State and Local Impacts of Automated Freight Transportation Systems (2023)

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.

156 Balancing Public Benefits and Public Costs Some potential public benefits from deployment of FAVs have been hypothesized. For OTR FAVs, the anticipated public benefits include the following: • Improved traffic safety beyond that offered by ASS/ADAS systems. • Lower freight transportation costs through labor reductions and more efficient operation. • Reduced emissions through more efficient operations. • Reduced traffic congestion due to better V2V cooperation and fewer incidents. Offsetting public costs could include the following: • Public infrastructure and IT investment. • VMT increases. • Job loss. For SADRs, the anticipated public benefits include the following: • Reduced street traffic and curb use. • Additional no-touch delivery options. • Economic development. Offsetting costs could include the following: • Public infrastructure and IT investment. • Sidewalk congestion. • Job loss. For UAVs, anticipated public benefits could include the following: • Reduced street traffic and curb use. • Improved access to remote locations. • Additional no-touch delivery options. Offsetting costs could include: • Additional ATC. • Adverse privacy, noise, and nuisance impacts. • Job loss. To date, none of these expected benefits have been quantified or shown to be significant. Moreover, there is no reliable information available on what level of OTR FAV deployment is necessary to achieve those benefits or when that level might be reached. The interdependency of FAV and PAV deployment and benefits complicates benefits assessments further. A P P E N D I X D Conceptual Benefit-Cost Analysis

Conceptual Benefit-Cost Analysis 157   There are thus two conceptual challenges to benefit-cost comparisons besides the general lack of quantitative estimates. • The need to distinguish the incremental benefits and costs of progressive automation levels. • The need to distinguish the incremental benefits and costs of FAV deployment versus PAV deployment for OTR applications. In addition to exercising regulatory authority to manage the public right-of-way, public enti- ties plan, invest, and often construct and operate needed infrastructure to promote equity in transportation and enable competitiveness in the jurisdiction. Related regulatory levers may be adapted and adopted to promote and manage AVs to improve the public outcome (e.g., safety, congestion reduction, system efficiency) and provide support for the marketplace. Public responsibilities most frequently include land-use planning, zoning, regulating, licensing, and investing or maintaining physical and information technology infrastructures. Many agencies responsible for these tasks use BCA as a tool to help determine the social and economic ben- efits of transportation infrastructure investments. A BCA standardizes comparisons of different types of benefits over time. The BCA can provide a more systematic perspective, not necessarily constrained by localized impacts. The more objective nature of BCAs also helps public agencies clarify their goals and communicate their perspectives to multiple stakeholder groups. Mathematically, a BCA helps an agency answer the question: “Are the benefits of this project worth the cost to the agency and its stakeholders?” The BCA can also assist an agency in identi- fying the ideal time to undertake a project, comparing alternative projects, and identifying the project with the highest beneficial impact on the jurisdiction. Transportation infrastructure project BCAs also accommodate the changes in travel demand, including collateral or induced changes resulting from the potential project’s improvements to the network. Levels of Automation. While “vehicle automation” and “autonomous vehicles” are often used interchangeably, they do not mean the same thing, and they have markedly different implica- tions for public benefits and costs. The SAE automation levels generally focus on the driver’s or operator’s role, which becomes supervisory in Level 3 and optional in Levels 4 and 5. ASS/ADAS and related OTR safety systems, however, can be fully present and engaged at Level 2, as can V2V, V2I, and V2X communications. The presumed incremental safety benefits of Levels 3-5 are therefore based on reducing the driver role and the potential for driver error with ASS/ADAS and communications fully engaged. Since Level 2 automation requires an engaged driver at the controls, there would be minimal if any public costs associated with Level 2 OTR FAV deployment. Moreover, costs for improved striping and signage, and investments in communications infrastructure should yield public safety (and potentially congestion) benefits for both PAVs and FAVs at Levels 1 and 2. It is instruc- tive to note that one automaker’s self-driving system is intended for use at Level 2 and that accidents have occurred when drivers use it at Levels 3-5. The public cost of accommodating Level 3-5 OTR FAVs could be significantly higher if those vehicles require higher quality and more extensive pavement markings, signage, communica- tions infrastructure, or roadway modifications. The likelihood of extensive Level 3-5 operations by the freight industry depends on the availability of tangible, incremental cost, safety, or operational benefits that have not been demonstrated or quantified to date. Unoccupied Level 4-5 operations would trigger incremental public sector costs to accommo- date unoccupied vehicles. These costs could include measures to enable unoccupied vehicles to negotiate work zones and incidents; means for law enforcement and first responders to control Level 4-5 vehicles or communicate with operators; and measures to cope with stranded or dis- engaged vehicles. These same costs, however, would be triggered by Level 4-5 passenger vehicles

158 State and Local Impacts of Automated Freight Transportation Systems repositioning between trips, parking, or otherwise operating unoccupied. Unoccupied Level 4-5 PAV operation would become common if self-driving taxi, ride-sharing, or shuttle services are widely deployed. As of mid-2021, FAV development across OTR, SADR, and UAV modes (and AV develop- ment overall) is focused on operating within existing public infrastructure without relying on V2V or V2I communications. It is thus possible, even likely, that OTR FAVs will be deployed and yield public safety benefits with no incremental public cost. FAVs versus PAV Benefits and Costs. Whether widespread deployment of FAVs precedes or follows widespread deployment of PAVs is, as of mid-2021, a matter of speculation. FAVs are in regular Level 3 operation on a few pilot highway routes, with manual operation at origin and destination. PAVs are widely operated at Level 2 and in a few pilot services at Level 3. One com- pany is reportedly operating Level 4 taxis in one service. There are no known Level 3 (hands-off auto mation) trucks in regular use. Regardless of the relative pace of technology deployment, freight trucks constitute a small portion of registered vehicles and a small part of typical traffic streams. It thus appears likely that public benefits and costs from PAVs will emerge before analogous benefits and costs for FAVs. Conceptual Spreadsheet Model The purpose of a spreadsheet model is to facilitate a simplified BCA of proposed or required infrastructure investment by focusing on transport efficiency benefits that are likely to accrue to the infrastructure users after project completion. Public agencies will likely need to make decisions regarding infrastructure investments or accommodations to enable AVs for both passengers and freight. These decisions will often incur some public cost, which must be weighed against the anticipated public benefits. Potential public investments regarding FAVs may be relatively straightforward, including projects such as improvement of IT infrastructure (e.g., installing fiber-optic cable), updating legacy infrastruc- ture to current standards, and renewing/maintaining road markings, including striping and signage. New costs for accommodating FAVs may include digitization of traffic light signals and changing work zone marking practices. It is essential to note that based on research to date, these costs are unlikely to be higher for FAVs than PAVs. That is, it is unlikely that public infrastruc- ture investment will be higher for FAVs than what will be provided for PAVs. The conceptual spreadsheet tool uses output from existing travel demand modeling. The tool can be used to prioritize and assess anticipated costs and benefits of projects over time. Note that this tool is not intended to make a final decision on investing in infrastructure to enable AVs but to inform the broader decision-making process. This BCA tool provides a simplified analysis linking outputs from an agency’s travel demand model to build and no-build scenarios in both the present and future to compute a benefit- cost ratio and a net present value. This tool is intended for use in estimating public benefits of AV-related investment projects, based on the public entity’s travel demand model. The tool assesses the anticipated quantified project benefits, provides input for choosing among alterna- tive expenditures, and identifies projects for prioritization. Input Variables For OTR applications, it is essential to include travel demand model output for all vehicles to reflect the full safety benefit of FAV deployment. Specifically, the output variables from the

Conceptual Benefit-Cost Analysis 159   travel demand model required for the model include, for both base and forecast years, those listed in Table 24. Model Parameters A conceptual spreadsheet model would use numerous parameters, with examples shown in Table 25. These could appear as default values in some cases and would be customized by sub- stituting the agency’s or jurisdiction’s values. Most of these parameters are familiar to those who have previously conducted BCAs for trans- portation infrastructure projects. The new and unique parameters of this FAV model are AV penetration for automobiles and trucks, and a modal shift from automobiles and trucks to AVs. Output This spreadsheet tool would generate output values for the following categories (see Table 26). • Emissions-derived cost savings. • Safety improvements. • Travel-time reduction. • Vehicle-operation-related cost savings. The output will also include two project cost-effectiveness measures: the benefit-cost ratio and the net present value (NPV). “The benefit-cost ratio is calculated by dividing the total discounted benefits by total discounted costs. The net present value is the difference between the discounted net present value of the benefits and the discounted present value of costs. [Generally], a positive NPV indicates that benefits exceed costs” (82). Parameter Detail Unit of Analysis Default Value Default Value Source % of Capital Costs (Project) to FAV % of O&M Costs (Project) to FAV Capital Costs (Project) Discount Rate O&M Costs (Project) Project Timing Base Year Year (Date) Project Timing Forecast Year Year (Date) Safety Improvement/Accident Reduction Fatalities % Reduction Safety Improvement/Accident Reduction Injuries % Reduction Safety Improvement/Accident Reduction Property Damage Only % Reduction Truck % VHT Base Year VHT Forecast Year VMT Base Year VMT Forecast Year Year of Model Input Model Data Date Year (Date) O&M: operations and maintenance. VHT: vehicle hours traveled. Table 24. Analysis variables.

GHG: greenhouse gas. Parameter Project Total Annual Average Incremental Public Benefits Safety Improvements Emissions Reductions GHG Reductions Travel Time Savings Infrastructure Capital Savings Infrastructure Maintenance Savings Total Public Benefits Incremental Public Costs Increased Regulation/ Enforcement Infrastructure Capital Cost Infrastructure Maintenance Cost Total Public Benefits Benefit-Cost Ratio Net Public Benefits Net Present Value Table 26. Conceptual spreadsheet output. Parameter Detail Unit Default Value Default Value Source Accident Costs Fatalities Per Accident Accident Costs Injuries Per Accident Accident Costs Property Damage Only Per Accident Accident Rate Fatalities Per Million Miles Accident Rate Injuries Per Million Miles Accident Rate Property Damage Only Per Million Miles AV Penetration Automobile % of All VMT AV Penetration Truck % of All VMT Cost of Emissions CO Per Gram Cost of Emissions NO(x) Per Gram Cost of Emissions PM(10) Per Gram Cost of Emissions SO(2) Per Gram Cost of Emissions VOC Per Gram Cost of Emissions GHG Per Gram Fuel Cost Automobile Per Gallon Fuel Cost Truck Per Gallon Induced Travel Automobile % of All VMT Induced Travel Truck % of All VMT Modal Shift: SADRs Automobile % of Auto VMT replaced by SADR Modal Shift: SADRs Truck % of Truck VMT replaced by SADR Modal Shift: UAVs Automobile % of Auto VMT replaced by UAV Modal Shift: UAVs Truck % of Truck VMT replaced by UAV Non-Fuel Cost Automobile Per Vehicle Mile Non-Fuel Cost Truck Per Vehicle Mile Value of Time Automobile Per Person Hour Value of Time Truck Per Person Hour CO: carbon monoxide. GHG: greenhouse gas. NOx: nitrogen oxides. PM: particulate matter. SO2: sulfur dioxide. VOC: volatile organic compounds. Table 25. Examples of input parameters.

Conceptual Benefit-Cost Analysis 161   Market Penetration Scenarios Market and fleet penetration by FAVs (and by PAVs in parallel) are pivotal factors in public sector planning for FAV deployment and BCA of infrastructure investments. The potential public safety and congestion relief benefits of PAV and FAV operation depend on how much of the vehicle fleet has and uses the technology. The costs and benefits of public infrastructure and IT investment likewise depend on how many AVs there are and where they will operate. OTR FAVs of any automation level will enter the truck fleet either as replacements for exist- ing trucks and their capacity or as new capacity. The durability of commercial trucks and, in particular, the durability of diesel engines leads to long working lives for commercial trucks com- pared with passenger autos. Fleet turnover, and thus deployment of new technology, therefore tends to be slower in freight fleets. The gradual influx of electric and hybrid electric commercial vehicles may slow fleet turnover further if, as expected, electric vehicles have longer lives. The research team developed an illustrative market penetration scenario for PAVs and FAVs. Key assumptions based on available data include the following: • An average life of 12 years for autos, light trucks, and commercial trucks yielding an average annual fleet replacement rate of 8.3%. • Annual fleet increase rates of 0.7% for autos and light trucks and 2.3% for commercial trucks, based on Bureau of Transportation Statistics (BTS) fleet data for 2011 to 2019. • An initial introduction rate similar to that evidenced by electric vehicles, which rose from negligible sales in 2010 to about 2% of annual sales in 2020. • An AV annual replacement rate rising from 0% in 2025 to the fleet norm of 8.3% in 2037. • The first commercial sales of PAVs and FAVs in 2025. • An S-shaped sales penetration curve for PAVs and FAVs (Figure 32), with rapid penetration in 2035 to 2042, followed by slower penetration to a plateau of about 90% in 2050. This scenario corresponds to widespread adoption after favorable results of small-scale intro- duction, but residual purchase and use of non-AV vehicles where AVs are not cost-effective or Source: Research team analysis. Figure 32. Conceptual market penetration scenario.

162 State and Local Impacts of Automated Freight Transportation Systems not desirable for other reasons. An attractive business case for Level 3-5 adoption due to signifi- cant labor savings would likely accelerate fleet turnover and FAV deployment. Conversely, lack of significant savings or offsetting costs would reduce the incentive to replace existing trucks with FAVs. Table 27 displays the scenario data. As shown there, and in Figure 32, this scenario yields market and fleet penetration of about 65% for PAVs and 77% for FAVs by 2050. Because pas- senger vehicles outnumber commercial vehicles, the overall fleet would be about 50% by 2047 and 70% by 2050. It must be noted that this scenario would apply only to the introduction of vehicles with given ADS capabilities and does not dictate where or how often those capabilities would be used. If the scenario reflected the introduction of Level 4 ADS technology, use of the systems would still be limited by the ODD. If the Level 4 ODD were restricted to limited-access freeways, the scenario would indicate what percentage of the fleet could operate autonomously only on those free- ways. On the other hand, if the scenario represented the penetration of advanced ASS/ADAS technology, that technology would be applicable anywhere. The introduction and penetration rates in the scenario suggest that public benefits from improved safety, particularly from reduced traffic congestion, may be slow to develop. As of 2014, the average age of autos and light trucks on U.S. roads was 11.3 years, and that number rose to about 12 years in 2021. Moreover, in 2012, 9.8% of U.S. vehicles were 21 years old or older, and 42% were 11 to 20 years old. There is no body of empirical knowledge on the safety benefits of individual AVs operating in mixed traffic flow versus AVs operating in more nearly homogenous AV traffic. Data to Support the Decision Framework In addition to an agency’s existing travel demand model and forecasts, many datasets can be helpful when using the proposed decision framework. These datasets include data categories such as built environment, emergency, infrastructure, navigation, traffic control systems, and 2024 2025 2030 2035 2040 2045 2050 AV Sales Share 0.0% 0.2% 1.2% 3.0% 22.8% 81.1% 86.7% Auto & Light Truck 201,437,146 202,885,539 210,285,223 217,954,790 225,904,084 234,143,305 242,683,029 Sales 18,234,821 18,365,935 19,035,781 19,730,059 20,449,657 21,195,502 21,968,548 New AVs - 36,732 228,429 591,902 4,658,688 17,179,254 19,053,454 AV Replacement Rate 0.0% 3.5% 6.9% 8.3% 8.3% 8.3% AV Replacements - 7,932 41,104 388,224 1,431,605 1,587,788 AV Fleet - 36,732 771,682 2,571,598 13,604,038 72,036,129 157,569,319 AV Share 0.0% 0.0% 0.4% 1.2% 6.0% 30.8% 64.9% Commercial Truck 81,198,383 83,047,806 92,946,292 104,024,581 116,423,293 130,299,811 145,830,274 Sales 8,615,955 8,812,197 9,862,524 11,038,041 12,353,667 13,826,104 15,474,040 New Avs - 17,624 118,350 331,141 2,814,320 11,206,253 13,420,728 AV Replacement Rate 0.0% 3.5% 6.9% 8.3% 8.3% 8.3% AV Replacements - 4,109 22,996 234,527 933,854 1,118,394 AV Fleet - 17,624 399,190 1,442,400 8,574,291 49,151,300 112,925,593 AV Share 0.0% 0.0% 0.4% 1.4% 7.4% 37.7% 77.4% Total Vehicle Fleet 282,635,529 285,933,344 303,231,515 321,979,371 342,327,377 364,443,116 388,513,303 AV Fleet - 54,356 1,170,872 4,013,999 22,178,328 121,187,429 270,494,912 AV Share 0.0% 0.0% 0.4% 1.2% 6.5% 33.3% 69.6% Table 27. Conceptual market penetration approach.

Conceptual Benefit-Cost Analysis 163   real-time travel time. Sources of data for the built environment include data from vehicle cameras, street cameras, satellite imagery, and traffic control cameras. Data on emergency conditions are often publicly available and include weather information, providing input into a real-time hazard or closure alerts. Infrastructure-related data can be from public or private sources. Examples include curb space information, fueling and charging stations, on-/off-street parking, and road sign inventories. Navigation datasets include both public and private sources such as public inventory of roadways and private network company data. Traffic control systems are typically operated by the public agency and include data from ATMS. Real-time travel-time datasets are typically owned by private companies. The available sources for this data type include Bluetooth data, cell phone data, in-vehicle navigation systems, and other data companies. Public agencies would need to study the continuing stream of information on supply chain trends, particularly those related to automation. This information can be aggregated from industry associations and industry trade publications. Relevant industry associations include but are not limited to the Council of Supply Chain Management Professionals (CSCMP), National Industry Transportation League (NITL), Retail Industry Leaders Association (RILA), and Warehouse Education and Research Council (WERC). Additionally, industry trade publica- tions will continue to provide the latest information on AV evolution and market penetration. These publications include but are not limited to DC Velocity, Development Quarterly, Inbound Logistics, Logistics Management, Supply Chain Quarterly, Supply Chain Management Review, and Transport Topics.