.76

7.3

2005

Suburb

10

832.5

32.4

1100

1.31

26.3

0.77

6.2

2035

Suburb

10

450.98

16.1

1100

1.31

26.3

0.77

6.2

Table 2.2.1.9. Peak Ambient CO Concentrations from Start Modeling

Year

Setting

Worst Case Peak Hourly Conc (ppm)

8-hr value (ppm)

Increment above/below 8-hour standard (ppm)

2005

Urban

16.8

11.8

2.8

2035

Urban

9

6.3

-2.7

2005

Suburb

26

18.2

9.2

2035

Suburb

13.9

9.7

0.7

As can be seen from Table 2.2.1.5, the total moving emissions at 48.3 mph, as used in the freeway modeling, in all scenarios ranges from about -14% to about 140%. In the majority of cases, MOBILE6 has larger emissions factor values than MOBILE5. For 2005 only, MOBILE6 was greater than MOBILE5 in all cases, with a minimum difference of about 70% and an average difference of about 103%. The 2035 MOBILE6 scenarios show an average increase of only about 10%. Other than some small rounding off, the relative concentration change at the receptors is equivalent to the relative emissions changes, as expected. Also, at increasing distance from the freeway, the ambient concentration differences diminish. At 50 feet from the freeway edge, the ambient concentration changes ranged from about 1.2 to about 5.5 ppm for the various 2005 scenarios. For the 2035 scenarios, the ambient concentration changes at 50 feet from the freeway edge ranged from about -0.25 to about 1.1 ppm. For both years, the high temperature ambient concentration change was significantly less than the low temperature change. For the two base cases (Scenario 1), the low temperature values both show increases in migrating from MOBILE5 to MOBILE6, as does the 2005 high temperature value, but the 2035 high temperature value shows a slight decrease. Generally, low temperature base year simulations show a large increase in peak concentration in migrating from MOBILE5 to MOBILE6, with changes of about 4 ppm or more at 50 feet and 0 °F, followed by 2005 simulations at higher temperatures with changes of about 1 ppm at 50 feet and 70 °F.9Low temperature future year simulations show smaller concentration increases, with values ranging between 0.5 and 1 ppm at 10 °F, while high temperature future year simulations typically showed small concentration decreases. In all cases, MOBILE6 produced higher concentrations than MOBILE5 for 2005, while in 2035, MOBILE6 was more comparable to MOBILE5, but produced higher concentrations for lower temperatures. Thus, application of MOBILE6 for freeways in the near future years coupled with high traffic volumes and high background concentrations, could demonstrate potential problems in meeting the CO standard.

As shown by Table 2.2.1.6 for the suburban intersection scenarios, MOBILE6 produced lower ambient CO concentration values for every combination than did MOBILE5. The difference in the worst case ambient concentrations over all the scenarios ranged from about -0.3 to about -12 ppm (full range of about 40% to about 80 % reductions). For the base scenario (Scenario 1), the differences were fairly central to the range as a whole, with about 4-6 ppm for the base year, 2005, and about 6-10 ppm for the future year, 2035. These differences in concentration are more correlated with the change in idling emissions than with the change in moving emissions across the various scenarios and levels of service. While idling emissions are always lower in the MOBILE6 model than in MOBILE5, the moving emissions alternate having larger and smaller values across the scenarios. Note that for the suburban intersection, the volume to capacity ratio ranged from about 1.3 to about 1.5 for the left turn lane and about 0.77 to about 0.79 for the right turn-through queue.

For the CBD intersection, too, the concentrations produced by the MOBILE6 model were always lower than those from MOBILE5, ranging from 34% to about 78%. The same general trends observed for the suburban intersection also hold for the CBD intersection. The volume-to-capacity ratio for the turn lanes range from about 1.0 to about 1.5 and for the right turn-through queues, from about 0.76 to about 0.86 for the urban intersection. The overall reductions in ambient concentration are somewhat less than for the suburban intersections.

For the start scenarios, both the urban and suburban cases showed reductions of about 47% in 2035 for the peak CO concentrations over the base year, 2005. For 2035, the 1-hour peak values are about 9 and 14 ppm for the urban and suburban intersections, respectively. For 2005, the one-hour peak values are about 17 and 26 ppm. In all cases, the colder, suburban intersection showed higher concentrations than its urban counterpart. For comparison to the eight-hour CO standard, these values were adjusted to an eight-hour concentration value using the persistence factor of 0.7. Of the simulated intersections, only the urban intersection in the 2035 future year was not in exceedance of the eight-hour standard. In all cases, these exceedances are associated with the high idle emissions factors associated with the high number of starts. Thus, intersections with high start fractions appear to have the potential for exceeding the CO standard, given high traffic volumes and low temperatures.

2.2.2 Changes in MOBILE6 Impacting the Process for Project-Level Analysis

Use of MOBILE6 has the potential to affect the process in which project-level analysis is performed. The potential process-impacting effects may be organized into three subject areas:

These three areas are explored primarily through interviews conducted during the study. A total of 24 individuals affiliated with state DOTs, MPOs and researchers/consultants who have experience working with MOBILE6 on project-level analyses were interviewed. These interviewees represented a total of 14 groups conducting project-level analysis, nine state DOTs/MPOs, two groups responsible for facilitating project-level process and three researchers/consultants. The nine state DOTs/MPOs interviewed were: New York, Illinois, Alaska, Washington, Utah, New Mexico, Montana, Colorado and Florida. Results from these interviews are summarized in the attached document. Interview questions and a summary of the interview results are provided in Appendix C and D, respectively.

Need for Additional Information and Additional Agencies

The need to gather added information and involve additional agencies to conduct project-level analysis as a result of using MOBILE6 varies from state to state. Some states have found no additional information or agency involvement is necessary (generally these are the states using the defaults provided by the model), while other states have found there are additional needs and in several cases, some significant additional effort (e.g., one agency quoted that as much as 100% more information is required). There is approximately an equal split between agencies indicating that MOBILE6 required more data and additional agency contact and those that said that little or no additional resources were required. For those agencies requiring additional contact and coordination, most were local air pollution control districts or MPOs and in one case, a state energy agency. These agencies cited the need for additional agencies because they were making use of a number of the MOBILE6 options for which they had previously relied upon national defaults.

Agencies also commented that additional information was needed on model sensitivity to changes in inputs relative to MOBILE5. Several state agencies requested that a statistical analysis of model sensitivity be performed for vehicle mix, vehicle distribution and vehicle class.

Almost all state agencies contacted indicated that more time was required to complete project-level analysis using MOBILE6 compared to using MOBILE5, ranging from several hours to 40 hours. It was generally agreed that this would decrease over time and with practice.

Affect on Local or State Procedures Including Background

Most state agencies reported that use of MOBILE6 has not had a direct change on their procedures for project-level analysis. However, a number of consultants and researchers indicated that changes were seen in the data collection process, which included gathering additional facility-specific data and assessing the need for start specifications. For example, previously in New York City, specific percentages were used for particular neighborhoods. The new procedures drafted for New York City will use the same emission factors forall New York City neighborhoods.

Most states have also found that future background concentrations of CO should be lowered as MOBILE6's downward CO emission trends are, for most locations, larger than the regional VMT growth. As a result, a number of locations have, or are looking at, adopting new procedures for determining future background CO levels. For example, New York state is switching from the relatively common approach of calculating background CO concentrations using a three-year average of local CO monitoring data to a roll forward technique.

Several researchers and consultants reported that, based on their experience working with MOBILE6.2, areas will need to change their "worst case" modeling receptors from intersection-based to a mid-block location. This is caused by MOBILE6's higher speed emission factors, coupled with much lower emission factors for near idle conditions, which leads to shifting the maximum CO concentration away from the intersection corners where idle emissions are most dominate.

Other comments noted by state agencies and researchers that affect the applicability of MOBILE6.2 in project-level analysis is the limitation of the tool for the modeling of freeway ramps, as the user cannot change the model's average speed from the national average default freeway ramp speed of 34.6 mph.

Impact on Mitigation Strategies

States have reported that use of MOBILE6 appears to have had an impact on CO mitigation strategies. In some cases, the use of the model in place of MOBILE5 has eliminated the need for a mitigation strategy, as the intersections no longer appeared to have problems. However, the traditional approach of increasing intersection capacity to achieve higher average speeds, given today's vehicle emission control technology, will no longer reduce overall CO emissions, but may actually increase emissions.

More studies are needed to better quantify possible benefits derived from adopting certain CO mitigation strategies, (e.g., what are optimal emissions mitigation strategies and scenarios for a given location?) Historically, one key strategy in most CO mitigation plans was to reduce idle emissions. Studies that can identify which mitigation strategy would work best to reduce CO emissions given a certain set of local conditions (e.g., high altitude, high volume, narrow streets, etc.) are highly desirable. For example, one researcher suggested that optimizing signal timing will have to be restudied to better understand the possible CO mitigation benefits.

2.2.3 Changes in MOBILE6 Impacting Screening Assessment Procedures

Use of MOBILE6 has the likely potential to affect the screening assessment procedures for project-level analysis. The potential processes and resulting needs that may affect these procedures may be organized into three subject areas:

These three areas are explored first through interviews conducted during this study, followed by an investigation into air quality screening procedures based on MOBILE6 and CAL3QHC simulations.

Identification of Efforts to Date Suggesting the Need for Revised Screening-Level Procedures

All of the state agencies interviewed base their screening assessment procedures on the transportation conformity rule, which requires a project-level analysis for federally funded projects in CO nonattainment and maintenance areas. This approach is extended to all projects, including state- or privately funded projects, as well as projects needing an assessment under the National Environmental Policy Act (NEPA). The conformity rule requires that a quantitative analysis, (e.g., using CAL3QHC) is required for projects: 1) in or affecting locations identified in the state implementation plan (SIP) as sites of potential or actual violations of the CO National Ambient Air Quality Standards (NAAQS); 2) affecting intersections that are at or will be at LOS10D or worse; or 3) affecting intersections identified in the SIP as having the three highest volumes or three worst levels of service in the nonattainment or maintenance area. Five of the nine state agencies use a modification of the LOS C screening approach (that is LOS C passes screening) which consists of LOS and traffic volume thresholds. Four of the nine state agencies use LOS C only, but several are looking at revisiting this procedure in light of MOBILE6. Three of the nine agencies are updating their screening procedures because of MOBILE6 changes. A more extensive review of these procedures is presented by Houk and Claggett in an FHWA paper,Survey of Screening Procedures for Project-Level Conformity Analyses.

In Illinois, new pre-screening analyses have been developed for use with MOBILE6. The new procedure features "cut-off" criteria that are built into the analysis, which are based on the worst case inputs and the distance to a receptor.

New York City has modified its screening method procedure because of MOBILE6. The selection of modeling receptors has changed from an intersection location to a mid-block location.

Researchers and consultants recommend that state agencies revisit the current screening procedures, as MOBILE6 coupled with CAL3QHC does not yield the same results. Some of the key differences that may affect the current screening procedures include: speed curves exhibiting increased emissions following a low point around 30-35 mph, idle emission decreases and moving emission increases and a shift in worst case receptor concentration towards mid-block.

Development of an Approach for Setting a Threshold Screening-Level Procedure

To examine the potential for CO air quality violations for project-level settings, an investigation was performed for a combination of levels of service D, E and F; two intersection configurations; and one freeway configuration for two speeds and two dispersion settings (urban and rural). The modeling was performed using the MOBILE6 emission factors for the years 2005, 2015, 2025 and 2035.

For the two generalized intersections, one represented the intersection of two streets with three lanes plus a left turn bay on every approach. This was assumed to be in a suburban setting. The second was the intersection of streets with two lanes plus a left turn bay on each approach, and it was assumed to be in an urban setting. In addition, concentrations were also determined for sites in proximity to a six-lane freeway. These are a subset of the configurations, as described in Section 2.2.1.4.1

Three traffic conditions were developed for each intersection, representing LOS D, E, and F. Signal operations were estimated assuming a total of four phases, allowing for separate operation of the left turns and the through movement (along with right turns) of each roadway. It was assumed that 15% of the approach volume was turning left. Receptors were located at each corner, ten feet off each roadway, with seven more spaced at 40 feet along each leg of the intersection, for a total of 60 sites.

The freeway was assumed to have a narrow right-of-way, with a total width of 96 feet between the outside edges of the travel lanes. With an assumed volume of 2,000 vehicles per hour per lane, the highway would be operating at LOS E. The average speed based on the highway capacity manual estimate was 55 mph. Receptors were placed at an assumed right-of-way line, 78 feet from the center line (30 feet from the nearest travel lane), with additional sites located at 80, 180 and 480 feet from the nearest travel lane.

These general modeling scenarios were intended to represent typical "worst case" conditions. A practical range of vehicle volumes was developed and then classified by LOS. Signal operations for intersections were estimated using the Highway Capacity Software (HCS), and vehicle speeds on freeways were evaluated also using HCS. In both cases, common assumptions were applied (e.g., lane widths of 12 feet).

For intersections, current EPA guidance suggests that locations operating at LOS C or better will not require detailed analysis. The results of this review of MOBILE6 continue to illustrate that concentrations increase with decreasing levels of service, even though intersections at LOS C or better were not analyzed. This outcome can be attributed to the increase in the density of vehicles and duration of queues (the cause and the effect) as one moves from LOS D to LOS F. A higher condition (e.g., LOS A) can be expected to have freer flow and hence, higher speeds, but as shown on Figure 2.2.7 and 2.2.8, emissions start to increase only slightly at higher approach speeds and only after 35 mph. The slight increase in emission factor will be more than offset by the lower vehicle volumes resulting in overall lower emissions compared to the cases of LOS D to LOS F presented here.

Figure 2.2.7. 2005 MOBILE6.2 CO Emission Factors versus Average Speed

This chart depicts the CO emissions factors in grams/mile for speeds from 0 to 60 mph, for temperatures between 0 and 90 degrees F. This chart illustrates that higher emissions factors are at low speeds decreasing to 35 mph then increasing slightly at higher speeds. This chart also illustrates that for all speeds, emissions factors are higher for lower temperatures.

Figure 2.2.8. 2035 MOBILE6.2 CO Emission Factors versus Average Speed

This chart depicts the CO emissions factors in grams/mile for speeds from 0 to 60 mph, for temperatures between 0 and 90 degrees F. This chart illustrates that higher emissions factors are at low speeds decreasing to 35 mph then increasing slightly at higher speeds. This chart also illustrates that for all speeds, emissions factors are higher for lower temperatures.

For freeways, the assumed conditions (including 2,000 vehicles per hour per lane) result in LOS E, with a computed speed of 55 mph. This "worst case" scenario has been extended by also considering the same volume with a speed of 65 mph, which has a greater emission rate. This case would represent very aggressive drivers. Using the same typical assumptions, HCS assessments of LOS A, B or C indicates flows of 333, 667, or 1,333 vehicles per hour per lane, respectively, with average speeds of 67 mph. Therefore, the emission rates at a higher LOS would be equivalent to the "aggressive driver" LOS E assumption in this analysis, but the number of vehicles and therefore, the overall emissions would be substantially reduced.

General results of the CAL3QHC modeling effort are summarized in Figure 2.2.9. The values shown represent the highest predicted eight-hour CO concentration in the vicinity of the intersection assuming a persistence11factor of 0.7 to estimate the adjustment from the one-hour to the eight-hour concentrations. For the intersection, the location of the maximum concentration varied from the corner to a more mid-block location, but the location at the nearest receptor to the right-of-way always had the highest freeway concentration. The overall modeling approach applied the typical worst case assumptions. A wind speed of 1 m/s was evaluated at every wind angle in 10° increments. An atmospheric stability class of D was used with a mixing height of 1,000 meters and an ambient temperature of 10°F. MOBILE6 was applied using national default values to represent typical conditions.

The results shown are based on a background concentration of zero. In some states, location-specific monitoring data is used, and as discussed in the previous section, a rollback technique may be used for future CO background concentration estimates. However, these values must be considered carefully since, in many cases, the measurements represent both regional background and local traffic conditions. An FHWA guidance document (1986) suggested that a background concentration of 1 ppm would be appropriate for rural settings, and 2 to 3 ppm would be typical in urban areas. These estimates appear to be reasonable estimates of today's typical urban and rural CO background values, as EPA's most recent trend data (http://www.epa.gov/airtrends/carbon.html, USEPA, 2003) shows that, for 2002, the national average 2nd highest high eight-hour average CO concentration is around 3 ppm (likely a typical urban setting) and the lowest 2nd highest high eight-hour average CO concentration is around 1.5 ppm (likely a rural setting)  . Additionally, as a first approximation, it is suggested that for screening purposes attainment areas assume a background concentration of 3 ppm. Non-attainment and maintenance areas should use previously developed methods for establishing background concentrations.

When looking at the values reported in Figure 2.2.9, the highest predicted eight-hour concentration is slightly above 9.0 ppm for traffic traveling at 65 mph on a freeway in 2005 in an area without an I/M program. This would imply a potential violation of CO NAAQS (9.0 ppm) at this level of service in the near-term. For this case, a more refined modeling approach using hour-by-hour traffic and meteorology would be recommended. Assuming a "worst case" background concentration of 3 ppm, the implication is that a project-level eight-hour concentration of 6 ppm or less is needed to satisfy the eight-hour NAAQS.

The results displayed in Figure 2.2.9 indicate a limited potential for violations of the NAAQS at typical locations in the near-term, and by 2015, the potential effectively disappears. An interesting feature in Figure 2.2.9 is that the freeway scenario operating at LOS E has the potential for higher CO levels than an intersection operating at LOS F. This implies that the freeway scenario should be examined, along with intersections, in setting a screening threshold assessment procedure.

Results from modeling also showed that, for intersections, the corner receptor no longer has the highest concentration; the greatest concentration is now typically found about 200 feet behind the front of the queue. In both the intersection and freeway cases, the benefits of I/M programs are limited (on the order of 20% or less), but improvements due to fleet turnover between 2005 and 2035 are substantial (on the order of 50%), with most of that improvement in the first ten years. This latter effect reflects the introduction of today's technology into the remaining portion of the fleet.

To more fully asses the applicability of this screening approach to the more extreme roadway settings, two additional roadway configurations were evaluated. The intersection was expanded to consist of two five-lane approaches and included a dual left-turn lane, four through lanes, and a right-turn lane for each approach. A ten-lane freeway was also evaluated. As with the previous case, this freeway was assumed to have a narrow right-of-way. However, concentrations at the right-of-way site (30-feet from the nearest travel lane) and a site 80 feet from the nearest travel lane (approximately 150 feet from the center line) have been shown in Figure 2.2.10. As with the earlier modeling runs, nearly all of the predicted levels are below 6 ppm, excluding modeling runs for 2005 and the sites near freeway rights-of-way.

The current data has been presented with an "open format" since it includes no background concentration. This value will vary depending on the guidance a given state might provide or the nature of the location (e.g., rural or urban). Although the modeling was performed for one-hour periods, the figures have been converted to eight-hour estimates using an assumed persistence factor of 0.7. Again, different states might suggest different values for this adjustment, but the eight-hour standard is more susceptible to being exceeded; therefore, the 0.7 persistence factor-adjusted results are shown in the figures.

Because application of the results of this study might vary by state, it is not possible to propose a universal screening policy at this stage. However, it appears that a straight-forward screening based on the LOS will be a practical approach to future air quality evaluations. Some states have used LOS C as an initial screening assessment and then applied a quantitative approach (e.g., simplified modeling or "look-up" tables) if a lower LOS was found. The current analysis indicates that a relatively low LOS (LOS E) will still meet the air quality standards in most cases, and it is unlikely that a proposed project would be advanced if it were not expected to improve operations. In terms of project-level studies, this approach would be similar for an attainment or a nonattainment area.

As with other screening methods, it will be important to develop an appropriate set of disclaimers. The assumptions applied in this study attempted to identify "worst case" assumptions that would address a wide range of projects. Still, it was determined that rare settings and conditions might lead to air quality concerns, for example, an urban intersection in close proximity to a parking garage or a site with limited offset from a high-volume freeway. Nevertheless, LOS alone might have wide applicability as a screening tool for project-level air quality assessments.

Limitations and Applicability of this Threshold Screening Approach

This assessment indicates that the potential screening threshold for project-level studies would, for most typical conditions, rarely require the need for a detailed analysis with CAL3QHC. In the past, LOS C has been widely used as a screening threshold to reduce the need for detailed modeling. If there were no signalized intersections associated with a project where the operations would be classified as LOS D or worse, then it was determined that there would be no air quality impacts. Due to the changes from MOBILE 5 to MOBILE6, the relative role of cruise emissions has increased, while the idle emission factors have been substantially reduced. Based on the assessment completed in this study, it appears likely that detailed modeling can be excluded for both intersection and freeway locations with LOS E or better under a wide variety of conditions, especially when looking beyond the near-term period (2015 or later).

The applicability of this screening threshold is dependent on the circumstances of a given project and how closely they resemble the "normal" conditions used and the other assumptions applied here. This effort has focused on applying a reasonable worst case condition that would capture the vast majority of real-world conditions. However, several exception type cases can be noted:

These type cases would need to be examined on a case-by-case basis. Nevertheless, it appears likely, based on the modeling efforts shown, that the vast majority of typical projects will not require detailed modeling if the traffic analysis indicates that all signalized intersections and freeway sections will operate at LOS E or better.

Figure 2.2.9. Maximum CO Concentrations near Typical Intersections and Freeways

This bar chart depicts the CO concentrations (ppm) for various traffic conditions and years, with and without I/M programs for basic intersections. Concentrations with LOS F, E and D are shown for typical suburban and urban intersections for years 2005, 2015, 2025, and 2030. The chart is used to illustrate that CO concentrations are highest in the near term at worse LOS, without I/M, and improve in later years. The results indicate a limited potential for violations of the NAAQS at typical locations in the near term, and by 2015, the potential effectively disappears.

This bar chart depicts CO concentrations (ppm) for various speeds and years, with and without I/M programs for base freeway at LOS E. Concentrations are shown at 65 mph and 55 mph for years 2005, 2015, 2025, and 2030. In addition, the 8-hour CO NAAQS of 9.0 ppm is also depicted. The chart is used to illustrate that CO concentrations are highest in the near term at higher speeds, without I/M, and improve in later years. The results indicate a limited potential for violations of the NAAQS at typical locations in the near term, and by 2015, the potential effectively disappears.

Figure 2.2.10. Maximum CO Concentrations near Major Intersections and Freeways

This bar chart depicts the CO concentrations (ppm) for various traffic conditions and years, with and without I/M programs for major intersections. Concentrations with LOS F, E and D are shown for a major intersection for years 2005, 2015, 2025, and 2030. The chart is used to illustrate that CO concentrations are highest in the near term at worse LOS, without I/M, and improve in later years. The results indicate a limited potential for violations of the NAAQS at major locations in the near term, and even less after 2015.

 

This bar chart depicts CO concentrations (ppm) for various speeds and years, with and without I/M programs for major freeway at LOS E. Concentrations are shown at 65 mph and 55 mph for years 2005, 2015, 2025, and 2030. The chart is used to illustrate that CO concentrations are highest in the near term at higher speeds, without I/M, and improve in later years. The results indicate a limited potential for violations of the NAAQS at major locations in the near term, and even less after 2015.

Previous Contents Next

FHWA Home | HEP Home | Feedback
FHWA
United States Department of Transportation - Federal Highway Administration