Estimating the Value of Truck Travel Time Reliability (2019)

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.

60 The Reliability Valuation Framework was used to study truck travel along a freeway corridor in downtown Austin, Texas. This chapter first describes the existing traffic conditions in eco- nomic terms, estimating the user costs resulting from recurring congestion and unreliability to identify bottlenecks that caused disproportionally high costs to the movement of freight. This bottleneck identification methodology is general and can be applied to corridors, regions, and even states by using publicly available data. The Reliability Valuation Framework was used to evaluate a hypothetical project to improve traffic on this corridor. Estimates of the monetary benefits of the project in decreasing travel time and unreliability were developed; these can be compared with capital costs to assess the economic favorability of the project. These analyses demonstrate how to use the value of reli- ability (VOR) figures estimated in Chapter 4 to answer common freight-planning questions. 6.1 Bottleneck Identification Travel time data from a 9-mile segment of I-35 in downtown Austin were analyzed to demonstrate the methodology for identifying truck bottlenecks. The travel time data analyzed came from the National Performance Management Research Data Set (NPMRDS) for 2017. These data were cleaned as described in Section 5.2.2 to eliminate sources of variation that obfuscate reliability measurement. Table 6-1 shows how to calculate the user costs for a 0.88-mile segment that runs south from Manor Road to 12th Street [Traffic Message Channel (TMC) code 112N04895]. Truck volumes were obtained from automatic traffic recorders on this segment. The cleaned data were used to calculate the average travel time and 95th percentile travel time for each hour of the day. Subtracting these two values provides an estimate of the 95th percentile delay. As can be seen from this table, unreliability increases from 1 p.m. to 9 p.m., with the highest values occurring during the afternoon peak of 4 to 8 p.m. The average travel time cost per mile per day and the unreliability cost per mile per day were calculated with the formulas shown in the table. User costs were normalized by segment length (centerline miles) to develop an estimate of the rate of cost accumulation per mile that can be compared between different segments. Freight users spent $37,874 per day in average travel time costs and $49,777 in unreliability costs traversing this segment. This totals $87,651 in user costs, which, normalized by segment length, leads to a rate of user cost accumulation of $99,377 per day per mile of roadway. This averages $10 per truck per mile (including all truck operation costs and supply chain costs resulting from unreliability), although trucks that travel during the afternoon peak hour, when it could take more than 10 minutes to traverse this short segment, account for most of these costs, increasing to more than $20 dollars per mile. C H A P T E R 6 Case Study

Case Study 61 The calculations described in Table 6-1 were implemented for the 75 segments that make up the I-35 corridor. The results for southbound travel are summarized in Figure 6-1 and the results for northbound travel are summarized in Figure 6-2. For southbound travel, user costs spiked between Mileposts 237 and 235, with unreliability contributing to roughly 65 per- cent of these costs. This unreliability was likely caused by trucks facing recurring congestion from commuters entering downtown Austin during the morning peak. When traveling in this direction, users faced greater unreliability north of the corridor, with unreliability representing almost 70 percent of user costs. For northbound travel on I-35, the analysis found that congestion increased significantly on and around the bridge crossing the Colorado River. This increase was likely caused by commuters trying to reach downtown Austin during the morning peak. User costs were Hour of Day Length (miles) Volume (trucks/h) NPMRDS Analysis User Costs Average Travel Time (min) 95th Pctl. Travel Time (min) 95th Pctl. Delay (min) VOT ($/h) Average Travel Time Cost ($/day- mile) VOR ($/h) Unreliability Cost ($/day-mile) Total Costs ($/day-mile) A B C D E = D–C F G = B*C*F/(A*60) H I = B*E*H/(A*60) J = G+I 0.882 66 0.901 0.967 0.067 67 75 160 13 88 0.882 50 0.870 0.926 0.056 67 55 160 8 63 0.882 48 0.863 0.925 0.063 67 52 160 9 61 0.882 64 0.855 0.922 0.067 67 69 160 13 82 0.882 125 0.850 0.909 0.058 67 134 160 22 156 0.882 259 0.882 0.963 0.081 67 288 160 63 351 0.882 440 0.981 1.095 0.114 67 544 160 152 695 0.882 511 1.009 1.123 0.114 67 649 160 176 825 0.882 489 1.052 1.290 0.238 67 648 160 352 1,000 0.882 487 1.092 1.524 0.432 67 670 160 636 1,306 0.882 522 1.233 2.406 1.173 67 810 160 1,851 2,661 0.882 568 1.686 3.338 1.652 67 1,206 160 2,836 4,041 0.882 599 2.477 4.483 2.006 67 1,868 160 3,632 5,500 0.882 650 2.786 4.958 2.172 67 2,280 160 4,267 6,547 0.882 749 3.910 6.548 2.638 67 3,687 160 5,971 9,658 0.882 861 5.591 8.410 2.820 67 6,060 160 7,338 13,398 0.882 908 7.641 10.455 2.814 67 8,734 160 7,723 16,458 0.882 764 8.687 11.756 3.068 67 8,356 160 7,085 15,441 0.882 545 5.958 10.000 4.041 67 4,088 160 6,658 10,746 0.882 404 2.590 5.682 3.092 67 1,317 160 3,776 5,093 0.882 317 1.458 3.463 2.005 67 582 160 1,922 2,503 0.882 238 1.252 2.735 1.483 67 375 160 1,067 1,442 0.882 168 1.190 2.806 1.617 67 252 160 821 1,073 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 0.882 105 1.075 1.217 0.143 67 142 160 45 187 Total 0.882 9,937 — — — $67 $42,941 $160 $56,436 $99,377 Note: pctl. = percentile; VOT = value of time; $/mi/day = dollars per mile per day. Table 6-1. Bottleneck identification for southbound segment (TMC 112N04895).

62 Estimating the Value of Truck Travel Time Reliability Figure 6-1. User costs traveling southbound. Segment values summarized at centerline mileposts; plots consider whole corridor analyzed. Figure 6-2. User costs traveling northbound. Segment values summarized at centerline mileposts; plots consider whole corridor analyzed.

Case Study 63 found to increase dramatically between Mileposts 235 and 232, with unreliability represent- ing 60–70 percent of these costs. The share of user costs due to unreliability increased from 35 percent at the north end of the corridor to more than 70 percent at the south end. The fact that unreliability can account for most user costs and vary significantly throughout a corridor suggests that it needs to be considered in the identification of bottlenecks. The five segments with the highest user costs for trucks are listed in Table 6-2. The segment that generates the highest cost is a short 480-ft segment that runs southbound at the interchange with Dean Keeton Street (TMC 112N04891). For this segment, 65 percent of the user costs were generated by unreliability. The segment that generates the second-highest costs to truck users (TMC 112-04890) is located just downstream of the top-ranked segment, which suggests that this part of the corridor sees significant congestion problems. The costs calculated in this section do not exactly represent the costs faced by trucks travers- ing this corridor, because segment-level correlations in travel time were not considered. The 95th percentile delay for any route will be lower than the summation of the 95th percentile delays of the segments along this route (as detailed in Section 5.2.3), which is implicitly what is assumed by the calculations in this section. However, this bias will be smaller in shorter corridors with homogeneous traffic conditions, as tend to occur mostly on access-controlled highways such as the one analyzed in this example. 6.2 Project Evaluation The Reliability Valuation Framework can also be used to estimate the benefits to trucking of improving traffic conditions. This section evaluates a hypothetical project that is assumed to decrease average travel times by 10 percent on a congested portion of I-35 (4 miles in length). This section describes how to calculate the cost savings to trucks (benefits) from faster average travel conditions and improved reliability. The impacts of the project on reliability are modeled by using a statistical relationship estimated previously in the literature. The percentage improve- ment in reliability is then applied to existing levels of reliability (from NPMRDS) so that the improvement can be monetized by using the VOR parameters. The calculations for estimating the impact of the project on traffic conditions are described in Table 6-3 and Table 6-4. The processed NPMRDS data were used to calculate the average travel time, the 95th percentile travel time, and the 10th percentile travel time (assumed to represent free-flow conditions) for every hour of the day. The difference between the 95th per- centile travel time and the average travel time provides the 95th percentile delay. This measure captures most directly the uncertainty that truck operators face when scheduling deliveries. Rank TMC Average Travel Time Cost ($/day-mile) Unreliability Cost ($/day-mile) Total Costs ($/day- mile) Percentage of Cost Average Travel Time Percentage of Cost Unreliability 1 112N04891 48,461 88,411 136,872 35 65 2 112-04890 47,516 84,389 131,905 36 64 3 112-04891 45,224 80,619 125,843 36 64 4 112-04889 47,225 74,155 121,380 39 61 5 112+05043 51,481 69,730 121,211 42 58 Table 6-2. Five segments with highest user costs.

64 Estimating the Value of Truck Travel Time Reliability The approach described in Section 5.3.2.1 was used to model changes in reliability as a func- tion of changes in average travel time. In this approach, the mean Travel Time Index is defined as where t– is the average travel time and tf is the free-flow travel time. As seen in Tables 6-3 and 6-4, the TTIm was calculated for the no project scenario and the project scenario, which assumes a 10 percent reduction in travel time (but not below free-flow travel time). Then, the 95th percentile Travel Time Index was calculated for each scenario by using the statistical relationship This relationship, which was estimated in a previous study that used data from Maryland, essentially indicates the 95th percentile travel time that is typically associated with differ- ent mean values for the Travel Time Index (Maryland Department of Transportation, State Highway Administration. 2018. Implementation of SHRP 2 Reliability Data and Analysis (49)TTI t t m f = 17.626 (50)95 2.872 TTI e TTIm= × −    Hour of Day NPMRDS Analysis No Project After Project Average Travel Time (min) 95th Pctl. Travel Time (min) 95th Pctl. Delay (min) 10th Pctl. Travel Time (min) Observed Modeled Modeled 95th Pctl. Delay (min) Assumed Modeled Modeled 95th Pctl. Delay (min) A B C = B–A D E = A/D F = Eq. 50 G = (F–E)*D H = max(A r/D,1) F = Eq. 50 J = (I–H) D 3.08 3.40 0.32 2.72 1.13 1.40 0.72 1.02 1.06 0.10 2.90 3.14 0.24 2.72 1.07 1.20 0.36 1.00 1.00 0.00 2.87 3.09 0.22 2.72 1.05 1.16 0.28 1.00 1.00 0.00 2.81 3.02 0.21 2.72 1.03 1.10 0.17 1.00 1.00 0.00 2.79 2.95 0.16 2.72 1.03 1.07 0.12 1.00 1.00 0.00 2.92 3.10 0.18 2.72 1.08 1.22 0.40 1.00 1.00 0.00 4.33 5.16 0.83 2.72 1.59 2.91 3.57 1.43 2.38 2.57 5.99 7.58 1.60 2.72 2.20 4.79 7.02 1.98 4.14 5.86 6.21 8.27 2.06 2.72 2.29 5.02 7.42 2.06 4.36 6.26 5.41 7.33 1.91 2.72 1.99 4.17 5.92 1.79 3.55 4.78 4.83 6.48 1.65 2.72 1.78 3.50 4.69 1.60 2.93 3.61 5.29 7.35 2.07 2.72 1.95 4.03 5.66 1.75 3.42 4.53 6.14 9.17 3.03 2.72 2.26 4.95 7.31 2.04 4.30 6.15 6.78 10.58 3.80 2.72 2.50 5.58 8.37 2.25 4.91 7.23 8.42 12.88 4.47 2.72 3.10 6.97 10.53 2.79 6.29 9.52 12.85 20.46 7.61 2.72 4.73 9.60 13.24 4.26 8.98 12.82 18.73 26.93 8.20 2.72 6.89 11.62 12.84 6.20 11.09 13.29 18.99 28.01 9.02 2.72 6.99 11.69 12.76 6.29 11.17 13.24 12.40 22.41 10.02 2.72 4.56 9.39 13.12 4.11 8.76 12.64 6.09 10.78 4.69 2.72 2.24 4.90 7.21 2.02 4.25 6.05 4.31 8.37 4.07 2.72 1.58 2.88 3.51 1.43 2.35 2.52 3.80 7.38 3.57 2.72 1.40 2.26 2.35 1.26 1.80 1.48 3.60 6.43 2.84 2.72 1.32 2.01 1.88 1.19 1.58 1.06 3.25 3.72 0.46 2.72 1.20 1.60 1.10 1.08 1.23 0.41 * * 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 Table 6-3. Project impact northbound.

Case Study 65 Tools (L38) in Maryland. Unpublished final report.). If possible, analysts are encouraged to reestimate this relationship with local data so that the results are more representative of local conditions. Figure 6-3 shows how the project would affect travel time in the northbound direction. The area between the solid curves represents the reduction in vehicle minutes of delay during an average day. The modeled 95th percentile was calculated as the difference between the average travel time (solid curves) and 95th percentile travel time (dashed curves). The modeled 95th percentile delay before and after the project were compared to estimate the percentage improvement in unreliability. The benefits to users from these improvements are calculated in Table 6-5 and Table 6-6. Improvements in average travel time and unreliability were calculated in proportion to the changes predicted by the model. The improvements were then multiplied by truck volumes and the VOT and VOR estimates to monetize these improvements over the stream of trucks using the corridor. Hour of Day NPMRDS Analysis No Project After Project Average Travel Time (min) 95th Pctl. Travel Time (min) 95th Pctl. Delay (min) 10th Pctl. Travel Time (min) Observed Modeled Modeled 95th Pctl. Delay (min) Assumed Modeled Modeled 95th Pctl. Delay (min) A B C = B–A D E = A/D F = Eq. 50 G = (F–E) D H = max(A r/D,1) F = Eq. 50 J = (I–H) D 1 4.42 6.72 2.29 3.74 1.18 1.55 1.00 1.06 1.18 0.33 2 3.98 4.24 0.26 3.74 1.06 1.19 0.33 1.00 1.00 0.00 3 3.96 4.28 0.32 3.74 1.06 1.16 0.29 1.00 1.00 0.00 4 3.90 4.14 0.24 3.74 1.04 1.12 0.21 1.00 1.00 0.00 5 3.87 4.09 0.22 3.74 1.03 1.10 0.17 1.00 1.00 0.00 6 4.02 4.29 0.27 3.74 1.07 1.22 0.39 1.00 1.00 0.00 7 4.34 4.73 0.39 3.74 1.16 1.48 0.87 1.04 1.12 0.21 8 4.41 4.78 0.38 3.74 1.18 1.54 0.97 1.06 1.17 0.30 9 4.56 5.33 0.77 3.74 1.22 1.67 1.22 1.10 1.28 0.50 10 4.77 6.33 1.56 3.74 1.27 1.85 1.57 1.15 1.44 0.80 11 5.42 9.10 3.69 3.74 1.45 2.42 2.65 1.30 1.94 1.74 12 7.07 12.82 5.75 3.74 1.89 3.85 5.33 1.70 3.25 4.21 13 9.75 16.32 6.57 3.74 2.60 5.85 8.82 2.34 5.18 7.70 14 10.84 17.98 7.14 3.74 2.90 6.54 9.89 2.61 5.85 8.83 15 14.54 22.45 7.91 3.74 3.88 8.41 12.31 3.50 7.75 11.56 16 20.00 27.92 7.92 3.74 5.34 10.30 13.46 4.81 9.70 13.29 17 26.43 35.15 8.72 3.74 7.06 11.73 12.70 6.35 11.22 13.21 18 30.08 41.23 11.14 3.74 8.04 12.33 11.67 7.23 11.85 12.54 19 21.01 32.39 11.38 3.74 5.61 10.56 13.46 5.05 9.98 13.40 20 10.03 19.37 9.34 3.74 2.68 6.04 9.12 2.41 5.36 8.00 21 6.20 12.28 6.08 3.74 1.66 3.11 3.96 1.49 2.57 2.93 22 5.69 12.53 6.84 3.74 1.52 2.66 3.10 1.37 2.16 2.15 23 6.13 18.57 12.44 3.74 1.64 3.05 3.84 1.47 2.51 2.81 24 5.69 14.82 9.14 3.74 1.52 2.66 3.10 1.37 2.16 2.15 * * * Table 6-4. Project impact southbound.

66 Estimating the Value of Truck Travel Time Reliability 0 5 10 15 20 25 30 35 0 5 10 15 20 T ra ve l T im e (m in ) Hour of the Day No Proj. Avg. Travel Time No Proj. 95th PCT Travel Time Proj. Avg. Travel Time Proj. 95th PCT Travel Time Figure 6-3. Modeled impact on northbound travel time. Hour of Day Volume (trucks/h) Improvement per Average Truck User Benefits Change in Average Travel Times (min) Change in 95th Pctl. Delay (min) VOT ($/h) Travel Time Benefits ($/day) VOR ($/h) Reliability Benefits ($/day) Total Benefits ($/day) K L =B (H/E–1) M = C J/G–1) N O = K L N/60 P Q = K M P/60 R = O+Q 1 64.1 0.308 0.276 66.6 21.9 159.9 47.2 69.1 2 48.6 0.187 0.237 66.6 10.1 159.9 30.7 40.7 3 46.6 0.148 0.224 66.6 7.7 159.9 27.8 35.5 4 62.2 0.093 0.214 66.6 6.4 159.9 35.4 41.8 5 122.4 0.070 0.161 66.6 9.4 159.9 52.5 61.9 6 253.5 0.208 0.176 66.6 58.5 159.9 118.9 177.4 7 430.2 0.433 0.232 66.6 206.8 159.9 265.7 472.5 8 499.2 0.599 0.263 66.6 331.6 159.9 349.5 681.1 9 477.8 0.621 0.321 66.6 329.3 159.9 408.6 737.9 10 475.9 0.541 0.367 66.6 286.0 159.9 465.6 751.6 11 509.9 0.483 0.381 66.6 273.3 159.9 517.3 790.6 12 555.5 0.529 0.412 66.6 325.9 159.9 609.7 935.6 13 585.6 0.614 0.479 66.6 399.4 159.9 747.2 1,146.6 14 635.1 0.678 0.518 66.6 478.0 159.9 877.2 1,355.3 15 732.2 0.842 0.429 66.6 684.2 159.9 838.1 1,522.2 16 842.0 1.285 0.240 66.6 1,201.0 159.9 538.1 1,739.0 17 887.6 1.873 –0.284 66.6 1,845.4 159.9 –671.6 1,173.8 18 746.8 1.899 –0.341 66.6 1,574.2 159.9 –677.8 896.4 19 533.2 1.240 0.370 66.6 733.6 159.9 525.5 1,259.1 20 395.3 0.609 0.752 66.6 267.2 159.9 791.8 1,059.1 21 309.8 0.431 1.153 66.6 148.1 159.9 951.9 1,100.0 22 232.1 0.380 1.329 66.6 98.0 159.9 822.2 920.2 23 164.1 0.360 1.229 66.6 65.5 159.9 537.6 603.1 24 102.0 0.325 0.293 66.6 36.8 159.9 79.5 116.4 Total 9,711.5 — — $66.6 $9,398.4 $159.9 $8,288.5 $17,686.9 * * ( * * * * Table 6-5. User benefits northbound.

Case Study 67 Overall, it was found that if reliability is not considered in the analysis—the approach pre- ferred by analysts at the moment—the benefits to freight users from the hypothetical project would be estimated at $6.0 million per year (assuming 260 workdays in the year and combining both directions of travel). However, if improvements in reliability are considered, the benefits almost double to $11.8 million per year. This large difference could make the difference between whether a project is funded or not. Hour of Day Volume (trucks/h) Improvement per Average Truck User Benefits Change in Average Travel Times (min) Change in 95th Pctl. Delay (min) VOT ($/h) Travel Time Benefits ($/day) VOR ($/h) Reliability Benefits ($/day) Total Benefits ($/day) K L = B (H/E–1) M = C (J/G–1) N O = K L N/60 P Q = K M P/60 R = O+Q 1 65.0 0.442 1.548 66.6 31.9 159.9 268.0 299.9 2 49.2 0.240 0.259 66.6 13.1 159.9 34.0 47.0 3 47.2 0.214 0.325 66.6 11.2 159.9 40.9 52.1 4 63.0 0.153 0.243 66.6 10.7 159.9 40.8 51.5 5 124.0 0.129 0.215 66.6 17.7 159.9 71.2 88.9 6 256.9 0.280 0.269 66.6 79.8 159.9 184.0 263.9 7 436.0 0.434 0.293 66.6 209.9 159.9 340.1 550.1 8 505.9 0.441 0.260 66.6 247.4 159.9 350.3 597.8 9 484.3 0.456 0.451 66.6 245.0 159.9 581.4 826.4 10 482.3 0.477 0.762 66.6 255.5 159.9 979.4 1,234.8 11 516.8 0.542 1.265 66.6 310.8 159.9 1,742.1 2,052.8 12 563.0 0.707 1.202 66.6 441.6 159.9 1,803.6 2,245.2 13 593.5 0.975 0.838 66.6 642.5 159.9 1,325.5 1,968.0 14 643.7 1.084 0.770 66.6 774.5 159.9 1,321.1 2,095.6 15 742.2 1.454 0.480 66.6 1,198.0 159.9 950.1 2,148.1 16 853.4 2.000 0.100 66.6 1,894.8 159.9 228.2 2,122.9 17 899.7 2.643 –0.349 66.6 2,639.5 159.9 –837.8 1,801.7 18 756.9 3.008 –0.840 66.6 2,527.7 159.9 –1,693.8 833.8 19 540.4 2.101 0.053 66.6 1,260.0 159.9 76.1 1,336.2 20 400.6 1.003 1.140 66.6 446.1 159.9 1,216.9 1,663.0 21 314.0 0.620 1.585 66.6 216.2 159.9 1,325.9 1,542.1 22 235.2 0.569 2.110 66.6 148.5 159.9 1,323.0 1,471.5 23 166.3 0.613 3.316 66.6 113.2 159.9 1,469.8 1,583.0 24 103.4 0.569 2.817 66.6 65.3 159.9 776.0 841.2 Total 9,843.0 — — $66.6 $13,800.8 $159.9 $13,916.8 $27,717.6 * * * ** Table 6-6. User benefits southbound.