Voyage Optimization
Voyage Optimization
Voyage Optimization
Master Thesis
4 )=,2)0+,26 9'24> @'49)4.(4=
.(% ,9? '4 9A'4= '2'
%
& &
'&
(
& &)
&*
+
%,
**-'&-,./01
2
&&&
&
3&
-4 5
5
5
&
&&
&
-6
&
&&
&
7
&
&
-
4 &
3&
& 5
&&
5
- 4
&2&
'
895
:&
&
5
-4
3& &
&&
5
&5
&
&
&
5-4
&
5
&
&-$
:;
-(
5&
&
&
7
&
&
5
<
"-
"-:;
&
&
-
'
&
& 5 &
&
- 4 &
&
&
5
5
&
&-4 &
5
&
& 5
5
& -
&&
5
7 &
&
-
6
'
&
&2&
' &&
Preface
This master’s thesis is completed as a part of Nordic Master in Maritime Engineering double degree
program. The thesis is submitted to two universities; the Technical University of Denmark (DTU)
and Aalto University, Finland.
The supervisors of the thesis are associate professor Ulrik Dam Nielsen from the DTU Department
of Mechanical Engineering, Section of Fluid Mechanics, Coastal and Maritime Engineering and
professor Pentti Kujala from Aalto University, Department of Mechanical Engineering, Marine
and Arctic Technology. The thesis was written in a co-operation with Finnish ship design and
operation software house NAPA. The academic advisor from NAPA is Pekka Pakkanen, the director
of development in NAPA Shipping Solutions.
Miika-Matti Ahokas
i
Acknowledgements
I would like to thank my thesis supervisors, associate professor Ulrik Dam Nielsen in the Technical
University of Denmark and professor Pentti Kujala in Aalto University for the participation in
this project. Your patience and flexibility when guiding my work to complete this thesis has been
extremely valuable. I would like to give special thanks to my advisors Pekka Pakkanen and Risto
Kariranta for the project supervision at NAPA and providing an interesting thesis subject to work
with. I would also like to thank all my team mates at NAPA Shipping Solutions for all the guidance
and valuable information around the thesis subject.
Moreover, I would like to thank my friends and fellow students for the journey during the master’s
studies. Without your support, I would not be able to overcome all the challenges encountered during
these two years.
Finally, the greatest thanks go to my newborn daughter for enchanting me during the final stages of
this project and to my wife, Heini, for the whole-hearted support and cheer during this process.
ii
Contents
1 Introduction 1
2 Shipping process 3
2.1 Stakeholders in the freight market . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1 Charterer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.2 Shipowner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Contracts and clauses used in shipping process . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Charter parties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Clauses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Shipping costs in voyage charter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.1 Voyage costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.2 Operating costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.3 Cargo handling costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.4 Capital costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Voyage cashflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4.1 Freight rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4.2 Time-Charter Equivalent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.5 Hypothetical scenarios of voyage charter . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.5.1 Scenario I - Operation with original ETA . . . . . . . . . . . . . . . . . . . . . . 10
2.5.2 Scenario II - Just-in-time operation and implementation of Virtual Arrival . . . . . 12
3 Ship energy efficiency 14
3.1 MARPOL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.1 EEDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.2 SEEMP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.3 EEOI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Operational efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2.1 Slow steaming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.2 Just-in-Time operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Voyage optimization 18
4.1 Motivation for voyage optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.1.1 Fuel consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.1.2 Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1.3 Weather awareness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Voyage planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.1 Voyage scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.2 Passage planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.3 Voyage optimization algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2.4 Voyage data sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3 Voyage optimization strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3.1 Voyage as sailed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3.2 Route as sailed and optimized speed with just-in-time ETA . . . . . . . . . . . . . 23
4.3.3 Optimized route and speed with just-in-time ETA . . . . . . . . . . . . . . . . . . 23
4.4 Expected benefits of voyage optimization . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4.1 Just-in-Time operation and re-scheduling during voyage . . . . . . . . . . . . . . 24
4.4.2 Weather routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
iii
5 Automatic Identification System 25
5.1 AIS broadcast information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.2 Voyage simulation utilizing AIS data . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.3 AIS data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.3.1 Portcall identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.3.2 Data filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.3.3 Draught separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.3.4 Ship payload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3.5 Anchorage identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6 Performance calculation 33
6.1 Acknowledgement of the case company . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.1.1 Ships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.2 Ship performance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6.2.1 Hull resistance model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
6.2.2 Wave added resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.2.3 Wind resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.2.4 Shallow water resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
6.2.5 Propeller model and propulsion coefficients . . . . . . . . . . . . . . . . . . . . . 37
6.2.6 Propulsion power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
6.2.7 Fuel consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.2.8 Emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
6.3 Sea-going voyage data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
6.4 Weather data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.5 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7 Simulations 42
7.1 Simulation methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.1.1 NAPA Voyage Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.1.2 DTU Ship Simulation Workbench . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7.2 Voyage as sailed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
7.3 Just-in-Time Arrival . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
7.3.1 Overview of speed optimization algorithm . . . . . . . . . . . . . . . . . . . . . 48
7.4 Weather optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
7.4.1 Overview of the weather optimization algorithm . . . . . . . . . . . . . . . . . . 48
8 Analysis of the results 49
8.1 Optimization of laden voyages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8.2 Optimization of all voyages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
8.3 Voyage cashflow estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
8.4 Comparison of selected voyages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
8.4.1 Voyage 1 - Eastbound Atlantic crossing . . . . . . . . . . . . . . . . . . . . . . . 54
8.4.2 Voyage 2 - Westbound Atlantic crossing . . . . . . . . . . . . . . . . . . . . . . . 56
8.4.3 Voyage 3 - Eastbound Pacific crossing . . . . . . . . . . . . . . . . . . . . . . . . 57
8.4.4 Voyage 4 - Southbound Philippine Sea crossing . . . . . . . . . . . . . . . . . . . 59
9 Benefit sharing 61
9.1 Voyage optimization benefits for shipping stakeholders . . . . . . . . . . . . . . . . . 61
9.1.1 Benefits from Just-in-Time operation . . . . . . . . . . . . . . . . . . . . . . . . 61
9.1.2 Benefits from Weather routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
iv
9.2 Possible solutions for benefit share . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
9.2.1 Implementation of Virtual Arrival policy . . . . . . . . . . . . . . . . . . . . . . 63
9.2.2 Sharing the saving in bunker cost . . . . . . . . . . . . . . . . . . . . . . . . . . 64
9.3 Contractual changes for benefit share . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
10 Discussion and conclusions 66
10.1 Further studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
11 Bibliography
A Fleet information
B Optimization results of laden voyages
C Optimization results of laden and ballast voyages
v
Abbreviations and symbols
Abbreviations
AIS Automatic Identification System
EEDI Energy Efficiency Design Index
EEOI Energy Efficiency Operational Indicator
ETA Estimated Time of Arrival
HFO Heavy fuel oil
IMO International Maritime Orgnization
IT TC International Towing Tank Conference
MGO Marine gas oil
MMSI Maritime Mobile Service Identity
MR Medium Range tanker
NAPAVO NAPA Voyage Optimization
RPM Revolutions per minute
RTA Required Time of Arrival
SEEMP Ship Energy Efficiency Management Plan
SFOC Specific fuel oil consumption
SOLAS Safety of Life at Sea
SSW DTU Ship Simulation Workbench
ST M Sea Traffic Management
TCE Time Charter Equivalent
V LCC Very Large Crude Carrier
VTS Vessel Traffic Service
vi
Roman symbols
AF ship wind area
B breadth
Cb block coefficient
Cb midship section coefficient
Cp prismatic coefficient
Cwp waterplane area coefficient
CX wind resistance coefficient
D propeller diameter
DW T ship deadweight
EAR propeller expanded area ratio
g acceleration due to gravity
Hs significant wave height
h water depth
Kt propeller thrust coefficient
KQ propeller torque coefficient
LPP length between perpendiculars
LW L waterline length
LOA length overall
MCR maximum continious rating
MFC fuel consumpiton mass
P engine power
PD delivered power
PE effective power
PL ship payload
PoD propeller pitch/diameter ratio
R resistance
S wetted surface area
T draught
Td design draught
TP wave period
t thrust deduction
V velocity
w wake fraction
Z number of propeller blades
vii
Greek symbols
ηH hull efficiency
η0 propeller open water efficiency
ηRR relative rotative efficiency
ηtrans transmission efficiency
ρ sea water density
viii
1 Introduction
Reducing the emissions in shipping is a common goal for all stakeholders involved in the shipping
process. Development of the Energy Efficiency Operational Indicator (EEOI) and the Ship Energy
Efficiency Management Plan (SEEMP) by IMO urges to improve energy efficiency not only by the
installation of more energy efficient machinery and equipment, but also saving energy by operations
(IMO 2009). However, the contractual mesh between the shipping stakeholders is complicated, thus
leading to situations where some stakeholders may have incentives for inefficient operation. As
an example of this, a ship operator may arrive to a busy harbor at an inefficient speed just to wait
to unload anchored for several days or weeks while receiving compensation (demurrage) from a
charterer. Typically, the demurrage rate is higher than the extra fuel cost which brings an incentive for
the ship operator to arrive to the harbour as early as possible (IMO 2009).
The key objective of the voyage optimization processes is to enhance the ship’s operational per-
formance towards energy efficient shipping. In recent years, several studies discussing the ship
voyage optimization problem have been published. For instance, studies by Lu et al. (2015) and
Chang et al. (2013) employ different voyage optimization strategies such as finding the shortest
route, avoiding bad weather and utilizing strong ocean currents to reduce the fuel consumption and
shipborne emissions on specific ships and voyages. Another study by Jia et al. (2017) estimates the
fuel consumption and emission reduction in a global context by optimizing ship speed to meet the
ETA just in time (Just-in-Time arrival). These studies, among others, conclude that implementation
of voyage optimization processes have a noteworthy benefit on ship’s energy efficiency.
While earlier research on voyage optimization focus mainly on energy efficiency on voyage and ship
level, the objective of this thesis is to study what are the overall benefits of voyage optimization for
a fleet of a specific shipowner. The research problem can be expressed by the following research
questions:
1. What are the energy efficiency, monetary and safety benefits for voyage optimization for a fleet
of specific shipowner?
2. How to share the monetary benefits to different shipping stakeholders, so that there is motivation
to change operational processes?
The scope of this thesis is limited to consider Medium Range tanker (45 000 - 55 000 DWT) voyages
operated under a voyage charter party. For understanding the big picture of shipping processes, the first
part of the thesis discusses the role of shipping stakeholders, operational processes, contractual terms
used in shipping and the present situation of benefit sharing between different shipping stakeholders.
The aim is to respond to the first research question by determining the expected benefits of voyage
optimization to the stakeholders. The problem of conflicting incentives in operational processes and
the present situation of monetary benefit sharing are exemplified by hypothetical operational scenarios.
While the energy saving and emission reduction is the starting point of the voyage optimization,
the consequential benefits e.g. reduction in fuel costs and improved safety are discussed as well.
Particularly, the importance of these benefits is reviewed for different shipping stakeholders.
The second part of the thesis covers the case study where the supposed fuel saving and CO2 reduction
by voyage optimization processes is assessed. The case study uses the operational Automatic
Identification System-data of 49 Medium Range tankers collected between January 2018 and March
2019. All ships are operated by the same shipowner, allowing an overview on the performance of
1
the majority of the company fleet. The ship performance simulations were performed with NAPA
Voyage Optimization tool by utilizing operational AIS-data, weather data, nautical charts and a ship
performance model. The simulations of actual voyages were then compared with optimized voyages
which utilize different voyage optimization strategies. The performance calculation methods and
data-flow of the optimization are presented in the outline. Supplementary simulations of selected
voyages were performed with DTU Ship Simulation Workbench tool and the results were compared
with actual performance records from noon reports.
The third part of the thesis reflects on the first part and the analysis of the case study to discuss
the encounter of the expected benefits and optimization. The aim is to respond to the second
research question by defining possible changes in contractual terms in order that they motivate both
stakeholders. A simple voyage cashflow calculation is included to express in which level the shipping
costs could be reduced on a yearly basis on a fleet level.
2
2 Shipping process
The shipping industry is one of the most important global industry networks connecting several
stakeholders from shipbuilders to ship operators, and eventually to ship scrappers. Due to its global
position, the shipping industry is highly affected by global market cycles. The universal shipping
industry can be divided into four separate markets (Stopford, 2009):
1. Freight market
2. Newbuilding market
3. Sale and purchase market
4. Demolition market
The analysis in this thesis is focused on the freight market and to the stakeholders involved in sea
transport. For simplicity, the sea transportation process in the freight market is referred hereinafter to
as shipping process and the stakeholders in the freight market is referred to as shipping stakeholders.
The two key stakeholders in shipping are the charterer and the ship operator which conclude the
charter-party to be followed in operation. Charterer is the stakeholder arranging the carriage of cargo.
The ship operator is the stakeholder responsible for the operation of the ship and often the ship
operator is also the stakeholder who owns the ship. A broker is an individual player whose task is to
communicate between the cargo owner and shipowner and find ships for the cargoes when a shipping
contract is about to be negotiated. Since the voyage optimization is affected mainly by the charter
party between the charterer and the shipowner, the following subsections concentrate on the roles of
these stakeholders.
3
2.1.1 Charterer
Charterer is the company who arranges the carriage of cargo by a ship owned by them or hiring a
ship. Depending on the charter party, the charterer either hires a ship with a fixed daily or monthly
price (time charter), or fixed price per ton of cargo transported (voyage charter). Different types of
charter contracts are described further in section 2.2.1.
2.1.2 Shipowner
Shipowner is the company or individual who owns the ship. The shipowner may operate their own
ships or hire ships to a dedicated ship operator company. Generally, the shipowner is responsible
for all shipping related costs, however, under different charter-party arrangements the voyage and
operating related costs may be shared differently between the shipowner, operator and charterer.
Voyage charter
In a voyage charter, the freight is determined with a fixed price per ton of cargo transported. The
voyage charter is typically used in tanker and bulk shipping where homogeneous goods are transported
in large quantities. Generally all voyage-related costs are paid by the shipowner (Stopford, 2009). A
typical scenario of the voyage charter is, for instance, when a cargo owner has a need to transport
petroleum products from port A to port B. The cargo owner contacts a broker who finds an available
shipowner with a suitable ship and timetable to complete the voyage. The terms for operation will
be set out in a charter party. If the operation is completed within the terms agreed, the process is
complete. However, if the voyage is not completed within the terms agreed in the charter party, a
claim can be raised by either stakeholder. For example, the shipowner is entitled to raise a claim for
demurrage if ship’s laytime at port exceeds the limit stated in the charter party. Respectively, the
charterer is entitled to dispatch if the ship spends less time at port than required in the charter party.
Clauses where these claims are defined are further discussed in section 2.2.2.
Time charter
In time charter, the operational control of the ship is handed over to the charterer with a fixed daily
or monthly price. In this arrangement, the shipowner pays the operating and capital costs of the
ship and the charterer is responsible of voyage and cargo related costs such as fuel, port charges and
loading/discharging fees. (Stopford, 2009).
4
In bare boat charter, only the ship is chartered to a third party who manages the complete freight
operation by organizing the crew and paying the operating and voyage costs. (Stopford, 2009)
Contract of affreightment
Contract of affreightment is typically used on long-term chartering and it can be described as a single
contract for multiple voyage charters. The contract can be concluded, for example, for a certain
period or certain amount of cargo, and the rate can be set by tons of cargo transported or number of
voyages completed. (Järvenpää, 2016).
2.2.2 Clauses
Additional clauses are used together with a charter party, to set the detailed terms and the modus
operandi in specific occasions. The key clauses in the context of this study are described in the
following.
Laytime
Laytime is the agreed period of time for the loading/discharge operation at the port.
Laycan
Laycan is the time period between the earliest and the latest occasion when the ship can commence
the loading/discharge operation.
Demurrage
Demurrage is a compensation paid to the shipowner for the delay in loading/discharge operation for
which he is not responsible (Stopford, 2009). For example, a loaded ship may arrive to a port by
the time agreed in the charter party, but due to a congestion in the port the ship cannot be received
for a discharge within the laycan agreed in the charter party. Therefore, the shipowner receives a
demurrage from the charterer for each extra day spent in operation.
Dispatch
Dispatch is the compensation paid for the charterer if the loading/discharge operation is completed in
less time than the laytime allows. The rate is typically half of the rate of demurrage (Schofield, 1997).
Notice of Readiness
The shipowner sends notice of readiness (NOR) to the charterer when the ship is ready for loading/dis-
charge. Typically, at the moment when Notice of Readiness is announced, the laytime comes into
effect. To announce a valid NOR, the ship needs to be arrived at the port and physically ready for
loading/port operations. (Laajarinne, 2014).
Virtual Arrival
The virtual arrival clause is intended to enable speed adjustment and later arrival time than the ETA
agreed in the charter party, if there is a known delay in the arrival port. The aim of virtual arrival
is to reduce waiting times at port and allow ships to utilize the extra time at voyage instead and,
5
thereby, receive savings in fuel consumption. The virtual arrival clause may include reduction in the
demurrage rate to share the financial benefits between the shipowner and charterer.
According to Stopford (2009), the shipping costs can be categorized into four main groups – operating
costs, periodic maintenance, voyage costs and capital costs. Additionally, the shipping costs include
cargo-handling costs which is omitted in fig. 2 due to its inaccuracy.
The economy of shipping is dependent on the relationship between freight cost and ship size. There-
fore, a suitable way to measure the total cost of shipping process is the unit cost, where the cost is
determined per unit transportable (Stopford, 2009):
OCtm + PMtm +VCtm +CHCtm + Ktm
Ctm = (2.1)
DW Ttm
where:
6
C is the cost per dwt per annum
OC is the operating cost per annum
PM is the periodic maintenance per annum
VC is the voyage cost per annum
CHC is the cargo handling cost per annum
K is the capital cost per annum
DW T is the ship deadweight
Subscript t stands for the year and m for the ship number of the fleet.
In a voyage charter the revenue depends on the quantity and the rate paid per unit of the cargo. For
example, in a case of product tankers, the cargo capacity is typically between 30 000 - 120 000 dwt
which is considerably lower compared to VLCC tankers with capacity over 200 000 dtw. However,
the rate paid per cargo unit in refined petroleum products is higher compared to crude oil which
makes the business cost-effective.
Fuel costs
Fuel costs is the largest variable in voyage related costs. As the power to move the ship increases by
the cubic of the speed, the fuel consumption is assumed to increase proportionally. A commonly used
fuel consumption cubic rule is presented in eq. (2.3) (Stopford, 2009).
S 3
∗
F =F (2.3)
S∗
where:
7
F is the actual fuel consumption
F ∗ is the design fuel consumption
S is the actual speed
S∗ is the design speed
In laden voyages, the operating speed is agreed in the voyage charter party. The ballast voyages are
sailed without a contract, hence the ship operator can choose a suitable operating speed to reach the
next loading port. In both cases it is beneficial to operate with as slow speed as possible to minimize
the fuel costs. However, the prevailing market situation is the crucial factor which dictates the final
operating speed. For example, if the freight market is increasing during ballast voyage, there is no
reason to speed up to arrive the loading port. On the contrary, when the freight market is decreasing,
it is important to arrive the loading port before the cargo rates decrease significantly. Furthermore, in
ballast voyages it is important for the ship operator not to miss the time period for laycan and losing
the cargo by arriving late to the loading port.
where:
The way how a ship is used in daily operations covered by voyage costs has an impact on the
operating costs as well. For example, reduction in fuel costs may have a consequential cut in lubricant
consumption and less frequent repairs and maintenance.
where:
8
CHC is the cargo handling cost
L is the cargo loading cost
DIS is the cargo discharge cost
CL is cargo claims
Typically cargo related costs under a voyage charter party are paid by the charterer, except the cargo
handling costs (Stopford, 2009). Cargo handling costs vary largely depending on the cargo type and
is therefore not included in the example calculation.
The number of days is determined as the time between the ship leaving port after discharge of previous
cargo until the discharge of current cargo. Slower speed yields to a longer time spent at sea and
lowers the Time-Charter Equivalent. Therefore, the speed should be reduced only in cases when
the savings in fuel costs exceeds the reduction of TCE caused by the increase in voyage duration.
(Hellenic Shipping News Worldwide, 2015)
9
2.5 Hypothetical scenarios of voyage charter
The focus when determining the expected benefits is targeted on the voyage charter contract. The
benefits are simplest to be defined when cargo is fully loaded at point A and fully unloaded at point
B. The unbalance in benefit sharing in shipping operations can be clarified by hypothetical example
scenarios described by the following. The fuel consumption and CO2 emission estimations were
calculated by using NAPA Voyage Optimizaton tool.
After 10 days of steaming according to the original voyage itinerary, port authorities at Marseilles
announce that due to shortage in storage tank space at the port, the 50 000 tons of petroleum products
can not be unloaded from the ship until five days after the original ETA. The Ship master decides to
proceed with original ETA and the ship arrives at port after 20 days from the departure. The ship is
anchored near the port to wait for the port authority’s permission to maneuver to the berth. After the
five days of wait, the ship is maneuvered to the berth for discharge. According to the voyage charter
agreement, the charterer is bound to compensate the ship operator 18 000 $ demurrage rate per day
for the time which could have been utilized in other operations if the ship was discharged according
to the original agreement. The total fuel consumption in this route is 390 tons of HFO based on the
performance calculation with NAPA Voyage Optimization tool. With a bunker price of 400 $/ton the
total cost of fuel per voyage would be 156 000 $. However, by reducing average steaming speed to
8 knots for the remaining voyage, the ship would arrive just-in-time to the port for discharge. The
calculated total fuel consumption would be then 340 tons. Therefore, the cost of bunker would be
10
136 000 $. By receiving 18 000 $ of demurrage each day spent at anchor, the ship operator would
gain 90 000 $ in total during the five days of wait. If the demurrage is used straight to compensate the
fuel costs, the monetary benefits for the ship operator would be:
Therefore, the percentual saving compared to voyage sailed in 25 days in fuel cost is 51% for operating
inefficiently. This comparison assumes that any demurrage is not paid during the extra five days
spent en route when speed is reduced after 10 days. The 20-day voyage produces 15 % more CO2
emissions compared to the 25-day voyage. Summary of the scenario is included in fig. 4.
The benefits for the scenario focuses on the ship operator which receives demurrage from waiting
at the anchor. Typically, the demurrage rate is higher than the expenses of fuel cost when steaming
efficiently, so the ship operator has incentive to operate inefficiently. Additionally, the ship operator
could have time for maintenance jobs etc. while at anchor.
The disadvantages of the scenario fall upon the charterer since they are bound to pay demurrage to the
ship operator for the delay in port. General disadvantages are more emissions in total and emissions
11
produced in port areas. Generally it can be also discovered that the benefits are not in balance between
stakeholders.
The ship operator reduces the voyage speed and new voyage itinerary is made to meet the Required
Time of Arrival. The ship operator has assessed the ship’s performance and fuel consumption to
ensure that the speed reduction is reasonable.
The ship is acknowledged as “virtually arrived” at Marseilles port 20 days after the departure of the
ship while, in reality, the ship is still en route. Five days later there is space in storage tank and the
ship is arrived at the port accordingly to the Required Time of Arrival and berthed for discharge. The
total calculated fuel consumption by steaming with average speed of 8 knots for the remaining 15
days would be 340 tons. The ship operator receives 75 % of the demurrage rate for each extra day
spent en route. The monetary benefits for the ship operator would then be:
Therefore, the shipowner saves 50% in the fuel costs by operating efficiently and the CO2 emissions
are 15% less compared to the 20-day voyage. Additionally, the demurrage which charterer is bound
to pay is 25% less than in previous scenario. The monetary benefit for ship operator is 3.5% less
compared to the previous scenario but yields to more efficient operation and balanced benefits between
both stakeholders. Summary of just-in-time operation with Virtual Arrival is presented in fig. 5.
12
Figure 5 – Summary of the just-in-time and virtual arrival scenario.
Distinguishable benefit for the charterer, compared to the previous scenario, is the reduction in
demurrage rate paid to the ship operator for the delay. Furthermore, the ship operator has an incentive
to operate efficiently, since the energy saving and reduction of emissions is cost effective. The
reduction of speed yields to improved safety and less congestion at port areas. Reduction in crew
fatigue is also conceivable benefit.
In general, the share of benefits is more balanced than in previous scenario. Less emissions are
produced especially in port areas when implementing just-in-time arrival.
13
3 Ship energy efficiency
The Third IMO GHG study 2014 has estimated that shipping produces 2.2% of anthropogenic CO2
emissions (IMO, 2015b). In recent years, the shipborne emissions have become more and more
controlled by regulative measures such as Energy Efficiency Design Index for newbuildings and Ship
Energy Management Plan set by International Maritime Organization. This chapter concerns the
technical requirements to improve ship energy efficiency and the improvement in energy efficiency
by operations.
3.1 MARPOL
The International Convention for the Prevention of Pollution from Ships (MARPOL) is a convention
developed to protect the pollution of marine environment by shipborne pollutants. The convention
was approved by IMO in 1973 and came into force in 1983. Today, the convention consists of
six annexes, and the latest amendment Annex VI – Prevention of air pollution from ships and its
regulations considering ship energy efficiency are discussed in the following.
3.1.1 EEDI
Energy efficiency index for newbuildings came into effect as an amendment under the regulation
of MARPOL Annex VI Chapter 4 in July 2011 (IMO, 2011). The purpose of EEDI is to ensure
that energy efficiency of newbuild ships is in an appropriate level. In practice this would result as
installation of more energy efficient appliances and adjustments in design principles to improve the
technical abilities towards energy efficient operations. The EEDI measures energy efficiency by
dividing the CO2 emissions produced with the tonne-mile transport work done. The condition to
fulfill the energy efficiency requirements is to satisfy the minimum energy efficiency level for specific
ship type and size category. Therefore, the EEDI is more a measure of CO2 emissions rather than a
direct indicator of actual energy efficiency. In simple terms, the EEDI can be represented by eq. (3.1)
(IMO, 2016a).
CO2 emission
EEDI = (3.1)
transport work
The actions to improve EEDI in newbuildings can be for instance hull optimization, propeller
optimization, adjustment of design speed, usage of energy saving devices and installation of energy
efficient machinery.
3.1.2 SEEMP
The MARPOL regulation on Ship Energy Efficiency Management Plan entered into force on January
1st 2013 and it requires all ships of more than 400 gross tonnage to carry a ship specific Ship
Energy Efficiency Management Plan. The main purpose of the SEEMP is to create a mechanism
for shipowners to monitor and improve the operational efficiency of a ship. The framework when
developing ship specific SEEMP can be realized by four steps: planning, implementation, monitoring,
and self-evaluation and improvement (IMO, 2016a). In this manner the SEEMP could be developed
by improving the performance continuously.
14
3.1.3 EEOI
The Energy efficiency operational indicator is a optional monitoring index to measure the operational
energy efficiency of a specific ship. The EEOI may be used as an index when monitoring the execution
of SEEMP. The voyage specific EEOI is calculated as:
∑ j FC j ·CF j
EEOI = (3.2)
mcargo · D
The ship specific average EEOI for the period under review is calculated as:
∑i ∑ j (FCi j ·CF j )
EEOIaverage = (3.3)
∑i (mcargo,i · Di )
where:
While EEOI can be a suitable indicator to measure the produced greenhouse gases in ship operations,
the EEOI has drawn criticism for being imprecise when presenting the ship’s energy efficiency.
According to Bimco (2017), "Operational efficiency indices, such as the IMO Energy Efficiency
Operational Indicator, are overly simplistic or even misleading on an individual ship basis and should,
therefore, not be considered for regulatory purposes". The problems of EEDI’s inaccuracy are mainly
due to unavoidable situations of bad weather conditions and ballast voyages which decrease the EEOI
value (Psaraftis, 2019).
15
3.2.1 Slow steaming
As described in section 2.3.1, a reduction in operational speed can decrease the fuel consumption
significantly. Slow steaming refers to operating ship at a significantly lower speed than the design
speed. The general definition of slow steaming is when main engine’s load is less than 60% of the
maximum continious rating (MCR) (IMO, 2016b). The crucial factor of a ship’s ability for slow
steaming is the main engine’s capability to run at low loads. From the increasing trend of slow
steaming, the engine manufacturers have developed slow steaming upgrade kits to allow main engines
to be run at lower loads (Wiesmann, 2010).
When considering slow steaming in the context of the complete shipping process, several benefits
and disadvantages can be found. Table 3 summarizes the benefits and disadvantages from the
perspective of the ship operator and charterer when operating under voyage charter party, and from
an environmental perspective as well (IMO, 2016b).
As summarized in table 3, slow steaming has numerous benefits for the ship operator but also
consequential disadvantages from slow steaming such as higher maintenance costs which diminish
the monetary benefits of the ship operator. When operating under a voyage charter party, the charterer
has no benefit from slow steaming – instead the operator needs to tolerate from longer delivery time of
cargo. From an environmental point of view slow steaming reduces the production of GHG emissions,
but increases the impact of carbon monoxide and PM concentration. Slow steaming also increases the
amount of sludge oil due to the poor combustion by the operating the main engine at low loads.
As discussed in section 2.1, the shipping process is influenced by many factors which are based on the
demand of different shipping stakeholders. Current practices and charter party constraints in shipping
16
industry are still encouraging ship operator to schedule their ships to arrive to the port as early as
possible. Additionally, the slotting issue in ports complicates the scheduling if berth or storage
space is not available. These circumstances create barriers against the successful implementation of
just-in-time operation.
17
4 Voyage optimization
The traditional way of planning ship routes in the past has relied mainly on avoiding bad weather
(Chen, 2013). However, voyage planning by focusing only on weather forecast has limitations on
optimizing efficient ship performance and the reliability of weather forecast decreases as the time
period of the forecast increases. A more advanced voyage optimization technology aims to take into
account the optimal routing, speed adjustment and port scheduling to reduce emissions produced
by shipping. Furthermore, a sophisticated voyage optimization takes into account the ship motions
and responses to enhance operational safety. A recent research, such as Jia et al. (2017) and Lu
et al. (2015), has showed that there is potential to improve ship’s operational performance and reduce
shipborne emissions by the implementation of voyage optimization technology.
Table 4 – Example on the prioritization of voyage optimization objectives under voyage charter party.
In some ship performance optimization cases, shipping stakeholders could achieve benefits in en-
vironmental, monetary and safety related terms by prioritizing just one objective. For example,
optimizing for minimum fuel consumption enhances the ship energy efficiency and reduces the
shipborne emissions. The consequential speed slow-down can enhance the ship safety and comfort
onboard by reducing risk of excessive ship motions, propeller racing or slamming in heavy weather.
18
to the recorded route. Since the reduction in fuel consumption has a consequential effect on reducing
the emissions and bunker cost, the result is remarkable in both environmental and monetary terms.
The research by Jia et al. (2017) uses speed optimization and implementation of just-in-time operation
in the assessment of potential reduction in fuel consumption and emissions. The hypothesis in the
research is that if the waiting time at the arrival port could be minimized and the ship can be berthed
at short notice after her arrival, the voyage can be sailed with a slower average speed and the savings
in fuel consumption and emissions arise from the impact of speed reduction. The research evaluates
5066 voyages operated by 483 Very Large Crude Oil Carriers. The results show that if 50% of the
assessed waiting time at port could be reduced, the average port time saved is 37 hours and the average
fuel consumption saving is 12.5% per voyage. Therefore, optimal port scheduling could benefit the
stakeholder who pays the fuel bill and helps to avoid congestion at the port areas.
4.1.2 Emissions
When estimating the EEOI described in section 3.1.3, the prioritized objective is the reduction
of shipborne emissions. The reduction of emissions is also the key objective in a global scale
since emissions have an impact on environmental sustainability. Because the emissions produced is
directly dependent to the amount of fuel burned, prioritizing minimum fuel consumption or minimum
emissions always avail one another.
IMO has committed to reduce the total annual greenhouse gas emissions at least 50% by the year
2050 compared to 2008 (IMO, 2018). The research by Jia et al. (2017) assess that if only 50%
of the estimated waiting time at the port could be avoided in the operation of VLCC tankers, the
consequential speed slow-down yields to an average reduction of 422 tons of CO2 emissions per
voyage. Relatively, this would be 12.5% of all the emissions produced in VLCC operations. Therefore,
the implementation of just-in-time operation could be one potential solution towards the IMO’s 2050
strategy.
The weather awareness in the voyage optimization process aids in the decision whether to avoid a
certain route due to bad weather conditions or to avoid excess ship motion in waves by voluntary
speed reduction. Considering ship safety, hazards such as parametric resonance, propeller racing,
slamming or shipping of green water should be avoided. (Chen, 2013)
A research by Chang et al. (2013) studies the utilization of strong ocean currents in route planning.
The object in the study is to use favourable ocean currents to reduce fuel consumption and voyage
time. With the information on ocean currents, the currents following the route direction could be used
as an advantage and the currents opposite the route direction can be avoided. The study focuses on
voyages sailed with super-slow steaming, i.e. speed of approximately 12 knots, at Kuroshio region
in East Asia. In the conclusion of the research Chang et al. state that utilizing the favourable effect
19
of Kurushio current could theoretically save 1-8% in voyage time on a example route from Taipei
to Tokyo. Alternatively, by avoiding unfavourable currents on the opposite route could theoretically
save 5-7% in voyage time.
As described in section 2, the starting point in all operations is that there is a demand for transportation.
For example, the cargo owner needs to transport petroleum products from port A to port B. The cargo
owner contacts a shipowner with a suitable ship and timetable to complete the voyage. The voyage
is operated under a voyage charter and the cargo owner pays a fixed rate per ton of cargo for the
complete transport service. Therefore, the voyage planning remains solely as the ship operator’s task
in voyage charter operations.
Initial voyage planning may be done in co-operation with the ship operator’s on-shore planners and
the crew onboard. However, the proper passage planning is done onboard, typically by the second
officer or the captain. The passage plan should include information, for instance, on course and
speed alterations, potential hazards and restricted water depth areas along the route (IMO, 1999).
Voyage optimization services enter into the voyage planning at this stage, since the resources for
estimating the optimum ship performance and proper weather routing onboard may be limited. After
the voyage is initiated, operator and cargo owner monitor the voyage during its execution, for example
by receiving noon reports or using AIS-tracking service. When the ship has reached the arrival port
and the voyage is complete, voyage reports will be prepared as agreed in the voyage charter party.
20
4.2.1 Voyage scheduling
Voyage scheduling can be a complex exercise in the terms of both ship and port operations. Some
shipowners run their operations on a liner shipping service where one or more ships sail on a fixed
route, departing and arriving to port at regular intervals. In such operations keeping the schedule is
essential for maintaining efficient operation. Breskin (2018). The contrary to liner shipping service is
the so called tramp shipping service. Ships operated under the tramp shipping market do not operate
on a regular schedule or routes. Tramp shipping operations are typically run under voyage charter and
the preference in such operation is to minimize the time spent in ballast. Operating under a voyage
charter is more flexible in a sense that the charter party applies to a single voyage. However, the
tramp shipping markets can be highly complex and scheduling operations based on the demand for
tramp shipping services can be troublesome.
The physical limits of the ports and the difference in port operation customs around the world
complicate the voyage scheduling further. For example, low tide, local holidays, the capacity of
cargo handling gear and berthing queue policy can cause unforeseen delays in port operations if these
issues are not taken into consideration when scheduling the voyage (Breskin, 2018). The uncertainties
in port operations and the prevailing first-come first-served berthing policy impede especially the
implementation of just-in-time operation strategy.
In tramp operations, the objectives in voyage scheduling depend on the current market situation.
In ballast voyages, when the ship is empty of cargo and she is sailing to the next loading port, the
fluctuation of the market plays a major part when deciding the operation speed. For example, when
the market is increasing there is no special reason to rush to the next port. However, when the market
is falling, it is important to arrive to the loading port before the cargo price decreases significantly.
Furthermore, the ship operators do not want to take the risk of losing cargo by arriving to the loading
port at the end of the laycan. In laden voyages, when the ship is carrying cargo and the schedule
is already agreed in the charter party, the situation is more stable and the operation speed can be
more carefully adjusted to meet the Just-in-Time ETA. Therefore, it can be stated that the room for
optimization falls upon mainly on laden voyages.
The Appraisal stage should take all voyage relevant information into consideration, such as seawor-
thiness of the ship, operational limitations, potential risks along the route and characteristics of the
cargo.
The planning stage involves detailed planning of the passage from berth A to berth B, including the
areas where the presence of a pilot is required. Voyage optimization services may be used at this
stage to estimate an optimal route in terms of fuel consumption, voyage time, ship and cargo safety,
or comfort onboard.
In the execution stage, the voyage is executed in accordance with the passage plan.
In the monitoring stage, the progress of the voyage is closely monitored. Any changes made to the
passage plan should be clearly marked and recorded.
21
4.2.3 Voyage optimization algorithms
To include the weather awareness and the desired optimization objective in the voyage planning,
sophisticated methods are required to find the best route available. A study by Walther, Rizvanolli,
Wendebourg and Jahn (2016) focuses on the different state-of-the-art methods to model the weather
routing problem and the algorithms to obtain the optimal route. To model the weather routing
problem, the research highlights the approaches where the problem is considered as a constrained
graph problem, a constrained nonlinear optimization problem or as a combination of both. Based
on the formulation of the optimization problem, different methods are then used to solve it. The
highlighted methods are Dijkstra’s algorithm, dynamic programming and optimal control methods,
isochrone methods and iterative approaches. In the conclusion, Walther et al. (2016) state that the
methods to be used strongly depend on the requirements of optimization objectives. For example, the
Dijkstra’s algorithm can lead to accurate results, but it is best suitable for short route calculations due
to a relatively long computation time. Therefore, for reliable results it should be carefully considered
what type of methods are used to optimize a specific objective, but also consider the computational
effort.
Starting point of the case study is to simulate several previous sailed voyages by utilizing AIS data
points from the route, generic ship model and past weather data. The simulation gives output for time
and distance sailed, speed, fuel consumption and produced CO2 emissions. In executable simulations,
it is essential to identify whether the ship has anchored near the port to wait her turn for berthing or
sailed straight to the berth for discharge. Typically ship operators will arrive port as soon as possible
22
within the contractual limits to avoid any due dispatch penalties (IMO, 2016b). Therefore, in most
cases there is some time spent at anchor anyway if there is congestion at the port or port is not ready.
The simulation in the case study utilizes the Required Time of Arrival to calculate the optimal speed
along the route to obtain lowest fuel consumption and CO2 emissions. In the case study, the time
adjustments between the original ETA and RTA are expected to be much shorter compared to the two
hypothetical scenarios.
23
Weather routing can be implemented also without involvement of port authorities. Therefore, in
analysis of the results, the benefits by using the just-in-time ETA and weather routing will be separated.
Weather optimization brings benefits mainly to the ship operator while just-in-time optimization
brings monetary benefits to the charterer and produces less emissions at the port.
24
5 Automatic Identification System
Shipborne Automatic Identification System (later referred to as AIS) is primarily used for transferring
ship identification and tracing information for the ships nearby and shore-based Vessel Traffic Service
(VTS) centers. IMO requires AIS transponders to be fitted in ships of 300 gross tons or above engaged
on international voyages, cargo ships of 500 gross tons or above not engaged on international voyages
and in all passenger ships. The initial objectives of AIS is to avoid collisions and enhance operational
safety by providing supplementary information together with navigational systems. (IMO, 2015a).
General description of shipborne AIS usage is presented in Figure 8.
Primarily ships share AIS information via dedicated VHF radio channels. In busy traffic areas, the
information is shared also with a local VTS center. VTS centers may send additional information
to ships, such as local navigational warnings, traffic management information and port information.
(IMO, 2015a). Along with enhanced situation awareness, information sent by AIS can be used in
many other applications, such as in search and rescue operations or analyzing vessel performance.
25
Table 5 – Information sent by ship with the AIS (IMO, 2015a).
The MMSI (Maritime Mobile Service Identity) is a unique data broadcast identification number which
is input manually. Occasionally, some ships may mistakenly share their MMSI with another ships
which causes errors in their position and possibly very unrealistic transitions on a map (Wu et al.,
2017).
Additionally, the ship can send short safety-related messages addressed either to a specified destination
or all ships in the area. The content of these messages should relate to the safety of navigation, e.g. an
iceberg sighted. The message should be kept at short as possible for clarity reasons. (IMO, 2015a).
According to the IMO (2015a) guidelines for the onboard use of AIS systems, the automatic reporting
interval for AIS information depends on the information type. Static and voyage-related information
should be updated every six minutes and on request. Dynamic information reporting interval is
dependant on ship’s speed and course alteration from two to twelve seconds according to table 6.
When the ship is at anchor, the reporting interval of three minutes is required. Safety-related text
messages are sent only if necessary.
26
(private) provide access for real-time AIS data and also collect and store historical AIS data. The
typical coverage range of onboard transceivers is between 15 to 20 nautical miles and for extending
the range, onshore AIS stations and satellite receivers are used. Typical coverage for onshore AIS
station is 40 nautical miles and satellite receiver can cover a range of 1000 nautical miles.(Tu et al.,
2018)
Unfortunately, the quality of stored AIS data does usually not come with such integrity as described
in table 6. A typical reason for this could be an absence of a data transmission link at the open sea
(Jaskolski, 2017). The data providers may also choose to use a lower time stamp resolution for data
storage. Apart from the resolution, the data may include some erroneous entries as well, for example,
the speed over ground entries may include extremely large or negative values (Tu et al., 2018).
The historical AIS data used in the case study is requested from various commercial providers. Access
to the data is provided by NAPA. The data is collected with terrestrial and satellite receivers. This
increases the data accuracy and integrity compared to using only either method. Overview of AIS
data collecting and storing is presented in Figure 9.
The key information for the case study are ship’s IMO number, ship’s position, time, speed over
ground, navigational status, draught and destination. In sake of efficient data extracting, it is important
to manage with as little information as possible. Therefore, some of the information can be left out
in the data extracting process from the provider and filled in later by other means. For example, the
IMO number identifies the ship in question and the call sign and relevant static information can then
be determined by using web services such as IHS Maritime Portal.
27
5.3 AIS data processing
The raw AIS data requires careful filtering and processing in order to analyze the voyage specific
information and to eventually simulate the ship’s performance during the voyage. For the evaluation
in the case study, the following steps are required to obtain voyage specific information by AIS data
processing:
1. Identify portcalls
2. Filter out short voyages
3. Filter out voyages with bad AIS data resolution
4. Identify ship’s loading condition
5. Identify time spent at anchorage
6. Estimate ship payload
The following subsections cover the details about these AIS data processing steps.
The voyage ends when the two conditions in ship’s speed and the proximity to the nearest port
are fulfilled. The information on existing ports is fetched from the port information database. The
conditions when starting the next voyage are that the next departure port is the same as the previous
arrival port. Respectively, the speed when the ship is not longer stationary needs to exceed the set
28
stationary threshold value. The portcall identification algorithm and port information database used
in the case study simulations are provided by NAPA.
As stated above, the quality of raw AIS data depends on the data provider and the proximity of
AIS data capturing transceivers and stations. Figure 11 illustrates the issue of bad AIS data point
resolution, where the missing data causes the connection between datapoints to egregiously cross
large land areas.
Especially the South China Sea and the Bay of Bengal are problematic areas in terms of AIS data
quality as seen in Figure 11. The data resolution at open sea is poor in many occasions and good
quality resolution is observed only near coastal areas. Unrealistic voyage data can also be detected
from extremely straight connections between datapoints in open sea. The land crossing voyages and
voyages with clearly unrealistic routes were filtered out by comparing the distance of the AIS-based
route with the shortest route from NAPA Voyage Optimization route library. All AIS-based routes
with distance less than 0.9 times the shortest route from the route library were eventually filtered out.
NAPA route library uses the typical sailed routes as the basis of the shortest possible route, thus the
factor of 0.9 was carefully adjusted to include the AIS-based routes which are slightly shorter than
the shortest route from NAPA route library.
29
to be operated in laden condition and half in ballast condition. However, when considering product
tanker operations, it may be profitable enough to sail with part laden condition, instead of fully laden
condition.
The AIS-reported draught of the ships considered in the case study varies from 4.5 to 13.1 meters.
Once again, it should be acknowledged that the draught is reported to the AIS system by the deck
officer and it may be incorrect. The draught ratio limit was set that voyages with a draught ratio <70%
were considered as ballast voyages. An example of draught separation is shown in Figure 12.
The distribution between the ballast and laden voyages in Figure 12 supports the assumption that half
of the voyages are sailed in laden and the other half in ballast condition. However, in the case of
product tankers, in a draught of 8.8 meters some cargo may still be carried between close ports.
PL = CB L pp Bρ T −CBd L pp Bρ Td + DW T (5.1)
where:
30
B is the ship breadth
ρ is the sea water density
T is the ship draught during operation
CBd is the ship design block coefficient
Td is the ship design draught
DW T is the ship deadweight in tons
Ship block coefficient during operation can be approximated with the prevailing draught according to
MAN (2011):
!1/3
Td
CB = 1 − (1 −CBd ) (5.2)
T
The ship payload estimation is later used when calculating the Energy Efficiency Operational Index
for the case study fleet.
A practical way to identify the anchorage time is to define port specific anchoring areas as a subset of
individual ports in port database by using geospatial analysis techniques. The ship can be considered
in anchorage when the ship is stationary inside the anchoring area. Some ports, such as Singapore,
define their anchoring areas precisely as sections for different ship categories and closely manage
the ships in anchorage (Maritime and Port Authority of Singapore, 2019). Therefore, in the analysis
of AIS data, the preferable way is to define the anchoring areas as polygons instead of circle with
specific radius.
The identification of individual anchorages in the case study was done manually for the 50 most
visited ports. The anchoring areas were defined in this case by observing the coordinate points of AIS
data with a geospatial analysis tool QGIS. A ship in anchorage is required to have a certain scope
in the anchor cable length. The scope is defined by the ratio of the length of cable to the depth of
water and the ratio ranges typically from 6 to 10 (IACS, 2017). This scope allows ship to have a
slow circular motion while in anchorage which is shown as a circular patterns when the AIS data
coordinates are plotted on a map. The anchoring areas can be then manually defined by framing the
circular patterns in specific area as visualized in Figure 13. The AIS data points inside this polygon
are used to create a sub-time series and when the following conditions are fulfilled, the ship can be
identified as anchored with a fair degree of certainty: (Holmslykke, 2019)
31
Figure 13 – Identifying the anchorings with geospatial mapping tool.
A more sophisticated way to obtain the anchoring areas would be to automate the process by using big
data methods to track precise ship movements in port areas in order to create the polygons defining
the anchoring areas. However, a development in this kind of approach requires substantial effort and
it is left out from the scope of this thesis.
32
6 Performance calculation
This section covers the methods how ship’s performance can be simulated by using ship performance
model, weather data, depth data and sea-going voyage data. In the case study, a set of realized
voyages are simulated by using NAPA Voyage Optimization tool (hereinafter referred to as NAPA
VO). Additionally, a set of selected voyages are simulated with a DTU Ship Simulation Workbench
tool (hereinafter referred to as SSW) and the results are compared to the actual values based on noon
reports. Flow chart of the performance calculation is presented in Figure 14.
The NAPA VO includes performance models of 55 000 ships in their database. The data sources for
depth data and nowcast weather data are integrated to the system, thus in the context of this thesis
only the sea-going voyage data is needed to be input by using the information from AIS-data to
recreate the sailed voyages. The simulations are run for three different voyage optimization scenarios
as described in section 4.3.
In SSW, the ship model is created from the user defined ship details. The weather data used in the
SSW simulations is extracted from the NAPA VO results, thus the weather conditions are similar in
both cases. In SSW simulations no optimization is applied, thus the simulation is done for the Voyage
as sailed scenario only. The results of these selected voyages are then compared with the NAPA VO
results and the actual voyage information based on noon reports in section 8.4.
33
6.1.1 Ships
The ships used in the case study are 49 Medium Range tankers (45 000 – 55 000 DWT). For
representational purposes, the ships are categorized into 15 ship groups so that all sister ships belong
to the same group. The ship groups are identified with the IMO number of the first ship in the group.
The main dimensions of each ship group are presented in table 7. Although the ships in the same
group are sister ships to each other, some of the dimensions may vary slightly. In these situations the
value of first ship in the group is displayed in the table. Full list with all individual ship information is
included in the appendix A. The information on ship main dimensions is collected from IHS Maritime
Portal.
34
Figure 15 – Ship performance model creation. Redrawn. (Napa, 2019).
The ship model creation in SSW follows similar procedure as described in Figure 15. Instead of
generic models, in SSW the models are defined in more detail by the user. Therefore, much more
details about the ship are needed to be known or estimated when creating the ship performance model.
A well-known approach to approximate hull resistance is the statistical method originally introduced
by Holtrop and Mennen (1982) where the total resistance is calculated by using basic hull dimensions.
The method is developed through a regression analysis based on model experiments and full-scale
data. The total stillwater resistance in Holtrop & Mennen method is subdivided into following
components (Holtrop and Mennen, 1982):
35
Rstill = RF (1 + k1 ) + RAPP + RW + RB + RT R + RA (6.1)
where:
The total stillwater resistance of the ship depends of the ship’s loading condition. Greater draught
increases the wetted surface area and displacement of the ship, yielding to a greater frictional and
wave-making resistance. Generally, the ship is encountering greater stillwater resistance in laden
condition where draught is greater than in ballast condition.
1 2
Rwind = ρairVair CX (ψ )AF (6.2)
2
where:
36
Vair is the wind speed in m/s
CX is the wind resistance coefficient respect to the wind direction
ψ is the relative wind direction; 0 being heading wind
AF is the wind area of the ship exposed to the wind in m2
1
∆VS AM gh 2
= 0.1224 − 0.05 + 1.0 − tanh (6.3)
VS h2 VS2
where:
PE = Rtotal ·V (6.4)
where:
The ship model in NAPA VO includes a 10% increase in total resistance due to the hull fouling.
Additional 15% sea margin is applied as well to provide allowance for significant increase in added
37
resistance.
The actual delivered power is the power measured at the propeller shaft. The delivered power takes
into account the hull, propeller, propulsion and transmission efficiencies. The delivered power is
calculated by:
PE
PD = (6.5)
ηH η0 ηRR ηtrans
where:
Propeller open water efficiency is the ratio of thrust power to the power absorbed by the rotating
propeller without the hull affecting to the propeller inflow. When the hull interferes the inflow to the
propeller, the flow becomes rotational. The relative-rotative efficiency takes account the change of
efficiency due to the rotational flow. The transmission efficiency takes account the energy losses due
to shaft alignment and friction in shaft bearings. If the shaft and engine are coupled with a gearbox,
the transmission efficiency takes account the gearbox efficiency as well.
The fuel consumption per voyage consist from the summation of fuel consumption per each voyage
leg. The principle of voyage legs is described further in section 6.5.
38
6.2.8 Emissions
The production of shipborne emissions is proportional to the amount of fuel burned. The amount of
produced emissions can be evaluated by using emission coefficients for each fuel type. The values of
the constants depend on various circumstances such as the fuel quality and engine conditions. The
emission coefficients used in the simulations are based on the results of the third IMO GHG study
and are shown in table 8.
Figure 16 – All available AIS data of Torm ships from January 2018 to the end of March 2019.
39
6.4 Weather data
Nowcast weather data is utilized in simulations to estimate the wave added resistance, resistance due
to wind and effect of sea currents. Also, in the case of weather routing, the weather data is used to
calculate the optimum voyage including the positive influence of weather. The weather data used in
the performance calculations includes entities for significant waveheight, wave zero-crossing period,
wave direction, wind speed, wind direction, sea current direction and sea current speed. The weather
data provider in Napa VO is TideTech.
The sea current data uses entities for current speed and current direction. Wave components are
separated into wind waves and swells. Both wave types have individual entities for significant wave
height, wave zero-crossing period and wave direction for each data point. The wind data uses entities
for wind speed and wind direction.
6.5 Output
The output results will be calculated for the three scenarios described in section 4.3. The results
in simulations return waypoints based on the AIS coordinates with corresponding timestamps. For
each waypoint, the nowcast weather data is extracted from the weathed data provider. The extracted
data for the waypoints include values for the wind speed and direction, significant waveheight, wave
direction, sea current speed, sea current direction. Water depth is extracted from nautical charts.
The adjacent waypoints are connected to each other with legs which follow the great circle. Each
leg includes calculated results for leg distance, speed over ground, propeller rpm, propulsion power,
fuel consumption and CO2 emissions. The total fuel consumption and CO2 emissions per voyage are
summed from the leg values. Additionally, the results output the histogram of Beaufort distribution
during the voyage. Visualization of waypoints and legs in NAPA VO is presented in Figure 17.
The visualization in Figure 17 is captured from NAPA VO staging platform, where leg specific results
40
are presented along with surrounding weather information such as wind barbs and the different areas
of significant waveheight. The output results in the case study are analyzed and visualized with
MATLAB.
41
7 Simulations
Because the operational details in different voyages can differ considerably, a set of assumptions
is needed to be made to ensure comparable results. For example, in reality different fuel types and
qualities are used and the draught can change during the voyage. The assumptions in all the simulated
voyages are the following:
When estimating the excess time on the voyage, it was assumed that all the time spent at anchorage
is spare time which could be utilized to extend the time spent for steaming. However, some ports,
such as Fujairah in United Arab Emirates and Lagos in Nigeria, use ship-to-ship cargo exchange
operations and offshore loading/discharge stations, when the loading/discharge operations can be
done at anchorage instead of mooring the ship at berth (Holmslykke, 2019). These kind of operations
cannot be identified by the methods used to identify the anchorage as described in section 5.3.5.
42
Table 9 – Calculation methods for ship performance parameters.
In the simulations run with NAPA VO the corresponding ship model is fetched automatically based
on the IMO number. In SSW simulations, the user defines the calculation methods for different ship
performance parameters and the ship model is created based on the input values defined by user.
Table 10 – The input attributes for different simulation scenarios in NAPA VO.
43
In Just-in-Time Arrival and Weather Routing & Just-in-Time Arrival-simulations the operation method
can be chosen from constant speed over ground, constant propeller rpm and specific arrival time. In
both simulations, the operation method was chosen as specific arrival time. The operation method
with specific arrival time uses constant rpm with a varying speed profile when calculating the required
time of arrival.
Table 11 – Resistance calculation methods used in DTU Ship Simulation Workbench model.
The required information to perform the resistance calculation with the selected methods are tabulated
in table 12.
44
Table 12 – The user defined parameters used in DTU Ship Simulation Workbench model for the ship
IMO 9392470.
Hull
L 174.5 Lwl 174.5
LOA 181.99 B 32.2
T 12.0 Cb 0.81
Cm 0.99 Cp 0.82
Cwp 0.915 V 7.458
S 8412.7
Propeller
D 6.64 Z 4
PoD 0.658 EAR 0.556
rpmdesign 113
Propulsion
w 0.31 relative rotative eff. 1.05
t 0.21 transmission eff. 0.98
Engine
PMCR 11620 rpmMCR 124
gear_ratio 1
The ship main dimensions and engine details in table 12 are collected from IHS Maritime Portal.
The propulsion details are estimations based on the NAPA VO model. The propeller diameter is
an estimation based on design draught and the propeller model is based standard B-series propeller
calculated automatically in SSW by using the propeller diameter and number of propeller blades. Gear
ratio is set as 1, since the main engine is assumed to have a straight shaft coupling to the propeller.
45
Figure 18 – All voyages of ship 9392470.
To ensure valid results, the data filtering methods described in section 5.3.2 were applied. The filtering
of simulation results was executed with the following conditions:
The total number of voyages for ship 9392470 decreased from 96 to 46 after the filtering process.
The remaining voyages were then simulated and the results were imported to MATLAB for analysis.
Descriptive statistics of the valid voyages are presented in table 13.
No. of voyages: 46
Duration [h] HFO daily [t] CO2 daily [t] Avg. speed [kn] Port time [h]
Mean 149.7 21.7 67.6 10.4 50.5
Median 124.2 23.0 71.6 11.1 38.0
Min 28.0 4.0 12.5 3.0 0.0
Max 471.8 31.1 96.8 12.8 299.5
Figure 19 shows the relationships between fuel consumption, sailing time and average speed.
46
Figure 19 – Fuel consumption plots of ship 9392470.
From the fitted curves in Figure 19 it can be observed that the relationship between fuel consumption
and sailing time have linear correlation whereas daily fuel consumption and average speed follow
third order polynomial curve.
1. The voyage is sailed with constant rpm to meet the required time of arrival.
2. The voyage speed is not limited by any regional restrictions.
3. All arrival ports utilize the scheduled queue system.
4. All arrival ports have time slot available for the just-in-time arrival.
5. All the ships are capable for slow steaming with minimum of 6 knots.
6. All the time spent in anchor has been excess time, i.e. no loading or discharge has been
performed in anchorage.
The inputs to run just-in-time arrival simulations as described in table 10 are route coordinates,
voyage start time, ship draught, and the chosen operation method is voyage arrival time. Based on
the interviews with Holmslykke (2019) all ships in the case study are capable to slow steam with
minimum speed of 6 knots. For maintaining the minimum speed of 6 knots in all the voyages, the
voyage as sailed simulations with the average speed under 6 knots were assumed to arrive the port
with an average speed of 6 knots and then wait the rest of the excess time at port. In these cases
the required arrival time was first obtained by running the simulations with the speed over ground
operation method set to 6 knots. The simulations were then run again by using the corresponding
arrival time with constant rpm to get the final results.
47
7.3.1 Overview of speed optimization algorithm
To meet the required arrival time the voyage is sailed with a constant rpm. Therefore, the speed profile
of the ship varies along the route from the effect of added resistance due to waves, wind, currents and
shallow water. In general, the required arrival time with a constant rpm is obtained with an iterative
process where several options of constant rpm value are calculated and the most suitable to meet the
arrival time is selected.
48
8 Analysis of the results
Total number of voyages simulated with NAPA VO is 1768, where 927 were laden and 840 were
ballast voyages. The results were calculated for two scenarios – one where only the laden voyages
were optimized and other where laden and ballast voyages were optimized. The two cases are
examined separately in the following subsections. The fuel consumption in with different voyage
simulation scenarios are compared by plotting the fuel consumption in relation to sailing time.
Descriptive statistics are used to express the reduction in CO2 emissions and fuel consumption and in
the consequential savings in fuel cost. The cost of the bunker is estimated by using the global 20 ports
average between January 2018 and end of March 2019, which is 430.5 USD for high sulfur IFO380
fuel oil (Ship and Bunker, 2019).
Voyage as Sailed
Duration [h] Fuel consumption [t] CO2 [t] Port time [h] Bunker cost [USD]
Sum 306 851.9 196 776.9 612 763.2 119 469.2 84 909 227.3
Mean 6 262.3 4 015.9 12 505.4 2 438.1 1 732 841.4
Median 6 174.5 3 834.9 11 941.8 2 365.2 1 654 744.6
Min 4 594.8 2 067.6 6 438.4 1 269.5 892 160.4
Max 8 526.4 6 218.3 19 363.8 3 990.8 2 683 194.7
Std. Dev 779.2 958.0 2 983.2 609.6 413 374.8
Just-in-Time arrival
Fuel saving [t] CO2 reduction [t] Port time saving [h] Bunker cost saving [USD]
Sum 22 539.0 70 146.2 25 355.3 9 725 576.8
Mean 460.0 1 431.6 517.5 198 481.2
Median 449.7 1 399.7 464.5 194 049.3
Min 138.1 429.2 0.0 59 584.7
Max 841.7 2 619.0 1 721.6 363 206.3
Std. Dev 179.4 558.4 412.2 77 398.4
49
Just-in-Time arrival & Weather Routing
Fuel saving [t] CO2 reduction [t] Bunker cost saving [USD]
Sum 32 315.4 100 596.2 13 944 099.8
Mean 659.5 2 053.0 284 573.5
Median 654.9 2 038.4 282 572.3
Min 246.9 768.3 106 553.5
Max 1 048.0 3 262.4 452 205.9
Std. Dev 209.6 652.4 90 424.7
For the fleet of 49 ships the total saving in fuel costs by using just-in-time arrival is over 9.7 million
USD or 11.5%. The total saving in port time for the fleet is 25 355 hours or approximately 1 056 days
if all the waiting time at anchorage can be minimized. On average, each ship has stayed approximately
21.5 days at anchorage during the 15-month period. However, significant differences can be observed
between different ships as one ship did not get any port time saving and maximum saving was
approximately 71.7 days with a standard deviation of 17.2 days.
The total saving in fleet fuel costs by using just-in-time arrival and weather routing is 13.9 million
USD or 16.4%. The total saving in port time is the same as in just-in-time arrival scenario. By using
weather routing services, the savings compared to just-in-time arrival alone is 4.9 percentage units
greater, thus the weather routing alone has a potential to achieve significant savings.
In Figure 20 the fuel consumption and the sailing time of all simulated voyages are plotted. Reference
lines for each simulation scenarios are used to represent the change in fuel consumption when the
sailing time increases. In the right hand side plot, the change in Energy Efficiency Operational
Indicator is shown in different simulation scenarios.
The average reduction in EEOI compared to Voyage as Sailed scenario is 10.3 % by using just-in-time
arrival strategy. When weather routing is added, the reduction is 9.7%. The result is a good example
of the inconsistent nature of the EEOI as discussed in section 3.1.3 when simulations with weather
50
routing have remarkably lower fuel consumption, but higher average EEOI value because the total
distance sailed is longer.
Voyage as Sailed
Duration [h] Fuel consumption [t] CO2 [t] Port time [h] Bunker cost [USD]
Sum 306 851.9 196 776.9 612 763.2 119 469.2 84 909 227.3
Mean 6 262.3 4 015.9 12 505.4 2 438.1 1 732 841.4
Median 6 174.5 3 834.9 11 941.8 2 365.2 1 654 744.6
Min 4 594.8 2 067.6 6 438.4 1 269.5 892 160.4
Max 8 526.4 6 218.3 19 363.8 3 990.8 2 683 194.7
Std. Dev 779.2 958.0 2 983.2 609.6 413 374.8
Just-in-Time arrival
Fuel saving [t] CO2 reduction [t] Port time saving [h] Bunker cost saving [USD]
Sum 36 464.4 113 489.4 38 049.0 15 734 377.2
Mean 744.2 2 316.1 776.5 321 109.7
Median 758.9 2 361.3 687.3 327 468.1
Min 319.6 994.0 83.3 137 886.5
Max 1 213.0 3 776.2 2 297.0 523 412.5
Std. Dev 218.7 680.9 494.5 94 361.8
Just-in-Time arrival & Weather Routing
Fuel saving [t] CO2 reduction [t] Bunker cost saving [USD]
Sum 55 115.7 171 579.1 23 782 409.9
Mean 1 124.8 3 501.6 485 355.3
Median 1 095.4 3 410.4 472 669.0
Min 582.5 1 813.2 251 352.5
Max 1 797.1 5 594.8 775 453.0
Std. Dev 296.3 922.5 127 858.3
The total saving in fuel costs for the fleet by using just-in-time arrival is 15.7 million USD or 18.5%.
The total saving in port time when the waiting time in ballast voyages is added is 38 049 hours
or approximately 1 585 days if all the waiting time at anchorage can be minimized. Therefore, if
just-in-time arrival can be implemented in ballast voyages as well, the increase in total port time saving
is 33% compared to the previous scenario where only laden voyages were optimized. On average,
each ship has stayed approximately 32 days at anchorage during the 15-month period.The differences
between different ships vary from the minimum being approximately 3.5 days and maximum 96.7
days with a standard deviation of 25.4 days.
51
The total saving in fuel costs by using just-in-time arrival and weather routing is 23.8 million USD
or 28.0%. The difference when weather routing is added is 9.5 percentage units compared to the
just-in-time arrival alone. Figure 21 presents the relationship of fuel consumption and sailing time,
and the change in EEOI when both laden and ballast voyages are optimized.
The average reduction in EEOI compared to Voyage as Sailed scenario is 17.5% by using just-in-time
arrival strategy. When weather routing is added, the reduction is 16.9%. As in previous scenario, the
EEOI has lower average value for simulation results with higher fuel consumption. For comparison,
UMAS (2016) evaluates the EEOI value index reduction of MR tankers between 11-54% in speed
range of 12 to 6 knots compared to the 12.8 knots baseline when only speed is reduced. The results in
this study fall within this estimation range.
52
Figure 22 – Torm Time Charter Equivalent for the period under review (Torm, 2019).
The average TCE for Torm MR tankers for the period under review is 10 525 USD per day. The
freight earnings are estimated directly by using the fleet averate TCE, voyage costs, operating costs
and voyage duration. Figure 23 shows the change in fleet average TCE value when different voyage
optimization strategies are implemented. Ship specific values of the voyage cashflow estimation are
included in appendix B and appendix C.
Figure 23 – Esimated Time Charter Equivalent improvement by the implementation of voyage optimiza-
tion.
The increase in average TCE value when only laden voyages are optimized is 7.2% with just-in-time
arrival and 10.4% when weather routing is added. Respectively, when laden and ballast voyages are
optimized, the increase in average TCE value is 11.9% with just-in-time arrival and 17.8% when
weather routing is added.
53
simulation tools. The selected voyages are Eastbound Atlantic crossing, Westbound Atlantic crossing,
Eastbound Pacific crossing and Southbound Philippine Sea crossing. All voyages were sailed by a
different ship. Figure 24 visualizes the routes of the selected voyages.
(a) Voyage 1: Panama canal - Rotterdam. (b) Voyage 2: IJmuiden - Panama Canal.
All selected voyages were sailed in laden condition and the simulations were run with an assumption
that the fuel used was high sulphur Heavy Fuel Oil with a lower calorific value of 42.7 MJ/kg.
According to the noon reports, in Emission Control Areas such as coast of United States, the ships
have steamed by using Marine Gas Oil instead of Heavy Fuel Oil. In such cases a conversion factor
of 1.06 has been used to convert the energy content of MGO to correspond the energy content of HFO
(Brynolf, Andersson and Fridell, 2011).
The results from both simulations and the actual values from the noon report are presented in table 17.
The percentual difference between simulations and noon report values is included in the right column
next to the calculated value.
54
Table 17 – Voyage 1 comparison statistics.
The most significant difference between NAPA VO and Noon Report values is in the average rpm,
NAPA VO value being 16.1% lower. The most significant difference between SSW is the 23.1% lower
fuel consumption in SSW value. Figure 25 shows the time series of speed over ground, propeller rpm,
propulsion power and fuel consumption.
Figure 25 – Voyage 1 speed over ground, propeller rpm, propulsion power and fuel consumption.
The fuel consumption time series shows a distinct peak in NAPA VO value as the speed over ground
decreases the same time. This is most likely due to a bad weather conditions when a voluntary speed
loss is introduced. The plots of SSW simulation follow the speed profile and do not point out the
similar peak in the fuel consumption. Figure 26 shows the simulation histograms of propeller rpm,
propulsion power and Beaufort distribution during the voyage.
Figure 26 – Voyage 1 propeller rpm, propulsion power and Beaufort distribution histograms.
55
The Beaufort scale in Figure 26 shows that the ship has encountered a broad scale of weather
conditions and Beaufort number 7 has been the most prevalent condition. In the definition of Beaufort
scale the number 7 represents a wind speed between 13.9 and 17.1 m/s and significant waveheight
between 4.0 and 5.5 meters (WMO, 2012).
The results from both simulations and the actual values from the noon report are presented in table 19.
The percentual difference between simulations and noon report values is included in the right column
next to the calculated value.
The difference in fuel consumption between NAPA VO and the Noon Report is 11.3% whereas the
fuel consumption in SSW is 21.7% lower than in Noon Report. Figure 27 shows the time series of
speed over ground, propeller rpm, propulsion power and fuel consumption.
56
Figure 27 – Voyage 2 speed over ground, propeller rpm, propulsion power and fuel consumption.
The propeller rpm, propulsion power and fuel consumption follow a similar pattern between NAPA
VO and SSW simulations, but the propulsion power has higher magnitude in NAPA VO simulation.
Since the speed profile used in both simulations is the same, the total resistance in NAPA VO model
is higher compared to the SSW model, resulting higher average propulsion power. Figure 28 shows
the simulation histograms of propeller rpm, propulsion power and Beaufort distribution during the
voyage.
Figure 28 – Voyage 2 propeller rpm, propulsion power and Beaufort distribution histograms.
The propeller rpm is close to each other in both simulations. However, the propulsion power is higher
in NAPA VO simulations as discussed before. The weather conditions vary from Beafuort scale 1 to
6 which could be described as good weather conditions in the Atlantic Ocean between August and
September.
57
Table 20 – Voyage 3 details.
The results from both simulations and the actual values from the noon report are presented in table 21.
As seen from table 21, the fuel consumption calculated with NAPA VO is nearly the same as the
actual fuel consumption from the Noon Report, but the average power is significantly larger in NAPA
VO. Similar results are found between the SSW simulations and the noon report. Figure 29 shows the
time series of speed over ground, propeller rpm, propulsion power and fuel consumption.
Figure 29 – Voyage 3 speed over ground, propeller rpm, propulsion power and fuel consumption.
Figure 29 shows that the most significant difference between the two simulations is the magnitude in
propeller rpm. Compared to the values from Noon Report, the propeller model in NAPA VO is more
accurate than the propeller model in SSW.
58
Figure 30 shows the simulation histograms of propeller rpm, propulsion power and Beaufort distribu-
tion during the voyage.
Figure 30 – Voyage 3 propeller rpm, propulsion power and Beaufort distribution histograms.
Figure 30 shows that the difference in propeller rpm is distinct, but the results in propulsion power
are close to each other. The prevalent Beaufort number is 3 and overall the weather can be described
as a good weather conditions in the Pacific Ocean in September.
The results from both simulations and the actual values from the noon report are presented in table 23.
Figure 31 shows the time series of speed over ground, propeller rpm, propulsion power and fuel
consumption.
59
Figure 31 – Voyage 4 speed over ground, propeller rpm, propulsion power and fuel consumption.
Figure 32 shows the simulation histograms of propeller rpm, propulsion power and Beaufort distribu-
tion during the voyage.
Figure 32 – Voyage 4 propeller rpm, propulsion power and Beaufort distribution histograms.
Figure 32 shows that the propeller rpm and propulsion power is higher in SSW simulation. The SSW
simulation results are also closer to the actual values from Noon Report. The Beaufort distribution
varies from 0 to 6, number 4 being the prevalent value.
60
9 Benefit sharing
This section discusses the overall benefits from the implementation of voyage optimization and the
possible solutions to share the monetary benefits to different shipping stakeholders. The material is
collected qualitatively by interviewing experts from Torm A/S and from STM Validation Project.
According to Siwe (2019) the three conditions to successfully implement just-in-time operation are:
The greatest barrier preventing the implementation of just-in-time arrival operation is the first-come,
first-served policy used by some of the global ports. If the ship would arrive to a port where this
policy is used later than than the original ETA, she will lose her place in the queue. This can lead to a
situation where no excess time is saved and future operations will be adjourned. Therefore, the basis
of implementation of just-in-time arrival operations is to change the first-come, first-served habit in
individual ports. The situation can be problematic even in ports where time slots for port operations
are scheduled well beforehand. The contract determining the port operations and schedule is typically
agreed between the charterer and port, so the ship operator does not have control when negotiating
the change in arrival time. Therefore, the motivation to negotiate about change in arrival time should
come from the charterer’s side.
The ship performance prediction has an important role when estimating the energy saving from
just-in-time operation. It is important that the voyage optimization service provider is an independent
player so that the results are not biased by either stakeholders involved in the shipping process.
Reliable estimations would require reliable weather forecasts and reliable performance calculation
methods.
The most significant benefit for the ship operator when operating under a voyage charter party is the
reduction in bunker costs and emissions. The estimation in case study shows a reduction in bunker
costs up to 18.5% if all the voyages can be operated with just-in-time arrival strategy. Overall, the
slower speed can lead to improved safety and increase crew well-being as the ship is less likely to be
exposed to excessive motions.
Charterer
When operating under a voyage charter party, the charterer has no noteworthy monetary benefit
when implementing just-in-time operation, unless the ship operator and charterer agree to share the
61
monetary benefits acquired from the reduction in bunker cost. The charterer should have motivation
to propose a suitable benefit sharing scheme, since under a voyage charter party, it is for them the only
way to gain monetary benefit from just-in-time operation. When considering safety, the improved
cargo safety can be considered as a benefit for the charterer as well when there is lower risk for cargo
damage or loss.
Port
The direct benefit from just-in-time operation for ports is that there are less emissions produced at
port areas from running auxiliaries at anchorage. Sharing of reliable information on arrival times
could help ports to schedule port stays more carefully which can help to reduce the congestion in
ports by having fewer ships at same time in the area.
With reliable weather routing, ship operators can perform voyage optimization with a desired op-
timization objectives. In the case study where minimizing fuel costs was the primary objective,
the reduction in bunker cost shows up to 28% when weather routing is combined with just-in-time
operation. Weather awareness can help improving operational safety when excessive motions can
be predicted in bad weather conditions. Furthermore, weather awareness helps to ensure the crew
well-being and lowers the risk of cargo damage and loss.
Charterer
The benefit of weather routing for the charterer is the lower risk of cargo damage or loss. If charterers
are able to negotiate lower insurance costs, because the ship operator is using weather routing services,
the charterer could achieve monetary benefit as well.
Port
Reliable weather routing services can lead to more accurate information on arrival times. With more
accurate arrival information, the port schedule management could be done more carefully, resulting
in shorter waiting times.
62
Table 24 – Comparison of BIMCO voyage charter clauses (BIMCO, 2019a).
BIMCO
Charterer Ship operator
clause
- Gives the right to request operator to ad-
Virtual - Entitled to refuse charterer’s request if it
just ship’s speed to meet the specified time
arrival clause e.g. might compromise vessel’s safety
of arrival
- Compensates the additional time used on
the sea voyage with an agreed demurrage - Receives savings from lower bunker costs
rate (If there is no agreement, default 50% and receives demurrage from the charterer
of the initial demurrage rate shall apply)
- Shall do the utmost to obtain and share - Shall do the utmost to obtain and share
STM clause
information about ship’s arrival time information about ship’s arrival time
- Gives the right to request operator to ad-
- Receives savings from lower bunker costs
just ship’s speed to meet the specified time
and receives demurrage from the charterer
of arrival
- Compensates the additional time used on
the sea voyage with an agreed demurrage
rate (If there is no agreement, default 50%
of the initial demurrage rate shall apply)
In BIMCO Sea Traffic Management (STM) clause, the key element is the communication and
information sharing between the charterer and the ship opertator as it is described in BIMCO
(2019b): "The Sea Traffic Management system is designed to contribute to a safer, more efficient and
environmentally friendlier maritime sector through the development and use of information shared
between different stakeholders". Otherwise, the features of the STM clause follow the features with
BIMCO Virtual Arrival clause.
In the case study there was no information about the situations where demurrage claims were risen.
Hypothetically, if all voyages in the case study where excess port time was observed were cases
where demurrage claims were risen, the total demurrage received from the charterer would have
been approximately 28.5 million USD. If Virtual Arrival clause with the default 50% reduction in
demurrage rate was used in these cases, the combined saving in bunker cost and received demurrage
would have been approximately 28.2 million USD. The calculation is shown in table 25.
63
Table 25 – Calculation of monetary benefit sharing with virtual arrival policy (laden voyages optimized).
Total bunker cost without optimization: Total bunker cost with virtual arrival clause
Total demurrage received 28.54 M$ Total demurrage received 14.27 M$
Total bunker cost 84.91 M$ Total bunker cost after optimization 70.97 M$
Bunker cost after compensation 56.37 M$ Bunker cost after compensation 56.70 M$
In the calculation shown in table 25, even the default share of 50% reduction in demurrage rate comes
close to the demurrage received in a case where no optimization was implemented. Therefore, by
negotiating the demurrage reduction to satisfy both stakeholders, the benefits from the implementation
of Virtual Arrival would be more balanced.
Table 26 – Calculation of monetary benefit sharing by sharing the saving in bunker cost (laden voyages
optimized).
From the ship operator’s point of view, this procedure would be more cost effective, whereas the
charterer can gain more benefit from the previous scenario, when there is a reduction in demurrage
rate. This procedure would require a reliable estimation of the saving potential when the arrival time
is changed. Therefore, services from an independent voyage optimization service provider would be
recommended in the calculation of the bunker saving potential.
64
on trust. Both stakeholders need to trust each other in order to arrange a benefit distribution scheme
which benefits both stakeholders. As voyage charters are sailed typically on a short term basis, there
is no possibility to create trustworthy relationships. (Siwe, 2019).
The advantage of benefit sharing with a reduction in demurrage rate is that the savings are not based
directly on an estimate of potential savings in bunker cost. The ship operator and charterer have,
therefore, more room to negotiate a reasonable reduction in the reduction in demurrage rate without
involvement of third party service provider estimating the saving potential in the bunker. The most
cost effective model for benefit sharing for the ship operator would be by simply sharing the savings
in bunker cost between the charterer. However, this would require involvement of a reliable service
provider to estimate the saving potential.
The international regulations on shipping should have amendments to ensure that ship operators
share their voyage plans with a relevant trusted parties. In practice, this could be enforced by adding
amendments to the SOLAS and IMO regulation related to voyage planning (Singhota, 2019). For
example, SOLAS Chapter V - Regulation 34 states:"Prior proceeding to sea, the master shall ensure
that the intended voyage has been planned (IMO, 2014)." However, currently the voyage plan is kept
as ship operator’s own information, although, the technology to share the information with trusted
parties exists.
To get all global ports to change from first-come, first-served policy to a scheduled port operations a
change in local customs and operation models is required. This may be the hardest barrier to overturn
when trying to change the current habits in shipping to allow more efficient operations. Nevertheless,
information sharing and more transparent shipping operations could change the habits in shipping
in a right direction. According to Siwe (2019), for authorities such as Vessel Traffic Services the
implementation of STM would be the biggest turning point since the implementation of AIS – with
the AIS the authorities became aware of the name of the ship and by the implementation of STM the
authorities will become aware of the intentions of the ship. Especially, a change of course in some
major global ports could have a knock-on effect on changing the habits globally.
65
10 Discussion and conclusions
The first part of this thesis introduced the big picture of shipping processes by discussing the role
of shipping stakeholders, operational processes and contractual terms used in shipping. The present
situation of benefit sharing and the problem of conflicting incentives between shipping stakeholders
were exemplified through hypothetical operational scenarios. The aspects of voyage optimization in
terms of passage planning and voyage scheduling were introduced together with different voyage
optimization strategies.
In the second part, voyage simulations were successfully run by using a ship performance model,
weather data, sea-going voyage data and data from nautical charts. Different methods to estimate total
ship resistance and to create sub-models, such as propeller model and wind model, were introduced.
Simulations were run for 1767 voyages sailed by 49 ships with three different simulation scenarios by
using NAPA Voyage Optimization-tool with ship performance models integrated in NAPA system.
Additionally, four selected voyages were simulated by using DTU Ship Simulation Workbench tool by
using the same weather data and sea-going voyage data as in simulations performed with Napa Voyage
Optimization. In this case, the ship performance models were created by the user entered parameters.
The simulation results of selected voyages from both simulation tools were then compared with actual
values from Noon Reports.
The results from the case study shows a 11.5% reduction in emissions and bunker cost when just-in-
time arrival is implemented to laden voyages. When weather routing is applied to these voyages, the
reduction increases to 16.4%. The room for optimization focuses mainly on laden voyages, because
there are less uncertainties when the details on the voyage schedule are determined in the charter
party. When both, the laden and ballast voyages are optimized with just-in-time arrival, the reduction
in bunker cost and emissions are 18.5%. When weather routing is applied, the reduction increases to
28.0%. However, to obtain savings of such magnitude in tramp shipping can be extremely difficult
and requires accurate prediction of the market development and the demand for transportation.
The result comparison on selected voyages showed relative difference from 0.4 to 16.2% in fuel
consumption between NAPA Voyage Optimization simulations and values from Noon Reports. For
simulations ran with DTU Ship Simulation Workbench, the relative difference in fuel consumption
varied from 5.2 to 23.1%. Therefore, there is room for further validation in the ship performance
calculation to achieve more accurate results. However, for approximate ship performance models, the
results do not seem overly alarming and the accuracy in the ship performance model can be enhanced
by using additional data sources, such as noon reports and automation measurements onboard.
The overall benefits of voyage optimization are related to lower emissions, lower bunker cost,
improved safety and less congestion in the port areas. For the ship operator, the major benefit when
operating under a voyage charter party is the reduction in bunker costs and emissions. The charterer
can benefit from voyage optimization if the savings can be shared e.g. by entering the virtual arrival
clause or by using other benefit distribution scheme. The benefits for ports are less emissions produced
in the port area and less congestion. The accurate information on ship arrival times can benefit the
port time scheduling.
The successful implementation of voyage optimization in a global scale requires a change in the
prevailing habit of first-come, first-served policy. Reliable estimations on energy saving and real-time
information on arrival times are considered as the key factors towards the change. However, when
66
the voyage plan is done by using a reliable weather routing service, the ship operator can achieve
significant savings without changes in the contracts between different shipping stakeholders.
Further validation in ship performance models can be done, for example, by using additional data
sources and big data methods. The enhanced models would increase the accuracy of the results and
provide more reliable performance predictions. Especially, detailed information on hull fouling can
increase the accuracy of the resistance model. Furthermore, the anchoring identification described
in section 5.3.5 can be automated, increasing the accuracy of extra time spent in anchorage. The
information of ports utilizing ship to ship loading/discharge integrated in the port database would
improve the overall reliability of the excess time estimation further.
67
11 Bibliography
References
Arribas, F. Pérez. 2007. “Some methods to obtain the added resistance of a ship advancing in waves.”
Ocean Engineering 34(7):946 – 955.
BIMCO. 2019b. “Sea traffic management (STM) clause for voyage charter
parties.” https://www.bimco.org/contracts-and-clauses/bimco-clauses/
sea-traffic-management-clause-for-voyage-charter-parties. Accessed: 2019-
2-27.
Breskin, I. 2018. The Business of Shipping. 9th ed. Schiffer Publishing, Ltd.
Brynolf, Selma, Karin Andersson and Erik Fridell. 2011. “A comparative life cycle assessment
of marine fuels.” Proceedings of the Institution of Mechanical Engineers Part M-Journal of
Engineering for the Maritime Environment 225:97–110.
Chang, Yu-Chia, Ruo-Shan Tseng, Guan-Yu Chen, Peter C Chu and Yung-Ting Shen. 2013. “Ship
Routing Utilizing Strong Ocean Currents.” Journal of Navigation 66(6):825–835.
Chen, Henry. 2013. “Weather routing vs voyage optimisation.” Digital Ship . accessed: 2019-5-7.
Gerritsma, J and W Beukelman. 1972. “Analysis of the resistance increase in waves of a fast cargo
ship.” International Shipbuilding Progress 19:285–293.
Hellenic Shipping News Worldwide. 2015. “Time Charter Equivalent (TCE) Earn-
ings.” https://www.hellenicshippingnews.com/wp-content/uploads/2015/04/
Copy-of-Special-Report-TCE-ENG.pdf. Accessed: 2019-6-26.
Holtrop, J and G.G.J. Mennen. 1982. “An approximate power prediction method.” International
Shipbuilding Progress 29:166–170.
IMO. 2009a. “Guidelines for voluntary use of the ship energy efficiency operational indica-
tor (EEOI).” http://www.imo.org/en/OurWork/Environment/PollutionPrevention/
AirPollution/Documents/Technical%20and%20Operational%20Measures/MEPC.1_Circ.
684_Guidelines%20for%20Voluntary%20use%20of%20EEOI.pdf. Accessed: 2019-3-8.
IMO. 2009b. “Second IMO GHG Study 2009.” http://www.imo.org/en/
OurWork/Environment/PollutionPrevention/AirPollution/Documents/
SecondIMOGHGStudy2009.pdf. Accessed: 2019-2-27.
IMO. 2014. Solas 2014: International Convention for the Safety of Life at Sea, 1974. IMO Publishing.
IMO. 2016a. “IMO Train the Trainer Course on Energy Efficient Ship Operation – Module 2.”
http://www.imo.org/en/OurWork/Environment/PollutionPrevention/AirPollution/
Documents/Air%20pollution/M2%20EE%20regulations%20and%20guidelines%20final.
pdf. Accessed: 2019-7-2.
IMO. 2016b. “IMO Train the Trainer Course on Energy Efficient Ship Operation – Module 3.”
http://www.imo.org/en/OurWork/Environment/PollutionPrevention/AirPollution/
Documents/Air%20pollution/M3%20from%20management%20to%20operation%20final.
pdf. Accessed: 2019-2-27.
ITTC. 2014. “Recommended procedures and guidelines – Analysis of speed/power trial data.”.
Jaskolski, Krzysztof. 2017. “Automatic Identification System (AIS) dynamic data estimation based
on discrete Kalman Filter (KF) algorithm.” Zeszyty Naukowe Akademii Marynarki Wojennej .
Jia, Haiying, Roar Adland, Vishnu Prakash and Tristan Smith. 2017. “Energy efficiency with the
application of Virtual Arrival policy.” Transportation Research Part D: Transport and Environment
54:50–60.
Jia, Haiying, Vishnu Prakash and Tristan Smith. 2019. “Estimating vessel payloads in bulk shipping
using AIS data.” International Journal of Shipping and Transport Logistics 11:25.
Laajarinne, M. 2014. Notice of Readiness and Tanker Charter Parties. Master’s thesis Novia University
of Applied Sciences.
Lackenby, H. 1963. “The Effect of Shallow Water on Ship Speed.” The Shipbuilder and marine
engine-builder 70(672).
Larsson, Lars, Raven Hoyte C. 2010. Principles of Naval Architecture Series - Ship Resistance and
Flow. Society of Naval Architects and Marine Engineers (SNAME).
Lu, Ruihua, Osman Turan, Evangelos Boulougouris, Charlotte Banks and Atilla Incecik. 2015. “A
semi-empirical ship operational performance prediction model for voyage optimization towards
energy efficient shipping.” Ocean Engineering 110:18 – 28. Energy Efficient Ship Design and
Operations.
Schofield, J. 1997. “Laytime and demurrage, 3rd edition.” Journal of Maritime Law and Commerce
28(3).
Sen, Debabrata and Chinmaya P. Padhy. 2015. “An approach for development of a ship routing
algorithm for application in the North Indian Ocean region.” Applied Ocean Research 50:173 –
191.
Ship and Bunker. 2019. “Global 20 ports average bunker prices.” https://shipandbunker.com/
prices/av/global/av-g20-global-20-ports-average. Accessed: 2019-7-8.
Tu, E., G. Zhang, L. Rachmawati, E. Rajabally and G. Huang. 2018. “Exploiting AIS Data for
Intelligent Maritime Navigation: A Comprehensive Survey From Data to Methodology.” IEEE
Transactions on Intelligent Transportation Systems 19(5):1559–1582.
UMAS. 2016. “CO2 Emissions from International Shipping. Possible reduction targets and their
associated pathways.” https://u-mas.co.uk/LinkClick.aspx?fileticket=na3ZeJ8Vp1Y%
3D&portalid=0. Accessed: 2019-6-28.
Walther, Laura, Anisa Rizvanolli, Mareike Wendebourg and Carlos Jahn. 2016. “Modeling and
Optimization Algorithms in Ship Weather Routing.” International Journal of e-Navigation and
Maritime Economy 4:31 – 45.
Wiesmann, Andreas. 2010. “Slow steaming – a viable long-term option?” Wärtsilä Technical Journal
pp. 49–55.
Wu, Lin, Yong-Jun Xu, Qi Wang, Fei Wang and Zhiwei Xu. 2017. “Mapping Global Shipping Density
from AIS Data.” Journal of Navigation 70:67–81.
DTU Mechanical Engineering
Section of Fluid Mechanics, Coastal and Maritime Engineering
Technical University of Denmark
www.mek.dtu.dk
July 2019