International Journal of Information Management Data Insights 1 (2021) 100010

Contents lists available at ScienceDirect

International Journal of Information Management Data

Insights
journal homepage: www.elsevier.com/locate/jjimei

An algorithmic approach for sustainable and collaborative logistics: A case

study in Greece
Grigorios D. Konstantakopoulos∗, Sotiris P. Gayialis, Evripidis P. Kechagias,
Georgios A. Papadopoulos, Ilias P. Tatsiopoulos
School of Mechanical Engineering, National Technical University of Athens, Sector of Industrial Management and Operational Research, Iroon Polytecnhiou 9, 15780
Zografos, Athens, Greece

a r t i c l e i n f o a b s t r a c t

Keywords: City logistics have always been significant for companies, since they constitute the last-mile distribution of goods
Collaborative logistics to consumers and retailers. Since the needs of consumers have increased in recent years, the exacerbation of traffic
Logistics pooling congestion, noise, and pollution in city centers has emerged. Therefore, logistics companies need to implement
City logistics
strategies to maintain their efficiency and accuracy, while simultaneously supporting green logistics. This paper
Case study
describes a sustainable approach where logistics companies collaborate, by sharing fleet and resources, in the
Vehicle routing
Scheduling of deliveries routing and scheduling operations. To estimate the improvements, in terms of pollution and cost reduction, the
Third party logistics case where companies operate independently, is compared to the collaborative case. The data exploited in the
research, originated from the daily distribution cases, faced by Third-Party Logistics companies in Greece. These
daily cases are addressed by a heuristic algorithm, either individually, for studying the way they operate today,
or jointly, for determining the collaborative case’s advantages.

1. Introduction
A few years ago, logistics companies, city inhabitants and consumers
would have been satisfied with just accurate (on-time) and low-cost de-
liveries, as they constitute the main goals in distribution, for these par-
ties. However, in the current state, where the reduction of traffic con-
gestion and air pollution in city centers have gained increased priority
for all stakeholders (Xiao & Konak, 2016), new technologies, applica-
tions (Kechagias, Gayialis, Konstantakopoulos & Papadopoulos, 2020b)
and strategies (Kechagias, Gayialis, Konstantakopoulos & Papadopou-
los, 2020a) are needed to be implemented, for satisfying the needs of
all parties involved. In this direction, companies could share their re-
sources, such as their vehicles, for reducing logistics costs, and mini-
mizing the environmental impact of the last-mile distribution process.
This strategy is either known as logistics pooling or collaborative logis-
tics (Wang et al., 2018), and has been implemented in multiple cases all
over the world (Rodrigues et al., 2018).
In the case of Athens, Greece, collaborative logistics was by no means
a common strategy, preferred by logistics and distribution companies, in
the past. The logistics companies where not keen with the idea of sharing
information, software tools and platforms, that becomes essential when
sharing their vehicles. This stance stems from the lack of openness and
spirit of cooperation, and has been the main reason for pushing away
the idea of synergies. However, the stance against logistics pooling and
collaborative routing of vehicles and scheduling of deliveries has been
recently changed, as companies have started to recognize its benefits.
These benefits are mainly related to the reduction of distribution costs
and greenhouse gas emissions, along with the increase of competitive-
ness.
In the present paper, the prevailing conditions of the distribution
of goods from 3PL companies to customers located in city centers, are
presented, along with the concept of collaboration between these com-
panies. Fig. 1 presents both cases, and describes that companies located
in the same area, organize and execute routes separately, in the exist-
ing situation. On the other hand, according to the proposed concept,
shared vehicles that receive goods from the warehouses of multiple com-
panies, execute routes that serve all customers. While at first sight, the
main benefit seems to be the decrease of needed vehicles, this cannot
be thought as de-facto without conducting further analysis. Therefore,
in order to enhance the research, and extract more accurate and precise
results, real-life distribution cases of three 3PL companies need to be
analyzed.
More specifically, the benefits arising from the transition from cur-
rent state to collaborative logistics, is estimated and calculated through
an improved heuristic algorithm. The developed algorithm, in the AS-IS
scenario, solves for each company separately, multiple daily distribu-
tion cases. On the contrary, in the collaborative routing and scheduling

∗
Corresponding author.
E-mail address: gkonpoulos@mail.ntua.gr (G.D. Konstantakopoulos).

https://doi.org/10.1016/j.jjimei.2021.100010
Received 13 November 2020; Received in revised form 12 March 2021; Accepted 12 March 2021
2667-0968/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al.
Fig. 1. (a) Description of the current state prevailing on most 3PL companies in Greece. (b) Description of the proposed collaborative strategy, that shared vehicles
load goods from all companies before executing the deliveries.
scenario, the daily distribution cases of all companies are united, form-
ing a larger daily distribution case. In this case, the vehicles involved
in the distribution, originate from all 3PL companies, and may need to
load goods from all warehouses before executing the plan of routes and
the schedule of deliveries.
In order to create efficient schedules, researchers have formed the
Vehicle Routing Problem (VRP) which refers to the problem of distri-
bution of goods. This is the general problem, while multiple variants
of the VRP have been created over the years, each of which corre-
sponds to different real-life cases faced during the distribution processes
(Konstantakopoulos, Gayialis & Kechagias, 2020). Additionally, in real-
life cases such as the ones studied in this paper, different variants of
the VRP may be combined, in order to form a complete mathematical
model, that is exploited for addressing the vehicle routing problem. In
the current research, when the 3PL companies operate separately, the
variants that are considered and combined are (i) the Heterogeneous
Fleet VRP (HFVRP), in which vehicles are different in terms of capac-
ity, fixed and variable costs (Subramanian, Penna, Uchoa & Ochi, 2012),
(ii) the VRP with Time Windows (VRPTW), that customers indicate time
slots in which vehicles may arrive and start serving (de Oliveira & Vas-
concelos, 2010; Konstantakopoulos, Gayialis, Kechagias, Papadopoulos
& Tatsiopoulos, 2020), and (iii) the VRP with Simultaneous Pickups and
Deliveries (VRPSPD), that customers may have both a demand for goods
to be delivered to them, as well as to be collected from them (Montané
& Galvão, 2006; Polat, Kalayci, Kulak & Günther, 2015). On the other
hand, in the synergy distribution cases, the Collaborative VRP (CVRP)
(Gansterer & Hartl, 2018) is also considered, along with the previous
variants. This means that all the above characteristics and requirements
exist, along with the fact that the resources, vehicles and deliveries of
multiple companies are combined. The aim in both cases is to minimize
the total distribution cost, while all constraints are respected.
The heuristic algorithm can handle all the above characteristics, re-
strictions and variables for minimizing the distribution cost, no matter
the case. More specifically, the algorithm, despite its functionality to
find routes and create schedule of deliveries that minimize the total dis-
tribution cost, calculates the emissions produced from the vehicles. The
emissions are not engaged neither in the objective function, nor to the
restrictions, but are exploited in order to estimate the difference in the
two cases.
Concluding, the research objectives of the present study are:
• the development of a heuristic algorithm that addresses efficiently
the collaborative VRP, along with the other variants that are consid-
ered, and
2
International Journal of Information Management Data Insights 1 (2021) 100010
• the definition of the economic and environmental benefits from such
synergies.
In the remainder of the paper, the Collaborative VRP with time win-
dows, heterogeneous fleet and simultaneous pickups and deliveries, as
well as the heuristic algorithm are described and analyzed in Section 2.
In Section 3 the case study is presented, along with the results obtained
from the comparison of the two cases. Finally, Section 4 includes the
conclusions of our research based on the results of the case study.
2. Collaborative logistics and VRP variants
The highly competitive industry of logistics and distribution, forces
companies to find efficient solutions in order to either maintain their
position in business or expand it. These solutions may range from ad-
vanced information systems, algorithms and new technologies, to col-
laborations between companies, or even to the combination of those.
In addition, while technological solutions are most often preferred by
companies, due to their efficiency, their cost is sometimes prohibitive,
leading companies to other solutions. On this premise, and due to the
increased demands and orders in urban areas, logistics companies plan
specific logistics processes jointly.
The route planning constitutes an operation that belongs in the spe-
cific category of supply chain cooperations. However, a prerequisite in
order logistics companies to collaborate in the route planning, is to per-
form similar logistics processes and functions, as well as operate at the
same level in a supply chain (Ferrell, Ellis, Kaminsky & Rainwater, 2020;
Verdonck, Caris, Ramaekers & Janssens, 2013)). This type of collabora-
tion is identified by researchers as horizontal (Pan, Trentesaux, Ballot
& Huang, 2019; Santos, Martins, Amorim & Almada-Lobo, 2021). On
the other hand, the case in which multiple stakeholders (manufactur-
ers, suppliers, distribution centers, carriers, customers etc.) that operate
in different levels in a supply chain collaborate, is identified as vertical
(Cleophas, Cottrill, Ehmke & Tierney, 2019; Santos, Curcio, Amorim,
Carvalho & Marques, 2021). Both types of collaboration can be benefi-
cial for companies and improve operational planning, but they focus on
different aspects of a supply chain.
In the current research, the horizontal collaboration between three
3PL companies in the routing and distribution process is studied.
This means that the customer requests of all companies, that repre-
sent a shipment either from the distribution center of the company
to the customer or the opposite, are shared, along with their re-
sources (vehicles). These variables define in a great extent the joint
vehicle routing problem that can also be mathematically formulated.
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al.
Soysal, Bloemhof-Ruwaard, Haijema and van der Vorst (2018) and
Montoya-Torres, Muñoz-Villamizar and Vega-Mejía (2016) proposed
such models and concluded that joint route planning offers significant
reduction not only in the distribution cost, but in emissions as well. How-
ever, multiple factors may influence this reduction, such as the size of
orders, the number of collaborating companies, as well as the variables
and constraints that are considered in the routing procedure (Cruijssen
& Salomon, 2004). That is why defining the variants of the VRP that are
involved in the routing process is very important.
More specifically, in the Collaborative VRP researchers focus on
the multiple depots that exist, since each logistics company has its
central depot. Due to this fact, many researchers consider the CVRP
an extension of the Multi-depot VRP (MDVRP) (Karakatič & Pod-
gorelec, 2015; Krajewska, Kopfer, Laporte, Ropke & Zaccour, 2008;
Montoya-Torres, López Franco, Nieto Isaza, Felizzola Jiménez & Herazo-
Padilla, 2015), where additionally resources and orders are shared. In
the current research, we follow this approach and formulate the Col-
laborative VRP as a multi-depot routing case. Additionally, we consider
more variants of the VRP that are correlated to the variables and con-
straints that arise from the studied 3PL companies, and their operations.
More specifically, the fleet of vehicles that each 3PL company owns or
uses is limited and composed of different types, in terms of capacity,
CO2 emissions, variable and fixed costs. This variant is known as the
HFVRP (Baldacci, Battarra & Vigo, 2008; Jiang, Ng, Poh & Teo, 2014;
Koç, Bektaş, Jabali & Laporte, 2014).
Moreover, one of the most frequently encountered variants of the
VRP according to Konstantakopoulos et al. (2020) and Elshaer and Awad
(2020) is the VRPTW. In this case, customers define a time slot in which
vehicles may arrive in order to serve the customers. If the vehicle ar-
rives before the start of the time window, it must wait till the time win-
dow opens, while in case the vehicle arrives after the end of the time
window, it cannot execute the delivery (Belmecheri, Prins, Yalaoui &
Amodeo, 2013; Chiang & Hsu, 2014; Ombuki, Ross & Hanshar, 2006).
In this case the time windows are considered hard, while in cases that
the vehicle may arrive after the end of the time window, and still serve
the customer, the time window is considered soft (Chu, Yan & Huang,
2017; Xu, Elomri, Pokharel & Mutlu, 2019). In the current research,
the time windows are considered hard since many of the customers of
the 3PL companies are demanding in this aspect as they receive multi-
ple orders in a daily basis. Finally, in this study customer requests are
not exclusively deliveries, but can also be returns or both. The case in
which deliveries and pickups are executed simultaneously is identified
as VRPSPD (Konstantakopoulos, Gayialis, Kechagias, Papadopoulos &
Tatsiopoulos, 2021; Montané & Galvão, 2006; Polat et al., 2015). In
every stage of the distribution process, including the departure of the
vehicle from the central depot, as well as its return, the goods on board
cannot exceed the vehicle’s capacity.
The identification of the multiple parameters and constraints that
affect the routing of vehicles and the scheduling of deliveries in each
company, as well as the mathematical formulation, constitute the first
steps for efficiently addressing the distribution process. The nest step in-
cludes the selection and use of advanced routing systems that integrate
optimization algorithms and offer efficient route planning and deliveries
scheduling. This is not an easy task as multiple commercial and free rout-
ing packages have been developed (Bräysy & Hasle, 2014; Wang et al.,
2015), especially in the last few years, where advanced technologies
and methods have been widely commercialized. Consequently, the se-
lection of the most suitable routing system according to the needs and
requirements of a specific company is essential and needs to be carefully
analyzed and implemented.
Additionally, the algorithms integrated into the information sys-
tems can be considered as black boxes, in most cases, as they con-
stitute the main part of the system and the one that generates rev-
enues to software companies. Despite the limited knowledge regard-
ing the integration of algorithms into real-life applications, it is well
established that researchers most commonly develop and propose
3
International Journal of Information Management Data Insights 1 (2021) 100010
bio inspired (Chakraborty & Kar, 2017; Kar, 2016) and local search
(Mekamcha, Souier, Bessenouci & Bennekrouf, 2019) metaheuristics,
while heuristics are studied less often. Both in research studies and real-
life applications, the algorithms need to be thoroughly tested in order
to validate their efficiency. Our research has reached this exact phase
and testings’ results will define whether the algorithm is suitable to be
integrated into a system. It should also be noted that, while studying
to the literature, we have observed a lack of algorithmic approaches in-
tegrated into routing systems proposed by researchers, most notably in
case of collaborative logistics. One of the very limited research studies
that focused both on the development of a routing system, and on the
algorithmic approach is the one of Erdoğan (2017) that developed an
open source routing system using Microsoft Excel.
To sum up, all the above variables and constraints, that derive from
the 3PL companies, are considered in this study and included in the
mathematical formulation of the next section. More specifically, the cus-
tomer requests and the resources of logistics companies are gathered in
a central pool (Verdonck et al., 2013) and the collaboration strategy is
formulated as a multi-depot vehicle routing problem with a heteroge-
neous fleet of vehicles, simultaneous pickups and deliveries, and time
windows. The aim of this problem, as in most VRP variants, is to min-
imize the total distribution cost by defining the fleet composition, and
the specific schedule of deliveries in each route (Laporte, 2007). This ef-
fort is enhanced by a heuristic algorithm that handles all these variables
and constraints, aiming to offer efficient and sustainable routes.
3. Problem description and solution method
3.1. Problem description and formulation
In this class of VRP variants that are studied in the current paper,
the set of depots (warehouses) D, as well as the set of customers N need
to firstly be defined. The set 𝐺 = 𝐷 ∪𝑁 defines the entire set, while 𝐴 =
{(𝑖, 𝑗 ) ∶ 𝑖 ≠ 𝑗 ∈ 𝐺}, the set of arcs. Each arc (i, j), which is a path from
i to j, is characterized by the distance 𝑑𝑖𝑗 , and by the travel time 𝑡𝑖𝑗 .
Each customer i is associated with a demand for deliveries 𝑟𝑖 , a demand
for pickups 𝑝𝑖 , a service time 𝑠𝑖 , and a time window [𝑒𝑖 , 𝑙𝑖 ]. The depots
also have time windows which are emerged by the working hours of the
logistics center. In each case, a vehicle may arrive before the start of the
time window 𝑒𝑖 , and may have to wait until it opens, but never after the
end of the time window, 𝑙𝑖 .
In addition, the fleet of vehicles, is represented by set K, while each
vehicle 𝑘 ∈ 𝐾 is characterized by its capacity 𝑄𝑘 , its variable cost 𝑣𝑘 and
its fixed costs 𝑓𝑘 . In each route, and consequently in each vehicle, the
goods on board can in no case, as well as in no stage of the distribution
process, to exceed the vehicle’s capacity. Therefore, in each customer,
both the delivered and the collected quantity of goods must be consid-
ered. All these variables, along with the limitations, make the studied
problem complex and hard to solve.
In the current state, each 3PL company has a single depot, and man-
ages the data of customers and vehicles itself. On the other hand, in the
case of Collaborative VRP, the depots come from the 3PL companies that
cooperate, while the data of vehicles and customers of the cooperating
companies are being united, forming problem of bigger size. However,
the mathematical formulation remains the same, irrespective of the case,
and is given below.
∑∑∑ ∑ ∑
𝑥𝑘𝑖𝑗 ⋅ 𝑓𝑘 + 𝑑𝑖𝑗 ⋅ 𝑥𝑘𝑖𝑗 ⋅ 𝑣𝑘 (1)
𝑘∈𝐾 𝑖∈𝐷 𝑗∈𝑁 𝑘∈𝐾 (𝑖,𝑗 )∈𝐴
∑∑
𝑥𝑘𝑖𝑗 ≤ 1, ∀𝑘 ∈ 𝐾 (2)
𝑖∈𝐷 𝑗∈𝑁
∑ ∑
𝑥𝑘𝑖𝑙 − 𝑥𝑘𝑙𝑗 = 0, ∀𝑙 ∈ 𝑁, 𝑘 ∈ K (3)
𝑖∈𝐺 𝑗∈𝐺
∑
𝑥𝑘𝑖𝑗 = 1, ∀𝑖 ∈ 𝐺, 𝑘 ∈ K (4)
𝑗∈𝑁
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al.
∑( )
𝑢𝑘𝑖𝑗 − 𝑟𝑗 + 𝑝𝑗 ⋅ 𝑥𝑘𝑖𝑗 ≤ Q𝑘 , ∀𝑖 ∈ 𝐺, 𝑘 ∈ K (5)
𝑗∈𝐺
∑
𝑢𝑘𝑖𝑗 ≤ Q𝑘 , ∀𝑖 ∈ 𝐷, 𝑘 ∈ K (6)
𝑗∈𝑁
∑
𝑢𝑘𝑖𝑗 ≤ Q𝑘 , ∀𝑗 ∈ 𝐷, 𝑘 ∈ K (7)
𝑖∈𝑁
𝑒𝑖 ≤ 𝑎𝑘𝑖 + 𝑤𝑘𝑖 ≤ 𝑙𝑖 , ⋅∀𝑖 ∈ 𝐺, 𝑘 ∈ 𝐾 (8)
𝑥𝑘𝑖𝑗 ∈ {0, 1}, (𝑖, 𝑗 ) ∈ 𝐴, ∀𝑘 ∈ K (9)
𝑢𝑘𝑖𝑗 ≥ 0, ∀(𝑖, 𝑗 ) ∈ 𝐴, 𝑘 ∈ K (10)
The objective function (1) seeks to minimize the total distribution
cost, that consists of the fixed and variable costs of all routes. Inequal-
ity (2) ensures that each vehicle starts from one of the available depots
(if multiple exist, such as in the Collaborative VRP), while Eqs. (3) and
(4) that each customer is served only once and by one vehicle. Con-
straints (5), (6) and (7) ensure that the load of goods in every stage of
the distribution process, including the departure from the depot and the
return to the depot, cannot exceed the vehicle’s capacity. Constrain (8)
ensures that each vehicle cannot start serving any customer, before the
time window starts, or after its close. Finally, the decision variable 𝑥𝑘𝑖𝑗 is
equal to 1, if the route from i to j is executed, 𝑢𝑘𝑖𝑗 is the load on vehicle
k and on arc (i, j), while 𝛼𝑖𝑘 is the arrival time of vehicle k at customer
i, and 𝑤𝑘𝑖 the waiting time of vehicle k at customer i.
3.2. Solution method
The algorithm developed for creating efficient schedules of deliver-
ies, belongs to the category of heuristics. Initially, a construction algo-
rithm operates for creating an initial solution, that is then further im-
proved by applying multiple local search heuristics. The optimization
objective is the minimization of the total distribution cost. The construc-
tion algorithm applied in the present paper is the Time-Oriented Nearest
Neighbor (TONN) algorithm, that was first proposed by Solomon (1987).
The specific procedure belongs in the sequential construction methods,
as one route at a time is being built.
More specifically, every single vehicle first visits the depots of the
companies (in current state only the company’s depot) and afterwards,
the customers. The customer that is added each time in the route, is the
“closest” to the last visited one, according to the metric 𝑐𝑖𝑗 , that involves
both time and distance. Although this procedure is very efficient in terms
of computational time, the same cannot be said for the quality of the so-
lution. Therefore, local search procedures are needed to be implemented
for modifying the existing solution S, to another solution, S’. In order to
create another solution, one or more customers are being removed and
reinserted at different positions. The operators that are implemented in
the current research are the node relocation, the k-opt moves, the Or-
Opt operator, and the exchange operator (Bräysy & Gendreau, 2005).
All these operators let the algorithm to search for better solutions in a
wider search space, while the computational time needed remains low.
If the total distribution cost of the new solution S’, has been decreased
due to the operator applied, then the optimal solution is renewed. This
procedure continues till the computational time meet the time limit that
was set, which is 5 min for each distribution case. The algorithm was
developed in Python programming language, and all cases were tested
on an Intel Core i7 8550 U 1.8 GHz with 16 GB memory computer. Fi-
nally, the steps that the algorithm follows, and that were also described
above, for finding the optimal route, are presented in Fig. 2.
4
International Journal of Information Management Data Insights 1 (2021) 100010
Fig. 2. General Framework of the proposed heuristic algorithm.
4. Results
4.1. Computational data – case studies
The need of companies to reduce both their distribution costs, as well
as their environmental footprint, has lead researchers and practitioners
to search for new strategies in order to achieve these goals. The col-
laborative VRP that is studied in the current research can manage both
factors, and enhance this effort. The data that are essential for address-
ing the problem are exploited from three medium size 3PL companies.
All companies are located in the area of Mandra, a suburb in the western
area of Athens, Greece, and distribute goods in city center. Therefore,
their cases can offer significant results, concerning collaborations be-
tween companies in their distribution operations.
Initially, the distribution cost and the CO2 emissions are calculated
for each company, and for numerous daily distribution cases that each
company addresses, separately. However, since the three companies dis-
tribute goods in Athens, the daily distribution cases that each company
faces, are also combined producing a single daily case, with multiple de-
pots, shared vehicles, and customer data resulting from all companies,
in order to form the case of collaborative logistics. The distribution cost
and the CO2 emissions are additionally calculated, and the results are
compared to those exploited from the first case, in order to estimate the
benefits of this strategy.
In order to create the schedule of deliveries, as well as to calculate the
distribution cost and the total emissions, multiple data are needed. These
data originate and are related to both the customers and the vehicles,
and are used either as variables, or as restrictions of the vehicle routing
and scheduling problem, as described in Section 2.1.
Concerning the customers, each one separately is characterized by
his ID number, address, the delivery and pickup quantity, and time win-
dow, as shown in Table 1. The addresses are essential for extracting all
the necessary distances and travel times, between customers (includ-
ing the depots). For creating the distance and travel time matrices, on-
line services are exploited. More specifically, these services, including
geocoding, as well as distance and travel time calculation, are offered
by HERE Technologies (www.here.com). All these data, along with the
delivery and pickup quantities, as well as with the time windows, are
further exploited by the heuristic algorithm as variables and constraints
of the problem for optimizing the plan of routes and schedule of deliver-
ies. More specifically, the delivery and pickup quantities are considered
in every stage of the distribution process so that the goods on board will
not exceed the vehicle’s capacity, while the time windows are consid-
ered so that to avoid delays.
As for the vehicles, the variable and fixed costs have been calcu-
lated based on the data provided by each one of the 3PL companies.
More specifically, the fixed cost of each vehicle is estimated by its an-
nual and capital cost, that include depreciation, repairs, maintenance
and repairs. According to the annual use of each vehicles, which is the
working days per year, and by considering that each vehicle executes a
single route each day, the fixed cost is then estimated in a daily basis.
Moreover, the variable cost is based on the fuel consumption of each
vehicle (liters/ 100 km) and the cost of fuel (€/ liter). Representative
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al. International Journal of Information Management Data Insights 1 (2021) 100010

Table 1
Data of customers.

ID Address Delivery quantity (kg) Pickup quantity (kg) Start of time window End of time window

1001 12 Arrianou, Athens, Attica 116 35, Greece 24.3 8.1 07:30 08:30
1002 1 Angelon, Peristeri, Attica 121 36, Greece 105.2 12 09:15 10:15
1003 12 Ithakis, Athens, Attica 112 51, Greece 31.8 2.5 08:00 09:00
1004 3 Troias, Athens, Attica 113 62, Greece 11.2 33.2 13:00 15:00
xxxx 6 Ippokratous, Athens, Attica 114 72, Greece 78.4 38.6 12:30 14:00

Table 2
Data of vehicles.

Type Gross weight (tn) Load Capacity (tn) Fixed cost (€) Variable cost (€/km) CO2 emissions (gr/tkm)

Petrol small van 1.2 0.795 14 0.101 143.4

Diesel small van 1.2 0.728 19 0.055 109.2
Large van 1.6 1.2 22 0.129 137.4
Semi-light trucks 3.0 3.5 32 0.189 107.6
Light trucks 5.0 7.5 42 0.252 86.0

Table 3
Computational results.

Case # of customers Total distance Total cost # of vehicles Emissions (kg of CO2 ) Load Factor

3PL C1 137 1722 500.37 € 16 443.48 90.88%

3PL C2 385 4251 1299.86 € 33 1244.73 84.57%
3PL C3 226 2918 812.73 € 21 793.60 88.47%
Cumulative 748 8891 2612.96 € 70 2481.81 87.16%
Collaborative 748 7992 2527.18 € 66 2438.36 83.38%
Deviation – −10.11% −3.28% −5.71% −1.75% −4.35%

data of vehicles used in the distribution processes of all 3PL companies
are given in Table 2.
As for the CO2 emissions due to the use of vehicles, most man-
ufacturers offer relevant information, such as the grammars of CO2
emitted for every tone transported over a distance of one kilometer,
in urban environment. Consequently, the emissions depend on the fac-
tor 𝑐 𝑘 (gr/tkm), the gross weight of the vehicle 𝑤𝑘 , the load of goods
on board on every arc (i, j), 𝑢𝑘𝑖𝑗 , as well as on the distance of the
arc, 𝑑𝑖𝑗 . The mathematical expression that calculates CO2 emissions is
∑
[(𝑢𝑘𝑖𝑗 + 𝑤𝑘 ) ⋅ 𝑐 𝑘 ⋅ 𝑑𝑖𝑗 ⋅ 𝑥𝑘𝑖𝑗 ], and is valid for every vehicle 𝑘 ∈ K. This
(𝑖,𝑗 )∈𝐴
approach offers a good estimation of CO2 emissions, even if in fact the
speed, the age, as well as other factors affect the emissions (Koç et al.,
2014).
Concluding, all the data that are presented in Tables 1 and 2, are
essential and significant for distribution companies that address the
HFVRP, the VRPSPD and the VRPTW. The specific VRP variants corre-
spond to real-life problems and cases, that most of the Greek companies
operating in the field of logistics face. Irrespective of whether compa-
nies operate individually and execute single routes, or collaborate and
share their resources, the needed data remain the same for optimizing
the routing and scheduling problem.
4.2. Computational results
Initially, multiple daily distribution cases, that each 3PL company
faces, have been addressed and solved by the proposed heuristic algo-
rithm. The daily schedules and results produced from each one of the
cases, are analyzed, offering the average cost, emissions, number of ve-
hicles, vehicle load etc. Therefore, the results and the mean values, that
refer to each 3PL company individually, correspond to a daily base, and
are indicated as 3PL C1, 3PL C2, and 3PL C3 respectively, as shown in
Table 3. Moreover, the aggregated results, of the three 3PL companies,
are presented in the row “Cumulative results” of Table 3.
In addition, the daily distribution cases of all companies are united
and form more complex ones, that are addressed by the algorithm as the
collaborative distribution cases. The results of the multiple collaborative
logistics cases, in which the depots and the customers, of all companies,
are simultaneously considered, also offer mean values (in cost, emis-
sions, vehicles, load factor etc.) that are given in “Collaborative case”
row. Finally, the results exploited both from the collaborative case and
from the cumulative case, are compared, while their deviations, in mul-
tiple factors, are also calculated.
As shown in Table 3, the first company (3PL C1) serves on average
137 customers, the second one (3PL C2) 385, and the third one (3PL C3)
226, in a daily base. From the results of the three cases, it is easy to say
that as the number of customers increases, the distribution cost, the total
traveled distance, the number of vehicles needed, and the emissions also
increase. Instead, the mean load factors in all 3PL companies remain at
the same level, about 75–80%. On the other hand, on the collaborative
case, the average number of customers served is 748, which is equal to
those in the cumulative case, while the depots from which vehicles load
the goods, are three.
The comparison of results, obtained from the cumulative and the col-
laborative case, indicate that the collaboration between 3PL companies
can offer improvements in multiple levels. The first aspect that is greatly
improved, and constitutes the main objective that 3PL companies need
to minimize, is the distribution cost, in which a reduction of 3.28% is
observed. In addition, the emissions are also decreased (−1.75%), reduc-
ing them by 42.45 kg daily. Despite the fact that the reduction seems
small, and considering that the companies operate at least 300 days per
year, annually 12.735 tons of CO2 can be avoided. This would be hard to
happen without the reduction in the total traveled distance (−10.11%),
and in the number of vehicles needed (−5.71%). Finally, the only fac-
tor that is worsening through collaboration is the load factor, and more
specifically from 87.16%, the load factor decreased to 83.38%, which
means that there is more unused space on vehicles.
Additionally, there is a significant decrease in the total distribution
costs, as the three companies save cumulatively 85.78 € every day, that
correspond to 25,725.00€ yearly. However, the sharing of each com-
pany in the total distribution cost, and consequently to the savings dif-
fers. Multiple factors may affect the sharing of each company, such as

5
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al.
the number of customers and the vehicles belonging in each company,
and involving in the collaborative routing and scheduling procedures.
Therefore, an agreement between the collaborative companies, that de-
termines the sharing of each company to the total cost, is essential.
5. Discussion
5.1. Contribution to theory
As found in this study, the competition between logistics compa-
nies in Greece, as well as globally is severe. This has led companies to
explore the processes that can be optimized and the technologies that
need to be applied, in order to survive and operate more efficiently.
The routing of vehicles is such an operation that, in most cases, is ad-
dressed by advanced systems. However, many companies have come
to the conclusion that for maximizing cost savings, they need to explore
new methods for more efficient operations. Collaborative routing is such
an approach, that while initially was not accepted by practitioners, has
now been recognized and companies have a positive attitude towards it
(Rodrigues et al., 2018).
Most collaborative routing studies focus on developing complex
mathematical models and advanced optimization methods in order to
minimize the distribution cost and the travelled distance. Zhang et al.
(2020) for example, propose a collaborative model similar to ours, with-
out considering the time windows and the environmental impact, as
well as an extended variable neighborhood search algorithm. The ex-
perimental results showed that the algorithm addressing the specific
collaborative case offers cost reduction that ranges from 3.19% up to
37.18% depending on the case. Similar results are obtained in the re-
search of Dahl and Derigs (2011), where the data of 50 express carriers
were simulated, indicating that the cost can reduce up to 14% by shar-
ing customer requests. Other studies also come to the conclusion that
the distribution cost, as well as the total travelled distance can be re-
duced, offering higher profit margin to the companies that plan routing
jointly (Pradenas, Oportus & Parada, 2013). However, all the aforemen-
tioned studies have not been tested in real life-cases. Moreover, over the
last few years the need for reducing the emissions has also emerged due
to the global warming effect. This has led researchers to focus on this
attribute and define how collaborations can enhance sustainable dis-
tribution and logistics (Perez-Bernabeu, Juan, Faulin & Barrios, 2015;
Pradenas et al., 2013). In this content, Danloup et al. (2015) studied the
effects of collaborative distribution on the environment. Their research
focuses on food distribution and their mathematical formulation aims
to minimize the CO2 emissions instead of the cost. The computational
results indicate that the average emissions and cost reduction is 9.0 and
12.0 percent respectively.
5.1. Implications for practice
Our research gives equivalent results to any of the aforementioned
studies but at the same time provides a novel that collaborative rout-
ing can have great impact both on cost (25,725.00€ every year) and
emissions (12.735 tons of CO2). Additionally, besides the variables and
constraints that are considered in each case, as well as the sector, the
results can greatly be affected by the method of calculating the cost
and emissions. In the current research, we consider that both meth-
ods are quite accurate since all the factors affecting the cost and emis-
sions are considered and integrated into the formulation. As for the im-
plications to practice, while over the years the interest on collabora-
tive logistics increases and researchers offer novel mathematical mod-
els and algorithms, software solutions and platforms to support this
concept have not followed the same path. Therefore, the next step of
our research is to exploit this exact research gap in order to investigate
whether the algorithm and the model can be integrated into a routing
system.
6
International Journal of Information Management Data Insights 1 (2021) 100010
6. Conclusions
The conducted research focused on a routing and scheduling ap-
proach that offered significant reductions, both in terms of distribution
cost, and CO2 emissions. Our approach proposed the collaboration be-
tween logistics companies that are located in the same area, and simul-
taneously distribute goods in the same city. Their collaboration, focus-
ing on the last mile distribution, included sharing customer requests,
resources and information, and aimed to minimize the cost.
In order to estimate the results of the proposed approach, we con-
sidered the cases of three medium size 3PL companies, located in the
western suburbs of the city of Athens. Their daily distribution cases were
tested in two different scenarios, one where each company operates and
distributes goods separately, and a second one where companies share
their fleets and combine their customers’ demands. The results showed
that addressing the routing and scheduling collaboratively, can offer
significant reduction in cost and emissions. In fact, this synergy can be
beneficial not only for logistics companies, but also for city inhabitants,
as pollution (CO2 emissions), noise and traffic congestion (number of
vehicles) in city centers are also decreased.
A further research study should focus on the profit-sharing mecha-
nism, that the collaborative companies need to define before their co-
operation. In each collaboration case, the profit-sharing approach needs
to be fair for all parties, while also being profitable. However, the fact
that profit-sharing constitutes another research topic, and multiple ap-
proaches have already been proposed, led us in not considering it as part
of our research study.
Finally, in order for companies to further benefit from such a syn-
ergy, an advanced system needs to be exploited. The system needs to
be efficient both in terms of computational time and solution quality,
since an increased number of customers’ deliveries are considered in
a daily phase. Additionally, the system must handle data from multiple
logistics providers, in compliance with the General Data Protection Reg-
ulation (GDPR). This is essential since data of customers from different
companies, need to be uploaded in a shared platform for addressing the
Collaborative VRP. This further research can provide logistics compa-
nies with a useful tool for more efficient last-mile distribution.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
References
Baldacci, R., Battarra, M., & Vigo, D. (2008). Routing a heterogeneous fleet of ve-
hicles. In B. Golden, S. Raghavan, & E. Wasil (Eds.), The Vehicle Routing Prob-
lem: Latest Advances and New Challenges (pp. 3–27). Boston, MA: Springer US.
10.1007/978-0-387-77778-8_1.
Belmecheri, F., Prins, C., Yalaoui, F., & Amodeo, L. (2013). Particle swarm optimiza-
tion algorithm for a vehicle routing problem with heterogeneous fleet, mixed
backhauls, and time windows. Journal of Intelligent Manufacturing, 24, 775–789.
10.1007/s10845-012-0627-8.
Bräysy, O., & Gendreau, M. (2005). Vehicle routing problem with time windows, Part I:
Route construction and local search algorithms. Transportation Science, 39, 104–118.
10.1287/trsc.1030.0056.
Bräysy, O., & Hasle, G. (2014). Software tools and emerging technologies for vehicle rout-
ing and intermodal transportation. In P. Toth, & D. Vigo (Eds.), Vehicle routing, prob-
lems, methods and applications (pp. 351–380). 10.1137/1.9781611973594.ch12.
Chakraborty, A., & Kar, A. K. (2017). Swarm intelligence: A review of algorithms. In
S. Patnaik, X.-S. Yang, & K. Nakamatsu (Eds.), Nature-inspired computing and optimiza-
tion: theory and applications (pp. 475–494). Cham: Springer International Publishing.
10.1007/978-3-319-50920-4_19.
Chiang, T. C., & Hsu, W. H. (2014). A knowledge-based evolutionary algorithm for the
multiobjective vehicle routing problem with time windows. Computers & Operations
Research, 45, 25–37. 10.1016/j.cor.2013.11.014.
Chu, J. C., Yan, S., & Huang, H. J. (2017). A multi-trip split-delivery vehicle routing prob-
lem with time windows for inventory replenishment under stochastic travel times.
Networks and Spatial Economics, 17, 41–68. 10.1007/s11067-015-9317-3.
G.D. Konstantakopoulos, S.P. Gayialis, E.P. Kechagias et al.
Cleophas, C., Cottrill, C., Ehmke, J. F., & Tierney, K. (2019). Collaborative urban trans-
portation: Recent advances in theory and practice. European Journal of Operational
Research, 273, 801–816. 10.1016/j.ejor.2018.04.037.
Cruijssen, F., & Salomon, M. (2004)Empirical study : Order sharing between transportation
companies may result in cost reductions between 5 to 15 percent, CentER Discussion
Paper.
Danloup, N., Mirzabeiki, V., Allaoui, H., Goncalves, G., Julien, D., & Mena, C. (2015).
Reducing transportation greenhouse gas emissions with collaborative dis-
tribution: A case study. Management Research Review, 38, 1049–1067.
10.1108/MRR-11-2014-0262.
de Oliveira, H. C., & Vasconcelos, G. C. (2010). A hybrid search method for the vehicle
routing problem with time windows. Annals of Operations Research, 180, 125–144.
10.1007/s10479-008-0487-y.
Elshaer, R., & Awad, H. (2020). A taxonomic review of metaheuristic algorithms for solv-
ing the vehicle routing problem and its variants. Computers & Industrial Engineering,
140, Article 106242. 10.1016/j.cie.2019.106242.
Erdoğan, G. (2017). An open source spreadsheet solver for vehicle routing problems. Com-
puters & Operations Research, 84, 62–72. 10.1016/j.cor.2017.02.022.
Ferrell, W., Ellis, K., Kaminsky, P., & Rainwater, C. (2020). Horizontal collaboration:
Opportunities for improved logistics planning. International Journal of Production Re-
search, 58, 4267–4284. 10.1080/00207543.2019.1651457.
Gansterer, M., & Hartl, R. F. (2018). Collaborative vehicle routing: A survey. European
Journal of Operational Research, 268, 1–12. 10.1016/j.ejor.2017.10.023.
Jiang, J., Ng, K. M., Poh, K. L., & Teo, K. M. (2014). Vehicle routing problem with a
heterogeneous fleet and time windows. Expert Systems with Applications, 41, 3748–
3760. 10.1016/j.eswa.2013.11.029.
Kar, A. K. (2016). Bio inspired computing – A review of algorithms and scope of applica-
tions. Expert Systems with Applications, 59, 20–32. 10.1016/j.eswa.2016.04.018.
Karakatič, S., & Podgorelec, V. (2015). A survey of genetic algorithms for solv-
ing multi depot vehicle routing problem. Applied Soft Computing, 27, 519–532.
10.1016/j.asoc.2014.11.005.
Kechagias, E. P., Gayialis, S. P., Konstantakopoulos, G. D., & Papadopoulos, G. A. (2020a).
An application of a multi-criteria approach for the development of a process reference
model for supply chain operations. Sustainability, 12, 5791. 10.3390/su12145791.
Kechagias, E. P., Gayialis, S. P., Konstantakopoulos, G. D., & Papadopoulos, G. A. (2020b).
An application of an urban freight transportation system for reduced environmental
emissions. Systems, 8. 10.3390/systems8040049.
Koç, Ç., Bektaş, T., Jabali, O., & Laporte, G. (2014). The fleet size and mix pollution-
routing problem. Transportation Research Part B: Methodological, 70, 239–254.
10.1016/j.trb.2014.09.008.
Konstantakopoulos, G. D., Gayialis, S. P., & Kechagias, E. P. (2020). Vehicle routing prob-
lem and related algorithms for logistics distribution: A literature review and classifi-
cation. Operational Research - An International Journal. 10.1007/s12351-020-00600-7.
Konstantakopoulos, G. D., Gayialis, S. P., Kechagias, E. P., Papadopoulos, G. A., & Tat-
siopoulos, I. P. (2020). A multiobjective large neighborhood search metaheuristic for
the vehicle routing problem with time windows. Algorithms, 13. 10.3390/a13100243.
Konstantakopoulos, G. D., Gayialis, S. P., Kechagias, E. P., Papadopoulos, G. A., & Tat-
siopoulos, I. P. (2021). Delivering and picking goods under time window restric-
tions: An effective evolutionary algorithm. Journal of Intelligent & Fuzzy Systems, 1–14
Preprint. 10.3233/JIFS-202129.
Krajewska, M. A., Kopfer, H., Laporte, G., Ropke, S., & Zaccour, G. (2008). Horizontal
cooperation among freight carriers: Request allocation and profit sharing. Journal of
the Operational Research Society, 59, 1483–1491. 10.1057/palgrave.jors.2602489.
Laporte, G. (2007). What you should know about the vehicle routing problem. Naval Re-
search Logistics, 54, 811–819. 10.1002/nav.20261.
Mekamcha, K., Souier, M., Bessenouci, H. N., & Bennekrouf, M. (2019). Two
metaheuristics approaches for solving the traveling salesman problem: An Al-
gerian waste collection case. Operational Research: An International Journal.
10.1007/s12351-019-00529-6.
Montané, F. A. T., & Galvão, R. D. (2006). A tabu search algorithm for the vehicle rout-
ing problem with simultaneous pick-up and delivery service. Computers & Operations
Research, 33, 595–619. 10.1016/j.cor.2004.07.009.
International Journal of Information Management Data Insights 1 (2021) 100010
Montoya-Torres, J. R., López Franco, J., Nieto Isaza, S., Felizzola Jiménez, H., & Herazo-
Padilla, N. (2015). A literature review on the vehicle routing problem with multiple
depots. Computers & Industrial Engineering, 79, 115–129. 10.1016/j.cie.2014.10.029.
Montoya-Torres, J. R., Muñoz-Villamizar, A., & Vega-Mejía, C. A. (2016). On the impact
of collaborative strategies for goods delivery in city logistics. Production Planning &
Control, 27, 443–455. 10.1080/09537287.2016.1147092.
Ombuki, B., Ross, B. J., & Hanshar, F. (2006). Multi-objective genetic algorithms for ve-
hicle routing problem with time windows. Applied Intelligence, 24, 17–30.
Pan, S., Trentesaux, D., Ballot, E., & Huang, G. Q. (2019). Horizontal collaborative trans-
port: Survey of solutions and practical implementation issues. International Journal of
Production Research, 57, 5340–5361. 10.1080/00207543.2019.1574040.
Perez-Bernabeu, E., Juan, A. A., Faulin, J., & Barrios, B. B. (2015). Horizontal coop-
eration in road transportation : A case illustrating savings in distances and green-
house gas emissions. International Transactions in Operational Research, 22, 585–606.
10.1111/itor.12130.
Polat, O., Kalayci, C. B., Kulak, O., & Günther, H.-. O. (2015). A perturbation based variable
neighborhood search heuristic for solving the vehicle routing problem with simulta-
neous pickup and delivery with time limit. European Journal of Operational Research,
242, 369–382. 10.1016/j.ejor.2014.10.010.
Pradenas, L., Oportus, B., & Parada, V. (2013). Mitigation of greenhouse gas emissions
in vehicle routing problems with backhauling. Expert Systems with Applications, 40,
2985–2991. 10.1016/j.eswa.2012.12.014.
Rodrigues, M., Zampou, E., Zeimpekis, V., Stathacopoulos, A., Teoh, T., & Ayfan-
topoulou, G. (2018). Cooperative models for addressing urban freight challenges: The
NOVELOG and U-TURN approaches. In E. Taniguchi, & E. Thompson (Eds.), City Lo-
gistics 3: Towards Sustainable and Liveable Cities (pp. 215–234). Ltd: John Wiley & Sons.
10.1002/9781119425472.ch12.
Santos, M. J., Curcio, E., Amorim, P., Carvalho, M., & Marques, A. (2021). A bilevel ap-
proach for the collaborative transportation planning problem. International Journal of
Production Economics, 233, Article 108004. 10.1016/j.ijpe.2020.108004.
Santos, M. J., Martins, S., Amorim, P., & Almada-Lobo, B. (2021). A green lateral collabo-
rative problem under different transportation strategies and profit allocation methods.
Journal of Cleaner Production, 288, Article 125678. 10.1016/j.jclepro.2020.125678.
Solomon, M. M. (1987). Algorithms for the vehicle routing and scheduling problems with
time window constraints. Operations Research, 35, 254–265. 10.1287/opre.35.2.254.
Soysal, M., Bloemhof-Ruwaard, J. M., Haijema, R., & van der Vorst, J. G. A. J. (2018).
Modeling a green inventory routing problem for perishable products with
horizontal collaboration. Computers & Operations Research, 89, 168–182.
10.1016/j.cor.2016.02.003.
Subramanian, A., Penna, P. H. V., Uchoa, E., & Ochi, L. S. (2012). A hybrid algorithm
for the Heterogeneous Fleet Vehicle Routing Problem. European Journal of Operational
Research, 221, 285–295. 10.1016/j.ejor.2012.03.016.
Verdonck, L., Caris, A. N., Ramaekers, K., & Janssens, G. K. (2013). Collaborative logistics
from the perspective of road transportation companies. Transport Reviews, 33, 700–
719. 10.1080/01441647.2013.853706.
Wang, X., Battarra, M., Golden, B., & Wasil, E. (2015). Routing, vehicle, and scheduling.
London, U. K.: Routledge Handbook. 10.4324/9781315756684.
Wang, Y., Zhang, J., Assogba, K., Liu, Y., Xu, M., & Wang, Y. (2018). Collabo-
ration and transportation resource sharing in multiple centers vehicle routing
optimization with delivery and pickup. Knowledge-Based System, 160, 296–310.
10.1016/j.knosys.2018.07.024.
Xiao, Y., & Konak, A. (2016). The heterogeneous green vehicle routing and scheduling
problem with time-varying traffic congestion. Transportation Research Part E: Logistics
and Transportation Review, 88, 146–166. 10.1016/j.tre.2016.01.011.
Xu, Z., Elomri, A., Pokharel, S., & Mutlu, F. (2019). A model for capacitated green ve-
hicle routing problem with the time-varying vehicle speed and soft time windows.
Computers & Industrial Engineering, 137, Article 106011. 10.1016/j.cie.2019.106011.
Zhang, W., Chen, Z., Zhang, S., Wang, W., Yang, S., & Cai, Y. (2020). Composite
multi-objective optimization on a new collaborative vehicle routing problem with
shared carriers and depots. Journal of Cleaner Production, 274, Article 122593.
10.1016/j.jclepro.2020.122593.