Proceedings of the International Conference on Industrial Engineering and Operations Management

Rome, Italy, August 2-5, 2021

A Practical Green Vehicle Routing Solution –

A Case in Ho Chi Minh City
Quang Pham Ngoc, Ha Thi Xuan Chi, Uyen Ngo Thi Thao,
Nguyen Huy Thien Phuc, Anh Duong Vo Nhi
School of Industrial Engineering and Management
International University – Vietnam National University Ho Chi Minh City
Ho Chi Minh City, Vietnam
[Link]@[Link], htxchi@[Link], ngothithaouyen94@[Link],
[Link]@[Link], dvnanh@[Link]

Abstract

Energy overuse in transportation is recently generating harmful effects to the environment. Green logistics is becoming
a prospective trend in the growing logistics industry. This paper investigates the green vehicle routing problem (Green-
VRP), with two objectives: minimizing combined fuel emissions and minimizing distance traveled with capacitated
vehicles. An improved K-means clustering method is employed to group demand nodes to reduce the size of the
problem. The model is then verified using a set of customers in Ho Chi Minh city, computational results show 3
clusters of customers. In addition, a real application for mobile phones namely “Control Fleet Vehicle” is created to
help users control their fleets of vehicles and connect easily to customers.

Keywords
Green vehicle routing problem, Improved k-means clustering, Fuel emission minimization

1. Introduction
Transportation really plays a particularly vital role in logistics and this role will increase because the cost of transport
accounts for an increasing proportion of logistics. Therefore, the logistics organizations try to bring higher production
efficiency and competitiveness of logistics by reducing transportation costs. However, quick development of
transportation sector also comes along negative impacts on society, health, and environment which are the
consequences of increases in the abundance of hazardous emissions. Due to the rapid development of e-commerce,
which leads to the presence of the transport occur more and more, the transportation industry has been one of the main
sources of increased energy consumption and carbon emission in recent years. This urges the governments and
scholars to deploy a new green-logistic supply chain system, reducing fuel emissions in transportation service.

Inspired by the traditional Vehicle Routing Problem (VRP), which was first formulated mathematically by (Dantzig
& Ramser, 1959) thereby studied by numerous researchers throughout the field of transportation, this study extends
the problem to Green Vehicle Routing Problem (G – VRP), aiming at contributing the recent environmental awareness
in the field of transportation in logistic by combining fuel consumption and travelled distance optimization. More
specifically, the G-VRP is applied within the scope of transportation service case in Ho Chi Minh city – one of the
biggest cities in Vietnam. The is now producing huge wellspring of greenhouse gas emissions (GHG, accounting for
about 16% of national emissions, in which GHG from the transport sector accounts for 45% ((MONRE) & DCC,
2015).

Objectives:
The purpose of the study is to minimize the total distance travelled and the total fuel consumption by vehicles to
increase not only customer satisfaction but also sustainable development of company’s transportation service. The
fuel utilization minimization is the significant goal along with the distance minimization. Therefore, appropriate
weights are given to each of the sub-objective. The solution is a set of routes by that each customer will be served
only once and concerning vehicle capacity constraints.

© IEOM Society International 2432

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

2. Literature Review
The vehicle routing problem is dealing with the problem of identifying the optimal series of routes with possible
minimal travelling time or travelling total cost, considering a set of customer nodes and depot(s) from which the
vehicle departs to visit each customer once, subjected to a set of constraints. The VRP has been developed to many
variants such as Capacitated VRP, VRP with time-windows, VRP with Backhauls, VRP with Pickup and Delivery
and some others. The Green Vehicle Routing Problem is also a variant of VRP, focusing on managing environmental
effect of vehicles during transportation. There are a number of literatures working on the Green Vehicle Routing
Problem which has been performed since 2006.

The VRP with green transportation was first developed by Sbihi & Eglese (2007) who consider the basic Vehicle
Routing and Scheduling Problem model that relate to environmental problems with the time-dependent, the
transportation of hazardous materials. Kara et al., (2007) introduced the so-called Energy-Minimizing Vehicle
Routing Problem which is an extension of the VRP where a weighted load function (load multiplied by distance),
rather than just the distance, is minimized. The authors present a model for this problem, based on a flow formulation
of the VRP with a load-based objective function derived from simple physics. However, the cost function lacks further
measurement of fuel consumption or emission. Palmer (2007) developed an integrated routing and carbon dioxide
emission model for a goods vehicle. In the transportation problem, the main objective is to minimize the total distance
travelled by trucks, ships, planes, and trains to transport the goods. This kind of transportation generates a lot of carbon
dioxide in the environment. The especially urban city is more polluted due to large carbon dioxide, which has been
received significant attention from social participants.

Xiao et al., (2012) assumed the fuel consumption rate be a linear function of load and built an integer programming
model containing fixed cost and fuel cost. Bektaş et al., (2016) indicated that the amount of pollution emitted by a
vehicle depends on its load and speed. They also considered the fuel cost, CO2 emission cost and driver's salary in the
objective, and proposed an integer linear programming model along with a heuristic algorithm to deal with the
Pollution-Routing Problem. Elbouzekri et al., (2013) reported the formation to estimate the emission of the trucks.
Tiwari & Chang (2015) used clustering and genetic algorithms to solve GVRP. The objective function considered the
minimum distance travelled by each vehicle as well as the total emitted CO2, which was directly assumed to be the
product of mean distance, load and conversion factor. Jabali et al., (2012) proposed a model that considers CO2
emission-related costs, including fuel consumption, travel time, and time-dependent emissions, and the model is
solved using a TS procedure. Qian and Eglese (2016) addressed the optimization of fuel emissions in VRPs with time-
varying speeds, considering the capacities of the vehicles and the time constraints on the total length of each route.
As the G-VRP is NP-hard problem just like traditional VRP, the small-sized problem could be solved effectively
within short time by exact solution methods, which have been researched and proposed by many scholars (Baldacci
et al., 2010; Kallehauge, 2008; Qureshi et al., 2009; Toth & Vigo, 1998). For node sets of large-size G-VRP, various
heuristics as well as metaheuristics have been investigated to achieve near-optimal results (Andelmin & Bartolini,
2019; Felipe et al., 2014; Montoya et al., 2016; Normasari et al., 2019; Olgun et al., 2021). Recently, many studies
have been considered to the G-VRP by using various approaches : the hybrid heuristic algorithm of Hooshmand &
MirHassani (2019) for a time-dependent G-VRP, differential evolution algorithm and Particle swarm optimization
(Fallah et al., 2019), ant-colony optimization for multi-depot and multi-objectives G-VRP (Li et al., 2019), Non-
Dominated Sorting Genetic Algorithm II (Dutta et al., 2021).

3. Mathematical Model
The G-VRP in this study can be expressed as a routes optimization problem with objective of minimizing fuel
emissions and travelled distance over a set of customer service network G = (V, A), where V = {0, 1, …, n} is a set
of n customer nodes, n = 0 is the depot from which each truck must start and return at the end of the journey after
visiting and satisfying demand of customer nodes in set N = {1, …, n}. Each customer node is served exactly once
by one truck, and all the trucks have the same capacity. Therefore, customer demand of each node must be less than
the capacity of the truck. A daytime is split into many time periods p with different vehicle speeds due to different
traffic conditions.

3.1 Comprehensive Modal Emission Model

In this section, fuel consumption is calculated by using the comprehensive modal emission model (CMEM - Niu et
al., (2018))

© IEOM Society International 2433

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

𝑘𝑘∗𝑁𝑁∗𝑉𝑉∗𝑑𝑑
𝐹𝐹 = 𝜆𝜆 ∗ ( + 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 ∗ 𝛾𝛾 ∗ 𝛼𝛼 ∗ 𝑑𝑑 + 𝛽𝛽 ∗ 𝛾𝛾 ∗ 𝑑𝑑 ∗ 𝑣𝑣 2 )
𝑣𝑣
𝜉𝜉 1
where 𝜆𝜆 = ; 𝛾𝛾 =
𝜅𝜅∗𝜓𝜓 1000∗𝑛𝑛𝑡𝑡𝑡𝑡 ∗𝜂𝜂
and 𝛼𝛼 = 𝜏𝜏 + 𝑔𝑔 ∗ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑔𝑔 ∗ 𝐶𝐶𝑟𝑟 ∗ 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝐶𝐶 ∗𝜌𝜌∗𝐴𝐴
𝛽𝛽 = 𝑑𝑑
2
τ is acceleration and 𝜽𝜽 is the road angle inclination, assuming that acceleration and road inclination are zero (Niu et
al., 2018).

Weight means the total vehicle weight including the weight of vehicle and goods, while 𝒗𝒗 represents the vehicle speed,
it can be calculated as the distance over the travelling time. Vehicle speed 𝒗𝒗 is obtained by traveling distance and
traveling time. Parameters are used to calculate fuel consumption are shown in Table 1. The specific parameters of
the vehicle are listed in Table 2 (Niu et al., 2018). In the original model of Niu and colleagues, 3 types of vehicle duty
are tested: light, medium and heavy duty. The results show that for 60-nodes problem and smaller, light-duty trucks
give better results of smaller cost. As our problem is considering set of nodes smaller than 60, light-duty vehicles are
assumed, given specific parameters in Table 2.

Table 1: Parameters for fuel consumption calculation

Notation Description Typical values
ξ Fuel-to-air mass ratio 1
g Gravitational constant (m/s2) 9.81
ρ Air density (kg/m2) 1.2041
Cr Coefficient of rolling resistance 0.01
η Efficiency parameter for diesel engines 0.45
fc Fuel and CO2 emissions cost (RMB/liter) 12.0165
κ Heating value of a typical diesel fuel (kj/g) 44
ψ Conversion factor (g/s to L/s) 737
𝑛𝑛𝑡𝑡𝑡𝑡 Vehicle drive train efficiency 0.45

Table 2: Vehicle specific parameters

Notation Description Typical values
w Curb weight (kg) 3500
Q Maximum payload (kg) 4000
k Engine friction factor (kj/rev/liter) 0.25
N Engine speed (rev/s) 38.34
V Engine displacement (liter) 4.5
Cd Coefficient of aerodynamics drag 0.6
A Frontal surface area (m ) 2
7

3.2 Green Vehicle Routing Mathematical Model

The model was developed to satisfy all constraints about the customers’ demand, the fuel consumption depending on
the total distance vehicles travel and the time restrictions for trucks in Ho Chi Minh city. The fuel consumption
objective is inherited from the CMEM of Niu et al., (2018) and adapted from Feng et al., (2017), being shown in the
following section.

Let G = (V, A) be a customer service network, where V = {0, 1, …, n} is a set of n customer nodes, n = 0 is the
warehouse.
N = {1, …, n} is a set of customer nodes
A = {(i, j) ∈ V x V} is the set of arcs
D = {dij | (i, j) ∈ A} is the set of travelling distance
M: a huge number
T: a set of period time

© IEOM Society International 2434

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

𝑞𝑞𝑖𝑖 : demand of customers

capa: Capacity of the truck

Decision variables:
𝑝𝑝
𝑥𝑥𝑖𝑖𝑖𝑖 = 1 if vehicle visits from customer i to j in period p
𝑝𝑝
𝑥𝑥𝑖𝑖𝑖𝑖 = 0 otherwise
𝑝𝑝
𝑢𝑢𝑖𝑖𝑖𝑖 : the quantity goods retained by vehicle from customer i to j in period p
𝑡𝑡𝑖𝑖 : the departure time at node i
𝑝𝑝
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚_𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑖𝑖𝑖𝑖 : the travelling time from customer i to customer j in period p

Objective function:
𝑝𝑝 𝑝𝑝
𝑀𝑀𝑀𝑀𝑀𝑀 𝑍𝑍1 = 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = � � � 𝜆𝜆 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖 ∗ 𝑘𝑘1 ∗ 𝑁𝑁 ∗ 𝑣𝑣1∗ 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚_𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑖𝑖𝑖𝑖
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇

𝑝𝑝
+ � � � 𝑑𝑑𝑖𝑖𝑖𝑖 ∗ 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 ∗ 𝛾𝛾 ∗ 𝜆𝜆 ∗ 𝑔𝑔 ∗ 𝐶𝐶𝑟𝑟 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇
𝑝𝑝 1 𝑑𝑑𝑖𝑖𝑖𝑖 2
+ � � � 𝑑𝑑𝑖𝑖𝑖𝑖 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖 ∗ ∗ 𝐶𝐶𝑑𝑑 ∗ 𝜌𝜌 ∗ 𝐴𝐴 ∗ 𝛾𝛾 ∗ 𝜆𝜆 ∗ ( 𝑝𝑝 )
2 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚_𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑖𝑖𝑖𝑖
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇
𝑝𝑝
+ � � � 𝑑𝑑𝑖𝑖𝑖𝑖 ∗ 𝑢𝑢𝑖𝑖𝑖𝑖 ∗ 𝛾𝛾 ∗ 𝜆𝜆 ∗ 𝑔𝑔 ∗ 𝐶𝐶𝑟𝑟
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇

𝑝𝑝
𝑀𝑀𝑀𝑀𝑀𝑀 𝑍𝑍2 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = � � � 𝑑𝑑𝑖𝑖𝑖𝑖 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇
Subject to:
𝑝𝑝
� � 𝑥𝑥𝑖𝑖𝑖𝑖 = 1 (∀ 𝑗𝑗 ∈ 𝑉𝑉, 𝑖𝑖 ≠ 𝑗𝑗) (1)
𝑖𝑖∈𝑉𝑉 𝑝𝑝∈𝑇𝑇

� � 𝑥𝑥𝑖𝑖𝑖𝑖
𝑝𝑝
= 1 (∀ 𝑖𝑖 ∈ 𝑉𝑉, 𝑖𝑖 ≠ 𝑗𝑗, 𝑖𝑖 ≠ 0) (2)
𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇
1
� 𝑥𝑥𝑖𝑖𝑖𝑖 = 1 (∀ 𝑖𝑖 ∈ 𝑉𝑉, 𝑖𝑖 ≠ 𝑗𝑗, 𝑖𝑖 = 0) (3)
𝑗𝑗∈𝑉𝑉
𝑝𝑝
� � 𝑢𝑢𝑖𝑖𝑖𝑖 − � � 𝑢𝑢𝑗𝑗𝑗𝑗 = 𝑞𝑞𝑖𝑖 (∀𝑗𝑗 ∈ 𝑁𝑁)
𝑝𝑝 (4)
𝑖𝑖∈𝑉𝑉,𝑖𝑖≠𝑗𝑗 𝑝𝑝∈𝑇𝑇 𝑖𝑖∈𝑁𝑁,𝑖𝑖≠𝑗𝑗 𝑝𝑝∈𝑇𝑇
𝑝𝑝
𝑝𝑝 (5)
� � � 𝑢𝑢 ≤ 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖 ( 𝑖𝑖 ≠ 𝑗𝑗)
𝑖𝑖∈𝑉𝑉 𝑗𝑗∈𝑉𝑉 𝑝𝑝∈𝑇𝑇 𝑖𝑖𝑖𝑖
𝑡𝑡0 = 0 (6)
𝑝𝑝
𝑡𝑡𝑖𝑖 − 𝑈𝑈𝑝𝑝−1 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖 ≥ 0 (∀𝑖𝑖 ∈ 𝑁𝑁, ∀𝑗𝑗 ∈ 𝑉𝑉, ∀𝑝𝑝 ∈ 𝑇𝑇, 𝑖𝑖 ≠ 𝑗𝑗) (7)
𝑝𝑝
𝑡𝑡𝑖𝑖 + 𝑀𝑀 ∗ 𝑥𝑥𝑖𝑖𝑖𝑖 ≤ 𝑈𝑈𝑝𝑝 + 𝑀𝑀 (∀𝑖𝑖 ∈ 𝑁𝑁, ∀𝑗𝑗 ∈ 𝑉𝑉, ∀𝑝𝑝 ∈ 𝑇𝑇, 𝑖𝑖 ≠ 𝑗𝑗) (8)
𝑝𝑝 𝑝𝑝
𝐼𝐼𝐼𝐼 𝑥𝑥𝑖𝑖𝑖𝑖 = 1, 𝑡𝑡𝑗𝑗 = 𝑡𝑡𝑖𝑖 + 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚_𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑖𝑖𝑖𝑖 (9)
𝑝𝑝
𝑥𝑥𝑖𝑖𝑖𝑖 ∈ {0; 1}(∀𝑖𝑖 ∈ 𝑉𝑉, ∀𝑗𝑗 ∈ 𝑉𝑉, 𝑖𝑖 ≠ 𝑗𝑗) (10)
𝑡𝑡𝑖𝑖 ≥ 0 (∀𝑖𝑖 ∈ 𝑉𝑉) (11)
𝑝𝑝
𝑢𝑢𝑖𝑖𝑖𝑖 ≥ 0 (∀𝑖𝑖 ∈ 𝑉𝑉, ∀𝑗𝑗 ∈ 𝑉𝑉, ∀𝑝𝑝 ∈ 𝑇𝑇, 𝑖𝑖 ≠ 𝑗𝑗) (12)

The first objective is to minimize the fuel consumption considering the speed of the vehicle related to the traffic jams
on the road, while the second one minimizes the total distance that vehicle moves between customers during the trip.
Constraint 1 defines that the vehicle must return to depot from the last customer. Constraint 2 shows that each customer
is served exactly once. Constraint 3 defines that the first customer must be visited from the depot in period one.
Constraint 4 shows the flow of goods containing on the vehicle when it leaves depot until it returns depot. Constraint
5 guarantees that the goods retain on a vehicle always less than its capacity. Constraint 6 sets the departure time at the
depot is 0. Constraint 7 and constraint 8 identify that the departure time at customer i belongs to the upper and lower

© IEOM Society International 2435

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

bound of a period time. Constraint 9 calculates the time of the departure time of customer depending on the state of
the traffic jams. While constraint 10, constraint 11 and constraint 12 restrict the condition of all decision variables.

4. Methods
Instead of solving a large problem, Geetha et al. used the approach for capacitated clustering for dividing the big
cluster to many smaller clusters (Geetha et al., 2009). To find the routing for each vehicle, the weight-sum technique
is applied to solve the multi-objective model. The weight of each objective can be defined by decision makers
depending on their strategies.

5. Numerical Example
In this study, the instance set is provided to illustrate the efficient of model. This data set includes 76 customers who
are in Ho Chi Minh City and only 4-ton truck is considered. The demand customers are shown in Table 3.

Figure 1: Traveling time from node A to node B with different departure time
Table 3. Customer demand

Customer demand Customer demand Customer demand

No. No. No.
(ton) (ton) (ton)
1 0.13 27 0.15 52 0.15
2 0.19 28 0.2 53 0.12
3 0.12 29 0.12 54 0.13
4 0.17 30 0.18 55 0.17
5 0.11 31 0.12 56 0.15
6 0.1 32 0.19 57 0.19
7 0.16 32 0.18 58 0.19
8 0.19 33 0.17 59 0.11
9 0.11 34 0.12 60 0.18
10 0.15 35 0.11 61 0.18
11 0.19 36 0.14 62 0.14

© IEOM Society International 2436

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

12 0.11 37 0.16 63 0.13

13 0.12 38 0.15 64 0.2
14 0.11 39 0.11 65 0.11
15 0.17 40 0.16 66 0.13
16 0.12 41 0.16 67 0.16
17 0.17 42 0.16 68 0.15
18 0.14 43 0.19 69 0.19
19 0.11 44 0.11 70 0.19
20 0.16 45 0.12 71 0.19
21 0.1 46 0.15 72 0.11
22 0.11 47 0.1 73 0.16
23 0.19 48 0.18 74 0.13
24 0.1 49 0.17 75 0.19
25 0.2 50 0.2 76 0.17
26 0.14 51 0.18

Illustrated by Figure 1, the departure time of vehicle changes during the daytime and starts at [Link]. The travelling
time from depot to customers and from customers to customers with different departure time is obtained through
Google Application Programming Interface (Google API). Traveling time may be changed depending on vehicle
departure time because of traffic –jam. If a vehicle leaves node A from 7 a.m. to 8 a.m., traveling time is 30 minutes
to reach node B. While the departure time at node A is from 8 a.m. to 9 a.m., it takes 40 minutes for a trip AB (see
Fig 1).

6. Results
By using Improved K-Mean Clustering approach, 76 customers are grouped to 3 clusters around the city. There are
25 nodes in cluster 1 and 2, while 27 nodes in group 3. In this case, as the focus of this G-VRP is more about the
environment, the weight chosen for the first objective is 0.7, while the second one is 0.3. The objective weights can
be defined based on the importance of objective. The optimal routings for vehicles, the minimized distances and
minimized fuel consumptions from 3 vehicles throughout 3 clusters are shown in Table 4. The first vehicle is expected
to finish more than 90 kilometers when generating 9.619 litters of gases, while the second is expected to generate 6.28
litters of gases after traveling a route of 63.8 kilometers. For cluster 3, even though the number of nodes is the highest
among 3, travel distance and emission amount is lower than that of cluster 1. However, computing time of the
optimizer for this instance is 3.39 minutes, the longest among 3 cases.

© IEOM Society International 2437

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

Table 4. The optimal routings of the 3 clusters

Minimize Minimize fuel
Routing Distance consumption
(km) (litter)
depot, 48, 56, 70, 64, 52, 49, 65, 77, 75, 36, 42, 61, 41,
Cluster 1 90.794 9.619
39, 43, 35, 33, 31, 34, 66, 38, 37, 28, 25, 10, depot
depot, 44, 73, 4, 3, 1, 60, 68, 59, 9, 12, 45, 76, 47, 74,
Cluster 2 63.836 6.28
7, 67, 58, 57, 8, 69, 46, 51, 50, 72, 62, depot
depot 32, 29, 27, 30, 40, 53, 55, 71, 54, 17, 19, 18, 16,
Cluster 3 82.771 8.14
26, 14, 15, 5, 13, 24, 20, 22, 23, 21, 6, 63, 11, 2, depot

6. Practical Implications
To support for drivers and managers, an android application called “Control Fleet Vehicle” is created. Interface of this
application is shown in Figure 2.

Figure 2. Login and Registration Screen for Drivers and Manager

“Control Fleet Vehicle” allows users to register to become a driver or a manager. To be a member, emails and phone
numbers are required and stored into firebase Authentication and Database. Firebase automatically creates UID for
each account (see Fig 3).

Figure 3. Firebase Real-time Database Authentication

© IEOM Society International 2438

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

Figure 4. Manager’s screen Login

Function for Manager:

After logging in with the function of a manager, a new screen appears for a manager to select a driver who will serve
the route. Managers input the sequence locations of the route which obtained from multi-objective model to the
applications (see Fig 5, 6).

Figure 5. Drivers’UID in FRD Figure 6. Putting driver UID into space

Figure 7. All information added by managers

A manager confirms all input information by pressing “CONFIRM”. These data are automatically update into
database (see Fig 8) and routing for delivery is shown for drivers (see Fig.9).

© IEOM Society International 2439

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

Figure 8. Information have put into keys in FRD Figure 9. Routing for drivers

Function for Drivers:

Drivers can see the routing for picking up or delivery products after login. Drivers can accept or reject the
journey. If the journey was finished, a driver selects “Update” to update it to a manager (see Fig 10,11,12).

Figure 10. Screen after Login

Figure 11. Screen “Updating The Journey”

Figure 12. Update information successfully

In addition, “Control Fleet Vehicle” assissts managers keep tracking any drivers by clicking “ checking location of
drivers” function (see Fig 13) and delivery status (see Fig 14) .

© IEOM Society International 2440

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

Figure 13. Checking location

of any drivers

Figure 14. Delivery status

7. Conclusion
In conclusion, sustainable transportation and green logistics have been a widespread topic in the field, attracting
continuous attempts of both industry and academia, to find better approaches to go towards environmentally friendly
economy. This paper contributes a solution method for the Green Vehicle Routing Problem, compromising vehicle
travelled distance and fuel emissions optimization. The computational experiment on a case of Ho Chi Minh city
customer set shows that the above objective is achievable given acceptable computing time, bringing motives for
future investment. This study also provides an application for supporting transportation companies implement
reducing CO2 emission, saving cost and keep tracking fleets of vehicles.

References
(MONRE), M. of N. R. and E., & DCC, D. of C. C. (2015). Project to Support the Planning and Implementation of
NAMAs in a MRV Manner (SPI-NAMA). Japan International Cooperation Agency (JICA).
[Link]
Andelmin, J., & Bartolini, E. (2019). A multi-start local search heuristic for the green vehicle routing problem based
on a multigraph reformulation. Computers & Operations Research, 109, 43–63.
Baldacci, R., Bartolini, E., Mingozzi, A., & Roberti, R. (2010). An exact solution framework for a broad class of
vehicle routing problems. Computational Management Science, 7(3), 229–268.
Bektaş, T., Demir, E., & Laporte, G. (2016). Green vehicle routing. In Green transportation logistics (pp. 243–265).
Springer.
Dantzig, G. B., & Ramser, J. H. (1959). The truck dispatching problem. Management Science, 6(1), 80–91.
Elbouzekri, A., Elhassania, M., & Alaoui, A. E. H. (2013). A hybrid ant colony system for green capacitated vehicle
routing problem in sustainable transport. J. Theor. Appl. Inf. Technol, 54(2).
Felipe, Á., Ortuño, M. T., Righini, G., & Tirado, G. (2014). A heuristic approach for the green vehicle routing
problem with multiple technologies and partial recharges. Transportation Research Part E: Logistics and
Transportation Review, 71, 111–128.
Feng, Y., Zhang, R.-Q., & Jia, G. (2017). Vehicle routing problems with fuel consumption and stochastic travel
speeds. Mathematical Problems in Engineering, 2017.
Geetha, S., Poonthalir, G., & Vanathi, P. T. (2009). Improved k-means algorithm for capacitated clustering problem.
INFOCOMP Journal of Computer Science, 8(4), 52–59.
Jabali, O., Van Woensel, T., & De Kok, A. G. (2012). Analysis of travel times and CO2 emissions in time‐

© IEOM Society International 2441

Proceedings of the International Conference on Industrial Engineering and Operations Management
Rome, Italy, August 2-5, 2021

dependent vehicle routing. Production and Operations Management, 21(6), 1060–1074.

Kallehauge, B. (2008). Formulations and exact algorithms for the vehicle routing problem with time windows.
Computers & Operations Research, 35(7), 2307–2330.
Kara, I., Kara, B. Y., & Yetis, M. K. (2007). Energy minimizing vehicle routing problem. International Conference
on Combinatorial Optimization and Applications, 62–71.
Montoya, A., Guéret, C., Mendoza, J. E., & Villegas, J. G. (2016). A multi-space sampling heuristic for the green
vehicle routing problem. Transportation Research Part C: Emerging Technologies, 70, 113–128.
Niu, Y., Yang, Z., Chen, P., & Xiao, J. (2018). Optimizing the green open vehicle routing problem with time
windows by minimizing comprehensive routing cost. Journal of Cleaner Production, 171, 962–971.
Normasari, N. M. E., Yu, V. F., & Bachtiyar, C. (2019). A simulated annealing heuristic for the capacitated green
vehicle routing problem. Mathematical Problems in Engineering, 2019.
Olgun, B., Koç, Ç., & Altıparmak, F. (2021). A hyper heuristic for the green vehicle routing problem with
simultaneous pickup and delivery. Computers & Industrial Engineering, 153, 107010.
Palmer, A. (2007). An integrated routing model to estimate carbon dioxide emissions from freight vehicles. Logist
Res Network Conference Proceedings; Lalwani, C., Mangan, J., Butcher, T., Coronado-Mondragon, A., Eds,
27–32.
Qian, J., & Eglese, R. (2016). Fuel emissions optimization in vehicle routing problems with time-varying speeds.
European Journal of Operational Research, 248(3), 840–848.
Qureshi, A. G., Taniguchi, E., & Yamada, T. (2009). An exact solution approach for vehicle routing and scheduling
problems with soft time windows. Transportation Research Part E: Logistics and Transportation Review,
45(6), 960–977.
Sbihi, A., & Eglese, R. W. (2007). The relationship between vehicle routing and scheduling and green logistics-a
literature survey.
Tiwari, A., & Chang, P.-C. (2015). A block recombination approach to solve green vehicle routing problem.
International Journal of Production Economics, 164, 379–387.
Toth, P., & Vigo, D. (1998). Exact solution of the vehicle routing problem. In Fleet management and logistics (pp.
1–31). Springer.
Xiao, Y., Zhao, Q., Kaku, I., & Xu, Y. (2012). Development of a fuel consumption optimization model for the
capacitated vehicle routing problem. Computers & Operations Research, 39(7), 1419–1431.

Biographies
Dr. Ha Thi Xuan Chi is a lecturer in School of Industrial Engineering and Management at International University
–VNU HCMC- VietNam. She received her PhD from the National Taiwan University of Science and Technology,
Taipei, Taiwan in 2014. Her main research interest is Fuzzy Theory, Mcdm, VRP, Business Analytic.

MSc. Duong Vo Nhi Anh is a lecturer in School of Industrial Engineering and Management- International University-
VNUHCMC-Vietnam. His research interests are Operation Management, Logistics and Supply Chain Management,
Quality Management and Lean 6-sigma.

MSc. Ngo Thi Thao Uyen is a lecturer in School of Industrial Engineering and Management, International University
- Vietnam National University of HCMC. Her research interests are international logistics and transportation, Data
analytics in Logistics, Logistics games, Supply chain design, Supply chain coordination.

Pham Ngoc Quang received his bachelor’s degree at School of Industrial Engineering and Management, International
University - Vietnam National University of HCMC and now pursuing a master’s degree in National Taiwan
University of Science and Technology.

Nguyen Huy Thien Phuc is a senior student at School of Industrial Engineering and Management, International
University, Vietnam National University, where she is pursuing in a bachelor’s degree in Logistics and Supply Chain
Management. ([Link].). She has accomplished her undergraduate thesis.

© IEOM Society International 2442