Skip to content
Paper

Intersections and Non-Intersections: A Protocol for Identifying Pedestrian Crash Risk Locations in GIS

 
Download PDF  (1.35 MB)
Publication: International Journal of Environmental Research and Public Health
Volume: 16 (19)
Pages: 3565
Publication Date: 2019
Summary:

Intersection and non-intersection locations are commonly used as spatial units of analysis for modeling pedestrian crashes. While both location types have been previously studied, comparing results is difficult given the different data and methods used to identify crash-risk locations. In this study, a systematic and replicable protocol was developed in GIS (Geographic Information System) to create a consistent spatial unit of analysis for use in pedestrian crash modeling. Four publicly accessible datasets were used to identify unique intersection and non-intersection locations: Roadway intersection points, roadway lanes, legal speed limits, and pedestrian crash records. Two algorithms were developed and tested using five search radii (ranging from 20 to 100 m) to assess the protocol reliability. The algorithms, which were designed to identify crash-risk locations at intersection and non-intersection areas detected 87.2% of the pedestrian crash locations (r: 20 m). Agreement rates between algorithm results and the crash data were 94.1% for intersection and 98.0% for non-intersection locations, respectively. The buffer size of 20 m generally showed the highest performance in the analyses. The present protocol offered an efficient and reliable method to create spatial analysis units for pedestrian crash modeling. It provided researchers a cost-effective method to identify unique intersection and non-intersection locations. Additional search radii should be tested in future studies to refine the capture of crash-risk locations.

Authors: Haena Kim, Mingyu Kang, Anne Moudon, Linda Ng Boyle,
Recommended Citation:
Kang, Mingyu, Anne Vernez Moudon, Haena Kim, and Linda Ng Boyle. 2019. Intersections and Non-Intersections: A Protocol for Identifying Pedestrian Crash Risk Locations in GIS. International Journal of Environmental Research and Public Health 16, no. 19: 3565. https://doi.org/10.3390/ijerph16193565
Article

A Framework to Assess Pedestrian Exposure Using Personal Device Data

 
Download PDF  (1.72 MB)
Publication: Human Factors and Ergonomics Society
Volume: 66 (1)
Pages: 320 - 324
Publication Date: 2022
Summary:

Capturing pedestrian exposure is important to assess the likelihood of a pedestrian-vehicle crash. In this study, we show how data collected on pedestrians using personal electronic devices can provide insights on exposure. This paper presents a framework for capturing exposure using spatial pedestrian movements based on GPS coordinates collected from accelerometers, defined as walking bouts. The process includes extracting and cleaning the walking bouts and then merging other environmental factors. A zero-inflated negative binomial model is used to show how the data can be used to predict the likelihood of walking bouts at the intersection level. This information can be used by engineers, designers, and planners in roadway designs to enhance pedestrian safety.

Authors: Haena Kim, Grace Douglas, Linda Ng Boyle, Anne Moudon, Steve Mooney, Brian Saelens, Beth Ebel
Recommended Citation:
Douglas, G., Boyle, L. N., Kim, H., Moudon, A., Mooney, S., Saelens, B., & Ebel, B. (2022). A Framework to Assess Pedestrian Exposure Using Personal Device Data. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. https://doi.org/10.1177/1071181322661319
Paper

Bowtie Analysis without Expert Acquisition for Safety Effect Assessments of Cooperative Intelligent Transport Systems

 
Download PDF  (1.83 MB)
Publication: ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering
Publication Date: 2018
Summary:
Estimating the safety effects of emerging or future technology based on expert acquisitions is challenging because the accumulated judgment is at risk of being biased and imprecise. Therefore, this semiquantitative study proposes and demonstrates an upgraded bowtie analysis for safety effect assessments that can be performed without the need for expert acquisition. While bowtie analysis is commonly used in, for example, process engineering, it is novel in road traffic safety. Four crash case studies are completed using bowtie analysis, letting the input parameters sequentially vary over the entire range of possible expert opinions. The results suggest that only proactive safety measures estimated to decrease the probability of specific crash risk factors to at least “very improbable” can perceptibly decrease crash probability. Further, the success probability of a reactive measure must be at least “moderately probable” to reduce the probability of a serious or fatal crash by half or more. This upgraded bowtie approach allows the identification of (1) the sensitivity of the probability of a crash and its consequences to expert judgment used in the bowtie model and (2) the necessary effectiveness of a chosen safety measure allowing adequate changes in the probability of a crash and its consequences.

 

 

Authors: Dr. Ed McCormack, Ute Christine Ehlers; Eirin Olaussen Ryengm Faisal Khan, and Sören Ehlers
Recommended Citation:
Ehlers, U. C., Ryeng, E. O., McCormack, E., Khan, F., & Ehlers, S. (2018). Bowtie Analysis without Expert Acquisition for Safety Effect Assessments of Cooperative Intelligent Transport Systems. ASCE-ASME Journal of Risk and Uncertainty in Engineering Systems, Part A: Civil Engineering, 4(4), 04018036.
Report

Curbing Conflicts: Curb Allocation Change Project Report

 
Download PDF  (4.05 MB)
Publication Date: 2019
Summary:

Like many congested cities, Seattle is grappling with how best to manage the increasing use of ride-hailing services by Transportation Network Companies (TNCs) like Uber and Lyft. According to a 2018 Seattle Times analysis, TNC ridership in the Seattle region has grown to more than five times the level it was in the beginning of 2015, providing, on average, more than 91,000 rides a day in 2018. And the newspaper reports Uber and Lyft trips are heavily concentrated in the city’s densest neighborhoods, where nearly 40,000 rides a day start in ZIP codes covering downtown, Belltown, Capitol Hill and South Lake Union.

This University of Washington (UW) study focuses on a strategy to manage TNC driver stops when picking up and dropping off passengers to improve traffic flow in the South Lake Union (SLU) area. SLU is the site of the main campus for Amazon, the online retail company. The site is known to generate a large number of TNC trips, and Amazon reports high rates of ride-hailing use for employee commutes. This study also found that vehicle picking-up/dropping-off passengers make up a significant share of total vehicle activity in SLU. The center city neighborhood is characterized by multiple construction sites, slow speed limits (25 mph), and heavy vehicle and pedestrian traffic.

Broad concerns about congestion, safety, and effective curb use led to this study, conducted by researchers at the UW’s Urban Freight Lab and Sustainable Transportation Lab. Amazon specifically was concerned about scarcity of curb space where TNC drivers could legally and readily stop to pick up and drop off passengers. Without dedicated load/unload curb space, TNC vehicles stop and wait at paid parking spots, other unauthorized curb spots, or in the travel lane itself, potentially blocking or slowing traffic. To try to mitigate the impacts of passenger pick-up/drop-off activity on traffic, the city proposed a strategy of increasing passenger loading zone (PLZ) spaces while Uber and Lyft implemented a geofence, which directs their drivers and passengers to designated pick-up and drop-off locations on a block. (Normally, drivers pick up or drop off passengers at any address a rider requests via the ride-hailing app.)

By providing ample designated pick-up and drop-off spots along the curb, the thinking goes, TNC drivers would reduce the frequency with which they stop in the travel lane to pick up or drop off passengers and the time they stay stopped there. By these measures, this study’s findings show the approach was successful. But it is important to note that the strategy is not a silver bullet for solving traffic congestion—nor is it designed as such. It is also important to note that any initiative to manage use of curbs and roads (by TNCs or others) is part of a city’s broader transportation policy framework and goals.

For this study, researchers analyzed an array of data on street and curb activity along three block-faces on Boren Ave N in December 2018 and January 2019. At a minimum, data were collected during the morning and afternoon peak travel times (with some collected 24 hours a day). The research team collected data using video and sensor technology as well as in-person observation. Researchers also surveyed TNC passengers for demographic, trip-related and satisfaction data. The five Amazon buildings in the area studied house roughly 8,650 employees. Researchers collected data in three stages. Phase 1, the study baseline, was before PLZs were added and geofencing started. Phase 2 was after the new PLZs were added, expanding total PLZ curb length from 20 feet (easily filled by one to two vehicles) to 274 feet. Phase 3 was after geofencing was added to the expanded PLZs. The added PLZ spaces were open to any passenger vehicle—not just TNC vehicles—weekdays from 7am to 10am and 2pm to 7pm. (Permitted food trucks were authorized from 10am to 2pm.)

Note that while other cities can learn from this analysis, the findings apply to streets with comparable traffic speed, mix of roadway users, and street design.

The study’s main findings include:

  • A significant percentage of vehicles performing a pick-up/drop-off stop in the travel lane. Those in-lane stops appear connected to the lack of available designated curb space: Adding PLZs and geofencing increased driver compliance in stopping at the curb versus stopping in the travel lane to load and unload passengers. But it was not lack of curb space alone that influenced driver activity: Between 7 percent and 10 percent of drivers still stopped in the travel lane even when PLZs were empty. After adding PLZs and geofencing, in-lane stops fell from 20 percent to 14 percent for pick-ups and from 16 percent to 15 percent for drop-offs.
  • Adding PLZs and geofencing reduced the average amount of time drivers stopped to load and unload passengers. For example, 90 percent of drop-offs took less than 1 minute 12 seconds, 42 seconds faster than the average with the added PLZs alone.
  • While curb occupancy increased after adding PLZs and geofencing, occupancy results show the current allocation of PLZ spaces is more than what is needed to meet observed demand: Average PLZ occupancy remained under 20 percent after PLZ expansion, even during peak commute hours.
  • Vehicles picking-up/dropping-off passengers account for a significant share of total traffic volume in the study area: during peak hours the observed average percentage of vehicles performing a pick-up/drop-off with respect to the total traffic volume was 29 percent (in Phase 1), 32 percent (in Phase 2) and 39 percent (in Phase 3).
  • High volumes of pedestrians (400-500 per hour on average) cross the street at points where there was no crosswalk. Passengers picked-up/dropped-off constituted a fraction (five to seven percent) of those pedestrians, but high rates of passengers (30 to 40 percent) cross the street at non-crosswalk locations.
  • Adding PLZs and geofencing did not have a significant impact on traffic safety. Researchers found no significant change in the number of observed conflicts from baseline to the addition of PLZs and geofencing. Conflicts are situations where a vehicle, bike, or pedestrian is interrupted, forced to alter their path, or engaged in a near-miss situation. Conflicts include vehicles passing in the oncoming traffic lane. • Adding PLZs and geofencing also did not produce a significant impact on roadway travel speed.
  • Of the 116 TNC passengers surveyed in the study area:
    • Roughly 40 percent to 50 percent said their trip was work related. More than half said they used ride-hailing service at least once a week and 70 percent or more used TNC alone (versus in combination with other transportation options) to get from their origin to their destination.
    • Most responded positively to the added PLZs and geofence: 79 percent rated their pick-up satisfactory and 100 percent rated their drop-off satisfactory as compared to 72 percent and 89 percent in the baseline.
    • Nearly half said they would have taken transit and one-third would have walked if ride-hailing was not available.
    • 40 percent requested a shared TNC vehicle in Phase 1 and 47 percent in Phase 3.

The study suggests that while vehicles picking-up/dropping-off passengers account for a significant share of traffic volume in SLU, they are not the primary cause of congestion. Myriad factors impact neighborhood congestion, including high vehicle volume overall and bottlenecks moving out of the neighborhood onto regional arterials. As researchers observed in the afternoon peak, these bottlenecks cause spillbacks onto local streets. Amazon garages exit vehicles onto streets that then feed into these clogged arterials.

Regarding traffic safety in SLU, this study was not designed to assess whether TNC driver behavior on average is safer or less safe than that of other vehicles. It is important to understand the safety and speed findings in the context of the SLU traffic environment. Drivers tend to drive at relatively slow speeds, navigating around high pedestrian and jaywalking volumes, and seem relatively comfortable stopping in the middle of the street for short periods of time. Due to the nature of area traffic, this seems to have relatively little impact on other drivers. Drivers appear to anticipate both this behavior and the high volumes of vehicles moving onto/off the curb and into/out of driveways and alleys.

Whether the strategy this study analyzed is recommended depends on a city’s transportation goals and approach. The researchers found the increased PLZ allocation and geofencing strategy worked in that it improved driver compliance, reduced dwell times, and boosted TNC user satisfaction. However, this may encourage commuters to use TNC. The passenger survey clearly shows that TNC service is attracting passengers who would have otherwise walked or used transit. While in the short term the increased PLZs and geofencing had a positive effect on traffic, if this induces TNC demand, there could be larger, more negative long-term consequences. If the end goal is to reduce traffic congestion, measures to reduce—rather than encourage—TNC and passenger car use as the predominant mode of commuting will yield the most substantial benefits.


In the news:

Geekwire: As Uber and Lyft pick-ups and drop-offs clog traffic, new study calls load zones a move in right direction

The Seattle Times: Seattle Uber and Lyft drivers often stop in the street to pick up or drop off riders. Here’s a way to reduce that.

Recommended Citation:
Goodchild, Anne. Giacomo dalla Chiara. Jose Luis Machado. Andisheh Ranjbari. (2019) Curb Allocation Change Project.
Paper

A Description of Fatal Bicycle Truck Accidents in the United States: 2000 to 2010

Publication: Transportation Research Board 95th Annual Meeting
Volume: 16-5911
Publication Date: 2016
Summary:

Bicycling is being encouraged across the US and the world as a low-impact, environmentally friendly mode of transportation. In the US, many states and cities, especially cities facing congestion issues, are encouraging cycling as an alternative to automobiles. However, as cities grow and consumption increases, freight traffic in cities will increase as well, leading to higher amounts of interactions between cyclists and trucks. This paper will describe where and how accidents between cyclists and trucks occur. From 2000 to 2010, 807 bicyclists were killed the United States in accidents involving trucks. In 2009, trucks accounted for 9.5% of fatal bicycle accidents, despite trucks only accounting for 4.5% of registered vehicles. The typical fatal bike-truck accident happens in an urban area on an arterial street with a speed limit of 35 or 45 mph. It is about equally likely to occur mid-block or at an intersection. Most accidents involved trucks going straight (56%), and right-turning trucks were involved in a much larger number of accidents (24%) than left turning trucks (7%). Methods such as providing bicycle lanes, or even physically separated bicycle tracks, will not be sufficient to address bicycle-truck collisions, as a significant number of accidents (49%) occur in intersections or are intersection related. Cities with a higher mode-share of bicycling had a lower rate of bicycle-truck fatality accidents.

Authors: Dr. Anne Goodchild, Jerome Drescher
Recommended Citation:
Drescher, Jerome and Anne Goodchild. (2016), "A Description of Fatal Bicycle Truck Accidents in the United States: 2000 to 2010," Accepted for presentation at the 95th Transportation Research Board Annual Meeting, Washington DC, January 10-14. [Paper # 16-5911]
Paper

Assessing the Safety Effects of Cooperative Intelligent Transport Systems: A Bowtie Analysis Approach

Publication: Accident Analysis and Prevention
Volume: 99 (A)
Pages: 125-141
Publication Date: 2017
Summary:

The safety effects of cooperative intelligent transport systems (C-ITS) are mostly unknown and associated with uncertainties, because these systems represent emerging technology. This study proposes a bowtie analysis as a conceptual framework for evaluating the safety effect of cooperative intelligent transport systems. These seek to prevent road traffic accidents or mitigate their consequences. Under the assumption of the potential occurrence of a particular single-vehicle accident, three case studies demonstrate the application of the bowtie analysis approach in road traffic safety. The approach utilizes exemplary expert estimates and knowledge from literature on the probability of the occurrence of accident risk factors and of the success of safety measures. Fuzzy set theory is applied to handle uncertainty in expert knowledge. Based on this approach, a useful tool is developed to estimate the effects of safety-related cooperative intelligent transport systems in terms of the expected change in accident occurrence and consequence probability.

Authors: Dr. Ed McCormack, Ute Christine Ehlers, Eirin Olaussen Ryeng, Faisal Khan, Sören Ehlers
Recommended Citation:
Ehlers, Ute Christine, Eirin Olaussen Ryeng, Edward McCormack, Faisal Khan, and Sören Ehlers. "Assessing the safety effects of cooperative intelligent transport systems: A bowtie analysis approach." Accident Analysis & Prevention 99 (2017): 125-141.