=Paper=
{{Paper
|id=Vol-3651/BMDA_paper6
|storemode=property
|title=Measuring the Impact of Road Removal on Vehicular CO2 Emissions
|pdfUrl=https://ceur-ws.org/Vol-3651/BMDA-6.pdf
|volume=Vol-3651
|authors=Simone Baccile,Giuliano Cornacchia,Luca Pappalardo
|dblpUrl=https://dblp.org/rec/conf/edbt/BaccileCP24
}}
==Measuring the Impact of Road Removal on Vehicular CO2 Emissions==
https://ceur-ws.org/Vol-3651/BMDA-6.pdf
Measuring the Impact of Road Removal on Vehicular CO2
Emissions
Simone Baccile1 , Giuliano Cornacchia2 and Luca Pappalardo3
1
University of Pisa, Pisa, Italy
2
University of Pisa, ISTI-CNR, Pisa, Italy
3
ISTI-CNR, Scuola Normale Superiore, Pisa, Italy
Abstract
Transportation networks face escalating challenges to cater to increased mobility demand while addressing traffic congestion.
Traditional remedies, such as adding roads, can paradoxically worsen congestion, as seen in Braess’s paradox. This study
emphasizes the potential benefits of strategically closing roads to alleviate congestion and carbon emissions. Milan serves
as a case study, where various road closure strategies were tested to identify scenarios where strategic removal not only
eased congestion but also significantly reduced CO2 emissions. The findings provide practical insights for urban planners
and policymakers, offering a roadmap to develop more efficient and eco-friendly urban transportation systems.
Keywords
CO2 emission, air pollution, mobility, urban simulation, urban sustainability, SUMO, braess paradox
1. Introduction sions, we use SUMO [9], an open-source and widely used
microscopic traffic simulator that allows for controlling
Traffic congestion is one of the most pressing problems different aspects of urban traffic, intervening in the road
in urban traffic management, increasing air pollution network, and simulating the impact of these interven-
and greenhouse gas emissions [1, 2]. Although several tions in terms of CO2 emissions.
policies have been applied so far to mitigate these issues Our analysis reveals roads that, if removed, could re-
[3, 4], evaluating their effectiveness is challenging be- duce CO2 emissions by approximately 10%, as well as
cause the interaction between the road network and the roads whose removal could result in an alarming increase
mobility demand is non-linear, making it hard to pre- in emissions of nearly 50%. Our work provides valuable
dict the behavior of numerous agents. This non-linearity insights and enables policymakers and city planners to
stems from various factors, such as varying traffic vol- analyze the potential outcomes of various road closure
umes, road conditions, and driver behaviors. strategies through what-if scenarios.
The complexity of the urban system can result in phe- The code for full replication of this work at https://
nomena like Braess’s paradox [5, 6, 7, 8]: adding roads github.com/Simoniuss/Braess-Paradox-Framework.
may inadvertently exacerbate congestion because each
driver, aiming to minimize travel time, may contribute to
slowdown traffic rather than alleviate it. The Braess para- 2. Related Work
dox has been studied on toy examples and validated em-
pirically in several cities and through simulations [6, 7, 8]. The Braess paradox (BP) [5] states that adding one or
In the real world, adding a new road requires a thorough more roads to a road network can cause a redistribution of
analysis of physical space, the available land, terrain char- the traffic flow such that the overall travel time increases.
acteristics, and the integration with the existing road net- When the road network is small, the BP can be solved
work. In this study, we focus on road closure since it is as a convex optimization problem [10, 11, 12]. When the
a less invasive and easy-to-make process. We assess the road network is large, as in the case of real-world road
impact of road closure on urban CO2 emissions in Milan, networks, the BP does not necessarily take place, i.e.,
Italy, assuming that closing a road may have beneficial adding a new road is not always detrimental [13].
effects, leading to a more uniform distribution of traffic Various studies have explored the impact of road clo-
across the remaining available roads. sure. In [14], the authors employed convex optimiza-
To evaluate the effects of road closures on CO2 emis- tion techniques to reveal that certain routes in Mont-
gomery County could lead to decreased travel times for
Published in the Proceedings of the Workshops of the EDBT/ICDT 2024 drivers if closed. In [15], the authors identified roads
Joint Conference (March 25-28, 2024), Paestum, Italy using heuristic with a genetic algorithm in Winnipeg
$ s.baccile@studenti.unipi.it (S. Baccile); (Canada), whose removal can reduce the total travel time
giuliano.cornacchia@phd.unipi.it (G. Cornacchia);
luca.pappalardo@isti.cnr.it (L. Pappalardo)
by 12%. Other studies focus on toy examples, modifying
Copyright © 2024 for this paper by its authors. Use permitted under Creative Commons License the type of agents, or incorporating additional informa-
Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
tion: Buscema et al. [6], add a line of sight to choose • Some contiguous edges are named with differ-
a route in NetLogo, Faccin et al. [16] use Belief-Desire- ent names (pseudonyms) but are the same road.
Intention agents to soften the BP effect, Zhuang et al. [7] We cannot automatically change these names be-
explore the effects in a dynamic traffic case, Ziemke and cause they are culturally or regionally dependent,
Nagel [17] simulate vehicular traffic using MATSim. so after detection, we change them manually.
In Barcelona, Sánchez-Vaquerizo and Helbing [8] dis-
Mobility demand. We model the vehicle flows
covered using SUMO that closing a few roads just to a
through the city’s mobility demand where a pair (o,d)
certain type of vehicle can reduce travel time by 14%
identifies each vehicle’s trip definition, representing the
and the emissions by 8%. Padrón et al. [18] perform a
origin and destination location. First, we divide the urban
simulation in SUMO and find that, in Valencia, closing
environment into tiles, each of which will be a possible
one of the major roads increases CO2 emissions by 3-
origin or destination location. We choose hexagonal tes-
4%. Other policies try to mitigate urban CO2 emissions,
sellation H3 developed by Uber and available in library
encouraging public transport, creating green areas, and
scikit-mobility [21]. Then, we use real mobility data, such
reducing heavy-duty vehicles [19]. A notable example is
as vehicles’ GPS traces, to estimate the flows between
the “three-in-one" policy applied in Jakarta [20], requir-
tiles. In practice, we build an Origin-Destination (OD)
ing all private cars, during peak hours, carrying at least
matrix where an element indicates the number of vehicle
three passengers to cross major roads. After the policy
trips from an origin tile to a destination tile.
was abandoned, there was an increase in delays from 2.8
Routes. For each (𝑒𝑜 , 𝑒𝑑 ) pair in the mobility de-
to 5.3 minutes per km during peak hours.
mand, we use a routing algorithm from SUMO, called
Duarouter1 , that connects two edges on a road network
3. Simulation Framework following the fastest path, i.e., the path that minimizes the
expected travel time. The fastest path can be perturbed
Our simulation framework generates urban traffic sce- using a randomization parameter 𝑤 ∈ [1, +∞) where
narios under various road closure strategies. The frame- the higher the 𝑤, the more the path is randomized (𝑤 = 1
work consists of two phases, each serving a specific pur- is exactly the fastest path). Duarouter dynamically dis-
pose. The first phase simulates urban traffic under normal torts edge weights (i.e., travel time) by a chosen random
conditions without road closures (baseline). The second factor drawn uniformly in [1, 𝑤). A 𝑤 value greater than
phase simulates vehicular traffic by systematically clos- 1 allows us to model imperfections in human driving
ing one or more roads through diverse closure strategies. behavior, reflecting the lack of complete knowledge of
Road network. We model the road network as a di- the road network while driving.
rected graph 𝐺 = (𝑁, 𝐸) where the set of edges 𝐸 con- Traffic simulation. We generate traffic from the ve-
tains the road or the streets, and the set of nodes 𝑁 con- hicle routes using SUMO (Simulation of Urban MObility)
tains the intersections (junctions) between roads. Each [9]. SUMO is a microscopic model, which means that
road may be composed of multiple edges, i.e., distinct each vehicle is modelled explicitly, has its route, and
segments of roads distinguished by specific attributes. moves individually through the network. We estimate
We aggregate edges in roads using road names. Given CO2 emissions using the HBEFA4 model inspired by the
that some edges are without road names and the lack of Handbook of Emission Factors from Road Transport [22].
a standardized naming criterion, we apply the following HBEFA4 provides emission factors for a wide list of ve-
preprocessing steps: hicle types, different pollutants, and several traffic situa-
tions. We use this model through the SUMO simulator,
• Each unnamed edge 𝑒 has an incoming and an
which automatically compute the CO2 (mg) emissions
outgoing edge, with their respective road names.
for each vehicle.
If those two road names are the same, we assign
Road closure. To simulate the effects of road closures
that name to 𝑒.
on CO2 emissions, we define a road closure simulation
• We use the ArcGIS service to get the midpoint
framework that allows us to select a set of roads and
coordinates (latitude, longitude) of each unnamed
remove them from the road network, maintaining the
edge using reverse geocoding, thus obtaining the
mobility demand as similar as possible. The steps of the
missing road name.
framework are the following:
• Since the road network may include more than
one municipality, there may be some edges with 1. Select a set of roads to close.
the same name (aggregated as roads) but geo- 2. Modify the edge parameters in the road network
graphically distant from each other. We identify using the “disallow" SUMO attribute to prevent
these roads using the boundaries of municipal- the passage of vehicles on the roads from being
ities and assign to each edge its name and the closed.
corresponding municipality. 1
https://sumo.dlr.de/docs/duarouter.html
3. Recompute the mobility demand. We iterate running a SUMO simulation with the mobility demand
through all the (𝑒𝑜 , 𝑒𝑑 ) pairs and see if the ve- derived from the GPS data.
hicle trip needs to be changed or not as follows: In the absence of the actual capacities of the roads,
• If 𝑒𝑜 or 𝑒𝑑 belongs to an edge of the re- we estimated them using the 2000 Highway Capacity
moved roads, we recompute a new trip for Manual [29]. Then, we compute the VOC for each edge
that vehicle. as:
𝑉 (𝑒)
• If 𝑒𝑜 and 𝑒𝑑 are not removed from the road, 𝑉 𝑂𝐶(𝑒) = (2)
𝐶(𝑒)
we verify if the two edges are still reach-
able after the road closure. If so, we keep To obtain the VOC associated with each road, we compute
the vehicle’s trip of the baseline mobility the weighted average of the VOC of each edge in the road:
demand. ∑︀
𝑒∈𝑟𝑜𝑎𝑑 𝑉 𝑂𝐶(𝑒) · 𝑙𝑒𝑛𝑔𝑡ℎ(𝑒)
• If the two edges are no longer connected, 𝑉 𝑂𝐶(𝑟𝑜𝑎𝑑) = ∑︀ (3)
we recompute a new trip. 𝑒∈𝑟𝑜𝑎𝑑 𝑙𝑒𝑛𝑔𝑡ℎ(𝑒)
4. Compute the routes for all the vehicles again, where a 𝑟𝑜𝑎𝑑 is composed by multiple edges 𝑒.
considering the new mobility demand.
5. Simulate traffic and gather results: we run a Kroad . The 𝐾𝑟𝑜𝑎𝑑 is a metric that measures the attrac-
SUMO simulation on the new road network with tiveness of each road segment. It quantifies how many
road closures and recomputed routed paths. Then, city areas (neighborhoods or tiles) contribute the most
we collect the results on CO2 emissions. to the traffic flow on that specific road segment [24]. To
compute the 𝐾𝑟𝑜𝑎𝑑 of each edge in the road network, we
first need to define the network of road usage, a bipartite
4. Road Classification network where each road edge 𝑒 is connected to its ma-
Inspired by previous works [23, 24], we classify roads jor driver areas. The major driver areas are the ranked
using a combination of theoretical measures derived from neighborhoods that produce 80% of the traffic flow for
graph theory, such as betweenness centrality, and data- an edge.
driven metrics like Volume-Over-Capacity (VOC) and We develop two concept related to the 𝐾𝑟𝑜𝑎𝑑 , the
(𝑠𝑜𝑢𝑟𝑐𝑒) (𝑑𝑒𝑠𝑡)
𝐾𝑟𝑜𝑎𝑑 , a metric indicating the degree of road usage. 𝐾𝑟𝑜𝑎𝑑 and the 𝐾𝑟𝑜𝑎𝑑 . An area 𝑇 is a driver source
for an edge 𝑒 if at least one vehicle, which passes through
Road betweenness centrality. It is a theoretical mea- the edge 𝑒, starts its trip from an edge 𝑒𝑠 ∈ 𝑇 . Similarly,
sure of centrality in a graph (road network) based on the an area 𝑇 is a driver destination for 𝑒 if at least one ve-
shortest paths. For our aim, we use the edge between- hicle, which passes through the edge 𝑒, ends its trip to
ness centrality [25, 26] to obtain the road betweenness an edge 𝑒𝑑 ∈ 𝑇 . An area can be both a source and a des-
centrality using a weighted average with the length of tination. For each edge, 𝑒, the driver sources and driver
the edge as weight: destinations can be ranked based on how many vehicles
traverse 𝑒, starting or ending in the respective area.
We compute the weight of each driver source as fol-
lows:
∑︀
𝑒∈𝑟𝑜𝑎𝑑 𝐵𝐶(𝑒) · 𝑙𝑒𝑛𝑔𝑡ℎ(𝑒)
𝐵𝐶(𝑟𝑜𝑎𝑑) = ∑︀ (1)
𝑒∈𝑟𝑜𝑎𝑑 𝑙𝑒𝑛𝑔𝑡ℎ(𝑒)
{︃
where 𝑟𝑜𝑎𝑑 is the road on which to compute the be- (𝑠𝑜𝑢𝑟𝑐𝑒) 1 if 𝑣 passes through 𝑒 from 𝑇
𝐼𝑒,𝑇 (𝑣) = (4)
tweenness, and it is composed of multiple edges 𝑒, and 0 otherwise
𝑙𝑒𝑛𝑔𝑡ℎ(𝑒) represents the length of the edge 𝑒.
∑︁ (𝑠𝑜𝑢𝑟𝑐𝑒)
𝐷𝑆𝑒 (𝑇 ) = 𝐼𝑒,𝑇 (𝑣) (5)
Volume-Over-Capacity. The Volume-Over-Capacity 𝑣∈𝑉
(VOC) is the ratio between the traffic flow on a road and
where 𝑒 is the edge on which compute the driver sources,
the capacity of the road. It is a standard metric to evaluate
𝑇 is a neighborhood, and 𝑉 is the set of vehicles crossing
a road’s service level and indicates how congested a road
the city. Similarly, we can define the driver destinations
is [27, 28]. When 𝑉 𝑂𝐶 < 1, the road can still contain
𝐷𝐷𝑒 (𝑇 ). We rank the list of driver sources (DS) and
other vehicles without undergoing particular slowdowns.
driver destinations (DD) for each edge and keep only
A road with 𝑉 𝑂𝐶 ≥ 1 suffers from congestion.
the DS and the DD responsible for 80% of the traffic
First, we compute the VOC to the edge level and then
flow on the edge. These new lists are the major driver
aggregate them to the road level. We compute the volume
sources (MDS) and the major driver destinations (MDD)
edge of each road segment using a data-driven approach,
for each edge 𝑒. After identifying the MDS and the MDD,
we can build the bipartite road usage networkFinally, If we aim to remove six roads, the list of removed roads
and 𝐾𝑟𝑜𝑎𝑑 from the bipartite would be 𝑅 = [𝑎1 , 𝑎2 , 𝑏1 , 𝑏2 , 𝑐1 , 𝑐2 ], such that the list 𝑅
(𝑠𝑜𝑢𝑟𝑐𝑒) (𝑑𝑒𝑠𝑡)
we compute 𝐾𝑟𝑜𝑎𝑑
network 𝐵𝐺: has the same amount of roads from each category.
The rationale behind employing a mixed strategy lies
in the road categorization based on their significance or
(𝑠𝑜𝑢𝑟𝑐𝑒) 𝑙
𝐾𝑟𝑜𝑎𝑑 (𝑒) = |{𝑙 | ∃𝑒 ← − 𝑇, ∀ 𝑇 ∈ 𝐴, 𝑙 ∈ 𝐿}| (6) attractiveness in the network. This approach increases
the likelihood of obtaining clusters with similar types of
roads, such as highways or freeways. Simply removing
(𝑑𝑒𝑠𝑡) 𝑙 all roads from a single category may have a detrimental
𝐾𝑟𝑜𝑎𝑑 (𝑒) = |{𝑙 | ∃𝑒 − → 𝑇, ∀ 𝑇 ∈ 𝐴, 𝑙 ∈ 𝐿}| (7) impact because vehicles previously using those roads
are now directed to alternative routes lacking similar
where 𝐴 is the set of area-nodes of the bipartite graph
characteristics, such as reduced capacity or lower speed
𝐵𝐺, 𝐿 is the set of links of 𝐵𝐺, 𝑒 is an edge ∈ 𝐸 the
limits. Therefore, removing roads from diverse categories
set of edge-nodes of the bipartite graph, and 𝑙 is an in-
helps mitigate this effect.
going or an out-going link from an area-node 𝑇 . Actually,
(𝑠𝑜𝑢𝑟𝑐𝑒) (𝑑𝑒𝑠𝑡)
𝐾𝑟𝑜𝑎𝑑 and 𝐾𝑟𝑜𝑎𝑑 are respectively the in-degree and
the out-degree of each 𝑒 ∈ 𝐸 of the bipartite graph. 6. Experimental Setup
Ultimately, we aggregate these measures to road level
with a weighted average using the length of each edge Road Network. We apply the simulation framework
as weight. to a squared area of almost 380 𝑘𝑚2 from the centre of
Milan and download the corresponding road network
Road clustering. We use a clustering approach to di- from OpenStreetMap, obtaining 24, 063 intersections
vide the roads into categories and obtain an effective and 46, 488 edges, which we aggregate into 7,654 roads.
closure strategy to evaluate the impact on CO2 emis- Mobility Demand. We split the selected area into
sions. We use the road betweenness centrality, the VOC, tiles using H3 hexagonal tessellation with a resolution
, and the 𝐾𝑟𝑜𝑎𝑑 as features for clustering to of 8, covering the selected Milan area with 680 tiles. We
(𝑠𝑜𝑢𝑟𝑐𝑒) (𝑑𝑒𝑠𝑡)
𝐾𝑟𝑜𝑎𝑑
obtain several road categories. We use the K-Means clus- then use GPS data describing 17,087 vehicles travelling
tering2 to group the roads and a grid search to find the between April 2nd and 8th, 2007 (658k points after pre-
optimal number of clusters. processing) to extract the flows of vehicles between tiles.
We test 𝑘 = 2, . . . , 10, choosing the best 𝑘 using the Next, we create a realistic synthetic OD matrix that mir-
elbow method [30]. rors real-world mobility patterns, maintaining the typical
distribution of trip distances and the power-law behav-
ior in the number of trips between two locations. We
5. Closure Strategies compute the mobility demand for 30, 000 vehicles as
it minimizes the Jensen-Shannon divergence between
We design different road closure strategies and evaluate the travel time distribution in real data and that of the
their impact on CO2 emissions. simulated vehicles [31, 32, 33].
CO2 policy. It removes roads based on their level of Routes. We generate the routes from the mobility
CO2 emissions. We normalize emissions by road length, demand using Duarouter, which computes the perturbed
obtaining CO2 per meter of road. fastest path using a randomization parameter 𝑤 = 7.5,
Category policy. It closes roads based on a classifica- which allows us to model the real behavior of drivers
tion of roads. We analyze the impact of CO2 emissions who typically do not follow the fastest route [34].
when roads are closed from each category, closing roads Closure strategy. We simulate the traffic demand
with the highest CO2 per meter first. using SUMO, obtaining the CO2 emissions for each edge
Mixed policy. We choose a list of roads for removal, and aggregating the results by road. We first simulate
ensuring that it comprises an equal number of roads the original road network as a baseline and then select a
from every category. These roads were ranked in de- closure strategy to remove a set of roads from the orig-
scending order based on their CO2 emissions per meter inal road network. A closure strategy consists of two
in each category. Initially, we eliminate the road with the parallel steps. The first is the informed strategy, where
highest emissions within each category, progressively we remove the roads from the road network based on
working towards those with lower emissions. As an ex- the defined strategy. The second is an uninformed policy
ample, consider three road categories denoted as 𝐴 = in which we close random roads from the road network.
[𝑎1 , 𝑎2 , 𝑎3 , 𝑎4 ], 𝐵 = [𝑏1 , 𝑏2 , 𝑏3 , 𝑏4 ], 𝐶 = [𝑐1 , 𝑐2 , 𝑐3 , 𝑐4 ], The removed roads in the uninformed strategy preserve
where each list is arranged in decreasing order of CO2/m. the same length as the informed one. For each policy,
2
we close 1, 10, 20, . . . , 100 roads. Independently of the
https://scikit-learn.org/stable/modules/clustering.html
policy applied, the road closures are selected with re-
spect to the baseline experiment, the original road net-
work. Thus, each set of roads is a subset of the next
set 1 ⊆ 10 ⊆ 20 ⊆ . . . ⊆ 100. We rank roads within
each strategy based on their CO2/m in descending order.
Subsequently, we incrementally select the set of roads
for closure, employing a precise approach to optimize
the reduction of carbon emissions. While this decision
is inherent in the CO2 policy by its definition, for both
the category and mixed policies, we follow the strategy
outlined in Section 5. This involves ranking roads based
on emission levels (CO2/m) in descending order and pri-
oritizing the closure of the most polluted roads from each
road category.
Road classification. We classify the roads of Milan to
identify similarities and differences between roads that
can be impactful in terms of CO2 (mg) emissions. As
suggested in [24], we characterize roads with their be-
(𝑠𝑜𝑢𝑟𝑐𝑒)
tweenness centrality and 𝐾𝑟𝑜𝑎𝑑 [24], as well as addi-
tional features such as the VOC, a useful metric to classify
(𝑑𝑒𝑠𝑡)
to quantify road congestion, and 𝐾𝑟𝑜𝑎𝑑 to capture the
attractiveness of each road based on the destination of Figure 1: Road distribution of each category defined from
the flows. We then apply a clustering algorithm using the classification of the roads in the road network of Milan.
the Silhouette score’s progression to determine that the (a)-(b) Roads with high 𝐾𝑟𝑜𝑎𝑑 . (c)-(d) Roads with low 𝐾𝑟𝑜𝑎𝑑 .
best number of clusters is four. We name the clusters as
follows:
• (HF): High 𝐾𝑟𝑜𝑎𝑑 , high relative VOC and emis- sure strategy. Now, we discuss each policy individually.
sions (in yellow); CO2 policy. It exhibits a beneficial impact on reduc-
• (HE): High 𝐾𝑟𝑜𝑎𝑑 , low relative VOC and emis- ing CO2 emissions up to a certain threshold (Figure 2a).
sions (in green); Closing up to 50 roads leads to a decrease in the overall
• (LF): Low 𝐾𝑟𝑜𝑎𝑑 , high relative VOC and emis- emission compared to the baseline scenario. After this
sions (in orange); threshold, the emissions increase exponentially with the
• (LE): Low 𝐾𝑟𝑜𝑎𝑑 , low relative VOC and emissions closing of every additional set of ten roads. The CO2
(grey). policy also highlights that an informed strategy based on
the level of CO2/m is more effective than an uninformed
First, we run a SUMO simulation on the original Milan strategy with the same emissions level as the baseline.
road network, i.e., without any road closure. Through Category policy. The impact of this policy depends
this simulation, we identify the roads with higher emis- on the road category. Closing HF roads (Figure 2b) leads
sions levels. Figure 1 shows the roads from each cluster to similar results to the CO2 policy: there are beneficial
previously identified for the 𝐾𝑟𝑜𝑎𝑑 and the mixed policy. effects until a certain threshold (40 closed roads). After
In Figure 1a and 1b, we observe the roads characterized closing 100 HF roads, CO2 emissions decrease, even if re-
(𝑠𝑜𝑢𝑟𝑐𝑒) (𝑑𝑒𝑠𝑡)
by high 𝐾𝑟𝑜𝑎𝑑 and 𝐾𝑟𝑜𝑎𝑑 . The yellow roads within maining above the baseline level. After closing 50 roads,
these figures can be differentiated from the green roads the random closure strategy outperforms the informed
based on their CO2/m (mg/m) and VOC levels. Notably, strategy regarding CO2 emissions, although still worst
the yellow roads exhibit higher levels of both CO2/m and concerning the baseline. Closing HE roads (Figure 2c)
VOC than the green roads. In Figure 1c and 1d, we show consistently results in a negative impact, with CO2 emis-
the roads classified with low 𝐾𝑟𝑜𝑎𝑑 . We can distinguish sions increasing up to 50% above the baseline. Even in
between two groups based on their VOC values: high the case of the uninformed strategy, emissions levels rise,
VOC for the orange roads and low for the dark grey roads. although to a lesser extent. Closing LF roads (Figure 2d)
always leads to an increase in CO2 emissions, although
to a lesser degree than other removal strategies. The
7. Results closed roads cause a maximum increase of 15% compared
Figure 2 provides an overview of total CO2 emissions to the baseline. Lastly, closing LE roads does not change
(mg) across the entire road network for each applied clo- emissions levels compared to the baseline (Figure 2e).
Mixed policy. Even if the impact of this policy does we are not simply redistributing the vehicles through
not exhibit a clear trend (Figure 2f), it is worse compared the road network but also removing roads from it. Road
to the baseline scenario. Focusing on the closure experi- removal limits the available route options for drivers,
ment of 20 roads, we find no linear relationship between leading to a situation that replicates or worsens the initial
the roads closed and the resulting CO2 emissions. conditions.
To gain insights into the impact of road closures on the
routed paths of vehicles, we analyze how the vehicles are
rerouted after a closure. Figure 4 shows an example of
vehicle rerouting after a road closure. The closed roads
are represented in black, while the routed paths are de-
picted in blue and orange. The orange path represents
the routed path in the baseline scenario; the blue path is
the rerouted path in the closure experiment. In this case,
we consider the CO2 policy with the removal of ten roads.
The rerouted path does not differ significantly from the
original path because the roads used in the closure exper-
iment have a smaller capacity than the baseline roads. As
a result, when simulating all traffic flows, this capacity
discrepancy leads to earlier congestion levels.
Figure 2: The results on CO2 emissions (mg) obtained from
different road closure strategies. (a) CO2 policy. (b-e) Category
policy. (f) Mixed policy.
Comparing strategies. Figure 3 provides a deeper
analysis of the different closure strategies from multiple Figure 3: The impact of different road closure strategies on
perspectives. In Figure 3a, we show the percentage of vehicles and the travelled road network edges. (a) Percentage
vehicles impacted by each closure strategy. A vehicle is of vehicles impacted by each road closure strategy. (b) Per-
considered impacted if its previous route contains at least centage of road network travelled.
one road closed in the closure strategy. The closure of
HE roads yields a higher percentage of impacted vehicles
due to two key factors. Firstly, HE roads encompass
7.1. Discussion
highly frequented routes, including highways, which
affect a larger proportion of vehicles. Secondly, HE roads Road closures impact CO2 emissions. Emissions de-
comprise a larger number of roads, some longer, thereby crease for specific road categories up to a certain thresh-
amplifying the impact on vehicle paths. old of removed roads. However, increasing the number
Figure 3b shows the percentage of the traveled road of removed roads, CO2 emissions increase exponentially,
network to the total available road network. In theory, reaching peaks of 50% above the baseline scenario with
the more vehicles are spread on the road networks, the the original road network. This happens because highly
fewer the emissions as we reduce the likelihood of road polluted roads are also typically those with higher ca-
congestion. However, this effect is not observed because pacity, and better equipped to handle larger traffic flows.
- Urban Artificial Intelligence” (CUP B53D23012770006),
Finanziato dall’Unione Europea - Next Generation EU.
This research has also been partially supported by EU
project H2020 SoBigData++ G.A. 871042; and NextGener-
ationEU—National Recovery and Resilience Plan (Piano
Nazionale di Ripresa e Resilienza, PNRR), Project “SoBig-
Data.it—Strengthening the Italian RI for Social Mining
and Big Data Analytics”, prot. IR0000013, avviso n. 3264
on 28/12/2021.
References
Figure 4: Example of vehicle route change after a road closure. [1] L. Pappalardo, E. Manley, V. Sekara, L. Alessandretti,
Black roads are the closed ones. In orange is the route of the Future directions in human mobility science, Na-
vehicle before the closure. In blue is the new route considering
ture computational science 3 (2023) 588–600.
the closed roads.
[2] D. Pedreschi, L. Pappalardo, R. Baeza-Yates, A.-
L. Barabasi, F. Dignum, V. Dignum, T. Eliassi-
Rad, F. Giannotti, J. Kertesz, A. Knott, Y. Ioan-
Consequently, traffic is rerouted through lower-capacity nidis, P. Lukowicz, A. Passarella, A. S. Pentland,
roads, thus increasing congestion and emissions. J. Shawe-Taylor, A. Vespignani, Social ai and
Not all roads count equally. Certain roads are piv- the challenges of the human-ai ecosystem, 2023.
otal in maintaining smooth traffic flow and preventing arXiv:2306.13723.
congestion. Those roads cannot be removed without in- [3] P. Kuss, K. Nicholas, A dozen effective interventions
creasing CO2 emissions. We also identify roads whose to reduce car use in european cities: Lessons learned
removal does not significantly affect CO2 emissions. from a meta-analysis and transition management,
Removing sparse roads is ineffective. Emissions Case Studies on Transport Policy 10 (2022). doi:10.
only decrease for a specific subset of roads, while most 1016/j.cstp.2022.02.001.
closure experiments result in increased emissions. In [4] P. Filigrana, J. Levy, J. Gauthier, S. Batterman,
other words, removing scattered roads throughout the S. Adar, Health benefits from cleaner vehicles and
road network may not be optimal. The underlying reason increased active transportation in seattle, wash-
lies in the rerouting of vehicles onto alternative roads, ington, Journal of Exposure Science & Envi-
which are often close to the removed roads and have ronmental Epidemiology 32 (2022). doi:10.1038/
lower capacity. Consequently, this necessitates consider- s41370-022-00423-y.
ing a “zone" closure strategy to mitigate this effect and [5] D. Braess, A. Nagurney, T. Wakolbinger, On a para-
optimize emissions reduction efforts. dox of traffic planning, Transportation Science 39
(2005) 446–450. doi:10.1287/trsc.1050.0127.
8. Conclusion [6] D. Buscema, M. Ignaccolo, G. Inturri, A. Pluchino,
A. Rapisarda, C. Santoro, S. Tudisco, The impact of
Using our simulation framework, we found that closing real time information on transport network routing
some roads is beneficial, but the removal becomes detri- through intelligent agent-based simulation, TIC-
mental above a certain threshold of closed roads. The STH’09: 2009 IEEE Toronto International Confer-
closure of other roads led to an increase in CO2 emis- ence - Science and Technology for Humanity (2009)
sions, while certain roads have negligible influence on 72 – 77. doi:10.1109/TIC-STH.2009.5444404.
CO2 emissions. Our work can be further improved in [7] D. Zhuang, Y. Huang, V. Jayawardana, J. Zhao,
several directions. For instance, another road closure D. Suo, C. Wu, The braess paradox in dynamic traf-
strategy could be to close all roads in specific “eco-zones" fic, 2022.
rather than closing single roads, which may lead to traffic [8] J. Argota Sánchez-Vaquerizo, D. Helbing, Less can
redirection onto adjacent roads with lower capacity. be more: Pruning street networks for sustainable
city making, 2022.
[9] D. Krajzewicz, G. Hertkorn, C. Feld, P. Wagner,
Acknowledgements Sumo (simulation of urban mobility); an open-
source traffic simulation, 2002, pp. 183–187.
Questo lavoro è stato finanziato dal PNRR (Piano [10] T. Roughgarden, Designing networks for selfish
Nazionale di Ripresa e Resilienza) nell’ambito del pro-
gramma di ricerca 20224CZ5X4_PE6_PRIN 2022 “URBAI
users is hard, 2001, pp. 472– 481. doi:10.1109/ big data resources, Transportation Research
SFCS.2001.959923. Part C: Emerging Technologies 58 (2015) 162–
[11] Q. Tu, L. Cheng, D. Li, J. Ma, C. Sun, Traffic paradox 177. URL: https://www.sciencedirect.com/science/
under different equilibrium conditions considering article/pii/S0968090X15001631. doi:https://doi.
elastic demand, PROMET - Traffic & Transportation org/10.1016/j.trc.2015.04.022, big Data
31 (2019) 1–9. doi:10.7307/ptt.v31i1.2795. in Transportation and Traffic Engineering.
[12] H. Yang, M. Bell, A capacity paradox in network [24] P. Wang, T. Hunter, A. M. Bayen, K. Schechtner,
design and how to avoid it, Transportation Re- M. C. González, Understanding road usage pat-
search Part A: Policy and Practice 32 (1998) 539–545. terns in urban areas, Scientific Reports 2 (2012).
doi:10.1016/S0965-8564(98)00017-2. URL: https://doi.org/10.1038%2Fsrep01001. doi:10.
[13] R. Steinberg, W. Zangwill, The prevalence of braess’ 1038/srep01001.
paradox, Transportation Science 17 (1983) 301–318. [25] U. Brandes, On variants of shortest-path between-
doi:10.1287/trsc.17.3.301. ness centrality and their generic computation, So-
[14] M. Burov, C. Kizilkale, A. Kurzhanskiy, M. Arcak, cial Networks 30 (2008) 136–145. doi:https://
Detecting braess routes: an algorithm accounting doi.org/10.1016/j.socnet.2007.11.001.
for queuing delays with an extended graph, 2021. [26] L. Lu, M. Zhang, Edge Betweenness Centrality,
[15] S. Bagloee, A. Ceder, M. Tavana, C. Bozic, A heuris- Springer New York, New York, NY, 2013, pp. 647–
tic methodology to tackle the braess paradox detect- 648. doi:10.1007/978-1-4419-9863-7_874.
ing problem tailored for real road networks, Trans- [27] O. Hrytsun, Impact of traffic volume and com-
portmetrica A: Transport Science 10 (2014) 437–456. position on the change in the speed of traffic
doi:10.1080/23249935.2013.787557. flow, Transport technologies 2023 (2023) 12–20.
[16] J. Faccin, I. Nunes, A. Bazzan, Understanding doi:10.23939/tt2023.01.012.
the behaviour of learning-based bdi agents in [28] D. Othayoth, K. K. Rao, Investigating the relation
the braess’ paradox (2017) 187–204. doi:10.1007/ between level of service and volume-to-capacity ra-
978-3-319-64798-2_12. tio at signalized intersections under heterogeneous
[17] T. Ziemke, K. Nagel, Braess’s paradox in an agent- traffic condition, Transportation Research Procedia
based transport model, Procedia Computer Science 48 (2020) 2929–2944. doi:https://doi.org/10.
83 (2016). doi:10.1016/j.procs.2016.04.190. 1016/j.trpro.2020.08.190, recent Advances
[18] J. Padrón, M. Terol, J. Zambrano-Martinez, and Emerging Issues in Transport Research – An Ed-
C. Calafate, J.-C. Cano, P. Manzoni, Assessing itorial Note for the Selected Proceedings of WCTR
the impact of road traffic constraints on pollution 2019 Mumbai.
(2021) 1–5. doi:10.1109/VTC2021-Fall52928. [29] W. Kittelson, R. Roess, Highway capacity analysis
2021.9625237. after highway capacity manual 2000, Transporta-
[19] H. Zhang, J. Peng, R. Wang, J. Zhang, D. Yu, Spa- tion Research Record 1776 (2001) 10–16. doi:10.
tial planning factors that influence co2 emissions: 3141/1776-02.
A systematic literature review, Urban Climate [30] H. Humaira, R. Rasyidah, Determining the appropi-
36 (2021) 100809. doi:10.1016/j.uclim.2021. ate cluster number using elbow method for k-means
100809. algorithm, 2020. doi:10.4108/eai.24-1-2018.
[20] R. Hanna, G. Kreindler, B. A. Olken, Citywide ef- 2292388.
fects of high-occupancy vehicle restrictions: Evi- [31] G. Cornacchia, M. Böhm, G. Mauro, M. Nanni, D. Pe-
dence from three-in-one in jakarta, Science 357 dreschi, L. Pappalardo, How routing strategies im-
(2017) 89–93. doi:10.1126/science.aan2747. pact urban emissions, in: Proceedings of the 30th In-
[21] L. Pappalardo, F. Simini, G. Barlacchi, R. Pellun- ternational Conference on Advances in Geographic
grini, scikit-mobility: A Python Library for the Information Systems, 2022, pp. 1–4.
Analysis, Generation, and Risk Assessment of Mo- [32] G. Cornacchia, M. Nanni, L. Pappalardo, One-shot
bility Data, Journal of Statistical Software 103 traffic assignment with forward-looking penaliza-
(2022) 1–38. URL: https://www.jstatsoft.org/index. tion, SIGSPATIAL ’23, Association for Comput-
php/jss/article/view/v103i04. doi:10.18637/jss. ing Machinery, New York, NY, USA, 2023. doi:10.
v103.i04. 1145/3589132.3625637.
[22] . Z. Beat Stutzer, Handbook emission factors for [33] G. Cornacchia, M. Nanni, D. Pedreschi, L. Pap-
road transport, 2023. URL: https://www.hbefa.net/ palardo, Navigation services and urban sustainabil-
e/index.html. ity, Fluctuation and Noise Letters (2023) 2450016.
[23] J. L. Toole, S. Colak, B. Sturt, L. P. Alexan- [34] A. Tawfik, S. Miller, Driver route choice behavior:
der, A. Evsukoff, M. C. González, The path Experiences, perceptions, and choices, 2010, pp.
most traveled: Travel demand estimation using 1195 – 1200. doi:10.1109/IVS.2010.5547968.