Region 10 University Transportation Center

PacTrans

Region 10 University Transportation Center

Safety and Economic Implications of Time-of-Day Shifts in Truck Freight Movement

July 2015-June 2016

(1)Collaborators and Affiliations

Lead Institution

Oregon State University

Civil and Construction Engineering

PI:Dr. Salvador Hernandez

Assistant Professor - Civil and Construction Engineering

Collaborating Institution

Washington State University

School of Economic Sciences

Freight Policy Transportation Institute

PI:Dr. Jeremy L. Sage

Research Assistant Professor- School of Economic Sciences

Assistant Director – Freight Policy Transportation Institute

/ 509.335.5536

(2)Project Goal

This proposal seeks to examine the supposed link between time-of-day travelling for large trucks in urban areas and the ability to effectively reduce the incidents of serious crashes involving these trucks and other users of the roadway. This project directly addresses the PacTrans theme of Safe Travel on Mixed-Use Roads. Specifically, this proposed project seeks to:

• Identify potential safety gains generated through a shift of large truck traffic away from peak volume periods;
• Evaluate the economic factors contributing to the current time-of-day travel patterns of large trucks;
• Evaluate the costs and benefits of time-of-day travel shifts to lower volume periods. Specifically included are the anticipated safety benefits of such a shift;
• Where warranted, identify potential policy or logistic recommendations that may be considered by private and public stakeholders.

(3)Relevancy of Institutional Partnership

Evaluation and forecasting of the potential outcomes generated by transportation investment and management strategies is inherently a multidisciplinary venture. The collaborating partners on this proposal have a long and successful history of melding transportation economics, planning, and engineering to form actionable results of high utility for our transportation agency partners. We found our research and output on the merits that it must meet three primary objectives:

Results must be Economically Viable for the private enterprise and transportation user perspective;

Results must be Socially Acceptablefrom the perspective of the managing public agencies and the populations/users they represent; and

Results must be Technically Feasiblefrom the perspective of the professional engineer and/or planner.

The research team assembled for this project provides the tools to sufficiently meet these objectives and deliver them to stakeholders.

(4)Research Background and Problem Statement

Problem Statement

As transportation infrastructure budgets are squeezed throughout the country, the capacity to build one’s way out of congestion becomes unrealistic, and often economically inefficient, thus incentivizing transportation agencies and fleet operators to identify alternative means to alleviate the impacts of congestion. Throughout many urban areas, reoccurring congestion is highly and predictably correlated to time of day (e.g. rush hour). Building on the knowledge gained from the literature, including the work of both PI’s[1], one may reasonably assume that by moving a portion of the large trucks out of high volume travel times of day may produce both safety and economic benefits. However, backing up this suggestion has currently has two major data limitations. First, while data readily shows that fewer large truck involved incidents occur outside of peak travel times, it as of yet fails to fully account for exposure and the ability to generate relative crash rates by hour-of-day. Second, large trucks are frequently found in high volume traffic for a reason; they are attempting to meet contractual delivery windows in a highly orchestrated logistic system. Not all trucks on the roadway may have the flexibility to alter delivery times in a manner that avoids high density traffic. Incentive structures thus may be needed to induce a time-of-day shift.

Background

Urban Goods Delivery

In a meta-analysis of urban freight studies, Allen et al[2], identify major portions of goods delivery taking place in the morning hours; before, through and just after the a.m. peak travel times. Allen et al[3] suggest this phenomenon to be attributable in part to the desire by the retailer for morning deliveries so that they can begin their working day ensuring shelves are stocked prior to their higher volume hours. McKinnon[4] suggests that the peak for deliveries to food retailers ranges from 05:00 to 09:00, while that for non-food delivery and other business establishments is from 08:00 to 12:00.

Allen et al2 further identify variation and even uncertainty as to who within the supply chain controls the time at which deliveries are made. Suggested control ranges from a majority of respondents believing the receiving establishment had most control in one reviewed survey, to a majority of respondents believing it is the supplier with the most control over timing. In each reviewed study, only small portions believed the carrier to have much influence over delivery time. Independent of the delivery time decision maker, the preferred delivery times indeed place many large trucks in high traffic volume time-of-day travel.

Congestion is Economically Inefficient

The economic vitality and livability of the Pacific Northwest substantially depends on reliable, responsible, and sustainable transportation. Maintaining the transportation system at a level that allows for the safe, efficient movement of freight is an important component of this sustainable system. In addition to the direct cost of accidents, congestion has been demonstrated to cause freight-dependent businesses, such as manufacturing, retail and wholesale trade, agriculture, construction, and timber/wood products, to operate less efficiently by increasing the amount of time for each truck trip and increasing the time that trucks (and drivers) spend in traffic;thereby, spending time in an unproductive manner[5]. Taylor et al. estimated that a 20% increase in congestion experienced by commercial trucks would result in over $14billion of increased operating costs to Washington’s freight-dependent industries. Since many freight industries have the ability to pass on their rising transportation costs in the form of higher cost goods, consumers and service industries may feel the biggest impacts from increasing congestion. When multiplied into economic impacts, this translates into losses of over 27,250 jobs (0.70% of statewide total) and $3.3billion (0.51% of statewide total) in economic output (2011 dollars).

Time-of-Day Travel and Incidence of Serious Crashes

Crashes involving large trucks are a costly experience for all parties involved; even non-injury causing incidents may average $15,114 in property damage, and fatal incidents easily top $3 million[6]In 2013, the U.S. Department of Transportation adopted a $9.1 million value per life saved for use in regulatory analyses[7].A 2007 publication by the Federal Motor Carrier Safety Administration (FMCSA) identified traffic flow interruption (congestion, previous crash) as the second leading associated factor, among hundreds collected, for serious crashes involving large trucks.

Belzer et al[8] survey 1,000 truck drivers in the Great Lakes region and figure out that drivers work 22% of their hours and 21% of the miles between midnight and 6:00 a.m. Sandberg et al. (2011) identify both sleepiness and darkness to adversely affect lateral positioning, in which a driver who is more tiresome and who maneuvers a vehicle at night is more likely to position the truck to the center of the road. Specifically, Massie et al[9] report a peak in fatal crash involvements from 4 a.m. to 7 a.m. and a peak from 3 a.m. to 7 a.m. for less severe involvements and driver fatigue-related non-fatal crashes. However, Dingus et al[10] find out that the critical incident rates, in which a truck driver exhibits a lapse in performance or potentially hazardous driving behavior as manifested in naturalistic studies, are the highest in the afternoon (12:00 to 17:59) than night time (22:00 to 03:59) and morning (04:00 to 11:59). While the results are contrary to the widespread hypothesis that driving is the riskiest at late night due to circadian-rhythm induced fatigue, the authors are cautious about any limitation on nighttime driving because the interaction with heavier traffic at daytime may posit a greater impact on the occurrence of critical incidents than late night with less traffic on the road. Blower and Campbell[11] reconcile the seemingly conflicting results: although less than 10% of the accidents occur at late night, crashes occurring during that period are more severe with a higher chance of injury and causality than the rest of the day, because about one-third of the other drivers involved in fatal crashes with a long-haul truck are either fatigued or drunk.

From a crash severity modeling perspective, Islam & Hernandez[12]used a random parameter ordered probit specification to study the impact of human, vehicleand road environmental factors on large truck crash injury severityusing the National Automotive Sampling System General EstimatedSystem (NASS-GES) database from 2005 to 2008. In contrast tothe insight obtained from other literature, the likelihood of lowerinjury severity was higher when crashes occurred in darkerconditions.Furthermore, Islam & Hernandez[13]developed mixed logit models for truck crashes inTexas using data from the Texas Peace Officer’s Crash Reportsdatabase for the year 2006–2010. The likelihood of fatal,incapacitating and possible injuries was found to reduce during the afternoon peak period due to congestion effects. The likelihoodof fatal and incapacitating injuries increased during dark conditions. Pahukula et al.[14] developed a framework to analyze time of day crashes involving large truck in urban areas using the Texas Peace Officer’s Crash Reportsdatabase for the year 2006–2010. In their study, it was determined that separate random parameters logit models were warranted to five times of day periods. The results of the individual models demonstrate considerable differences among the five time periods. These differences show that the different time periods do in fact havedifferent contributing factors to each injury severity further highlighting the importance of examiningcrashes based on time of day. Traffic flow, light conditions, surface conditions, time of year and percentage of trucks on the road were found as key differences between the time periods.

Incentivizing Large trucks out of High Volume Traffic

Despite the known factors leading to serious crash incidents, and the economic inefficiencies generated as a result of congestion, commuters and many large trucks alike continue to travel at these times. This observation is a result of value pricing. In other words the private cost of using a particular roadway at a given time is often significantly lower than the social cost, as uses do not directly consider the impact of their travel on congestion and its subsequent impacts. According Holguín-Veras[15], carrier costs would be 28% cheaper for deliveries made in the off-peak hours than the regular hours. To incentivize delivery during off-peak hours to ease the congestion problem, Holguín-Veras[16]propose a tax-subsidy scheme, in which the toll levied on peak time road users are transferred to off-peak receivers to compensate for their incremental costs of overtime compensation, electricity, technology and security investments, because the decision on delivery schedules rests in the hands of the receivers, who stand on the upside of the bargaining table, but not the carriers. This is corroborated by Vilain & Wolfrom[17] who suggest that firms operating under the constraints driven by that of inventory costs minimization incur a higher cost of freight delivery after the start of business day that outweighs any current congestion surcharge levied on peak-hour delivery. Subsequently, the questions of facilitating a time-of-day shift in large truck movement to aide in the alleviation of congestion and its negative impacts becomes one of correctly negotiating the value pricing decisions of users such that safety may be enhanced.

(5)Approach

Proposed Study

The proposed study will be a twofold exploration drawing on the relative strengths of the researchers at the partnering institutions. Phase one of the study will identify potential data sources on large truck crash rates and risks by time-of-day.This includes crash distributions, crash harm distributions, mileage exposure distributions, time-of-driving distributions, and any other data metrics representing either a numerator or denominator in an overall time-of-day risk assessment.From these sources and developed models, the study will derive time-of-day crash risk functions to inform carrier dispatching and other operational decisions.These functions may also serve Federal, State, and local government, as well as the motoring public.

Concurrent with Phase one, Phase two will develop the potential economic implications of a time-of-day shift in distribution channels throughout the Pacific Northwest. These implications will be developed through a survey of freight dependent industries, including shippers, carriers, and customers. The survey focus will include the development of a tool to estimate a proportion range of potential truck traffic that may feasibly adjust its time-of-day operations, and the expected cost to accomplish such a shift. These costs will then be compared to the savings garnered through serious crash reduction as well as economic efficiency gains. Culminating from this comparison will be an estimation of the net economic benefits and impacts of time-of-day shifts within the truck freight transportation sector.

5(a)Phase I

Phase I (a) – Extensive Literature Review

The overall objective of this task is to review the most recent projects and literature of successful paradigms and lessons learned by other state agencies and other research activities with regards to large truck time-of-day crash risk analyses. In addition, review and synthesize information on the various safety policies and initiatives undertaken by local, state, and national governments. This will provide the research team with a list of policies and initiatives that can be studied for possible recommendation in Phase II.

Phase I(b) – Collect and Analyze Data

This task will seek to collect and inventory current data (e.g., Washington and Oregon) related to large truck crashes, truck traffic, historical crash data analysis (will include a hot spot analysis to uncover hazard clusters both spatially and temporally), and synthesize and analyze the information collected from part (a) to identify best practices and practice yet to be deployed. The analysis will consist of the standard descriptive statistics (e.g., means, standard deviations, mileage exposure distributions, time-of-driving distributions, and synthesize, etc.). For specific sites, detailed police reports will be requested to better understand the circumstances of crashes around rest areas and to uncover any factors that may be contributing to the specific incident. The goal of the statistical analysis is to calculate statistics that will reveal characteristics of the data and identify the key variables that contribute to large truck crash rates and risks by time-of-day.

Phase I(c) – Large Truck Crash Rates and Risks by Time-of-Day Analyses

The objective of this task is to assess the safety risks associated with large truck involved crashes by time-of-day. Advanced statistical analysis coupled with crash severity and crash frequency analyses will be used to provide insights related to the impact ofcrash factors and the complex interactions of these factors on crashes by time-of-day. Additionally, in this study we will also utilize the “Crash Harm” metric. Crash harm is defined in Knipling[18] as, “A quantitative measure of the combined human and material loss from traffic crashes based on economic valuation. Using crash “harm” as a metric, permits objective comparisons across different vehicle types, crash types, crash severity levels, and ways of assessing risk.

Phase I(d) – Output preparation for Phase II

Phase II will apply the findings of Phase I into a benefit-cost analysis, revealing the relative merit of potentially feasible time-of-day shifts, thus the conclusion of Phase I generates the necessary outputs for Phase II. These outputs, in the form of various severity and risk metrics, will inform the discrete choice models developed in Phase II(d) that are subsequently input into the benefit-cost models.

5(b)Phase II

Phase II(a) – Extensive Literature Review

The research team will conduct an extensive literature review of the conditions guiding the current time-of-day decisions of freight dependent industries. Focus of this review will be applied to the cumulative decisions made by the three phases of the supply chains; shippers, carriers and customers, along with any intermediary actors. Additional emphasis will be on the current policy basis, if any, for efforts to direct large trucks to off peak driving.

Phase II(b) – Scenario Development

Prior to survey implementation, and guided by the findings of the literature review and Phase I, the research team will establish a series of potential scenarios to be proposed to survey respondents. These scenarios will elicit stated preferences by the respondent in response to policy changes that alter the private costs experienced by the user. Preliminarily identified scenarios include: