National Academies Press: OpenBook

Travel Demand Forecasting: Parameters and Techniques (2012)

Chapter: Chapter 7 - Case Studies

« Previous: Chapter 6 - Emerging Modeling Practices
Page 100
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 100
Page 101
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 101
Page 102
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 102
Page 103
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 103
Page 104
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 104
Page 105
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 105
Page 106
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 106
Page 107
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 107
Page 108
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 108
Page 109
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 109
Page 110
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 110
Page 111
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 111
Page 112
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 112
Page 113
Suggested Citation:"Chapter 7 - Case Studies." National Academies of Sciences, Engineering, and Medicine. 2012. Travel Demand Forecasting: Parameters and Techniques. Washington, DC: The National Academies Press. doi: 10.17226/14665.
×
Page 113

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

100 7.1 Introduction As discussed in Chapter 5, there are two primary uses for the data provided in this report: • Developing travel model components when no local data suitable for model estimation are available and • Checking the reasonableness of model components devel- oped using local data. In the first case, local data should be collected to validate the models or model components developed based on this report. In the second case, the data in this report can be used to supplement and support the validation and reasonableness checking process. This chapter presents two case studies to illustrate the use of the report for these purposes. In the first case study, the MPO for a large metropolitan area, Gtown, has recently conducted a household activity/travel survey, and has recalibrated its model using the new data. The information from this report is used to verify that the model parameters and results from this recalibration are reasonable. Note that this case study does not represent the entire validation effort for such a model, which must include other checks (for example, sensitivity tests and checks of forecasts). The second case study is for a small urban area, Schultzville, that has never had a travel forecasting model and does not have any area-specific travel data. The MPO for this area has borrowed the model structure from another small area and is using that structure to develop a model for its area. 7.2 Model Reasonableness Check Gtown is a large metropolitan area with more than 5 million residents and a diverse public transportation system that includes various rail and bus services. A household activity/ travel survey was completed 3 years ago; and data from that survey, transit surveys, and traffic counts have been used by MPO staff to recalibrate the trip-based travel forecasting model for the area. The MPO staff wants to make sure that the newly calibrated model is reasonable and has decided to compare model parameters and selected model results with information contained in this report. In this section, parameters from the recalibrated Gtown model are compared to those provided in Chapter 4 of this report. The information provided in Chapter 4 often does not use the same variables or uses them at different levels of aggregation. Therefore, throughout this section, either parameters from Chapter 4 or the Gtown data are aggregated to make them comparable. One prime example of this dif- ference relates to trip purpose. The Gtown model has five trip purposes: home-based work (HBW), home-based shop (HBS), home-based other (HBO), nonhome-based work (NHBW), and nonhome-based other (NHBO). Parameters and data in Chapter 4 are provided for three purposes: HBW, home-based nonwork (HBNW), and nonhome based (NHB) (alternatively, for four purposes, including home-based school, but this purpose is not used in the Gtown model). Therefore, for Gtown parameters to be compared to those in this report, the Gtown data for the five trip purposes must be collapsed to the classic three trip purposes. 7.2.1 Trip Generation Trip Production Rates Trip production rates for Gtown for all trip purposes are applied using a three-dimensional, cross-classification model with household size, number of vehicles, and income level as variables. All person trips are modeled, including non- motorized trips. Table C.5 in Appendix C provides HBW trip rates derived from NHTS data, based on three different cross-classifications; two of which are household size by number of vehicles and C h a p t e r 7 Case Studies

101 household size by income level. However, the income defini- tions in the Gtown model are significantly different than those in the NHTS data summaries. It was therefore decided to com- pare the rates using the household size by number of vehicles classification, as shown in the middle section of Table C.5. Table 7.1 shows this comparison. Note that the Gtown model uses only four household size categories (the largest is 4 or more persons), while the NHTS data summary in Table C.5 uses five categories (the largest is 5 or more persons). As shown in Table 7.1, the Gtown trip production rate is 1.7 HBW trips per household, compared to 1.4 trips per household from Chapter 4; a difference of about 20 percent. This difference seems to be concentrated in smaller households, which predominantly are childless households. The Gtown MPO theorized that the difference may be due to a lower than average rate of retired people living in the region. In addition, Gtown has higher than average transit usage, and there may be more direct trips between home and work than in other areas since auto trips are more likely to include stops on the way to or from work (leading to more HBNW and NHB trips in place of HBW trips). The basic question for the MPO is whether the trip rates derived from their local survey are more reliable than those from the NHTS, which has a higher sample size but is a national sample collected mostly outside Gtown. Certainly, the difference indicates that checks of the Gtown survey data are warranted. Table C.6 provides HBNW trip rates derived from NHTS data, based on three different cross-classifications, two of which are household size by number of vehicles and household size by income level. Separate rates are presented for areas with populations more than 500,000 and less than 500,000. The appropriate rates to use for this comparison are those for the areas of less than 500,000. It was decided to compare the rates using the household size by number of vehicles classification, as shown in the third section of Table C.6. Table 7.2 shows this comparison. As shown in Table 7.2, the Gtown trip production rate is 4.6 HBNW trips per household, compared to 5.6 trips per household from Table C.6; a difference of nearly 20 percent. For HBNW trips, the differences seem to be across all house- hold size and vehicle availability categories. Again, the differ- ences indicate that further checks of the Gtown survey data are warranted. Table C.7 provides NHB trip rates derived from NHTS data, based on three different cross-classifications, two of which are household size by number of vehicles and household size by income level. It was decided to compare the rates using the household size by number of vehicles classification, as shown in the middle section of Table C.7. Table 7.3 shows this comparison. As shown in Table 7.3, the Gtown trip production rate is 2.3 NHB trips per household, compared to 3.0 trips per household from Table C.7; a difference of nearly 25 percent. For NHB trips, the differences seem to be across most house- hold size and vehicle availability categories, although the differences are higher in larger households. Again, the differ- ences indicate that further checks of the Gtown survey data are warranted. NHTS Data (from Table C.5) Autos Persons 1 2 3 4 5+ Average 0 0.2 0.7 1.1 1.0 0.9 0.5 1 0.6 0.8 1.2 1.7 1.5 0.8 2 0.7 1.3 2.0 2.0 2.3 1.6 3+ 0.9 1.4 2.6 2.9 3.3 2.3 Average 0.5 1.2 2.0 2.3 2.4 1.4 Gtown Trip Rates Autos Persons 1 2 3 4 Average 0 0.9 1.3 1.4 1.5 1.1 1 0.9 1.4 1.8 1.8 1.3 2 1.0 1.6 2.0 2.1 1.8 3+ 1.0 1.7 2.4 2.7 2.2 Average 0.9 1.5 2.1 2.2 1.7 Table 7.1. Comparison of Gtown HBW trip production rates to NHTS data from Table C.5.

102 NHTS Data (from Table C.6) Vehicles Household Size 1 2 3 4 5+ Average 0 1.4 3.8 5.6 7.5 10.0 3.2 1 1.9 3.9 6.5 9.0 11.8 3.7 2 2.4 4.0 6.5 11.0 14.0 6.8 3+ 2.5 4.0 7.3 11.0 14.5 8.6 Average 1.8 4.0 6.7 10.6 13.4 5.6 Gtown Trip Rates Autos Persons 1 2 3 4 Average 0 1.6 2.3 2.9 3.4 1.9 1 1.6 3.2 4.4 7.4 2.8 2 1.7 3.3 5.4 8.3 5.1 3+ 1.9 3.4 5.5 9.2 6.2 Average 1.6 3.2 5.1 8.4 4.6 Table 7.2. Comparison of Gtown HBNW trip production rates to NHTS data from Table C.6. NHTS Data (from Table C.7) Vehicles Household Size 1 2 3 4 5+ Average 0 0.7 1.7 2.0 3.7 3.9 1.3 1 1.4 2.3 3.5 3.9 3.9 2.0 2 1.6 2.6 3.9 5.5 5.6 3.5 3+ 1.6 2.7 4.5 5.8 7.1 4.4 Average 1.3 2.5 3.8 5.3 5.7 3.0 Gtown Trip Rates Autos Persons 1 2 3 4 Average 0 1.0 1.3 1.7 2.1 1.2 1 1.4 2.0 2.3 3.1 1.8 2 1.7 2.1 2.3 3.2 2.5 3+ 2.0 2.4 2.5 3.6 2.9 Average 1.5 2.1 2.3 3.2 2.3 Table 7.3. Comparison of Gtown NHB trip production rates to NHTS data from Table C.7.

103 When total trips per household by all purposes from the Gtown model are compared to the information presented in Tables C.5 through C.7, the overall rate for Gtown is 8.6 trips per household, 14 percent lower than the total of 10.0 trips per household derived from the NHTS in Chapter 4. Based on this analysis, Gtown rates are lower than the national average. NHTS rates are averages based on urban areas with different characteristics, and the rates for individual areas can be dif- ferent. Furthermore, the higher Gtown rate for HBW trips, which are generally longer, may compensate for the lower overall rate. Trip Attraction Rates Table 4.4 summarizes average trip attraction rates from the MPO Documentation Database for the classic three trip purposes. The Gtown trip attraction model differs from the models shown in Table 4.4 in several ways. First, the employ- ment categories used for the Gtown HBNW and NHB attrac- tion models are defined differently than those in Table 4.4. For comparison purposes, the categories in the Gtown model were redefined to approximate those shown in Table 4.4. Second, the Gtown model stratifies trip attraction rates by area type. Weighted averages of Gtown’s area type-specific models were used to compare to the models in Table 4.4. The resulting comparison of trip attraction models is shown in Table 7.4. The models chosen for comparison from Table 4.4 were Model 1 for HBW, Model 3 for HBNW, and Model 2 for NHB. As can be seen in Table 7.4, the Gtown trip attraction rates are lower than the rates shown in Table 4.4, especially those for HBNW trips. The Gtown trip attraction models will generate fewer attractions than the models shown in Table 4.4. Since trip attractions are typically balanced to match produc- tions, the effects of the lower trip attraction rates might be small, but it makes sense to further check the trip attraction model estimation results, as well as the balancing of produc- tions and attractions. If the balancing process requires factoring up attractions to match productions, perhaps the rates could be adjusted upward. 7.2.2 Trip Distribution The reasonableness of the Gtown trip distribution model can be assessed by comparing the friction factors used in the Gtown gravity model and the resulting average trip lengths with comparable values provided in Section 4.5. Average Trip Length Table C.10 provides average trip length by mode (travel times in minutes) for urban areas of different sizes. The Gtown model results should be compared to the figures from Table C.10 cor- responding to areas of “1 million or more with subway or rail.” The Gtown trip distribution model produces a compos- ite travel time that reflects highway and transit travel times. Table 7.5 compares the average trip times for all modes by trip purpose from Table C.10 and compares those trip lengths to the times resulting from the Gtown model. The average trip duration for HBW trips from the Gtown model is 48 minutes, compared to an average HBW trip duration from the NHTS of 32 minutes. While most large metropolitan areas experience high levels of congestion during peak hours, the Gtown highway network is very congested during the peak periods, which can last 4 or more hours. Since most HBW trips are made during the peak periods, it can be expected that the travel time for those trips will be longer in Gtown than in other areas with a popula- tion over 3 million. Furthermore, Gtown encompasses a very large geographic area, also contributing to longer work trips. Another consideration is that Gtown has a relatively high transit share, and transit trips are longer than auto trips, as shown in Table C.10. Households Employment Basic Retail Service Total Home-Based Work Gtown Model 0.9 Model 1 from Table 4.4 1.2 Home-Based Nonwork Gtown Model 0.4 0.9 3.4 Model 3 from Table 4.4 0.7 0.7 8.4 3.5 Nonhome Based Gtown Model 0.1 3.3 0.7 Model 2 from Table 4.4 1.4 6.9 0.9 Table 7.4. Comparison of Gtown trip attraction rates to those shown in Table 4.4.

104 Nonetheless, the large discrepancy between the Gtown average trip length for HBW trips and that of other large areas does warrant some further review. The 48-minute average travel time resulting from the model was compared to the time reported in the household travel survey and the 2000 CTPP. The average travel time reported for HBW trips in the house- hold survey was also 48 minutes; and in the 2000 CTPP, it was 45 minutes, thus, confirming the modeled time. The average travel time for HBNW and NHB trips result- ing from the Gtown model compared more favorably to those shown in Table C.10. The mean HBNW travel time for Gtown is 17 minutes, compared to 18 minutes from the NHTS data. NHB travel times also compared favorably with both the Gtown and NHTS averages at approximately 20 minutes. The total travel time for all trips is 24 minutes from the Gtown model, which is 2 minutes longer than the time reported in Table C.10. If the Gtown trip generation rates and travel times are viewed together, they seem more reasonable. Studies have shown that people will only travel a certain amount of time for all pur- poses during a given day. Thus, the longer-than-usual amount of time spent making work trips can result in fewer and shorter trips for other purposes. Thus, the lower HBNW and NHB trip generation rates in the Gtown model may result from higher HBW trip rates and longer travel times. Gamma Function and Friction Factors The Gtown model distributes trips separately for each of four income groups and five purposes. A useful reasonableness check is to compare the Gtown estimated model parameters to those developed in other regions. The estimated friction factors calibrated for Gtown are represented by gamma functions that can be compared to those reported by areas of similar size. Table 4.5 provides trip distribution gamma func- tion parameters for eight MPOs, three of which are large. One way to compare friction factors used in the Gtown model to those resulting from the gamma functions for large MPOs in Table 4.5 is to compare the resulting graphs of friction factors to see if they are comparable. Figure 7.1 is a graph of the HBW friction factors for Gtown compared to those for the three large MPOs reported in Table 4.5. Friction factors for the three large MPOs and for the four HBW income groups in the Gtown model are shown in Figure 7.1. The Gtown friction factors for the two higher incomes are almost exactly the same as those for MPO 3. The friction factors for the two lower incomes are not as steep but are comparable to those for the three sample MPOs. Figure 7.2 is a graph of the HBS and HBO friction factors for Gtown compared to the HBNW friction factors for the three large MPOs. All of the Gtown friction factors lie between the values for MPO 1 and MPO 3, and the slopes for almost all purposes and income groups are very similar to that for MPO 1. Figure 7.3 is a graph of the NHB friction factors for Gtown compared to those for the three large MPOs reported in Table 4.5. The Gtown friction factors for NHBO trips are similar to the NHB values for MPO 2. The Gtown friction factors for NHBW trips are not as steep as those for any of the MPOs. Since neither the NHBO or the NHBW friction factors are as steep as those from any of the large MPOs, it is unlikely that friction factors for a combination of NHBO and NHBW trips would match the values for any of the MPOs. However, since the average travel times for NHB trips from the Gtown model are the same as those from the NHTS, the difference in friction factors may not be significant. 7.2.3 Mode Choice The Gtown model uses a nested logit mode choice model with coefficients for the classic three trip purposes. Auto submodes include drive alone and shared ride; and transit submodes include local, premium, and rail submodes (as well as separate models for auto and walk access). Variables used in the Gtown model include in-vehicle time, out-of-vehicle time, and a single cost variable. The coefficients of these vari- ables are summarized in Table 7.6. Tables 4.8, 4.11, and 4.14 present mode choice model parameters, by purpose, that are used by MPOs included in the MPO Documentation Database. For HBW trips, Models B, C, D, F, G, and I from Table 4.8, all of which are for urban areas All Modes (Minutes) Average All Trips HBW HBNW NHB Gtown 48 17 20 24 NHTS Averages from Table C.10 32 18 20 22 Difference 16 −1 0 2 Percentage Difference 50% −6% 0% 9% Table 7.5. Comparison of Gtown average trip length to NHTS data from Table C.10.

105 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 1 3 5 7 9 1 1 1 3 1 5 1 7 1 9 2 1 2 3 2 5 2 7 2 9 3 1 3 3 3 5 3 7 3 9 4 1 4 3 45 Travel time (min) MPO 2 MPO 1 MPO 3 MPO 3 Gtown Income 1 Gtown Income 2 Gtown Income 3 Gtown Income 3 Gtown Income 4 Gtown Income 4 Figure 7.1. Home-based work trip distribution friction factors. 0.1 1.0 10.0 100.0 1,000.0 10,000.0 100,000.0 1,000,000.0 10,000,000.0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 HBNW MPO 1 HBNW MPO 2 HBNW MPO 3 HBS Gtown Income 1 HBS Gtown Income 2 HBS Gtown Income 3 HBS Gtown Income 4 HBO Gtown Income 1 HBO Gtown Income 2 HBO Gtown Income 3 HBO Gtown Income 4 Travel time (min) Figure 7.2. Home-based nonwork trip distribution friction factors.

106 with populations of more than 1 million, have comparable variables to those in the Gtown model. Models F, G, and I are nested logit models. The coefficients of the Gtown HBW mode choice model are not too different from those of Models F, G, and I, although the Gtown cost coefficients are lower in absolute value. Looking at the relationships between coefficients, Table 7.7 shows that the ratio of the out-of-vehicle time and in- vehicle time coefficients in the Gtown model is comparable to those for Models F, G, and I, as shown in Table 4.9. The value of time in the Gtown model, however, is significantly higher than in the models from other areas. This compari- son holds for most of the other models shown in Tables 4.8 and 4.9. For HBNW trips, Models E, G, I, and K from Table 4.11 are for urban areas with populations of more than 1 million and have comparable variables. The in-vehicle time coefficient of the Gtown HBNW mode choice model is higher than those in the models from Table 4.11, while the Gtown cost coeffi- cients are lower in absolute value. Looking at the relationships between coefficients, Table 7.8 shows that the ratio of the out- of-vehicle time and in-vehicle time coefficients in the Gtown model is a bit lower than those of the other models, as shown in Table 4.12. The value of time in the Gtown model, however, is significantly higher than in the models from other areas. This comparison holds for most of the other models shown in Tables 4.11 and 4.12. For NHB travel, models F, G, and I from Table 4.14 are most comparable to Gtown. The coefficients in the Gtown HBNW mode choice model are fairly comparable. Looking at the relationships between coefficients, Table 7.9 shows that the ratio of the out-of-vehicle time and in-vehicle 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 1 3 5 7 9 11 13 1 Travel Time (min) 5 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 NHB MPO 1 NHB MPO 2 NHB MPO 3 NHW Gtown NHO Gtown Figure 7.3. Nonhome-based trip distribution friction factors. HBW HBNW NHB Parameter In-Vehicle Time Out-of-Vehicle Time Cost (low income) Cost (high income) Derived Relationships Out-of-Vehicle Time/ In-Vehicle Time Ratio 2.0 2.0 2.0 Value of In-Vehicle Time $9.08/hour (low income) $25.44/hour (high income) $8.80/hour (low income) $22.00/hour (high income) $1.76/hour HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based. −0.0212 minute −0.022 minute −0.029 minute −0.043 minute −0.0449 minute −0.0572 minute −0.0014 cent −0.0015 cent −0.0099 cent −0.0005 cent −0.0006 cent −0.0099 cent Table 7.6. Gtown mode choice model parameters.

107 Model Out-of-Vehicle Time/ In-Vehicle Time Value of In-Vehicle Time Gtown 2.0 $9.08 to $25.44/hour Model F (Table 4.9) 2.0 $3.94/hour Model G (Table 4.9) 2.3 $3.05/hour Model I (Table 4.9) 2.0 $3.00/hour Table 7.7. Relationships between coefficients from home-based work mode choice models for Gtown and from Table 4.9. Model Out-of-Vehicle Time/ In-Vehicle Time Value of In-Vehicle Time Gtown 2.0 $8.80 to $22.00/hour Model E (Table 4.12) 3.0 $3.69/hour Model G (Table 4.12) 4.6 $0.21/hour Model I (Table 4.12) 3.1 $0.48/hour Model K (Table 4.12) 3.0 $1.40/hour Table 7.8. Relationships between coefficients from home-based nonwork mode choice models for Gtown and from Table 4.12. Model Out-of-Vehicle Time/ In-Vehicle Time Value of In-Vehicle Time Gtown 2.0 $1.75/hour Model F (Table 4.15) 2.0 $4.04/hour Model G (Table 4.15) 11.3 $0.46/hour Model I (Table 4.15) 2.1 $2.00/hour Table 7.9. Relationships between coefficients from nonhome-based mode choice models for Gtown and from Table 4.15. time coefficients and value of time in the Gtown model are (as shown in Table 4.15) fairly comparable to those in Models F and I, but Model G appears to be an outlier. The other models shown in Tables 4.14 and 4.15 have coefficient values that vary widely, but the coefficients from Gtown fit well within this range. In summary, the value of time, indicating the willingness to pay for travel timesavings by switching modes, seems high for home-based trips in the Gtown model. The related model coefficients, mainly the cost coefficients for these trip purposes, should be reviewed. 7.2.4 Automobile Occupancy The Gtown mode choice model forecasts auto driver and auto passenger trips by purpose separately. Table 7.10 provides a comparison of the resulting Gtown auto occupancy rates compared to the values reported from the NHTS in Table 4.16. As Table 7.10 shows, the Gtown home-based auto occupan- cies are within 5 percent of those from the NHTS. Gtown NHB auto occupancies are noticeably lower than those from the NHTS. The NHB mode choice model should be checked regarding how auto driver and passenger choices are made. HBW HBNW Nonhome Based All Trips NHBW NHBO Gtown 1.05 1.64 1.10 1.48 1.39 Table 4.16 1.10 1.72 1.66 1.55 HBW = home-based work; HBNW = home-based nonwork; NHBW = nonhome-based work; NHBO = nonhome-based other. Table 7.10. Comparison of average daily vehicle occupancy by trip purpose.

108 The household survey is another source against which auto occupancy rates by purpose can be checked. 7.2.5 Time of Day Table 7.11 provides a comparison between the modeled times of day for auto trips in the Gtown model with those derived from NHTS data that are shown in Table C.11. As Table 7.11 shows, the percentage of travel occurring in peak periods is lower in Gtown than in the national sur- vey, and the nighttime percentage of travel is substantially higher in Gtown. As mentioned earlier, the Gtown highway system is very congested, and the peaks are much longer than in other comparable cities. It would seem reasonable, therefore, that peak spreading would be more prevalent in Gtown. This finding could be confirmed using other data sources such as traffic counts. 7.2.6 Summary This section provides a comparison of model parameters and results produced by the model for a hypothetical large MPO and the values in this report. Overall, the Gtown model parameters and results appear to be reasonable when compared to the values in Chapter 4 of the report, although some Gtown model parameters, such as cost coefficients in the mode choice models for home-based trip purposes, should be checked further. The congested nature of Gtown does appear to result in fewer nonwork trips, very long work trips, and extended peak periods. 7.3 Model Development Case Study for a Smaller Area without Data for Model Estimation This case study is for a small urban area that never had a travel forecasting model and does not have any local data from which to estimate model parameters. The MPO for this hypothetical city, Schultzville, borrowed the model structure from another small area and used that structure to develop its own model. Schultzville is an urban area of about 100,000 people. It has very little in the way of pub- lic transportation, so the MPO decided to develop a daily (i.e., no time of day), three-step model with auto trips only, using the classic three trip purposes. 7.3.1 Zone and Highway Network Definition Highway Network Definition A highway network for the Schultzville area was developed to obtain acceptable volumes on minor arterials; therefore, collectors and local roads were included in the network. Digital street files available from the U.S. Census Bureau (TIGER/ Line files) were used to create the highway network shown in Figure 7.4. Freeways, major arterials, minor arterials, collector links, and some local roads were coded into the network. The following are examples of some of the fields coded for nodes and links in the network: Time Period Gtown Table C.11 Difference Percent Difference 14.4% 17.1% 34.4% 35.6% 27.4% 32.1% 23.8% 15.2% 8.6% 57% Total 100.0% 100.0% −2.7% −16% −1.2% −3% −4.7% −15% 6:00 a.m.−9:00 a.m. 9:00 a.m.−3:00 p.m. 3:00 p.m.−7:00 p.m. 7:00 p.m.−6:00 a.m. Table 7.11. Comparison of time of day for auto trips. Figure 7.4. Schultzville highway network.

109 • XY coordinates—Geographic coordinates for nodes; • Node identifiers (anode/bnode)—Unique numbers assigned to each end of a link; • Distance—Distance in miles between anode and bnode; • Functional (link) classification—Type of facility (e.g., major arterial, minor arterial, etc.); • Traffic count volume—Average daily volume of traffic on link (where available); • Number of lanes; • Facility type; • Area type—Location and development characteristics of area that link serves (e.g., urban, suburban, rural, etc.); and • Link capacity and free-flow speed—Link capacities are a function of the number of lanes on a link. Area type and facility type were used to define per-lane default capacities and speed. The number of lanes was also checked using field verification or aerial imagery to ensure accuracy. Transportation Analysis Zone Definition A map of Schultzville transportation analysis zones is shown in Figure 7.5. Each TAZ has a centroid, which is a point that represents all travel origins and destinations in a zone. 7.3.2 Socioeconomic Data Socioeconomic data—household and employment data for the modeled area—were organized into the TAZs. Esti- mates of base-year socioeconomic data by TAZ were devel- oped for use in model development. The population and household data for Schultzville came from the decennial census. Data such as income and vehicle availability were derived from the ACS. Basic socioeconomic data by TAZ were derived for Schultzville, including households, population, total employ- ment, retail employment, service employment, manufacturing employment, nonmanufacturing employment, and school enrollment. More detailed data, such as number of persons per household, household income, workers per household, and vehicles owned per household, as well as cross-classifications of households by zone, were also derived from the U.S. Census and ACS. Employment data by TAZ were derived from data pro- vided by the state employment commission. Each employer was identified by a federal identification number, number of employees, and a geocodable address, which were allocated to TAZs. Since these data were keyed to where the payroll is prepared for employees, the MPO made adjustments to allocate employment to the proper TAZ, where necessary. School enrollment data by school were provided by the Schultzville School District and allocated to the appropriate TAZs; this information was supplemented by information the MPO collected directly from the larger private schools in the region. 7.3.3 Trip Generation Trip Productions The MPO was able to develop estimates of households cross- classified by household size and number of vehicles, and by workers by number of vehicles for each zone. The information in Tables C.5 through C.7, which shows trip rates derived from 2009 NHTS data, was used to estimate productions by trip purpose. The HBNW trip rates for areas with less than 500,000 residents in Table C.6 were used. These trip generation rates were applied to the socioeconomic data for each zone to create total productions by purpose by zone. An example calculation is provided for home-based work trips in Table 7.12. Trip production rates from Table C.5 were multiplied by the households cross-classified by workers and vehicles to obtain a total of 1,092 HBW trip productions occurring in the sample zone. (Note that Table C.5 provides rates for households with three or more vehicles, while data for Schultzville were only available for households with two or more vehicles; therefore, the rates for two vehicle and three vehicle households were averaged for use in Schultzville.) Trip Attractions The values for trip attraction rates for motorized trips, shown in Table 4.4, were used as a trip attraction model for Schultzville. Model 1 from this table was used for each trip purpose. An example calculation is provided for home-based Figure 7.5. Schultzville TAZs.

110 work trips in Table 7.13. Data for households, employment, and school enrollment for each Schultzville TAZ were multiplied by the trip attraction rates from Table C.7 to achieve a total of 130 HBW, 583 HBNW, and 306 NHB trip attractions occur- ring in the sample zone. 7.3.4 Trip Distribution The doubly constrained gravity model, described in Equation 4-5, was used as the trip distribution model for Schultzville. The inputs to the trip distribution model include: • The trip generation outputs—productions and attractions by trip purpose for each zone; • Highway travel time, as the measure of travel cost between each pair of zones; and • Friction factors, as discussed in the following section. The outputs are trip tables, production zone to attraction zone, for each trip purpose. Because trips of different purposes have different levels of sensitivity to travel time and cost, trip distribution is applied separately for each trip purpose, with different model parameters. Development of Travel Time Inputs Zone-to-zone (interzonal) travel costs. This case study used the simplest cost variable, highway travel time, which is an Number of Autos Workers Total 0 1 2 3+ Home-Based Work Trip Production Rates 0 0.0 1.1 2.0 4.0 1 0.0 1.1 2.5 4.3 2+ 0.0 1.3 2.6 4.5 Example TAZ Data 0 20 30 10 0 1 65 155 75 4 2+ 4 90 170 24 Example Zone Trip Productions 0 0 33 20 0 1 0 171 188 17 2+ 0 116 442 106 Total Productions 0 319 650 123 1,092 Table 7.12. Example trip production calculation. Trip Pu rp os e Households School Enroll me nt Em ploy me nt Trip Attractions Basic Retail Service Total Ho me -Based Work Model 1 1.2 Sa mp le TA Z Va lu e 108 Tr ip A ttractions 130 130 Ho me -Based Nonwork Model 1 0.4 1.1 0.6 4.4 2.5 Sa mp le TA Z Va lu e 320 210 34 10 64 Tr ip A ttractions 128 231 20 44 160 583 Nonh om e Base d Model 1 0.6 0.7 2.6 1.0 Sa mp le TA Z Va lu e 320 34 10 64 Tr ip A ttractions 192 24 26 64 306 Total Trips Attr acted to Sa mp le TA Z 1,019 Table 7.13. Trip attractions calculation for sample TAZ.

111 adequate measure for a small area such as Schultzville. This area does not have a significant level of auto operating cost beyond typical per-mile costs—for example, relatively high parking costs or toll roads—or extensive transit service. The zone-to- zone highway travel time matrix was developed through “skim- ming” the highway network using travel modeling software. The highway assignment process does not require that times be coded on the centroid connectors since those links are hypo- thetical constructs representing the travel time within zones. Initial skim times from the network assignment did not include time representing travel within zones, or terminal time. Intrazonal time. Intrazonal times were defined as one- half of the average of the skim times to the three nearest neighboring zones. Terminal time. Terminal times, which represent the time required to park a vehicle and walk to the final destination, or vice versa, were added to the intrazonal time. Terminal times of 4 minutes were added to the time for any trip where a trip end was in the business district, and 2 minutes were added for trip ends elsewhere. Friction factors. Friction factors were derived for each purpose (HBW, HBNW, and NHB trips) using a gamma function (described in Equation 4-6) using the b and c values shown in Table 4.5 for Small MPO 1. The gamma func- tion parameters, including the scaling factor a, are shown in Table 7.14. The resulting friction factors are plotted in Figure 7.6. The resulting average travel times by trip purpose from this first application of the gravity model were evaluated to determine if the distribution was acceptable. Friction factors were calibrated to match average travel times using an iterative process. No local data existed regarding average travel times, so the best option in this situation was to start with parameters from another modeling context. Average trip lengths by trip purpose are presented in Table C.10, and were used as a basis of comparison with trip lengths resulting from the initial trip distribution in Table 7.15. As can be seen in Table 7.15, the average trip lengths resulting from this initial set of friction factors are lower than the average travel times reported in Table C.10. Since Schultzville is a small geographic area with little congestion, one might expect that the average trip length would be lower than the NHTS aver- age reported for all areas with a population less than 500,000. However, the initial mean travel times were judged too low. The initial friction factors were adjusted iteratively to test variations Parameter HBW HBNW NHB a 26,000 130,000 260,000 b c HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based. −0.265 −1.017 −0.791 −0.04 −0.079 −0.195 Table 7.14. Gamma function parameters for Schultzville. 0.10 1.00 10.00 100.00 1,000.00 10,000.00 100,000.00 1,000,000.00 61 11 16 21 26 31 36 41 46 51 56 NHB HBNW HBW HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based. Travel time (min) Figure 7.6. Schultzville case study initial friction factors.

112 that achieved a higher average trip length for all purposes. The friction factors resulting from this fitting process are shown in Figure 7.7. The comparison of the mean travel times resulting from the use of these revised friction factors with those from Table C.10 is shown in Table 7.16. The final friction factors are not as steep as those that were initially used and result in mean travel times closer to those shown in Table C.10. 7.3.5 External Trips The best source of data for estimating external trips (EI and EE) is a roadside survey conducted at external stations; however, no such survey was available for Schultzville. The state in which Schultzville is located has a statewide travel model that provided information on EE trips and EI trips for the study area. The statewide model provided the origin and destination station, as well as the volume for EE trips. For EI trips, a select link assignment from the statewide model provided the number of trips entering and leav- ing each external station allocated to the statewide model zones. These needed to be suballocated to the Schultzville model zones based on the relative internal attractions and productions in each TAZ compared to the total in the larger statewide model zones. 0.10 1.00 10.00 100.00 1,000.00 10,000.00 100,000.00 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 Initial HBW Initial HBNW Initial NHB Final HBW Final HBNW Final NHB HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based. Travel time (min) Figure 7.7. Schultzville case study final friction factors. HBW HBNW NHB Urban Area Population from Table C.10 Less than 500,000 All population ranges Other urban area Value from Table C.10 20 minutes 18 minutes 18 minutes Schultzville 15 minutes 12 minutes 9 minutes Difference 5 minutes 6 minutes 9 minutes HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based. Table 7.15. Initial evaluation of Schultzville mean travel times. Table 7.16. Evaluation of Schultzville mean travel times using adjusted friction factors. HBW HBNW NHB Urban Area Population from Table C.10 Less than 500,000 All population ranges Other urban area Value from Table C.10 20 17 3 minutes 18 15 3 minutes 18 15 3 minutes Schultzville minutes minutes minutes Difference minutes minutes minutes HBW = home-based work; HBNW = home-based nonwork; NHB = nonhome based.

113 7.3.6 Vehicle Occupancy The highway assignment step, discussed in Section 7.3.7, requires tables of vehicle trips, while the output of early model steps was in person trips. Person trips made by auto from the earlier steps were converted to vehicle trips using the factors provided in the first row of Table 4.16, which represent all auto modes for daily travel. These factors—1.10 for HBW, 1.72 for HBNW, and 1.66 for NHB—were applied to the auto passenger trip tables produced by the trip distribution step, as described in Section 7.3.4. 7.3.7 Highway Assignment Trip tables from origins to destinations (O-D format) are required for the daily highway assignment; however, the HBW and HBNW trip tables resulting from the previous steps pro- vide trip tables from productions to attractions (P-A format). The P-A trip tables were converted to O-D trip tables by splitting the value in each cell in half to create two duplicate matrices, transposing the values in one of the matrices, and adding the two matrices together. The resulting O-D trip tables were then ready to be assigned to the highway network. A user equilibrium assignment using the BPR formula for capacity restraint was used for assigning vehicle trips to the highway network. Values for the a and b parameters were needed for application of the BPR formula (described in Section 4.11.1). Table 4.26 presents BPR function parameters used by 18 MPOs. The most appropriate values for Schultzville are those shown for areas with a population less than 200,000: a = 0.15 for freeways, 0.45 for arterials; and b = 8.8 for freeways, 5.6 for arterials. The results of the traffic assignment are shown as a band- width plot in Figure 7.8. In this diagram, the width of each link in the network is proportional to the volume on that link. An assessment was made of the quality of the traffic assign- ment on links where traffic counts were available by comparing the root mean square error (RMSE) of assigned values to traffic counts by facility type. As can be seen in Table 7.17, the RMSE is within an acceptable range for all facility types, except local roads. Since the goal of the model was to get acceptable values for minor arterials, the results were deemed acceptable. Figure 7.8. Schultzville case study final assigned volumes. Table 7.17. RMSE comparison of modeled volumes with traffic counts. Functional Class Links ADT Error Percentage Error Acceptable Error Freeways 18 228,340 15,021 6.6% +/−7% Principal Arterials 90 538,210 37,674 7.0% +/−10% Minor Arterials 226 730,030 80,303 11.0% +/−15% Collectors 218 304,110 66,904 22.0% +/−25% Locals 14 20,000 10,400 52.0% +/−25%

Next: References »
Travel Demand Forecasting: Parameters and Techniques Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 716: Travel Demand Forecasting: Parameters and Techniques provides guidelines on travel demand forecasting procedures and their application for helping to solve common transportation problems.

The report presents a range of approaches that are designed to allow users to determine the level of detail and sophistication in selecting modeling and analysis techniques based on their situations. The report addresses techniques, optional use of default parameters, and includes references to other more sophisticated techniques.

Errata: Table C.4, Coefficients for Four U.S. Logit Vehicle Availability Models in the print and electronic versions of the publications of NCHRP Report 716 should be replaced with the revised Table C.4.

NCHRP Report 716 is an update to NCHRP Report 365: Travel Estimation Techniques for Urban Planning.

In January 2014 TRB released NCHRP Report 735: Long-Distance and Rural Travel Transferable Parameters for Statewide Travel Forecasting Models, which supplements NCHRP Report 716.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!