sustainability
Article
Container Shipping Optimization under Different Carbon
Emission Policies: A Case Study
Xiangang Lan 1 , Xiaode Zuo 1, * and Qin Tao 2
Citation: Lan, X.; Zuo, X.; Tao, Q.
Container Shipping Optimization
under Different Carbon Emission
Policies: A Case Study. Sustainability
2023, 15, 8388. https://doi.org/
10.3390/su15108388
Academic Editors: Kum Fai Yuen,
Mingyang Zhang and Ran Yan
Received: 3 April 2023
Revised: 9 May 2023
Accepted: 17 May 2023
Published: 22 May 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Sustainability 2023, 15, 8388. https://doi.org/10.3390/su15108388
1 School of Management, Jinan University, Guangzhou 510632, China
2 Faculty of Business, City University of Macau, Macau, China
* Correspondence: tzuoxd@jnu.edu.cn
Abstract: Climate change is a major environmental issue facing humanity today, and the International
Maritime Organization has accelerated the formulation of greenhouse gas emission policies. This
study considers different carbon emission policies to construct an optimization model for container
shipping, design an improved Whale Swarm Algorithm to solve related issues, and use the marginal
carbon abatement cost method to analyze the deep-seated reasons for the optimization of liner
shipping according to different carbon emission policies, thereby revealing the underlying reasons of
emission-reduction decisions. The conclusions reveal that both kinds of carbon emission policies will
reduce the profits of companies, the average speed of shipping, and carbon emissions. The carbon
tax model has the greatest impact on the profits of shipping companies, and carbon cap-and-trade
is easier to obtain support from enterprises. Sensitivity analysis shows that the implementation of
carbon cap-and-trade or a carbon tax policy is closely and complexly related to the carbon trading
price, carbon tax rate, fuel price, and ship size, and there is uncertainty.
Keywords: carbon emission policies; container shipping optimization; marginal carbon abatement
cost method; Whale Swarm Algorithm
1. Introduction
Maritime transport is responsible for more than 80% of the freight volume of inter-
national trade [1], with container liner shipping accounting for 24% of global maritime
business. In recent decades, the pollution problem of the shipping industry has attracted
increasing attention. The fourth Greenhouse Gas Study was released in August 2020 (by
the International Maritime Organization) [2], which reported that global CO2 emissions
from shipping amounted to 794 million tons in 2020, accounting for 3.1% of the world’s
total CO2 emissions. Climate change is a major environmental issue facing humanity today.
By 2030, global average temperatures are expected to be 1.5 degrees warmer than before the
Industrial Revolution or even higher [3]. Therefore, it is particularly important to transition
the shipping industry toward an environmentally sustainable mode of development. The
IMO has accelerated the development of a greenhouse gas emissions policy, and in 2018, it
proposed its first greenhouse gas reduction strategy. By the middle of the century, global
shipping carbon dioxide emissions will be cut by 50 percent from 2008 levels and gradually
move closer to zero carbon emissions [4]. With the end of the COVID-19 pandemic, the
shipping market has returned to a downturn; simultaneously, the IMO is accelerating the
formulation of greenhouse gas emissions policies, with shipping gradually transitioning
toward the goal of zero carbon emissions. This renders it even more necessary to optimize
transportation based on analyzing the internal mechanism of emissions policies to provide
countermeasures and suggestions for companies and policymakers. There are two main
potential carbon emissions policies available: the carbon tax and carbon cap-and-trade [5–7].
In this context, this paper compares the regulation effects of two common carbon emission
policies, as well as enterprise transportation optimization strategies under different regula-
tion policies. Xing [5] studied the fleet configuration and speed optimization of container
https://www.mdpi.com/journal/sustainability
Sustainability 2023, 15, 8388 2 of 20

liners based on two carbon emission policies (carbon tax and carbon cap-and-trade). Some
interesting conclusions have also been drawn from related studies, which provide the
research basis for this study.
However, few scholars have revealed the internal mechanism of the two emission poli-
cies by comparing them and analyzing their marginal costs, adaptability, and advantages
and disadvantages. In response to these two carbon emission adjustment policies, what
kind of transportation strategy should companies adopt, and how should they optimize
transportation in a timely manner considering market changes and policy adjustments?
Against the backdrop of green shipping, it is necessary to consider protecting the envi-
ronment without affecting the healthy development of the shipping market. How, then,
should policymakers choose the optimal policy? To solve these problems, we propose the
following research objectives:
• Profit maximum, type selection, ship number, and speed as decision variables, build
the two common carbon emissions under the policy of the liner optimization model
and design the algorithm to solve.
• Through empirical analysis, sensitivity analysis, and marginal cost analysis of common
carbon emission adjustment policies, the internal mechanism of emission policies is
revealed, and the regulation effect, adaptability, and advantages and disadvantages of
the two kinds of carbon emission policies are analyzed.
• According to the analysis results, it provides reference suggestions for transportation
optimization of enterprises and policy formulation of relevant departments.
To achieve the above goals, we incorporated the cost of carbon emissions into the cost
structure of marine transportation. On an intercontinental container liner route comprising
multiple ports, the shipping company provides liner transportation services for cargo
owners according to a pre-defined schedule that includes ports of call and their sequences
and charges for freight according to the liner tariff. When the shipping company’s supply
of goods is stable, and the freight rate is stable, it can be considered that its transportation
revenue is unchanged, and profit is equal to revenue minus cost. The nonlinear optimization
model of shipping speed with a maximum profit was constructed by taking the choice
of vessel type, input quantity of vessels, and speed as decision variables. An improved
Whale Swarm Algorithm for solving the model was designed, and real data were collected
for a real case analysis. The results of the empirical analysis were used to compare the
two policies’ controlling of carbon emissions and analyze the impact of fluctuations in the
carbon tax, carbon price, and fuel price on transportation optimization and profit. Finally,
the marginal carbon abatement cost method was used to analyze the impact of both types
of carbon emission policies on the optimization of liner shipping and reveal the internal
mechanism of emission reduction decisions.
The remainder of this paper is organized as follows. Section 2 presents a literature
review. The models are presented in Section 3. The proposed algorithm is designed in
Section 4. The case study is presented in Section 5. Marginal abatement cost analysis and
summary conclusions are presented in Sections 6 and 7, respectively.

2. Literature Review
2.1. Liner Transportation Optimization
Kontovas C. [8] introduced the concept, model, and scenarios suitable for speed opti-
mization. Fagerholt [9] set up a ship speed optimization model for a route composed of
a series of ports with the goal of minimizing fuel consumption, optimized the speed of
each route to achieve a large amount of cost savings, and used heuristic algorithms to solve
the problem. Shuaian Wang [10] calibrated the functional relationship between fuel con-
sumption and speed. Using the historical data of a global shipping company for regression
analysis, the cubed functional relationship between fuel consumption and speed was found
to be relatively accurate. Ellen [11] studied optimal speed on the basis of reducing fuel
consumption and emissions without increasing the number of ships in the fleet. Wang [12]
studied the determination of refueling ports and the optimization of sailing speed under
Sustainability 2023, 15, 8388 3 of 20

different oil prices. Wang [13] set revenue maximization as the goal, considered the influ-
ence of containers of different sizes and seasonal changes in transportation demand, and
built a liner transportation revenue management model to reasonably optimize shipping
speed, reduce fuel consumption, and increase revenue. Guericke [14] constructed a cargo
distribution model for container liner transport considering speed optimization and service
level. Norlund [15] studied ship speed optimization under different weather conditions
and built simulation models to estimate the weekly average fuel consumption of ships.
Wen [16] studied ship navigation path and speed selection from multiple perspectives, such
as voyage time, cost, and environment. Maricruz [17] built a discrete simulation model
to study ship deceleration of dry bulk cargo fleet considering normal speed, deceleration
route, and ultra-low speed sailing. Adland [18] studied the selection of dynamic speed
for bulk shipping by taking into account factors such as organizational constraints, crew
quality, trading mode, loading conditions, technical constraints, and special ship type with
18,000 voyage data collected from AIS. Karsten [19] built a comprehensive optimization
model considering the sailing speed, ship path, and voyage time limits of container ships.
Corbett [20] proposed the problem of CO2 emission and ship speed optimization. Taking
voyage profit maximization as the objective function, they studied the influence of ship
deceleration on CO2 emission under fixed routes. Zheng [21] addresses single-line shipping
services for ship speed and fleet size optimization. Zhao [22] proposed an optimization
model considering the minimum operation cost of the fleet (including voyage cost, opera-
tion cost, and capital cost). Speed is a key determinant of fuel cost. Adding just a few knots
can lead to a dramatic increase in fuel consumption. Speed has a significant impact on
operating costs. Speed optimization has been a hot issue in the liner shipping research field
for the last ten years, and scholars have performed a lot of research in this aspect. These
studies provide a solid foundation for our study, but most scholars only take the speed as
the decision variable and use the algorithm to solve it; few studies consider the selection
of ship type, ship investment, and speed optimization as the decision variables from the
perspective of liner shipping companies operating a route.

2.2. Liner Transportation Optimization Considering Carbon Emission Regulation Policies

Kim [23] builds a model to determine the fleet size, ship speed, and the number
of charters according to maritime environmental regulations (including carbon tax and
carbon emission trading). Lee [24] studied the influence of carbon tax policy (GTAP-E)
on container shipping. Huang [25] proposed a ship emission trading scheme with CO2
as its main content and the impact of CO2 emission trading based on the deceleration
of annual profits and CO2 emissions of container ships. Cullinane [26] summarized the
current policy system of shipping carbon emissions and believed that the combination
of policy and technological innovation could better promote the emission reduction of
ships. Kim [27] studied the speed optimization of ships on routes with designated call
port sequences by considering environmental regulations to reduce carbon emissions and
fuel consumption. Wang [28] established three types of carbon emission tax (no carbon
emission tax, limited carbon emission tax, and carbon emission tax) voyage speed decision
model. Wang [29] studied the impact of carbon tax policy on ship refueling strategy, speed,
and deployment. Zhu [30] studied fleet planning under the uncertainty of carbon tax policy
by a multi-stage stochastic integer programming model. Cheaitou [31] proposed a multi-
objective optimization model of liner transportation based on CO2 emission minimization,
SOx emission minimization, and profit maximization in which all objective functions of the
model are functions of ship sailing speed. Xin et al. [32], considering the characteristics of
modern crude oil supply systems, discussed the green scheduling problem of tanker fleets
considering variable speed and carbon tax.
The shipping industry was relatively prosperous from 2000 to 2007, especially in
2007, which represented a golden period of shipping development, with fuel prices for
ships being significantly lower than today’s prices. Therefore, to maximize profits, most
ships have adopted high-speed sailing; hence, relatively limited literature has addressed
Sustainability 2023, 15, 8388 4 of 20

speed optimization. With the impact of the 2008 financial crisis, the shipping market has
undergone a 180-degree shift, and coupled with the rising fuel prices, shipping companies
have adopted low-speed sailing, resulting in an increasing amount of literature related
to shipping transportation optimization. Xing et al. [33] sorted out a series of methods
to reduce shipping carbon emissions. In addition to the widely studied technical and
operational measures, they also discussed market-level measures and independent emission
reduction behaviors. MA et al. [34] carried out multi-objective optimization of route and
speed under weather changes and emission control policies. LAN [35] consider carbon
emission and emission control area built a liner transport network design model, including
route design, route allocation, and cargo transport plan.
Carbon emission reduction has become an inevitable trend in the development of the
shipping industry. In order to reduce the carbon emissions of the shipping industry, the
international community has put forward different carbon emission policies. Carbon tax
and carbon emission quota are the two main carbon emission policies. Therefore, carbon
tax policy is bound to have an impact on transportation optimization decisions. Most of
the existing relevant studies only focus on the impact of carbon emission policies on ship
speed and only consider carbon emission regulation policies to study the optimal speed of
traditional transportation optimization, while few scholars reveal the internal mechanism
of emissions policies and analyze their adaptability and advantages by comparing common
carbon emission adjustment policies and conducting marginal cost analysis. This study,
based on the original research, takes ship type selection, ship input quantity, and speed
as decision variables and considers different carbon emission regulation policies to build
an optimization model for container liner transportation. Through empirical analysis,
sensitivity analysis, and marginal cost analysis of common carbon emission regulation
policies, the internal mechanism of emission policies is revealed. The regulation effect,
adaptability, and pros and cons of the two carbon emission policies are analyzed.

3. Models
On an intercontinental container liner route, the shipping company provides trans-
portation services for cargo owners according to a pre-defined schedule, including ports of
call and their sequence and charges for freight according to the liner tariff. Profit is equal to
revenue minus costs, which include fixed vessel operating costs, main engine and auxiliary
engine fuel costs, carbon emissions costs, and port costs. A nonlinear optimization model
of shipping speed with the maximum profit of the shipping company is constructed by
taking the speed, input quantity of vessels, and choice of vessel type as decision variables.
Assumptions: (1) Excess transport capacity, regardless of ship input cost; (2) No special
circumstances shall be taken into account in the normal navigation of the ship, and no
accident shall occur in the course of navigation; (3) The ship will operate in the form of
return voyage in the past, with weekly shift frequency for each route and the same speed
for the same route; (4) During the study period, the demand for cargo transportation was
stable, and the freight volume between ports remained basically unchanged; (5) The size of
the ship type on the route is determined by the freight volume, while the freight volume
does not change, so the ship type invested in the same route is the same, but the ship type
selection of the route is decided by model optimization; (6) The infrastructure of all ports
has been mature, assuming that the loading and unloading efficiency, charging standard,
ship arrival and departure time of all ports selected in this study are the same.
k denotes the shipping segment, and Yk denotes the total amount of cargo (tons)
on segment k. qij represents a container transport from port i to port j, as presented in
Figure 1: When port 2 has container q12 to be transported to port 4, it needs to pass
through Sections 2 and 3. When port 1 has container q13 to be transported to port 3, it must
pass through Sections 1 and 2. In that case, the total number of containers transported
on segment 3 is q24 + q13 . Similarly, the cargo volume Yk on segment k is denoted as
Yk = ∑ N k
j=k +1 ∑i =1 qij . Figure 1 represents the round-trip route. As presented in Figure 2,
 segment representsaacontainer
segmentk.k.qqijijrepresents containertransport
transportfromfromport
portiitotoport
portj,j,as
aspresented
presentedin inFigure
Figure
1: When port 2 has container q12 to be transported to port 4, it needs to pass through Sec-
1: When port 2 has container q12 to be transported to port 4, it needs to pass through Sec-
tions 22 and
tions and 3.3. When
When port
port 11 has
has container
container qq1313 to
to be
be transported
transported to to port
port 3,3, itit must
must pass
pass
throughSections
through Sections11andand2.2.In
Inthat
thatcase,
case,the
thetotal
totalnumber
numberof ofcontainers
containerstransported
transportedon onseg-
seg-
ment 3 is q 24 + q13. Similarly, the cargo volume Yk on segment k is denoted as
ment 3 is q24 + q13. Similarly, the cargo volume Yk on segment k is denoted as
Sustainability 2023, 15, 8388 5 of 20
Nj =j =kk++11
ik=i=11qqijij..Figure
N k
YYkk == Figure11represents
representsthe
theround-trip
round-triproute.
route.As
Aspresented
presentedin
inFigure
Figure2,2,
theexport
the exportroute
routeisis(to):
(to):Ningbo
NingboPort—Shanghai
Port—ShanghaiPort—Busan
Port—BusanPort—Long
Port—LongBeach
BeachPort.
Port.The
The
the export
import route
route is is (to):Long
(back): Ningbo Port—Shanghai
Beach Port—Busan Port—Busan
Port—NingboPort—Long
Port when Beach
N = 6. Port. The
import route is (back): Long Beach Port—Busan Port—Ningbo Port when N = 6.
import route is (back): Long Beach Port—Busan Port—Ningbo Port when N = 6.

Section2
Section2
Section1
Section1
Section3
Section3

11 22 33

44
55
NN
Sectionk
Sectionk Section4
Section4

Figure1.1.
Figure
Figure Relationshipbetween
1.Relationship
Relationship betweenthe
between thecargo
the cargovolume
cargo volumeand
andthe
theOD
ODdemand
demandbetween
betweenports.
ports.

Figure
Figure2.2.
Figure COSCO Shipping
COSCOShipping
2.COSCO Trans-Pacific
ShippingTrans-Pacific Service—AAC4.
Trans-PacificService—AAC4.
Service—AAC4.

The
The totaltime
Thetotal timeT
time TTof
ofaaaround
of roundtrip
round tripincludes
trip includes
includes seasea
sea travel
travel time,
time,
travel arrival
arrival
time, and
arrival and
and departure
departure
departuretime, time,
and

time,
N
loading
and and
loading unloading
and time.
unloading Q is the
time. loading
Q i
and loading and unloading i time. Qi is the loading and unloadingis and
the unloading
loading of
andport i Q =
unloading
i ∑ of
j=of q
1 port +
port
ij q jiii
 (( ))
N
the expression of the total voyage time T is:

QQi i == j =j =11 qqijij ++qqjiji the
N
theexpression
expressionof ofthe
thetotal
totalvoyage
voyagetimetimeTTis:
is:
N
N

T ==DD vv++∑ QQi i ll++NN⋅⋅ttpil


T =T D/v + N Q /l + N · t (1)

i =1i =1
i =1
pil
i pil
(1)
(1)

Supposethat
Suppose
Suppose the number
thatthe number
numberof of
ofships
shipsper
ships perweek
per weekof
week ofthe
of theroute
the route nn
routeisisis mnm,m,and
andthe
,and total
the
the number
total
total number
number of
ships
of per
ships week
per is
week 168
is h,
168 then
h, the
then expression
the expression of the
of number
the number of ships
of
of ships per week is 168 h, then the expression of the number of ships per week nm is:ships per week
per weekn m nis:
m is:

T= TT = 1 11  D 
N
!
 
NN
n ( x v
 ) +
D ((xxmmmvvmmm)) ++ ∑QQi i i /l
Q l +
N ⋅ t 
168 m∑
nm = n mm = = = D/ l ++NN⋅ t ·pilpilt pil (2)
(2)
(2)
168168168
168168 ∈M
m
m∈M
∈M i =
i =1
i =11  
Suppose
Suppose the
Supposethe daily
thedaily
daily revenue
revenue ofof
of
revenue thethe
the fleet
fleet isisf(r),
fleet f(r), thedaily
is fthe
(r), daily revenue
revenue
the daily andfreight
and
revenue freight ratemul-
rate
and freightmul-
rate
tiplied
tiplied bythe
by
multiplied the weekly
byweekly volume
volume
the weekly divided
divided
volume by7,7,by
by
divided and
and theexpression
7, the
and expression is: is:
is:
the expression

N −1 N −1
f(r ) = ∑ ∑ qij rij /7 (3)
i =0 j =0

The average daily total cost of the fleet includes port charges, fuel consumption
charges, and operating costs. Let the weekly operating costs of the fleet be Cop , and the
average port charges per week be CP (Including port usage fee, mooring fee, berthing fee,
and other fixed charges). Then, the expressions Cop , CP :

C op = ∑ ( xm Cm nm ) (4)
m∈ M
Sustainability 2023, 15, 8388 6 of 20

N N N
CP = ∑ ∑ (qij Cl + q ji Cu )/7 + N · ∑ (Gm0 xm )/7 (5)
i =1 j =1 i =1

According to the cubed function between speed and fuel consumption [10] F (v/v0 )3 .
The main engine uses heavy oil, and the auxiliary engine uses light oil. Then, the expression F.
 3
vm D
F= ∑ xm Fm
v0m 168vm m∑
+ Am · xm · nm (6)
m∈ M ∈M

Ship carbon emissions depend on ship fuel consumption in a certain period and the
carbon emission factor of fuel λ. According to Fourth Greenhouse Gas Study 2020, this
paper uses the factor λ = 3.114. That is, 1t of marine fuel produces 3.114t of CO2 . Qco2 can
be expressed as:
 3
v D
Qco2 = λ · F0 +λ·A·n (7)
v0 168v

3.1. Optimization Model of Liner Shipping under a Carbon Tax Policy

Carbon tax refers to the policy of imposing a certain tax on shipping enterprises
according to their carbon emissions. Volume-based carbon taxes are levied based on
the ship’s CO2 emissions, assuming that a quantitative carbon tax δ is collected per ton
(USD$/ton), and the carbon emission cost Cco2 is expressed as (8):

Cco2 = Qco2 · δ (8)

Objective function:
N N q ·r  3
vm D
max f day (tp) = ∑ ∑ ij ij
7 − ∑ xm Fm v0m
· vm ·168 · PHFO − ∑ Am · xm · nm · PMFO
i =1 j =1 m∈ M m∈ M
N N (9)
− ∑ ( xm Cm nm ) − ∑ ∑ (qij C l + q ji C u )/7 − ∑ ( xm Gm
0 ) · N/7 − Q
co2 · δ
m∈ M i =1 j =1 m∈ M

Constraints:
∑ xm = 1, xm = 0 or 1 (10)
m∈ M

max(Yk ) ≤ ∑ Bm xm (11)
m∈ M

vmin max
m ≤ vm ≤ vm (12)

k = 1, 2, · · · , N − 1 i, j = 1, 2, · · · , N (13)

m∈M (14)
With the vessel speed, choice of vessel type, and input quantity of vessels as the
decision variables, the objective function (9) has two parts. The first component is the
freight revenue, while the second component is the cost, including the average daily fuel
cost of the main engine, the average daily fuel cost of the auxiliary engine, the average
daily operating cost (daily rent of the vessel), loading and unloading fees, port fees, and
carbon emission cost. (10) The constraint is that ships of the same type are deployed on the
route. (11) The constraint is the vessel capacity restriction. (12) The constraint is the vessel
speed limits. (13) This includes non-negative and integer constraints. (14) The constraint is
the vessel type.
Sustainability 2023, 15, 8388 7 of 20

3.2. Optimization Model of Liner Transportation under the Carbon Cap-and-Trade Policy
The carbon allowance is the production and operation of enterprises in accordance
with the limited carbon emission quota. Companies are not allowed to enter the carbon
cap-and-trade market to buy or sell carbon credits when carbon emission quotas are
insufficient or excessive. Therefore, the carbon allowance policy is a mandatory constraint
for manufacturing companies, assuming that carbon emissions are limited to no more than
θ (tons). Based on the above analysis of each cost item, the optimization model of liner
shipping under the multi-vessel carbon cap-and-trade policy is expressed as follows:
Objective function:

N N q ·r  3
vm D
max f day (tp) = ∑ ∑ ij ij
7 − ∑ xm Fm v0m
· vm ·168 · PHFO − ∑ Am · xm · nm · PMFO
i =1 j =1 m∈ M m∈ M
N
N q C l +q C u 0 N
xm Gm
− ∑ ( xm Cm nm ) − ∑ ∑ ij 7 ji − ∑ 7 − ( Qco2 − θ ) · pe
m∈ M i =1 j =1 m∈ M
(15)
Constraints: (10)–(14)
With speed, choice of vessel type, and input quantity of vessels as the decision vari-
ables, the objective function (16) has two parts. The first component is the freight revenue,
while the second component is the cost, including the average daily fuel cost of the main
engine, the average daily fuel cost of the auxiliary engine, the average daily operating cost
(daily rent of the vessel), loading and unloading fees, port fees, and carbon emission cost.

4. Arithmetic Design
Using the MATLAB platform, an improved Whale Swam Algorithm (WSA) was
designed to solve the planning model. Swarm behaviors, such as predation of the whale
swarm with ultrasonic waves as the information medium, are simulated, and each solution
is compared to a whale. The movement of each whale is guided by the nearest whale
among the whales that are better than it (judged by their fitness value), and this leading
whale is defined as the “better and nearest” whale.
The flow chart of the algorithm is presented in Figure 3.

Description of Improved Whale Swarm Algorithm

iter ·(rhmax−rhmin)
(1) The intensity of the ultrasonic source decreases linearly rho0 = rhmax − iterations .
(2) normalization: As the range of independent variables in this study is considerably
large and performing optimization in a large range is significantly more difficult, this
algorithm will normalize the independent variables. For example, an individual whale is
pop j,k , where pop j,k ∈ [0, 1], and its corresponding actual value is (maxvalue − minvalue) ·
popi + minvalue.
(3) The ultrasonic attenuation coefficient. The intensity of the ultrasonic source is
attenuated as a multiple of  the attenuation coefficient, and the attenuation system formula
0.25
is eta = − log( Farthest_ f it where Farthest_ f it is the fitness value of the farthest individual
from the j-th whale and the intensity of the ultrasonic source is rho0 = eta · rho0.
(4) The improved population update method. If the nearest individual is better than the
current individual, the individual is updated according to the formula
pop j = pop j + rho · rand() · ( Nearest_pop − pop j ). If it is worse, the new individual is
generated in reverse newpop j = pop j − rho · rand() · ( Nearest_pop − pop j ), and if the new
individual is better than the current individual, it replaces the latter; otherwise, the original
individual is retained.
Sustainability 2023, 15, x FOR PEER REVIEW 8 of 20
Sustainability 2023, 15, 8388 8 of 20

Start

Set parameters

Import data
Initialize whale population
individuals pop
Calculate the fitness value
fitness

zbest-fitness zbest

Euclidean
distance

Neares_pop

Nearest_fit

Judge the pros

cons Produce new
and cons whales
pros
Renewal whale

Reach the maximum

No number of iterations?

Yes
Output the global
optimal solution

end

Figure 3. Flow chart of improved Whale Swarm Algorithm.

Figure 3. Flow chart of improved Whale Swarm Algorithm.
5. Case Study
Description of Improved Whale Swarm Algorithm
5.1. Data Collection
In order
(1) to improve
The the research
intensity of significance,
the this paper studies
ultrasonic sourcethe real case of the linearly
decreases

ho0 =assumes
rstudy
( )
COSCO Shipping Container Lines Co., Ltd., Shanghai, China, as shown in Figure 4. This
iter ⋅ rh max − rh min
− the ship departs from and. returns to the Port of Shanghai. Its western
rh maxthat
departure is as follows: iterations
Port of Shanghai—Port of Ningbo—Port of Xiamen—Port of
(2) normalization:
Yantian—Port As the range
of Singapore—Port of independent
of Felixstowe; variables
it returns towardin the
thiseast
study
as is considerably
follows:
Port of Rotterdam—Port of Wilhelmshaven—Port of Gdansk—Port of Felixstowe—Port
large and performing optimization in a large range is significantly more difficult, this al-
of Singapore—Port
gorithm of Yantian—Port
will normalize the independentof Shanghai; D = 2483
variables. For kn, N = 13, an
example, andindividual
Hong Kongwhale is
Feb 15 prices in USD are as follows: PIFO = 426.5 USD/TON, PVLSFO = 628 USD/TON;
pop j , , to
According data frompop
where j , k ∈Port:
Yantian [0,1]l =, 150and
TEU/h, itsCl =corresponding
Cu = 65 USD/TEU, actual
tpil = 2value
h; is

( max value
reference − min
ESMA Finalvalue
)
Report⋅[36]:pop∂ +=min value . θ = 1600 Ton/day; accord to Lowe’s
63 !USD/TON,
Marine Database, data of each shipi type are shown in Table 1.
(3) The ultrasonic attenuation coefficient. The intensity of the ultrasonic source is at-
tenuated as a multiple of the attenuation coefficient, and the attenuation system formula
0.25
is eta = − log( ） where Farthest _ fit is the fitness value of the far-
Farthest _ fit
thest individual from the j-th whale and the intensity of the ultrasonic source is
rho 0 = eta ⋅ rho 0 .
(4) The improved population update method. If the nearest individual is better than
the current individual, the individual is updated according to the formula

Sustainability 2023, 15, 8388
of Rotterdam—Port of Wilhelmshaven—Port of Gdansk—Port of Felixstowe—Port of Sin
gapore—Port of Yantian—Port of Shanghai; D = 2483 kn, N = 13, and Hong Kong Feb 1
prices in USD are as follows: P 426.5 USD⁄TON, P 628 USD⁄TON; According
⁄
to data from Yantian Port: l 150 TEU h , C C ⁄
65 USD TEU , t 2 h; referenc
ESMA Final Report [36]: ∂ 63 USD⁄TON, θ = 1600 Ton day ; accord to Lowe’s
9 of 20 Marin
Database, data of each ship type are shown in Table 1.

Figure 4. COSCO Far East—Mediterranean route chart.

Figure 4. COSCO Far East—Mediterranean route chart.

Table 1. Data of each ship type.

Table 1. Data of each ship type.
Bm Fm Am 0
vm 0 vmin max
min vm cm G0
Bm (TEU) Fm (T/day) (T/Day)
Am v vmmax(USD/Day)
m m
(Kn/h)vm (Kn/h)
Name of Vessel m
Name of Vessel m
(Kn/h)m
cm (USD/Day)
(USD/Call)
Gm0 (USD/Call
XIN ZHANG ZHOU 1 (TEU) 4253(T/day)
139.5 (T/Day)
6.33 (Kn/h)
18.2 11.34(Kn/h)25.15 (Kn/h) 9000 3001
XIN WENZHOU 2 4738 82 4.3 18 11.04 24.7 10,026 3344
XIN ZHANG
XIN YAN ZHOU1
TIAN 3 4253 5668 139.5 202 6.33
7.81 18.2
17.7 12.0511.34 26.7
25.15 11,994 9000 4000 3001
XIN WENZHOU
COSCO 2
THAILAND 4 4738 8501 82 250 4.3
10.47 18.618 12 11.04 26.6
24.7 17,989 10,026 6000 3344
XINXIN
YANSHANGHAI
TIAN 3 5
5668 9572 202 248.2 10.43
7.81 17.2
17.7 11.22
12.0526.73
26.7 20,255 11,994 6204 4000
COSCO ASIA 6 10,036 250 12.75 16.8 11.04
21,238 25.86505
COSCO THAILAND
COSCO 4
FAITH 7 8501 13,114 250 274.9 10.47
13.2 18.6
16.7 11 12 26.2
26.6 27,751 17,989 8500 6000
CSCLJUPITER
XIN SHANGHAI 5 8 9572 14,074248.2 262 14.51
10.43 16.1
17.2 11.1811.2226.62
26.73 29,783 20,255 9122 6204
CSCLPACIFIC OCEAN 9 18,982 195.5 13.768 18 10
40,169 24.6
13,000
COSCO
COSCO ASIA VIRGO
SHIPPING 6 10 10,036 20,119 250 168 12.75
10.263 1916.8 8.461511.04 22.5
25.8 42,575 21,238 13,040 6505
COSCO FAITH 7 Data
13,114 274.9
source: Lowe’s 13.2
Marine Database. 16.7 11 26.2 27,751 8500
CSCLJUPITER 8 14,074 262 14.51 16.1 11.18 26.62 29,783 9122
CSCLPACIFIC OCEAN 9 18,982 195.5
Drewry Shipping 13.768 The weekly
Consultant 18 10 from 24.6
demand 40,169
Asian ports to 13,000
a European port
COSCO SHIPPING VIRGO 10 (westbound)
20,119 is about 1000
168 TEU. Drewry
10.263 19 Shipping Consultant
8.4615 22.5 The weekly demand from
42,575 13,040
Asiansource:
Data ports toLowe’s
a European
Marine port (westbound) is about 1000 TEU.inter-port rate according to
Database.
China International Shipping Network, See Table 2.
Drewry Shipping Consultant The weekly demand from Asian ports to a European
Table 2. Interport traffic and freight rate (TEU, freight rate USD/TEU).
port (westbound) is about 1000 TEU. Drewry Shipping Consultant The weekly demand
from Asian
Outward ports to aSingapore
Voyage European port (westbound)
Felixstowe is about 1000 TEU.inter-port rate ac
Rotterdam
cording
Shanghaito China International Shipping
(570,150) Network, See Table 2.
(1040,800) (1040,850)
Ningbo (570,150) (1050,750) (1050,750)
Xiamen (380,140) (1040,745) (1040,745)
Yantian (570,100) (1050,600) (1050,650)
Singapore (0,0) (1060,550) (1050,550)
Return voyage Singapore Xiamen Shanghai
Rotterdam (570,550) (640,900) (640,900)
Gdansk (570,550) (650,900) (650,900)
Wilhelmshaven (380,550) (640,900) (640,900)
Felixstowe (570,550) (650,900) (650,900)
Singapore (0,0) (660,150) (650,150)
Data source: China International Shipping Network.

5.2. Analysis of Results

From the optimization results (Table 3), low-speed sailing is the optimal choice under
both policies controlling carbon emissions, with the optimal choice of vessel type being a
13,000 TEU ship and an input of seven vessels. An analysis of policies controlling carbon
emissions reveals that the carbon tax policy has a large impact on the profit of shipping
Sustainability 2023, 15, 8388 10 of 20

companies, while carbon cap-and-trade has a smaller impact on their profit. The carbon
tax policy results in a large emission reduction, while that of the carbon cap-and-trade
model is small, mainly because the price in the carbon cap-and-trade model is too low, with
relatively weak market regulation.

Table 3. Analysis of results.

Carbon Cap-and-Trade Carbon Tax

daily profit of the fleet 884,841.6 809,853.48
carbon emission 1851.27 1840.09
Carbon emission cost 5977.27 92,004.56
Main engine fuel cost 168,868.10 165,568.10
Fuel cost of auxiliaries 104,541.10 104,541.10
Daily operating cost 360,766.0 360,766.0
port dues 15,785.71 15,785.71
speed 12.31 12.26
speed 13 13
ship type 7 7

5.3. Sensitivity Analysis

In order to study the influence of carbon tax and carbon trading price changes on
enterprises’ choice of emission reduction strategies, whether high or low fuel prices affect
enterprise emission reduction decisions, and whether ship size is conducive to shipping
emission reduction. In this section, the influences of carbon tax, carbon trading price,
fuel price, and ship size on carbon emission adjustment policies and fleet profitability are
studied through sensitivity analysis.

5.3.1. Sensitivity Analysis of Carbon Tax and Carbon Price

Carbon taxes and carbon cap-and-trade have been adopted as the main policy tools
for carbon reduction in most countries emphasizing carbon reduction. With all other
Sustainability 2023, 15, x FOR PEER REVIEW 11 of 20
parameters unchanged, the model was operated by changing the carbon tax and the carbon
price, with the results presented in Figure 5.

1,200,000.00 2200
2100
Carbon emission/Ton
1,000,000.00
2000
800,000.00
Profit/USD

1900
600,000.00
1800
400,000.00
1700
200,000.00 1600
0.00 1500
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100

Carbon tax policy（carbon emissio）

Carbon cap-and-trade（carbon emissio）
Carbon tax policy（Average daily profit）
Carbon cap-and-trade（Average daily profit）

Figure5.5.Sensitivity
Figure Sensitivityanalysis
analysisofof carbon
carbon taxtax
andand carbon
carbon price.
price.

The
Theresults
resultssuggest
suggestthat carbon
that carbonemissions continued
emissions to decrease
continued in both
to decrease the carbon
in both the carbon
cap-and-trade and carbon tax models with the increase in carbon price and
cap-and-trade and carbon tax models with the increase in carbon price and carbon carbon tax. tax.
When
Whenthe
thecarbon
carbonprice and
price andcarbon taxtax
carbon reached $75$75
reached USD/ton,
USD/ton,the the
carbon emissions
carbon of both
emissions of both
models tended to become consistent. From the perspective of company profit, company
models tended to become consistent. From the perspective of company profit, company
profit first declines under the carbon cap-and-trade model. When the carbon price exceeds
$50 USD/ton, the company profit starts to rise (as shown in Figure 6). Conversely, under
the carbon tax model, company profit decreases, and hence, the comparative analysis of
the two models suggests that shipping companies like cap-and-trade, which also allows
 Figure 5. Sensitivity analysis of carbon tax and carbon price.

The results suggest that carbon emissions continued to decrease in both the carbon
Sustainability 2023, 15, 8388 cap-and-trade and carbon tax models with the increase in carbon price and11carbon of 20 tax.
When the carbon price and carbon tax reached $75 USD/ton, the carbon emissions of both
models tended to become consistent. From the perspective of company profit, company
profitfirst
profit firstdeclines
declines under
under thethe carbon
carbon cap-and-trade
cap-and-trade model.
model. When When the carbon
the carbon price exceeds
price exceeds
$50 USD/ton, the company profit starts to rise (as shown in Figure 6). Conversely, under under
$50 USD/ton, the company profit starts to rise (as shown in Figure 6). Conversely,
thecarbon
the carbontax taxmodel,
model, company
company profit
profit decreases,
decreases, and and hence,
hence, the comparative
the comparative analysis
analysis of of
thetwo
the twomodels
modelssuggests
suggests that
that shipping
shipping companies
companies like like cap-and-trade,
cap-and-trade, whichwhich also allows
also allows
themtotocut
them cutemissions.
emissions.

915,000
910,000
Average daily profit (USD)

905,000
900,000
895,000
890,000
885,000
880,000
875,000
870,000
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Carbon price (USD/ton)

Figure6.6.Average
Figure Average daily
daily profit
profit curve
curve of the
of the fleetfleet
underunder carbon
carbon cap-and-trade
cap-and-trade policy.policy.

As
Asthe
thecarbon
carbonmarket
marketcontinues
continuestotomature
matureand
and the transaction
the transactionprice continues
price to to in-
continues
increase, results indicate that company profits will increase rather than decrease when
crease, results indicate that company profits will increase rather than decrease when com-
companies strive to reduce emissions. For companies, this will provide a greater incentive
panies strive to reduce emissions. For companies, this will provide a greater incentive to
to optimize transportation to actively reduce emissions. In contrast, the incentive to reduce
optimize transportation to actively reduce emissions. In contrast, the incentive to reduce
emissions from a carbon tax is weaker because the tax is compulsory, and the emission
emissionsbehavior
reduction from a driven
carbonbytax is weaker
a carbon tax isbecause
passive.the tax is compulsory, and the emission
reduction behavior driven by a carbon tax is passive.
5.3.2. Changes in Fuel Prices
Currently, most shipping companies use speed changes to control costs. Figure 7
depicts that, in the short term, low oil prices can cut costs for shipping companies. Choosing
a high speed when the oil price is low can reduce ship investment or increase the voyage
number, while choosing a low speed when the oil price is high can reduce cost. The
rising oil price leads to a decrease in ship speed and an increase in ship investment cost.
Additionally, Figure 7 indicates that when the oil price drops, the carbon emission of ships
is high, so the low oil price will have a negative impact on the carbon emission reduction
of ships. For example, the COSCO Shipping Group has invested a substantial amount of
money in management software and fuel additives for emission reduction. If low oil prices
persist, these investments are largely meaningless from a corporate earnings perspective.

5.3.3. Changes in Vessel Size and Carbon Emissions

Table 4 indicates that larger ships enjoy the dual advantage of lower carbon emis-
sions and lower fuel costs. Adopting larger vessels improves their shipping capacity and
significantly reduces the energy consumption of shipping and carbon emissions. Driven
by the economy and environmental protection, container ships are transitioning toward
larger sizes.
 Sustainability 2023, 15, x FOR PEER REVIEW 12 of 20

5.3.2. Changes in Fuel Prices

Sustainability 2023, 15, 8388 12 of 20
Currently, most shipping companies use speed changes to control costs. Figure 7 de-
picts that, in the short term, low oil prices can cut costs for shipping companies. Choosing
a high speed when the oil price is low can reduce ship investment or increase the voyage
Table 4. Optimization results of different vessel types.
number, while choosing a low speed when the oil price is high can reduce cost. The rising
oil price leads to a decrease in ship speed and an
Carbon Taxincrease in ship investment
Carbon cost. Addi-
Cap-and-Trade
tionally, Figurem
7 indicates that when 9
the oil price drops,
10
the carbon
9
emission of10ships is
high, so the Speed
low oil price will have a11.36
negative impact12.31on the carbon
11.36emission reduction
12.31 of
ships. Carbon
For example,
Emissionsthe COSCO Shipping
2640.28 Group has invested2640.28
2162.93 a substantial 2162.93
amount of
money
Averageindaily
management software
profit of the fleet and fuel additives
453,292.70 for emission559,325.02
559,063.81 reduction. If low oil prices
653,162.53
persist, these investments are largely meaningless from a corporate earnings perspective.

6000 25

5000
Carbon emissions (tons)

20
4000

Speed (kn)
15
3000
10
2000

1000 5

0 0
20.25
40.5
81
121.5
162
202.5
243
283.5
324
364.5
405
445.5
486
526.5
567
607.5
648
688.5
729
769.5
Fuel price(USD/Ton)
Carbon cap-and-trade Carbon tax policy
Carbon cap-and-trade Carbon tax policy

Figure7.7.The
Figure Thecurve
curveof
ofoptimization
optimizationresults
resultsunder
underdifferent
differentfuel
fuelprices.
prices.

6.5.3.3.
Marginal
ChangesAbatement
in VesselCost
Size Analysis
and Carbon Emissions
The marginal
Table abatement
4 indicates cost isships
that larger the cost of each
enjoy additional
the dual unit ofofCO
advantage 2 abatement.
lower The
carbon emis-
classical
sions and marginal abatement
lower fuel cost function
costs. Adopting was
larger proposed
vessels by Nordhaus,
improves who provided
their shipping the
capacity and
relationship between marginal abatement cost and the corresponding emission
significantly reduces the energy consumption of shipping and carbon emissions. Driven reduction
rate.
by the Combined
economywith
and the evaluation index
environmental of abatement
protection, cost
container perare
ships unittransitioning
CO2 gas proposed
toward
by S. E. Magnus,
larger sizes. the marginal abatement cost MAC(ship) of a ship is calculated with the
following equation [37]:
Table 4. Optimization results of different vessel types.
MAC (ship) = ∆C ( A)/∆CO2 ( A) (16)
Carbon Tax Carbon Cap-and-Trade
∆C ( A) is themincrease in cost resulting
9 from the10 adoption of Measure
9 A; 10
∆CO2 ( A)Speed
is the reduction of CO211.36
emissions after
12.31the adoption of
11.36 Measure A.
12.31
1. Carbon
Marginal Emissions
Abatement Costs of the Carbon2640.28Cap-and-Trade
2162.93 Model2640.28 2162.93
Average daily profit of the fleet 453,292.70 559,063.81 559,325.02 653,162.53
∆C ( A) = CCosts o f implementing a carbon Cap− and−Trade policy − CNo carbon emission limit costs
F · D ·( PHFO +6.λ HMarginal
· pe ) 2
∆C ( A) = 0 vCarbon Abatement
Cap− and− Trade −
Cost
F0 · D ·Analysis
PHFO 2
v + D·( A·( PMFO168
+λ M · pe )+C ) −1
· vCarbon Cap− and− Trade
168(vo )3 168(vo )3 NO (17)
The marginal abatement 
N cost is the cost of each additional unit of CO2 abatement.
D ·( A· PMFO +C ) −1 λ · p · A Qi
− 168 The· vclassical
NO −
M e
marginal
168 · ∑ abatement
l + Nt pil cost − θfunction
· pe was proposed by Nordhaus, who pro-
i =1
vided the relationship between marginal abatement cost and the corresponding emission
Simplify
reduction rate.to Combined
obtain with the evaluation index of abatement cost per unit CO2 gas
proposed by S. E. Magnus, the marginal abatement cost MAC(ship) of a ship is calculated
F · D ·λ H · pe 2 with the following F0 · D ·equation
 [37]: 
PHFO D · A·λ M pe −1
∆C ( A) = 0 3 v Carbon Cap − and − Trade + 3 v 2
Carbon Cap − and − Trade − v 2
NO + 168 · vCarbon Cap− and− Trade
168(vo ) 168(vo )
 
A·λ M · pe

N
 (18)
+ D·( A· P168
MFO +C ) −1
· vCarbon Cap− and− Trade − v NO +
−1
168 · ∑ Ql i + Nt pil − θ · pe
i =1
 ΔC ( A ) =
F0 ⋅ D ⋅ λH ⋅ pe
168 ( v )
o 3
2
vCarbon Cap − and −Trade +
F0 ⋅ D ⋅ PHFO
168 ( v )
o 3
(v 2
Carbon Cap −and −Trade
2
− vNO )
+
D ⋅ A ⋅ λM pe -1
168
⋅ vCarbon Cap−and −Trade
(18)
D ⋅ ( A ⋅ P + C ) -1 A ⋅ λM ⋅ pe  N Q 
+ 15, 8388 MFO
Sustainability 2023, ⋅ ( vCarbon Cap−and −Trade − vNO
-1
) + ⋅   i + Nt pil  − θ ⋅ pe 13 of 20
168 168  i =1 l 

ΔCO2 ( A ) =
∆CO2 ( A) =
λH ⋅ F0 ⋅ D
( )·
· F0 · v
λ H168 Do 3
2
2
NO(
⋅ v − v 2Carbon Cap−and −Trade +
2
) λM ⋅ A ⋅ D
+ 168
λM · A · D
-1
⋅ vNO (
− v -1Carbon Cap−and −Trade
· v −1 − v −1
)  (19)
v NO − vCarbon Cap− and−Trade NO Carbon Cap− and− Trade (19)
168 vo 3
( ) 168
Considering the optimal speed determined without a policy controlling carbon emis-
Considering the optimal speed determined without a policy controlling carbon emis-
sions, that is, vnull is a known quantity, the model curve was plotted by MATLAB as fol-
sions, that is, vnull is a known quantity, the model curve was plotted by MATLAB as follows
lows Figure 8:
Figure 8:

GraphofofMAC
Figure8.8.Graph
Figure MACtrend.
trend.

The carbon cap-and-trade model creates a market for emission rights based on emis-
sion control targets, and through carbon trading, an equilibrium market price can be
formed such that the efficiency of resource allocation in the shipping market reaches Pareto
optimality. Carbon credits are a resource that can bring economic benefits to all shipping
companies, with the marginal value of carbon credits varying among them. Shipping
companies with higher marginal abatement costs (MAC) can reduce their abatement cost
by purchasing carbon credits, while shipping companies with relatively low MAC can earn
income by selling carbon credits. From the perspective of MAC, the MAC of shipping
companies under the carbon cap-and-trade is smaller than those under the carbon tax
policy. In other words, the carbon cap-and-trade makes the emission reduction of ships
more profitable and can further encourage shipping companies to reduce emissions.
As shown in the Figures 9–11, the MAC of a ship under the carbon cap-and-trade
policy is closely and complexly related to its vessel speed, fuel price, vessel size, and carbon
allowance. MAC decreases as the fuel price increases, and the larger the ship type, the
larger the MAC.
2. Marginal Abatement Cost of the Carbon Tax Model

∆C ( A) = CCost o f implementing carbon tax policy − CNo carbon emission limit costs
F · D ·(λ H δ+ PHFO ) 2 D ·( A·( PMFO +λ M δ)+C )
∆C ( A) = 0 vCarbon tax − F0 · D· PoHFO 2
3 v NO +
−1
· vCarbon
168(vo )3 (v )
168 
168 tax
(20)
N
D ·( A· PMFO +C ) A·λ M δ Q
− 168 · v− 1
NO + 168 · ∑ l i + Nt pil
i =1
 
F0 · D · PHFO
∆C ( A) = F0 · D ·λ H δ 2
v + v2Carbon tax − v2NO + D· A168
·λ M δ −1
· vCarbon
168(vo )3 Carbon tax
168(vo )3 tax
  
N
 (21)
+ D·( A· P168
MFO +C ) −1
· vCarbon tax − v− 1
NO + 168 ·
A·λ M δ Q
∑ l i + Nt pil
i =1
 formed such targets,
sion control that the and efficiency
throughof resource allocation
carbon trading, an in the shipping
equilibrium market
market pricereaches
can bePa-
reto
formedoptimality.
such thatCarbon credits of
the efficiency areresource
a resource that caninbring
allocation economic
the shipping benefits
market to all Pa-
reaches ship-
ping companies, with the marginal value of carbon credits varying
reto optimality. Carbon credits are a resource that can bring economic benefits to all ship- among them. Shipping
companies
ping companies, with higher
with themarginal
marginalabatement costs credits
value of carbon (MAC)varying
can reduceamongtheir abatement
them. Shipping cost
Sustainability 2023, 15, 8388 by purchasing carbon credits, while shipping companies with
companies with higher marginal abatement costs (MAC) can reduce their abatement cost relatively low14 MAC
of 20 can
earn income by carbon
by purchasing selling carbon
credits, credits. From thecompanies
while shipping perspective of MAC,
with the MAC
relatively low MAC of shipping
can
companies under the carbon cap-and-trade is smaller than those
earn income by selling carbon credits. From the perspective of MAC, the MAC of shipping under the carbon tax pol-
icy. In other
companies under words, the carbon cap-and-trade
the carbon cap-and-trade makes
 λ · A · Dis smaller the emission reduction of ships
than thoseunder the carbon tax pol- more
λ H · F0 ·and D can 
M shipping companies
−1 the emission−1reduce
∆CO2 ( A)profitable
icy.
= In other3 words, further
· v2Carbon
thetax − vencourage
carbon2
NO cap-and-trade
+ ·makes
v to
Carbon tax − v NO
emissions.
reduction of ships
(22)more
168 ( v o ) 168
As shown in the Figures 9–11, the MAC of a ship
profitable and can further encourage shipping companies to reduce emissions. under the carbon cap-and-trade
policyAsisshown
closelyin
Considering and
the the complexly
Figures
optimal related
9–11,
speed the to
determined itswithout
MAC vessel
of a shipspeed,
under
a policyfuel price,
the vessel
carbon
controlling size, and car-
cap-and-trade
carbon emis-
sions,
bon
policy that is, vnull is
allowance.
is closely MACa known
and quantity,
decreases
complexly asthe
themodel
related fuel
to curve increases,
its price
vessel was plotted
speed, by
fueland MATLAB
thevessel
price, assize,
larger follows,
the and
shipcar-
type,
Figures
the
bonlarger12 and
allowance. 13:
the MAC.
MAC decreases as the fuel price increases, and the larger the ship type,
the larger the MAC.

Figure9.9.Graph
Figure Graphofof MAC
MAC against
against different
different fuel fuel prices
prices underunder the carbon
the carbon cap-and-trade.
cap-and-trade.
Figure 9. Graph of MAC against different fuel prices under the carbon cap-and-trade.

Figure10.
Figure
Figure 10.Graph
Graph
Graphofof
of MAC
MAC
MAC against
against
against different
different
different carbon
carbon
carbon allowances.
allowances.
allowances.
Sustainability 2023, 15, x FOR PEER REVIEW 15 of 20
Sustainability 2023, 15, 8388 15 of 20

Sustainability 2023, 15, x FOR PEER REVIEW 16 of 20

Figure 11.
Figure Graph of
11. Graph of MAC
MACagainst
againstdifferent
differentvessel types
vessel under
types the cap-and-trade
under policy.
the cap-and-trade policy.

2. Marginal Abatement Cost of the Carbon Tax Model

ΔC ( A ) = CCost of implementing carbon tax policy -C No carbon emission limit costs

F0 ⋅ D ⋅ ( λ H δ + PHFO ) F0 ⋅ D ⋅ PHFO D ⋅ ( A ⋅ ( PMFO + λM δ ) + C )
ΔC ( A ) = 2
v Carbon tax -
2
vNO + -1
⋅ v Carbon
168 ( v ) 168 ( v )
tax
o 3 o 3 168 (20)

D ⋅ ( A ⋅ PMFO + C ) -1 A ⋅ λM δ  N Q 
- ⋅ vNO + ⋅   i + Nt pil 
168 168  i=1 l 
F0 ⋅ D ⋅ λH δ F0 ⋅ D ⋅ PHFO D ⋅ A ⋅ λM δ -1
ΔC ( A ) = 2
v Carbon tax + (v 2 2
− vNO )+ ⋅ v Carbon tax
168 ( v ) 168 ( v )
Carbon tax
o 3 o 3 168
(21)
D ⋅ ( A ⋅ PMFO + C ) -1 A ⋅ λM δ  N Q 
+ ⋅ ( vCarbon tax − vNO
-1
) + ⋅   i + Nt pil 
168 168  i =1 l 
λH ⋅12. ⋅ D ofof2MAC
F0Graph λMfuel
⋅ Aprices
⋅ Dunder
ΔCO2 ( A )Figure
=
Figure 12. Graph⋅ ( vCarbon
MAC taxCarbon tax − vNO ) +
against
against 2
different fuel
different prices
⋅ (under tax − vNO )
-1the carbon
vCarbon -1 policy.
tax
the carbon tax policy.
168 ( v )
o 3
168 (22)
Environmental economic policies mainly follow the Pigovian and Coase approaches.
As the carbon tax is essentially an environmental tax, the theoretical basis of the carbon
Considering
tax policy is groundedthe optimal speed determined
on the Pigovian tax theory,without a policyby
which operates controlling
adjustingcarbon
relativeemis-
sions, that is, v is a known quantity, the model curve was plotted by MATLAB
prices and aims to correct environmental externalities by eliminating the difference between
null as fol-
lows, Figures
private 12 and
and social 13: caused by pollutant emissions. If a Pigovian tax is used to tax
prices
carbon emissions from ships, then shipping companies will evaluate their options for
action under the effect of economic incentives. When the MAC of the shipping company is
lower than the unit carbon tax rate δ, abatement is profitable, and the shipping company
can reduce tax payment by abatement. When the MAC of the shipping company is
equal to the emission tax rate, the emission tax rate δ is optimal. When the MAC of the
shipping company is greater than the emission tax rate, the abatement policy is less effective
or ineffective.
Sustainability 2023, 15, 8388 16 of 20
Figure 12. Graph of MAC against different fuel prices under the carbon tax policy.

13. Graph
Figure 13.
Figure GraphofofMAC
MACagainst vessel
against typetype
vessel under the carbon
under tax policy.
the carbon tax policy.
As presented in the figure, the MAC of a vessel is closely related to its speed and size
Environmental
and the economic
price of fuel. MAC policies
increases withmainly follow
an increase the Pigovian
in vessel speed and and Coase approaches.
decreases with
As the carbon
an increase taxprice
in fuel is essentially
and vesselan environmental
size. tax, price,
The lower the fuel the theoretical basis
the higher the of the carbon
marginal
tax
costpolicy is grounded
of abatement, on the
and hence, Pigovian
lower taxnegatively
fuel prices theory, which
impactoperates by adjusting
energy conservation andrelative
prices
emissionand aims to correct environmental externalities by eliminating the difference be-
reduction.
tween private and social prices caused by pollutant emissions. If a Pigovian tax is used to
Analysis of the Case Calculation Results
tax carbon emissions from ships, then shipping companies will evaluate their options for
actionWith
underthe other parameters
the effect held constant,
of economic the carbon
incentives. Whentax (carbon
the MACprice)
of thewas increased
shipping company
for research.
is lower than
Figure 14 the unit carbon
indicates tax rate
that, under
δ , abatement
the carbon tax policy,isthe
profitable, and the
MAC of carbon shipping com-
emissions
pany can reduce
with optimized taxwas
speed payment
greaterby abatement.
than When
0 and less than the the
carbonMAC
tax.of the shipping
Optimizing company is
the vessel
equal to the emission tax rate, the emission tax rate δ is optimal.
speed to reduce emissions is, therefore, profitable, and shipping companies are willing
When the MAC of the
to adopt this method to reduce emissions and taxes. Under the carbon cap-and-trade
shipping company is greater than the emission tax rate, the abatement policy is less effec-
policy, when the trading price is less than USD$50, the marginal emission reduction cost is
tive or ineffective.
greater than zero, indicating that the company needs to pay some economic cost to reduce
As presented
emissions in thethe
by optimizing figure, the MAC
shipping speed.ofTherefore,
a vessel isthe
closely related
company willto itsconsider
not speed and size
and the price
reducing of fuel.
emissions MAC increases
by optimizing with an
the shipping increase
speed. in vesselthe
Furthermore, speed and
smaller thedecreases
carbon with
an increase in fuel price and vessel size. The lower the fuel price, the higher theby
price, the larger the MAC, and the shipping company will reduce the abatement cost marginal
purchasing carbon credits to maximize profits. When the carbon price is less than USD$50,
the marginal emission reduction cost is greater than zero. The shipping company does
not need to pay economic costs to optimize the shipping speed and can obtain certain
benefits. Therefore, the shipping company is willing to reduce emissions by optimizing the
shipping speed. In the case of low carbon trading prices, the carbon cap-and-trade policy
has less incentive for shipping enterprises to reduce emissions than the carbon tax policy.
With the continuous maturation of the carbon trading market and the continuous increase
in carbon trading price, the carbon cap-and-trade policy will be welcomed by shipping
enterprises that actively reduce emissions, and the enthusiasm of shipping companies
to reduce emissions will be greatly enhanced. From the perspective of the long-term
development of the green shipping market, a carbon cap-and-trade policy is better than a
carbon tax policy and more in line with the sustainable development of shipping.
 policy has less incentive for shipping enterprises to reduce emissions than the carbon tax
policy. With the continuous maturation of the carbon trading market and the continuous
increase in carbon trading price, the carbon cap-and-trade policy will be welcomed by
shipping enterprises that actively reduce emissions, and the enthusiasm of shipping com-
panies to reduce emissions will be greatly enhanced. From the perspective of the long-
Sustainability 2023, 15, 8388 17 of 20
term development of the green shipping market, a carbon cap-and-trade policy is better
than a carbon tax policy and more in line with the sustainable development of shipping.

3
MAC
2

0
30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125
-1

carbon cap-and-trade carbon tax

Figure14.
Figure 14. Graph
Graph of
of calculated
calculated MAC
MAC results
results under
under the
the carbon
carbon tax
tax and
and carbon
carbon cap-and-trade
cap-and-trade policies.
policies.

7.
7. Conclusions
Conclusions
This
This study
study considers
considers different
different carbon
carbon emission
emission policies
policies to
to construct
construct an
an optimization
optimization
model
modelforfor container
containerliner
liner shipping,
shipping, design
design anan improved
improvedWhale
WhaleSwarm
Swarm Algorithm
Algorithmto to solve
solve
related
related issues,
issues, and
and use
use the
the marginal
marginal carbon
carbon abatement
abatement cost
cost method
method to to analyze
analyze the
the deep-
deep-
seated
seatedreasons
reasonsforforthe
theoptimization
optimization ofof
liner shipping
liner according
shipping accordingto different carbon
to different emission
carbon emis-
policies, thereby revealing the underlying reasons of emission-reduction decisions.
sion policies, thereby revealing the underlying reasons of emission-reduction decisions. The
following conclusions were derived:
The following conclusions were derived:
• The nonlinear optimization model constructed in this paper and the improved Whale
Swarm Algorithm designed in this paper can well solve the container transporta-
tion optimization problem under different carbon emission policies and can provide
enterprises with better ship speed, ship investment, and ship type selection reference.
• Both kinds of carbon emission regulation policies will lead to a reduction in the profits
of liner transport companies, a reduction in average speed, and a reduction in carbon
emissions. The carbon tax model has the greatest impact on the profits of shipping
companies, and carbon cap-and-trade is easier to obtain support from enterprises.
However, the current carbon trading price is too low, and its market regulation effect
is weak.
• Under the carbon tax policy, the marginal cost of reducing carbon emissions by means
of optimizing shipping speed is less than that of a carbon tax, and the emission
reduction method of optimizing shipping speed is profitable. The shipping company
is willing to reduce emissions by this method to reduce tax. Under the condition
that the current carbon trading market is not mature enough and the trading price
is low, compared with the carbon tax model, the carbon cap-and-trade model has
less incentive for shipping enterprises to reduce emissions. As the carbon trading
market continues to mature and the carbon trading price continues to rise, shipping
companies will greatly improve their enthusiasm for reducing emissions. From the
perspective of the emission reduction effect and enterprise enthusiasm, the carbon
cap-and-trade model is better than the carbon tax model, which is more in line with
the sustainable development of shipping.
• The implementation of a carbon cap-and-trade policy or carbon tax policy is closely
and complexly related to carbon trading price, carbon tax rate, fuel price, and ship
size, and there are uncertainties. The uncertainty caused by these factors makes the
uncertainty of the implementation effect of each policy should arouse the attention
of relevant decision-makers. The optimal measures should be formulated under the
premise of fully understanding the uncertainty factors.
Sustainability 2023, 15, 8388 18 of 20

• Policy suggestions: (1) It is the general trend to formulate carbon emission policies
reasonably and achieve carbon emission reduction through dual means of regulation
and market; (2) In the case of immature carbon trading markets, carbon taxes are more
flexible, allowing enterprises to respond to market signals in the most economical way,
choose between paying taxes and reducing carbon emissions, and significantly elimi-
nate the environmental externalities of carbon emissions. However, the imposition
of a carbon tax will increase the financial burden of shipping companies and inhibit
the development of the shipping market to some extent. Therefore, the setting of a
carbon emission tax rate is particularly important. We can refer to the current marginal
emission reduction cost MAC of the shipping industry to set the tax rate in line with
the actual development. (3) The emission reduction effect of carbon cap-and-trade
in total cap control is clear. No matter what behavior choice shipping companies
adopt, the total emission reduction control target in the equilibrium shipping market
is constant and can promote the development of low-carbon production technology.
Therefore, the establishment of a carbon emission trading market in the shipping
market should be accelerated.
This study does not consider the weather factor, the weather has a certain impact on
ship navigation, and the impact has significant randomness, real-time and dynamic. In the
future, environmental factors can be taken into account to establish an optimization model
of variable speed transportation affected by random factors.

Author Contributions: Conceptualization and methodology, X.L., X.Z. and Q.T.; writing—original
draft preparation, X.L. and Q.T.; writing—review and editing, X.L.; software, X.L. and X.Z.; supervi-
sion, X.Z. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Marine Economy Development Special Fund Project of
Guangdong Province—Guangdong Natural Resources Cooperation [2020], grant number 071.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.
Description of the Variables:
Decision variables
nm is the number of m vessels assigned;
is variable 0, 1, and its value is 1 when m vessel is configured on the route and 0
xm
otherwise;
Vm is ship speed (kn);
Port-related parameters
D Total route distance (kn);
qij is the weekly freight volume between ports j and i (TEU);
N Number of ports on the route;
rij is the freight rate between ports j and i (USD$/TEU).
tpil is the sum of berthing and unberthing time of the ship in and out of the port (hour);
l The overall average handling efficiency of the port (TEU/hour);
Cm Fixed daily rate of m-vessel (USD$/day);
Cl Port charges for loading containerized cargo (USD$/TEU);
Cu Port charges for unloading containerized cargo (USD$/TEU);
0 is the entry fee of the m vessel (USD/time), including berthing fee, mooring and
Gm
unmooring fee, port clearance fee and other charges incurred in port;
Sustainability 2023, 15, 8388 19 of 20

Ship-related parameters
M is the set of ship types.
QCO2 Weekly average CO2 emissions of the fleet (tons).
CCO2 Costs of carbon emissions.
f day (tp) is the average daily profit of the fleet (USD$).
Fm Main engine daily fuel consumption of m-vessel at designed speed (t/day).
Am Auxiliaries daily fuel consumption of m-vessel at designed speed (t/day).
PIFO Heavy crude oil prices (USD$/ton).
PVLSFO Very low sulfur fuel oil prices (USD$/ton).
λ Carbon conversion factor of ships.
T The expression of the total voyage time (day).
Yk The total amount of cargo (tons) on segment k.
Qi is the loading and unloading of port i.
Cop The weekly operating costs of the fleet (USD$).
CP The average port charges per week (USD$).

References
1. Cheng, T.; Farahani, R.; Lai, K.; Sarkis, J. Sustainability in maritime supply chains: Changes and opportunities for theory and
practice. Transp. Res. E Logist. Transp. Rev. 2015, 78, 1–2. [CrossRef]
2. Fourth Greenhouse Gas Study 2020. (11/10/2022). Available online: https://www.imo.org/en/OurWork/Environment/Pages/
Fourth-IMO-Greenhouse-Gas-Study-2020.aspx (accessed on 1 March 2023).
3. John, B. Intergovernmental Panel on Climate Change Special Report on Global Warming of 1.5 ◦ C Switzerland: IPCC. Popul. Dev.
Rev. 2018, 45, 251–252.
4. IMO Agrees to Cut Emissions by at Least 50% by 2050. Lloyd’s List. Available online: https://lloydslist.maritimeintelligence.
informa.com/LL1122195/IMO-agrees-to-cut-emissions-by-at-least-50-by-2050 (accessed on 2 March 2023).
5. Xing, Y.; Yang, H.; Ma, X.; Zhang, Y. Optimization of ship speed and fleet deployment under carbon emissions policies for
container shipping. Transport 2019, 34, 260–274. [CrossRef]
6. Aldy, J.E.; Pizer, W.A. The competitiveness impacts of climate change mitigation policies. J. Assoc. Environ. Resour. Econ. 2015, 2,
565–595. [CrossRef]
7. Yang, H.; Ma, X.; Xing, Y. Trends in CO2 emissions from China-oriented international marine transportation activities and policy
implications. Energies 2017, 10, 980. [CrossRef]
8. Kontovas, C. Reduction of emissions along the maritime intermodal container chain: Operational models and policies. Marit.
Policy Manag. 2011, 38, 451–469. [CrossRef]
9. Fagerholt, K.; Laporte, G. Norstad. Reducing fuel emissions by optimizing speed on shipping routes. J. Oper. Res. Soc. 2010, 61,
523–529. [CrossRef]
10. Wang, S.; Qiang, M. Sailing speed optimization for container ships in a liner shipping network. Transp. Res. Part E Logist. Transp.
Rev. 2012, 48, 701–714. [CrossRef]
11. Norlund, E.K.; Gribkovskaia, I. Corrigendum to “Reducing emissions through speed optimization in supply vessel operations”
[Transp. Res.—Part D 23 (2013) 105–113]. Transp. Res. Part D Transp. Environ. 2013, 24, 135. [CrossRef]
12. Wang, S.; Wang, X. A polynomial-time algorithm for sailing speed optimization with containership resource sharing. Transp. Res.
Part B Methodol. 2016, 93, 394–405. [CrossRef]
13. Wang, Y.; Meng, Q.; Du, Y. Liner container seasonal shipping revenue management. Transp. Res. Part B Methodol. 2015, 82,
141–161. [CrossRef]
14. Guericke, S.; Tierney, K. Liner shipping cargo allocation with service levels and speed optimization. Transp. Res. Part E Logist.
Transp. Rev. 2015, 84, 40–60. [CrossRef]
15. Norlund, E.K.; Gribkovskaia, I. Environmental performance of speed optimization strategies in offshore supply vessel planning
under weather uncertainty. Transp. Res. Part D Transp. Environ. 2017, 57, 10–22. [CrossRef]
16. Wen, M.; Pacino, D.; Kontovas, C.; Psaraftis, H. A multiple ship routing and speed optimization problem under time, cost and
environmental objectives. Transp. Res. Part D Transp. Environ. 2017, 52, 303–321. [CrossRef]
17. Cepeda, M.A.F.; Assis, L.F.; Marujo, L.G.; Caprace, J.-D. Effects of slow steaming strategies on a ship fleet. Mar. Syst. Ocean.
Technol. 2017, 12, 178–186. [CrossRef]
18. Adland, R.; Jia, H. Dynamic speed choice in bulk shipping. Marit. Econ. Logist. 2018, 20, 253–266. [CrossRef]
19. Karsten, C.; Ropke, S.; Pisinger, D. Simultaneous optimization of container ship sailing speed and container routing with transit
time restrictions. Transp. Sci. 2018, 52, 769–787. [CrossRef]
20. Corbett, J.J.; Wang, H.; Winebrake, J.J. The effectiveness and costs of speed reductions on emissions from international shipping.
Transp. Res. Part D Transp. Environ. 2009, 14, 593–598. [CrossRef]
21. Zheng, J.; Ma, Y.; Ji, X.; Cheng, J. Is the weekly service frequency constraint tight when optimizing ship speeds and fleet size for a
liner shipping service? Ocean. Coast. Manag. 2021, 212, 105–815. [CrossRef]
Sustainability 2023, 15, 8388 20 of 20

22. Zhao, Z.; Wang, X.; Wang, H.; Cheng, S.; Liu, W. Fleet Deployment Optimization for LNG Shipping Vessels Considering the
Influence of Mixed Factors. J. Mar. Sci. Eng. 2022, 10, 2034. [CrossRef]
23. Kim, J.; Kim, H.; Lee, P.T. Optimising containership speed and fleet size under a carbon tax and an emission trading scheme. Int.
J. Shipp. Transp. Logist. 2013, 5, 571–590. [CrossRef]
24. Lee, T.; Chang, Y.; Lee, P. Economy—Wide impact analysis of a carbon tax on international container shipping. Transp. Res. Part A
Policy Pract. 2013, 58, 87–102. [CrossRef]
25. Huang, Z.; Shi, X.; Wu, J. How Will the Marine Emissions Trading Scheme Influence the Profit and CO2 Emissions of a
Containership. In Proceedings of the 4th International Conference (ICCL 2013), Copenhagen, Denmark, 25–27 September 2013;
Springer: Berlin/Heidelberg, Germany, 2013; Volume 8197, pp. 45–57.
26. Cullinane, K.; Cullinane, S. Atmospheric emissions from shipping: The need for regulation and approaches to compliance. Transp.
Rev. 2014, 33, 377–401. [CrossRef]
27. Kim, J.-G.; Kim, H.-J.; Lee, P.T.-W. Optimizing ship speed to minimize fuel consumption. Transp. Lett. 2014, 6, 109–117. [CrossRef]
28. Wang, C.; Xu, C. Sailing speed optimization in voyage chartering ship considering different carbon emissions taxation. Comput.
Ind. Eng. 2015, 89, 108–115. [CrossRef]
29. Wang, C.; Chen, J. Strategies of refueling, sailing speed and ship deployment of containerships in the low-carbon background.
Comput. Ind. Eng. 2017, 114, 142–150. [CrossRef]
30. Zhu, M. Modeling the impacts of uncertain carbon tax policy on maritime fleet mix strategy and carbon mitigation. Transport
2018, 33, 707–717. [CrossRef]
31. Cheaitou, A.; Cariou, P. Greening of maritime transportation: A multi-objective optimization approach. Ann. Oper. Res. 2018, 273,
501–525. [CrossRef]
32. Xin, X.; Wang, X.; Tian, X.; Chen, Z.; Chen, K. Green scheduling model of shuttle tanker fleet considering carbon tax and variable
speed factor. J. Clean. Prod. 2019, 234, 1134–1143. [CrossRef]
33. Xing, H.; Spence, S.; Chen, H. A comprehensive review on countermeasures for CO2 emissions from ships. Renew. Sustain. Energ.
Rev. 2020, 134, 110222. [CrossRef]
34. Ma, W.; Lu, T.; Ma, D. Ship route and speed multi-objective optimization considering weather conditions and emission control
area regulations. Marit. Policy. Manag. 2020, 48, 1053–1068. [CrossRef]
35. Lan, X.; Tao, Q.; Wu, X. Liner Shipping Network Design with ECAs: A Real Case Study. Sustainability 2023, 15, 3734. [CrossRef]
36. ESMA Final Report (2022, ESMA70-445-38). Available online: https://www.esma.europa.eu/databases-library/esma-library?f%
5B0%5D=basic_%3A44&page=2 (accessed on 3 March 2023).
37. Eide, M.S.; Endresen, O.; Skjong, R. Cost-effectiveness assessment of CO2 reducing measures in shipping. Marit. Policy Manag.
2009, 36, 367–384. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
Reproduced with permission of copyright owner. Further reproduction
prohibited without permission.