Source-linked AI summary
Resilience and efficiency in transportation networks
Alexander A. Ganin, Maksim Kitsak, Dayton Marchese, Jeffrey M. Keisler, Thomas Seager, Igor Linkov
TL;DR
Urban roadway planning emphasizes efficiency, while analytic support for resilience investments remains limited. This paper models efficiency and disruption resilience across 40 U.S. urban road networks, finding that efficient cities can be fragile while inefficient cities can be resilient. The findings support considering both characteristics jointly in roadway investment decisions.
Problem
Transportation planning emphasizes normal-condition efficiency, while practical analytic support for evaluating roadway resilience to disruption remains limited.
Method
The study builds and calibrates a roadway traffic model for 40 major U.S. cities using network geometry, commuting patterns, and observed commuter-delay data.
Results
Urban-area resilience differs persistently across stress levels, with some cities efficient in normal conditions becoming sharply less resilient under disruption.
Takeaways & Limitations
Roadway project selection should consider efficiency and resilience jointly because reducing normal-condition delays may leave cities vulnerable under mild stress.
Takeaways & Limitations
Because the model considers only roadway transit, its results are unlikely to be informative for New York City, where nonroad transportation options are extensive.
Abstract
from arXiv · showhide
Urban transportation systems are vulnerable to congestion, accidents, weather, special events, and other costly delays. Whereas typical policy responses prioritize reduction of delays under normal conditions to improve the efficiency of urban road systems, analytic support for investments that improve resilience (defined as system recovery from additional disruptions) is still scarce. In this effort, we represent paved roads as a transportation network by mapping intersections to nodes and road segments between the intersections to links. We built road networks for 40 of the urban areas defined by the U.S. Census Bureau. We developed and calibrated a model to evaluate traffic delays using link loads. The loads may be regarded as traffic-based centrality measures, estimating the number of individuals using corresponding road segments. Efficiency was estimated as the average annual delay per peak-period auto commuter, and modeled results were found to be close to observed data, with the notable exception of New York City. Resilience was estimated as the change in efficiency resulting from roadway disruptions and was found to vary between cities, with increased delays due to a 5% random loss of road linkages ranging from 9.5% in Los Angeles to 56.0% in San Francisco. The results demonstrate that many urban road systems that operate inefficiently under normal conditions are nevertheless resilient to disruption, whereas some more efficient cities are more fragile. The implication is that resilience, not just efficiency, should be considered explicitly in roadway project selection and justify investment opportunities related to disaster and other disruptions.
INTRODUCTION
Roadway design and evaluation have traditionally prioritized efficiency, while resilience lacks a shared, practical framework for guiding transportation infrastructure decisions. This study examines the relationship between efficiency and resilience in road networks across 40 major U.S. cities using calibrated delay modeling and simulated roadway disruptions.
- Motivation: Existing roadway standards primarily evaluate networks by efficient vehicle movement, including shortest or fastest routes and congestion minimization.Efficiency is the main criterion for modeling road networks and considering design alternatives.
- Problem: Transportation resilience lacks a shared definition suitable for guiding roadway design, operation, and reconstruction.Definitions variously emphasize adaptive response, maintaining or restoring service, and preparing for, absorbing, recovering from, and adapting to disturbances.
- Problem: Existing resilience evaluations use indicators such as total traffic delay and economic loss but may rely on uncertain performance data and omit unquantifiable indicators.The passage identifies framework development and quantification methods as central efforts in transportation resilience research.
- Contribution: The study analyzes interconnections between resilience and efficiency in road networks across 40 major U.S. cities.It develops an urban roadway efficiency model calibrated with observed annual delay per peak-period auto commuter and applies it to all 40 cities.
- Method: Random roadway disruptions are modeled to recalculate expected delays and measure each city’s sensitivity to lost roadway linkages.The results are intended to inform roadway improvement proposals that maintain efficiency under stress conditions.
METHODS
The study develops urban roadway efficiency and resilience measures from Census Bureau, OpenStreetMap, and commuter-pattern data. Its model uses readily available public data rather than requiring particular datasets that may be impractical to obtain.
- METHODS: The researchers defined urban boundaries, constructed road networks, and evaluated city population density using Census Bureau and OpenStreetMap datasets.These data supported assessment of commuter patterns used to measure roadway efficiency and resilience.
- METHODS: The model assessed roadway efficiency and resilience using commuter patterns derived from readily available public data.The cited data sources were Census Bureau and OpenStreetMap datasets.
- METHODS: The approach differs from alternatives based on percolation, cascading failures, human mobility patterns, queueing, or historical traffic data.These alternative approaches are reviewed in the Supplementary Materials.
Geospatial boundaries and population density
The study defined transportation-network boundaries from U.S. Census Bureau urban-area data, approximating each urban-area polygon and including roadways within 40 km. Population near each intersection was estimated by overlaying Voronoi cells with Census Tract data.
- Geospatial boundaries: Urban-area boundaries used U.S. Census Bureau geospatial data, with manually simplified polygons and roadways included within 40 km (25 miles).The boundaries represented densely developed residential, commercial, and other nonresidential areas.
- Population density: Population near each intersection i was estimated from Census Tract data by partitioning the map into Voronoi cells centered at intersections.The population estimate was denoted Ni for each cell.
- Population density: Nt denotes the population of Census Tract t, while Pi and Pt denote the corresponding Voronoi-cell and tract polygons.These quantities were used to estimate intersection-adjacent population from tract data.
Transportation model
The transportation model used a gravity-based framework to estimate commuting flows, assigned commuters to shortest-time paths, and calculated road-segment loads as traffic-based centrality measures. It then estimated cumulative commuter delay from free-flow and actual speeds using the Daganzo traffic-speed model.
- Commuting-flow generation: The model generated commuting patterns with the classical gravity model, which is widely used in U.S. metropolitan and statewide travel-demand models.Destination-choice models require detailed data that are often unavailable at large scale.
- Commuting-flow generation: Commuter flow from origin region o to destination region d was assumed proportional to destination population N_d and dependent on origin–destination distance x_od through P(x_od).These assumptions define the fraction of individuals commuting between each origin and destination region.
- Route assignment: Drivers were assumed to minimize travel time, so commute paths and x_od were calculated using shortest-time paths based on inferred free-flow speeds.Nodes farther than 30 km from the urban-area boundary were used only as destinations when estimating commuters traveling away from the urban area.
- Road-segment loads: Commuter loads L_ij on road segments were treated as traffic-based centrality measures estimating the number of individuals using each segment during peak periods.The load calculation used q_od(ij), a binary indicator of whether link ij lies on the shortest path from o to d.
- Delay estimation: Cumulative commuter time loss was calculated from segment free-flow speed V_ij, actual speed v_ij, segment length l_ij, signal correction l_0, and coefficient b.Actual traffic speeds were derived with the Daganzo model, incorporating minimum traffic speed v_min, vehicle-size correction v_veh, and coefficient a.
Efficiency and resilience metrics
The study measures efficiency using average annual delay per peak-period auto commuter and resilience using additional traffic delays under roadway stress. Lower delays indicate higher efficiency or resilience, while the resilience measure emphasizes network topology rather than recovery resources or coordination.
- Efficiency: Efficiency is measured as average annual delay per peak-period auto commuter, with lower delay indicating higher efficiency.The study uses delays themselves rather than inverse or negative transformations to avoid ambiguity.
- Resilience: Resilience is operationalized as the change in traffic delays relative to stress modeled as loss or impairment of roadway linkages.The analysis focuses on topological features of cities rather than available recovery resources.
- Resilience: Lower additional delay corresponds to higher resilience, so resilience is quantified through additional delays.A complete resilience assessment would also examine recovery resources and coordination among relevant authorities.
RESULTS · Efficiency
The calibrated traffic model estimated urban travel delays using three parameters and showed strong agreement with observed delays in calibration cities, while validation-city estimates were also significantly correlated with actual delays. Model performance supported its use for comparing modeled and observed transportation conditions across urban areas.
- Efficiency: Three parameters—proportionality coefficient a, minimum speed vmin, and finite vehicle size correction vveh—summarized the traffic-delay model.The model estimated total delay for commuters in a suburban area or average delay per commuter.
- Efficiency: 9 km/hour and 5 km/hour were used for vveh and vmin, respectively, while a was calibrated against real annual-average-delay data.These fixed parameter values were paired with calibration of a to match observed delay data.
- Efficiency: 20 urban areas were used for calibration and 20 for validation, dividing the 40 urban areas into equally sized groups.The calibration and validation groups each contained half of the modeled urban areas.
- Efficiency: −0.01 to 0.83 was the R-squared range across the 20 calibration urban areas.This calibration analysis supported setting model parameters a and b.
- Efficiency: a = 4.30 × 10^4 hour−1 and b = 10.59 were the resulting model parameter values, corresponding to a Pearson coefficient of 0.91 (P = 2.17 × 10−8).The reported Pearson coefficient describes the calibration relationship.
- Efficiency: R = 0.63 and P = 3.00 × 10−3 showed significant correlation between estimated and actual travel delays in the 20 validation urban areas.The validation result was reported as evidence supporting the transportation model.
Resilience
Resilience was evaluated by randomly disrupting road segments and recalculating traffic delays, with delay increases varying substantially across cities. Some cities that are efficient under normal conditions nevertheless have low resilience under disruption.
- Resilience: Each stress simulation randomly affects a finite fraction r of road segments, with failure probability proportional to segment length.Failed segments receive free-flow speeds of 1 km/hour before loads and traffic delays are recalculated; statistics are collected over 20 realizations.
- Resilience: Traffic delays grow rapidly as the affected-segment fraction r increases and saturate as r approaches 1.At saturation, all routes move at 1 km/hour.
- Resilience: Salt Lake City, UT, is the most resilient urban transportation network, while Washington, DC, is the least resilient among the 40 modeled metropolitan areas.
- Resilience: Some cities with low normal-condition delays nevertheless exhibit low resilience, marked by sharp traffic-delay increases under stress.Virginia Beach, VA; Providence, RI; and Jacksonville, FL are examples.
- Resilience: 5% link disruption was used to compare resilience responses with normal-condition efficiency across the 40 modeled urban areas.Resilience response denotes additional delays due to 5% link disruption.
DISCUSSION
The discussion finds that roadway resilience and efficiency are distinct, complementary characteristics that should be considered jointly. Resilience differences persist across stress levels, and prioritizing normal-condition delay reductions alone may leave cities vulnerable to disruption.
- Resilience and efficiency are not correlated and should be considered jointly as complementary characteristics of roadway networks.
- Resilience differences among urban areas persist under mild, medium, and widespread stress, with rank ordering generally insensitive to disruption severity.San Francisco is the most fragile at stress levels r < 20%, but is surpassed by Boston and Washington, DC.
- Investments that reduce traffic delays under normal conditions may nevertheless leave urban areas vulnerable to delays under mild stress.The discussion identifies crashes, construction, special events, extreme weather, equipment malfunctions, and deliberate attack as inevitable stressors.
- 63 hours of delay are reduced by public transit for peak-period auto commuters in New York, compared with 23 hours in Chicago and less than 20 hours elsewhere.The roadway-only model does not account for these substitutes for roadway transportation, helping explain New York City's exceptional position.
- Few analytic tools guide investments in roadway resilience despite the long-lasting consequences of expensive infrastructure decisions.The discussion links decisions about rights-of-way, alignment, crossing, and access made over decades to traffic data observed today.
SUPPLEMENTARY MATERIALS
Supplementary materials provide additional methodological details on transportation-network modeling, including network construction, population assignment, travel likelihood, traffic-speed estimation, calibration, ramp-speed sensitivity, and disruption-related delay.
- Supplementary methods: The supplementary material covers alternative transportation-modeling approaches and the mapping of OSM Foundation shapefiles to network nodes and links.It also documents population assignment and the distance factor governing travel likelihood between nodes.
- Supplementary methods: Additional methods address traffic-speed estimation from vehicle density, model calibration, sensitivity to ramp speeds, and delay as a function of disruption severity.
(12), e1701079. 3 Sci Adv
This section provides the paper’s DOI and a link to its bibliography. The bibliography lists 35 cited articles, including one freely accessible article.
- The paper’s DOI is 10.1126/sciadv.1701079.
- The bibliography is available at the Science Advances article page.
- 35 articles are cited, including 1 freely accessible article.