ArticlePDF Available

Abstract and Figures

Urban deliveries are traditionally carried out with vans or trucks. These vehicles tend to face parking difficulties in dense urban areas, leading to traffic congestion. Smaller and nimbler vehicles by design, such as cargo-cycles, struggle to compete in distance range and carrying capacity. However, a system of cargo-cycles complemented with strategically located cargo-storing hubs can overcome some limitations of the cargo-cycles. Past research provides a limited perspective on how demand characteristics and parking conditions in urban areas are related to potential benefits of this system. To fill this gap, we propose a model to simulate the performance of different operational scenarios—a truck-only scenario and a cargo-cycle with mobile hubs scenario—under different delivery demand and parking conditions. We apply the model to a case study using data synthesized from observed freight-carrier demand in Singapore. The exploration of alternative demand scenarios informs how demand characteristics influence the viability of the solution. Furthermore, a sensitivity analysis clarifies the contributing factors to the demonstrated results. The combination of cargo-cycles and hubs can achieve progressive reductions in kilometers-traveled and hours-traveled up to around densities of 150 deliveries/km ² , beyond which savings taper off. Whereas the reduction in kilometers-traveled is influenced by the the carrying capacity of the cargo-cycle, the reduction in hours-traveled is related to to the cargo-cycle ability to effectively decrease the parking dwell time by reducing, for instance, the time spent searching for parking and the time spent walking to a delivery destination.
Content may be subject to copyright.
Research Article
Transportation Research Record
2020, Vol. 2674(5) 553–562
ÓNational Academy of Sciences:
Transportation Research Board 2020
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0361198120917162
journals.sagepub.com/home/trr
Exploring Benefits of Cargo-Cycles
versus Trucks for Urban Parcel Delivery
under Different Demand Scenarios
Giacomo Dalla Chiara
1
, Andre
´Romano Alho
2
, Cheng Cheng
2
,
Moshe Ben-Akiva
3
, and Lynette Cheah
4
Abstract
Urban deliveries are traditionally carried out with vans or trucks. These vehicles tend to face parking difficulties in dense
urban areas, leading to traffic congestion. Smaller and nimbler vehicles by design, such as cargo-cycles, struggle to compete in
distance range and carrying capacity. However, a system of cargo-cycles complemented with strategically located cargo-
storing hubs can overcome some limitations of the cargo-cycles. Past research provides a limited perspective on how demand
characteristics and parking conditions in urban areas are related to potential benefits of this system. To fill this gap, we pro-
pose a model to simulate the performance of different operational scenarios—a truck-only scenario and a cargo-cycle with
mobile hubs scenario—under different delivery demand and parking conditions. We apply the model to a case study using
data synthesized from observed freight-carrier demand in Singapore. The exploration of alternative demand scenarios
informs how demand characteristics influence the viability of the solution. Furthermore, a sensitivity analysis clarifies the con-
tributing factors to the demonstrated results. The combination of cargo-cycles and hubs can achieve progressive reductions
in kilometers-traveled and hours-traveled up to around densities of 150 deliveries/km
2
, beyond which savings taper off.
Whereas the reduction in kilometers-traveled is influenced by the the carrying capacity of the cargo-cycle, the reduction in
hours-traveled is related to to the cargo-cycle ability to effectively decrease the parking dwell time by reducing, for instance,
the time spent searching for parking and the time spent walking to a delivery destination.
Urban deliveries are traditionally carried out using motor
vehicles such trucks, a term that we will use to encompass
the use of vehicles such as vans, lorries, or others that are
similar. Such conventional urban delivery method is
increasingly under pressure. Because of the increase in
urban population and the rise of e-commerce, delivery
demand has not only increased, but also shifted toward
smaller consignments shipped at a higher frequency (1).
However, urban transport infrastructure supporting city
logistics distribution has often remained unchanged. In
particular, trucks often lack available curb-space for
parking and loading/unloading in urban areas, with the
consequence of increasing cruising for parking, dwell
times (as a result of increase in walking distance), and
unauthorized parking (2). With the increase in urban
population density, traffic congestion levels have also
increased, negatively affecting truck travel times (3). To
reduce the negative externalities caused by trucks in
urban areas, including road blockages caused by
unauthorized parking and noise and air pollution, city
authorities have often taken regulatory actions imposing
parking and travel restrictions for trucks in urban areas
(4). These factors lead to an increase in delivery costs,
negative social and environmental impacts, or both.
An alternative to the use of conventional trucks to
perform urban last-mile deliveries that has been deployed
in several cities around the world (5–7) relies on the com-
plementary use of cargo-cycles, sometimes supported by
mobile hubs. We use the term cargo-cycles as encom-
passing of two, three and four wheelers, which have
some component of human-based propulsion, often with
1
Department of Civil and Environmental Engineering, University of
Washington, Seattle, WA
2
Future Urban Mobility Interdisciplinary Research Group, Singapore-MIT
Alliance for Research and Technology (SMART), Singapore
3
Department of Civil and Environmental Engineering, Massachusetts
Institute of Technology, Cambridge, MA
4
Department of Engineering Systems and Design, Singapore University of
Technology and Design, Singapore
Corresponding Author:
Andre
´Romano Alho, andre.romano@smart.mit.edu
electric motor support. Mobile hubs are small de-
consolidation centers which, differently from traditional
urban consolidation centers, often do not require a fixed
infrastructure, except pre-existing parking space. In this
case, trucks carry consolidated delivery loads from a
depot to central areas, more specifically to the hubs,
from which cargo-cycles pickup the goods and delivery
locally to the final destinations.
The use of cargo-cycles is commonly assumed to be
comparatively more agile than trucks, as these vehicles
require a smaller parking space, can access areas and
routes not accessible to trucks, and can park closer to
destination. This is particularly advantageous in settings
of heavy parking and traffic congestion, as these vehicles
can lead to reduced cruising for parking and unauthor-
ized parking if due care is taken in their use and adequate
regulations are in place. On the other hand, the cargo-
cycle is disadvantaged with respect to trucks as it able to
carry less cargo and it reaches a lower maximum travel
speed, being more suited to deal with smaller consign-
ments and higher delivery densities (deliveries per km
2
).
In fact, past research showed mixed results on whether a
cargo-cycle-based distribution system can bring about
improvements compared with the traditional truck-based
system. According to a literature review, 47% of the
studies reviewed proposed the use of bicycles/tricycles as
alternative vehicles to perform last-mile deliveries instead
of the conventional alternative (8).
In this study we seek to answer the following research
question: how do delivery demand and parking conditions
relate to the benefits of using a cargo-cycle-based distribu-
tion system when compared with the conventional truck-
based system? In this setting, we investigate how demand
characteristics, particularly delivery density, parcel
weight, and parking conditions, particularly the excess
dwell time resulting from cruising for parking and walk-
ing, influence the performance of a system of cargo-cycles
with mobile hubs, when compared with truck-based
deliveries, in ‘‘last-mile’’ operations. For this purpose, we
develop a simulation method that allows replicating to a
suitable extent the delivery tours of a freight carrier in
urban areas for both a truck-based system and a cargo-
cycle-based system with mobile hubs, under different
delivery demand and parking conditions. The simulation
method is based on detailed data for delivery trips per-
formed by trucks of a parcel delivery company in down-
town Singapore. The simulation method first takes trip
records as an input to generate a delivery demand with
given characteristics (location, quantity, time window).
Then, given a delivery demand, it simulates two opera-
tional scenarios: (a) a truck-only distribution scenario
and (b) a cargo-cycle with mobile hubs distribution sce-
nario. The truck-based distribution estimates are com-
pared with the real data for validation. The data are also
used to infer the traveling, cruising, and parking dura-
tions. For the latter two we assume a reduction when
using cargo-cycles. For varying demand characteristics,
we compare the two scenarios using several performance
metrics (total daily travel distance and time; total daily
operating time, which includes travel time and dwell time;
and the number of vehicles required). We also perform a
sensitivity analysis on three key model parameters as to
better understand their contribution to predicted total
travel distance and operations time.
In the following section we review the relevant litera-
ture. Then, we describe the truck trips data and the simu-
lation model developed. Further, we present the results
of the model validation, of the simulated scenarios,
and of the sensitivity analysis. We conclude with a dis-
cussion of the results.
Literature Review
Whereas some research focuses on the potential and fac-
tors that can lead to or restrict the adoption of cargo-
cycles (9,10) our focus is on the operational impact of
adopting this alternative mode. In this domain, a large
part of the research focuses on environmental impacts
compared with conventional delivery operations. Sadhu
et al. analyzed the data collected from 2,000 cycle rick-
shaw trolley (CRT) drivers in Delhi in 2011 and showed
that CRTs for city goods movements reduced fuel use
and vehicle emissions (11). Saenz et al. compared the car-
bon footprint of a cargo-cycles logistics service with that
of a traditional system (12). The results showed that total
greenhouse gas emissions reduced between 51% and
72% in the cargo-cycles system. Figliozzi and
Tipagornwong presented a lifecycle emissions minimiza-
tion model, concluding that using small cargo-cycles in
dense service areas, for relatively small loads, and when
the depot is located close to the delivery area, can obtain
the lowest lifecycle emissions (13).
There is also considerable attention to financial
impacts and operational impacts. Melo and Baptista
developed a microsimulation model to replicate traffic
conditions and estimate cargo-cycles environmental and
operational impacts (14). The results showed that the use
of cargo-cycles improves mobility, environment, energy
consumption, and running costs if deployed at the appro-
priate spatial scale within the city. Marujo et al. exam-
ined the cost and environmental impacts of mobile-
depot-based delivery using GPS data (15). They showed
that the use of cargo-cycles and mobile depots in the last-
mile delivery led to significant reduction of greenhouse
gas emissions and local air pollution. In addition, the
mobile-depot-based delivery setup yielded a slight cost
reduction over the traditional setups. Nascimento et al.
assessed the economic viability of cargo-cycles from the
554 Transportation Research Record 2674(5)
carrier’s perspective, concluding possible benefits despite
several contextual barriers to adoption in Brazil (16).
Arnold et al. simulated a cargo-cycle system for B2C e-
commerce distribution in a comparative setting with
pickups from lockers, and with a hybrid system of cargo-
cycles and lockers against conventional deliveries (17).
The authors focused on a comparison of internal and
external costs across the delivery strategies. The study
revealed that diminishing external costs requires an
increase of internal costs by the carriers. Gruber and
Narayanan compared the travel time difference between
cargo-cycles and conventional vehicles in commercial
transport, applying a regression model (9). The results
indicate that when the trips distance is short, the cargo-
cycles and cars performances overlap. The increase of the
trip distance made cars more advantageous. However,
the time to park, which might be comparatively larger
for trucks and contributes to the vehicle dwell time, was
a neglected element in their research. Tipagornwong and
Figliozzi analyze the competitiveness of cargo-cycles
against diesel vans in urban areas using a cost model and
considering two demand scenarios (18). The results indi-
cate that (a) cargo-cycle competitiveness is affected by
drivers’ costs assumptions, and (b) specific local condi-
tions such as parking difficulties, low travel speeds, and
time-constrained delivery windows can contribute to
widen the gap between the delivery modes. Similar con-
clusions about the scenarios in which cargo-cycles are
more promising are achieved by Conway et al. (5) and
Choubassi et al. (19). The latter highlighted factors such
as high delivery density and congested areas as well as
the presence of depots within the delivery areas as contri-
buting factors to the comparatively better performance
of the cargo-cycles. To the best of our knowledge no
study could be found that explores in detail the relation-
ship between the performance which the system of cargo-
cycles can achieve and demand characteristics relating to
delivery density and parcel/delivery weight distribution.
Methods
Data Description
For estimating delivery demand, we used the records of
about 20,000 deliveries performed over 3 months by a
parcel delivery company in the central business district
(CBD) of Singapore, a 2.3 km
2
urban area. For each
delivery, the delivery weight (in kg), delivery destination
GPS location, and delivery priority level are available
(‘‘high’’ priority consignments have to be delivered by
noon, ‘‘low’’ priority consignments can be delivered any-
time). Moreover, each record reports the date when the
delivery took place, two timestamps at the start/end of
the delivery time, and a unique ID of the vehicle used to
perform the delivery.
Note that when a truck is parked, the driver might
perform more than one delivery, walking between differ-
ent delivery locations, before returning to the truck. We
therefore rely on the concept of ‘‘delivery groups,’’
explained as follows. Delivery locations are first mapped
to buildings. Then, deliveries that took place consecu-
tively in time, within the same building, and performed
from the same vehicle are aggregated into a single ‘‘deliv-
ery group,’’ which indicatively corresponds to a group of
deliveries performed in a single parking event, that is,
without moving the vehicle from the chosen parking
location. This process leads to a set of 8,300 delivery
groups from the original dataset of about 20,000 deliv-
eries. The total shipment weight delivered in a delivery
group is the sum of the individual weights of the deliv-
eries associated with that group. For simplicity, we
address delivery groups as ‘‘deliveries’’ throughout this
paper.
Travel distances between consecutive deliveries are
obtained by computing the shortest path between each
pair of locations, and the respective travel times are
obtained by querying the Here Maps API (here.com) for
the historical travel time between each pair of locations.
We estimated the vehicle dwell times from the individ-
ual deliveries start/end timestamps as show in Figure 1.
Consider two deliveries, named i1and i, performed at
two different locations. First, we estimate the arrival
time to location i(tarr
i) by adding to the previous deliv-
ery departure time ( tdep
i1) the estimated travel time from
location i1to i(we estimated travel times querying the
here.com API). Then, the dwell time for delivery iis
computed by subtracting from the recorded delivery end
time the new estimated delivery start time (dwell time =
tdep
itarr
i).
The resulting dwell time distribution is right skewed
with a mean and median dwell time of 7.6 and 16 min,
respectively (see histogram in Figure 2). It is important
to note that the resulting dwell time estimates include not
only the times to deliver the consignments but also other
activities including cruising for parking (i.e., time spent
searching for available parking), queuing, and walking
time to the consignee. Later in the paper we will assume
that, by using a cargo-cycle, the vehicle dwell times are
reduced—because a cargo-cycle can park closer to a
Figure 1. Method used to estimate vehicle dwell times.
Dalla Chiara et al 555
destination, therefore reducing walking time, and does
not cruise for parking—with respect to trucks.
Overview of Simulation Model
The simulation model developed consists of three sub-
models (Figure 3):
a demand generation model that generates deliv-
ery demand scenarios based on (i) a real-world
dataset of daily deliveries performed by a freight
carrier and (ii) a set of parameters (demand den-
sity and delivery weight distribution) characteriz-
ing the desired demand scenario;
a (tour) scheduling model that organizes the gen-
erated delivery locations into time- and distance-
efficient freight vehicle tours;
a vehicle assignment model that calculates the
minimum number of vehicles (either trucks or
cargo-cycles) required to perform all scheduled
tours.
Both demand-specific inputs characterizing the
demand scenarios and the desired operational scenario
(i.e., the way deliveries are performed) need to be speci-
fied to operationalize the model. It should be noted that
the proposed method to generate tours does not aim to
compute optimal delivery tours, for example, tours with
the minimum travel distance and time. Our intention is
to replicate, to some extent, the actual tours observed to
be performed by the freight carrier. Details on this com-
parison are provided in the Operational Scenarios
section.
Simulation Sub-Models
Sub-model 1: Demand Generation. Delivery demand for a
hypothetical weekday is generated based on the records
of 8,300 delivery groups. A demand scenario is charac-
terized by (i) a demand density (number of deliveries per
km
2
for the study area), and (ii) a delivery weight distri-
bution. To generate demand with a given number of
deliveries, we re-sample without replacement the original
dataset of delivery records. To simulate different delivery
weight distributions, we multiply the original weights of
the sampled deliveries by a positive constant. We assume
the dwell time remains as per prior estimates.
Sub-model 2: Tour Scheduling. In the tour scheduling sub-
model, deliveries are organized into tours. The procedure
to construct distance- and time-efficient feasible tours
consists of three steps: (1) Clustering (2), Feasible tour
search and (3) Tour merging. The procedure takes two
main inputs: first, the daily delivery demand generated in
the previous demand generation sub-model, and second,
the characteristics of the operational scenario simulated,
including the type of vehicle used (truck or cargo-cycle)
and the related vehicle constraints (e.g., vehicle weight
capacity, maximum speed). The Operational Scenarios
section provides a detailed description of the different
operational scenarios simulated.
In the Clustering step the objective is to group a given
set of deliveries into disjoint clusters, each representing a
potential tour. We use a hierarchical clustering algorithm
to generate a hierarchy of nested clustering solutions. At
the lowest level of the hierarchy, there are as many clus-
ters as the number of deliveries; at the highest level, all
deliveries are grouped into a single cluster; any intermedi-
ate level is generated by merging two clusters generated
at the lower level. We aim to generate clusters which are
not only geographically compact, but also which contain
Figure 2. Empirical distribution of the estimated dwell times.
Figure 3. Simulation flow (dashed: inputs/outputs, solid border:
models).
556 Transportation Research Record 2674(5)
deliveries of the same priority level. For this, the algo-
rithm requires as input a distance matrix Dwith elements
dij, which is a measure of dissimilarity between deliveries
iand j. We define dij as the weighted average of (i) the
geographical distance between iand jand (ii) a binary
variable taking value 0 if iand jhave the same priority
level, and 1 otherwise.
Once clusters are defined, in the next step of Feasible
tour search, a tour is feasible when it satisfies the follow-
ing constraints: (i) the total weight of all deliveries in the
tour is less than or equal to the vehicle weight capacity,
(ii) the total time it takes for a vehicle to perform the
tour is less than or equal to the maximum vehicle operat-
ing time (assumed shift time), (iii) the total tour-distance
traveled is less than or equal to the maximum vehicle
travel distance (important for cargo-cycle range), and
(iv) all high-priority deliveries are scheduled before noon.
The upper bounds in constraints (i)–(iii) are vehicle-
specific, and therefore depend on the mode of transport
(i.e., trucks or cargo-cycles). The tour travel distance and
time are computed by finding the shortest path covering
all delivery points of the tour, starting and ending at the
depot (for trucks) or at a consolidation hub location (for
cargo-cycles), and assuming an average travel speed. The
shortest path is found by solving a traveling salesman
problem for each cluster. To satisfy constraint (iv), if a
tour contains at least one delivery with high priority, we
limit the maximum tour time to the length of the morn-
ing shift, such that the whole tour can be completed
before noon. We iteratively search for the largest clusters
satisfying all the above constraints from the hierarchy of
clustering solutions. We therefore start from the top of
the hierarchy and iteratively check, for each cluster in
the clustering solution associated with the given level
of the hierarchy, whether its associated tour is feasible.
Feasibility leads to the cluster’s deliveries being assigned
to a tour; otherwise, we proceed to a lower level of the
hierarchy, that is, splitting the cluster into smaller nested
clusters. The search stops when all deliveries are assigned
to feasible tours. In the operational strategy that involves
the use of the cargo-cycles, deliveries need to be assigned
to the respective hub location. To do so, we start the
search for feasible tour not from the top of the hierarchy
of clustering solution, but from a lower level, specifically
from the level containing as many clusters as the desired
number of hubs, and then search for feasible tours within
each cluster.
Lastly, for Tour merging the objective is to overcome
tours that are undesirably small (containing few deliv-
eries), therefore forcing the vehicle to return to the depot
(or the consolidation hub in case of cargo-cycles) more
often than needed. To form larger tours, we consider all
possible pairwise combinations of the identified feasible
tours and identify those in which merged tours are (i)
feasible tours and (ii) their total tour travel distance and
time are smaller than those of the original tours com-
bined. We then model the possible tour combinations as
a bipartite graph in which nodes represent the tours and
links connecting pairs of tours represent the feasible com-
binations. Then, the maximum matching of the graph is
obtained to find the largest possible number of combina-
tions. We then merge the selected matches into single
tours. We repeat the pairwise merging of tours until there
are no additional feasible combinations.
Sub-model 3: Vehicle Assignment. After we generate efficient
and feasible tours, we find the minimum number of vehi-
cles needed to perform all tours. We start from an initial
number of vehicles equals to the number of tours. We
then solve two bin-packing problems sequentially, in
which tours are assigned (‘‘packed’’) to a minimum num-
ber of vehicles (‘‘bins’’), in which the ‘‘capacity’’ of each
vehicle consists of the length of a driver’s working shift.
We assume a driver shift starts at 7 a.m. and ends at
5 p.m. In the first bin-packing problem we assign high-
priority tours (i.e., tours that contain at least one high-
priority delivery, meaning the whole tour needs to be
completed by noon) to vehicles, assuming that each vehi-
cle has an initial ‘‘capacity’’ to perform the high-priority
tours within 5 h (assuming vehicles start at 7 a.m.), there-
fore guaranteeing that high-priority tours are dealt with
within the morning shift. In the second phase we add to
all vehicles 5 h of ‘‘extra capacity’ (assuming the after-
noon shift starts at 12 p.m. and ends at 5p.m.) and assign
the remaining low-priority tours to vehicles solving a sec-
ond bin-packing problem. Finally, we retain only those
vehicles to which at least one tour was assigned. As a
result, we obtain the minimum fleet size needed to per-
form all tours, such that all high-priority tours can be
completed in the morning.
Operational Scenarios
We simulate the two operational scenarios: a truck-only
scenario (which is validated against the baseline scenario)
and a truck + cargo-cycle scenario. The first scenario
simulates a traditional distribution system relying solely
on trucks (scenario truck-only). Truck tours start and
end at a depot, located about 20 km from the area where
deliveries are performed. These tours are validated
against real data by computing the total travel distance
and operating time and comparing them against the
same metrics derived from the carriers’ data. We refer to
the real tours performed by the delivery company as the
baseline (scenario baseline) which we expect our algo-
rithm to replicate to a reasonable extent (scenario truck-
only).
Dalla Chiara et al 557
The alternative scenario simulates the use of a hybrid
system of trucks and cargo-cycles (scenario
truck +cargo-cycle): trucks are used to carry consoli-
dated loads from the depot to a mobile hub (hereafter
called ‘‘hub’’) located in the CBD; cargo-cycles pick up
the consolidated loads at the hubs and perform the indi-
vidual deliveries. Note that in this scenario, trucks are
used only to deliver consolidated loads to the hubs and
are not used for delivering to the final destinations, and
only cargo-cycles are used to perform the last leg of the
chain. Therefore, we do not consider the case of a mixed
cargo-cycle and truck scenario in which both vehicle
types are used for the last mile. Hubs are assumed to be
similar to containers in form and fit within a car parking
space. For the presented results, we assume the deploy-
ment of two hubs. Each hub contains a maximum of
nine cargo-cycle loads. The geographical locations of the
hubs change depending on the daily demand scenario
and are determined by the simulation model. Each hub
is placed at the geographical center of the demand cluster
of deliveries served by the hub. Given that cargo-cycles
are more compact and nimbler and therefore able to find
a parking space easily, their dwell times are assumed to
be 40% less compared with those of trucks. This is a
conservative assumption, as past research reports dwell
time savings of 75% (20). It should be noted that, as
mentioned earlier, our dwell time estimation includes the
time for cruising and queuing for parking. The original
maximum weight a cargo-cycle can carry is 205 kg.
However, recognizing that the heterogeneous volumes
and shapes of the delivery packages might not allow the
cargo-cycle to be loaded to maximum capacity, we
assume that only 70% of the maximum carried weight
can be loaded. Table 1 summarizes the input parameters
for each mode.
The two alternative operational scenarios are tested
on different demand scenarios, characterized by a given
demand density and delivery weight distribution. In the
baseline case, daily demand densities range between 25
and 90 deliveries per km
2
, the mean delivery weight is
5 kg and median weight is 1.1 kg. Each demand scenario
is simulated 50 times, and the resulting outputs are aver-
aged across simulation runs.
Performance Metrics and Sensitivity Analysis
The model outputs a set of metrics describing the opera-
tional performance of the simulated freight distribution.
These are: (a) total daily travel distance and time; (b)
total daily operating time, which includes travel time and
dwell time; and (c) the number of vehicles required.
Further, we analyze the sensitivity of some output
metrics—total travel time and operating time—to input
parameters. We test two vehicle-specific inputs: dwell
time reduction and capacity for the cargo-cycles and a
demand-specific input, the prior mentioned constant that
multiplies the original delivery weights. We take a one-
factor-at-a-time approach, changing one of the four
parameters each time, from 250% to 50% with a 10%
increment to the baseline value reported in Table 1, keep-
ing the other parameters fixed.
Results
Validating the Baseline Case
Figure 4 shows the output metrics for validation scenar-
ios (baseline and truck-only), using as input the observed
delivery demand. The baseline scenario illustrates the dis-
tance and time that the freight carrier took to perform
the deliveries, following the observed vehicle assignment
and delivery sequences in the data. The truck-only sce-
nario is obtained by simulation, using the same real
demand as input. Note in this case the demand genera-
tion model is not run and results are averaged by day of
the week. The daily travel distance of the simulated
truck-only scenario is on average 28% less, and the total
operating time is 12% less than in the baseline scenario.
This difference can potentially be explained by some fac-
tors which cannot be accounted for in the simulation,
such as:
Deviation in real-world delivery tours performed
by the drivers from optimal tours. These might
have been because of receiver-related factors (e.g.,
the receiver was not available at the moment of
the delivery) or driver-related factors (e.g., driver
might prefer to deliver bulky items first even if
they are not high priority as that would free space
in the van and ease subsequent load/unloads)
Exceptional traffic congestion conditions, parking
congestion conditions
Operational practices such as having smaller vans
re-loading from larger vans during daily operations
Naı
¨ve assumption of shortest path routes
Table 1. Assumptions of Mode-Specific Parameters
Type of vehicle
Model parameter Truck Cargo-cycle
Max. carried weight (kg) 600 kg 140 kg
Average speed (km/h) 30 km/h 15 km/h
Maximum distance
traveled (km) per tour
400 km 40 km
Maximum operating time
per day (h)
10 h 10 h
Dwell times change (with
respect to original dwell
times)
No change Reduced by 40%
558 Transportation Research Record 2674(5)
Despite this, we conclude that the simulation model
can replicate to a reasonable extent the performance of
carrier tour planning. Note that this comparison is per-
formed for averaged metrics across days of the week,
whereas for the following analysis we draw random deliv-
ery groups from any given day of the workweek.
Simulation-Based Analysis
Figure 5 shows the mean percentage change in travel dis-
tance and operating time obtained by using cargo-cycles
(truck +cargo-cycle scenario), versus the truck-only
scenario, for varying demand density conditions and dif-
ferent delivery weight multipliers. We observe a steep
decrease of both travel distance and operating time as
the demand density increases from 25 delivery/km
2
to
about 150 delivery/km
2
, reaching a 75% decrease in
travel distance and a 40% reduction in total operating
time (given the original weight distribution). After this
point, the savings in total operating time and travel dis-
tance stabilize as demand density increases.
We also observe that heavier weights influence the
performance, especially if combined with larger demand
densities. As the delivery weight distribution shifts
Figure 4. Daily total distance traveled (left) and the daily total operating time (right) for the baseline (observed data) versus truck-only
scenario (simulated).
Figure 5. Travel distance (left) and time (right) savings for cargo-cycle solution with respect to a truck-based solution.
Dalla Chiara et al 559
toward heavier weights, the improvements brought
about by the cargo-cycle system become less significant.
Figure 6 compares the average minimum fleet size
required to fulfill daily demand, for the two distribution
scenarios and for varying demand density. The smaller
fleet of cargo-cycles can be explained as a product of the
method, which minimizes the distance traveled and oper-
ating time but not the fleet size itself. Firstly, cargo-cycles
tours are shorter as these pick up the goods from hubs
located in the CBD, and do not require traveling back to
the depot. Since our method attempts to add multiple
tours to each vehicle, shorter tours can lead to a more
efficient use of vehicles. On the other hand, the method is
detrimental to the metrics of truck fleet required, because
longer tours leave less opportunities for matching in light
of existing schedules and maximum operating time per
day parameter.
Sensitivity Analysis
We perform a sensitivity analysis for three key para-
meters to better understand their contribution to pre-
dicted total travel distance and operations time, with
results illustrated in Figure 7. The selected parameters
are (1) the capacity of cargo-cycles, (2) cargo-cycle dwell
time, and (3) weight multiplier factor of parcels. In these
tests, we fix the demand density at 100 deliveries per km
2
.
Note that concerning total travel distance, this result is
highly dependent on the assumption of the depot loca-
tion; for example, in the truck +cargo-cycle scenario,
trips to/from the depot contributed on average 70% of
the total distance traveled.
The results for the selected parameters’ influence on
total travel distance demonstrate a major effect by the
assumed carrying capacity and minor effects from the
assumed dwell times and delivery weight. Looking at
reductions in capacity, a 50% reduction (to 70 kg) can
lead to 17% increase in travel distance. On the other
hand, increases in capacity have a small impact. With
regards to total operational time, the only critical factor
is the assumption of dwell time savings associated with
cargo-cycle use. This is justified mainly as the dwell time
contributes, on average, to more than 75% of total
operational time. In addition, the reduction in dwell
times can also increase the number of deliveries that can
Figure 6. Mean fleet size for truck-only (blue) and
truck +cargo-cycle (red) scenario.
Figure 7. Sensitivity analysis results.
560 Transportation Research Record 2674(5)
be completed in a single tour, which to some extent influ-
ences the fleet size required to fulfill the demand.
Conclusion
We developed and applied a quantitative method to
assess how delivery demand conditions relate to the ben-
efits of using a cargo-cycle-based distribution system
when compared with the conventional truck-based sys-
tem. We estimated the benefits in relation to total dis-
tance traveled, total operating time, and fleet size.
Decreases in all metrics can be justified as proxies for a
more sustainable freight distribution strategy. This
method is applied in a simulation context informed by a
dataset of real deliveries performed in the highly urba-
nized setting of Singapore’s CBD.
In line with recent experiments we found that the
introduction of cargo-cycles for last-mile distribution,
supported using mobile hubs, can lead to considerable
benefits. However, this is clearer under specific demand
scenarios, particularly at higher demand densities and
lower shipment weights. As delivery density increases
from low density to up to around 150 deliveries per km
2
the cargo-cycle based system shows increasing improve-
ments (reduction in total operating time and distance tra-
veled). Beyond this delivery density we observe a
stabilization of the cargo-cycle-derived performance
improvements compared with a truck-only system.
However, these results depend on several factors. Note
that actual delivery density would be equal or higher
than the used values, as we grouped deliveries performed
by the same vehicle in the same stop as the unit of analy-
sis. Furthermore, the starting baseline distribution of
observed delivery weights shows that most of the deliv-
eries recorded are light (median weight is 1.1 kg). Most
of the deliveries remain relatively light even when we
simulated heavier weights (by multiplying the original
delivery weights by a constant), but these hinder the ben-
efits of a system using cargo-cycles and mobile hubs. The
distribution of heavier weights was shown to sometimes
hinder any improvement, which supports the hypothesis
that this distribution system is more suited for parcel
deliveries, which is timely considering e-commerce
growth. Finally, we assumed that the cargo-cycles are
associated with a reduction in delivery times of 40% with
respect to trucks. Albeit conservative, this reduction
should be expected if the cargo-cycles are able to park
closer to the delivery destinations and are not affected by
parking congestion. To conclude, if demand density is
low, the benefits are expected to be fewer, and if a city is
not affected by heavy parking congestion, or the cargo-
cycles are affected as much as truck by parking conges-
tion, the benefits obtained by the use of cargo-cycles are
expected to be more limited.
Taking the baseline case for the freight carrier (with a
demand density between 25 and 90 deliveries per km
2
),
the total fleet size needed can be reduced by up to 36%.
Moreover, the total distance traveled can be reduced up
to 70%, and the total operating time up to 40%.
Considerations for operating costs and other influen-
cing factors were not explored in this research. For
instance, an important factor to take into consideration
when switching to a cargo-cycle with mobile hubs system
is the need for extra space closer to the delivery loca-
tions. Although these need not take the form of a fixed
infrastructure, the cargo-cycles still need to be parked
overnight and require charging stations. Moreover,
throughout the day the mobile hubs need to occupy the
space equivalent to a single parking slot. We cannot
ignore that larger scale deployment of cargo-cycles might
shift the traffic and parking management concerns to
curb-space management. We put forward as future
research the exploration of the impacts of wide-scale
adoption of cargo-cycles from a traffic flow and curb-
space management perspective, as well as giving consid-
eration to the impacts of street network design on the
presented results.
Acknowledgments
This work is supported in part by the SUTD-MIT International
Design Centre (IDC). We thank our partners, RYTLE GmbH
and United Parcel Service Singapore Pte Ltd (UPS), for sharing
data and insights. We also thank Ming Hong Chua and
Zhiyuan Chua for the support provided in data analysis and
processing. Any findings, conclusions, recommendations, or
opinions expressed are those of the authors only.
Author Contributions
The authors confirm contribution to the paper as follows: study
conception and design: Dalla Chiara, Alho, Cheah and Ben-
Akiva; simulation model development: Dalla Chiara and
Cheng; analysis and interpretation of results: Dalla Chiara,
Alho, Cheng and Cheah; draft manuscript preparation: Dalla
Chiara, Alho, Cheng and Cheah. All authors reviewed the
results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port for the research, authorship, and/or publication of this
article: This work is supported in part by the SUTD-MIT
International Design Centre (IDC), under the grant
IDG21800101, under a project titled ‘Data-Driven Design of
Dalla Chiara et al 561
Last-Mile Urban Logistics Solutions to Address E-Commerce
Growth’.
Data Availability
The data that support the findings presented here were used
under license for the current study. As such, the data are not
publicly available.
References
1. Shah, N., and J. K. Brueckner. Price and Frequency Com-
petition in Freight Transportation. Transportation
Research Part A: Policy and Practice, Vol. 46, 2012,
pp. 938–953. https://doi.org/10.1016/j.tra.2012.02.014.
2. Jaller, M., J. Holguı
´n-Veras, and S. Hodge. Parking in the
City: Challanges for Freight Traffic. Transportation
Research Record: Journal of the Transportation Research
Board, 2013. 2379: 46–56.
3. Figliozzi, M. A. The Impacts of Congestion on Time-Defi-
nitive Urban Freight Distribution Networks CO2 Emission
Levels: Results from a Case Study in Portland, Oregon.
Transportation Research Part C: Emerging Technologies,
Vol. 19, No. 5, 2011, pp. 766–778. https://doi.org/10.1016/
j.trc.2010.11.002.
4. Quak, H. J. (Hans), and M. (Rene
´) B. M. de Koster. Deli-
vering Goods in Urban Areas: How to Deal with Urban
Policy Restrictions and the Environment. Transportation
Science, Vol. 43, No. 2, 2009, pp. 211–227. https://doi.org/
10.1287/trsc.1080.0235.
5. Conway, A., J. Cheng, C. Kamga, and D. Wan. Research
in Transportation Business & Management Cargo-Cycles
for Local Delivery in New York City: Performance and
Impacts. Research in Transportation Business & Manage-
ment, Vol. 24, No. July, 2017, pp. 90–100. https://doi.org/
10.1016/j.rtbm.2017.07.001.
6. Akeb, H., B. Moncef, and B. Durand. Building a Colla-
borative Solution in Dense Urban City Settings to
Enhance Parcel Delivery: An Effective Crowd Model in
Paris. Transportation Research Part E: Logistics and Trans-
portation Review, Vol. 119, No. April, 2018, pp. 223–233.
https://doi.org/10.1016/j.tre.2018.04.007.
7. Lachapelle, U., M. Burke, A. Brotherton, and A. Leung.
Parcel Locker Systems in a Car Dominant City: Location,
Characterisation and Potential Impacts on City Planning
and Consumer Travel Access. Journal of Transport Geogra-
phy, Vol. 71, No. June, 2018, pp. 1–14. https://doi.org/
10.1016/j.jtrangeo.2018.06.022.
8. de Oliveira, C. M., R. A. De Mello Bandeira, G. V. Goes,
D. N. S. Goncxalves, and M. De Almeida D’Agosto. Sus-
tainable Vehicles-Based Alternatives in Last Mile Distribu-
tion of Urban Freight Transport: A Systematic Literature
Review. Sustainability, Vol. 9, No. 8, 2017. https://doi.org/
10.3390/su9081324.
9. Gruber, J., and S. Narayanan. Travel Time Differences
between Cargo Cycles and Cars in Commercial Transport
Operations. Transportation Research Record: Journal of
the Transportation Research Board, 2019. 2673: 623–637.
10. Thoma, L., and J. Gruber. Drivers and Barriers to the
Adoption of Cargo-Cycles: An Exploratory Factor Analy-
sis. Proc., City Logistics Conference 2019, Dubrovnik,
Croatia, 2019.
11. Sadhu, S. L. N. S., G. Tiwari, and H. Jain. Impact of Cycle
Rickshaw Trolley (CRT) as Non-Motorised Freight Trans-
port in Delhi. Transport Policy, Vol. 35, No. 2014, 2014,
pp. 64–70. https://doi.org/10.1016/j.tranpol.2014.05.015.
12. Saenz, J., M. Figliozzi, and J. Faulin. Assessment of the
Carbon Footprint Reductions of Tricycle Logistics Ser-
vices. Transportation Research Record: Journal of the
Transportation Research Board, 2016. 2570: 48–56.
13. Figliozzi, M., and C. Tipagornwong. Impact of Last Mile
Parking Availability on Commercial Vehicle Costs and
Operations. Supply Chain Forum: An International Journal,
Vol. 18, No. 2, 2017, pp. 60–68. https://doi.org/10.1080/
16258312.2017.1333386.
14. Melo, S., and P. Baptista. Evaluating the Impacts of Using
Cargo-cycles on Urban Logistics: Integrating Traffic, Envi-
ronmental and Operational Boundaries. European Trans-
port Research Review, Vol. 9, No. 30, 2017. https://doi.org/
10.1007/s12544-017-0246-8.
15. Marujo, L. G., G. V Goes, M. A. D. Agosto, A. Fer-
nandes, M. Winkenbach, and R. A. M. Bandeira. Asses-
sing the Sustainability of Mobile Depots: The Case of
Urban Freight Distribution in Rio de Janeiro. Transporta-
tion Research Part D: Transport and Environment, Vol. 62,
2018, pp. 256–267. https://doi.org/10.1016/j.trd.2018.
02.022.
16. Nascimento, C. O. L., I. B. Riggation, and L. K. de Oli-
veira. Characterization and Analysis of Economic Viability
of Cycle Logistics Transport in Brazil. Proceedings, City
Logistics Conference 2019, Dubrovnik, Croatia, 2019.
17. Arnold, F., I. Cardenas, K. So
¨rensen, and W. Dewulf.
Simulation of B2C E-Commerce Distribution in Antwerp
Using Cargo Bikes and Delivery Points. European Trans-
port Research Review, Vol. 10, No. 1, 2018. https://doi.org/
10.1007/s12544-017-0272-6.
18. Tipagornwong, C., and M. Figliozzi. Analysis of Competi-
tiveness of Freight Tricycle Delivery Services in Urban
Areas. Transportation Research Record: Journal of the
Transportation Research Board, 2014. 2410: 76–84.
19. Choubassi, C., D. P. K. Seedah, N. Jiang, and C. M. Wal-
ton. Economic Analysis of Cargo-Cycles for Urban Mail
Delivery. Transportation Research Record: Journal of the
Transportation Research Board, 2016. 2547: 102–110.
20. Ploos Van Amstel, W., S. Balm, J. Warmerdam, M. Boer-
ema, M. Altenburg, F. Rieck, and T. Peters. City Logistics:
Light and Electric. Amsterdam University of Applied
Sciences, The Netherlands, 2018.
562 Transportation Research Record 2674(5)
... For example, studies exist on the life cycle environmental impact of electrified mopeds [9,10] or stand-up scooters [11,12]. Existing studies on e-cargo bikes focus on the emission reduction potential of substituting conventional delivery vans with e-cargo bikes in urban areas for case studies in Europe [1,[13][14][15][16], the United States [17][18][19][20] and Singapore [21]. However, those studies only consider well-to-wheelemissions. ...
... According to the study's results, the adoption of e-cargo bikes is both environmentally and financially beneficial. [15] [21] Singapore Urban ...
... Concludes that cargo cycles can reduce travel time and distance travelled Similar case studies have been conducted for urban delivery systems in other European cities, such as Antwerp [16], and outside of Europenamely in Boston, Massachusetts [17], Seattle, Washington [18,19], Austin, Texas [20], and Singapore [21]. These studies find that the implementation of e-cargo bikes decreases GHG emissions. ...
Article
Full-text available
The population and economic growth in African countries such as Ghana and the resulting increase in the need for mobility are raising unsolved challenges. An increasing number of people and goods have to be transported within an inadequate energy supply and transport infrastructure. However, the increasing number of vehicles also cause challenges such as air pollution or traffic congestion. Electrified cargo bikes are considered a space-efficient and eco-friendly alternative for freight transportation in cities. Due to their low energy demand and battery capacities, they also offer the potential to be supplied with electricity via the mini-grids commonly used in African countries, as a full-coverage electricity supply is not yet available everywhere. The growth of electric cargo bikes for the delivery of parcels, groceries and other goods in urban regions raises the question of how environmentally friendly they are compared to other urban transport modes, considering the whole life cycle, and which modal share is required to have a significant impact on the overall transport system. This paper aims to answer this question by conducting a Life Cycle Assessment on the use case of an electric cargo bike used in Ghana based on different operating and dissemination scenarios. The design of the analyzed cargo bike is adapted, e.g. in terms of the used suspensions and tires, for a usage on Ghanaian roads and climatic conditions. The results show that cargo bikes have a smaller environmental impact on global warming potential in terms of ton-kilometers as diesel vans, especially if long product lifetimes are realized. The most important parameters for the reduction of greenhouse gas emissions are the use of renewable energies for the charging of the batteries, the utilization of the payload capacity and the production of the battery. It could be shown that the GHG emissions over the life cycle are significantly lower when the batteries are powered by solar energy. Including solar powered e-cargo bikes in the modal split for freight transport in the accra region would result in greenhouse gas emission savings of 4-8%.
... A range of other research publications 13 look instead at the combination of trucks and automated delivery robots. While some (16,17,18) 14 formulate problems where trucks are steered optimally to pick up and drop off delivery robots, 15 which perform the last-mile deliveries, Jennings and Figliozzi (19) take a macroscopic perspective 16 by proposing a continuous approximation of the problem. These use cases have in common that 17 they focus on the operational aspect of the system and are based on entirely artificially generated 18 demand data. ...
... The modeling pipeline presented in this paper follows a chain of distinct steps, which are displayed 14 in Figure 1. First, raw data sets are used to generate synthetic travel demand data, describing the 15 households, persons, and their daily mobility patterns in the study area. Second, marginal infor- 16 mation on the generation of parcels based on sociodemographic information is used to generate 17 the demand of parcel deliveries on an average day. ...
... The survey asked a total of about 2,000 respondents about their consump- 14 tion behavior with respect to out-of-home purchases. While the detailed survey data is not available 15 for this study, the marginal data provided by Gardrat (33) can be used. The marginal data gives the 16 annual number of out-of-home household purchases of a reference person which is described by 17 certain sociodemographic categories. ...
Preprint
Full-text available
The paper describes an open and fully replicable pipeline to assess the performance of parcel delivery systems. Starting with publicly available data in France, a range of open source tools is used to create a synthetic population of the region from which its parcel delivery demand is derived. Using JSprit and MATSim for solving the resulting Vehicle Routing Problem and simulating vehicle movements, a test case for automated delivery robots in the Confluence neighborhood of Lyon is set up and analyzed. Discussion on generalization and potential adaptation of the pipeline is provided.
... There are several studies concerned with the user characteristics of conventional bikes and e-bike users and potential users, whereas only a few of them are related to e-cargo bikes or cargo cycles in general. This is mainly because cargo cycle studies have been predominantly focused on cycle logistics as an innovative mode of transport in urban areas (Ben-Akiva et al., 2020;Büttgen et al., 2021;Caggiani et al., 2021;Conway et al., 2017;Elbert & Friedrich, 2020;Gruber et al., 2022;Gruber & Rudolph, 2017;Gruber & Thoma, 2019Gunes, 2021;Hagen & Scheel-Kopeinig, 2021;Katsela et al., 2022;Llorca & Moeckel, 2021;Nascimento et al., 2020;Niels et al., 2018;Nürnberg, 2019;Robichet et al., 2022;Sheth et al., 2019;Zhang et al., 2018). There is growing research looking into e-cargo bikes as private transportation, although, it is still very limited. ...
Article
Full-text available
Regarding climate change and energy resource problems, mobility transition is essential to shift from private fossil cars to sustainable modes in both the passenger and freight sectors. One of the reasons that citizens depend on their private cars is using this mode for carrying parcels, shopping, and transporting children. However, new electrized mobility modes like electric cargo bikes (e-cargo bikes) can provide mobility services to decrease citizens' car dependency. Recently, their share in the modal splits of cities has been increasing through technological advancement, infrastructure development, and government support in some countries. This paper aims to provide a comprehensive and international review of the role of e-cargo bikes in urban transportation. This paper reviewed and synthesized international studies in the field of e-cargo in terms of typology, technical parameters , users' characteristics, and readiness of cities for this emerging transport mode. Moreover, the barriers and driving factors for developing e-cargo bikes were analyzed and compared in terms of social, economic, environmental parameters, and regulations in different countries.
... However, the positive effects of using SEV are shaded by the unwanted private sector risks and additional costs. Another simulation system for comparison of two operational schemes (where deliveries are performed only by trucks or by CB with mobile hubs) is presented in the paper [59]. Considering the varying characteristics of demand, the evaluation of some key transport efficiency indicators was performed. ...
Article
Full-text available
The impact of the use of cargo bicycles for delivery processes on the environment is undeniably positive: it leads to the reduction of pollutants, noise, and vibrations caused by traditional vehicles; decreases traffic jams; causes more effective use of public space; and others. But how should such an effect be measured? What tools should be used to justify the necessity for change to more sustainable means of transport? How can we improve the state of the environment considering the interests of logistics service providers? There is a large amount of scientific literature dedicated to this problem: by using different modeling approaches, authors attempt to address the issue of sustainable transport. This paper conducts a literature review in the field of green cargo deliveries, investigates the benefits and drawbacks of integrating cargo bikes in urban logistics schemes, and examines methodologies and techniques for evaluating the impact of using cargo bicycles on the environment. By providing an opportunity to get acquainted with the situation in the sphere of green deliveries, the authors aim to encourage a breakthrough in the field of sustainable transport that may be achieved by using cargo bikes in modern cities. We review the existing approaches and tools for modeling transport emissions and state the significant positive environmental consequences.
... Many infrastructure, regulation, and technology measures have been proposed to promote sustainable urban deliveries (De Marco, et al., 2018) and to bring economic, environmental, and social benefits to city logistics actors and citizens (Russo & Comi, 2020), most notably the use of lowcarbon vehicles: e.g., battery electric vans, electric cargo scooters and cargo bicycles (Lebeau, et al., 2016) (Giordano, et al., 2018). Differently from battery electric vans, two-wheeled low-carbon vehicles have the potential to improve the livability of cities by reducing traffic congestion and delivery times, especially in densely populated urban areas (Della Chiara, et al., 2020). Furthermore, commercial cargo bicycles could bring benefits to the mental and physical health of the riders (Mueller, et al., 2015) (de Hartog, et al., 2010) and promote the transformation of city streets into less noisy and more attractive public spaces to all users. ...
Article
Commercial cargo bicycles have the potential to reduce greenhouse gas emissions, air pollution, traffic congestion, and delivery times. However, urban physical barriers and small cargo capacities could limit their uptake in city logistics. This analysis explores the effects of payload, air density, wind, and topography on commercial cargo bicycle energy use in cities with a focus on six European capitals (Berlin, Paris, Rome, Lisbon, Oslo, and London), and is supported by an empirical model using GPS traces and hourly weather data. The study finds that city hilliness could deplete up to 13-34% of the riders’ energy use, while wind speed can negatively affect the bikes’ energy use by up to 12-23%. Air-density effects are small; however, in cold cities like Oslo, they could reduce riders’ energy use by up to 8% in wintertime. These estimates can enable future studies to assess the operational feasibility of commercial cargo bicycles and help policy makers to promote their use by subsidizing the purchase of batteries to mitigate energy use differences across cities due to topographic or weather effects. Results revealed that, across the cities in this study, this “ideal battery size” varies from 0.1-0.8 kWh in Berlin to 1.6-2.1 kWh in Lisbon.
... Many infrastructure, regulation, and technology measures have been proposed to promote sustainable urban deliveries (De Marco, et al., 2018) and lower city logistics emissions in urban centers (BESTFACT , 2016), namely battery electric vans, electric cargo scooters and cargo bicycles (Lebeau, et al., 2016), besides other technological and route optimization strategies aimed at improving city logistics flows. Differently from battery electric vans, two-wheeled low-carbon vehicles have the potential to improve the livability of cities by reducing traffic congestion and delivery times, especially in densely populated urban areas (Della Chiara, et al., 2020). Furthermore, commercial cargo bicycles could bring benefits to the mental and physical health of the riders (Mueller, et al., 2015) (de Hartog, et al., 2010) and promote the transformation of city streets into less noisy and more attractive public spaces to all users. ...
Article
The deployment of low-carbon vehicles, such as cargo bicycles, will play a key role in reducing road transport emissions and contributing to city logistics’ sustainability. However, it is crucial to assess their ability to replace urban delivery operations and perform a fair carbon footprint comparison with diesel vans. This study aims at providing a replicable methodology to compare the environmental and operational performance of commercial vehicle technologies in cities. Results based on six cities reveal that while battery electric (BEV) vans may guarantee 100% replacement of the fleet, the replaceable mileage using only combinations of two-wheeled low-carbon vehicles varies from 24% in Lisbon to 62% in Berlin/London. Furthermore, when food needs are considered, human-powered cargo bicycles’ greenhouse gas (GHG) emissions are larger than for their electric models, and that the “type of diet” is critical to determine whether their deliveries have lower carbon footprint than electric scooters or BEV vans.
Article
Full-text available
Electric cargo cycles are often considered a viable alternative mode for delivering goods in an urban area. However, cities in the U.S. are struggling to regulate cargo cycles, with most authorities applying the same rules used for motorized vehicles or traditional bikes. One reason is the lack of understanding of the relationship between existing regulations and transport infrastructure and cargo cycle parking and driving behaviors. In this study, we analyzed a cargo cycle pilot test in Seattle and collected detailed data on the types of infrastructure used for driving and parking. GPS data were augmented by installing a video camera on the cargo cycle and recording the types of infrastructure used (distinguishing between the travel lane, bicycle lane, and sidewalk), the time spent on each type, and the activity performed. The analysis created a first-of-its-kind, detailed profile of the parking and driving behaviors of a cargo cycle driver. We observed a strong preference for parking (80 percent of the time) and driving (37 percent of the time) on the sidewalk. We also observed that cargo cycle parking was generally short (about 4 min), and the driver parked very close to the delivery address (30 m on average) and made only one delivery. Using a random utility model, we identified the infrastructure design parameters that would incentivize drivers to not use the sidewalk and to drive more on travel and bicycle lanes. The results from this study can be used to better plan for a future in which cargo cycles are used to make deliveries in urban areas.
Article
Urban freight studies of e-commerce activities and impacts tend to focus on last-mile movements of goods. Little information is available to quantify e-commerce activity for residential deliveries of parcels upstream of the last mile. In this exploratory study, we investigate the usefulness of a novel type of data—individual package tracking records—to measure end-to-end e-commerce activity. Using a small case study, we measure distances, times, and speeds of service to characterize components of e-commerce supply chains and to understand related impacts on transportation networks. Our results detail the strengths, weaknesses, and challenges expected in using this type of data, as well as provide insights for future data collection efforts. Overall, our results indicate that package tracking records do provide information that is potentially useful to inform transportation planning, but that further efforts are needed to establish appropriate technical and institutional frameworks to support large-scale analysis.
Article
Full-text available
Electric cargo cycles (E-cargo cycles) have the potential to reduce the negative impacts of transportation systems, when utilised for the right application. There is a growing literature pertaining to them, especially for commercial transport. It is essential to have a comprehensive outlook of the prevailing literature to synthesise the existing knowledge base. Therefore, this work comprehensively consolidates the studies in the growing field of E-cargo cycles. The primary focus is on commercial transport. However, insights obtained from the very limited number of studies pertaining to private transport is also presented. Within commercial transport, the focus is on typology, penetration, impacts, and operational and policy requirements. The typology is developed based on (i) Trip type, (ii) Network configuration, and (iii) Fleet composition. The factors influencing the penetration of E-cargo cycles are identified and classified into six groups, namely (i) Operation, (ii) Vehicular, (iii) Infrastructural, (iv) Workforce, (v) Organisational, and (vi) Policy. Furthermore, the impacts of E-cargo cycles are categorised as (i) Economic, (ii) Environmental, (iii) Societal, (iv) Traffic and safety, (v) Operational, and (vi) Governance. Besides, factors that influence the impact on cost-of-using E-cargo cycles are found and grouped under (i) Customer, (ii) Operator, and (iii) Policy maker. To facilitate the selection of network configuration and fleet composition, when beginning to utilise E-cargo cycles, a simplified decision making scheme is provided. Policy requirements discussed in the pertinent literature are also described and categorised as (i) Regulatory, (ii) Incentive, (iii) Infrastructural, and (iv) Awareness creation. Finally, a typology is provided for private transport, along with a summary on penetration and impacts.
Article
Urban freight by motorized transport poses a threat to the sustainable development of modern city centers through extensive emissions of greenhouse gasses, contributing to congestion, and increasing the risk of traffic accidents. In addition, further densification of European city centers demands an effective use of urban areas. As urban planners are shifting their focus from automobiles to cyclists and pedestrians, the cargo bike has been used as an alternative to motorized vehicles for urban freight in several European cities. Flexible use of cargo bikes permits efficient delivery of goods in congested urban areas, with zero emissions. Cargo bikes are also becoming a part of the city logistics in northern countries like Norway, where they are prone to adverse weather conditions, low temperatures and roads covered with snow and ice. The Armadillo, a four-wheeled cargo bike from the Swedish company Velove, was used in Norway for its first winter in 2018/19. Based on the lack of research on how cargo bikes operate in northern climates and a hypothesis that cargo bikes operate with less efficiency on winter roads than in summer conditions, the two following research questions were formulated: 1. What challenges and obstacles are cargo bikes facing on winter roads? 2. How can we facilitate and ensure efficient operation of cargo bikes in northern climates? To answer these questions, two methodological approaches were applied. First, observations and GPS-tracking of a working cargo cyclist were performed on winter roads in Trondheim, Norway. Second, seven qualitative interviews were conducted among experienced cargo cyclists in several Norwegian cities. Calculated velocities for the observed cargo cyclist showed that the average velocity in general was 27.6% lower for winter trips than spring trips. The observations and the interviews show that even though snow, ice, slush and low temperatures reduces the cargo bikes efficiency, using cargo bikes for urban delivery on winter roads is fully feasible. The design of the Armadillo, sufficient clothing and good planning solves many of the challenges faced by cargo bikes on winter roads. In addition, a cargo bike with more durable parts, larger wheels and better illumination, as well as an increase in the amount of bicycle lanes, would further facilitate the efficient operation of cargo bikes in northern climates. The biggest challenge concerning the use of cargo bikes for urban delivery in northern countries, seems to be urban consolidation centers or micro depots as an essential prerequisite for effective operation. Such centers would also provide the cargo cyclists a warm place to take breaks and to load, charge and maintain the cargo bikes. The findings in this study are supposed to provide further knowledge for urban planners, municipalities, road owners and logistic companies on how to ensure efficient operation of cargo bikes on winter roads.
Article
Full-text available
Cargo cycles are gaining more interest among commercial users from different business sectors, and they compete with cars in urban commercial transport. Though many studies show the potential of cargo cycles, there is still a reluctance to deploy them. One possible reason for this is the lack of knowledge regarding their suitability in relation to travel time. Therefore, this study aims to explore cargo cycles’ travel time performance by quantifying the travel time differences between them and conventional vehicles for commercial trips. The authors compare real-life trip data from cargo cycles with Google’s routed data for cars. By doing this, the authors explore the factors affecting the travel time difference and propose a model to estimate this difference. The attributes for the model were selected keeping in mind the ease of obtaining values for the variables. Results indicate cycling trip distance to be the most significant variable. The study shows that expected travel time difference for trips with distances between 0 and 20 km (12.4 mi) ranges from -5 min (cargo cycle 5 min faster) to 40 min with a median of 6 min. This value can decrease if users take the optimal cycling route and the traffic conditions are worse for cars. Although what is an acceptable amount of travel time difference depends on the user, practitioners can be certain of the travel time difference they can expect, which enables them to assess the suitability of cargo cycles for their commercial operations.
Article
Full-text available
Automated systems enabling customers to pick-up e-purchased goods from designated 24 h Parcel Locker pick-up points are proliferating. While they offer the promise of improved access to goods and reduced travel for consumers and delivery vehicles, the integration of these systems in cities, and their impact on access, generally and by mode are understudied. This paper explores the development, site characteristics and regional location characteristics of Parcel Lockers in five South East Queensland (SEQ) cities, Australia, and discusses implications for urban and transport planning in a car-oriented city. Micro level site analyses for 45 locker locations provide information on facilities (e.g. access by modes and universal accessibility, ground level, tenure type, safety). Hierarchical cluster analysis helps identify categories of locations. Suburb-level census and land use data as well as distance from transportation infrastructure and other Australia Post facilities were gathered and used to compare regional suburbs with and without Parcel Lockers. Logistic regression identifies the suburb level characteristics associated with the presence of a Parcel Locker (n = 273). Parcel Lockers in Brisbane are generally located at Australia Post locations, on commercial streets, shopping centres, suburban arterials and industrial parks. Automobile access and nearby parking is nearly ubiquitous, whereas alternative transport access is limited. Though site locations are constrained by commercial decisions, proximity to highways, to public transport, population density, a good balance of jobs and population, and higher rates of household Internet access are associated with the presence of a Parcel Locker. Documenting and assessing site characteristics and regional location of Parcel Lockers can help ensure its optimal and adequate integration and distribution across the region to satisfy city and transport planning goals. In its current state, Australia Post's Parcel Lockers partially meet the objectives of a sustainable city.
Article
Full-text available
Parcel delivery, one of the major components of e-commerce growth, has become increasingly challenging with the increase of urban logistics issues. During home delivery, many parcels may be returned if the customer is away from home, generating additional costs. In this paper, we propose a solution based on the crowd that consists to collect and deliver parcels using individuals (neighbors). The method uses circle packing to estimate the number of neighbors needed, the number of parcels they have to manage and the corresponding reward. An experiment was conducted on the 12th district of Paris (France) dataset. Encouraging results were obtained.
Article
Full-text available
The growth of e-commerce is accompanied by an increasing distribution of parcels in cities resulting in externalities like traffic congestion or emissions. As a consequence, different delivery concepts like bike deliveries or delivery points have been suggested. Naturally, companies will only accept these changes, if they do not result in higher costs. However, it is difficult to predict the impact of a certain delivery concept in a certain city. This leads to the research question, how different delivery scenarios for a certain area can be assessed and compared, especially if some of them have not been implemented.
Article
Full-text available
The advent of new technologies in last mile deliveries is about to cause a disruption in the traditional business model applied in urban cargo transportation, thus presenting innumerous research opportunities in this field of knowledge. In this context, identifying new operation models and vehicles that could be applied for last mile deliveries in urban areas becomes crucial. Therefore, this paper aims to identify, through a systematic literature review, the main types of vehicles addressed in the literature that could be used in the last mile of urban freight distribution in order to increase the sustainability of this type of operation. The results indicate a trend for the implementation of smaller and lighter vehicles for last mile deliveries in urban areas: 47% of the studies suggest, among other alternatives, the use of bicycles and tricycles; while 53% of the articles support the use of light commercial vehicles. Another trend observed in this type of distribution, indicated in 64% of the studies, is the shift from conventional (fossil fuels) to alternative sources of energy (electricity).
Conference Paper
A variety of drivers and barriers to the adoption of cargo cycles has been described in contemporary academic literature. This paper aims at reducing this complexity by identifying their underlying factor structure. To this aim, 389 organizations interested in cargo cycles rated their agreement towards the adoption of cargo cycles with 23 literature-derived drivers and barriers. An exploratory factor analysis yielded three driver factors (soft factors, cost benefits and urban advantages) and four barrier factors (vehicle limitations, worries and perils, riders’ concerns and infrastructure constraints) which are interpreted and discussed.
Conference Paper
Cargo cycles are gaining more interest among commercial users from different business sectors, and they compete with cars in urban commercial transport. Though many studies show the potential of cargo cycles, there is still a reluctance to deploy them. One possible reason for this is the lack of knowledge on their suitability in terms of travel time. Therefore, this study aims to explore cargo cycles´ travel time performance by quantifying the travel time differences between them and conventional vehicles in commercial transport. The authors compare real-life trip data from cargo cycles with Google´s routed data for cars. By doing this, the authors explore the factors affecting the travel time difference and propose a model to estimate this difference. The attributes for the model were selected keeping in mind the ease of obtaining values for the variables. Results indicate cycling trip distance to be the most significant variable. The study shows that expected travel time difference for trips with distances between 0 and 20 km (12.4 miles) ranges from -5 (cargo cycle 5 minutes faster) to 40 minutes. This value can decrease if users take the optimal cycling route and the traffic conditions are worse for cars. Though what is an acceptable value for the travel time difference depends on the user, practitioners can be certain of the travel time difference they can expect, which enables them to assess the suitability of cargo cycles for their commercial operations.
Article
To encourage the provision and use of more sustainable means of transportation, cities and companies are implementing a variety of measures, such as strengthening the use of public transportation infrastructure and services to alleviate traffic congestion and to democratize the urban space. In these cases, literature shows that the combined use of smaller vehicles and mobile depots is a practice to be explored more deeply. This paper focuses on the use of motorized cargo tricycles alongside conventional trucks in a mobile-depot-based procedure to accommodate the restrictions imposed on conventional freight vehicle access in densely populated areas. Therefore, a new method is proposed to identify the impact on service level, emissions footprint and delivery cost of this distribution strategy. Moreover, we assess the environmental benefits of this new distribution strategy by estimating the reduction in various pollutant emissions attributable to the adoption of smaller, more agile last-mile delivery vehicles. The analyses have shown that greenhouse gas emissions and local air quality pollutants can be significantly cut by the use of cargo tricycles and mobile depots in the last mile delivery. With respect to cost, we can show that the mobile-depot-based delivery setup yields slight cost advantages over the traditional setups for neighborhoods that are characterized by low average delivery drop sizes.
Article
Logistics, queuing and optimisation models are combined to study the impact of last-mile parking availability on commercial vehicle costs and operations. Parking availability levels affect commercial vehicle parking costs and operations and has an impact on route characteristics and commercial vehicle fleet sizes. The magnitude of the parking availability impacts on costs is a function of customer and route characteristics. Elasticity values indicates that only a few variables have a significant impact on commercial vehicle parking behaviour. Productivity improvements like service time reductions may result in undesirable changes in commercial vehicle parking behaviour.
Article
This study aimed to evaluate the operating performance of human-powered cargo cycles compared to motorized delivery vehicles for local delivery in Manhattan, New York City (NYC) in the United States of America (USA). While many recent studies have more broadly examined cost and impact tradeoffs in the use of cargo cycles for last-mile parcel delivery, the primary focus of this study was to understand how human-powered cargo cycle moving and parking behaviors differ from those of motorized vehicles in NYC. The study employed spatial and statistical analysis methods to directly estimate traffic performance measures using GPS data collected from two local operators. Traffic operating characteristics - including corridor travel speeds, trip delays, delivery times, and travel distances were examined to compare goods movements and parking behaviors for each vehicle type. Street and parking space, CO2, and particulate matter emissions savings were also estimated for alternate urban delivery vehicle replacement scenarios for one of the participating operators. Overall, results from this study suggest that cargo cycles can provide environmental benefits and offer a competitive last-mile delivery option for some local operators in very congested cities like NYC.