Cost-effective mitigation of urban congestion with adaptive traffic signal control

doi:10.21203/rs.3.rs-3176883/v1

Download PDF

Article

Cost-effective mitigation of urban congestion with adaptive traffic signal control

https://doi.org/10.21203/rs.3.rs-3176883/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Urban congestion is a widespread issue with detrimental effects on urban efficiency, energy consumption, and pollution levels. Traditional approaches to mitigating congestion, such as increasing transport infrastructures or reducing travel demands, can be costly or inequitable for residents. Adaptive traffic signal control is a less understood method that may potentially improve intersection efficiency and reduce congestion without changing travel demands or transport infrastructures. By analyzing the top 100 congested cities in China, here we show that adaptive traffic signal control reduces trip time by 11% and 8% during peak and off-peak hours, respectively, compared to pretimed traffic signals. This reduction in congestion also results in a decrease in fuel consumption and CO2 emissions of 12 million tonnes (Mt) and 40 Mt, respectively. Although implementing adaptive traffic signal control requires an annual cost of US$1.5 billion, the resulting societal benefits, mainly fuel savings and CO2 reduction, amount to US$40.4 billion per year for the 100 congested cities in China. To encourage the adoption of this method, a central system is necessary to facilitate traffic data sharing across sectors, which will require policy and technological innovation.

Physical sciences/Engineering/Civil engineering

Social science/Business and management

In developing countries such as China, urban populations and private car ownership have increased sharply in recent decades¹. This has led to severe traffic congestion in mega-cities². During congested periods, travel time across city downtown areas can even double. Traffic congestion also leads to increased fuel consumption and air pollution^3,4, and can negatively impact the physical and mental health of urban residents. Congestion poses additional challenges to several global sustainable development goals (SDGs), including SDG 11 “Sustainable Cities and Communities”⁵, SDG 9 “Industry, Innovation and Infrastructure”, and SDG 13 “Climate Action”. To mitigate urban congestion, mega-cities such as Beijing and Shanghai have implemented measures to limit residents’ access to downtown areas through controls on annual vehicle licensing or restrictions on the number of trips during peak hours⁶. However, these travel restriction policies can exacerbate inequality as wealthier residents can afford multiple cars while poorer ones cannot. In recent years, fast-growing cities like Hangzhou have sought to mitigate congestion by building new freeways and subways. While these projects improve transport infrastructure supply, they are often costly. These measures do not increase traffic efficiency but rather reduce trip demand or increase the supply of transport infrastructure.

Road traffic is typically managed by traffic police departments using traffic signals at intersections. These signals are crucial for maintaining traffic system efficiency as they control multiple traffic streams^7,8. Most intersection traffic signals are pre-timed due to their low implementation cost and lack of need for real-time data. However, this lack of adaptivity to changing traffic patterns can result in reduced intersection efficiency. In contrast, adaptive signal timing methods aim to match the demand for green times with the supply for specific traffic streams using real-time traffic data^9–12. This adaptive method has been recently adopted by several cities such as Nanchang and Hangzhou, facilitated by the City Brain^13,14 and other smart city platforms. Hangzhou has been reported using camera data for adaptive traffic signal control at more than 1,000 intersections, reducing trip delay by more than 15%¹⁵. Using similar measures, Nanchang canceled its traffic restriction policy implemented over 10 years but has maintained an acceptable level of congestion due to its advanced traffic control methods. Although these explorative practices have received positive feedback, there is still a lack of systematic understanding of this method. Is adaptive traffic control effective for all cities? Considering both benefits and costs, which cities can expect higher cost-effectiveness? From a policy and technical perspective, how can the implementation of the adaptive method be enabled?

In response, we conducted a series of simulated experiments on the top 100 congested cities in China. We used road network data from OpenStreetMap^16,17 and travel speed data from Gaode Map (a transport service company). We also examined how traffic congestion mitigation potential related to socioeconomic and infrastructure indicators across the 100 cities, using multivariate linear regression and additional data such as population, GDP and road network length. To assess the cost-effectiveness of this congestion mitigation measure, we quantified the benefits and costs, and compared the trip time, fuel consumption and transport emission benefits with the corresponding implementation costs. Finally, we proposed feasible data sharing solutions and supportive policies to enable the implementation of adaptive traffic signal control in different cities.

How adaptive traffic signal control mitigates congestion?

Traffic signals play an essential role in mitigating congestion at intersections by assigning green times to traffic streams (Fig. 1a). Signal timing methods can be categorized into two groups: pre-timed and adaptive. The pre-timed method provides a fixed amount of green time to each traffic stream over a set period, such as an hour or a peak hour period¹⁸. Real-time traffic data is not required for this method. In contrast, the adaptive method assigns green time to each traffic stream based on anticipated arrivals during a signal cycle. As arrivals change from cycle to cycle, green times are updated using real-time traffic data. While the pre-timed method is widely implemented, the adaptive method is relatively new and has been reported to outperform pre-timed methods at intersections and arterial streets.

To illustrate how the adaptive method can reduce congestion, we use a typical four-leg intersection as an example. At an intersection, a traffic phase is defined as the green and yellow intervals assigned to pairwise non-conflicting traffic streams in a signal cycle (Fig. 1a). Efficient phase timing requires appropriate setup of green time and signal cycle lengths for all phases. We adopted Webster’s method^19,20 to compute the signal cycle length and updated the lengths of green times based on traffic volume ratios in the previous signal cycle (see Method). Our proposed adaptive method minimizes demand-supply gaps of green times by continuously adjusting their lengths based on arrival patterns of traffic streams (Fig. 1b). It is reasonable to expect that overall system efficiency could be improved if more intersection signals were implemented with our proposed adaptive method in a city.

We conducted a series of simulation-based experiments to explore the relationship between system performance and the implementation rate of our proposed adaptive method. The business-as-usual (BAU) scenario serves as a baseline and assumes that all intersection signals in a city are pre-timed. Using Shanghai as an example, we found that both travel speed and trip time reduction increase logarithmically with the implementation rate during both peak and off-peak hours (Extended Data Fig. 1). Interestingly, implementing our proposed adaptive method at the first 20% of intersections reduced trip time by 8% during peak hours, while the reduction was negligible for the last 20% of intersections. The implementation order was from high to low traffic volume intersections, which is generally optimal since high volume intersections are often more critical²¹. These findings suggest that in practice, we should prioritize high traffic volume intersections when implementing our proposed adaptive method in a city.

Congestion mitigation potential across China

Our results show that the potential for congestion mitigation is universal and not limited to a few mega-cities like Shanghai. We studied the top 100 congested cities in China ranked by Gaode in 2022. By altering all traffic signals in each city from pretimed to adaptive, we found that 97 of the 100 cities could expect a non-trivial change in average speed (greater than 1%), with 49 cities expecting an increase of more than 10% during peak hours. We also calculated the per-hour trip time change for each city and found that trip time would be reduced by 11% (79,695 hours) during peak hours across all 100 cities. During off-peak hours, trip time reduction was about 8% (42,080 hours) (Extended Data Fig. 2). The bottom three cities (Jining, Yili, and Urumqi) could expect negligible changes in speed (less than 1%) (Fig. 2c). These results approximate the country-level congestion mitigation potential due to the trivial benefit to cities other than the top 97 in China. Geographically, we found that cities with high mitigation potential were mainly coastal or provincial capitals, which are important for economic development in China (Fig. 2a-c).

We ranked 100 cities based on reductions in trip time and found that Shanghai, Beijing, Harbin, Qingdao, and Shenzhen were the top 5 cities. Shanghai alone could save up to 6,581 hours (an 18% reduction) in total trip time during peak hours. This reduction is significant, as each trip could save up to 9 minutes on average (Fig. 2d). Interestingly, our analysis revealed a truncated power relationship between trip time change and city rank. Specifically, we observed a power function for the top 50 cities and a linear function for the bottom 50 cities (Fig. 2d). This finding suggests that the potential for congestion mitigation diminishes rapidly for high-ranked cities but more slowly for low-ranked cities. For example, the trip time reduction in Beijing (4,339 hours) would be 66% of that in Shanghai, whereas the difference between Shenzhen and Shenyang (ranked 5th and 10th, respectively) is only 10%. We hypothesize that the differences among low-ranked cities are primarily due to variations in travel demand influenced by socioeconomic factors such as population. In contrast, for high-ranked cities, the sharp differences in trip time reduction are jointly determined by both travel demand and speed change.

To explore the infrastructure and socioeconomic determinants of congestion mitigation potential in 100 cities, we conducted a multivariate linear regression analysis (Table 1). Our findings reveal that baseline BAU speed, road length and population negatively impact speed improvement. Specifically, a 1% increase in baseline speed, logarithms of road length and population corresponded to reductions of approximately 1.1%, 3.3%, and 2.3% in potential speed improvement, respectively. This is consistent with the expectation that higher baseline speeds leave less room for improvement and that higher population and road length, indicating greater travel demand towards downtown areas, make it more challenging to mitigate traffic congestion. In contrast, cities with more straight streets and higher gross domestic product based on purchasing power parity (GDP(PPP)) exhibited greater potential for speed improvement. In summary, our results suggest that implementing the adaptive method would be most beneficial in cities with high congestion levels, strong economic development, and well-designed streets.

Table 1

**Regression analysis between speed improvement and city indicators.**
	Coefficient	P
BAU speed (km/h)	-1.129	< 0.001
Log[road length (m)]	-3.330	0.076
Log[Population]	-2.295	0.298
Street straightness	106.770	< 0.001
Log[GDP (PPP)]	2.113	0.079
Pseudo/adjusted R²	0.486
Method	OLS
This table presents the result of a multivariate linear regression analysis of speed improvement (%) against baseline BAU speed, road network length, population, street straightness (ratio of straight-line distances between adjacent street nodes to street lengths)³¹, and GDP (PPP). Negative and positive coefficients indicate that city indicators have negative and positive impacts on congestion mitigation potential, respectively. The P-value indicates the significance of each city indicator; the smaller the value, the more significant it is. The method used for linear regression is ordinary least squares (OLS).

To explore the infrastructure and socioeconomic determinants of congestion mitigation potential in 100 cities, we conducted a multivariate linear regression analysis (Table 1). Our findings reveal that baseline BAU speed, road length and population negatively impact speed improvement. Specifically, a 1% increase in baseline speed, logarithms of road length and population corresponded to reductions of approximately 1.1%, 3.3%, and 2.3% in potential speed improvement, respectively. This is consistent with the expectation that higher baseline speeds leave less room for improvement and that higher population and road length, indicating greater travel demand towards downtown areas, make it more challenging to mitigate traffic congestion. In contrast, cities with more straight streets and higher gross domestic product based on purchasing power parity (GDP (PPP)) exhibited greater potential for speed improvement. In summary, our results suggest that implementing the adaptive method would be most beneficial in cities with high congestion levels, strong economic development, and well-designed streets.

Reduction of fuel consumption and emission

Road traffic accounts for 11.9% of global greenhouse gas (GHG) emissions^22,23 from the combustion of petrol and diesel by cars, trucks, lorries, motorcycles, and buses. In China, road traffic emissions have been growing at an average annual rate of 20.5% between 2016 and 2019³. Therefore, the road traffic sector will contribute to China’s carbon neutrality goal by 2060⁴. According to the Urban Centre Database^24,25, the annual road traffic carbon dioxide (CO₂) emissions of 100 cities were approximately 667 Mt in 2015. Our results indicate that overall CO₂ emissions could be reduced by 40 Mt (a 6% reduction) each year (Fig. 3b), equivalent to a savings of 12 Mt of gasoline (Fig. 3c). Given that improvements in cities other than the top 100 congested ones are negligible, this result can be considered an approximation of the country-level impact.

At the city level, changes in transport emissions are jointly determined by the baseline BAU emissions and speed changes. In 2015, annual BAU CO₂ emissions across the 100 cities ranged from 0.8 Mt (Sanya) to 28.8 Mt (Chongqing). With the implementation of our adaptive method, reductions in CO₂ emissions varied from 0.9 Kt (Urumqi) to 3,081 Kt (Shanghai), indicating a sharp difference in emission reduction compared to BAU CO₂ emissions. After ranking cities by emission reduction, we found that the top 10 cities accounted for 33.5% of the overall CO₂ emission reduction, whereas the bottom 10 cities accounted for only 1.0%. In terms of percentage change, Shanghai would experience a 13.4% reduction in annual emissions. Cities such as Shenzhen (ranked 25th) and Zibo (ranked 50th) would also see reductions of 4.9% and 7.0% in annual emissions, respectively. Thus, our proposed adaptive method could contribute to China’s carbon neutrality goals, particularly if implemented in high-ranked cities such as Shanghai, Beijing, and Shenzhen.

Assuming no change in trip demand and route choice, the potential for emission reduction primarily arises from increases in the travel speed of individual vehicles. In urban areas, vehicles often operate at inefficient speeds, particularly under congested traffic conditions. Therefore, according to a speed-based emission model²⁶, even slight improvements in speed could result in significant reductions in emissions. In addition to CO₂, automobiles also emit other air pollutants such as nitrogen oxides (NO_x), ammonia (NH₃), and volatile organic compounds (VOC) from their tailpipes, particularly during congestion. These pollutants are important precursors of fine particulate matter (PM2.5) and near-surface ozone (O₃) pollution. Thus, reducing congestion with the adaptive method would also benefit air pollution mitigation and contribute to improvements in human health.

Cost-benefit analysis

To achieve congestion mitigation, our proposed adaptive method must be implemented considering both its costs and potential societal benefits (Fig. 4a-d). The primary implementation cost is traffic detection, as our method requires real-time traffic volume data. Approximately US$1.48 billion would be required to achieve the aforementioned potential goals in these 100 cities (Fig. 4e), including one-time installation and annual operation and maintenance costs. We estimated the one-time installation cost of traffic detectors using 145 recently reported field cases. Assuming an average lifespan of 71 months for traffic detectors²⁷, the annual installation cost would be US$1.31 billion yr^− 1. The operation and maintenance cost of traffic detection is estimated at US$0.17 billion yr^− 1 using the same reported field cases. Since implementation cost is proportional to the number of intersections, it is unsurprising that mega-cities account for high costs due to their large built-up areas and dense street networks.

In contrast to these implementation costs, the benefits of implementing our adaptive method include reductions in trip time, fuel consumption, and transport emissions. Benefits from trip time savings are estimated at US$1.56 billion yr^− 1 across the 100 cities. These benefits are calculated by investigating travelers’ willingness to pay for trip time savings, using results from a stated preference (SP) survey conducted in Nanjing²⁸. The increase in travel speed leads to reductions in fuel consumption and CO₂ emissions, with benefits estimated at US$20.32 billion yr^− 1 and US$18.54 billion yr^− 1, respectively. In total, implementing our proposed adaptive method could result in societal benefits of approximately US$40.42 billion yr^− 1 across the 100 cities, approximately 27.3 times the total implementation cost, suggesting it would be cost-beneficial to achieve ambitious targets for mitigating urban congestion. For individual cities, a natural question arises: is it worth implementing the proposed method considering both benefits and costs? We calculated the benefit-cost ratio for all 100 cities (Fig. 4f). Our results indicate that more than half of the 100 cities would expect a benefit-cost ratio higher than 30, and a quarter would expect a ratio higher than 50. The high benefit-cost ratios for individual cities mainly result from fuel savings and CO₂ emission reductions since trip time benefits and implementation costs are comparable in quantity.

Feasibility and supportive policy

Although the adaptive method presents a cost-effective measure for congestion mitigation, its feasibility relies on real-time traffic detection. The primary types of traffic detectors are inductive-loop detectors, surveillance cameras, and floating car detectors. Inductive-loop detectors are the most consistently accurate in terms of traffic volume; however, their maintenance cost is often high due to occasional (road works) or constant damage (heavy-duty vehicles) in China. Surveillance cameras are primarily used for traffic violation detection and security monitoring. Their implementation cost is low compared to inductive-loop detectors, but their performance may degrade under bad weather and nighttime conditions. Floating car detection data are collected from multiple sources, including navigation map apps and GPS trackers installed on hazmat trucks, buses, and taxis. Detailed vehicle trajectory information can additionally contribute to traffic signal coordination. However, floating car detection cannot directly measure traffic volume unless the penetration rate of floating cars is high enough.

Considering the strengths and weaknesses of the three types of detectors, connecting multiple data sources is recommended. From a performance perspective, this improves the robustness of the adaptive method through cross-validation across multiple data sources. From an implementation perspective, it enhances the applicability of the proposed adaptive method given the availability of alternative traffic data sources. For instance, floating car data can serve as substitutes for intersections without loop detectors or cameras. However, in practice, traffic detectors (and traffic signals) are often owned or governed by multiple public and private sectors (Extended Data Table 1).

We categorize solutions for connecting multiple data sources into three tiers: (1) Departmental solutions using data from a single department (often the Traffic Police Department), with low implementation barriers and technical thresholds; (2) Cross-sector solutions involving multiple public sectors such as the Traffic Police Department, Public Security Bureau, and Highway Agency, with medium implementation barriers and technical thresholds, requiring policy support to promote application; (3) Public-private cooperative solutions further involving private sectors such as taxi and ride-hailing companies, with high implementation barriers and technical thresholds, requiring policy and social support to promote application. This classification allows us to recognize increasing implementation challenges from lower to higher tiers. The lower the tier, the easier it is for city governments to adopt our proposed adaptive method, as it requires less cooperation among sectors. The higher the tier, the higher the cost-effectiveness due to the increased likelihood of finding low-cost data source substitutions. When considering other congestion mitigation measures such as multimodal transport planning and congestion pricing, the higher the tier, the better its future applicability, as it supports a more complete and detailed description of the entire traffic system.

Recent smart city practices such as City Brain^13–15 have demonstrated that a central system enables cross-sector data sharing and decision-making in traffic management. To facilitate high-tier solutions, future policies should support the development of a central system to enable traffic data sharing among sectors (Fig. 5). With a central system, a city could feasibly manage its traffic system as a whole and implement our proposed adaptive method along an optimal pathway (i.e., from heavy traffic intersections to light traffic ones). In addition to data-sharing, a central system also provides the computing capability to process the considerable real-time and historical data collected by multiple sectors and from different traffic detectors. Currently, each sector may maintain an isolated data management system; a central system could help link these systems and lay the foundation for so-called networked governance^29,30. Networked governance enables both data and knowledge sharing among sectors, facilitating the implementation (and evaluation) of our proposed method and other congestion mitigation measures in cities.

Data sources. We used data from OpenStreetMap to obtain road networks in cities (available at https://www.openstreetmap.org/). The urban area boundary, population, GDP, baseline transport CO₂ emission data were retrieved from the GHS Urban Centre Database^24,25 in 2015 (available at https://ghsl.jrc.ec.europa.eu/ghs_stat_ucdb2015mt_r2019a.php). We manually adjusted the errors in boundaries of some cities like Guangzhou and Hangzhou based on OpenStreetMap. The top 100 congested cities in China and the associated speed data were retrieved from Gaode during the period from 2022/11/02 to 2022/11/08 (available at https://report.amap.com/diagnosis/index.do).

To estimate the gasoline consumption, we used the result of CO₂ emission from gasoline from the Engineering Toolbox (available at https://www.engineeringtoolbox.com/co2-emission-fuels-d_1085.html). The price of gasoline is retrieved from Trading Economics on 2022/11/08 (available at https://tradingeconomics.com/commodity/gasoline). We collected data from 145 field projects that were recorded in the Intelligent Transportation Systems Joint Program to estimate implementation cost of adaptive traffic signal control (available at https://www.itskrs.its.dot.gov/).

Experiment setup. We used a series of simulation-based experiments to evaluate the performance of pretimed and adaptive traffic signal timing methods. Road network query, trip generation, and simulation environment calibration are the three main tasks for experiment setup (Extended data Fig. 3a).

Road network query. We used the OSMnx³¹ package to query road networks given a polygon boundary. When querying road networks, we set the “network_type” as the “drive” option to include drivable public streets (but not service roads) for vehicles in a city. The road network was then divided into 8×6 identical traffic zones for generating synthetic trips.

Synthetic trip generation. We used a two-step procedure to generate synthetic trips within a city downtown area: (1) trip demand estimation, and (2) trip route assignment. We used the gravity model^32,33 to estimate the trip demands (T) between pairwise traffic zones over the simulation period. A uniform random choice model was used to select intersection locations as origin and destination for each trip. With such a model, each intersection within a specified traffic zone was selected at an equal probability.

We used a two-step procedure to generate synthetic trips within a city downtown area: (1) trip demand estimation, and (2) trip route assignment. We used the gravity model to estimate the trip demands ($T$) between pairwise traffic zones over the simulation period. A uniform random choice model was used to select intersection locations as origin and destination for each trip. With such a model, each intersection within a specified traffic zone was selected at an equal probability.

$$T=\frac{{m}_{i}^{\alpha }\bullet {m}_{j}^{\beta }}{\gamma \bullet {r}_{ij}^{\delta }}$$

where $m$ denotes the population of a traffic zone. Since the raw population data is at city-scale, we estimated $m$ by multiplying overall population with the proportion of road length in a traffic zone. The subscripts $i,j$ are used to denote the origin and destination traffic zones, respectively. And $r$ denotes Euclidean distance between center-points of origin and destination traffic zones. The parameters ${\alpha },{\beta },{\gamma },{\delta }$ are estimated for each city at the simulation environment calibration step.

To assign trips ($T$) to travel routes, we first divide the overall simulation period into 10-minute intervals. In each interval, we use the expected travel time based shortest algorithm (Dijkstra’s algorithm) for trip assignment³⁴. The BPR function^35,36 is used to estimate the expected travel time ($tt$) of a road segment.

$$tt={tt}_{f}\bullet [1+0.15\bullet (\frac{x}{1000}{)}^{4}]$$

where, ${tt}_{f}$ is the time cost for passing a road segment when travelling at speed limit. And $x$ denotes the assigned trips to a road segment in previous time intervals, at beginning of simulation period, $x$ is set to be 0.

Simulation environment calibration. We input the road network and generated synthetic trips into CBEngine^37,38 for simulating traffic in a city. CBEngine is a traffic simulator that allows for city-scale traffic simulation. To narrow the gap between simulation and real-world, we used a trial-and-error procedure to iteratively adjust trip demand parameters (i.e., parameter-${\alpha },{\beta },{\gamma },{\delta }$) to minimize the difference between observed and simulated average speed at city-scale (Extended Data Fig. 3b). The observed speed data were retrieved from Gaode in 2022 during peak and off-peak hours.

Traffic signal control method. We used traffic volume data to compute the two signal timing variables: (1) cycle length and (2) length of green interval for each traffic phase. We used the typical phase setting as shown in Fig. 1a for all four-leg intersections. For three-leg intersections, one of the four signal phases is dropped based on the intersection configuration. Here, we used Webster’s method^20,39 to calculate signal cycle length. This method requires estimation of saturation flow rates, critical flow ratios and lost times of each traffic phase. Here, we use subscript $i$ to denote a traffic phase. For any traffic phase, saturation flow rate (${s}_{i}$) is estimated as the maximum volume passed an unsignalized highway, averaging over 10 rounds of simulation. To compute critical flow ratio (${y}_{i}$), we need to identify the critical traffic stream with maximum observed volume (${q}_{i}$), in a traffic phase. The critical flow ratio is then calculated as follows,

$${y}_{i}=\frac{{q}_{i}}{{s}_{i}}$$

In each traffic phase, a portion of beginning of each green interval (start-up lost time) and a portion of each yellow interval (clearance lost time) is not usable by vehicles, the sum of these two periods compromises the lost time ${l}_{i}$ for a phase. Here, the lost time for each phase is assume to be 5 seconds. Using critical flow ratios and lost times, the signal cycle length is computed as below.

$$\text{C}=\frac{1.5L+5}{1-Y}$$

where $Y=\sum _{i}^{n}{y}_{i}$ and$L=\sum _{i}^{n}{l}_{i}$ denote the summed critical flow ratios and lost times over all traffic phases in a signal cycle, respectively. And the superscript $n$ denotes total number of phases in a cycle, in general, $n=4$ for typical four-leg intersections.

Next, the length of green interval for each traffic phase is computed as below, and the yellow interval is assumed to be 3 seconds for all phases.

$${g}_{i}=(C-3n)\bullet \frac{{y}_{i}}{Y}$$

It is worth noting that the pretimed and adaptive methods differ in signal timing update frequency and the data used for update. Here, our proposed adaptive method uses real-time volume data collected in the previous signal cycle to update signal timings for the coming cycle. By contrast, the pretimed method uses historical volume data to update signal timing variables every 15 minutes. The historical volume data are generated by the same road network simulation but with variations on trip departure times.

Scenario design. To evaluate the performance of the proposed adaptive method, we designed a set of baseline BAU and target traffic scenarios for 100 cities. In the BAU scenario, all traffic signals in a city’s downtown area were pretimed. The preset traffic signal timings were derived from synthetic historical volume data using Webster’s method. The synthetic historical traffic volumes were comparable to target scenario traffic but with minor differences due to randomness in departure time choice of specified trips.

In our target scenarios, we implemented partial or all intersection signals using our proposed adaptive method. To explore congestion mitigation potential in a city, we used a target scenario with all adaptive traffic signals. To study the impact of implementation rate, we ranked all intersections by total traffic volumes of all traffic streams over an hour and selected intersections from high to low volume ones. We assume that traffic management agencies would follow such a quasi-optimal implementation order since high volume intersections are often more critical.

Performance indicators. We used total trip time, average travel speed, CO₂ emission reduction and fuel consumption savings to evaluate the benefits of the proposed adaptive method. Trip time and travel speed were selected considering both drivers’ and traffic police’s perspectives, respectively. We computed average travel speed and total trip time by querying trip trajectories from simulator. For a trip – $j$, the trip time (${TT}_{j}$) is computed by substracting arrival and departure times of the trip. The total trip time ($TTT$) is then obtained by summing over all trips. To compute average travel speed ($\varvec{v}$), we calculate the total trip distance ($TTD$) and then divide it by total trip time.

$$TTT={\sum }_{j}{TT}_{j}$$

$$TTD={\sum }_{j}{TD}_{j}$$

$$v=\frac{TTD}{TTT}$$

where, ${TT}_{j}$ and ${TD}_{j}$ are trip time and trip length for a trip - $j$. Such a quantification makes every kilometer travelled by vehicles count and equitable.

As noted before, the BAU transport emission data are obtained from GHSL-UCDB [Refs]. To estimate the emission change, we adopted a speed-based emission model from a previous study on real-world CO₂ emission under impacts of traffic congestion²⁶. This study models CO₂ emission (in natural logarithm) as a function of average travel speed.

$${ln}\left(E\right)= {b}_{0}+{b}_{1}\bullet v+{b}_{2}\bullet {v}^{2}+{b}_{3}\bullet {v}^{3}+{b}_{4}\bullet {v}^{4}$$

$$∆ TE={TE}_{0}\bullet (1-\frac{{E}_{1}}{{E}_{0}})$$

where $E$ denotes the amount of CO₂ emission (in kilogram or kg) per kilometer travelled by a vehicle. And $TE, ∆ TE$ denote total CO₂ emission and the change in emission, respectively. We used subscripts 0 and 1 to denote the emissions associated with BAU and target scenarios. And ${b}_{0}\sim{b}_{4}$ are parameters derived by regression fit of real-world experiment results. Here, we obtained the parameter values from the study on real-world CO₂ emission under impacts of traffic congestion²⁶.

The fuel consumption is estimated on the gasoline basis. We derived the amount of gasoline consumption based on the CO₂ emission. Typically, the combustion of 1 kg gasoline would produce about 3.3 kg CO₂ emission, according to The Engineering ToolBox. Hence, we estimated the reduction of gasoline consumption by multiplying amount of CO₂ emission reduction by 0.30.

Implementation cost and benefit. The implementation cost was estimated based on 145 field projects recorded in the Intelligent Transportation Systems Joint Program. The implementation cost includes one-time installation cost and annual operation and maintenance cost. The per-intersection installation cost varies between US$22,000 and US$82,300, with an average of US$48,069. The annual operation and maintenance cost is estimated at US$1,079 per intersection. The lifespan of the whole adaptive traffic signal control system was estimated to be about 71 months on average²⁷. Therefore, the annual per-intersection implementation cost is US$9,203, including both installation and operation costs.

The trip time benefit was estimated by investigating travelers’ willingness to pay for travel time savings (i.e., value of trip time). The results from a stated preference (SP) survey conducted in the city of Nanjing in 2018 were adopted²⁸. According to the survey, the mean estimated value of trip time is about US$4.84 (or 30.39 CNY) per hour. The trip time benefit was then estimated by multiplying the value of trip time with total trip time savings during both peak and off-peak hours.

We adopted the results from a study that used a global atmospheric model to simulate the benefits of global GHG reductions on air quality and human health⁴⁰. According to the study, monetized emission benefit estimates are US$70–840 per ton of CO2 for China, with an average of US$455 per ton. Gasoline was used as a surrogate for estimating benefits from fuel consumption savings. The price of gasoline was retrieved from Trading Economics, which is US$1,646 per ton (retrieved on 2022/01/20). The emission and fuel consumption benefits were estimated by multiplying the benefit per unit by total reductions of CO₂ emission and gasoline consumption, respectively.

Limitations. Our study has some limitations that could potentially affect the robustness of our findings. The main objective of this study was to assess the impact of our proposed adaptive traffic signal control method on urban congestion mitigation. The sample cities in this study only cover the top 100 congested cities in China (ranked by Gaode), and the time period for this study is limited to 6:00–23:00 each day. Despite these sample limitations, our findings are still valid since there might be very little difference between pretimed and adaptive methods under uncongested conditions. For example, there is less than a 1% change in speed in cities like Urumqi and Yili when implementing the proposed adaptive method.

In addition, we quantified trip time benefits based on a study conducted in 2018 and the baseline BAU CO₂ emissions using data collected in 2015 by assuming that socioeconomic conditions remain unchanged. This would actually underestimate the benefits to some extent, mainly due to the high-speed economic progress in China in recent years. Both urban residents’ value of time and travel demand (and associated transport emissions) have increased. However, it is challenging to get around this limitation mainly due to limited data and field studies on this related topic. Despite underestimating benefits, we can still conclude that the proposed adaptive method presents a cost-effective measure for congestion mitigation.

Our scenarios are affected by travel demand and trip length in a city downtown area. Apparently, higher travel demand and longer trip length often lead to more severe traffic congestion in cities. In practice, there might be induced traffic on highways with increased speed. This can occur when people choose to travel by car instead of public transport or decide to travel when they otherwise would not have. Additionally, shortening trip time may also encourage longer trips as reduced travel costs encourage people to choose farther destinations. Although this may not increase travel demands, it increases vehicle kilometers traveled. The abovementioned side-effects may compensate for impacts of our proposed adaptive traffic signal control. We can’t account for the induced changes in travel demand trip length because the data and model used for trip generation could not predict the impact of average speed on urban mobility. However, despite speed may not increase due to the induced traffic, the proposed adaptive method could possibly serve more trips. Other countermeasures like improving public transit systems^41,42 could compensate for induced travel demand by increasing the share of public transport in passenger traffic. In addition to impacts on motorized vehicles, further investigation is needed into the impact of adaptive traffic signal control on pedestrians and cyclists. We are aware of this limitation and identify key research to improve it in the future.

Despite these limitations, this research contributes to advancing the scientific understanding of how adaptive traffic signal control and congestion mitigation, emission reduction are connected. We highlighted that the adaptive method (and potentially other data-based measures) could offer a cost-effective pathway to mitigate traffic congestion and contribute to urban carbon emission reduction in cities. Our findings not only shed light on the impact of adaptive traffic signal control on congestion mitigation but also contribute essential new data-based solutions for attainment of the SDGs focusing on “sustainable cities and communities” and “climate action”, especially for cities in China and other countries globally that are facing similar issues. Additionally, the proposed central system to enable data-sharing among sectors is essential for both our proposed adaptive method and other data-based city management or governance solutions.

Data availability

Data supporting the findings of this study are available within the article and our Github repository at https://github.com/Kanstarry9T/Cong_ATSC/tree/master.

Code availability

We use CBEngine for simulating city-scale traffic in this study. CBEngine is freely accessible by users and can be downloaded at https://github.com/CityBrainLab/CityBrainLab. The code for this research is available at https://github.com/Kanstarry9T/Cong_ATSC/tree/master.

Acknowledgements

This study was supported by the Key R&D Project of Science Technology Department of Zhejiang Province (Grant No. 2022C01121) and Zhejiang Provincial Department of Transport research projects (Grant No. ZJXL-JTT-202223).

Author contributions

B.G., K.W., and G.Z. designed the study. K.W., J.D. and J.L. conducted the research. K.W. wrote the first draft of the paper, B.G. and Y.Z. revised the paper. Q.H. and T.X. processed the raw data. B.G., Y.Z., and G.Z. contributed to the discussion of the paper.

Declaration of competing interest

All authors have no conflicts of interest to report.

Yang, Z., Jia, P., Liu, W. & Yin, H. Car ownership and urban development in Chinese cities: A panel data analysis. Journal of Transport Geography 58, 127–134 (2017).
Li, Y., Xiong, W. & Wang, X. Does polycentric and compact development alleviate urban traffic congestion? A case study of 98 Chinese cities. Cities 88, 100–111 (2019).
Lu, Q. et al. Decarbonization scenarios and carbon reduction potential for China’s road transportation by 2060. npj Urban Sustain 2, 1–9 (2022).
Zhang, R. & Hanaoka, T. Cross-cutting scenarios and strategies for designing decarbonization pathways in the transport sector toward carbon neutrality. Nat Commun 13, 3629 (2022).
Keith, M. et al. A new urban narrative for sustainable development. Nat Sustain 6, 115–117 (2023).
Sun, C., Zheng, S. & Wang, R. Restricting driving for better traffic and clearer skies: Did it work in Beijing? Transport Policy 32, 34–41 (2014).
Wei, H., Zheng, G., Gayah, V. & Li, Z. A survey on traffic signal control methods. arXiv preprint arXiv:1904.08117 (2019).
Wei, H. et al. Presslight: Learning max pressure control to coordinate traffic signals in arterial network. in Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining 1290–1298 (2019).
Chow, A. H., Sha, R. & Li, Y. Adaptive control strategies for urban network traffic via a decentralized approach with user-optimal routing. IEEE Transactions on Intelligent Transportation Systems 21, 1697–1704 (2019).
Wang, Y., Yang, X., Liang, H. & Liu, Y. A review of the self-adaptive traffic signal control system based on future traffic environment. Journal of Advanced Transportation 2018, (2018).
Cai, C., Wong, C. K. & Heydecker, B. G. Adaptive traffic signal control using approximate dynamic programming. Transportation Research Part C: Emerging Technologies 17, 456–474 (2009).
Mirchandani, P. & Head, L. A real-time traffic signal control system: architecture, algorithms, and analysis. Transportation Research Part C: Emerging Technologies 9, 415–432 (2001).
Zhang, J. et al. City brain: practice of large-scale artificial intelligence in the real world. IET Smart Cities 1, 28–37 (2019).
Caprotti, F. & Liu, D. Platform urbanism and the Chinese smart city: the co-production and territorialisation of Hangzhou City Brain. GeoJournal 87, 1559–1573 (2022).
Erasmus, M. T., Leonie. Alibaba’s ‘City Brain’ is slashing congestion in its hometown | CNN Business. CNN https://www.cnn.com/2019/01/15/tech/alibaba-city-brain-hangzhou/index.html (2019).
Haklay, M. & Weber, P. Openstreetmap: User-generated street maps. IEEE Pervasive computing 7, 12–18 (2008).
Mooney, P. & Minghini, M. A review of OpenStreetMap data. Mapping and the citizen sensor 37–59 (2017).
Dion, F., Rakha, H. & Kang, Y.-S. Comparison of delay estimates at under-saturated and over-saturated pre-timed signalized intersections. Transportation Research Part B: Methodological 38, 99–122 (2004).
Cheng, D., Messer, C. J., Tian, Z. Z. & Liu, J. Modification of Webster’s minimum delay cycle length equation based on HCM 2000. in Paper Submitted to the Transportation Research Board for Presentation and Publication at the 2003 Annual Meeting in Washington. DC (2003).
Urbanik, T. et al. Signal Timing Manual, 2nd ed. (2015).
Hamedmoghadam, H., Jalili, M., Vu, H. L. & Stone, L. Percolation of heterogeneous flows uncovers the bottlenecks of infrastructure networks. Nature communications 12, 1–10 (2021).
Ritchie, H., Roser, M. & Rosado, P. CO² and Greenhouse Gas Emissions. Our World in Data (2020).
Emissions by sector - Our World in Data. https://ourworldindata.org/emissions-by-sector.
Florczyk, A. J. et al. Description of the GHS urban centre database 2015. Public Release 1, 1–75 (2019).
Florczyk, A. J. et al. GHS Urban Centre Database 2015, multitemporal and multidimensional attributes, R2019A. EUROPEAN COMMISSION & JOINT RESEARCH CENTRE (JRC)(eds.) (2019).
Barth, M. & Boriboonsomsin, K. Real-world carbon dioxide impacts of traffic congestion. Transportation research record 2058, 163–171 (2008).
Jung, Y. & Oh, J. Lifespan evaluation of traffic detector for Automated traffic recorders based on Weibull distribution. Journal of Transportation Engineering, Part A: Systems 143, 05017006 (2017).
Yang, Z., Liu, P., Xu, X. & Wang, W. Estimation of Value of Travel Time Savings Using Willingness-to-Pay Method. in (2015).
Rode, P. Urban planning and transport policy integration: The role of governance hierarchies and networks in London and Berlin. Journal of Urban Affairs 41, 39–63 (2019).
Rode, P. & da Cruz, N. F. Governing urban accessibility: moving beyond transport and mobility. Applied Mobilities 3, 8–33 (2018).
Boeing, G. Street network models and indicators for every urban area in the world. Geographical Analysis 54, 519–535 (2022).
Erlander, S. & Stewart, N. F. The gravity model in transportation analysis: theory and extensions. vol. 3 (Vsp, 1990).
Simini, F., González, M. C., Maritan, A. & Barabási, A.-L. A universal model for mobility and migration patterns. Nature 484, 96–100 (2012).
Fan, D. & Shi, P. Improvement of Dijkstra’s algorithm and its application in route planning. in 2010 seventh international conference on fuzzy systems and knowledge discovery vol. 4 1901–1904 (IEEE, 2010).
Wong, W. & Wong, S. C. Network topological effects on the macroscopic Bureau of Public Roads function. Transportmetrica A Transport Science 12, 272–296 (2016).
Simon Zhou, X. et al. A meso-to-macro cross-resolution performance approach for connecting polynomial arrival queue model to volume-delay function with inflow demand-to-capacity ratio. Multimodal Transportation 1, 100017 (2022).
Liang, C. et al. CBLab: Scalable Traffic Simulation with Enriched Data Supporting. arXiv preprint arXiv:2210.00896 (2022).
Liang, C. et al. CBLab: Supporting the Training of Large-scale Traffic Control Policies with Scalable Traffic Simulation. in Proceedings of the 29th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (2023).
Manual, H. C. HCM2010. Transportation Research Board, National Research Council, Washington, DC 1207, (2010).
West, J. J. et al. Co-benefits of Global Greenhouse Gas Mitigation for Future Air Quality and Human Health. Nat Clim Chang 3, 885–889 (2013).
Wu, K., Lu, M. & Guler, S. I. Modeling and optimizing bus transit priority along an arterial: A moving bottleneck approach. Transportation Research Part C: Emerging Technologies 121, 102873 (2020).
Wu, K. & Guler, S. I. Estimating the impacts of transit signal priority on intersection operations: A moving bottleneck approach. Transportation Research Part C: Emerging Technologies 105, 346–358 (2019).

(Not answered)

ExtendedData.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Cost-effective mitigation of urban congestion with adaptive traffic signal control

Status:

Version 1

Abstract

Figures

Full Text

Results and discussion

How adaptive traffic signal control mitigates congestion?

Congestion mitigation potential across China

Reduction of fuel consumption and emission

Cost-benefit analysis

Feasibility and supportive policy

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1